Yew-Kwang NG - Efficiency, Equality and Public Policy - With A Case For Higher Public Spending - Palgrave Macmillan (2000)

Efficiency, Equality and
Public Policy
With a Case for Higher Public Spending
Yew-Kwang Ng
Efficiency, Equality and Public Policy
Also by Yew-Kwang Ng
INCREASING RETURNS AND ECONOMIC ANALYSIS (editor with K. Arrow and

X. Yang)
MESOECONOMICS: A MICRO–MACRO ANALYSIS
SOCIAL WELFARE AND ECONOMIC POLICY
SPECIALIZATION AND ECONOMIC ORGANIZATION (with X. Yang)
WELFARE ECONOMICS
Efficiency, Equality and
Public Policy
With a Case for Higher Public Spending
Yew-Kwang Ng
Professor of Economics
Monash University
Victoria
Australia
First published in Great Britain 2000 by
MACMILLAN PRESS LTD
Houndmills, Basingstoke, Hampshire RG21 6XS and London
Companies and representatives throughout the world
A catalogue record for this book is available from the British Library.
ISBN 0–333–67165–1
First published in the United States of America 2000 by

ST. MARTIN’S PRESS, INC.,
Scholarly and Reference Division,
175 Fifth Avenue, New York, N.Y. 10010
ISBN 0–312–23208–X
Library of Congress Cataloging-in-Publication Data
Ng, Yew-Kwang.
Efficiency, equality and public policy : with a case for higher public spending /
Yew-Kwang Ng.
p. cm.
Includes bibliographical references and index.
ISBN 0–312–23208–X
1. Policy sciences. 2. Economic policy. 3. Expenditures, Public—Decision mak-
ing. I. Title.
H97 .N5 2000
338.9—dc21
99–056733
© Yew-Kwang Ng 2000
All rights reserved. No reproduction, copy or transmission of this publication may be made
without written permission.
No paragraph of this publication may be reproduced, copied or transmitted save with written
permission or in accordance with the provisions of the Copyright, Designs and Patents Act
1988, or under the terms of any licence permitting limited copying issued by the Copyright
Licensing Agency, 90 Tottenham Court Road, London W1P 0LP.
Any person who does any unauthorised act in relation to this publication may be liable to
criminal prosecution and civil claims for damages.
The author has asserted his right to be identified as the author of this work in accordance with
the Copyright, Designs and Patents Act 1988.
This book is printed on paper suitable for recycling and made from fully managed and sustained
forest sources.
10 9 8 7 6 5 4 3 2 1
09 08 07 06 05 04 03 02 01 00
Printed and bound in Great Britain by

Antony Rowe Ltd, Chippenham, Wiltshire
To
Siang, Aline, Eve, and Tess
This page intentionally left blank
Contents
Preface x
1 Introduction 1
1.1 A dollar is a dollar: a simple solution to the big
efficiency–equality tradeoff 1
1.2 A partial resurrection of the old ‘new welfare
economics’? 1
1.3 Introductory summary 2
1.4 A case for higher public spending 4
1.5 The misguided consensus 5
1.6 Three basic problems of social choice/public policy 7
1.7 Some specific points made 9
1.8 A methodological note 9
1.9 What this book is not about 10
Part I The Foundation of Public Economic Policy 13

2 The Necessity of Interpersonal Cardinal Utility 15
2.1 An intuitive explanation using a parent’s choice 15
2.2 Economists’ misplaced hostility against cardinal
utility 17
2.3 The impossibility of social choice based on ordinal
utilities 21
3 Welfarism 24
3.1 The Sen–Ng debate on welfarism: an appraisal 24
3.2 Another argument for welfarism 30
3.3 Rational individualism implies welfarism 32
4 Utility, Informed Preference, or Happiness 36

4.1 Non-affective altruism: the pure concern for the
welfare of others 37
4.2 Irrational preferences 45
4.3 Autonomous desires 52
4.4 Why is happiness fundamental? 54
5 Utilitarianism 57
5.1 Compelling arguments for utilitarianism 57
vii
viii Contents
5.2 Rational individualistic egalitarianism implies

utilitarianism 62
5.3 A defence of Harsanyi against some recent criticisms 66
5.4 A defence of using just perceptible increments 72
5.5 Utilitarianism and process fairness 76
6 A Dollar is a Dollar: Solution to the Paradox of

Interpersonal Cardinal Utility 82
6.1 The paradox of interpersonal cardinal utility 82
6.2 The proposed solution 83
6.3 Economists should be in favour of reversed weighting! 90
6.4 Some qualifications 92
6.5 Concluding remarks 97
7 Economics versus Politics 99
Part II How Much Should the Government Spend? 103

8 A Case for Higher Public Spending 105
8.1 Some simple theories of government spending 105
8.2 Inefficiency in public spending may increase its
optimal level 108
8.3 General taxation may be more corrective than
distortive 111
8.4 Diamond effects and burden-free taxes 111
8.5 Relative-income effects bias against public
expenditure 113
8.6 The unimportance of absolute income 115
8.7 Welfare-improving public expenditures 119
9 The Appropriate Benefit–Cost Ratio for Public Spending 124

9.1 The conventional view 124
9.2 Kaplow’s argument and the principle of ‘a dollar is
a dollar’ 126
9.3 Reconciling Kaplow and Feldstein 130
10 Concluding Remarks 135
Appendix A Electrical Brain Stimulation: a Case Showing the

Importance of Public Spending in Research 137
A1. Misconceptions on the costs of public spending 137
A2. Direct access to intense pleasure 137
Contents ix
A3. Enormous benefits 138

A4. Safe and long-lasting pleasure 138
Appendix B Pure Egalitarianism: a Critique 141

B1. Why is happiness the only ultimate consideration? 141
B2. Equality in the welfare weights 143
B3. The libertarians 143
B4. Rawls 144
B5. Pure and welfare egalitarianism 145
B6. Sen and capability egalitarianism 148
B7. Concluding remark 150
Appendix C Economic Growth Increases the Optimal Share of the

Public Sector 151
Appendix D Non-Affective Altruism: When the Pareto Principle is

Unacceptable 158
D1. The Pareto principle 158
D2. When the weak Pareto principle is unaccceptable 159
D3. The case of affective altruism 161
D4. Should voters be asked to suppress their non-affective
altruism? 162
Appendix E The Bergson–Samuelson Tradition Implies Individualism,

Independence, and Ordinalism 163
Appendix F Inefficiency in Provision May Increase Optimal Public

Spending 164
Notes 165
References 169
Name Index 184
Subject Index 188

Preface
This book addresses some important and fundamental problems of

public policy, arriving at remarkable and controversial conclusions. For
example, it is argued that pure efficiency (‘a dollar is a dollar’) should
rule in any specific issues, leaving the objective of equality to the general
tax/transfer system. While redistribution through the tax/transfer
system may generate substantial disincentive effects, doing so through
specific measures like taxing/subsidising goods consumed dispropor-
tionately by the rich/poor also has the same degree of disincentive
effects in accordance to its redistributive effects. In addition, the latter
measure has additional distortive effects and hence is inferior. I arrived
at this principle of ‘a dollar is a dollar’ in my attempt to prove it wrong.
A strong case for higher public spending, especially on research and
environmental protection is also made. Despite some inefficiency in
public spending, higher public spending is likely to be more welfare-
improving than private spending. In countries that are no longer poor,
further increases in private consumption fail to increase happiness.
The small consumption effect may be more than offset by the negative
environment disruption effect. People still engage in the rat race
for making more money mainly due to competition and the related
relative-income effects. Subjective well-being and objective quality-of-
life indicators correlate more with the increase in knowledge at the
global level than with income per capita. Public spending on research
and environmental protection undertaken by national governments is
much lower than optimal, partly because of the global and long-term
nature of these items. (A more detailed introduction and summary is
contained in Chapter 1.)
Elizabeth Kwok wordprocessed this book and purged the manuscript
of many mistakes and inconsistencies. Her patience and efficiency are
gratefully acknowledged. Tina Bell, Kate Orchard and other students in
my Welfare Economics class made useful presentational suggestions.
x
1
Introduction
1.1 A dollar is a dollar: a simple solution to the

big efficiency–equality tradeoff
As emphasised in the title of Okun’s (1975) book, efficiency versus

equality is the big tradeoff in public policy. It is probably the most
important problem in the foundation of public economic policy. It
appears in almost all fields of public economic policy. Yet, according
to my argument explained in this book, it should basically only be con-
fined to one area of public policy – the general tax/transfer system –
though it remains an important problem. In all other areas, efficiency
(the principle of ‘a dollar is a dollar’) should be allowed exclusive reign.
This provides the most efficient method of achieving whatever degree
of equality is aimed at, or whatever value of tradeoff allowed for. If
accepted, my proposal leads to a tremendous simplification in the for-
mulation of public policy in general and in cost–benefit analysis in
particular. Thus, no economist, at least no economist concerned with
policy issues, can afford to ignore this argument.
Before tackling the big efficiency–equality tradeoff, even more basic
issues on the appropriate ultimate objectives of public policy are first
discussed, as outlined in Section 1.3.
1.2 A partial resurrection of the old

‘new welfare economics’?
The old welfare economists (represented by Pigou, 1912/1932) assumed

cardinal utility and interpersonal comparability of utility. The ‘new’
welfare economists tried to dispense with both, but the attempt proved
unsuccessful (Suzumura, 1997). For example, the debate on compensa-
1
2 Efficiency, Equality and Public Policy
tion tests or welfare criteria in the 1940s and 1950s ended with the
ambivalent attitude towards the Little (1949, 1957) criterion. Though
I strongly defended (in Ng, 1979/1983, Chapter 3) the acceptability of
Little’s criterion (compensation test plus a distributional proviso), it
gives no answers to many cases and its distributional part requires inter-
personal comparison of cardinal utility. Thus, in its attempt to get rid
of cardinal utility and interpersonal comparison, the new welfare econ-
omics failed.
In this book, I argue that interpersonal comparison of cardinal utility
is necessary for making social choice. However, my proposal for
tackling the efficiency–equality tradeoff by treating a dollar as a dollar
whomsoever it goes to at every specific issue, leaving the pursuit of
equality to the general tax/transfer system means that cardinal utility
and interpersonal comparison are not needed except in determining
the optimal tradeoff between efficiency and equality in the general
tax/transfer system. Thus, in this sense, the objective of new welfare
economics in doing away with cardinal utility and interpersonal
comparison may be said to be largely (but not completely) met in our
proposal.
1.3 Introductory summary
An economic problem is typically the maximisation of the relevant

objective subject to certain constraints. This is true both for an indi-
vidual economic actor (consumer, worker, firm) as well as for the
whole society or the government. However, for the former, the relevant
objective is usually less controversial; the individual actor can decide
whatever objective she wants to maximise, though her decision may
be affected by legal, moral and social factors. For the case of the society,
what the objective should be is a heavily debated (not only by
economists but also by philosophers, political scientists, politicians,
religious leaders, and persons in the street) but apparently unsettled
issue.
The first objective of this book is to provide a compelling argument
that the ultimate social objective should be the maximisation of the
unweighted sum of individual cardinal utilities. (Here, we have largely
abstracted away complications of non-human sentients, national bound-
aries, and variable sets of people, problems which I have dealt with in
Ng, 1983, 1989a, 1995.) Chapter 2 argues that the beliefs of (apparently)
the majority of economists on the sufficiency of ordinal utilities and
the impossibility of interpersonal utility comparison are certainly mis-
Introduction 3
taken. Chapter 3 argues that the social objective or social welfare should
depend ultimately only on individual utilities (welfarism). Chapter 4
argues that, where individual utility (representing her preference) and
individual welfare (representing her happiness) differ, happiness should
be the ultimate concern. (Where I continue to use ‘utility’, it is partly
due to its more popular usage, partly to the presumption of no diver-
gence, and partly to distinguish it from ‘social welfare’ more clearly.)
Chapter 5 argues that social welfare should be an unweighted sum of
individual cardinal utilities/welfares (utilitarianism). Rational individu-
alistic egalitarianism alone is sufficient to yield utilitarianism. Harsanyi’s
(1953, 1955) arguments for utilitarianism are defended against their
recent critics and utilitarianism is defended against the criticism that it
ignores process fairness.
While interpersonal comparison of cardinal utilities is necessary for
social choice, it is in practice very difficult to obtain these utility indices
and to compare them interpersonally. A simple method to largely solve
this dilemma is proposed in Chapter 6. Basically, the willingness to pay
is used to reveal individual intensities of preference and the unweighted
sum of these figures (‘a dollar is a dollar’) is used as the criterion in order
to achieve efficiency, leaving the objective of equality to be tackled
in the general tax/transfer system where interpersonal comparison of
cardinal utility is still needed to achieve a rough trade-off (between
efficiency and equality). This achieves a tremendous simplification in
the formulation of public economic policy in general and in
cost–benefit analysis in particular. This principle of ‘a dollar is a dollar’
is consistent with the utilitarian objective but utilitarianism is not
necessary for the principle. This principle can be based just on the
normal Pareto principle. If my argument is correct, anyone accepting
the Pareto principle must accept the principle of ‘a dollar is a dollar’.
This rules out specific purely equality-oriented policies such as the use
of distributional weights in cost–benefit analysis, subsidies/taxes on
goods consumed mainly by the poor/rich, the first-come-first-served
method for car parking spaces to provide ‘equal access’ in city-centre
parking, etc. However, it does not rule out measures based on efficiency
considerations such as external effects, second-best factors, etc. For
example, subsidies on education and health care may be justified on the
grounds of external benefits but not on the ground of equality. It is not
in favour of more or less equality. Rather, for any degree of equality
aimed at, it is more efficient (in the Pareto sense) to achieve it by adopt-
ing the principle of ‘a dollar is a dollar’ and using the appropriate degree
of progressivity in the tax/transfer system.
The basic argument for ‘a dollar is a dollar’ is this. The objective of

equality is more efficiently achieved by the general tax/transfer system
since specific purely equality-oriented policies impose additional
inefficiency by distorting specific choices. It is true that the general tax/
transfer system also involves inefficiency in the form of disincentive
effects. However, specific purely equality-oriented policies also possess
the same disincentive effects (in addition to the specific distortive
effects). Rational individuals take account not only of the tax rates but
also the degree of preferential treatment in the expenditure system
(including the specific purely equality-oriented policies). For the same
degree of equality in real income (utility) achieved, the same degree of
disincentive effects applies whether it is achieved by tax/transfer alone
or complemented by specific equality-oriented policies. Thus, transfer-
ring some of the equality-oriented methods to specific efficiency-
inconsistent policies does not reduce the overall degree of disincentive
effect, it just adds on the unnecessary distortive costs. It is thus inferior.
My argument in favour of ‘a dollar is a dollar’ may be regarded as
extremely right-wing. However, I can prove that the conclusion was
derived by logic, not by ideology. I reached that conclusion starting out
from a somewhat left-wing presumption in favour of the poor (I was a
strong and active communist supporter as a teenager), while attempting
to prove ‘a dollar is a dollar’ wrong due to the diminishing marginal
utility of income, in my lunch-time disagreement with my colleague,
Professor Ross Parish. Moreover, ‘a dollar is a dollar’ is not quite ‘right-
wing’ in comparison to the apparently more outrageous argument made
in Chapter 6 (Section 6.3) that economists, as a group, should be in favour
of a reverse weighting system (counting a dollar to the poor as worth
less than a dollar to the rich). That they have not done so is explained
by their ignorance of the relevant argument for ‘a dollar is a dollar’.
Chapter 7, the last chapter of Part I, argues that the boundary of econ-
omics (where dollars rule), especially with politics (where ‘one person
one vote’ applies), is mainly sustained by the general insufficient under-
standing of economics. With the more widespread understanding of
economics, the sphere of economics should gradually expand, as it was
in the past.
1.4 A case for higher public spending
Part II discusses the appropriate size of government spending, an issue

regarded as the central problem in public finance by Feldstein (1997).
Contrary to the current world-wide trend towards privatisation and the
Introduction 5
attempt to reduce or to limit the growth in the size of government, a

strong case for higher public spending is made. Both happiness surveys
and quality-of-life indicators (discussed in Section 8.5) suggest that
higher (real and per capita) incomes are not as important to welfare as
the general advance in knowledge at the world level. The latter depends
more on government spending, especially on higher education and
research, than on private consumption. While absolute incomes are
unimportant, relative incomes remain important (if not increasingly
so), making people, the rich and the poor alike, still engage in the rat
race for making more money. Together with the significant environ-
mental disruptive effects of most consumption and production, this
may result in welfare-reducing growth. At higher levels of income, it is
likely that a higher proportion of the GNP should be used for public
spending to achieve efficiency, as environmental quality and most
public goods are not inferior goods. However, the importance of rela-
tive-income effects biases people in favour of private spending, making
the traditional estimate of the optimal size of public spending an under-
estimate. Together with the global public goods and long-term nature
of environmental quality and most basic research, this makes public
spending by myopic national governments well below the optimal
level. On top of this, the existence of other taxes with negative excess
burdens (or even burden-free taxes, e.g. on diamond goods – goods
valued for their values rather than their intrinsic consumption effects)
makes the true cost of public spending much less than commonly
believed by economists. In addition, Chapter 9 discusses the important
argument by Kaplow (1996). Since at least the time of Pigou (1928), vir-
tually all economists (myself included before I read Kaplow) accept the
view that the benefits of public goods should exceed their direct costs
by an amount sufficient to outweigh the excess burden or distortionary
costs of taxation. Kaplow argues that it is optimal to supply a public
good whenever the benefit–cost ratio exceeds one. While Kaplow’s
argument has to be subject to some qualifications, he is basically correct.
Large excess burdens may exist on the revenue side but are largely
offset by the negative excess burdens or distributional gains on the
spending side.
1.5 The misguided consensus
As noted by Amartya Sen (1979, p.537), Wassily Leontief has sum-

marised succinctly the normative properties ‘on which something like
a general consensus of opinion seems to exist’ in the formal discussion

of public economic policies:
In the discussion of public economic policies – the normative

character of the problem has been clearly and generally recognized.
There the mathematical approach has crystallized the analysis
around the axiomatic formulation of the (desirable or conventional)
properties of the ‘social welfare function’. Social utility is usually
postulated as a function of the ordinally described personal utility
levels attained by each of the individual members for the society in
question.
The only other property on which something like a general con-
sensus of opinion seems to exist is that ‘the social welfare is increased
whenever at least one of the individual utilities on which it depends
is raised while none is reduced’.
Leontief, 1966, p.27
The second property is of course the (strong) Pareto principle. The first
property is further factorised by Sen into the following three distinct
parts:
(1) Welfarism
Social welfare is a function of personal utility levels, so that any two
social states must be ranked entirely on the basis of personal utilities
in the respective states (irrespective of the non-utility features of the
states).
(2) Ordinalism
Only the ordinal properties of the individual utility functions are to
be used in social welfare judgements.
(3) Non-comparable Utilities
The social welfare ranking must be independent of the way the
utilities of different individuals compare with each other.
Sen, 1979, p.538
Sen then launched a vigorous attack on these four conditions. As usually

happens between friends, I agree partly with Sen but disagree with him
on the other half of his argument. Thus I wish to defend welfarism and
the Pareto principle but to question ordinalism and non-comparability.
In fact, the Pareto principle and welfarism are very close to each other.
No one accepting welfarism wants social welfare to be decreasing
instead of increasing in individual utilities. Thus, with this requirement
Introduction 7
on positivity, welfarism implies the Pareto principle. On the other hand,

the Pareto principle also implies welfarism. If individual utilities are not
the only elements affecting social welfare, then someone better off with
no one worse off cannot be a sufficient condition for an increase in
social welfare, since other factors (unless lexicographically less impor-
tant than individual utilities) affecting social welfare could decrease sub-
stantially to offset this. Thus, ignoring factors less important than 0.000
. . . 1 util, our defence of welfarism in Chapter 3 is also a defence of the
Pareto principle.
While ordinalism is sufficient (and hence preferable to cardinalism on
the ground of Occam’s razor) for the theory of consumer demand, it is
certainly insufficient for social choice (Chapter 2). Graduating from
cardinalism to ordinalism is a significant methodological advance in
consumer theory. Economists, in emphasising this advance, have taught
their students to reject cardinalism even in social choice. This is the
fallacy of misplaced abstraction (Ng, 1979/1983, pp.13–15). Denying
the use of cardinal utility in areas (social choice, optimal population,
valuation of life, choices involving risks) where it is useful is like insist-
ing that a person must shave off his moustache on the ground that it
is unnecessary for eating, ignoring the fact the he may want to keep it
to increase his sex appeal.
1.6 Three basic problems of social choice/public policy
If we regard issues like strategic behaviour, incentive compatibility,

mechanism design as advanced topics, we may identify the following
three basic social choice or public policy problems.
I. What types of information are relevant and needed in making

social choice?
Most economists accept the Fundamental Value Proposition of Individ-
ual Preferences (Bergson, 1938), rejecting the existence of a mythical
society or state independent of, or over and above the various individ-
uals composing the society, and thus regard the relevant information as
individual preferences. This must also be the case if we accept the widely
used Pareto Principle. However, as noted above, Sen (1979) calls this
welfarism and attacks it as ignoring principles as no-exploitation, no-
torture, etc. In Chapter 3, welfarism is defended as compelling if applied
at the fundamental level, viewing moral and legal principles as instru-
ments for the promotion of individual welfares.
Arrow’s (1951/1963) theorem established the impossibility of a

reasonable social choice based on individual ordinal preferences within
a multi-profile (or inter-profile) framework where alternative profiles of
individual preferences are considered. Kemp and Ng (1976) and Parks
(1976) (generalised by Roberts, 1980) show the same impossibility
within the single or intra-profile framework where individual prefer-
ences are unchanged. Sen (1970a) shows the impossibility of social
choice without interpersonal comparison. It is thus well-established that
reasonable social choice requires the interpersonal comparison of indi-
vidual cardinal utilities. This necessity of interpersonal cardinal utilities
is explained in non-technical terms in Chapter 2.
II. How can we obtain the required information?

Following from the solution to the first basic problem, the required infor-
mation is the interpersonally comparable individual cardinal utilities.
The problem of how these may be obtained has been largely ignored by
social choice theorists and economists in general. However, I have pro-
posed the use of a maximal indifference (akin to Edgeworth’s concept of
a just perceptible increment of pleasure) as an interpersonally compa-
rable unit of utility and discussed ways of practical measurement (Ng,
1975a, 1996a). I have also shown that the cardinal utility index obtained
by using the Neumann–Morgenstern expected utility method is in fact
the subjective utility index of the neoclassical economists and hence is
a suitable ingredient for normative social choice (Ng, 1984a). Contribu-
tions from non-social choice theorists may also help. (See, e.g. van Praag,
1968, 1991; Veenhoven, 1984, 1993.) Nevertheless, a lot more work
should be done to answer this second basic question.
III. Which form of social welfare function (SWF)

should be adopted?
Even assuming that we agree that the relevant information is interper-
sonal comparable cardinal utilities and somehow have obtained them,
there is still the problem of how these data are to be aggregated to form
the social preference or to define the social welfare index. Should the
SWF be an unweighted sum of these interpersonally comparable indi-
vidual cardinal utilities (defined as utilitarianism) or a product or some
other form? This third basic problem is answered in Chapter 5 where it
is argued that compelling axioms force us to accept utilitarianism.
Apparently compelling objections to utilitarianism are shown to be
unacceptable.
Introduction 9
1.7 Some specific points made
Some other specific points argued in this book include the following:
• More research should be done to perfect the methods of electrical

and other forms of brain stimulation (known for nearly half a
century) to make them safe for common use to achieve a quantum
leap in welfare. (See Appendix A.)
• Higher public spending may be needed if only to raise the remuner-
ation of those working in the public sector to attract top talents back
from the private sector. (See Section 8.8.)
• Non-affective altruism (or malice), or altruism in reason rather
than in emotion, not only makes it necessary to use individual
welfares instead of utilities as what public policies should ultimately
be concerned with (Section 4.1), but may also make the otherwise
compelling Pareto principle unacceptable in certain cases. (See
Appendix D.)
1.8 A methodological note
(General readers may skip this subsection.)

Most of the arguments made in this book are positive. (Substantial nor-
mative arguments are also made in Chapters 3 to 5.) However, many
specific points made also involve normative judgments, concerning
what should be done. For example, the case in favour of higher public
spending on research and environmental protection is clearly prescrip-
tive. However, the gulf between positive analysis and normative
prescription is not that big if we accept the following convention, as
I do for this book and most of my other writings. Whenever some
prescriptive/normative statement is made, e.g. public spending should
be increased, it means that the prescribed action (e.g. higher public
spending under whatever specified conditions) will increase social
welfare which is an increasing function of individual utilities/welfares.
Usually, this is sufficient (especially in view of the principle of ‘a dollar
is a dollar’; see Chapter 6). (Where a specific functional form is needed,
then the unweighted sum is used, as amply argued for in Chapter 5.)
With this convention, an apparently prescriptive statement (X should
be done) is made equivalent to a more positive one (X will increase
social welfare or X will increase the sum of individual utilities.) Whether
it is desirable to increase social welfare so conceived remains a norma-
tive issue. (Arguments in its favour are offered in Chapters 3 to 5.) On
the cardinal measurability and interpersonal comparability of individ-
ual utilities/welfares, see Chapter 2. With this convention, it means that

one may differ with my prescriptive/normative statements due to dis-
agreement with my positive analysis and/or to disagreement with my
concept of social welfare. However, with this convention, I cannot be
accused of confusing positive analysis with normative judgments. It is
elementary (Hume’s law) that any prescription must be based on some
normative judgments.
It is true that, due to the difficulties of measuring and comparing indi-
vidual utilities (substantially lessened however by the principle of ‘a
dollar is a dollar’), many prescriptive statements are not just supported
by the above convention and some positive analysis, but also by some
subjective judgments of fact. However, as I argued in Ng (1972), econ-
omists are more qualified in making those subjective judgments of
fact (which are usually confused with value judgements proper) that
are closely related to their field of study, than non-economists, though
not with respect to value judgments proper. Recognising the distinc-
tion between value judgments proper and subjective judgments of
fact thus substantially increases the role of economists in policy
recommendation.
1.9 What this book is not about
Generally speaking, this book is concerned with the more general issues
of public policy at the more fundamental level. Thus, it is concerned
with what public policy should ultimately aim at rather than with sec-
ondary principles and specific issues. It is concerned with the ultimate
ethical foundation of public policy rather than with issues at the
practical or political levels. While it discusses the problem of public
spending and names some specific areas (research and environmental
protection), it is still mainly concerned with the more general and fun-
damental issues than with specific items and complications of imple-
mentation and political feasibility. This does not mean that such issues
are regarded as unimportant; rather, they are beyond the scope of this
book and also largely beyond the expertise of the author. While issues
of feasibility and practical difficulties have to be faced at the stage of
actual policy decisions, the issue of basic objectives and fundamental
principles remains important and relevant even at the implementation
level.
This book avoids unnecessary formalism and technical analysis.
Virtually no mathematics is used except in some appendices. While by
no means against the use of rigorous mathematical techniques (as I have
Introduction 11
used them myself), I believe that their use has become excessive and is
unnecessary in many cases. Two examples related to the theme here
may be given. It is compelling, at the ultimate level, that the desir-
ability of a social state depends only on all its relevant variables or
characteristics (rather than that of another social state), due to the
exhaustive nature of ‘all relevant variables’ and the mutually exclusive
nature of social states. This makes the pure element of independence a
compelling condition. Then, analysis is sufficient by using the concept
of a social welfare function (with the social welfare of any social state
depending only on variables at that social state) instead of that of a
social welfare functional (with the social welfare of any social state
depending on the variables of all social states). Though it is formally
more general to conduct the analysis in terms of a social welfare func-
tional and then impose the compelling condition of independence, this
makes the analysis unnecessarily more complicated and less accessible
to people unfamiliar with the complicated technical framework. As
another example, when stripped of its technical facade, the essence of
the (valid) argument by Roemer (1986, 1996) that resource egalitarian-
ism implies welfare egalitarianism may be seen in a simple diagram (see
Appendix B). Thus presented, the rationale for the validity of the argu-
ment, the ethical unacceptability of both resource egalitarianism and
welfare egalitarianism are all clear. However, I am sympathetic towards
Roemer (and others) for their presentation of arguments in complicated
mathematics; it is increasingly difficult to publish academic papers not
using complicated mathematics. If Roemer’s argument were originally
presented in the transparent form of a simple diagram, it would almost
certainly have been rejected by the Quarterly Journal of Economics. When
formalism has progressed to such a level, a reconsideration is in order.
(For a critique of formalism in modern economics, see Blaug, 1998; for
a discussion, see Backhouse, 1998; Chick, 1998; Krugman, 1998 and
Weintraub, 1998.)
Part I
The Foundation of Public
Economic Policy
2
The Necessity of Interpersonal
Cardinal Utility
2.1 An intuitive explanation using a parent’s choice
Suppose you are a parent having to make certain decisions affecting

the well-being of the whole family. Assume that no one else (including
animals) other than your family members will be affected by your
decisions. You do not hate any of the members and you care about
all members. Thus, you believe that your decisions should depend on
their (including your own, being a family member yourself) utilities.
You also agree that if any one member is made better off with no
member made worse off, it is a better situation. In other words, you
accept the Pareto principle. This is sufficient for our presupposition.
Some readers may object that the Pareto principle may not be sufficient
here because, for example, one may want one’s children to have the
right moral principles and training, and be willing to sacrifice the utility
of some members in exchange. In my view, this is due to the favourable
effects of the right principles and training on one’s future well-being
and/or on that of other individuals. I believe most parents accept
this. Either including such favourable effects in the calculation or
abstracting from these complications here, the Pareto principle should
be acceptable to most people. In any case, this issue is discussed in
Chapter 3.
In one trivial set of cases, interpersonal (i.e. between family members
here) comparison of cardinal utility is not needed for your choice. This
is when your decisions always affect the well-being of all your family
member in the same direction. If you have only one decision of choos-
ing between x and y and all members are better off in x than in y, then
it is clear that you should choose x. Here, adherence to the Pareto prin-
ciple solves your problem of family choice.
15
In most cases, conflicts arise in the sense that the choice of one alter-
native over the other makes some members better off but some other
members worse off. It is such cases of conflict that make the problem
of social choice of substance and interesting. For simplicity, assume that
you and your spouse are always indifferent over the relevant alterna-
tives and hence you choose purely in accordance to the preferences of
the children. There is another trivial case where interpersonal compari-
son of cardinal utility is not needed. If one of the children is your
favoured child such that you let him/her choose, ignoring the prefer-
ence of other children, then your family choice problem is solved by
the dictatorial choice of this favoured child. (This is ruled out by Arrow’s
condition of non-dictatorship.)
You might adopt the simple rule that, for divisible choices, always
divide equally between all the children and for indivisible choices,
always toss a coin to decide. This may ‘work’ and appears ‘fair’. However,
it violates the weak Pareto principle (which says that if all individuals
prefer x to y, the society or the family should prefer x to y). For sim-
plicity, take the case of only two children, Jan and Kelvin, and only one
divisible good, money, to be allocated. The principle involved is applic-
able to general cases.
Suppose Jan prefers hot weather if she has enough money ($75) to
enable her to go surfing but would rather have cold weather with
less money than enduring the heat without surfing. On the other
hand, Kelvin prefers cold weather if he has enough money ($75) to
go skiing, but would rather have hot weather with less money than
watching the snow without being able to ski. Now consider the
following four alternatives H1, H2, C1, C2, each pair corresponding to
two different distributions of the given $100 in two alternative weather
conditions:
H1 = (Jan has $75, Kelvin has $25, Hot weather)

H2 = (Jan has $50, Kelvin has $50, Hot weather)
C1 = (Jan has $25, Kelvin has $75, Cold weather)
C2 = (Jan has $50, Kelvin has $50, Cold weather)
The preference patterns of Jan and Kelvin described above may be

illustrated in Figure 2.1 which readers may wish to use for reference.
The rule of dividing money equally between Jan and Kelvin means that
H2 is (socially) preferred to H1 and C2 is preferred to C1. However, both
Jan and Kelvin prefer C1 to H2 and both also prefer H1 to C2. Thus,
combining the egalitarian division rule with the weak Pareto principle,
The Necessity of Interpersonal Cardinal Utility 17
UK
C1
H2
H1
C2
UJ
Figure 2.1
we have the following preference cycle: C1 preferred to H2 (weak

Pareto), H2 preferred to H1 (egalitarian rule), H1 preferred to C2 (weak
Pareto), and C2 preferred to C1 (egalitarian rule). This means that the
use of the egalitarian rule violates the weak Pareto principle. This viol-
ation may be given the following intuitive explanation.
The preference patterns of Jan and Kelvin mean that money is more
important to Jan if the weather is hot, and more important to Kelvin if
the weather is cold. Thus, a Pareto-consistent or utility-efficient method
of dividing the $100 is to give more money to Jan if the weather is hot
and more to Kelvin if the weather is cold, rather than equal division
irrespective of the weather. Subject to practical difficulties, a good parent
should find out the different needs of the children under different
situations and hence be able to satisfy their needs better, rather than
just adopt some rigid rules such as equal division or random choice
(coin tossing). However, to learn about the different needs requires the
interpersonal comparison of cardinal utilities.
2.2 Economists’ misplaced hostility against cardinal utility
When I first studied Arrow’s impossibility theorem, I quickly attributed

the difficulty to the lack of cardinal utilities. (See Ng, 1971. Later, when
I studied Sen, 1970a and DeMeyer and Plott, 1971, I realised the necess-
ity of interpersonal comparison as well.) However, after the ordinalism
revolution in consumer theory, economists are very hostile to cardinal
utilities. A recent textbook example is Varian. (A non-textbook example
is Kolm, 1993.)
But how do we tell if a person likes one bundle twice as much as

another? How could you even tell if you like one bundle twice as
much as another? One could propose various definitions for this kind
of assignment: I like one bundle twice as much as another if I am
willing to run twice as far to get it, or to wait twice as long, or to
gamble for it at twice the odds. . . . Although each of them is a poss-
ible interpretation of what it means to want one thing twice as much
as another, none of them appears to be an especially compelling
interpretation.
Varian, 1993, pp.57–8
There is in fact an interpretation that is especially compelling. This is

whatever the individual concerned values ultimately. If we abstract
away effects on other individuals and sentients, what I ultimately value
is my net happiness (i.e. enjoyment minus suffering, including the sen-
suous as well as the spiritual). On the grounds of evolutionary biology,
daily experience, and interviews, I have reasons to believe that I am not
an exception here but rather quite representative. Since, for myself, it
is ultimately net happiness that I want, it has an especially compelling
interpretation (for cardinal utility).
For most of my sleeping time and much of my time awake, I
am neither enjoying (feeling good) nor suffering (feeling bad). My
happiness is zero over those periods. When I am sick, hurt (bodily
or emotionally), sad, etc. my happiness is negative and the extent
of the negative happiness may differ depending on the intensity of
the pain. When I enjoy myself either sensuously or mentally, my hap-
piness is positive, with again different degrees, as shown in Figure 2.2.
My net happiness (over any period) is the area above the line of
neutrality minus that below. Thus, despite different types of well-being
(poetry versus pushpin), the overall happiness is one-dimensional.
(See also Joseph and Lewis, 1998 on the unidimensionality of happi-
ness. While indirect, external, and long-term effects on oneself and
others complicate measurement, they do not affect the unidimension-
ality. On a recent discussion of the concept of well-being, see Qizilbash,
1998.)
happiness
in utils
0 time
unhappiness
in utils
Figure 2.2
I want money but not for its own sake, only to obtain, ultimately,
happiness. Thus, I have diminishing marginal happiness of money.
For big variations, my willingness to pay twice as much does not indi-
cate twice as much happiness or utility. (The two are the same if we
abstract away ignorance, concern for others, and irrational preference.
This is so with my definition of irrational preference as preferring some-
thing that decreases one’s own happiness or welfare, neither due
to ignorance nor to a concern for the welfare of others. Utility here is
taken as representing preference which can be cardinal. See Chapter 4
for details.) However, as I want happiness for its own sake, if bundle
A (or anything else) gives me twice as much happiness as bundle B,
it is perfectly sensible, natural, and informative to say that I prefer
(reminding the abstraction of ignorance, etc.) bundle A twice as much
as bundle B.
Even if we re-introduce factors other than personal happiness in the
preference function, I have no difficulty (except for imperfect infor-
mation to be discussed presently) in comparing the intensities of my
preference for different pairs of alternatives. Thus, if I take into account
the income or welfare levels of others, there is no problem in allowing
for that. For the simple case where both my and the only other person’s
welfare are functions of the log of own income, my preference or utility
function could be the log of my income plus alpha times the log of the
other person’s income, where alpha is a positive number slightly larger
than zero. I could also allow for alpha itself to be a function of the
income or welfare level of others. It is true that I often have difficulties

knowing the intensity of my preference for an alternative over another.
This is due to the lack of information (or lack of perfect memory for
past events) as to how my and other people’s welfare values will be
under different alternatives. Given this lack of information, I even have
difficulties knowing whether I prefer A to B or vice versa. Thus, not only
are the intensities of preference or cardinal utility made unclear but the
ordinal preference or ranking itself is also made unclear. If you put two
close enough quantities of water into two containers of different shape,
I may have difficulty judging which container has more water. But that
does not mean that the volume of water is not a cardinally measurable
quantity!
I have also no difficulty saying that my welfare level is positive, zero,
or negative. When I am neither enjoying nor suffering, my welfare is
zero. Thus, the value of my welfare is a fully cardinal quantity unique
up to a proportionate transformation. I am also sure that I am not
bestowed by God or evolution to have this special ability of perceiving
the full cardinality (both intensity and the origin) of both my welfare
and preference levels. In fact, from my daily experience, observation,
and conversation, I know that all people (including ordinalist econ-
omists) have this ability, except that economists heavily brain-washed
by ordinalism deny it despite actually possessing it. This denial is quite
incredible. If your preference is really purely ordinal, you can only
say that you prefer your present situation (A) to that plus an ant bite
(B) and also prefer the latter to being bodily thrown into a pool of sul-
phuric acid (C). You cannot say that your preference of A over B is less
than your preference of B over C. Can you believe that? (See also Ng,
1997).
I attempted to show the bias of economists against cardinal utility, at
least in comparison to the general public, in a small survey. While the
result may be said to confirm my expectation, it is not as clear-cut as I
expected. The questionnaire asks four questions.
Question A: 1. Compared to the corresponding period yesterday, my

net happiness (i.e. happiness minus unhappiness) for today so far is
higher/lower/about the same; 2. by 0%/1–10%/11–30%/31–100%/over
100%.
Question B: 1. Think of a person (other than yourself) whose
happiness/unhappiness you know best, and for a period you know
best. In comparison to that of a corresponding period for yourself
that you can easily compare, is her/his happiness over that period
higher/lower/about the same; 2. by 0%//1–10%/11–30%/31–100%/over
100%.
For all questions, respondents are also provided with boxes marked
‘cannot answer’ and ‘this question is meaningless’. I had expected far
more economists to regard Question A2 as either unanswerable or mean-
ingless than the general public. The difference is much smaller than I
expected. The sample size is 49 for the general public (from shopping
and entertainment centres) and 52 for economists, consisting of par-
ticipants of an econometrics conference (31) and a social choice con-
ference (21). For the general public, 16.3% (8 out of 49) answered either
‘cannot answer’ or ‘meaningless’ for Question A2. If we group all econ-
omists together, 32% (16 out of 50; 2 did not answer) ticked ‘cannot
answer’ or ‘meaningless’. While this is certainly a significant difference,
the difference is not as remarkable as I expected. I expected more than
half to tick so. On hindsight, it may be said that, the majority of
economists still retain the commonsense that happiness is cardinally
measurable, despite their professional training to the contrary. I also
expected the group from the social choice conference to be more sym-
pathetic to cardinal measurability than the group from the economet-
rics conference. But there is no significant difference (7 out of 21 and
9 out of 29 respectively).
For the general public, 14.3% (7 out of 49) and 18.75% (9 out of 48)
answered either ‘cannot answer’ or ‘meaningless’ to Question B1 and
B2 respectively. The corresponding figures for economists are 40.4% (21
out of 52) and 51.1% (23 out of 45; 7 did not answer) respectively.
Again, there are no significant differences between the two sup-groups
of economists. Less than 10% of any group answered ‘cannot answer’
or ‘meaningless’ to Question A1.
2.3 The impossibility of social choice based on

ordinal utilities
Samuelson (1967, 1977), following Little (1952), rejects the relevance

of Arrow’s theorem to economics. Essentially, they argue that Arrow’s
impossibility theorem was proved by requiring some reasonable
conditions over alternative sets of individual preferences. While this
may be relevant for politics, they argue that it is not relevant for wel-
fare economics. For the latter, we need a SWF for a given set of indi-
vidual preferences. If the set of individual preferences changes, we
have another SWF. By not requiring any consistency or reasonable
conditions between the different SWF’s corresponding to different
sets of individual preferences, Arrow’s impossibility result does not
apply. Honestly, I am not persuaded by this line of reasoning. As indi-
vidual preferences change, the social ordering or social choice may
change. But why should the rule used to aggregate individual prefer-
ences into social choice change? Nevertheless, for my purpose here, I
would rather have readers agreeing with Little and Samuelson, as this
would increase the importance of the following result that I was
involved in proving.
Kemp and Ng (1976) and Parks (1976) show that, even operating
within the same set of individual preferences or holding individual pref-
erences unchanged (the so-called single-profile or intra-profile frame-
work), impossibility results may still be proved using compelling
conditions, if we have only individual ordinal preferences to base on.
This impossibility result was quickly generalised (Hammond, 1976;
Pollak, 1979 and Roberts, 1980), establishing the existence of a corre-
sponding (impossibility or possibility) proposition in the single-profile
framework for every proposition in the multi-profile or inter-profile (i.e.
the original Arrowian) framework. This should seal the fate of ordinal-
ism but Samuelson objected.
Samuelson (1977) argues that an axiom (Anonymity Plus Orderings
Only) used by Kemp and Ng is objectionable. Since no one objects to
anonymity, the real issue concerns ordinalism. For the general social
choice problem, I myself regard ordinalism as unreasonable. However,
since the point here is about whether ordinalism is a sufficient basis
for the single-profile social choice, the axiom is a natural condition to
impose. If Samuelson objects to this axiom, he should not continue to
insist on the sufficiency of ordinalism.
Some years after the Kemp–Ng–Samuelson debate, I discussed the
problem with Arrow. He said that, if Samuelson insisted on ordinalism,
he had to give up independence. I replied that independence is com-
pelling. Later, I realised that independence is really implied in the
Bergson–Samuelson tradition of writing social welfare at any social state
x as a function of individual utilities at x only. This tradition is all right
if we have interpersonally comparable cardinal utilities. However,
Samuelson insists that the individual utility indices are only ordinal.
This is the crux of the impossibility. (See Appendix E for more detailed
argument.)
Despite the necessity of interpersonal comparison of cardinal utilities,

the need to make such difficult comparisons may be minimised (but not
avoided completely) by using the willingness to pay to reflect the car-
dinal intensities of preferences. (See Chapter 6.)
3
Welfarism
Welfarism is the belief/principle that social welfare depends (positively)

only on individual welfare (or utility) levels. (Abstracting away the dif-
ferences between utility and welfare, on which see Chapter 4. Questions
of animal welfare and national boundary are also ignored.) Most people
(including Sen) accept that individual welfares matter but many want
to add other desiderata. While one may use any number of secondary
instrumental principles, one can only have a single ultimate objective.
Given a sufficiently wide domain of choice and sufficient differences
in individual preferences, a conflict must arise with multiple ultimate
objectives/principles, as exemplified by Sen’s (1970b) impossibility of a
Paretian liberal. (See also Ng, 1982 on the impossibility of a Paretian X
where X is any other principle.)
Partly for its own interest and partly for expositional colour, the issue
of the acceptability of welfarism is first discussed in Section 3.1 with
reference to my debate with Sen. Two separate arguments for welfarism
are then presented in Section 3.2 and Section 3.3.
3.1 The Sen–Ng debate on welfarism: an appraisal
As noted in Section 1.4, Sen (1979) strongly criticises welfarism.

Welfarism may be criticised on two different levels. First, owing to lack
of complete information about individual welfares, we may wish to
base our social decision on the objective information about the various
social states or alternatives. On this point, I am in complete agreement
with Sen, provided he agrees that this is not an argument against
welfarism as such but is rather a consideration of the ways to achieve
our objective (even if welfarist) in the absence of perfect information.
24
Welfarism 25
For example, suppose we agree (for argument’s sake) that we should

maximise the SWF of the sum or the product or some other quasi-
concave function of individual welfares. If we also have complete
information about individual welfares in various social states, what we
need to do is just to compare these various sums or products, etc.
of individual welfares. However, usually we cannot do just this owing
to the lack of information. If we also believe in diminishing marginal
welfare of income and similar capacity to enjoy income (or that this
capacity is not strongly positively correlated with income and we do
not know who has high or low capacity; see Lerner, 1944 and Sen,
1973b), then a certain policy may be regarded as desirable if it increases
the incomes of the poor and reduces those of the rich without reduc-
ing total income. Then, a consideration that affects our social decision
appears to be that of income equality or some similar consideration,
among others. Our SWF appears to be a function not only of individ-
ual welfares but also of some objective indicators such as the equality
of income distribution of social states.1 But it is clear that we use such
objective indicators only as indirect measures or estimates of individual
welfares. Thus, far from being a rejection of welfarism, the use of such
objective indicators serves to achieve our welfarist objective. Such prin-
ciples as ‘giving priority to the interests of the poor over the interests
of the rich’ need not be non-welfarist if it is derived from a welfarist
SWF plus certain other assumptions such as diminishing marginal
welfare of income.
Sen’s objection to welfarism is, however, much more fundamental
since he ‘criticizes welfarism even when utility information is as com-
plete as it can conceivably be’ (Sen, 1979, p.547), even with completely
cardinal, unique and interpersonally comparable individual welfare
indices (which we shall be using here). Basically, Sen believes that we
may wish to place value on such principles as liberty, non-exploitation,
no-torture, etc. over and above or besides their contribution in pro-
moting individual welfares. Thus, even if two social states are exactly
the same with respect to individual welfares, we may prefer the one
that adheres to these principles to the one that violates them. For
illustration, Sen (1979, p.547) provides the following examples (Sen
remains very critical of welfarism along the similar line in his 1997
paper):
Consider a set of three social states, x, y and z, with the following

interpersonally comparable cardinal welfare numbers for a two-
person community:
x y z
Person 1’s welfare 4 7 7
Person 2’s welfare 10 8 8
In x, person 2 is eating a great deal of food while person 1 is hungry.

In y, person 1 consumes a bit more of the given food supply. While
2 is made worse off (in comparison to x), 1 is made better off by a
larger amount and the sum of welfare becomes larger (with dimin-
ishing marginal welfare). It is clear that y must be judged to be better
than x by utilitarianism, and also by virtually all the criteria that have
been proposed using data on individual welfares. Let us take y to be
socially better than x.
Consider now z. Here person 1 is still as hungry as in y, and person
2 is also eating just as much. However, person 1, who is a sadist, is
now permitted to torture 2, who – alas – is not a masochist. So person
2 does suffer, but resilient as he is, his suffering is less than the utility
gain of the wild-eyed 1. The utility numbers in z being exactly the
same as in y, welfarism requires that if y is preferred to x, then so
must be z. But y is socially preferred to x. So z is preferred to x as
well, thanks to welfarism.
Sen, 1979, pp.547–8
In a footnote, Sen adds: ‘It is assumed that there are no indirect conse-
quences of torture, e.g. in attitude formation. These indirect effects do
not change the nature of the difficulty, even though they can be prop-
erly accommodated only in a much more complex analysis.’
The above argument of Sen against welfarism appears very persuasive.
However, it seems to me whether one has to reject welfarism on the
grounds of such moral principles as no-torture depends on whether one
believes in these principles as his basic value judgements. ‘A value judge-
ment can be called “basic” to a person, if the judgement is supposed to
apply under all conceivable circumstances, and it is “non-basic” other-
wise’ (Sen, 1970, p.59).2 Those whose belief in no-torture is more basic
than welfarism must withdraw their commitment to welfarism in cir-
cumstances like z. But one, like myself, may also believe in no-torture
not as a basic value judgement but as derived from a welfarist SWF
because one may believe that in practically all circumstances, torture
leads, directly and indirectly, to more harm than good. However, most
people believe that it is wrong to inflict harm on others even if the
pleasure of doing so exceeds the direct and indirect sufferings caused.
Thus, it is very tempting to reject welfarism (and utilitarianism, a spe-
Welfarism 27
cific form of welfarism, in particular) at the immediate intuitive level.

But as argued by Hare (1976), the rejection of utilitarianism at the imme-
diate intuitive level is consistent with the support for utilitarianism at
a deeper ‘critical’ level.
One may accept a welfarist SWF as their most basic value judgment.
It is, however, difficult to do a precise welfare calculation for every de-
cision made. Thus, it is generally desirable to adopt certain rules such
as honesty, liberalism, non-exploitation, no-torture, etc. which gener-
ally contribute to the promotion of social welfare as defined by the wel-
farist SWF. In time, these rules become moral principles and tend to be
valued for their own sake even to the extent of persuading people to
reject the original welfarist SWF. But in my view, this is confusing the
more basic value with less basic values. This is like insisting on telling
a cruel, dishonest invading army the truth in order to stick to the prin-
ciples of honesty. However, I am not advocating giving up these moral
principles lightly owing to the enormous long-run implications through
attitude formation.
Some people may wish to regard certain principles such as freedom
from torture as being based on, say, human rights independent of or
over and above welfare considerations. However, if one presses oneself
hard enough with the question, ‘Why human rights?’, I believe that one
will most likely come up with a welfarist answer. I enjoy walking. But
this enjoyment is based on the violation of the ‘rights’ of stones, sands,
etc. for freedom from being stepped upon. So why don’t stones have
‘stone rights’? Why do we hear about ‘human rights’, ‘animal rights’,
etc. but not ‘stone rights’? An obvious answer is that stones do not feel
pleasure and pain. But this is clearly a welfarist answer. It is possible that
certain basic human rights are so definitely conducive to the improve-
ment of welfare and the violation of them on the grounds of short-run
welfare (or non-welfare) considerations so likely to produce effects very
unfavourable to long-run welfare, that the insistence on these human
rights on a political or practical level without regard to any welfare con-
sideration becomes defensible. But this does not negate the belief that,
on a deeper philosophical level, these rights are ultimately derived from
welfare considerations.
Sen believes however that:
any serious theory of human or civil rights will bring in other [than
sentient] aspects of people which also differentiate them from stones.
The respect and concern that one person owes another can hardly
be seen as a function only of the latter’s capacity for pleasure. Indeed,
even after a person dies and obviously has no further capacity for
pleasure or pain, right-based considerations vis-à-vis him do not all
cease to apply, e.g. they are relevant in disposing of his body, or in
defending him against vilification.
Sen, 1981, p.534
In my view, such post-mortems ‘rights’ can be explained by the prefer-

ences of living individuals and the effects on future behaviour. For
example, living persons may want to show respect for people in general
and specific individuals in particular by having proper burial in general
and national burial for heroes that, in addition, may encourage heroic
behaviour. It is true that people differ from stones in many other
aspects. For example, stones and even animals cannot exercise such
rights as voting in elections. But I am referring to the more elementary
rights of freedom from being unnecessarily hurt (which I believe should
apply to all sentients). Moreover, if there exist non-sentient objects
(super-machines?) that can ‘exercise’ certain rights (e.g. voting) oper-
ationally, I would treat them no more favourably than stones.
While I maintain my basic defence of welfarism given in Ng (1981a),
I am happy to make the following three concessions to Sen. First,
though I find welfarism a perfectly defensible principle as an ultimate
value judgement, it does not mean that those who attack welfarism
must thus be wrong in any logical or scientific sense. Since this con-
cerns questions of basic value beliefs, perfectly rational and logical
persons may differ ultimately. On the other hand, this does not mean
that such questions cannot be logically discussed or that the discussion
is useless. One may logically show which axioms (including some nor-
mative ones) imply certain normative principles or which normative
principles imply some desirable or undesirable outcomes. In the process,
one may persuade opponents to abandon their initial value judgements
which they mistakenly believe to be basic but turn out to be non-basic
upon critical analysis. (See the appendix of Ng, 1981b, on the use of
axiomatic value theories.)
Secondly, I think Sen is right in believing that the consensus of econ-
omists on welfarism is ‘based on not examining explicitly the problems
of conflict that can and do arise’ (Sen, 1981, p.532). If economists were
to seriously consider issues raised by Sen, a significant proportion, if not
a majority, of them would probably reject welfarism. This concession is
made partly due to the fact that moral philosophers, who have pon-
dered over such issues over a long time, are notoriously divided in their
views with respect to utilitarianism in particular and to consequential-
Welfarism 29
ism in general (for example, see Smart and Williams, 1973). ‘Ng’s own
view of the matter should not be confused with a claim that a consen-
sus does exist on this question’ (Sen, 1981, p.532), though I may attempt
to persuade economists to stick to welfarism, and for the defectors to
re-embrace welfarism after a more ‘critical’ examination.
The third concession I wish to make is based on practical consider-
ations including the lack of complete information about individual wel-
fares mentioned above. Partly because of this lack of information and
partly because of the cost of making a detailed welfarist calculation for
each decision, it is sensible to adopt some principles, procedures, etc.
that are generally consistent with the ultimate welfarist objective. (The
relevance of arguments, e.g. Suzumura, 1999, on the importance of the
appropriate procedures should be seen in this light. They are relevant
and important and yet do not affect the validity of welfarism at the ulti-
mate level. See also Section 5.5 which argues that utilitarianism is con-
sistent with process fairness.) In time, these principles are valued for
their own sake and these secondary or derived values have to be reck-
oned with even by a pure welfarist (who may however try at the same
time to persuade people not to value non-basic principles over and
above their instrumental values so that we can avoid being the slaves
of certain ‘moral principles’ which are no longer conducive to social
welfare due to new circumstances). Therefore, while I continue to
adhere to a pure welfarist objective at the basic philosophical or ‘criti-
cal’ level, I am quite willing to concede that, at the practical day-to-day
(or even year-to-year, decade-to-decade) basis, characteristic of most
economic and social decisions, non-welfarist (from the short-term view-
point) principles have to be reckoned with. At least in this sense, then,
Sen’s objection to a purely welfarist calculation is important.3
After making these concessions to Sen, I am sure I will not jeopardise
my friendship with him by adding a caution against his following con-
clusion: ‘Arrow’s impossibility theorem can be seen as resulting largely
from combining ‘welfarism’ (ruling out the use of non-utility infor-
mation) with remarkably poor utility information (especially because of
the avoidance of interpersonal comparisons)’ (Sen, 1979, p.554). I agree
whole-heartedly that the poor utility information (non-cardinalism
and non-comparability) is of paramount importance in producing the
impossibility result (see Chapter 2). I also agree that the introduction
of non-welfarist principles may help social choice in particular in-
stances. However, without richer utility information, the rejection of
welfarism itself does not provide a satisfactory solution to the paradox
of social choice, as shown in Chapter 2 above. (For more rigorous
demonstration of the inadequacy of ordinalism even with non-welfarist

principles, see Ng, 1982, 1985.)
3.2 Another argument for welfarism
Welfarism (including, of course, utilitarianism) subsumes consequen-

tialism – the belief that the moral goodness of an action depends only
on its consequences. However, it is possible to divorce welfarism from
consequentialism, as discussed below. (For an argument on consequen-
tialism, see Hammond, 1996.)
A main objection to consequentialism is that the motive behind an
action should not be completely ignored in assessing the moral good-
ness of an action. Two different reasons for this belief may be distin-
guished. First, good motives and their encouragement may contribute
to good consequences in the long run. But this is not inconsistent with
consequentialism provided a long-run, complete view is taken (Radford,
1988). Secondly, some people may regard good motives as good in
themselves irrespective of their consequences. Person A attempted out
of altruism to save someone dying at a great cost to himself. However,
due to ignorance, he used a wrong method which hastened the death.
Person B maliciously tried to kill someone just for fun. However, due to
some unlikely occurrences, some benefits were effected. Most people feel
that A is good and B is bad rather than the other way round. This is
partly because we know that A/B is likely to do many other good/bad
deeds most of which would lead to good/bad consequences. Even
abstracting from this factor, most people still believe that motives
cannot be completely ignored in assessing moral goodness either of a
person or of an action. This objection to consequentialism may be
resolved by the following distinction.
Let us apply moral goodness to motives and to consequences separ-
ately. We can then say that the motive behind action X is good though
its consequences are bad or vice versa. We may then confine welfarism
to moral assessment with respect to consequences. Thus revised, wel-
farism is the belief that what makes anything morally good or bad in
terms of its consequences ultimately depends only on its effects on indi-
vidual utilities. In other words, welfarism rules out non-utility conse-
quences as of ultimate relevance for moral goodness with respect to
consequences. Thus defined, welfarism can be seen to be compelling as
shown below.
Consider the following value axiom.
Welfarism 31
A1. In a world with no conscious beings, nothing is of any moral

consequence
Here, a ‘world’ is defined as a four-dimensional totality, encompassing
all space and all times and of course all beings and events therein to
eternity. Thus a world with no conscious beings means no conscious
beings even in the future. Possible goodness such as in the conser-
vation of resources or natural beauties for possible future beings to
appreciate does not therefore arise. Since there is nothing conscious in
such a world to perceive anything, it seems extremely reasonable, if not
compelling, to say that nothing is of any moral consequence.
From A1, it does not take a huge step to accept A2.
A2. In a world with no sentients, nothing is of any moral consequence
Here, a sentient is defined as a being that is capable of being happy or
miserable. Thus, a sentient must be a conscious being but a conscious
being need not be a sentient. For example, one can logically imagine
beings that can perceive lights or even see and feel (the existence of)
things without being able to feel subjective pleasure and pain. In fact,
many lower forms of animals in our world are likely to actually fit this
description. It is likely that the ability to perceive evolved earlier than
the ability to feel rewards and punishment, because the latter ability
serves no purpose without the former.4 However, while these animals
eventually evolve into sentients in our world, they shall never do so in
a world satisfying the precondition of A2.
Suppose that our world W0 is transformed into W1 by changing all
sentients (including ourselves) into non-sentient conscious beings (to
eternity). According to A2, nothing is of any moral consequence in W1.
However, we (i.e. human beings after being made non-sentient but still
conscious) may still regard some acts (such as lying) as immoral though
without the feeling of moral compulsion (otherwise we would still be
sentients). But this must surely be due to our memory of affairs in our
present world. Let us transform our present world W0 into W1 by not
only changing all sentients into non-sentient conscious beings but
also eliminating their memories of affairs in W0. Would lying still be
regarded as immoral in W1? Many people (including, say, Arthur) may
reply in the affirmative and believe themselves to have such a moral
view in W1. To see that such a view is untenable, consider a different
world W2 which is similar to ours in most aspects (it even has a counter-
part of each and every individual in our world) except that lying in W2
does not cause any real hardship, mistrust, or any other undesirable
effects. Rather they lie to add colours to their life. People that seldom
lie are regarded as rather dull. Lying in W2 is like telling a joke in W0.
Everyone in W2 regards lying as morally good. Now, let us transform
W2 into W3 by making all sentients into non-sentient conscious beings
and eliminate all their memories of affairs in W2. Moreover, make some
minor changes if necessary such as to make W3 exactly the same as W1
in every aspect. If we ask the counterpart of Arthur in W2 (call him
Arthur2) what he thinks or would think of the moral standing of lying
in W3, he will probably say that lying is still morally good even in W3.
But by construction W3 is identical to W1. How could the same thing
(lying) in the same world (W1 or W3) be both morally bad and good at
the same time? Such absurdity can be avoided by accepting A2.
The acceptance of A2 seems to compel the acceptance of A3.
A3. In a world with sentients, what is of moral consequence is ultimately
due to the effects on the feelings of happiness or unhappiness of the sen-
tients in that world.
If what is of moral consequence is not ultimately due to the effects on
the feelings of happiness or unhappiness of sentients (or sentient feel-
ings for short), then something may be of moral consequence even
without in any way, directly or indirectly, affecting sentient feelings. If
this is true, something should also be of possible moral consequence in
a world with non-sentient conscious beings since there is no morally
relevant difference between the two. But this violates A2. Hence we
must also accept A3. Using utility to represent sentient feelings, A3 is
welfarism. Ultimately, what is morally good or bad depends only on the
effects on individual utilities:
w = w (u1 , u2 , . . . , un )
3.3 Rational individualism implies welfarism
In this section, it is argued that the requirements of rational individu-

alism, appropriately defined, already imply welfarism. As we are con-
cerned with the appropriate form of social welfare functions (i.e. with
what the society should ultimately maximise), it is natural to take the
existence of a SWF as given. Also, as we are dealing with the general
principle applicable to all potential situations, the assumption of
universal domain (which allows for all logically possible patterns/
combinations of individual preferences/welfares) is warranted. These
assumptions are not really needed for the substance of the following
arguments, but only to rule out cases like: If the number of social alter-
natives and/or the diversity of individual preferences are severely
Welfarism 33
restricted, e.g. if there is only one alternative or if everyone has the same
preference/welfare levels, then our substantive axioms below will have
no scope to ‘bite’ and one may argue that all (or all Paretian) social
welfare functions satisfy our substantive axioms in a trivial sense. This
is, of course, not substantive. Those who want completeness may take
it that we are assuming:
Axiom 0. A social welfare function exists and the domain of social alter-
natives and individual welfare levels are not restricted.
In fact, we do not need the full force of this axiom. As it is purely
a non-substantive issue, we will not include this axiom in the follow-
ing discussion. It may also be noted that, in the absence of either car-
dinal utilities/welfares or their interpersonal comparison, no reasonable
SWF is possible, even if we operate with a fixed set of individual
preferences/welfares. (Kemp and Ng, 1976 and Parks, 1976 establish
the impossibility without cardinality; Sen, 1970a establishes the imposs-
ibility without interpersonal comparison.) Thus, for the question on the
appropriate form of SWF, we have to operate with the existence of inter-
personal comparable cardinal utility/welfare. (See Chapter 2.) We now
turn to the substantive axioms.
Axiom 1. Rational individualism: Where no other sentients outside
a given society are affected, what the society should ultimately maximise
should depend positively on what individuals in the society rationally and
knowledgeably want for themselves ultimately.
This axiom may be broken into the following four components.
1a. Individualism: The society consists of individuals and does not
exist independently of individuals. Thus, what the society wants should
depend on what individuals in the society want; what the society should
pursue ultimately should depend on what individuals in the society ultimately
want.
1b. Positive dependence (the Pareto principle): The dependence
of social choice on individual values should be positive other than negative.
1c. Proviso on others: Just as my pursuit of my welfare is subject to
certain constraints to avoid causing great harm to you, on my neighbours, on
animals, etc. a society’s maximisation is subject to a similar proviso.
However, since the appropriate constraints and/or tradeoff is beyond
the scope here, this proviso is just noted without discussing its detailed
implications, by confining to choices where no other sentients are
affected. Due to this confinement, any disagreement on this proviso has
no effect on our arguments. For example, suppose someone disagrees
that animal welfare should enter this proviso. This does not affect our
arguments which are concerned only with choices where animal welfare
is not affected. Another might believe that we should not have this
whole proviso at all. This only means that our conclusions would then
apply without subjecting to ‘no effects on others’.
1d. Rationalism: The individual preferences that should be used in
determining the social preferences should be those that are free from cor-
rectable ignorance, irrationality, and non-affective altruism or malice.
This consists of three components.
1d1. Informed preferences: To a certain extent, individual
preferences may be distorted by ignorance, imperfect foresight, or
misinformation. Many of these are not correctable as the society is
similarly ignorant. Others are correctable. The fluoridation of water
is partly justified on this ground. Harsanyi (1997) has convincingly
argued for the use of informed preferences over actual preferences.
(See Chapter 4.)
1d2. Exclusion of non-affective altruism: Non-affective altruism
is the pure concern (or malice if the concern is negative) for the welfare of
others (which need not be confined to human only) independent of or over
and above the effect on one’s own welfare.
I may donate money to the poor because I feel happier seeing them
in a less deprived condition (affective altruism) or because my donation
will increase my status in the community (not real altruism). However,
if my welfare loss in the money donated is larger than my welfare gain
through the above and all other effects and yet I prefer the donation,
non-affective altruism is involved. Non-affective altruistic preferences
should be excluded from social choice because the welfare of each
person (including that of the poor) should already be fully concerned
with on its own right in the relevant social welfare function. Including
non-affective altruistic preferences will lead to multiple counting. This,
together with a more detailed distinction of affective versus non-
affective altruism, the existence of the latter, and the unacceptability
of the Pareto principle in terms of preferences in the presence of non-
affective altruism are contained in Section 4.1 and Appendix D.
1d3. Rationality: Social choice should be based on individual prefer-
ences that are not distorted by irrationality.
The preference of an individual is defined to be irrational if they
prefer A to B while their welfare will be higher at B than at A and if their
preference is due neither to ignorance nor non-affective altruism.5 This
definition makes ignorance, non-affective altruism, and irrationality
three exhaustive reasons for the divergence between the preference and
the welfare of an individual. Section 4.2 in the next chapter explains
causes of irrational preferences and argues why they should be excluded.
Welfarism 35
Due to the above considerations (and that of Section 4.2), I believe

that most reasonable people realising our confinement to the ultimate
level including all indirect and long-term effects would agree after crit-
ical examination that Rational Individualism is compelling.
Definition 1. Welfarism: Where no others are affected, what a society
should maximise should be a function of (i.e. depend on) individual welfares
of members of that society, i.e.
W = W ( w 1, w 2 , . . . , w I )
where W = social welfare, wi = welfare of individual i, I = number of

individuals in the society.
Proposition 1. Rational individualism implies (‘positive’)
welfarism
Proof: Where no others are affected, Individualism requires the
society to pursue what individuals in the society ultimately want. An
individual may ultimately want: 1. their own welfare, 2. the welfare of
others, 3. any other things. Factor 2 is excluded by Axiom 1d2. Factor
3 could be due either to ignorance or irrationality which are respectively
excluded by Axiom 1d1 and 1d3. This leaves only the welfare of the
individual. Thus, Rational Individualism implies Welfarism. Axiom 1b
further requires the welfarist social welfare function to be ‘positive’ in
the sense of satisfying the Pareto principle, i.e. ∂W/∂wi > 0 for all indi-
viduals. Q.E.D.
In Chapter 4, it is argued that, where individual welfares (represent-
ing their happiness) differ from their utilities (representing their pref-
erences), social choices/public policies should ultimately be based on
individual welfares. Chapter 5 goes further than welfarism to argue in
favour of utilitarianism (social welfare should be an unweighted sum
of individual welfares). For a philosophical defence of welfarism, see
Sumner (1996). (However, Sumner uses a somewhat different concept
of welfare. Though he rejects the objective conceptions of welfare, he
also rejects the state-of-mind concept used here. Rather, he has a
mixture of state-of-mind and state-of-world concept of welfare as
‘authentic happiness’ or informed and autonomous happiness, some-
what similar to Harsanyi’s ‘informed preference’ discussed in Chapter
4.)
4
Utility, Informed Preference,
or Happiness1
Even accepting welfarism (argued to be compelling at the ultimate level

in Chapter 3), there is still the issue as to what the individual welfares
in the welfarist social welfare function should represent. In particular,
should the welfare of individuals be their (net) happiness as I define
welfare, or should it be their utility (representing their preferences)
as most economists seem to be assuming? Recently, Harsanyi (1997)
argues, among other things, that in welfare economics and normative
public policy, what are important are people’s informed preferences, rather
than either their actual preferences (as emphasised by modern econ-
omists) or their happiness (as emphasised by early utilitarians). The main
purpose of this chapter is to argue that pursuing Harsanyi’s argument
that allows him to move from actual to informed preferences to its
logical conclusion forces us to accept happiness as the ultimately impor-
tant thing. The early utilitarians were right after all! Since I personally
approve of Harsanyi’s basic argument, I regard myself as a follower who
has become more Catholic than the Pope. (On the importance of dis-
tinguishing between preference and happiness, or between ‘decision
utility’ and ‘experienced utility’, see Kahneman et al., 1997.)
It is not denied that, in practice, the practical difficulties and unde-
sirable side effects of the procedure of using happiness instead of prefer-
ences have to be taken into account. Thus, even if we ultimately wish
to maximise the aggregate happiness of people, it may be best in prac-
tice to maximise their aggregate preferences in most instances. This
important consideration will be largely ignored in this chapter.
The use of individual happiness instead of preference for social choice
may seem (especially to economists) to make the problem more dif-
ficult since happiness appears more difficult to discover. However, it
also makes the problem less intractable as happiness is more cardinally
36
Utility, Informed Preference, or Happiness 37
measurable and interpersonally comparable than preference. (See Ng,

1996a on a practical method of measuring happiness cardinally and
interpersonally comparably.) Also, the use of happiness instead of pref-
erence creates some difficult problems like evaluation in the presence
of preference changes (on the prevalence of which see Bowles, 1998)
tractable. Holding the welfare of others unchanged, whatever changes
in my preference, what I am ultimately concerned with is my happi-
ness which is cardinally measurable, inter-preference-pattern compara-
ble, and as concrete as my morning jam. All ultimate evaluation may
thus be made in terms of happiness.
4.1 Non-affective altruism: the pure concern for the

welfare of others
A person’s actual preferences are ‘indicated by his choice behaviour and

by his verbal statements’. His informed preferences are ‘defined as the
hypothetical preferences he would have if he had all the relevant infor-
mation and had made full use of this information’ (Harsanyi, 1997,
p.133). In addition, to qualify as informed preferences, ‘our preferences
should be genuine preferences rather than spurious ones’ (p.134). There
are various classes of spurious preferences including compulsive behav-
iour and self-deception.
Now, let us compare Harsanyi’s distinction outlined above with my
distinction between happiness (or welfare, defined to be the same) and
(actual) preference (Ng, 1979/1983; 1989b). (From now, ‘actual’ will be
dropped unless when emphasis is needed while ‘informed’ will be kept
throughout.) The preference of individuals may differ from their welfare
due to either one or more of the following exhaustive reasons: 1. ignor-
ance or imperfect foresight; 2. a concern for the welfare of others (the
definition of ‘others’ could be broad enough to include animal welfare);
and 3. irrational preferences. The last is defined to be any divergence
from welfare other than due to the first two factors. (This definition
makes the tripartite classification exhaustive, though some readers may
query the use of terminology.)
Comparing the above two schemes of distinction, it is clear that there
should be a complete agreement where the divergence is due to ignor-
ance, imperfect foresight, or misinformation. If a person’s preference
of medical treatment A over B is based on misinformation, the actual
superiority of B over A seems uncontroversial. For this type of diver-
gence, economists’ preference for using actual preferences may still be
sustained on the practical grounds of: 1. the difficulties of discovering

the informed preferences and 2. the possible unfavourable side-effects
of diverging from the actual preferences of people. Where these two
considerations are not important and/or the degree of ignorance is more
important, then it is uncontroversial that welfare or informed prefer-
ences should prevail over actual preferences, as witnessed by the preva-
lence and widespread support of fluoridation of water and prohibition
of hard drugs.
To a large extent, the spurious preferences of Harsanyi corresponds
with my concept of irrationality. (I will return to discuss some possible
differences later.) Thus, whether informed preference or welfare should
be used for social evaluation depends mainly on how the divergence
between preference (actual or informed) and welfare due to a concern
for the welfare of others (non-affective altruism/malice) should be dealt
with. I wish to argue strongly that such a concern should be ignored in
social evaluation. In other words, social evaluation should take account
of happiness or individual welfare rather than preferences (actual or
informed). This is so for the simple reason that otherwise double or
rather multiple counting will be involved. (Harsanyi allows for ‘truly
altruistic actions’ which seems to correspond to my ‘concern for the
welfare of others’ or non-affective altruism; however, he does not discuss
the implication of this for the choice of individual welfares versus util-
ities as the appropriate arguments in the social welfare function. Else-
where, Harsanyi and others appear to reject all forms of altruism from
consideration in social choice on the ground of multiple counting. As
argued below, affective altruism should not be so excluded.)
For simplicity, consider a society of two individuals. Individual 1 is
self-concerning and maximises his own welfare, with utility function
u1 = w1. Individual 2 has a substantial concern for the welfare of indi-
vidual 1, with utility function u2 = w2 + 0.5 w1. Also for simplicity (but
not essential for the argument), suppose that the relevant social welfare
function is utilitarian, maximising the unweighted sum of individual
values. The question here is whether the individual values should be
individual utilities or individual welfares. Note also that, for this exer-
cise (i.e. summing, and in fact any other reasonable method of aggre-
gating individual utilities or welfares to arrive at social welfare) to be
possible, we must have a framework of interpersonally comparable car-
dinal individual utilities or welfares, as argued in Chapter 2 above. Thus,
in Table 4.1, the welfare and utility figures are cardinal and interper-
sonally comparable.
Table 4.1
Social states w1 w2 u1 u2 Swi Sui
x 2 7 2 8 9 10
y 6 2 6 5 8 11
z 8 5 8 9 13 17
For the situation depicted, should the society choose x or y for

the choice between x and y only (with z not feasible)? If Swi is used,
x is preferred to y; if Sui is used, y is preferred to x. In my view, it is
clear that Swi should be used and x should be socially preferred to y.
In fact, since the individual welfare profile of x is (2, 7) while that of
y is (6, 2), in terms of individual welfares, x will be preferred by any
social welfare function that is anonymous and increasing in individual
welfares. Though x has a lower Sui, this is so only because the substan-
tial concern of individual 2 for the welfare of individual 1. If individ-
ual 2 were to make the choice herself, she would prefer z (if feasible)
to x. Though her welfare is lower at x than at z, her concern for the
welfare of individual 1 more than offsets this as w1 is much higher at
z than at x. However, her concern for the welfare of individual 2 is
not sufficient for her to prefer x to y. For social choice, the use of Swi
already takes full account of w1, i.e. w1 is already treated at a par with
w2. Thus, there is no further need to take account of the concern of
individual 2 for w1. To do so would involve the double counting
(multiple counting in the case of many non-self-concerning individ-
uals) of the welfare of individual 1. In using Sui, w1 is counted fully (i.e.
with the weight of unity) under u1, and then counted again at the
weight of 0.5 under u2. It is thus counted 1.5 times, while w2 is counted
only once.
It may be thought that, since w1 enters u2 at the weight of 0.5 while
w does not enter u1, it may be right to count w1 1.5 times while count-
2
ing w2 only once. This is inappropriate. Consider the concrete example

of parental choice. Usually, parents have sufficient concern for the
welfare of their children such that the society does not find it necessary
to interfere and just let parents choose for the whole family. (There are
also reasons of the practical difficulties and side-effects of interference
which we shall ignore for simplicity.) However, to tackle some special
cases of gross parental negligence, there are appropriate legislations.
Nevertheless, even ideally, such legislation should only aim to ensure

that the welfare of children are taken fully (i.e. at a par with that of the
parents) into account, not that the welfare of children should be more
important than that of the parents. To do the latter would be making
the mistake of correcting a mistake excessively.
A reason why some people may prefer using Sui rather than Swi (or
other functions of individual utilities rather than individual welfares)
is due to a failure to distinguish ‘minding’ from ‘concerning’. A self-
minding person is one whose welfare (or happiness) is not affected
by the welfare levels of others. A person may feel bad knowing that
there are people suffering in Africa. This is a ‘minding’ effect or
affective altruism. A self-concerning person is one whose utility (or
preference) is not affected by the welfare levels of others, except
possibly through their effects on his/her own welfare (i.e. through
the ‘minding’ effect). A person may choose x over y even if her welfare
is lower at x than at y because the welfare levels of others are higher
at y than at x. This is a ‘concerning’ effect or non-affective altruism.
A self-centring person is one who is both self-minding and self-
concerning.
In functional forms and ignoring ignorance, we may write the utility
of a rational person as:
u i = u i ( w 1, w 2 , . . . , w I )
where u = utility, w = welfare, and a superscript indicates the person

concerned, and I = the number of individuals concerned.2 And:
w i = w i (a, b , c , . . . , w 1 , w 2 , . . . , w i -1 , w i +1 , . . . , w I )
where a, b, c, . . . are some variables such as consumption that enter the

welfare function but are not the focus here. The ‘minding’ effect is
∂wi/∂wj, j π i and the ‘concerning’ effect is the direct effect of wj on ui
without going through ∂wi/∂wj.
It is true that these two effects are usually intertwined. If a person
is concerned with the welfare of another person, she is also likely to be
non-self-minding towards his welfare. (The reverse may also be true
though with less force, as the minding effect is usually stronger and
more prevalent than the concerning effect.) Thus, parents are typically
happy knowing that their children are happy (minding effect) and also
willing to choose something against their welfare for the welfare of their
children (concerning effect). It is true that they will feel good knowing
that they are sacrificing for their children and that the welfare of their
children increases, but if this is not sufficient to offset their welfare loss
due to other factors (such as a lower level of consumption, less leisure,
worse health), the concerning effect is involved over and above the
minding effect.
As the minding effect actually affects the welfare level, it is already
taken fully into account in the social decision in accordance to Swi, as
it should be. For the concerning effect, the welfare of the concerning
person is not affected, but her preference is affected by her concern for
the welfare of others. However, in using, say, the unweighted sum of
individual values, the welfare of others is already taken fully into
account, i.e. is already fully ‘concerned’ with. Hence, there is no further
need to take the concern of this person for the welfare of others into
account.
The existence of non-affective altruism (the concerning effect) may
be doubted (especially by economists, as mentioned to me on several
occasions). This attitude has some advantages as many instances of
apparent altruism are consistent with self interest due to the influence
of social pressure, obligations, fame-seeking, future rewards, and the
like. (For a collection of papers on the economics of altruism, see
Zamagni, 1995. See also Bagwell and Bernheim, 1996 and Harbaugh,
1998 on the status or prestige motive for making charitable contribu-
tions; Clark, 1998; Eshel et al., 1998; Laitner and Juster, 1996 and Rose-
Ackerman, 1996 for evidence and economic theories regarding altruism;
Lunati, 1997 on some ethical aspects; and Mulligan, 1997 on parental
altruism towards children. Altonji et al., 1997 show that a dollar increase
in parental income plus a dollar decrease in the child’s income only lead
to a $0.13 increase in transfer, contrary to the altruistic model. However,
this is so only for a specific altruistic model, not necessarily so for some
more general one.)
It may be argued that, if Ms 2 is willing to choose z over x in con-
sideration of the welfare of Mr 1, she must be happier with z than with
x. So, ui and wi must always go together such that the situation depicted
in Table 4.1 where u2 and w2 conflict each other with respect to the pair
(x, z) cannot logically arise. In other words, only ‘minding’ is possible;
‘concerning’ is not possible. This argument is incorrect. As conceded
above, the two effects usually intertwine with each other, and ‘minding’
is more prevalent than ‘concerning’. Many apparently ‘concerning’
effects may actually be ‘minding’ effects upon closer examination.
However, some truly pure ‘concerning’ effects are possible. To see this,
consider the following purely hypothetical construction which never-
theless illustrates the point most dramatically:
Suppose that you are commanded by the all-powerful Devil to press

either button A or B within 2 seconds. You know with certainty that
the following outcomes will happen depending on which button you
press.
A: You will go to ‘Bliss’ with your happiness level at 1,000 trillion
units. (Those who prefer concreteness may imagine a health-
enhancing island with all the material supplies you want plus
100 abiding partners of your dream.) All others on earth will go
to Hell with happiness level at minus 1,000 trillion units each.
B: You will go to ‘Bliss Minus’ with your happiness level at 999 tril-
lion units. (The same as above but with 99 partners.) All others
will go to ‘Niceland’ with happiness level at 100 trillion units
each.
C: If you do not press any button within the 2 seconds, you and all
others will go to Hell.
There will be no communication between Bliss, Niceland, Hell, etc. You

will lose all memory of the present world once you press either button
or fail to press within the 2 seconds. So you will not have any guilt feel-
ings in Bliss or Bliss Minus. Within the 2 seconds, you will be too pre-
occupied with pressing the right button that your welfare will be zero
whichever button you press. (It is too brief to experience any significant
amount of welfare anyway.)
Thus, by construction, your welfare will be higher with A than with
B. However, most people would choose B. (I would have not the slight-
est hesitation at all in choosing B, even if Bliss Minus involves only 500
trillion units of happiness.) The choice of B over A exhibits non-
affective altruism, though it may not be a very strong one. If you still
prefer A, change the happiness level of Bliss Minus into 999.999999
trillion units. If you still prefer A, then you may really be perfectly self-
concerning. But how could you condemn all others to Hell for a tiny
fractional increase in your own welfare?
Now consider more realistic choices. It is true that, as parents, we
usually feel happy doing something for our children (or other loved
ones) such as sacrificing our time, effort, money. Thus, the importance
of the ‘minding’ effect is uncontroversial. However, are we willing to
make such sacrifices only if our loss in welfare due to the sacrificed time,
effort, or money is more than made up by our warm-glow feeling of
helping our children? If the loss is not fully made up but yet the welfare
of our children will be significantly increased at a moderate or even
small net loss to our own welfare, will not at least some of us be willing
to do that?
If we have a true concern (over and above the ‘minding’ effect) for
the welfare of our children, can we not have a similar (though lesser in
degree) concern for our siblings, our relatives, our friends, our fellow
countrymen/women, the whole humankind, and eventually all sen-
tients? (For some evidence for true altruism, see Hoffman, 1981 and
Monroe, 1996.)
To further support the distinction between preference and happiness,
it may be noted that neuro- and behavioural scientists (e.g. Berridge,
1996) found that wanting (related to preference) and liking (related to
happiness) arise from two different neural systems, and therefore pref-
erence may not be synonymous with happiness.
Moreover, our behaviour and preferences are affected by our biologi-
cally determined inclinations. Natural selection results in the maxi-
misation of inclusive fitness (Hamilton, 1964; Wilson, 1975). Fitness
here is the ability of an organism to survive and transmit its genes to
the next generation. Inclusive fitness consists of the individual’s own
fitness plus her effects on the fitness of genetically related neighbours.
Thus, an individual may be expected to be altruistic towards her close
relatives, especially her own children. In addition, Trivers’ (1971) model
of reciprocal altruism shows that natural selection had to favour some
degree of altruism even between non-related individuals provided that
there is a significant chance of role reversal, as may be expected for
social species, including homo sapiens.
We care about our children/relatives because they carry/share our
genes. Sometimes one (or one’s spouse) may wonder why one is so
willing to help one’s siblings (or other relatives) whom one does not
really love much, not knowing that perhaps the genetically determined
subconscious inclination may play a role. Thus, the existence of some
non-self-centring effects is a biological necessity. However, whether the
biologically determined non-self-centring effects manifest mainly/only
in the ‘minding’ or the ‘concerning’ effects remains to be explored. Of
course, we are much more influenced by nurture than other animals
and also have a higher, if not exclusive, sense of morality. It may be
conjectured that, the more important is the influence of nurture and
the more important is the sense of morality, the more likely is it for the
true concerning effect to be present and significant.
From the above discussion, it may be concluded that true concern
(i.e. on top of or over and above the ‘minding’ effect) for the welfare
of others does exist. Moreover, when preferences (actual or informed)

differ from welfare due to a concern for the welfare of others, it should
be the individual welfare rather than the utility values that enter the
social welfare function.
While Harsanyi does not discuss the appropriate treatment of altru-
istic preferences in his 1997 paper, he discusses it elsewhere (Harsanyi,
1995, p.325). He suggests that all ‘external’ preferences, even if altruis-
tic, should be excluded in social welfare consideration, presumably
based on similar reasoning as our argument above. Here, external pref-
erences are ‘preferences for assignment of goods and opportunities to
others’ (Dworkin, 1977). The belief in the need to exclude altruism (or
its opposite, malice) is also held by a number of other authors discussing
the issue (e.g. Hammond, 1987/1995). My point here is that these
authors do not distinguish between affective (i.e. the minding effect)
and non-affective (the concerning effect) altruism and seem to suggest
that they should all be disregarded. As argued above, while non-
affective altruism should be excluded to avoid multiple counting, affec-
tive altruism should be included.
To see the reasonableness of including the minding effect, consider
the hypothetical example that either S or M has to die and a lot is drawn
to decide who has to die. Suppose that the two are equally old, capable,
etc. except that S is single with no close friends and M has many close
relatives who love her very much and will suffer a lot from her death.
Then, other things being equal, most people will hope that the lot will
turn out to let M survive rather than the other way round. The death
of M will cause more suffering. Counting this suffering does not involve
double counting. The minding effect should be treated differently from
the concerning effect. If I feel sad or happy for a certain event, such
feelings should certainly count in assessing the social desirability of that
event. On the other hand, if I have some non-affective concern for
the welfare of others, such a pure concern should not be counted if the
welfare of these others are already fully accounted for in the social
welfare function.
How could we say that the sorrow of a mother from the death of her
child should be disregarded as it is ‘external’ preference? It could cause
more suffering on her than starvation! Perhaps Dworkin would reply
that the death of a child would adversely affect the opportunities of the
mother, and hence the mother’s preference here is not wholly external.
However, for any definition of opportunities (unless it is defined to coi-
ncide with welfare), one can revise the example such that the prefer-
ence is external as defined by Dworkin (i.e. regarding the assignment of
goods and opportunities to others) and yet the person concerned (who
has the external preferences) genuinely suffers a lot subjectively. Such
external preferences should not be ignored. (However, we may have to
disregard many external preferences even when they genuinely affect
people’s welfare, on grounds of practical difficulties and undesirable side
effects; but this problem has been abstracted away in this chapter.)
The recognition that affective/non-affective altruistic preferences
should be included/excluded in social choice has practical implications.
For example, this is related to the issue on the extent to which gifts
should be taxed or subsidised. (There are of course other considerations
related to the tax/subsidy of gifts; see Kaplow, 1998. On the possible
unacceptability of the Pareto principle due to the presence of non-
affective altruism, see Appendix D.)
4.2 Irrational preferences
As mentioned above, Harsanyi regards ‘spurious’ preferences as not real

preferences and hence should not be used for normative purposes. He
gives two examples of spurious preferences. One is ‘compulsive behav-
iour’ where, for example, ‘some neurotics wash their hands far too many
times a day for no obvious reason. Their behaviour may be to some
extent voluntary and to that extent a result of their own preferences to
act in this way. But at a deeper level, it is obviously a result of a more
or less irresistible inner compulsion, very much contrary to their true pref-
erences’ (p.135). This may be regarded as some degree of psychological
sickness due perhaps to some subconscious need to wash away some
guilt or ‘dirtiness’. Such spurious preferences should not be treated on
a par as normal preferences. Thus, curing such sickness should be
regarded as an improvement but we do not want to change a person’s
normal preference of, say, apples over pears. However, before the sick-
ness is cured, we may still want to ‘respect’ his preference as, for
example, denying him the water to wash his hands may make him feel
very uncomfortable. But why is curing his compulsive behaviour good
while denying him water bad? In my view, this is due to different effects
on his (and perhaps also on others’) happiness. Thus, again, happiness
is more fundamental than preferences.
The second example of spurious preferences is ‘self-deception’:
Some people pretend to have, and may in the end even convince
themselves that they have, some preferences they think to be fash-
ionable and sophisticated – even though their real preferences may
be quite different, or even though after a point they may not really
know themselves what their real preferences are. Devotees of various
esoteric art forms of questionable aesthetic value often form coteries
that seem to display this kind of behaviour.
Harsanyi, 1997, p.135.
I do not know much about these esoteric art forms but it could be debat-
able to call the preferences of their devotees self-deception.
Harsanyi believes that informed preferences, by definition, will always
be based on all the relevant information, so that they will always be in
agreement with our real interests. However, it is debatable that people’s
preferences must always coincide with their interests if they are fully
informed. The meaning of ‘interests’ here is vague. If it means whatever
people prefer under full information, then it coincides with informed
preference tautologically but the terminology of ‘interest’ is misleading.
If it means welfare, happiness, or something similar, then it is not
true that people’s preferences when fully informed always coincide
with their interest, for at least two reasons. First, there is non-affective
altruism (and possibly also malice), as discussed in the last section.
Secondly, biological and psychological motivational factors may affect
preference or behaviour in ways not fully consistent with welfare, as
argued below.
There are a number of causes that may make preferences differ from
happiness other than ignorance and a concern for the welfare of others,
hence irrational according to my definition. There are at least three
sources of irrational preferences:
• Natural selection programmes us to maximise our biological

(survival and reproductive) fitness which may be different from our
welfare.
• Homo sapiens, the most intelligent species on earth, is still not per-
fectly rational as rationality is costly to programme.
• Side-effects (largely unavoidable) of generally good moral education
and upbringing.
The first source is not difficult to see. We are a biological species that
has evolved through many generations to survive natural selection.
(Otherwise I/you wouldn’t be here to write/read this book.) Natural
selection works through the preservation of those attributes (created
by random mutation) that are beneficial to the survival of the organ-
ism and the reproduction of offspring. In most lower species, their
behavioural patterns are completely determined by genetic programmes
and their interaction with specific environmental conditions. However,
when the environment became more complex (largely due to the evol-
ution of more species and more complex species), the fitness enhanc-
ing responses became more difficult to programme in advance. When
consciousness evolved to make choices, an important advantage is
gained by delegating some responses to be decided at the specific
moment. Individuals of species of such capability decides on ‘fight or
flight’ by sizing up the specific situation instead of relying on a fixed
response triggered by a given stimulus. However, when consciousness
has evolved to guide choices, such conscious choices have to be gener-
ally consistent with fitness for consciousness to survive natural selec-
tion (Ng, 1995). God (or natural selection) solved this problem by
endowing the conscious species with a reward/penalty centre (or
centres). Thus, activities consistent with fitness, such as eating nu-
tritious food when hungry or mating with fertile members of the op-
posite sex, are rewarded with pleasure and those harmful to fitness, such
as injuries to the body, are penalised with pain. (This makes the flexi-
ble species also ‘rational’ as defined in Ng, 1996b which shows that
complex niches favour rational species which make the environment
more complex, leading to a virtuous cycle that accelerated the rate of
evolution, partly explaining the dramatic speed of evolution based
mainly on random mutation and natural selection, a speed doubted by
creationists. Note that a more ‘rational’ species refers to a species with
less fixed behavioural patterns; hence that ‘rationality’ differs from the
concept of ‘rationality’ used in this chapter.) Due to this, there is a high
degree of consistency between fitness and welfare. However, the con-
sistency is not perfect.
On top of the ex-post rewards and penalties, we are also endowed with
inner drives to satisfy the fitness-enhancing functions like mating. On
the whole, these powerful temptations and drives work in the right
direction, making us do things that both enhance our biological fitness
and psychological welfare. However, since evolution is largely fitness-
maximisation and the welfare-enhancing aspect is only indirectly to
enhance fitness, some divergence between our behaviour and our
welfare is unavoidable, as our behaviour is not completely determined
by rational calculation but also partly by the programmed inclination,
including drives. For example, traits that incline individuals to produce
a number of offspring larger than that dictated by welfare maxi-
misation may be selected (Ng, 1995). Thus, we are not only born with
reproductive capabilities that we may rationally use to increase our
welfare and fitness together but we also possess biological drives that
incline us to behave or prefer activities not fully consistent with our
welfare, at least in many occasions.
As an example, adolescent girls and boys often engage in careless

sexual acts propelled by their sexual drives and tempted by sexual
pleasure even at high risk to their long-term welfare, such as having
unwanted pregnancies and contracting AIDS. While this is partly due
to ignorance, the role of biological drives cannot be denied.
Consider a specific example. Suppose that a person agrees that, for
choices involving risks, the correct thing to do is to maximise expected
welfare (assuming no effects on the welfare of others) and also actually
do so for most choices. However, for choices concerning seeking sexual
activities, he chooses x over y though his expected welfare is lower with
x than with y and that he knows this to be the case. Here, x may involve
having sex with an obliging partner of his dream or having sex with
many partners without clear knowledge (this knowledge is assumed
to be not feasible to obtain and hence not relevant) about her/their
infectability of AIDS. His (expected) welfare-reducing choice of x may
be due to the biological inclination to seek many sexual encounters. He
knows that doing so has a non-insignificant chance of contracting AIDS
and hence is welfare-reducing. He has all the relevant feasible infor-
mation and yet chooses (due to the powerful sex drive) x that he knows
to be of lower expected welfare. (This is not really a hypothetical
example. I am confident that, out of 100 average adult males, at least
10 have actually made such choices. If one wants more solid evidence,
one may look at the frequency of prostitution and extra-marital sex
despite the danger of sexually transmitted diseases including AIDS.)
Should we call this preference informed as the person has all the rele-
vant feasible information, or not informed because it is not in agree-
ment with his real interests? (See Elster, 1998 for a survey of the role of
our various emotions in our behaviour and welfare, especially in rela-
tion to economic analysis.)
For another example, traits that incline individuals to be excessively
in fear of death may also be selected. Undergoing enormous hardship
for a long time may enable one to bring one’s children to survivable
ages and perhaps also further chances of passing on one’s genes.
However, the life-time welfare may well be very negative.
Though we are the most intelligent and most rational species on
earth, we are still not perfectly rational. This is so because rationality is
costly to programme. For example, an important aspect of rationality
requires the individual not only to take account of current costs and
benefits but also those in the future (with appropriate discount for the
uncertainty on their realisation for costs and benefits in welfare terms,
or a discount at market interest rate for monetary costs). However, the
ability to purposefully take account of future costs/benefits appears to

be very rudimentary, if it exists at all, for most species. The storage of
food by ants and the burial of nuts by squirrels are hard-wired instincts,
not deliberate choices. If calculated choices are made by animals, they
are largely confined to sizing up the current situation to decide the
best move at the moment, like fight or flight. The ability to anticipate
rewards in the fairly distant future requires much more ‘reason’,
‘imagination’, and ‘telescopic faculty’ than normally cost-effective to
programme in most other species. However, we know that we are
endowed with some such faculty. (This is not to say that we are not also
partly driven by what Keynes called ‘the animal spirit’.) Nevertheless,
since this advanced faculty is almost completely absent in most other
species, it is natural to expect that it is not fully developed even in our
own species. Moreover, different members of our species may be
endowed with different degrees of such faculty. The existence of a sig-
nificant proportion of our species that do not possess a full telescopic
faculty is thus not surprising.
Many people tend to take inadequate care for the future or have
excessive discount rates. This is widely noted, including by economists.
For example, Pigou (1929, p.25) called it the ‘faulty telescopic faculty’,
Ramsey (1928, p.543) called it the ‘weakness of imagination’ about the
future, Harrod (1948, p.40) regarded it as the ‘conquest of reason by
passion’. A discount on future consumption, income, and any other
monetary values is rational as a dollar now can be transformed into
more than a dollar in the future. A discount on future utility may still
be rational if the realisation of the future utility is uncertain. (For
healthy people, this uncertainty is usually very small.) Discounting the
future for more than these acceptable reasons is probably irrational. A
manifestation of this irrationality is the insufficient amount of savings
for old age, necessitating compulsory and heavily subsidised superan-
nuation schemes. I came across an extreme example of such under-
saving during a survey regarding how much more people would be
willing to save if the rates of interest were higher (Ng, 1992a). The ques-
tion implicitly assumed that everyone did some saving, as the answers
were in terms of how many percentages more one would save. One
subject declared that he did not save anything. I then asked him to
change the answers to be chosen from ‘saving 20 per cent more’ into
‘saving $20 more per month’, etc. He still said that he could not be
induced to save anything even at annual interest rates of hundreds of
per cent. It was only when I said, ‘If a dollar saved now could become
a million dollar next year, would you save?’ that he admitted he would
save then. I was careful enough to find out that this healthy-looking
young man was not expecting early death from a terminal disease or
the like.
Related to both the above causes of irrational preference but border-
ing on ignorance is the tendency of people (and animals) to choose
mainly in accordance to current utilities, ignoring the effects on the
utilities of future choices, and to underestimate the effects of current
pleasures/pains in decreasing/increasing future enjoyment through the
adaptation effects. For example, most people believe that winning a big
sum of money will bring them great happiness and are thus willing
to spend a lot of money buying lottery tickets at unfavourable odds.
However, psychologists find out that lottery winners are typically
not happier than non-winner controls. (See Brickman et al., 1978;
Herrnstein and Prelec, 1992.)
Most people regard our own species as the only one that has moral-
ity. I am agnostic on this. (I cannot rule out some rudimentary sense of
morality in, say, the chimp.) However, I accept that the sense of moral-
ity evolved very late in the phylogenetic scale. The inborn feelings for
morality probably helped our survival by enhancing cooperation in our
species which relies less on hard-wired fixed response patterns. Our
sense of morality is not only inborn, we are brought up and educated,
and we also learn and teach others, to behave morally, to value justice,
etc. This is extremely important in the smooth functioning of the
society and in improving human relationships. Morality is not only
important for the welfare of the individual but also for others. If any-
thing, I believe that modern education is deficient in the moral aspect.
However, the inborn and learned strong sense of morality also, at least
occasionally, make us do and prefer things consistent with our moral
principles but contrary to our welfare. If an individual acts against his
own welfare in order to observe certain moral principles, this need not
be irrational. First, his sacrifice may benefit others. Secondly, he may
feel bad in violating principles he values. Thirdly, violating moral prin-
ciples may have long-term undesirable effects both for him and for
others, including weakening the degree of morality. Fourthly, his choice
may be due to ignorance. However, if an individual knowingly sacrifices
so much welfare not made up by all of the above considerations put
together, he is irrational according to my definition.
An example of welfare-reducing moral principle that had been
observed for a long time with disastrous consequences is the scruple (in
ancient China in particular) that a woman has to stick to a single
husband even after the early death of the husband. Statues were erected
to extol the virtue of not marrying again, observing the widowhood for
decades from youth to death. Enormous family and social pressures were
put on widows, with or without children, not to marry again, simply
because it is not virtuous to marry again. Most people agree now that
such blind insistence on moral virtue is irrational and indeed morally
wrong. Shouldn’t those who insist on whatever moral principles now
irrespective of or independent of their welfare implications take some
lessons from this example? (It may be thought that draconian anti-
welfare moral principles, such as the no-remarriage commandment, no
longer exist today, especially in the West. However, consider the prin-
ciple ‘life is sacrosanct’ and the related strict anti-euthanasia legislation
and the artificial prolonging of hopeless lives in pain by modern
machines. The amount of unnecessary suffering is still enormous.)
The above causes of irrational preference show that, due either to
imperfection in our endowed faculty, the biological bias in favour of
reproductive fitness, or certain side-effects of morality, we may do things
not quite consistent with our welfare. The issue here is that, for nor-
mative purposes, should we use welfare or actual preferences/behaviour.
It seems clear that, where individual irrational preferences are due to
the bias of biological drives or to imperfect telescopic faculty, social
choices should try to exclude such irrational preferences and be based
on individual welfares instead. An old Chinese dictum says, ‘Out of
the three unfilial acts, not having offspring is the greatest.’ However, for
the human species as a whole, we are certainly not getting smaller in
population size. Moreover, a long-run social welfare function account-
ing for the welfare of future generations should account for that. If we
go for biological fitness, we will prefer unlimited procreation even if that
means that we will all be suffering to a smaller population with a higher
aggregate welfare. We are the feeling selves that care ultimately about
our welfare (positive minus negative affective feelings). We are not them,
the unfeeling genes that, through random mutation and natural selec-
tion, programmed us to maximise fitness. Unlike other species who are
almost completely controlled by their genes and the environment, we
have learned to change our fate by using such measures as birth control.
For normative issues, it is our welfare, rather than the selected random
dictates of the unfeeling genes, that should count.
For irrational preferences due to rigid adherence to out-dated moral
principles, it may be more controversial. In fact, some people may be
against calling such preferences irrational. However, before the evol-
ution or development of morality and the like, we (perhaps still in the
form of apes) had no moral or other principles, no concept of commit-
ments and justice, etc. Self-interest dominated totally, though this does
not exclude genetically endowed ‘altruism’ for the maximisation of
inclusive fitness. As we evolved and relied more and more on our high
intelligence and social interaction for survival, the instinct for moral
feelings also evolved which helped our survival by enhancing cooper-
ation. This was enhanced by learning the importance of such moral
practices as honesty in improving our struggle against nature (includ-
ing wild animals) and against competing human groups. No one
can deny that the initial evolution/development of morality must be
purely instrumental (in enhancing either our welfare or our surviving
and propagation fitness) as there existed no morality to begin with.
We then learned and taught our children and students to value moral
principles, etc. first as a way to increase the degree of adherence to these
principles and hence our welfare. Eventually, some, if not most, people
came to value these principles in themselves by learning and probably
also by instinct. (The evolution of such commitment enhancing devices
as blushing can be fitness-enhancing; see Frank, 1987.) Failing to see
the ultimate values is a kind of illusion fostered by learning (I dare not
say indoctrination) and perhaps genetics. However, I personally have
great moral respect for people with such illusions. They most probably
make better citizens, friends and colleagues. But illusions they are
nevertheless, at least at the ultimate analytical or critical level. While
on the whole positive (in maintaining the moral standards), such
illusions do have some costs in delaying the rejection of certain out-
dated moral principles and in contributing to the wide disagreement in
normative discussion.
For those who agree with Harsanyi in rejecting spurious preferences
for normative purposes, it seems likely that irrational preferences due
to biological drives and imperfections and psychological conditioning
should similarly be rejected. However, there is a class of preferences
(‘autonomous desire not based on hedonistic considerations’) that may
be classified as irrational using my definition but insisted by Harsanyi
to be respectable, as discussed in the next section.
4.3 Autonomous desires
Instances of autonomous desires not based on hedonistic considerations

given by Harsanyi (1997, pp.132–3) includes: altruistic desires, desires
for accomplishments, our natural desire to satisfy our curiosity. I allow
for altruistic desires to be rational as a concern for the welfare of others.
This has been discussed in Section 4.1 above.
I quite agree that some of our desires are not based on hedonistic con-
siderations. However, I believe that the satisfaction of desires or prefer-
ences as such has no intrinsic normative value; it is the effects on
happiness that is ultimately valuable. Otherwise, why are spurious pref-
erences not normatively important?
We do have a natural desire to satisfy our curiosity. We are probably
both born and brought up to be so. As our species manages successfully
to survive mainly based on its superior intelligence and knowledge,
curiosity has a fitness-enhancing effect. Hence, we are also rewarded in
satisfying our curiosity by feeling very good. And the increased knowl-
edge contributes to future success. Thus, the satisfaction of this desire
is generally consistent with our welfare. However, there are cases where
the satisfaction of our curiosity actually makes us worse off. If a person
sees a box in her office, she will naturally open it to have a look. If a
colleague prevents her from doing so by seizing the box away from her,
she will be made a little unhappy. However, if the box contains a
poisonous snake, she will be very grateful to her colleague. It may be
said that, in this case, it is only the uninformed preference that is vio-
lated, not the informed one. But suppose the box contains a photograph
of her mother being raped by a soldier. Then, even after being informed
of that, she may still be unable to overcome her curiosity and will open
it to have a look. If this disturbs her a lot, she will be made much worse
off than by the seizing away of the box from her. Suppose that there are
no benefits, directly or indirectly, of so looking at the photograph, to
her or to others and that there are no side-effects of seizing the box away
from her. Most people will agree that, in this case, seizing away the box
from her is the right thing to do. If I were her, I would be most grateful
for that. The satisfaction of my (even if informed) preferences as such
has no normative significance for me; it is important only because, in
most cases, it makes me (and/or others) happier, directly or indirectly.
Now consider the desire for accomplishment. This may also have
some genetic basis and may be related to the accumulation instinct
possessed by many animal species. However, it is clear that this desire
is also affected by social and educational influences. We are not only
born and taught, but we also learn, to want to do useful things. Parents
teach and influence their children to be so for the good of themselves,
their children, and perhaps also of others. Having a desire to achieve is
generally good for the welfare of ourselves and of others. However, good
qualities may also be detrimental occasionally. Wise people try to avoid
this but not all people are perfectly wise. An extreme example of the
lack of wisdom is the saying ‘If I cannot be famous for my good deeds,
I will still try to be notorious for my bad ones.’ Once a person has a
strong desire to achieve, the satisfaction of the desire will make him
happy. However, he may be willing to undertake so much hardship
that far exceeds the happy feeling of achievement. This may still not be
irrational if his accomplishments greatly increase the welfare of others.
However, most of us have on some occasions been trying to fulfil our
desire for accomplishment (or for revenge) in such a way and to an
extent that decreases our own welfare without increasing that of others.
If this is not just due to ignorance, it is irrational according to my
definition. Though this definition may be debatable, we may achieve
some agreement by noting that there are different degrees of ir-
rationality. In any case, I do not find that there is any intrinsic norma-
tive significance in having the desire for accomplishment fulfilled,
except for the welfare effects of such fulfilment, including the indirect
ones through the accomplished deeds. Other autonomous desires (such
as preferring a less happy ‘real-world’ life to one of being attached to a
pleasure machine) may be similarly analysed.
Now, consider a concrete real example. As reported in Ming Pao Daily
(a reliable leading daily in Hong Kong) on 23 March 1997 (p.A11), a
man in Tienjin was sent to a hospital after fainting while cycling.
Further investigations revealed that he decided to buy a cellular mobile
phone costing more than $9,000, despite having a monthly salary of
only about $600 and a life saving of only $5,000 (all in Chinese dollars).
He thus cut down on all his expenses including food. After more than
four months of semi-fasting, he managed to buy the cellular mobile
phone with a loan of another $2,000 from relatives. The phone was not
for any business or other essential use. Rather, he used it to show off to
his friends, cycling from one house to another, ending up in the hos-
pital. Maybe he was ignorant of the possibility of fainting. However,
even if he did not faint, I do not think that the happiness he would
obtain from showing off his phone would be more than his welfare loss
from spending more than $9,000, including making himself rather
unhealthy from semi-fasting. His desire to have the phone, whether
autonomous or not, is likely to be irrational.
4.4 Why is happiness fundamental?
Why do I regard the satisfaction of preferences, desires, etc. as such not

of normative significance while happiness is? Why is happiness funda-
mental while other things important, ultimately speaking, only to the
extent that they directly or indirectly contribute to happiness? The
simple answer is that happiness/unhappiness is good/bad in itself and

no other thing is so in itself.
There are many things we want: money, job security, status, freedom,
etc. However, we do not want them for themselves, but to make us more
happy or less unhappy. But we want happiness for itself. It is true that,
being happy may also have some instrumental values such as making
us healthier and/or more successful in our jobs. However, being
healthier or being more successful in jobs are, ultimately speaking, only
valuable (in the normative sense) by contributing to happiness directly
or indirectly. Happiness is itself valuable without having to contribute
to anything else. If my happiness does not make any other sentient less
happy, it is valuable. I do not have to argue with philosophers for thou-
sands of years to establish this since I can directly perceive the enjoy-
able/painful feelings and know that they are intrinsically good/bad in
themselves. But for extreme philosophical solipsists, no one will object
to my presumption that other normal individuals have the same ca-
pacity for such enjoyment/suffering. (To be less controversial, let us
ignore animal welfare here.) Their enjoyment and suffering are intrin-
sically good/bad to themselves.
To see that happiness is more fundamental than preference, consider
advanced computers in the 21st or 22nd century that have preferences
but no affective subjective feelings. Clearly their preferences should not
count morally. If it is replied that only human (informed) preferences
should count, not machine preferences, then consider animals now and
advanced computers in the 25th century that do have subjective affec-
tive feelings, i.e. they have pain, joy, etc. then most morally sensitive
persons will agree that their welfare should also count. Thus, clearly
welfare is more important and fundamental than preferences, informed
or not, ultimately speaking.
If we follow (as I largely do) Harsanyi in rejecting actual and opting
for informed preferences, it is difficult not to go all the way to happi-
ness, in contrast to preferences, as what is ultimately normatively valu-
able. This is particularly so if we realise that much of our actual and
informed preferences and/or our actual behaviour are shaped by our
genes to increase our biological fitness which may be at variance with
our actual welfare. As pointed out above, it is our welfare that is nor-
matively valuable rather than the dictates of the unfeeling genes formed
by random mutation and natural selection. Lastly, as we can all natu-
rally feel that enjoyment/suffering is intrinsically good/bad but cannot
naturally feel the same for anything else that may be held to be intrin-
sically good/bad, the burden of argument rests with those who want to
replace/supplement happiness by/with something else as ultimately

good/bad.
At the risk of repetition, the argument for recognising happiness as
ultimately the valuable thing at the fundamental moral philosophical
level does not rule out the importance of insisting on such useful prin-
ciples as honesty, freedom, democracy, law and order, justice, human
rights, etc. at the practical, political, or day-to-day level. However, recog-
nising the real ultimate objective will help us in making decisions on
more fundamental issues like the trade-off of the useful principles es-
pecially when they are in conflict with each other, the long-term choices
of the appropriate institutions and principles to promote. The existence
of an objective more fundamental than preference also allows us to
analyse more fruitfully changes in preferences which have left econ-
omists largely voiceless. As economists assume given preferences and
take preferences as the ultimate objective, they are largely at a loss when
preferences changes. This partly explains their insistence on stable pref-
erences and try to explain most, if not all, of the changes in terms of
changes in constraints, especially incomes and prices. They have indeed
been very successful in many of these explanations. However, true
changes in preferences are also quite prevalent in the real world (see
Bowles, 1998). The use of happiness gives us the appropriate criterion
to judge the desirability of changes or policies in the presence of pref-
erence changes. (On the cardinal measurement and interpersonal com-
parison of happiness, see Ng, 1996a and Chapter 2 above.)
5
Utilitarianism
Utilitarianism is the belief that the ultimate objective of the society

should be the maximisation of the unweighted sum of individual wel-
fares (or utilities, ignoring the differences between the two; in the pres-
ence of such differences, it is argued in Chapter 4 that welfares should
be the one to use). A utilitarian SWF is the unweighted sum of indi-
vidual welfares. If we do not confuse non-ultimate considerations with
basic values, it seems natural that the right ultimate objective for the
society should be the unweighted sum of individual welfares, and was
taken to be so by all classical utilitarians. (The difference between pref-
erence and welfare, discussed in Chapter 4, is largely ignored in this
chapter.) Many modern philosophers and economists are sceptical.
However, arguments for the compellingness of utilitarianism are over-
whelming. In my view, utilitarianism has never been adequately shown
to be unacceptable, if the correct ‘ultimate’ view is taken. (And if we are
confining utilitarianism to the society rather than to the individual
level; see Goodin, 1995 for an argument that utilitarianism makes more
sense as a public than as a private moral philosophy.)
5.1 Compelling arguments for utilitarianism
Apart from the basic requirements that social welfare is a function of

individual utilities only (welfarism, defended in Chapter 3) and that
social welfare increases with individual utilities (the Pareto principle,
which is uncontroversial, given welfarism), the utilitarian SWF entails
three increasingly precise requirements:
1. Separability
That the (social) weight attached to any individual utility is inde-
57
pendent of the utility levels of other individuals. (This is called

separability because it means that SWF is then a separable function
of individual utilities.)
2. Linearity
That the weight attached to any individual utility is a constant (i.e.
unchanged), making social welfare a weighted sum of individual
utilities and increases linearly with each individual utility. This is a
more precise requirement than separability because the latter allows
the weight on any individual to be dependent on the utility level of
that individual.
3. Unweighted sum
That social welfare is the unweighted sum of individual utilities.
Compelling arguments supporting the above three steps toward the

full utilitarian SWF are presented below. (Only the direct substantive
justifications are included. Utilitarian results based on informational
restrictions, e.g. D’Aspremont and Gevers, 1977 and Maskin, 1978, are
not discussed.)
1. Separability – Fleming (1952) and Sugden and Weale (1979)
Fleming establishes separability, i.e. W = Sf i(W i) from a very reasonable
set of axioms: the existence of a SWF,1 the Pareto principle, and the
‘Elimination of Indifferent Individuals’ (strong version). For some prob-
lems, the existence of a SWF (which presupposes a social ordering) may
be a strong assumption to make. But this axiom imposes no restriction
at all on the question of which SWF we should adopt. Taking also the
Pareto principle as acceptable, let us thus concentrate on the crucial
‘Elimination of Indifferent Individuals’ (Fleming’s Postulate E).
Postulate E (Elimination of Indifferent Individuals): Given at least

three individuals, suppose that both individuals i and j are indiffer-
ent between x and x¢ and also indifferent between y and y¢ but
individual i prefers x (or x¢) to y (or y¢) and individual j prefers y
(or y¢) to x (or x¢). (See Figure 5.1.) Suppose also that all other
individuals are indifferent between x and y, and indifferent between
x¢ and y¢ (but not necessarily indifferent between x and x¢ and
between y and y¢). Then social preferences must always go in the same
way between x and y as they do between x¢ and y¢.
The ethics behind the postulate is the following. In the choice

between x and y, since all other individuals are indifferent, social pref-
erence should be decided by the preferences of individuals i and j.
Utilitarianism 59
Suppose the preference of individual i of x over y is judged on some

grounds (stronger, worthier, etc.) to outweigh the dispreference of j such
that x is socially preferred to y. Then in the choice between x¢ and y¢,
since all other individuals are again indifferent for both i and j, x is indif-
ferent to x¢ and y is indifferent to y¢, so the preference of i of x¢ over y¢
must also outweigh the dispreference of j to give x¢ socially preferred
to y¢.
The strength and possible objections to the postulate spring from the
fact that, while all other individuals are indifferent between x and y
and between x¢ and y¢ they are not necessarily indifferent between x
(or y) and x¢ (or y¢). Thus some people may want (probably based on
some misunderstanding, see below) to base the social judgement as to
whether the preference of individual i should outweigh the disprefer-
ence of j on the welfare levels of other individuals. One possible reason
for this is that welfare levels are misunderstood to be something else
such as income levels. Then, even if individual i (and also j) has the
same income levels between x and x¢ and between y and y¢, his welfare
change between x and y may be different from that between x¢ and y¢
if the income levels of other individuals differ and produce external
effects on i (and also j). However, Postulate E does not refer to income
levels but holds individual i (and j) indifferent between x and x¢ and
between y and y¢, taking everything into account. The objection to
Postulate E requires in fact the rejection of individualistic ethics. As
Harsanyi (1955) explains, Postulate E is a natural requirement of indi-
vidualistic ethics.
Fleming’s set of axioms ensures that social welfare is a separable func-
tion of individual welfares, i.e. W = Sf i(W i ). For any given set of the f i
functions, we may adopt monotonic transformations making W a sum
of the transformed individual welfare indices, as Fleming does. This,
however, is not permitted by the cardinal welfare framework. As
Fleming himself is emphatic in pointing out, the transformed indices
need not bear direct relationship to say a cardinal measure of happi-
ness. Thus, in our interpersonal comparable cardinal welfare framework,
we must view Fleming’s result as no more than separability.
Another contribution in support of a separable SWF is Sugden and
Weale (1979). They adopt a set of very compelling axioms within
the individualistic, contractarian framework which implies Fleming’s
Postulate E and the separability in social welfare function. I find their
argument so persuasive that I do not find their following conclusion
presumptuous: ‘In so far as our argument is valid, therefore, future
formulations of welfare economics should include this separability
requirement as a matter of course.’ I do not believe there is any escape

from this separability satisfying the individualistic ethics. Those
who prefer a non-separable SWF should have a hard time reconciling
with their simultaneous belief (as most of them apparently do) in
individualism.
However, separability may easily be regarded as a stronger result than
it really is. In fact, within the confine of a separable SWF, enormous
flexibility is still possible. For example, if we write W = S(Wi)1-a/(1 - a)
where a is some constant, we have a Paretian and anonymous SWF
which is separable. But by varying the value of a, we can accommodate
virtually all degrees of ‘egalitarianism’. With a = 0, we have the Bentham
SWF; with a = 1, we have the Nash SWF, with a approaching infinity,
we approach the maximin SWF. Thus, the result on linearity is an impor-
tant advance over separability.
2. Linearity – Harsanyi (1953, 1955)
Harsanyi shows that W = Ski(Wi) (where each ki is a constant) by as-
suming that social as well as individual preferences satisfy the set of
axioms for expected utility maximisation (Marschak, 1950) and that the
society is indifferent between any two prospects which are indifferent
to every individual.
3. Unweighted sum – Harsanyi (1953, 1955) and Ng (1975a, forthcoming,
a)
Harsanyi also has a more intuitive argument based on the impersonal
observer approach. Each individual is to indicate ‘what social situation
he would choose if he did not know what his personal position would
be in the new situation chosen (and in any of its alternatives) but rather
had an equal chance of obtaining any of the social position existing in
this situation,2 from the highest down to the lowest’ (Harsanyi, 1955,
p.316). Given this hypothetical choice, it is clear that if the person
wants to maximise his expected utility (which can be argued to be the
only rational objective; see Ng, 1984a), he would choose the alternative
that maximises the sum of individual utilities.
That an individual may have difficulties judging the utility levels of
other individuals only concerns the practical difficulties of obtaining
the interpersonally comparable individual utility indices. If we have
these utility indices, Harsanyi’s argument means that we should use the
sum of utilities as the appropriate social objective rather than any other
form.
Another case for utilitarianism is advanced in Ng (1975a). This is
based mainly on the following postulate:
Utilitarianism 61
Weak Majority Preference Criterion (WMP): For any two alternatives

(or social states) x and y, if no individual prefers y to x and (1) if n,
the number of individuals, is even, at least n/2 individuals prefer x
to y; (2) if n is odd, at least (n + 1)/2 individuals prefer x to y and at
least another individual’s utility level is not lower in x than in y, then
social welfare is higher in x than in y.
As Mueller observes:
WMP is obviously a combination of both the Pareto principle and

the majority rule principle that is at once significantly weaker [hence
more acceptable as a sufficient condition for a social improvement]
than both. In contrast to the Pareto criterion it requires a majority
to be better off, rather than just one [individual], to justify a move.
And, in contrast to majority rule, it allows the majority to be de-
cisive only against an indifferent minority. In spite of this apparent
weakness, the postulate nevertheless proves strong enough to
support a Benthamite social welfare function.
Mueller, 1989, p.435
The reason why WMP leads us to the utilitarian SWF is not difficult
to see. It requires that individual utility differences sufficient to give rise
to preferences of half of the population must be regarded as socially
more significant than differences not sufficient to give rise to prefer-
ences (or dispreferences) of another half. Since any group of individuals
comprising 50 per cent of the population is an acceptable half, this
effectively makes a just perceivable increment of utility of any individ-
ual equivalent to that of any other individual.
Ng (forthcoming, a) has a different argument for full utilitarianism.
Starting from separability as compellingly established by either Fleming
(1952) or Sugden and Weale (1979) (see Section 5.1 above), the follow-
ing condition is added:
Local Cardinal Individualism: For any four social alternatives, if all
individuals apart from any single individual j are strictly indifferent between
all the four alternatives, then the social welfare difference between a pair of
these alternatives is higher than that between another pair if individual j’s
utility difference between the first pair is higher than that between the second
pair. In symbols, for all a, b, c, d in the set of social alternatives, for any
individual j, if a I i b I i c I i d for all i π j (where Ii stands for strict indif-
ference by individual i), then:
Uj( a ) - Uj( b ) > Uj( c ) - Uj( d ) implies W ( a ) - W ( b ) > W ( c ) - W ( d )
The compellingness of Local Cardinal Individualism may be

explained. Essentially, it says that, where no one else is affected, the
cardinal intensity of preferences by any single individual must be
respected. Here, everyone else is strictly indifferent (not even any slight-
est unnoticeable preference or uncertainty or whatever) between all the
four alternatives concerned. Only individual j is affected by the choice
among these alternatives. If the society is to respect her preferences and
if she prefers a to b more strongly than she prefers c to d, the society
should also prefer a to b more strongly than c to d. From this reason-
able condition of local cardinal individualism and the compelling
separability condition, full utilitarianism (unweighted sum) may then
be proven.
The fourth case for full utilitarianism is provided in the next
section. Following on from the argument in Section 3.3 above that
rational individualism implies welfarism, it is shown that rational
individualistic egalitarianism (a strengthened version of rational
individualism plus a compelling version of egalitarianism) implies
utilitarianism.
5.2 Rational individualistic egalitarianism

implies utilitarianism
In this section, it is argued that rationality, (a strong version of) indi-

vidualism and egalitarianism imply utilitarianism. Thus, utilitarianism
is not only consistent with but is also demanded by egalitarianism at
the most fundamental level of ultimate objectives. Hence, it is also con-
sistent with some concept of ‘justice as impartiality’ (title of Barry,
1995). Whoever one is, one carries the same welfare weight at the most
fundamental level of the ultimate objective of individuals and society
– happiness.
In addition to Rational Individualism discussed in Section 3.3, we add
the following axioms.
Axiom 2. Egalitarianism: All individuals should be treated equally in
social choice.
In itself, Egalitarianism does not answer ‘equality in what?’. However,
together with Rational Individualism (which implies Welfarism as
shown in Section 3.3), it means that individual welfares should be
treated on an equal footing. However, this still does not imply (welfare)
utilitarianism (which maximises the unweighted sum of individual wel-
Utilitarianism 63
fares), if ‘equal footing’ is given only a general interpretation. Thus,

‘equal footing’ does not rule out giving w A (the welfare of individual A)
a higher weight than that of individual B when w B > w A, provided w B is
also given a similarly higher weight than w A when w A > w B. Egalitarian-
ism does imply anonymity or symmetry in the social welfare function.
The Nash SWF, W = w A.w B satisfies egalitarianism but the SWF, W = 2w A
+ w B does not.
Axiom 3. Strong Individualism: For any social alternatives x, x¢,
y, y¢, if all non-(individual)-welfare variables that may affect social
welfare are held unchanged over all these alternatives,3 and if all indi-
vidual welfares are exactly identical except for any two individuals A
and B, (i.e. wi(x) = wi(x¢) = wi(y) = wi(y¢) for all individuals i except A and
B); w A(x) - w A(y) = w A(x¢) - w A(y¢) > 0; w B(y) = w B(y¢) > w B(x) = w B(x¢); then
W(x) > W(y) implies W(x¢) > W(y¢).
The welfare levels of all individuals except A and B are exactly ident-
ical over all the four alternatives. The welfare levels of A and B are illus-
trated in Figure 5.1. (Note: for wA, x¢ and y¢ could be at the same level
as, or higher than, x and/or y.) The excess of wA in x over that in y is
exactly equal to the excess of wA in x¢ over that in y¢. This equality is in
terms of the welfare level, what the individual and the society rationally
value ultimately. If the equality is in terms of income or some other
non-ultimate variables, one may say that the excess of $100 when the
income level is 1 million dollars is less important than the same excess
x y
x¢ x y¢
y
x¢
y¢
i ′s indifference map j ′s indifference map
Figure 5.1
of $100 when the income level is only one thousand dollars. However,
here the excesses are already in terms of welfare. If there is diminishing
marginal welfare of income, the same welfare excess when income is
high must already mean larger income excess. The society may also
not want to place the same weight on the two excesses or differences if
other individuals are differently affected. However, here individual B is
affected by exactly the same amount and all other individuals are not
affected at all. Thus, there is no individualism-consistent reason not
to give the welfare excess w A(x) - w A(y) the same social weight as the
welfare excess w A(x¢) - w A(y¢). These are valued to be exactly the same
in welfare to individual A. Strong Individualism requires the society to
value these two welfare differences also equally. Then, if the society
views the welfare difference w A(x) - w A(y) as big enough to outweigh the
negative welfare difference w B(x) - w B(y) to decide W(x) > W(y), it must
also regard the welfare difference w A(x¢) - w A(y¢) as big enough to out-
weigh the negative welfare difference w B(x¢) - w B(y¢), as x and x¢ gives
exactly the same welfare level to B and so do y and y¢, and all other indi-
vidual welfare levels are exactly the same. Thus, although labelled
‘strong’ to distinguish it from Axiom 1a, Strong Individualism is really
compelling.
It might be thought that, despite the fact that w A(x) - w A(y) = w A(x¢) -
w (y¢), the same welfare difference w A(x¢) - w A(y¢) may be regarded as
A
socially more important than that of w A(x) - w A(y) for the following con-
sideration. Suppose that person A is poor (in income) at y¢ and below
average at x¢, rich at y and very rich at x, with other things being the
same. While A may regard the increase in income from rich to very rich
as of similar welfare significance to her as the increase from poor to below
average, others in the society may regard the latter increase as more
important. In fact, they may not like A getting very rich as much as they
do not like seeing A in poverty. If this is true, as is likely to be the case,
it is socially more important to increase A’s income from poverty to
below average than to increase her income from rich to very rich.
However, this does not violate Strong Rational Individualism which
holds the welfare levels of all individuals except A and B (which is used
for comparison with A) unchanged. In the above consideration, other
individuals will have lower welfare levels in y¢ than x¢ and lower welfare
levels in x than in y.4 If other individuals all have the same welfare levels,
there is no individualism-consistent reason to give the same weight to
welfare differences of the same value for the same person.
What if the person herself gives different weights to welfare differ-
ences of the same value? Certainty it will then be individualism-
Utilitarianism 65
consistent to give accordingly different social weights to welfare differ-

ences of the same value. However, while one can have diminishing
marginal welfare of income (as a unit of income satisfies more urgent
needs when the income level is low) or of any other non-ultimate vari-
able, one cannot logically have diminishing marginal welfare of welfare.
As non-affective altruism/malice, ignorance, and irrationality have all
been excluded by axioms 1d1–1d3 (see Section 3.3), the welfare of the
individual concerned is the only thing affecting her preference. It may
be thought that one may still have diminishing marginal utility (repre-
senting preference) of welfare. However, it can be shown that, provided
the welfare levels are measured appropriately, it violates compelling
rationality axioms to have diminishing (or increasing) marginal utility
of welfare, as shown in the next paragraph. The appropriate cardinal-
isation of welfare measurement is using the same welfare unit (say, one)
to represent a welfare difference of a just perceivable increment of
Edgeworth (1881). Edgeworth regarded it as ‘a first principle in-
capable of proof’ (pp. 7ff., 60ff.) that such a just perceivable increment
is equatable across all pleasures and across all persons. This equatabil-
ity is shown as a result from some compelling axioms in Ng (1975a).
(See Section 5.1 above.)
For simplicity, consider an atemporal framework. (If time is intro-
duced, the just perceptible increment of welfare have to be specified
over a just perceptible time duration.) As non-affective altruism/
malice, ignorance, and irrationality have all been abstracted away, we
may use welfare (representing happiness) and utility (representing
preference) interchangeably. The concept of just perceivable increments
recognises the fact (consistent with commonsense and well established
by psychologists) that human beings are not infinitely sensitive in
their capacity to enjoy welfare. Using one to measure the welfare dif-
ference of a just perceivable increment,5 it is not difficult to see that, if
two welfare differences of the same value are given different weight,
then a perceivable preference will be given a lower weight than an
imperceivable one. This is unacceptable. (For more details, see Ng,
1975a, 1984a.)
Definition 2. Strong Rational Individualism requires both
Rational Individualism (Axiom 1, Section 3.3) and Strong Individualism
(Axiom 2).
Proposition 2. Strong Rational Individualism and Egali-
tarianism imply Utilitarianism.
Proof: Rational Individualism implies welfarism. Where others are
not affected, social welfare is a positive function of and only of indi-
vidual welfares in the society. (See Section 3.3.) In the specification of

Strong Individualism, the case where w A(x) = w A(x¢) and w A(y) = w A(y¢)
is not excluded. Since Strong Individualism dictates that W(x) > W(y)
implies W(x¢) > W(y¢), as long as all individual welfare levels (other than
those of A and B) are the same over the four social states x, y, x¢, y¢ (i.e.
irrespective of being the same at what levels), this means that the rela-
tive social evaluation of the welfare levels of A and B does not depend
on the welfare levels of other individuals. Since A and B can be any pair
of all the individuals, this makes the social evaluation of the welfare
level of each individual independent of the welfare levels of other indi-
viduals. Hence, the social welfare function is a separable function of
individual welfares. Another way of demonstrating this is to note that,
for the case where w A(x) = w A(x¢) and w A(y) = w A(y¢), Strong Individual-
ism6 is equivalent to the axiom ‘Independence of Indifferent Individ-
uals’ of Fleming (1952) who proves that this axiom, together with the
Pareto principle and the existence of a social welfare function, imply
that social welfare is a separable function of individual welfares. Since
the Pareto principle (Axiom 1b above) and the existence of a social
welfare function (Axiom 0 above) are also assumed here, we also have
this separability result:
W = Â f i (w i )
From Strong Individualism, each and every f i must be linear in wi. This
is so because if any f i is not linear in wi, the social welfare significance
(effect on W) of some equal welfare differences will be different. Strong
Individualism must thus be violated. We must thus have W separable
and linear in individual welfares:
W = Â ai w i
where the ai are constants. From Egalitarianism, ai has to be the same

for all i. We lose no generality and make no difference in substance in
making all ai equal to one, yielding:
W = Â wi
which is utilitarianism. Q.E.D
5.3 A defence of Harsanyi against some recent criticisms
In the last dozen of years, Harsanyi’s utilitarianism results have been

subject to rather prominent criticisms. Sen (1986, pp. 1122–4) questions
Utilitarianism 67
the significance of Harsanyi’s impersonal observer result on the ground

that ‘There is no independent concept of individual utilities [other than
the von Neumann–Morgenstern values] of which social welfare is
shown to be the sum, and as such the result asserts a good deal less than
classical utilitarianism does’ (p.1123). This (or a similar) point is made
much clearer and in much stronger terms by Roemer, who asserts that
‘Harsanyi’s Impartial Observer Theorem has nothing to do with util-
itarianism’ (Roemer, 1996, p.149). Roemer sustains this strongly worded
assertion by arguing that the VNM (von Neumann–Morgenstern) util-
ities and the fully measurable and comparable utilities needed for
social choice need have no relation at all. Where the two sets of utility
functions are incompatible with each other, it is easy to show that social
decisions in accordance to the sum according to one set will in general
differ with the sum in accordance to another set. This criticism of
Harsanyi’s result is answered by my result (Ng, 1984a) that, using axioms
no stronger than those for the expected utility hypothesis, with the
recognition of finite sensibility (which is just commonsense and well
established by psychologists), the utility function derived by the
Neumann–Morgenstern method is the same as the subjective utility
function of classical utilitarianism and neoclassical economists such as
Edgeworth (1881). Given this result, Harsanyi’s result does give full
support to utilitarianism.
Apart from a few other axioms no stronger than those used by the
VNM expected utility hypothesis, I was able to establish the equivalence
mainly by the following axiom (which is itself implied by the VNM set
of axioms, assuming infinite sensibility, making explicit preference coi-
ncide with intrinsic preference):
Axiom A: "r, x, y, z: (rIx & zPy) implies [(r, z; 1/2, 1/2) P (x, y; 1/2,
1/2)] where r, x, y, z are alternative outcomes, (r, z; 1/2, 1/2) is the lottery
with 50/50 chance of obtaining r and z, P is intrinsic preference, I and
P are explicit indifference and explicit preference respectively. The need
to distinguish intrinsic and explicit preference/indifference is due to the
recognition of finite sensibility. If the optimal amount of sugar in your
coffee is 1.8, you may not be able to tell a difference between 1.8 and
1.79 and fail to register an explicit preference but 1.8 may be intrinsi-
cally preferred to 1.79. (See Ng, 1975a for more details.) This axiom says
that the same 50/50 probability mix of obtaining a (explicitly) preferred
and an indifferent outcome must be an intrinsically preferred lottery.
This is in the spirit of a semiorder that a preference should outweigh an
indifference to give at least an intrinsic preference. This is compelling
in an atemporal framework where time and future effects have been
abstracted away (on which see Section 5.4 below). Axiom A then implies
that a maximal indifference (continuous with a marginal preference,
or Edgeworth’s minimum sensibile, or just perceptible increment of
pleasure) must be represented by a same positive constant. The VNM
utility indices satisfying Axiom A then must be the same as the Edge-
worthian cardinal utility indices. It is true that the zero point is left
undefined in this. However, for utilitarianism, the zero points do not
matter, only utility differences count for making social choice.
Even without using the above rigorous axiomatic justification of the
equivalence of the VNM utility and the subjective cardinal utility of the
classical utilitarians and neoclassical economists, a commonsense argu-
ment may also establish this equivalence. We only need to note that
the VNM utility indices do not come from thin air only to represent
individual choices over lotteries. Rather, each individual already has her
subjective cardinal utility function to begin with. When I face choices
involving risks, I compare the probabilities and the associated subjective
cardinal utility gains and losses involved before making a choice. More-
over, I choose to maximise, as far as possible, subject to mistakes (which,
together with things like regrets, excitement, give rise to various para-
doxes and intransitivities; see, e.g. Munier, 1988), my expected subjec-
tive cardinal utility. Thus, these subjective cardinal utility functions
exist before the VNM construction is used. The latter is used to discover
the pre-existing individual subjective cardinal utility functions by ob-
serving their choices involving risks. The degree of the pre-existent
diminishing/increasing marginal (subjective and cardinal) utility (of
income or some other objective indices) determines the degree of risk
aversion/preference, not the other way round.
What the ‘pure representation’ economists do is to say that the VNM
utility is purely a representation of individual choices involving risks
and has nothing to do with the subjective cardinal utility of the same
individual. This would be true if individuals did not consult their sub-
jective utility in making choices involving risks. (But then on what basis
do they make rational choices involving risks is rather mythical.) There
is a sense in which these pure representation economists are right. First,
the well-known axioms of the VNM hypothesis does not ensure that
individuals do consult their subjective utilities and try to maximise
expected subjective utility in making choices involving risks. However,
this may be taken as a commonsense requirement for rational choice
or, if one wants to be rigorous, Axiom A above may be assumed to ensure
this. This then makes Harsanyi’s result a full utilitarian one. Secondly,
Roemer (1996, p.142) is correct in claiming that the knowledge of all
Utilitarianism 69
individual VNM preferences or utilities does not give us a meaningful

way of making interpersonal comparisons [necessary for any SWF, a
utilitarian one in particular]. However, Axiom A and the approach of
Edgeworth (1881) and Ng (1975a, 1996a) do give us interpersonal com-
parability. In Ng (1975a), I argue that a natural way of obtaining inter-
personally comparable individual cardinal utility indices is to use a
common number (e.g. one) across all individuals to indicate the utility
difference of a maximal (non-strict) indifference, recognising that
individuals are not infinitely sensitive. (See Section 5.1 above.) In Ng
(1996a), I use this method to actually obtain utility indices that are in-
terpersonally comparable, overcoming the non-comparability problem
of happiness surveys done by psychologists. There are other ways (some
discussed in Section 9 of Ng, 1975a) of obtaining interpersonal compar-
able utility indices. Thus, even if one does not concede that Harsanyi
(1953) is sufficient for utilitarianism, one must admit that Harsanyi
(1953) plus Ng (1975a, 1984a, 1996a) are.
Sen also plays down on the significance of the second Harsanyi’s
(1955) (weighted) utilitarian result, saying that it is primarily a ‘repre-
sentation theorem’, not really utilitarianism. Sen has two objections.
The first is regarding the choice of individual utility indices. This has
already been answered above. The second is that the result is within the
single-profile framework with a given set of individual preferences. My
defence is twofold. First, as the utilitarian result applies within each
single-profile, it really does not matter that it is only a single-profile
result. Secondly, Harsanyi’s result is easily generalised to the multi-
profile framework, as must already be implicit in Roberts (1980) who
establishes corresponding results in both frameworks, and as explicitly
shown in Mongin (1994).
After a long evaluation of the Harsanyi–Sen debate, Weymark con-
cludes that:
If utility only has meaning as representation of [ordinal] preference,

then Sen is correct in regarding Harsanyi’s theorems as social utility
representation theorems. . . . If utility does not simply measure
preference, Harsanyi’s Impartial Observer Theorem can be inter-
preted as an axiomatization of utilitarianism provided (i) well-being
is cardinally measurable and fully comparable, (ii) each person’s well-
being, including that of the impartial observer, is measured by a von
Neumann–Morgenstern utility function, and (iii) the Principle of
Welfare Identity is satisfied.
Weymark, 1991, p.315
This last principle is compelling in the context as it just requires that

the impartial observer’s ordering of the extended lotteries in which he
is individual i for certain agrees with individual i’s ordering of the simple
lotteries. Point (ii) is satisfied given my 1984a result cited above in ref-
erence to Sen’s objection. Point (i) is compelling to assume for the
problem of what form of SWF to take as the existence of any reason-
able SWF presumes the existence of interpersonally comparable indi-
vidual cardinal utilities as argued in Chapter 2. In fact, Weymark (1991,
p.299) himself argues that the ordinal concept of utility cannot ‘provide
an adequate basis for utilitarianism’, which can have ‘meaning [only] if
utility is cardinally measurable and unit comparable’ (p.303). Thus, for
the problem of the appropriate form of SWF, Harsanyi (or anyone else)
must be granted interpersonal cardinal utilities. Thus, despite these
queries regarding the relevance of Harsanyi’s results to utilitarianism, I
believe that Harsanyi (at least after having been strengthened with my
arguments) emerges completely unscathed.
A somewhat different objection to Harsanyi’s utilitarianism may be
considered. Mongin (1988, p.143) asks us to consider a society ‘of 100
identical individuals, whose utility functions are egoistical, i.e. depend
on the individual’s own endowment only. Endowments are distributed
once-and-for-all, and one unit of them is taken to be just above the sub-
sistence level’. Would you prefer:
Society A: 100 persons get 2 units each

or
Society B: 89 persons get 2 units each
1 person gets 1 unit
10 persons get 5 units each
and
Society C: 89 persons get 1 unit each
or
Society D: 90 persons get 1 unit each
Mongin has:
little doubt that some ethical observers would choose A rather

than B and D rather than C on the grounds that complete equality
should be preferred to any other social solution, but when complete
equality is impossible, significant differences in total endowment
Utilitarianism 71
should tilt the balance towards the best-endowed society. There

might even be ethical observers who would choose B rather than C
and C rather than D on the grounds that complete equality is dis-
tasteful, but moderate inequality should be preferred to extreme
inequality. Of course, none of those ethical observers could consis-
tently be utilitarian.
p.143
I wish to argue that one can have either type of ethical preference
and still be a utilitarian, if secondary effects are taken into account.
On the other hand, if secondary effects are abstracted away, as ap-
parently intended by Mongin (note his ‘once-and-for-all’ distribu-
tion and the dependence of utility on own endowments only), there
is no individualism-consistent grounds for either type of ethical
preference.
Mongin’s example is the adaptation of a famous Allais paradox
to the context of social choice. However, the intended paradox is
really there only if there is a confusion of fundamental values with
non-ultimate considerations by the ethical observer. Thus, I, a self-
proclaimed utilitarian at the fundamental level, prefer society A to B
and also prefer D to C, if the secondary effects of social cohesion in
A dominates the higher endowment of B (valuing the various effects
ultimately on individual utilities) and if the higher endowment in D
dominates the slightly more equality of C. However, if the 10 better
endowed persons in B are able to pursue worthwhile activities con-
tributory to future advance in welfare, I prefer B to A; if the extreme
inequality in D results in a polarised society with undesirable secondary
effects, I prefer C to D. Thus, if we take secondary effects into account,
none of the ethical preferences mentioned by Mongin need be contrary
to utilitarianism, nor is any of them inconsistent. On the other hand,
if these secondary effects have been abstracted away (as they should
be, otherwise they should be included in the description of the above
societies), there is no individualism-consistent grounds for having either
of the inconsistent types of ethical preferences mentioned by Mongin.
For example, since the utility of each and every individual depends only
on her own endowment, none of them finds complete equality dis-
tasteful. Then, how could the ethical observer choose ‘on the grounds
that complete equality is distasteful’? Surely, the hypothetical ethical
observer is supposed to choose on behalf of the 100 individuals, not on
behalf of herself! She is just a theoretical construct that does not really
exist.
5.4 A defence of using just perceptible increments
Edgeworth (1881, pp.7, 60) took it as axiomatic (‘a first principle

incapable of proof’) that a just perceptible increment of pleasure, for
all pleasures, are equitable across all individuals. Abstracting from sec-
ondary effects as we should and from different time durations (discussed
below), I find Edgeworth’s axiom compelling. Nevertheless, I derived
the axiom as a result based mainly on the Weak Majority Preference
criterion discussed in Section 5.1. However, the use of just perceptible
increments for interpersonal comparison has been strongly objected
to, including by Hammond (1991). This objection is taken as so con-
clusive that a commentator said that the ‘method of just perceptible dif-
ferences is dead, in the eyes of most economists’. If this were true, most
economists would be wrong, as the method can be strongly defended
below.
The objection by Hammond (see also Arrow, 1963, pp.115–18) is that
the method ‘tend[s] to favour the sensitive’ (pp.216–17). A similar objec-
tion I have heard many times at seminars is that the method would give
less weights to the incomes of the less sensitive (in enjoying income).
This objection is puzzling. A strong (if not the strongest) argument in
favour of higher equality is that a dollar to the poor meets more urgent
needs or yields higher marginal utilities. Thus, a person is more sensi-
tive to income when poor; each dollar yields more perceptible incre-
ments of happiness. So, we should not focus on changes in incomes
only. A million dollar increment to the rich may be worth much less
than half a million dollar increments to the poor. We should ultimately
focus on the subjective units of happiness or utilities. But the method
of just perceptible increments in happiness precisely makes us focus on
these subjective units rather than the objective amounts of incomes or
whatever. To shift focus from the purely objective amounts of pro-
duction, income, etc. to the subjective utilities and using marginal
analysis is an important advance of the neoclassical economists. How
ironic if we were to retreat back to the naive income fetishism!
It is true that income inequality may generate certain undesirable
effects such as less social cohesion, more crimes, etc. This means that
individual welfares in the future will be reduced by inequality now. A
long-run SWF will take account of that. Thus, this alleged problem arises
only because of the confusion of fundamental values with non-ultimate
considerations or because of the confusion of the ultimate objective
with the constraints (intertemporal utility feasibility in particular).
Will the application of the utilitarian SWF lead to a very unequal dis-
Utilitarianism 73
tribution of income? Apart from incentive effects, it is doubtful that

differences in sensibility will lead to a great inequality of income.
Psychological studies in pain sensation show that the pain thresholds
are very close for different individuals (e.g. averaging 230 ± 10
mc./sec./cm2 standard variation), as are the number of just noticeable
differences (Hardy et al., 1952, pp.88, 157). If there are more differences
in the capacity to enjoy income, these are probably due to ‘learning by
doing’, and a long-run SWF allowing for all relevant effects will take
account of that. If some inequality still persists after taking all effects
into account, I cannot see why this is not an optimal distribution if it
maximises aggregate welfare (especially within the scope of this chapter
starting from welfarism; for a defence of welfarism, see Chapter 3). Con-
sider the much-cherished principle, ‘From each according to his ability;
to each according to his needs’ (which I personally approve of, assum-
ing no disincentive effect). Why doesn’t it read, ‘An equal amount of
work from each; an equal amount of income to each?’. If a weak person
is tired by four hours of work, it is better for a strong person to work
longer to relieve him/her. Similarly, if a less-sensitive person does not
enjoy the extra income much, it is better that a more sensitive person
receives more of it. What prevents us from seeing such a simple
analogy?
Thus, I find the ‘weak equity axiom’ of Sen (1973a, p.18) unaccept-
able. The axiom says: ‘Let person i have a lower level of welfare
than person j for each level of individual income. Then in distributing
a given total of income among n individuals including i and j, the
optimal solution must give i a higher level of income than j.’ Thus,
even if giving more income to i does not make her appreciably or even
perceptively happier, while reducing j’s income does make him enor-
mously less happy (though still of higher welfare level than i), the axiom
insists that i should receive more income than j. In my view, what is
important is not the welfare levels of various individuals, but rather how
much increases in welfare, to whomsoever they go (anonymity), we can
have.
Another type of objection to WMP and the resulting use of just per-
ceptible increments may now be discussed. First, it is argued that a just
perceptible improvement may differ across different experiences even
for the same individual. ‘On many occasions, a just perceptible improve-
ment in musical performance means much more to me than a just per-
ceptible quantity of drink’ (Mirrlees, 1982, p.69). This I think may be
due to the difference in time period involved and/or to possible future
effects. A just perceptible quantity of drink may last only a fraction of
a second while the musical performance probably a couple of hours.

Moreover, Mirrlees may also value the recollection of a high-quality
musical performance. As it is, WMP is formulated in an atemporal
basis and hence does not cover the complication introduced by differ-
ences in time period. With time as a dimension, one should then select
a just perceptible improvement over a just perceptible length with
neither future nor external effect as the standard unit, and use indirect
measurement (Ng, 1975a, section 9) for experiences difficult to be so
measured.
Now consider the following objection raised in a personal discussion
by David Friedman (cf. Sen, 1970a, p.94; Pattanaik, 1971, p.150). If indi-
vidual utility levels are continuous (which is the assumption used in
Ng, 1975a), then even an indifference may involve a utility difference.
It is thus not clear that a just perceivable increment of utility should be
regarded as interpersonally equitable. For example, a person poor in per-
ception or reporting may fail to notice or report a utility difference more
significant in some subjective sense than one noticed or reported by
another more perceptive person. On the other hand, if utility levels are
discontinuous so that individual utility does not change until an incre-
ment of pleasure is perceived, then these discontinuous little jumps in
utility may differ across individuals in some sense and hence should not
be regarded as equivalent.
My reply to the above objection is several-fold. First, it may be noted
that the above objection is not against the principle of the utilitarian
SWF, but just against the way the various Ui should be best measured or
interpersonally compared. In other words, if the marginal preference of
a person half as perceptive (in terms of reporting a given amount of sub-
jective utility) as another is counted twice as important, the above objec-
tion will presumably no longer apply. Thus, this objection is properly
regarded as pointing to the difficulty of making interpersonal compari-
son of utility rather than as an objection against the utilitarian SWF
as such.
Secondly, if we shy away from making any interpersonal comparison
of cardinal individual utilities, then a SWF is impossible. Thus, for the
question of which SWF to adopt, we must accept some form of com-
parison. Most methods of interpersonal utility comparison are rough
estimates (if not guesstimates) perhaps partly influenced by personal
value beliefs. It seems to me that the method of measuring the levels of
just perceivable increments of pleasure offers to date the most promis-
ing objective way of making interpersonal comparison. (On some prac-
tical difficulties and ways to overcome them, see Ng, 1975a, Section 9.)
Utilitarianism 75
If we accept the above objection, it just means that even this best avail-
able method is not ideal. With finite sensibility (which is an indis-
putable fact) and without some form of interpersonal comparison, we
cannot make a reasonable social choice except in the unlikely cases
where every individual strictly prefers x to y. Even the indifference of
one individual is sufficient to cloud the issue unless we are prepared
to make the interpersonal comparison that any utility difference associ-
ated with his indifference is not strong enough to overshadow the pref-
erences of others. Thirdly, even if it is true that a just perceivable
increment of utility differs across individuals either due to different
‘perceptive’ powers or due to different ‘jumps’ in utility, how are we to
know whose just perceivable increment is larger and whose smaller? In
the absence of specific knowledge regarding this, it seems best to regard
them as equal (cf. Lerner, 1944, pp.29ff; Sen, 1973b). If we adopt the
convention of designing an average just perceivable increment of utility
as one, then the just perceivable increment of any particular individual,
in the absence of any specific knowledge, may be expected to have a
probability distribution such as depicted in Figure 5.2, that is, the indi-
vidual is just as likely (if at all) to have a lower as a higher than average
value. We can thus minimise our mistake by taking this increment to
be one, the average value (Ng, 1980).
50% 50%
–41 –21 1 2 4
Figure 5.2
Our arguments for taking a just perceivable increment of utility

as equitable across individuals (at least in the absence of an eudai-
monometer) seem to be supported by the following argument of
Harsanyi:
The metaphysical problem would be present even if we tried to

compare the utilities enjoyed by different persons with identical pref-
erences and with identical expressive reactions to any situation. Even
in this case, it would not be inconceivable that such persons should
have different susceptibilities to satisfaction and should attach dif-
ferent utilities to identical situations . . . , and identical expressive
reactions may well indicate different mental states with different
people. At the same time, under these conditions this logical pos-
sibility of different susceptibilities to satisfaction would hardly be
more than a metaphysical curiosity. If two objects or human beings
show similar behaviour in all their relevant aspects open to obser-
vation, the assumption of some unobservable hidden difference
between them must be regarded as a completely gratuitous hypoth-
esis and one contrary to sound scientific method. (This principle may
be called the ‘principle of unwarranted differentiation’.) In the last
analysis, it is on the basis of this principle that we ascribe mental
states to other human beings at all: the denial of this principle would
at once lead us to solipsism. . . . Thus in the case of persons with
similar preferences and expressive reactions we are fully entitled to
assume that they derive the same utilities from similar situations.
Harsanyi, 1955, p.317
5.5 Utilitarianism and process fairness
There is an apparently very persuasive objection to utilitarianism raised

by Diamond (1967; see also Kolm, 1998). I have not read any satisfac-
tory answer to this objection. Consider a society of two identical indi-
viduals, 1 and 2, facing a choice between two alternatives x and y, with
two equally probable states of nature, a and b; let the cardinal and inter-
personally comparable individual utility indices be as shown in Table
5.1 (ignoring z for the moment):
Harsanyi’s set of axioms implies that the society is indifferent between
x and y. But many people may find y (strictly) preferable to x. If we ask
why is y preferable to x, there may be two different answers. First, con-
sider Diamond’s explanation: ‘I am willing to accept the sure-thing prin-
ciple for individual choice but not for social choice, since it seems
Utilitarianism 77
Table 5.1
x y z
a b a b a b
1 1 1 1 0 0 0
2 0 0 0 1 1 1
reasonable for the individual to be concerned solely with final states

while society is also interested in the process of choice’ (Diamond, 1967,
p.766). But the society consists of individuals. If all individuals are con-
cerned with final states, why should the society be concerned with the
process of choice? I can think of only two reasons why the society may
be concerned with the process of choice that are consistent with indi-
vidualism. First, some individuals in the society have views about the
process of choice and will be hurt or feel happy by the adoption of
certain processes. In this case, what needs to be done is just to amend
the pay-off matrices or to construct a more complex one to take account
of this feeling. This is then not a valid argument against the utilitarian
SWF, provided the feeling about the process of choice has been taken
into account in the utility calculus. (Cf. a similar but different resolu-
tion of the problem raised by Diamond’s example by Broome, 1991,
pp.111–15. Even after abstracting from feelings about the process of
choice, Broome reconciles Diamond’s preference with the sure thing
principle by distinguishing the outcomes of ‘death with fair treatment’
and ‘death with unfair treatment’. However, Broome’s argument is not,
nor is it meant to be, a defence of utilitarianism. Thus, my argument
here and below differs from Broome’s but Broome’s defence of the sure-
thing principle is very thorough.)
Secondly, even if no individual in the society now has any feeling
about the process of choice as such, it may still be regarded as a legiti-
mate concern since the process of choice may generate secondary effects
and affect the utilities of individuals in the future who may not exist
now. In this case again, what needs to be done is just to take account
of the effects on individual utilities in the future as well. Thus, I cannot
find any ground consistent with individualism that is a valid objection
to utilitarianism on the grounds of the process of choice or the like. Let
us turn to a different objection.
Even if the feeling for the process of choice, the welfare of future gen-
erations, etc. has been taken into account or has been abstracted away,
some people may still regard y as preferable to x because y gives indi-
vidual 2 a ‘fair shake’ while x does not. To be socially indifferent between
x and y is thus regarded as being unfair against 2. But I fail
to see why it is unfair if the society is also indifferent between y and
z (see Table 5.1 above), as utilitarianism axioms dictate. Is it unfair
against both individuals (!) to be socially indifferent between x, y and z,
despite the fact that both individuals are indifferent between y and
a 50–50 chance of x and z? It is probably difficult to convince those who
still regard y as preferable to both x and z, but I hope that the following
argument will persuade at least some of them to have a second thought.
Let us accept the reasonable assumption that social welfare is con-
tinuous in individual utilities (an infinitesimal change in individual
utilities does not cause a big jump in social welfare). To prefer y over x
and z then implies that if we change the utility indices of y sufficiently
slightly, then this new alternative y≤ is still preferred to x. To see this
more intuitively, note that we have to change the utility indices by a
finite amount (due to continuity), say to y¢ (see Table 5.2; the precise
figure is immaterial) to make it indifferent to x. We can then increase
the utility indices back a little to y≤ which will then be preferable to x
and z. I wish to show that the adoption of a SWF dictating this prefer-
ence is detrimental to the utility of both individuals and hence ir-
rational at least from the viewpoint of individualism, when individual
preferences satisfy the Marschak (VNM) axioms.
We do not adopt a SWF to guide our social choice for one particular
known instance only. Rather, once a SWF is adopted, it is used for all
instances until, for some reason, it has been discarded in favour of
another. Moreover, even if we consider a once-and-for-all choice only,
we do not know which particular alternatives that choice will involve
at the ‘constitutional stage’ or the ‘original position’. Thus, on the
Table 5.2
x y z y¢ y≤
a b a b a b a b a b
1 1 1 1 0 0 0 0.89 0 0.9 0
2 0 0 0 1 1 1 0 0.89 0 0.9
Utilitarianism 79
abstract problem of the ideal SWF, we must consider all possible

instances of choice. For the particular instance of the choice between x
and y≤, the preference of y≤ over x does not appear to be clearly un-
reasonable. The expected utility (EUi ) of both individuals being 0.45
with y≤ (noting that a = b = 1/2) and EU1 = 1, EU2 = 0 with x. Since 2 gains
0.45 and 1 loses 0.55 in terms of expected utility by the choice of y≤
over x, it does not necessarily appear to be a social loss to the non-
utilitarians. Similarly, this is true for the preference of y≤ over z.
However, ex ante, we do not know whether we will be confronted with
the choice between y≤ and x or between y≤ and z; both must be regarded
as equally probable. It then becomes clear that if individual preferences
satisfy the Marschak axioms, both individuals would be in favour of the
adoption of a SWF preferring both x and z to y≤ rather than the reverse,
since the preference of y≤ over x and z gives EU1 = EU2 = 0.45, while the
reverse preference gives EU1 = EU2 = 0.5. Individualism thus clearly dic-
tates the adoption of the utilitarian SWF.
It may be objected that the above argument is based on regarding
(meta) social welfare as a function of expected individual utilities, i.e.
WE = W(EU1, . . . , EU I ) = W(SjqjUj1, . . . , SjqjUjI) where qj is the prob-
ability of a state of nature times the probability of that choice instance.
For example, confining only to the two equally probable choice
instances x against y≤ and z against y≤, the probability of a prevailing
in the choice instance x against y≤ is 1/2 ¥ 1/2 = 1/4. In the presence of risk,
perhaps it is better to maximise expected welfare as a function of ex-
post individual utilities, i.e. EW = SjqjW(Uj1, . . . , UjI). However, it can be
seen that, even using this latter criterion, the choice of y≤ over x and z
is still clearly undesirable since EW(y≤) = 1/2W(0.9, 0) + 1/2W(0, 0.9) and
EW(x, z) = 1/2W(1, 0) + 1/2W(0, 1). Thus, EW(x, z) > EW(y≤) if W is increas-
ing in individual utilities, i.e. if the Pareto principle is satisfied.
In the presence of risk, ignoring differences in (or incorrect) prob-
ability estimates (on which see Harris and Olewiler, 1979), there seem
to be good grounds for either maximising social welfare as a function
of individual expected utilities or maximising expected social welfare as
a function of ex-post individual utilities. Within the framework of indi-
vidualism, neither objective can be rejected as unreasonable. The
reasonableness of both WE and EW can in fact be used as an argument
for a linear SWF. If social welfare is a sum (unweighted or weighted with
constant individual weights ki) of individual utilities, then WE and EW
are equivalent. Thus, WE = SikiSjqjUji = SjqjSikiUji = EW. If social welfare
is not a sum of individual utilities, the maximisation of EW and WE
will not, in general, yield the same result.7 Thus, those who reject social
welfare as a sum of individual utilities must explain why either WE

or EW (but not both) is acceptable and the other is not. This is not
an easy task within the individualistic ethics, ignoring incorrect prob-
ability estimates.
Now consider the argument of Holcombe (1997) that the absence of
envy does not imply fairness as the fairness of the process of choice
itself has to be taken into account. Consider the simple case of divi-
ding two goods between two individuals involving no production and
no difference in desert. The divide-and-choose (one person divides the
total endowment into two bundles, the other chooses one bundle) may
not be perfectly fair (though it may be a fair enough method to be prac-
tically useable most of the times) as it may involve a divider’s advan-
tage. If the preference of each person is known to the other, then the
divider may divide the endowment in such a way that he will end up
with a better bundle than if he is a chooser. Thus, if person J strongly
prefers good X to Y and person K strongly prefers Y to X, then J, as
the divider, may divide the bundle (of, say, 10X and 10Y) into, say,
(2X + 6Y or 8X + 4Y ), knowing that K will prefer 2X + 6Y, leaving her
( J ) with 8X + 4Y. On the other hand, if K is the divider, he may divide
the bundle into (6X + 2Y or 4X + 8Y), knowing that J will prefer 6X +
2Y, leaving him with 4X + 8Y. Thus, whoever is the divider ends up
with 8 units of the preferred good and 4 units of the less preferred good,
in contrast with the 6 + 2 respectively for the chooser. This illustrates
the divider’s advantage. The divide-and-choose method is envy-free in
the sense that no one prefers the bundle of any other person. However,
it is not envy-free in the sense that the chooser prefers to be the divider.
Given the assumption of known preferences and hence the divider’s
advantage, the chooser has the reason to envy the divider for being able
to enjoy the divider’s advantage. Nevertheless, if we stick to the tra-
ditional concept of envy-freeness referring only to the bundle of goods
consumed, then Holcombe is correct that the absence of envy does
not imply fairness. He attributed the ‘problem with using freedom
from envy as a criterion for fairness’ to the fact ‘that the criterion exam-
ines only the outcome of the process rather than the process itself’
(Holcombe, 1977, p.801). In a sense, this is correct. However, in my
view, the reason why the process of divide-and-choose is regarded as
not a (perfectly) fair process and may not be envy-free in a wider sense
is precisely due to the outcomes. (For cases where either the two indi-
viduals have the same preferences or no one knows the other’s prefer-
ences, there is no divider’s advantage, and the above objection to the
divide-and-choose procedure does not apply. However, for the case of
Utilitarianism 81
unknown preferences, a Pareto-inferior outcome may eventuate unless

Pareto-improving exchanges are allowed after the divide-and-choose
exercise. In particular, a risk-averse individual may divide the bundle
equally which may not be Pareto optimal if preferences differ. The
divide-and-choose with exchange may then be regarded as fair. The
fairness or not of the rule depends on outcomes.) Hence, the point
does not negate the acceptability of consequentialism in general and
utilitarianism in particular, if these are not taken to narrowly refer only
to the consequences on the consumption bundles, but to all relevant
consequences.
6
A Dollar is a Dollar: Solution to
the Paradox of Interpersonal
Cardinal Utility
6.1 The paradox of interpersonal cardinal utility
As explained in Chapter 2, the impossibility theorems of Arrow (1951/

1963), Kemp and Ng (1976), Parks (1976), and Sen (1969, 1970a) prove,
and even just common-sense arguments demonstrate, that interper-
sonal comparisons of cardinal utility are necessary for making social
choice which cannot reasonably be based on individual ordinal prefer-
ences only.
On the other hand, there is a long tradition in economics of regard-
ing interpersonal cardinal utilities as impossible or meaningless and
scientifically inadmissible (see, e.g. Kolm, 1993). As Wicksteed (1933)
and Robbins (1938) put it, each mind is totally inscrutable to any other
mind, making interpersonal comparisons of utility mere value judg-
ments without any scientific status. I have argued elsewhere that inter-
personal comparisons of utility are not value judgments but subjective
judgments of fact and that economists are more capable in making
those judgments of fact closely related to their field of study (Ng, 1972).
I also argue (Ng, 1992b) that such views (apparently held by the
majority of economists) on the conceptual impossibility of interper-
sonal comparisons of utility are based on the existence of souls (while
a majority of economists, I believe, are philosophical materialists who
don’t believe in the existence of souls). I argue that individual utilities
are not only cardinal and interpersonally comparable but there are
also practicable (though imperfect) ways to measure and compare util-
ities (Ng, 1975a, 1979/1983, 1985, 1996a). However, I freely admit that
the practical difficulties associated with cardinal utility measurement
and comparison are orders of magnitude higher than those for ordinal
82
A Dollar is a Dollar 83
preferences (which still exist and are significant, though ignored by

most economists).
If we regard interpersonal cardinal utilities as impossible either in
principle or in practice, a paradox is created as they are needed for
making social choice. In the next section, a method is proposed that,
to a very large extent, resolves the paradox of interpersonal cardinal
utility.
6.2 The proposed solution
6.2.1 Using unweighted aggregate costs/benefits or

a dollar is a dollar
The proposal to solve the paradox is to simply use the individual will-
ingness to pay to obtain information on intensities of preferences, the
unweighted aggregate willingness to pay in making social choice, plus
the appropriate redistribution of total purchasing power to address
the issue of equality. This frees us from having to obtain information
on cardinal utilities and to compare them interpersonally, except in the
decision on the appropriate redistribution of total purchasing power.
Before justifying this solution, a number of clarifications are needed.
(Some of these touch on some technical concepts; general readers may
want to skip the rest of this section at least on first reading.)
1. There are some ambiguities and controversies associated with the

measure of the willingness to pay and the issue whether willingness
to pay (CV or compensating variation in income) or willingness to
accept (EV or equivalent variation) (in lieu of something) should be
the appropriate measure. This is related to the problems in the
measurement of consumer surplus. I have discussed and resolved the
issue elsewhere (Ng, 1979/1983, Chapter 4 and Appendix 4A). Basi-
cally, I argue that the differences between the various measures are
usually minor and negligible in comparison to the inaccuracies of data
collection and hence can be ignored in practice. Where the differences
are large, I propose a better measure than either CV or EV called mar-
ginal dollar equivalent, defined as the number of times the relevant
gain or loss is equivalent to that of a marginal dollar.1
2. Where external effects are important, the willingness to pay of the
individuals directly involved may not be sufficient. The willingness
to pay of the externally affected individuals has also to be included.
Similarly, where factors such as second-best interconnections are rel-
evant, they have to be appropriately taken into account. These ef-
ficiency issues do not change the principle of willingness to pay, they

just require the principle to be applied more broadly.
3. Where ignorance of the individuals concerned are involved, the issue
becomes more controversial. Personally, I believe that ignorance with
minor effects should simply be ignored, partly to save administrative
costs and partly because the violation of free individual choice has
indirect costs (for being inimical to freedom). However, where ignor-
ance results in big losses, individual willingness to pay may have to
be overruled or revised. Thus, we have such cases like the prohib-
ition of certain addictive drugs, the fluoridation of water, and sub-
sidised milk for school children. (See Ng, 1979/83, Appendix 10 A.3,
on the grounds for such merit and demerit goods.)
4. Even accepting the use of willingness to pay as the measure of inten-
sity of preferences at the level of the individual, the acceptability of
using aggregate (i.e. over a number of individuals) willingness to pay
is still controversial, even if the issue of equality or distribution is
ignored. This is so because, as shown by Boadway (1974) and
Blackorby and Donaldson (1990), a positive aggregate willingness to
pay (CV) need not ensure that the gainers can overcompensate the
losers even given the feasibility of lump sum transfers. This difficulty
is discussed in Ng (1979/1983, pp.96–8). Basically, it is argued that
this difficulty is due to a change in relative prices as compensation
takes place, possibly making aggregate willingness to pay not per-
fectly accurate as it is based on unchanged prices. Where changes in
relative prices are not huge, as is true for most specific projects or
measures, the inaccuracy involved is negligible. While aggregate will-
ingness to pay does not correspond perfectly with a potential Pareto
improvement, it corresponds closely for most cases.
6.2.2 The justification

The justification for using the unweighted (maximum) willingness to
pay as a measure of the preference intensity of any individual (called ‘a
dollar is a dollar’ principle for brevity) is based on the following two
arguments. First, abstracting away certain difficulties mentioned in the
previous section, the amount an individual is willing to pay to obtain
a certain item reflects the intensity of their preference. The more intense
is my preference for a certain performance, the higher is my maximum
willingness to pay for it. This is less controversial.
Secondly, the reason we can use the unweighted willingness to pay is
based on the argument that it is more efficient to do so and achieve
whatever degree of redistribution desired through the general tax/

transfer system.2 This point needs more elaboration.
The main reason people may be against the principle of a dollar is a
dollar is that a dollar to the poor or the more needy meets more urgent
needs than a dollar to the rich. In fact, I agree with this belief and am
in favour of helping the poor (provided the costs of doing so are not
excessive). However, unless factors such as ignorance, and external
effects are involved (making it efficient to subsidise education and
health care), it is more efficient to help the poor through the general
tax/transfer system instead of overriding the principle of a dollar is a
dollar in specific items, such as using distributional weights in cost–
benefit analysis, using first-come-first-served in allocating car parking
spaces. A counter-argument to this is that the tax/transfer system
involves excess costs in the form of disincentive effects (the problem of
administrative, compliance, and policing costs is discussed under ‘trans-
action costs’ in Section 6.4 below). It is thus believed that it is better to
achieve part of the redistribution through the progressive tax/transfer
system, and partly through specific equality-oriented measures such as
subsidising goods consumed disproportionately by the poor and using
distributional weights in cost–benefit analysis. However, this counter-
argument (held by most economists, including myself before I analysed
the problem carefully) ignores the fact that the use of such specific,
purely equality-oriented policies has the same disincentive effects as
the tax/transfer system, but also has additional efficiency costs by dis-
torting choice.
It is tempting for an economist to think that, since the substantial
redistributive tax/transfer system has significant (larger than marginal)
disincentive effects at the margin, it is better to shift some of the redis-
tributive burden to specific items such as taxing/subsidising items con-
sumed disproportionately by the rich/poor. Though some marginal
efficiency costs of distorting choice are created, they are thought to be
smaller than the reduction in disincentive effects due to relying less on
the progressive tax/transfer system. This belief is incorrect. The reason
is that, assuming rational individuals, the disincentive effects are in
accordance to the total system of tax/transfer, taxes/subsidies, plus all
other redistributive and preferential measures, instead of having a sep-
arate and independent increasing marginal disincentive effects sched-
ule for each of the separate measures. Rational persons, in their
work/leisure choice, do not just ask how much post-tax income they
can earn, but also have a rough idea the utility they can get from con-
suming goods and services purchased from the income. They are trading
off the utility of leisure with the utility from work (which consists of
the utility from consuming the higher income and the positive or
negative utility of work itself). Moreover, the utility of consuming the
higher income is affected by whatever specific redistributive or prefer-
ential measures are in place. Thus, the preferential treatment against the
rich in government expenditure and other areas will add on to the pro-
gressive tax/transfer system to determine the total disincentive effects.
Hence, even if only a marginal amount of specific equality-oriented
measures are used, the disincentive effects involved are not just
marginal. Thus, for the same degree of equality in real income (utility)
achieved, the same degree of disincentive effects is incurred whether we
use only the tax/transfer system or a combination of it and some
specific purely equality-oriented system. But the latter alternative has
the additional efficiency costs of distorting choice, and is thus inferior.
In Ng (1984b), the following proposition is proved.
Proposition A (A dollar is a dollar): For any alternative (desig-
nated A) using a system (designated a) of purely equality-oriented pref-
erential treatment between the rich and the poor, there exists another
alternative, B, which does not use preferential treatment, that makes no
one worse off, achieves the same degree of equality (of real income, or
utility) and raises more government revenue, which could be used to
make everyone better off.
Under alternative B, a more (than alternative A) progressive tax/
transfer system (designated b) may have to be used. By definition,
progressivity in the tax/transfer system is not classified as ‘preferential
treatment’ here. Section 6.3 below argues that complications such as
administrative costs, political constraints, ignorance of benefits dis-
tribution, etc. either strengthen the proposition or do not affect the
main thrust significantly.
In fact, proposition A can be generalised to any efficiency-
inconsistent alternative A, not just an equality-oriented preferential
treatment. The proof of the proposition is exactly unaltered if the pref-
erential treatment is not equality-oriented but inequality-oriented.
Instead of counting a dollar to the poor as worth less than a dollar to
the rich in cost–benefit analysis, it is better to tax the rich less and tax
the poor more (or subsidise them less). Similarly, an alternative based
on random treatment, one based on tradition, etc. can all be shown to
be inferior to some alternative B defined to compensate for the gains
and losses in dismantling the efficiency-inconsistent methods used.
Then, ignoring issues of practical difficulties, individual ignorance, ir-
rationality, and procedural preferences (on the last item, see Ng, 1988),
alternative B must be Pareto superior.
However, the existence of alternative B does not necessarily mean that
it can be identified and implemented. If system a is designed to take
account of second-best considerations, then system b can also be so
designed. But system a may only be consistent with second-best con-
siderations by chance rather than by design. In addition, the infor-
mational costs of designing a system consistent with the second-best
considerations may be prohibitive.3 Then we may not be able to iden-
tify system b. Thus, while alternative B may exist, it may not be feas-
ible to implement. Thus, if we wish to strengthen Proposition A to be
one about the existence of a feasible superior alternative B, it would
apply only in a probabilistic sense. That it (the strengthened prop-
osition) still applies in a probabilistic sense is due to the theory of third
best. Just as it may be consistent with second-best considerations,
system a may also be opposite to the requirement of second best. The
theory of third best (see Ng, 1977) can then be used to show that the
expected gain is negative. Hence, as far as the second-best consideration
is concerned, the use of system a involves negative expected gains.
Essentially, random divergence from first-best rules is as likely to be
inconsistent as to be consistent with the requirement of second best.
Due to the expected concavity in the relevant objective function, the
gain when it is consistent is less than the loss when it is inconsistent.
So the expected gain is negative. (See Ng, 1979/1983, Chapter 9 and Ng,
1984b for a more detailed demonstration of this.)
The half-a-century old argument of Little (1951) may be briefly dis-
cussed at this juncture. Basically, Little argues that one cannot say
that a direct tax (on income) is better than an indirect tax (on some
goods) on the ground that the former taxes all goods at the same rate.
The reason is that there is another good, leisure, which is also untaxed
under direct taxation. Taking the simple case of just two non-leisure
goods (see the top half of Figure 6.1), an indirect tax on one of the
goods creates a wedge (between relative costs and benefits) between
that good (good 1) and the other good (good 2) as well as between that
good and leisure. A direct tax causes a wedge between good 1 and leisure
and between good 2 and leisure. Hence, no conclusion can be made on
the desirability of direct versus indirect taxes without specific infor-
mation on the interrelationships between the various goods and leisure.
While this argument is formally correct in a certain sense, it is mis-
leading in its conclusion. Thus, it is certainly true that an indirect
tax on half of the (non-leisure) goods may be less distorting than a direct
tax (equivalent to a uniform tax on all non-leisure goods) of equal

revenue, if the former happens to concentrate on goods more compl-
ementary to leisure. However, if the former is just a random half, then
it may be taken to be worse than the direct tax, at least in terms of
average expectation. To see this, first consider a case where all goods are
of equal degree of complementarity/substitutability to leisure. An indi-
rect tax on half of the goods is then inferior to a direct tax of equiv-
alent revenue. On the other hand, if (the first) half of the goods are
more complementary to leisure than the other half, while an indirect
tax on the first half is more efficient than the direct tax, this efficiency
gain is less than the efficiency loss when the second half is taxed instead.
(Both points are illustrated in specific mathematical examples available
from the author.)
In terms of the ‘wedge’ argument of Little, we may say that, while a
direct tax (of say 10 per cent) drives a moderate wedge between both
goods and leisure, an indirect tax on half of the goods (typically in
excess of 20 per cent to yield the same amount of revenue) drives a
much bigger wedge between the taxed and untaxed goods, and between
the taxed goods and leisure. (See Figure 6.1.) It is thus inferior to the
direct tax at least in terms of expected value. If the goods taxed are ran-
domly selected with respect to their complementarities to leisure, this
bigger wedge between the taxed goods and leisure already induces, in
terms of expected value, as big a disincentive effect (as the case of the
direct tax). In addition, there is a distortion effect on individual choice
between goods. Thus, in spite of Little’s argument, a direct tax is better
than an indirect tax not designed to take advantage of second-best con-
siderations (complementarities with leisure). Our argument here is con-
sistent with the general view in public finance that, other things being
equal, a broader tax is a better tax.
Equality-oriented policies and other non-efficiency-oriented policies
in the real world are, as far as I am aware, without exception, not based
on the second-best consideration. Due to the non-taxability of leisure,
goods highly complementary to leisure should be taxed more on the
grounds of second best. Due to the higher income tax rates on the rich,
the second-best consideration may suggest imposing taxes on the rich
even higher than on others for the same leisure-complementary goods,
and imposing higher taxes on leisure-complementary goods that are
predominantly consumed by the rich. It may thus be thought that
higher taxes on luxuries (defined as goods consumed disproportionately
more by the rich) may be justified on efficiency grounds. However, it is
equally true that goods highly substitutable to leisure should be sub-
Little’s Argument
Direct Tax Indirect Tax
G1 G2 G1 ≠ G2
≠ ≠ ≠
L L
(The ≠ sign signifies a wedge between MRS and MRT, i.e. relative benefits and costs)
Our Argument
Direct Tax Indirect Tax
G1 G2 G1 ≠ G2
≠ ≠ ≠
L L
Figure 6.1
sidised (or at least taxed less). Hence, due to the higher income tax rates
on the rich, the second-best consideration also suggests giving more sub-
sidies to the rich than others for the same leisure-substitutary goods,
and giving higher subsidies on leisure-substitutary goods that are dis-
proportionately consumed by the rich. Thus, luxuries as such do not
justify higher taxes. (The proposal to impose high taxes on ‘diamond
goods’ which are valued for their value is a separate matter of pure effi-
ciency; see Ng, 1987a and Part II of this book which argues for more
public spending partly based on this point.)
In practice, it is also informationally, politically, and administratively
impossible to design and implement specific taxes/subsidies and other
specific economic measures in accordance to the requirement of second
best. This requires full information of the interrelationships of all rele-
vant sectors in the economy. It may be informationally feasible to take
account of the more important interrelationships, but political consid-
erations may make it infeasible. For example, the political feasibility of

giving higher subsidies on leisure-substitutary goods disproportionately
consumed by the rich is certainly suspect. In addition, the adoption of
the principle of a dollar is a dollar will save substantial administrative
costs and achieve much simplicity. Its political feasibility, though cer-
tainly better than the second-best policies of subsidising the rich more
than others for the same leisure-substitutary goods, has yet to be fos-
tered through a long process of education, starting with the under-
standing of the argument of this chapter by economists.
Our proposed solution largely, but does not completely, resolve the
paradox of interpersonal cardinal utility since interpersonal comparison
of cardinal utility is still needed in determining the optimal tradeoff
between efficiency (incentive) and equality in the general tax/transfer
system. Moreover, for cases where individual willingness to pay cannot
be used as a good reflection of welfare due to serious ignorance or ir-
rationality (e.g. hard drugs), interpersonal comparison of cardinal utility
may also be needed on specific decisions.
6.3 Economists should be in favour of reversed weighting!
At the risk of repetition, it should be emphasised that our argument for

using the unweighted sum of willingness to pay or for treating ‘a dollar
as a dollar’ is not based on valuing a dollar to the poor as worth no
more than a dollar to the rich. In fact, personally, I value the former as
worth much more than the latter. If manna were to fall from heaven
(and regarded as once only, or at least as unrelated to income), I would
certainly hope that they fall to the poor. However, if we adopt a policy
of treating a dollar to the poor as worth more than a dollar to the rich,
this will create disincentive effects and hence is inferior to using the
tax/transfer system to reduce inequality. I am not against the attempt
to reduce inequality, just against using inefficient methods to do so.
It is true that governments may not actively and optimally pursue dis-
tributive justice through income taxes and transfers. However, in the
long run, some degrees of distributive balance are maintained. Even if
this does not lead to an optimal tradeoff between equality and ef-
ficiency, it is incorrect to ignore that some degree of balance is being
maintained. As far as I know, all those in favour of using distributional
weights or inequality-averse criteria effectively ignore any degree of
such a balance, or, at least they have not shown their awareness of the
implication of such a balance on the appropriate distributional weights.
Of course, they are aware of the existence of such a balance. However,
this balance implies that the distributional weights should be less

unequal (e.g. a dollar to the poor counted as only $1.20 instead of $2)
than the case in the absence of any balance. In the presence of an
optimal balance, no weights (or only equal weights) should be used. In
the presence of an excessive balance (i.e. equality pursued at excessively
high incentive costs), the distributional weights should be reversed (i.e.
more weights to the rich than to the poor). Nevertheless, no advocate
(as far as I know) of distributional weights or inequality-adverse criteria
has explicitly shown some awareness in this respect. Many people prob-
ably believe that the distributional weights should be proportional to
the social marginal utility of a dollar. However, this can only be justi-
fied if the use of distributional weights have no disincentive effects
(which are efficiency costs over and above their direct distortive costs).
I strongly suspect that many of those in favour of distributional weights
simply ignore these disincentive effects.
It is possible that, despite some degree of balance between equality
and efficiency, some economists may regard the balance as being far
too inadequate. They may think that much more equality should be
achieved despite the inefficiency costs involved. While this is certainly
possible, the reverse case that an economist may believe that too high
efficiency costs are being incurred to achieve equality should be even
more likely, or at least as likely. As a group, economists are unlikely to
be more equality-inclined than politicians, bureaucrats, or voters gen-
erally whose inclinations influence government policies. (In fact, there
is evidence suggesting that economists are much more ‘right-wing’ in
comparison to the general public; see, e.g. Ng, 1988.) Secondly, eco-
nomists are more aware of the efficiency costs of pursuing equality,
including administrative, compliance, policing, and disincentive costs.
Thus, one should expect more economists to find the actual tradeoff
between equality and efficiency being excessively in favour of equality
than the other way round. So, one should expect more economists in
favour of using reversed distributional weights than in favour of using
the normal distributional weights. However, to my knowledge, not a
single economist has come out in favour of the reversed weighting. In
fact, the most radical in this respect is probably Posner (1981), a lawyer,
who is in favour of wealth maximisation. (My argument of treating ‘a
dollar as a dollar’ differs from Posner’s as I allow for the achievement
of equality through taxes/transfers beyond wealth maximisation.) These
considerations support my suspicion that many economists in favour
of distributional weights simply ignore the disincentive effects involved,
not just because they are more egalitarian.
What about the minority of economists who are genuinely more

egalitarian than the prevailing policy? In my view, they, as economists,
should not be in favour of using distributional weights and purely
equality-oriented policies. However, as citizens, they should campaign
to move government policies towards more equality (preferably by
using more efficient methods), while the majority of their fellow econ-
omists, as citizens, should campaign for less equality if, as argued above,
they view the equality-efficiency trade-off as having been pursued to an
excess.
6.4 Some qualifications
6.4.1 Political constraints on redistribution through taxation

It is argued above that, instead of using weighting, quotas, or other
kinds of preferential treatment to achieve the objective of equality (not
as second-best correctives); it is better to adopt a more progressive
income tax schedule. What if the taxation system cannot be changed
to the desired structure due to political constraints? If it is true that we
can’t change the taxation system but we can effect redistribution by
other means, my conclusion may have to be qualified accordingly,
though there is still the ethical question of the desirability of doing good
by stealth. Yet, why should the political constraint act only to prevent
redistribution through taxation and not redistribution by other means?
Maybe because the voters are rather irrational. I suspect, however, that
on this issue, voters are very rational and practical. The upper and
middle classes will not only vote a government out of office for carry-
ing out drastic changes in taxation but also for carrying out other drastic
redistributive measures. Especially in the long run, the forces that
operate to prevent redistribution through taxation will also operate to
prevent redistribution by other means. If we are thinking in terms of a
distributional equilibrium, the distribution should be considered not in
terms of money income but in terms of real income. Naturally, if the
rich are penalised in other ways they will have less tolerance with
regards to the progressiveness of taxation. For example, had Australia
been operating closer to the principle of ‘a dollar is a dollar’ in the past,
the reductions in the progressiveness of its income tax schedule under-
taken in 1978 and later would probably not have been required. Both
the rich and the poor would probably have been better off.
It is true that actual political decisions are affected by a host of factors
and not just by an impartial consideration of a balance between equal-
ity and efficiency. However, equality and efficiency are important con-
siderations and the fact that preferential treatment is an inefficient tool
to achieve equality and efficiency has to be pointed out.
6.4.2 Ineffectiveness in the tax/transfer system

If the tax/transfer system is not effective due to, say, tax avoidance/
evasion while redistribution through certain specific measures is more
effective, there may be a case for using such specific measures. This does
not violate Proposition A (a dollar is a dollar) since the case for using
such specific measures is not purely justified on purely equality-oriented
considerations but also on that of effectiveness in getting the result. It
may, however, be regarded as a practical qualification to the principle
of a dollar is a dollar. (The argument of Kelsey, 1988, for putting eggs
in many baskets due to uncertainty may be similarly viewed.) Never-
theless, the significance of the qualification may be questioned, at least
for many cases. For example, if one may hide one’s income to pay less
taxes and get more transfers, one may also hide one’s income to avoid
penalty prices or to qualify for subsidies on specific items.
6.4.3 Transaction costs

Our analysis has been based on the assumption that the additional
‘transaction’ costs associated with redistribution through more progress-
ive income taxation are not higher than the transaction costs of its
alternatives. Apart from the disincentive effect (which has been taken
into account and is not subsumed under transaction costs), the costs
associated with income taxation seem to fall mainly under: (i) costs of
administration on the parts of both the taxpayers and the collectors; (ii)
costs involved in tax evasion and enforcement; and (iii) costs of tax lob-
bying activities and the like. It is recognised that all of these forms of
costs are substantial. But the relevant amounts are not total costs, only
marginal costs. For good or for bad, income taxation will be with us in
the foreseeable future. The incremental costs of administration of a
more progressive tax system seem trivial. The costs of a change from
one system to another may not be trivial, but will probably not be sub-
stantial. However, at least in the long run, the relevant comparison
is the cost of administering two alternative systems, the difference
between which is probably quite negligible. A more progressive tax
system may however involve higher costs in encouraging more evasion
and more lobbying activities. But the increase in progressiveness is in
lieu of some system of preferential treatment which itself is a subject of
evasion and lobbying. While this is a subject where a precise con-
clusion can hardly be expected, it does not seem probable that costs
involved in the latter (i.e. preferential treatment) will be much lower
than costs involved in the former (i.e. a more progressive tax system).
On the other hand, the costs of administering a pure taxation system
are almost certainly significantly lower than those of administering
a system of taxation combined with preferential treatment in govern-
ment expenditure. Hence, consideration of transaction costs seems to
strengthen, not weaken, our central conclusion.
There may be specific cases whereby the use of an apparently ‘pref-
erential’ policy may be superior to making the income tax system more
progressive because of the significantly lower transaction (evasion, lob-
bying, and administrative) costs of the former. But such a policy is jus-
tified on its efficiency consideration of lower transaction costs (relative
to income taxation) and cannot be justified purely on the equality con-
sideration advanced by egalitarian lobbyists.
6.4.4 Ignorance of benefit distribution

Our argument is based on the assumption that individuals know the
distribution of costs and benefits in government expenditure across in-
come groups, or the details of preferential treatment, so that the incen-
tives are the same as an equivalent pure income taxation system. In
practice, this knowledge is unlikely to be perfect. On the other hand,
most individuals do know the scale of income taxation. Does this
asymmetrical knowledge mean that the disincentive effects of income
taxation are more severe than an equivalent preferential expenditure
system, as, in effect, argued by Martin Feldstein (1974, p.152)?
(However, de Bartolome, 1995 shows that at least half of the taxpayers
underestimate their real marginal rates by confusing it with their
average rates.)
In the absence of perfect knowledge, an individual has to base his
choice on his estimates. From the fraction of knowledge he possesses,
it seems that he is as likely to overestimate as to underestimate the
degree of progressiveness implied in a given preferential expenditure
system, depending on the psychology of the individual in question.
Hence, on the whole, the degree of incentives is likely to be similar
between the preferential expenditure system and the pure income
taxation system.
6.4.5 Redistributive effects of the project itself

The argument for treating a dollar as a dollar does not imply that a
billion dollars is always equal to a billion dollars. This point can be seen
clearly by considering a simple example. Consider two alternative pro-
jects: project M will increase the incomes (after allowing for cost share)
of one million individuals by $10 thousand each, and project N will
increase the income of one single random individual by $10 billion.
Ruling out costless lump sum transfers (that would make us indifferent
between the two projects), it is clear that project M will be preferred to
project N by all social welfare functions egalitarian in incomes. This
preference is not based on valuing a marginal dollar to the rich as lower
than a marginal dollar to the poor. Rather, it is based on treating the
first dollar as more valuable than the 10 billionth dollar, to whomever
they go. (Due to diminishing marginal utility, the same person typically
regards the loss of $10,000 as more significant in utility terms than the
gain of $10,000.) Hence, the equality-incentive argument used above
does not apply here. However, the equality-incentive argument can be
used to dispel the possible belief that, since a project that itself creates
inequality is inferior to one with the same aggregate net benefits but
which does not create inequality, a project that creates equality must
be preferable to one with the same aggregate net benefits but distri-
butionally neutral. Consider a third project O that will yield the same
aggregate net benefits of $10 billion but be distributed across the
economy in such a way that the poor will have much higher benefits
and the rich have negative benefits. While this may seem to be a good
thing in itself, the incentive argument will show that project O is in
fact inferior to project M. If project O were to happen as a natural event,
it would be preferable. But if it is chosen instead of project M, it will
produce disincentive effects.
From the above, it may be said that, for projects whose redistributive
effects are marginal, one can simply choose in terms of aggregate net
benefits; for projects whose redistributive effects are significant, we
should prefer the one with less redistributive effects, given the same
aggregate net benefits. This seems to lend support to the concept
of a conservative social welfare function discussed by Corden (1974,
p.107).
6.4.6 Preference for working

If an individual prefers to have his income by earning it instead
of receiving it as a transfer welfare payment, then a cost–benefit analy-
sis that does not take this preference into account may be misleading.
(This has been emphasised by Skolnik, 1970. For an analysis of the social
norm of living on one’s work, see Linbeck et al., 1999.) This kind of
complication can be taken care of by appropriate shadow pricing. For
example, in the particular case considered here, the main difference
is the possible preference of an individual for earning his income in-
stead of receiving some kind of dole money. This can largely be taken
care of by putting an appropriate shadow price on employment. For a
single person without dependents, his income from a low-paid job
is likely to be sufficient to preclude him from receiving a subsidy. All
that is needed is a low or zero income tax, so he will not have to suffer
the feeling of being on the dole. For families with dependents, the
subsidy can be effected in the form of, say, substantial child-endowment
payments differentiated according to income levels. A fixed child-
endowment is used in Australia with no one feeling ashamed of receiv-
ing it and the introduction of differentiation is unlikely to change this
substantially.
6.4.7 Unexpected emergencies

In times of unexpected emergencies such as earthquakes, wars, etc.
certain necessities may be in very short supply. In principle, we could
impose appropriately higher taxes on the rich and those who happen
to own the goods in short supply and pay subsidies to the poor and the
victims of the disaster. Then the policy of ‘a dollar is a dollar’ could still
be best. However, due to time lags, imperfect information, and the like,
it may be practically infeasible to effect the required changes in
taxes/subsidies in time. Rationing of basic necessities such as medical
supplies (which also involves external economies) may then be the best
practical solution. However, the possible desirability of violating the
principle of ‘a dollar is a dollar’ in such emergencies does not mean that
the same is true for normal times.
6.4.8 Non-income indices for preferential treatment

Our analysis concentrates on the use of income as the index for pref-
erential treatment and redistributive taxation. But surely income is an
imperfect measure of ‘deservingness’, and non-income variables such as
health and age status are likely to enter distributional objectives. In par-
ticular, the use of age as an index for preferential treatment will create
few, if any, disincentive effects, since one cannot change one’s age.
However, we can similarly use age as an index for the purpose of tax-
subsidy. The purpose of giving assistance to the aged, for example, can
be achieved without the additional efficiency costs of, say, giving
free milk as some may not wish to drink milk. (Subsidised milk to
schoolchildren may, however, be justified on the efficiency ground
of merit wants; on merit wants as a possible efficiency ground, see Ng,
1979/1983, Section 10A.3). The consideration of non-income factors
does suggest that a single tax based on incomes only may not be suf-
ficient; the tax-subsidy system may have to take non-income factors

into account.4
A related problem of using income as the basis for taxation is that
measured income may be a poor indicator of actual earning potential
due to savings, risk bearing, etc. Thus, persons of the same earning
potential may be taxed more if they are more willing to bear risk,
to save, etc. However, this imperfection applies also to the use of
measured income for the purpose of preferential treatment and hence
does not affect our argument. In general, to the extent that a better
index is available for use as a basis of preferential treatment, it can also
be used for the tax-subsidy purpose. Unless there are asymmetrical trans-
action costs (see 6.4.3 above), no qualification to our central argument
is necessary.
From the discussion above, it may be concluded that none of the com-
plications seems to change our central argument significantly.
6.5 Concluding remarks
The argument of this chapter has far-reaching implications. It justifies

the separation of equality and efficiency considerations in public policy.
On all specific areas, public policy needs only be concerned with ef-
ficiency, leaving the objective of equality to be achieved through the
general tax/transfer system. This creates a tremendous simplification
in the formulation of all public policies and in cost–benefit analysis in
particular. (On the case and actual uses of distributional weights in
cost–benefit analysis, see Little and Mirrlees, 1974, Adler and Posner,
1999, and references therein.)
However, the argument of this chapter does not deny the possibility
that some measures may improve efficiency and equality simul-
taneously and some measures may harm both simultaneously due, for
example, to such considerations as the physiological effect of income
on work effort (Strauss and Thomas, 1995), the principal–agent prob-
lems and imperfection in the credit market due to informational
asymmetry which may be lessened by the use of collaterals that the
very poor lack (Binswinger et al., 1995; Hoff and Lyon, 1995 and Hoff,
1996), moderation of tax progressivity on wage demands of unions
(Hersoug, 1984; Creedy and McDonald, 1990 and Lockwood and
Manning, 1993), the discouragement of investment of high inequality
which increases the difference between median and average incomes
and hence the median voter demand for higher tax rate (Persson and
Tabellini, 1994; but see Perotti, 1996 on the negative empirical evidence
for the mechanism, though the negative relationship between inequal-

ity and growth appears valid; see Birdsall et al., 1995), and the crime-
reduction effect of equality (Eaton and White, 1991 and Grossman,
1994). (For a summary of empirical evidence on the relationship
between inequality and growth, see Bénabou 1996. On other related
issues, see Haveman, 1988; Murphy et al., 1989; Schmid, 1993; Bowles
and Ginitis, 1996; Baland and Platteau, 1997; Breen, 1997; Le Grand,
1997; Devins and Douglas, 1998; Nagel, 1998; Putterman et al., 1998;
Aghion and Williamson, 1999; and Lee and Roemer, 1999. Also,
Ackerman et al., 1997 contains a large selection of readings in related
normative, social and other ‘extra-economic’ issues in summary form;
it provides enlightening reading, though an economist will typically
find most of the points made not inconsistent with economic analysis
and some are even misleading or just a difference in terminology.) Our
argument is also consistent with the idea of regulating or taxing more
heavily the incomes of superstars in sports and entertainment, as they
earn more than their marginal contribution since their role may be
assumed by the runners up without significant losses (Borghans and
Groot, 1998).5 We also ignore the possible role of increasing returns in
affecting the efficiency–equality relationship. (See, e.g. Brown and Heal,
1979 which shows that the efficiency of marginal-cost pricing may
depend on distribution in the presence of increasing returns.) Our argu-
ment also does not deny the need to take account of indirect effects,
including external effects and those arising from a change in prefer-
ences. (On preference changes, see Bowles, 1998.) It may also be noted
that Congleton (1997) has an interesting argument for the equal pro-
tection of the law due to its efficiency in avoiding the differential enforce-
ment of the law for the interest of the governing coalition. This
argument may be misleadingly regarded as a counter-example to the
principle of a dollar is a dollar. However, the equal protection of the law
is really a second-best solution to the violation of the principle of ‘a
dollar is a dollar’ (from the viewpoint of the whole population) by a
subset of the population, the governing coalition, at a greater cost to
the non-governing sections of the population. Thus, there is really no
violation of the principle.
7
Economics versus Politics
In Chapter 6, it is argued that a dollar should be treated as a

dollar to whomsoever it goes to achieve efficiency, leaving the objec-
tive of equality to the general tax/transfer system. This proposition
of ‘a dollar is a dollar’ can be generalised to show the Pareto inferi-
ority of any efficiency-inconsistent alternative A, not just an equality-
oriented one. Interpreting this generalisation widely, the principle
of ‘a dollar is a dollar’ can be applied to areas outside the tra-
ditional confines of economics. Then, what prevents the application
of the simple efficiency principle from being used in political
issues like the election of a president or members of a parliament,
where the principle of one person one vote seems to be universally
accepted?
It is tempting for one to believe that the answer lies in the need to
limit the power of the rich. Using the simple efficiency principle of ‘a
dollar is a dollar’ would allow the rich too much power in political de-
cisions. This answer is incomplete, to say the least. The main point can
be explained simply. If we limit the range in which the market or ef-
ficiency principle apply, we need to have more inequality in income dis-
tribution to maintain the same degree of incentives since the range of
things income can be spent on has been reduced. Thus, if the same
degree of incentives is to be maintained, the poor cannot be made better
off. In fact, every income group will be made a little worse off due to
the inefficiency of restricting the working of the market. On the other
hand, if we are prepared to reduce the degree of incentives in order to
make the poor better off, we could do this more efficiently by just
making the tax/transfer system more progressive, without restricting the
working of the market. This point is shown more precisely in a simple
mathematical example available from the author, showing that the
99
random restriction of the range of market operation makes both the rich
and the poor worse off.
In fact, the point can be seen without the mathematical example. To
see it more clearly, abstract from the effects of changes in relative prices
(which could go either way) by considering a case of fixed earning abili-
ties and fixed prices (as could be facilitated by constant returns, small
country, etc.). Starting from a situation where some of the goods
(private, no external effects) are not subject to market allocation but are
supplied in fixed amounts by the government to everyone (possibly
equally), dismantle the government provision and allow these goods to
be allocated in the market. The savings of the government from not
having to provide these goods are then used to pay each individual,
with each individual getting the worth of the previously allocated
amounts of these goods in money and free to buy whatever quantities
desired. Then obviously no individual will be worse off as she can buy
the previously allocated quantities if so desired. (It is assumed that
the transaction costs of market allocation are not higher than that of
government allocation.) Those choosing to buy different combinations
(including buying more or less of some or all goods) of these goods are
made better off.
The market economy can be shown to function perfectly efficiently
in the sense of Pareto optimality mainly under the assumptions of
perfect competition, adequate information by the market participants,
and the absence of external effects. Thus, possible reasons for govern-
ment intervention or even replacing the market include objectives
beyond Pareto efficiency on the one hand and certain serious violations
of the above assumptions (called ‘market failures’ below) on the other.
In my view, a valid objective beyond Pareto efficiency is distributional
equality. (See Ng, 1988, 1990a for arguments on the unacceptability of
many other proposed objectives.) However, as argued above, the objec-
tive of equality may justify a progressive tax/transfer system but does
not justify specific efficiency-inconsistent measures. It is true that this
argument assumes the feasibility of the appropriate tax/transfer system.
Where such a system is politically or administratively infeasible or
subject to very important uncertainty in its effects (Kelsey, 1988), it may
be efficient to supplement the tax/transfer system with other specific
equality-oriented measures, including commodity taxes (Cremer and
Gahvari, 1995). While such a possibility cannot be completely ruled out,
problems of administrative costs and political difficulties are likely to
be more serious for the specific equality-oriented measures, as argued in
Economics versus Politics 101
Section 6.4 above. Thus, we have to look to market failures for valid
reasons for replacing the market.
Different degrees of market failures prevail all over the place, most
requiring no or only limited intervention instead of supplanting the
market altogether. Thus, the existence of some market failures is not
enough; we have to look for very substantial, even overwhelming,
extent of market failures. I do not think we can find this with respect
to market power and ignorance. The wealthiest person may be able to
wield many times more influence than an average person if willingness
to pay is taken into account, but in most countries, no single person
earns a substantial fraction of the GNP. While substantial ignorance is
involved in political voting, it is not substantially alleviated by going
for the one-person-one-vote solution. This leaves external effects as the
possible source of substantial market failure.
Three distinct types of external effects may be associated with politi-
cal decisions. First, political decisions taken may externally affect people
beyond the jurisdiction and people in the future. This external effect
has little to do with the issue here, since the choice of either one-person-
one-vote or willingness-to-pay has negligible effects on these external
effects. Secondly, political decisions, being concerned with public issues,
involve the public-good external effect. Both the choice (usually in elec-
tions) of public officials and the decisions of the public officials on
public issues (including law and order, other regulations, provision of
public goods) are related to the public-good problems, including that of
free riding. Thus, if voters are allowed to actually pay money to express
their willingness to pay, most of them may pay nothing, preferring
to free-ride. However, this problem only makes the willingness-to-pay
principle difficult to put into effect, it does not make the principle
undesirable. If we have more sophisticated method of inducing self-
interested voters to express their true willingness to pay, such as the use
of the Vickrey–Clarke–Groves incentive-compatible mechanism on a
carefully selected sample of voters (Ng, 1979/1983, Section 8.3), the free-
riding problem could be overcome. Thus, I see the next external effect
as the most important stumbling block to the use of the efficiency prin-
ciple in political decisions.
The third type of external effects associated with political decisions is
less similar to the normal external effects familiar in economics; it may
even be controversial to describe it as an external effect. The function-
ing of a modern society, including its economy, requires political
stability, the maintenance of law and order, and preferably also a high
degree of social cohesion. These elements may also be desirable for the
psychological well-being of individuals in the society. These desirable
elements are fostered by a political system using the one-person-one-
vote principle in its political process instead of the willingness-to-pay
principle, except for a society where the majority of its citizens are as
good economists as you (i.e. the likely readers of this book), and this is
well nigh impossible, at least in the foreseeable future. The use of the
efficiency principle in political process is likely (if not certainly) to gen-
erate feelings of apathy, antagonism, and the like against the govern-
ment, the wealthy, and law and order among the poorer classes and
those in the political fringe. This will increase the difficulties or costs of
maintaining law and order and will likely reduce stability and cohesion.
Thus, the exercising of the preference intensities through the willing-
ness to pay in traditionally political sphere, especially by the rich, may
be said to entail important external costs. In most cases, this consid-
eration makes the use of one-person-one-vote principle optimal in the
political process despite the existence of the normal inefficiency as-
sociated with supplanting the market. This is the most important factor,
as far as I can see, in explaining the separation of political and economic
spheres. However, since this factor depends on the attitude of the
people, as this attitude changes with time and with the understanding
of economics, the optimal sphere of economics will gradually expand.
Nevertheless, knowing the difficulty of teaching basic economic prin-
ciples even to university students, I am not at all optimistic about the
speed of this expansion and about the ultimate conquest.
PART II
How Much Should the
Government Spend?
The central public finance question facing any country is the

appropriate size of its government.
Feldstein, 1997, p.197
It is not the intention here to discuss all issues relevant to this central
question of public finance. Important questions like the relative ef-
ficiency of government versus private supply, the feasibility of the
private production of public goods, informational asymmetry, etc. are
not addressed except in passing. Instead, some neglected and perhaps
more important issues are focused. After reviewing the basic theory on
the appropriate size of government spending on public goods and some
empirical considerations, Chapter 8 argues for higher public spending
based on the importance of diamond goods, relative-income effects, and
environmental disruption effects, the unimportance of absolute income
for welfare at the social level at least for developed economies, and the
high welfare potential of public spending especially in research (the
example of a possible quantum leap in welfare through brain stimu-
lation is discussed in Appendix A). With reasonable assumptions, it is
shown that the optimal relative (to the GDP) size of public spending
increases with per capita income (Appendix C). Chapter 9 addresses the
question: Should the benefit–cost ratios for public projects exceed one
by a sufficient margin to account for the distortionary costs of raising
government revenue?
103
8
A Case for Higher Public Spending
Co-author: Siang Ng
I like to pay taxes. With them I buy civilization.

Justice Holmes in Eisenstein, 1961, p.5
Saunders (1993) shows that the share of public expenditure reached a

peak in most countries in early to mid-1980s, stabilised and then
declined. (See also Bohl, 1996 and Payne and Ewing, 1996 on inter-
national evidence on the Wagner’s law on the increase in the share of
the public sector.) The decline is sharper in terms of the real size of the
public sector due to a higher relative price; see the end of Section 8.2
below. This decline is associated with the worldwide movement towards
privatisation and the use of the market mechanism. While accepting
that there are some valid reasons (such as the inefficiency of the
public sector) underlying this movement, this chapter discusses some
neglected factors that provide some offsetting considerations. Before
doing so, the basic theory on the appropriate size of public spending
on public goods and some related considerations are reviewed.
8.1 Some simple theories of government spending
8.1.1 Government spending as providing public goods

It is clear that a major part of the spending of most governments is on
items (e.g. the maintenance of law and order, defence, etc.) that are
uncontroversially public goods in the Samuelsonian sense (non-
rivalness in consumption). Others are more controversial. A major part
of government spending is paying the public employees. However, if
these employees are largely necessary for the government to perform
the functions of providing public goods, this part of government spend-
105
ing may also be regarded as the cost of providing public goods. In many
countries, governments spend a lot on transfer payments such as old-
age pensions and unemployment benefits. These are regarded as redis-
tributive spending rather than spending on public goods. While these
items certainly have redistributive effects, they also have some public-
goods aspect. Thus, keeping the unemployed paid reduces crimes and a
more equal distribution of income increases social cohesion and enters
positively into many people’s utility function. Hence, while the aspect
of public goods provision does not exhaust the function of government
spending, it is the major one for most countries. For simplicity, only
this major aspect is focused here.
A public good is non-rivalrous in consumption. My tuning to a radio
or TV broadcast in no way affects your reception of it. Thus, in contrast
to the equation of the marginal benefit of each and every consumer to
the marginal cost for a private good, the optimal supply of a public good
is based on equating the aggregate marginal benefit (vertical sum of indi-
vidual marginal benefit curves) to the marginal cost of provision, as is
(should I just say ‘should be’) now well known to all undergraduates in
economics. A simple theory on the appropriate size of public spending
may then be stated thus. Sum the aggregate marginal benefit curves of
all relevant public goods horizontally to give the aggregate marginal
benefit curve of public spending in general. The intersection of this
curve with the marginal cost curve of public spending then determines
the optimal size of public spending.
Due to the heterogeneity of different public goods, a convenient way
of measuring public spending is by the dollar amount spent. Then, a
change in the efficiency in supplying public goods will be reflected in
different physical quantities for the same spending. This will be reflected
in a shift in the aggregate marginal benefit curve rather than in the cost
curve. However, despite the dollar definition of public spending, the
marginal cost of a unit of public spending need not be one dollar. Taking
account of the excess burden (also called deadweight loss or distor-
tionary cost) of taxation and other excess costs of raising government
revenue (administrative, compliance, and policing cost), the marginal
cost of a dollar of public spending is typically regarded as exceeding one
dollar by a considerable margin. Feldstein (1997) gave an estimate as
high as $2.65. (On the marginal cost of public funds, see also Fortin and
Lacroix, 1994 and references therein.) With a higher marginal cost of
public spending, the optimal size of public spending is smaller, given
the marginal benefit curve. However, contrary to this traditional view,
A Case for Higher Public Spending 107
Kaplow (1996) argues that the benefit–cost ratio for public project only
needs to exceed one. Chapter 9 addresses this issue. (If we ask the wider
question whether a higher public sector increases unemployment
and/or reduces the rate of economic growth, it seems that the evidence
is inconclusive; see, e.g. Dowrick, 1993; Karras, 1996; Phelps and Zoega,
1998, pp.788–9.)
8.1.2 Growth in public spending and the median-voter model

Assuming that government spending is determined by the preferences
of the median voter, the standard model is:
log G = a + b log Ym + c log Pm
where G is government spending, a, b, c are constants (with b positive

and c negative), Ym is the income of the median voter, and Pm the mar-
ginal ‘price’ of public goods to her (related to the marginal tax rate on
her). (See, e.g. Jackson, 1993, p.29 on some problems of this model; de
Bartolome, 1995 on the point that there are as many individuals who
mistakenly use the average rates as the relevant marginal rates; Turn-
bull and Chang, 1998 on an empirical test supportive of the median
voter model.)
In terms of the above model, the long-term growth in the share
of government spending may be explained by (as summarised by
Atkinson and Stiglitz, 1980, pp.326–7):
A. Increasing per capita real income, with public expenditure being

a superior good (income elasticity being larger than one).
B. A redistribution of income that raises the median relative to the
average income.
C. A decrease in the perceived tax burden of the median voter. This
may be caused by a change in the tax structure or due to a higher
fiscal illusion.
D. A decrease in the relative price of public sector output.
Extending beyond the median-voter model, we have additional factors:
E. An increase in voting participation of lower income groups.

F. An increase in the activities of interest groups lobbying for more
public spending.
G. Changing ideology of political parties and shifts in sources of
financial support for political parties.
This list does not include some important considerations (discussed

below) that may explain the increase in public expenditure. (On the
growth of government spending, see also Baumol, 1967; Levitt and
Joyce, 1987; and West, 1991 on the increase in costs relative to the
private sector; Peltzman, 1980 emphasising the role of equality, Meltzer
and Richard, 1981 emphasising inequality and the democratic process
in increasing transfer payments; Kau and Rubin, 1981 on the increase
in efficiency in collecting taxes, Ferris and West, 1996 on the empirical
test of the various theories, Rodrik, 1998 on the risk-reduction role of
public spending for open economies, Marmola, 1999 on a utility-
interdependence explanation of the constitutional choice of the set of
public goods, Mueller, 1989, Chapter 17, for a discussion of several
public-choice theories including interest groups and bureaucracy, and
Lybeck and Henrekson, 1988 on some conceptual and empirical issues.)
8.2 Inefficiency in public spending may

increase its optimal level
As pointed out at the beginning of the last section, public spending is

measured by the monetary amount, and a change in the degree of ef-
ficiency in providing public goods is reflected in a shift in the marginal
benefit curve of public spending. Does this mean that the inefficiency
in public provision necessary shifts the marginal benefit curve down-
ward and hence decreases the optimal level of public spending? The
answer is no (at least ‘not necessarily’) even ignoring the possible more
subtle interaction between benefits and costs and holding the cost side
unchanged. (For simplicity, it is also assumed that the degree of pro-
vision inefficiency is independent of the level of public spending; the
relaxation of this complicates the analysis without changing the possi-
bility that optimal public spending may increase with provision inef-
ficiency.) The reason is that the inefficiency in public provision may
increase the marginal benefit at the relevant margin, even though it
decreases the marginal benefit at the initial range.
This simple point is shown in Figure 8.1 where the MB curve
measures the marginal benefit of public spending assuming a given
degree of efficiency (including that of perfect efficiency as a special case)
in public provision of public goods. With the given MC curve, the
optimal level of public spending is at g*. Now consider a decrease (say
halving) in the degree of efficiency in the provision in the sense that
the same amount of spending results only in half of the previously
amount of public goods provided. Then, the marginal benefit of the first
marginal unit is halved. The marginal benefit curve must thus shift
downward in its initial range to become MB¢. The value of (the new) MB
(i.e. the height of the MB’ curve) at any given level of public spending
g is only half of the original value of MB at half that level of public
spending g/2; e.g. the height of A¢ is only half that of A. This is so
because, by halving provision efficiency as assumed, the g amount of
public spending provides only g/2 amount of public goods previously
provided, and a marginal increase in g also results in half of the previ-
ous increment. This results in a downward shift in the marginal benefit
curve for the range where the curve is elastic in the sense of a demand
curve (i.e. if a more than 1% increase in g is associated with a 1 per cent
reduction in the marginal benefit, which must be true starting at g = 0;
note that the MB curve being elastic in the sense of a demand curve
really means that the fall in MB is proportionately small with respect to
an increase in g). However, for the range where the marginal benefit
curve is inelastic, a decrease in provision efficiency shifts the MB curve
upward. Thus, if the MC curve intersects the MB curve at the inelastic
range as shown in Figure 8.1, inefficiency in public provision may in
fact increase the optimal level of public spending. (This is subject to the
condition that the decrease in provision efficiency is not so large as to
make non-provision optimal or to make a shift to private provision
optimal.) We thus have (see Appendix F for a formal demonstration):
MC
A¢
MB MB¢
0 g
g* g*¢
Figure 8.1
Other things being equal, an increase in inefficiency in provision

increases/decreases the optimal level of public spending if the mar-
ginal benefit of public spending (or good) decreases proportionately
more/less than the increase in public spending (or the physical
amount of public good provided).
The part in parenthesis also holds because the (proportionate) respon-

siveness of the marginal benefit to the level of public spending (in dollar
terms) is the same as the responsiveness of the marginal benefit to the
physical amount of public good, under the simplification that the
degree of inefficiency does not depend on the level of public spending,
as shown formally in Appendix F.
The result may be seen in an alternative way. If we measure the physi-
cal amount of public spending along the horizontal axis (Figure not
shown here), a (uniform) decrease in provision efficiency may be taken
as an proportionate upward shift in the MC curve. This necessarily
reduces the optimal physical amount of public spending (ignoring
some possible interaction in the MB and the MC curves). However, if
the MB curve is inelastic, the reduction is proportionately less than the
increase in cost, leading to an increase in the monetary amount of
public spending.
The reason that inefficiency in provision may increase the optimal
level of public spending may be put in the following intuitive terms.
Inefficiency decreases the (total) benefit but increases (by under-
provision physically) the need for public spending. Needless to say, an
increase in efficiency is desirable for all cases (i.e. whether the MB curve
is elastic or inelastic).
Distinct from but similar to the inefficiency in provision is the
increase in cost of providing public goods relative to private goods. As
pointed out by Baumol (1967), this increase in relative cost increases
the size of the public sector even if it remains unchanged in real terms.
As calculated by Gemmell (1993, p.7), the share (as a percentage of GDP)
of general government consumption expenditure in the UK at current
prices increased from 16.5 per cent in 1950 to 21.7 per cent in 1980
(and declining back to 19.9 per cent in 1990) but decreased at constant
1985 prices from 25.7 per cent in 1950 to 21.9 per cent in 1980 (and
further decreased to 19.1 per cent in 1990). The difference is caused by
an increase in the relative price of government from 64.2 (using that of
1985 = 100) in 1950 to 99.1 in 1980 and 104.2 in 1990. In the presence
of the increase in cost, if the demand for public spending is inelastic,
the optimal size of public spending increases.
8.3 General taxation may be more corrective

than distortive
A reason why the marginal cost of public spending is regarded as rather

high is explained by Feldstein (1997, p.209): ‘the deadweight loss caused
by a change in tax rates is not a small triangle but a much larger trap-
ezoid because we start with an existing tax distortion’. This is a valid
argument within the orthodox framework where taxes are raised just to
pay for the public spending and serves no other useful purposes. Within
this framework, it is also quantitatively a very important consideration
as the existing level of tax revenue accounts for more than 30 per cent
of GDP for many countries. However, there is one consideration that
may offset this important factor. In the production and/or consump-
tion of most if not all goods and services, significant environmental
disruption effects (including pollution, deforestation, littering, and con-
gestion) are created either directly or indirectly (through intermediate
goods). Few, if any, corrective taxes have been imposed on these activ-
ities. Thus, the usual income and consumption taxes, while designed
for revenue-raising purposes, may serve as a rough correction to the
environmental disruption effects of production and consumption.
While the rate structures are far less than ideal in an abstract sense, this
is a much smaller problem when the issue of feasibility and/or admin-
istrative costs are taken into account, though higher tax-rates on more
disruptive activities should still be imposed. (See Ballard and Medema,
1993, on the different figures for the marginal cost of public fund raised
by output, consumption, income, and Pigovian taxes.) Given the sever-
ity of the environmental problems (recalling the recent report of the big
ice crack in the Antarctica due to global warming) and the very long-
term nature of the effects, an average corrective tax-rate of around 30
per cent may not be excessive, though more reliable estimates should
be done. At least, this consideration significantly offsets the trapezoid
effect of higher tax rates, if not making the distortionary costs of tax-
ation negative!
8.4 Diamond effects and burden-free taxes
It is widely known that corrective taxes on external costs such as pol-

lution may yield negative excess burden, as argued in the last section,
though the point is seldom brought up with respect to the optimal size
of public spending. It is almost completely unknown, even among econ-
omists, that there exists a class of goods, taxes on which impose not
only no excess burden but no burden at all. Ignoring administrative

costs, the extent of negative excess burden is 100 per cent.
Given the quality, the utility (i.e. the extent of usefulness) of a good
to a consumer depends on the quantity consumed. This is true for most
goods that yield intrinsic consumption effects, e.g. nutrition, warmth.
However, for many goods, it is the value (price times quantity) of the
good rather than the quantity that is important. A notable example is
diamond. Though beautiful, the utility of diamond to the (household)
consumer is mainly due to its value. Cubic zirconia looks exactly like
top-quality diamond but costs only a tiny fraction. Someone mixed a
few stones of cubic zirconia with top-quality diamonds and gave them
to diamond experts for verification. The expert declared all stones imi-
tations. The willingness of people to pay high prices for diamonds
cannot be explained by their intrinsic consumption utility, but by their
values. For convenience, I call goods valued purely for their values ‘pure
diamond goods’.
People value goods for their values for a number of reasons. First, they
can be used as items of conspicuous consumption, demonstrating one’s
wealth. Secondly, they are given as gifts of value. Thirdly, they are kept
as a convenient store of value. Refugees cannot carry their shops and
factories but could carry diamonds and gold when leaving in an emerg-
ency. One might add the speculative motive but these days speculation
is largely done in paper (or rather electronic) transactions.
When someone gives a diamond ring or some other gift of value, it
is the value of the ring rather than the size of the stone that is really
important. The same is true for items of conspicuous consumption
and for store of value. If gold is doubled in value, one may just half
the physical quantity to achieve the same purpose. In fact, it makes
carrying a large value of gold less cumbersome. Inconvenience in
dealing with tiny quantities for smaller transactions can be avoided by
using less valuable stones and metals like silver. Thus, ignoring transi-
tional and distributional issues, if pure diamond goods are heavily
taxed causing much higher prices, consumers suffer no loss in utility.
Thus, efficiency requires arbitrarily high taxes on pure diamond
goods, as demonstrated in Ng (1987a). (See also Ng, 1993 on mixed
diamond goods with upward compensated demand curves and welfare-
improving taxes.)
However, not many goods are pure diamond goods. Even diamond
itself has some intrinsic uses. Nevertheless, many goods have significant
components of the diamond effect. Thus, efficiency requires the impo-
sition of higher taxes on these goods. This seems to be ignored in the
present climate with emphasis on broad-based uniform taxation.
It should be emphasised that our case for higher taxes on goods with
significant diamond effects is purely efficiency based, not equality
based. In fact, as argued in Chapter 6, the objective of equality is more
efficiently achieved through the general tax/transfer system rather than
through specific measures.
8.5 Relative-income effects bias against public expenditure
Over a certain biologically minimum standard of living, the welfare of

people depends more on non-material factors such as personal re-
lationships and outlook in life. As far as personal or household incomes
are concerned, it is more the relative level than the absolute level that
is important. Since the early 1960s, per capita real income in OECD
countries has more than doubled. However, the average person now is
not really much happier than those in the 1960s. If they are, it is
unlikely due to the much higher per capita income. In fact, the average
Australian probably feels economically worse off now, mainly because
of a drop in their relative position internationally, despite their bigger
house, technically more sophisticated car, etc. than in the 1960s. For its
own interest and as an extreme manifestation, it may be noted that even
mortality may be more a function of relative than absolute income
(Wilkinson, 1997).
The importance of relative-income effects and the relative unim-
portance of absolute incomes (beyond a certain level) are of course well
known and have been discussed by economists and psychologists,
including Rae (1834), Veblen (1929), Duesenberry (1949), Easterlin
(1974), Hirsch (1976), Kapteyn and Wansbeek (1985), Frank (1985,
1999), and Cole et al. (1992). However, the implications of this on rel-
evant economic policy issues, especially on the optimal size of public
spending, have been largely ignored by economists. (There are some
specific policy implications made, e.g. Ireland, 1994 notes the possi-
bility of taxing status-seeking signals.)
For a single person, an increase in income increases both their
absolute income and relative income. It is thus perceived to be of sub-
stantial importance. However, for the whole society, an increase in
the relative income of some implies the reduction in others. If income
levels of all increase proportionately, relative-income levels remain
unchanged. Hence, from a social viewpoint, only the absolute compo-
nent is relevant. This component is not of very high welfare significance
for countries of high per capita incomes. Thus, in the presence of
relative-income effects, the usual method of estimating the optimal
level of public expenditure (equating the sum of individual marginal

valuations to the marginal cost) is likely to lead to a sub-optimal level
(Ng, 1987b). In most estimates, the marginal benefit of private expen-
diture is likely to be taken to include the intrinsic consumption effects
and the internal or direct relative-income effect (as these two taken
together constitute the worth of a private good as it appears to each
individual), but not to include the external or indirect (through re-
duction in the relative incomes of others) relative-income effects. This
creates an over-emphasis in favour of private expenditure, leading to a
sub-optimal level of public expenditure. This is shown more formally
in Appendix C where it is also shown that, under reasonable assump-
tions, the optimal size of public spending relative to GDP increases with
per capital income.
8.5.1 Policy implications ignored by economists

While economists recognise the importance of relative-income effects,
the relevant policy implications are ignored even when undertaking
cost–benefit analyses relevant to actual policy decisions in the real world.
As a specific example, we may consider the otherwise competent
cost–benefit analysis of Portney (1990) who concludes that ‘Congress
and the President are about to shake hands on a landmark piece of en-
vironmental law [The Clean Air Act] for which costs may exceed bene-
fits by a considerable margin.’ However, the analysis of costs and benefits
is based on ignoring the existence of relative-income effects, the full con-
sideration of which may change the cost–benefit ratio substantially.
8.5.2 Rat-race for material growth may reduce welfare

In fact, it may be reasonably argued that, unless environmental protec-
tion and other welfare-improving measures are increased, economic
growth may reduce welfare. Taking absolute income, relative income,
and environmental quality as important for individual welfare, an indi-
vidual places a higher value on the contribution of an increase in their
income to their welfare than the value by which a general increase in
incomes for the whole society could really increase social welfare. This
is so because for an individual, an increase in income contributes to
their welfare both through the absolute and the relative-income com-
ponents. For the whole society, only the absolute-income component
is relevant. Moreover, if this absolute component is relatively unim-
portant, and if economic growth is not accompanied by appropriate
environmental protection measures such that environmental quality
deteriorates significantly, the growth may be welfare reducing from the
world perspective, though each individual and even each nation may
find growth important, as argued in more details in Ng and Wang
(1993). Appendix C goes further to show that welfare-reducing growth
may prevail even if the government chooses the optimal level of public
spending and the optimal amount of spending on environment-
disruption abatement.
8.6 The unimportance of absolute income
While absolute income is not as important as indicated by the rat race

for making money, it seems to be still of considerable importance in
terms of utility or preference. In terms of happiness or welfare, the
importance of absolute income is even more doubtful.
It is true that happiness is more difficult to measure and compare
objectively than some more objectively measurable magnitudes
like GDP. However, conceptually, it is cardinally measurable (Chapter 2
above). A practical method has also been developed and used to measure
happiness cardinally and interpersonally comparably (Ng, 1996a).
Though most existing measures of happiness have more problems
with their comparability, they are not completely useless. Different
researchers arrive at largely consistent results which also correspond
with alternative measures like the opinions of friends. (See Veenhoven,
1984, 1993.) If one wants to be pedantic in insisting on perfect accu-
racy, even the measurement GDP is open to query on its accuracy and
comparability.
Studies by psychologists and sociologists show that, both within
a country and across nations, the happiness level of people increases
with the income level, though not very significantly. For example,
using regional and cultural classifications, the Northern European
countries with high income also score top on happiness, followed by
the group of English-speaking US, UK, Australia, and Ireland. Central
and South-American countries including Brazil come next, followed by
the Middle East, the Central European, Southern and Eastern European
(Greece, Russia, Turkey, and Yugoslavia), the Indian Sub-continent,
and Africa which does not, however, come last. Southern and Western
European (France, Italy, and Spain) score significantly lower than Africa.
And the last group is East Asia, including the country that leads
in income, Japan. Singapore has an income (per capita) level 82.4
times that of India. Even in terms of purchasing power parity instead
of using exchange rate, Singapore still has 16.4 times higher income
than India. However, the happiness scores of both countries are exactly
the same, both significantly higher than that of Japan. (See Cummins,
1998.)
While there are notable cases like Japan and France that are far off
the regression line, statistically significant positive relationship between
happiness and income exists cross-nationally globally. However, this is
due mainly to the inter-group difference between the high-income and
high-happiness advanced and free countries and the others. The analy-
sis by Schyns (1998) shows that there is no positive relationship
between income and happiness within either of these two groups.
Kenny (1999, p.12) reports a similar lack of relationship within each
group by splitting all countries into two groups of below and above
$8,000 in per-capita income.
The relationship between happiness and income level inter-
temporally within the same country (at least for the advanced countries
which have such data) is even less encouraging in terms of giving a posi-
tive relationship. For example, from the 1940s to 1994, the real income
per capita of the US nearly trebled. However, the percentage of people
who regard themselves as very happy fluctuated around 30 per cent,
without showing an upward trend; another measure of average happi-
ness fluctuated around 72 per cent. Over the period 1958–88, the per
capita real income level in Japan increased by more than five times
(Summers and Heston, 1991). However, its average happiness measure
fluctuated around 59 per cent, also without an upward trend. (See Veen-
hoven, 1993; Myers, 1996, p.445; Diener and Suh, 1997; Oswald, 1997;
Frank, 1999.) In fact, ‘if there is any causal relationship in rich
countries, it appears to run from happiness to growth, not vice-versa’
(Kenny, 1999, p.19). Also, recent research suggests that individuals who
strongly value extrinsic goals (e.g. fame, wealth, image) relative to
intrinsic goals (e.g. personal development, relatedness, community)
have less happiness (Ryan et al., forthcoming).
Kenny (1999, pp.4–5) also puts the point of fast diminishing marginal
utility of income in more objective terms thus:
Compare Mozambique, China and the USA. In turn, the countries’

GNPs per capita in 1992 were $80, $470 and $24,740. Infant mor-
talities were 145.6, 30.5 and 8.6 per 1,000 live births, respectively.
Life expectancies were 47, 69 and 76 years. Thus, going 1.6 percent
of the distance between Mozambique and the United States in terms
of wealth, so reaching China’s income, we move 84 percent of the
distance in terms of infant mortality and 76 percent of the distance
in terms of life expectancy.
On the other hand, there are factors that affect or at least correlate
with happiness much more significantly than income, including
being married or single (Myers, 1996, p.510), being employed or not
(Winkelmann and Winkelmann, 1998), and having a religious belief
and church attendance.
For those who do not trust the more subjective measure of happiness
and opt to use more objective indicators of the quality of life, the picture
is not much different. Analysing a panel dataset of 95 quality-of-life
indicators (covering education, health, transport, inequality, pollution,
democracy, political stability) covering 1960–90, Easterly (1997) reaches
some remarkable results:
1. While virtually all of these indicators show quality of life across

nations to be positively associated with per capita income, when
country effects are removed using either fixed effects or an esti-
mator in first differences, the effects of economic growth on the
quality of life are uneven and often nonexistent. ‘For the fixed effects
estimator applied to 95 indicators, the coefficient of income was signifi-
cant at the 5% level for 40 indicators. This is not so bad, except that only
23 of these 40 show improvement in the quality of life associated with
rising income. It is distressing that almost as many indicators show sig-
nificant deterioration in quality of life’ (Easterly, 1997, p.18).
2. Most of the exogenous time shifts (69 out of 95 indicators) improve
the quality of life and time shifts are more important than growth
effects in the majority (62% of the 79 available indicators) of indi-
cators. Even for the only 22 out of the 95 indicators with a signifi-
cantly positive relationship with income under fixed effects, time
improved 10 out of these 22 more than income did.
The surprising results are not due to the worsening income distri-
bution (there is some evidence that the share of the poor gets better
with growth). Rather, the quality of life of any country depends less on
its own economic growth or income level but more on the scientific,
technological, and other breakthroughs at the world level. These
depend more on public spending than private consumption.
If higher income does not really increase happiness, why do people
put so much emphasis in making money? Apart from the importance
of relative-income effects and the public bad aspect of environmental
disruption, individual ignorance and/or irrationality may also have a
role to play. Apart from the discussion in Chapter 4 (Section 4.2 in par-
ticular), some further remarks may be added here.
First, there is the animal spirit for accumulation mentioned by

Keynes. This is similar to the accumulation instinct of some animals,
like the storage of food by rats, the burial of nuts by squirrels. Such
instincts are selected as they generally help the species to survive. We
are a much more ‘rational’ species. (A species is defined to be more
‘rational’ than another if its behaviour is more affected, than the other
species, by the calculation of reward and punishment and less by hard-
wired instincts. See Ng, 1996b where it is argued that the evolution of
more ‘rational’ species makes the environment more complex which in
turn makes more ‘rational’ species more selectable, leading to a virtu-
ous circle culminating in our species of homo sapiens, partly explaining
the fast rate of evolution, the speed of which is doubted by creation-
ists.) However, we are still partly affected by our animal spirit of accumu-
lation instinct. In times of scarcity, this helps us to survive better.
However, in our present era of material plenty (but environmental-
quality scarcity), our animal spirit may make us irrational. Is it rational
to sacrifice friendship, love, and family, or to take the high risk of losing
freedom, fame, and jobs by engaging in corruption, robberies, murders,
etc. just to make some more money which is not that important to one’s
happiness, especially in the long run? Those who can see through this
may call themselves ‘not deluded’. (Confucius said, ‘I became indepen-
dent at thirty, not deluded at forty.’)
Secondly, there are studies by psychologists and sociologists suggest-
ing that people are typically ignorant and/or irrational. For example,
most people believe that winning a big sum of money will make them
very happy. Thus, many people spend a lot of money and time buying
lottery tickets. However, there is evidence that lottery winners are no
happier than non-winner controls (Brickman, et al., 1978). For another
example, most or at least many people also believe that losing one’s legs
or eyesight in an accident is worse than being killed. However, studies
show that quadriplegics are only slightly less happy than healthy people
(Brickman, et. al., 1978). After an initial period of adjustment, maimed
victims can still enjoy life and are glad that they were not killed. Also,
people typically underestimate the negative/positive effects of current
enjoyment/suffering on future ability for enjoyment. (See Headley and
Wearing, 1991.) Eating sufficiently salty, sweet, or tasty food now may
yield more utility now than the slight health hazard involved. Shifting
to healthier food may incur too big a loss in present pleasure. However,
our taste will adjust to the blander and coarser but healthier food.
I changed from white bread to wholemeal bread a long time ago on
health reasons. Initially, I was not really sure that the gain in being
healthier justified the loss in taste. However, after months of eating

wholemeal bread, I began to enjoy that more than white bread even
just on taste. Those still on white bread are strongly urged to shift. Also,
making children accustomed to white bread may be very unwise.
An ancient Chinese poet had a line: ‘Having seen the big blue sea, it
becomes difficult to appreciate waters [i.e. lakes become less impress-
ive]’. Thus, the long-term utility of seeing the blue sea may not be that
high as it lowers the utility of seeing other less impressive waters in the
future. However, most people fail to take adequate account of such
effects, including myself. When I was on my first leave, my family and
I visited the top attractions in the world like the Great Wall, Niagara
Falls, and the Grand Canyon. After that, sightseeing is no longer excit-
ing. ‘Except for that in the Wu Mountains, no cloud is attractive’, so
says the second line of the poem. Thus, another Chinese sage advised
that one should eat sugar-cane from the less tasty end, proceeding to
the more tasty parts gradually.
8.7 Welfare-improving public expenditures
While noting that the rat race for material growth may be welfare reduc-
ing, I am not anti-growth. Instead of hoping for a slower growth or even
a depression, as some environmentalists do, a healthy growth with an
appropriate increase in environmental protection measures seems a
much superior option. Moreover, if economic growth is conducive to
a higher degree of civilisation by providing more resources to support
scientific and technological advances conducive to welfare, it may be
welfare-improving despite some negative environmental effects. For
example, the advance in science and its applications to medicine, engin-
eering, etc. have contributed much to a comfortable and enriched life,
the relief of pain, and the cure of many illnesses. The development of
the Westminster and other forms of democratic government and the
rule of law, the prevalence of modern communications and social inter-
action have also helped create a freer and more peaceful world. The
understanding of the working of the market mechanism has contributed
to the collapse of the communist totalitarian system and to the end of
the Cold War. Higher education is becoming more widespread. Looking
into the future, one can be reasonably confident that further advances
in these and related areas will be forthcoming. Moreover, there are likely
to be significant or even dramatic improvements in welfare from sources
most people do not dream of now. For example, the techniques of elec-
trical brain stimulation (discussed in Appendix A) and genetic engin-
eering could be used, after more intensive experimentation and with

careful management, to increase welfare by quantum leaps.
Despite the above-mentioned promises, it remains true that a pure
increase in GNP (even without any deterioration in income distri-
bution) may be welfare-reducing unless environmental protection and
other welfare-improving measures are facilitated. In rich countries, what
is important is not simply growth in GNP and the resulting higher con-
sumption per head but how welfare-improving measures (environmen-
tal protection, scientific advances, etc.) can be increased. Thus, despite
the excess costs (excess burden, administrative, compliance, and poli-
cing costs) of raising public revenue, more public expenditures in the
right areas (including research) may yet be most welfare improving. This
is especially so in view of the importance of environmental disruption
effects, relative-income effects and diamond effects (with the associated
burden-free or at least less burdensome taxes) which may be expected
to be increasing in importance, absolutely and relatively.
It is shown more formally in Appendix C that, with reasonable as-

sumptions, the optimal relative (to the GDP) size of the public sector
increases with higher per capita income. Thus, one may say that the
historical increase in the relative size of the public sector is consistent
with and is partly explained by this. In fact, we may go further than
this and argue that, despite the higher relative size of the public sector,
it is still lower than the optimal level. This may be so for a number of
reasons.
Environmental quality is to a large extent a global public good. The
big mountain fire in Indonesia recently (1997) affected the air quality,
climate, and even visibility of neighbouring countries within days/
weeks and is fairly certain to affect the air quality (though to a lesser
extent) of the whole world for a very long time to come. Decisions by
separate national governments based on national interests are unlikely
to take full account of the global effects unless there is adequate inter-
national coordination. The latter is only starting to emerge and is far
from being adequate.
Environmental quality is a very long-term problem. Many pollutants
remain in the environment for a long time. The warm-house effect
could threaten the coastlines of low-lying cities perhaps decades or cen-
turies from now through the melting of polar ice. In contrast, poli-
ticians in both democratic and authoritarian countries are likely to have

much shorter time horizons.
In assessing the relative benefits of public versus private expenditures,
people are likely to overestimate the true importance of the latter. (See
Section 8.5 above.) Affected by the working of relative-income effects
including the desire to keep up with the Joneses, people find private
spending very important. If most of your child’s classmates are given
expensive birthday gifts you will then feel pressurised into giving your
child a substantial gift. However, most people ignore the fact that, if the
whole society devotes more resources to supply public goods, on average
everyone will have less to spend on private items. Then, despite a lower
post-tax income, one can still keep up with others.
Research, especially fundamental research, is also largely a global
public good and hence is under-funded by national governments.
However, once a basic level of consumption has been attained, happi-
ness cannot increase significantly with higher private consumption
(especially at the social level due to relative-income effects as already
discussed) given the level of knowledge. Many of us can easily double
our private consumption. (For those who cannot, many can halve their
consumption without much real loss in happiness.) But what is the
point? If you are already well-fed, well-clothed, well-accommodated,
increasing private spending does not increase your happiness signifi-
cantly but imposes significant negative externalities in the form of en-
vironmental disruption and relative-income effects. Is it really very
important to spend many weeks holidaying in far-away places living in
5-star hotels, when you can walk or jog in nearby parks and read fasci-
nating books or watch interesting programmes on TV? Is having a beer
in a local pub or a cup of tea with a few friends at home significantly
less enjoyable than dining in expensive restaurants or drinking wine at
US$500 a bottle? Rather than spending extravagantly, our welfare can
certainly increase much more if the cities become safer, the environ-
ment becomes cleaner, a cure is found for some illness, or the method
of stimulation of the brain can be improved for common use. (On the
quantum leap in welfare with the last method, see Appendix A.) Most
of such items need public spending.
Due to the inefficiency of public organisations, privatisation in certain
areas and de-regulation of counter-productive regulations may make a
lot of sense. It is also easy to find particular areas of waste and ineffec-
tiveness in public expenditures. (This is so since public expenditure is
unlikely to be optimally distributed among different items and also
unlikely to be provided in the most efficient way.) However, due to the

above considerations, the recent global trend towards the checking of
public expenditures, especially in the funding of research, may be very
negative in terms of social welfare, especially globally and in the long
term.
This negative effect may consist to a large extent in the failure to
attract talented people to work in the welfare-improving areas of
research and the public sector. Top students used to stay to do PhDs and
remain in the academic world. Now, many good students leave to work
for business firms where real talents are largely wasted in competitive
rivalry at both the production and consumption levels. True, talents are
also needed to have efficiency in production and innovation. However,
the very best talents are needed to do research and to serve in the public
sector, especially after the satisfaction of basic needs when increases in
private consumption without an increase in knowledge are not very
welfare-conducive. Many people are sceptical of the productivity of
increased funding for research, being aware of many ill-conceived pro-
jects and many publications of doubtful usefulness. However, one piece
of useful research out of a hundred may make it all worthwhile. More-
over, long-term social welfare may be improved by raising the rewards
and working conditions of researchers and other public-sector em-
ployees so as to attract real talents back from the business sector. The
market fails here partly because of the public goods externality and
partly because of factors like relative-income effects (also a form of
externality) discussed above.
Combining the analysis of this chapter (and that of Appendix C) with
the evidence reported in section 8.6 above (especially the remarkable
results of Easterly, 1997), a strong case may be made for international
cooperation to dramatically increase public spending on global public
goods including environmental protection and research. A few exam-
ples may be given to indicate that much more research is needed to
increase welfare.
• The very topic of the appropriate size of the public sector, regarded
by Feldstein (1997) as the central public finance question, is very
under-researched. For example, few, if any, researchers relate the
important issue of relative income to this central question. While I
have discussed this and other related issues above, a lot more quan-
titative studies are needed.
• While studies on the effects of specific drugs and ingredients have
been done, it seems that a general study tracing the different types
of food, drugs, and activities taken by a big enough sample of people

(at least thousands) of different ages and health conditions (not just
those hospitalised) over a long period (at least decades) to discover
the desirable and undesirable effects may be most rewarding. Though
the study would be very costly, we may gain very useful knowledge
on millions of factors simultaneously.
• The stimulation of certain pleasure centres in the brain can induce
intense pleasure without the effect of diminishing marginal utility.
This has been known since the mid-1950s. Why has the method not
been perfected for common use? (See Appendix A for details.)
Much useful research and development are undertaken in the private

sector. However, by and large, it is effective only at the level with
immediate or close to immediate application. Nowadays, most funda-
mental research is undertaken in the public sector:
It was a government research program that produced the math-

ematical algorithms that led to the computer. The Internet was
created by researchers in US universities under contract to the
Defence Department. The laser, the electric telegraph – research has
always been a public responsibility. And, as we move to a knowledge
economy, its role has become central to the economy
Joe Stiglitz, reported in the The Age, (Melbourne), 27 Mar. 1999
Despite the negative evidence discussed in Section 8.6, I believe that

happiness has increased. (It is likely that people’s standards for describ-
ing themselves as ‘happy’ or ‘very happy’ have increased, hiding an
increase in real happiness. This increases the importance of using the
proposed more interpersonally comparable measure of happiness (Ng,
1996a).) However, I remain convinced that happiness could be increased
by shifting resources from the largely competitive private consumption
to items of public spending that benefit the whole world for a long time.
9
The Appropriate Benefit–Cost Ratio
for Public Spending
The optimal level of public spending is related to what public

spending/projects are regarded as appropriate. This is in turn related to
the benefit–cost ratio used to assess public projects. The use of a
benefit–cost ratio more stringent (i.e. larger) than one would generally
lead to a smaller level of public spending. Recently, some prominent
economists have expressed apparently rather contrasting views on this
issue. This chapter reconciles and qualifies their arguments.
9.1 The conventional view
The conventional view of economists is that the optimal level of public

spending should be less than that indicated by the use of the simple
benefit–cost ratio of one, since the financing of public revenue involves
distortionary costs. This view dates from at least the time of Pigou
(1928), who stated that the benefits of public goods must exceed their
direct costs by an amount sufficient to outweigh the distortionary cost
(also called deadweight loss or excess burden) of taxation. An authori-
tative modern textbook (Stiglitz, 1988, p.140) puts the Pigovian princi-
ple this way: ‘Since it becomes more costly to obtain public goods when
taxation imposes distortions, normally this will imply that the efficient
level of public goods is smaller than it would have been with nondis-
tortionary taxation.’1
It is known that the Pigovian principle is subject to qualifications
due to the presence of considerations like second best, including com-
plementarity/substitutability between public and private goods. (See,
e.g. Atkinson and Stiglitz, 1980; King, 1986; Batina, 1990 and Chang,
1997.) Wilson (1991) goes as far as establishing a case where the second-
best public-goods level exceeds the first-best level. However, he assumes
124
The Appropriate Benefit–Cost Ratio 125
the feasibility of financing public spending through a general lump-sum

tax (or a reduction in the lump-sum subsidy) without which, Wilson
(1991, p.153) himself recognises, the abnormal case ‘normally will not
happen’.
Christiansen (1981) identifies a set of sufficient conditions for the
applicability of the simple benefit–cost ratio of one, namely similar
individual preferences except for the influence of different earning a
bilities (a standard assumption in the Mirrlees, 1971 framework),
weak separability of the public good and the numeraire private good
from leisure/work, the absence of commodity taxation (in whose
presence, an additional condition on the independence of the marginal
valuation of the public good is needed), and the operation of an optimal
(in general) non-linear income tax. Using the self-selection approach
(a high-ability person may choose to work less and pay less tax),
Boadway and Keen (1993) derive a modified (Samuelsonian) benefit–
cost rule for public goods, explaining and confirming Christiansen’s
important result for the specific case. Using his approach to analyse
commodity tax reform, Konishi (1995) generalises Christiansen’s result
to any smooth non-linear income tax, optimal or not. However, Chris-
tiansen’s case for the simple benefit–cost rule is meant as applicable to
a special case.
Feldstein (1997) presents the orthodox position on the distortionary
costs of taxation most comprehensively and clearly, reviewing some
very important contributions by himself and other researchers (includ-
ing Stuart, 1984; Ballard et al., 1985; Browning, 1987; Auten and Carroll,
1994; Feldstein, 1995a, 1995b; Feldstein and Feenberg, 1996 and
Slemrod and Yitzhaki, 1996). At the risk of over-simplification, Feld-
stein’s position may be summarised by the following quotation:
‘First, higher tax rates may reduce the supply of labor and, in the
longer run, the supply of capital. . . . Second, higher tax rates change
the forms in which individuals take their compensation. A higher
marginal tax rate on labor income induces a substitution of untaxed
fringe benefits and more pleasant working conditions for taxable
cash income . . . [Third,] higher marginal tax rates reduce taxable
income by inducing more spending on things that are tax deductible
(including . . . charitable gifts, and health care) . . . many economists
believe that an increase in tax rates would cause only a small dead-
weight loss . . . as a “small” triangle. . . . That line of reasoning is
wrong for four reasons. First, the deadweight loss caused by a change
in tax rates is not a small triangle but a much larger trapezoid because
we start with an existing tax distortion. . . . Second, the relevant labor

supply elasticity is much larger than the traditional estimates imply.
. . . Third, . . . other ways of reducing taxable income . . . [Fourth, the]
same kind of wasteful distortion is also true for spending on things
that are tax deductible. . . . The total cost per incremental dollar of
government spending, including the revenue and the deadweight
loss, is thus a very high $2.65. Equivalently, it implies that the mar-
ginal distortionary costs per dollar of revenue are $1.65.’
Feldstein, 1997, pp.201, 209–11
(Other estimates, e.g. Ballard et al., 1985, usually give a much

lower figure of around 30 per cent, but Browning, 1987 gives a bigger
range. See also Snow and Warren, 1996 for a more general analytical
analysis.)
9.2 Kaplow’s argument and the principle of

‘a dollar is a dollar’
While some recent literature has qualified the Pigou principle, a full-
scale onslaught is presented by Kaplow. He argues that public goods can
be financed without additional distortion by using an adjustment to the
income tax that offsets the benefits of the public good. The:
preexisting income tax schedule is adjusted so that, at each income

level, the tax change just offsets the benefits from the public good.
By construction, an individual’s net reward from any level of work
effort will be unaltered; any reduction in disposable income due to
the tax adjustment is balanced by the benefits from the public good.
Because an individual’s after-tax utility as a function of his work
effort will thus be unchanged, his choice of work effort – and utility
level – will also be unaffected.
Kaplow, 1996, p.514
For example, if the benefit of a public good is proportional to the

income level of the taxpayers, it may be financed by a (or an increase
in) proportional income tax. The proportional income tax itself may
involve a disincentive effect. However, the tax plus the public good
together involve no disincentive effect. For example, suppose the
benefit of police protection of properties is proportional to the income
level of the taxpayers. With a higher degree of police protection
financed by a higher proportional income tax, a person benefits more
(in comparison to a lower degree of protection and lower tax) and pays
more by the same amount if she earns more, leaving the incentive struc-
ture unaffected.
If the benefit of a public good is a constant amount across all income
levels but it is financed by a proportional income tax, then a disincen-
tive effect may be involved. But then ‘it will be accompanied by a
countervailing change in redistributive benefits that has been ignored’
(Kaplow, 1996, p.525). To this observation, it may be added that, in the
reversed case, e.g. where a public good with benefit proportional to
income is financed by a lump-sum tax, a negative redistributive benefit
will be involved, but it will be accompanied by a negative disincentive
effect, reasonably assuming that a tax increasing in income is already
in existence for redistributive purposes. If the income-tax schedule has
been designed to approximately achieve an optimal balance between
the redistributive benefits and the distortive costs of taxation, the redis-
tributive benefits (positive or negative) of non-benefit-proportional
financing of a (non-huge) public good will be approximately offset by
its distortionary costs (positive or negative). For a huge public good or
the cumulative effects of a large number of public goods, either benefit-
proportional financing may be used or a change to the income-tax
schedule may be made to redress any significant divergence between
redistributive benefits and the distortionary costs of taxation.
For the case where the benefits of the public goods increase with the
income level, it may be interpreted as a case where there is significant
complementarity between the public good and the taxed goods. In this
case, some recent analysis allows for certain offsetting effect on the
extent of distortionary costs. (Intuitively, an increase in public goods
then increases the demand for the taxed goods, generating more tax
revenue and offsetting the under-consumption of these taxed goods.)
However, while this complementarity is unambiguous in reducing the
distortionary costs, it is ambiguous in affecting the optimal size of
public good provision due to the following counteracting effect. An
increase in public good provision necessitates higher taxes on the taxed
goods, leading to higher prices of these goods which reduces the
marginal valuation on and hence the optimal size of the public good
through complementarity. Substitutability has the reversed effects. (See
Batina, 1990 and Chang, 1997.) Since the Kaplow principle is concerned
with the size of the required benefit–cost ratio for public goods rather
than on the level of public spending directly, it is not affected by the
counteracting effect through complementarity on the marginal valu-
ation on public goods. Moreover, most of the recent contributions in
optimal taxation/expenditure, using a single representative consumer

model, ignore the countervailing change in redistributive benefits for
cases where the distortionary costs are not offset by complementarity
to become zero.2
The argument of Kaplow may be compared with my argument
(Chapter 6) for treating a dollar as a dollar whomsoever it goes to or
comes from (in particular, irrespective of income groups) in the assess-
ment of any change, policy, project, etc. (except in the general income
tax/transfer system). If a project benefits the rich by $10 m and costs the
poor $8 m, the pure efficiency principle of a dollar as a dollar dictates
its adoption but most people will not regard it as desirable. However, it
is Pareto-superior to adopt the project and adjust the tax schedule such
that the rich have to pay $9 m more and the poor has to pay/receive
$9 m less/more. Most economists do not believe that this is Pareto
superior because making the tax schedule more progressive increases the
disincentive effects of taxation. This belief ignores the fact that the
policy of rejecting such a project also has higher disincentive effects.
Despite the similarity of my ‘a dollar is a dollar’ argument with
Kaplow’s argument (both involving offsetting income tax/transfer
adjustment), I must admit that I failed to see the Kaplow principle (that
the benefit–cost ratio for public goods only need to exceed one) for two
decades after formulating my ‘a dollar is a dollar’ principle (seminar pre-
sentations at Monash in 1976, at VPI, New York, and Yale in 1978; the
principle appeared to be so right-wing that the paper could not be pub-
lished until 1984) until I read Kaplow. In a sense, one may say that
the Kaplow principle is a specific application of the ‘a dollar is a dollar’
principle to the case of public goods/projects. Since a dollar should be
treated as a dollar whomsoever it goes to or comes from, the costs and
benefits of any project (private or public) should be counted equally and
hence the benefit–cost ratio needs only exceed one for any project
(private or public). However, before reading Kaplow, I thought that
requiring the benefit–cost ratio for a public good to exceed one by a suf-
ficient margin is based on the efficiency principle as a dollar of public
revenue costs the economy more than one dollar. In other words, I
accepted the conventional position here completely. Thus, I find
Kaplow’s paper extremely important and hope that it will be given the
attention it deserves.
One obvious qualification to the Kaplow principle and the principle
of ‘a dollar is a dollar’ is: What if the higher public spending cannot be
financed by offsetting benefit taxation and if the existing progressivity
in income tax/transfer system is not optimal or cannot be adjusted to
offset the shift to the efficiency principle, due to political constraints or

other factors? While we may try to do good by stealth in the short run
and proceed to use, say, distributional weights, in the long run this will
be known and cause disincentive effects. Moreover, the same political
forces that prevent the increase in progressivity may then work to
decrease the degree of progressivity as actually happened in many
countries in the past few decades, when incentives have been reduced
by the use of distributional weights and other non-efficiency, purely
equality-oriented policies. We then end up with less efficiency and no
more equality. (See Section 6.4.)
Kaplow also mentions two qualifications to his simple analysis which
also applies to my ‘a dollar is a dollar’ principle. However, neither quali-
fication affects the central thrust of either the Kaplow principle or the
principle of ‘a dollar is a dollar’. First, when second-best considerations
such as different degrees of complementarity with leisure are taken into
account, the simple cost–benefit rule or the ‘a dollar is a dollar’ prin-
ciple should be adjusted or be defined to include such indirect
costs/benefits through second best or external effects. It might then, for
example, be argued:
that there should be smaller public libraries than otherwise would

be efficient, because libraries make leisure more attractive, reinforc-
ing the adverse incentive effect of the income tax. Conversely,
there perhaps should be greater police protection of private property
than otherwise would be efficient, because this increases the value
of goods that are purchased from the fruits of labor. . . . This qualifi-
cation . . . does not justify the type of adjustment to cost–benefit
analysis.
Kaplow, 1996, p.518
This would require a higher benefit–cost ratio for all public goods or a
distributional weighting system favouring the poor and against the rich
for all specific cases. (For the case of libraries, I believe that there are
counteracting benefits, including indirect external economies and merit
goods, justifying higher levels of provision. But this is a separate issue.
On indirect externalities, see Ng, 1975b.)
The second qualification refers to heterogeneity of preferences for
people on the same income. It is not feasible to design income tax
schedules and public spending programmes that differ between people
on the same income but of different preferences. Thus, the construction
of compensating tax/transfer changes in either Kaplow’s or my argu-
ment can only make people on each income as a group no worse off.
People who differ in preference significantly from the average (at that
income level) one way or the other may either gain or lose. However,
the gainers within each income level could compensate the losers fully
(with some overall gains left over). This is called a quasi-Pareto social
improvement in Ng (1984b). Objections to the compensation tests are
mainly based on the distributional consideration that a gain valued at
$2 million by the rich need not be more important than a loss valued
at $1 million by the poor if the compensation is not actually paid. For
a quasi-Pareto social improvement, full compensation is possible within
each group of the same income level. The rich-poor issue does not apply.
(The minor problem of possible inconsistency in the application of
compensation tests is also discussed in Ng, 1984b.) Thus, the problem
of heterogeneity of preferences does not cause a big problem. A more
important qualification to the Kaplow principle (but not to the ‘a dollar
is a dollar’ principle) is discussed below.
9.3 Reconciling Kaplow and Feldstein
The positions of Kaplow and Feldstein (and that of most other econ-
omists, including myself before I read Kaplow) appear to be diametri-
cally opposite, with Kaplow arguing for the adequacy of the simple
benefit–cost principle and Feldstein emphasising the need to account
for the big deadweight loss of raising revenue for public spending.
However, I wish to argue in this section that we may adopt an inter-
pretation that makes both sides basically correct, subject to some impor-
tant qualifications.
Subject to an important comment mentioned below, Feldstein is basi-
cally correct that, given the various ways individuals adjust to the higher
tax rates, raising an additional dollar of public revenue (by, say, a mar-
ginal proportionate increase in tax rates across the board) on its own does
impose, in general, a significant (say, 100 per cent) distortionary cost
over and above the revenue collected. However, this does not mean that
the benefit–cost ratio for a public project has to exceed 2. There are off-
setting benefits at the spending side (on top of the direct benefits of the
public goods).3 If the valuation of the higher public spending is the same
as the extent of the higher taxes at each and every income level, then
the higher disincentive effects of the higher taxes will be exactly (subject
to the qualifications mentioned at the end of Section 9.2 above and
another qualification to be mentioned presently) offset to give no net
increase in disincentive effects, as shown by Kaplow. This is so because
when people reckon in terms of the total package of the higher taxes
and spending, the gains offset the losses at each income group. Putting
it differently, the fact that people at higher income levels place higher
values on the public goods makes the spending side possess negative dis-
incentive effects or positive incentive effects, offsetting the disincentive
effects of the higher taxes, when the higher taxes do cause disincentive
effects.4 On the other hand, if the degree at which the valuation of the
higher public spending increases with income is less than the degree at
which higher taxes increases with income, such that there are higher
disincentive effects for the total package (of higher taxes and higher
spending), there exist offsetting distributional gains (since the poor gain
and the rich lose in net terms). In the opposite case where the poor lose
and the rich gain in net terms, there are distributional losses. But then
the combined disincentive effects of the total package are negative. In
any of these three cases, the Kaplow principle applies.
However, apart from the qualifications mentioned at the end of
Section 9.2, the Kaplow principle is subject to another important quali-
fication. (See Appendix C of Ng, forthcoming b, on another more com-
plicated consideration.) Kaplow’s analysis ignores the behavioural
responses to higher taxes except through the income (consumption)/
leisure choice. As Feldstein (1997) emphasises, there are other responses
including substitution into less tax-assessable spending like luxurious
offices. This encouragement of inefficient choices does increase the dis-
tortionary costs of higher taxes despite Kaplow’s argument. Higher taxes
encourage people at each and almost all income levels (especially those
facing high marginal tax rates) to use less-beneficial though more tax
(avoidance)-effective ways of spending money. This extra distortion is
not offset by any extra benefit of the higher public spending. This quali-
fication to the Kaplow principle applies also to illegal means of tax
evasion: higher tax rates encourage more evasion. Here, the loss in tax
revenue itself is not a distortionary cost as it is offset by the gain of the
taxpayers. The distortionary cost consists in the fact that the taxpayers
would prefer to have the higher incomes legally, without having to go
into the dubious, time-consuming, and likely less beneficial means that
are more tax-effective. In practice, this may be the most important
qualification to the Kaplow principle.
A query may arise: Why do the higher benefits of the higher
public spending, financed by the higher compensatory taxes, offset
the higher tax-rates to produce neutrality (no extra disincentive effect)
in income/leisure choice, but do not offset the higher tax-rates to
also produce neutrality in tax evasion and other similar choices? This
is so because if an income earner decides to have more leisure and earn

less income, she forgoes not only the after-tax income but also the ben-
efits (e.g. protection of more property) associated with higher incomes.
(For the case where the benefits from public spending is related to the
publicly unobservable earning abilities rather than incomes, a qualifi-
cation may be needed as discussed in Ng, forthcoming b.) On the other
hand, when a person under-reports the income she actually does earn,
she does not forgo, by and large, the benefits (from higher public spend-
ing) associated with higher (actual) incomes. Those parts of the ben-
efits of higher public spending that only apply to reported incomes will
in fact offset the higher tax-rates and reduce the incentives for tax
evasion. If all the benefits depend on reported than on actual incomes,
then the qualification to the Kaplow principle being discussed is not
needed.
It may be noted that the above qualification to the Kaplow principle
due to non-leisure behavioural responses need not apply to the princi-
ple of ‘a dollar is a dollar’, for the following consideration. While higher
taxes/transfers on the rich/poor in lieu of specific equality-oriented pol-
icies tend to encourage more tax evasion and avoidance, the specific
equality-oriented policies themselves also encourage evasion and avoid-
ance. Just as one may pretend to be on low income to pay less tax, one
may also pretend to be on low income to have access to specific ben-
efits for low income-earners. Unless there is asymmetry in the extent of
losses in the ‘wrong’ direction, no qualification to the principle of ‘a
dollar is a dollar’ is needed here. I discussed this and similar issues under
‘transaction costs’ in Section 6.4 and concluded that the asymmetry is
actually in the ‘right’ direction, with an a fortiori case against specific
equality-oriented policies.
Another query may arise: Why is the Kaplow principle subject to
the qualification of non-leisure behavioural responses while the princi-
ple of ‘a dollar is a dollar’ is not? This may be explained thus. The more
progressive tax/transfer system used in the argument for the principle
of ‘a dollar is a dollar’ is in lieu of a system of specific purely equality-
oriented policies that is itself subject to similar behavioural responses.
For the Kaplow principle, the higher taxes are used to finance for extra
public spending that does not exist yet. The losses due to the behav-
ioural responses to the higher tax rates have thus to be taken into
accounted in assessing whether the extra public spending is worth
undertaking.
The argument of Feldstein regarding the extra distortionary costs due
to the various non-leisure responses of individuals has itself to be
subject to an important qualification. Two different types (for sim-

plicity, we take the two polar types, though various degrees of mixture
of the two types may be involved in practice) of (legal or illegal) tax-
free (or lower tax-rates) spending should be distinguished. First, there
are various items that the government wants to (or should) encourage
for some social purposes, including the efficiency consideration of exter-
nal benefits (e.g. health care, the prevention of communicable diseases
in particular) and the distributional one of poverty reduction (e.g. char-
itable gifts). If the higher taxes encourage people in genuinely spend-
ing more on these items that the society wants to encourage, there is
no extra deadweight loss. (If the increase in taxes is so huge that the
extra spending becomes too excessive, the degree of tax deductibility
may need to be reduced to revert to optimality.) This is so because the
reduced benefits to the taxpayers are offset by the increased gains of
external benefits or distributional benefits.
Secondly, the higher tax rates may also encourage people to spend
more on those items that the government does not really want to
encourage but nevertheless has to treat them as tax-deductible due to
the difficulties of distinguishing them from items that genuinely should
qualify for tax-deductibility. These include (but may not be confined
to): 1. pretended, non-genuine spending on deductible items, e.g.
claiming private dining as business expenses; 2. excess spending on
deductible items, e.g. big offices. These types of spending create extra
distortionary costs as explained by Feldstein. They do not produce com-
pensating benefits like external or distributional benefits like the first
type of tax-free spending.
Even without accepting Kaplow’s argument, the estimation of the dis-
tortionary costs of taxation should reflect those due to behavioural
changes of the second (encouragement-not-intended) type but not the
first (encouragement-intended) type. Feldstein includes items like chari-
table gifts and health care in his discussion of behavioural responses.
These items clearly largely belong to the encouragement-intended type.
Also, his method of estimating taxable income elasticities (used in turn
to estimate the distortionary costs) by the blanket comparison of actual
tax revenues before and after tax-rate changes necessarily lump the two
types of responses together. Thus, if the method is used, a separate esti-
mate of the part due to the first type of response should be made and
deducted from the total to give an appropriate estimate of the true dis-
tortionary costs, before considering Kaplow’s point. The need for the
separate treatment of the two types of responses is established in a more
formal analysis in Ng (forthcoming b).
Putting the above arguments and those of Chapter 8 together, it means

that, in estimating the distortionary costs of taxation for determining
the appropriate excess (of unity) benefit–cost ratio for public spending,
the relevant positive (i.e. in favour of a positive excess) effects are the
non-income responses of type two (those not purposefully encouraged
by the society) and the relevant negative effects include the bias due to
the importance of relative-income effects and the corrective nature of
most existing taxes due to environmental disruption of most pro-
duction and consumption. Before a reasonable estimation of these
opposing effects, it is difficult to say whether the appropriate
benefit–cost ratio for public spending should exceed or fall short of
unity. I agree with Feldstein (1997, p.197) that the ‘central public
finance question facing any country is the appropriate size of its gov-
ernment’ and that economists should put more effort in trying to help
answer this question. Our discussion indicates that both a comprehen-
sive consideration taking all relevant issues into account and much
more quantitative estimation of the relevant factors are needed.
10
Concluding Remarks
Discerning readers may have noticed that Part I of this book is some-
what ‘rightwing’ while Part II is ‘leftwing’. Thus, Part I argues strongly
for utilitarianism (Chapter 5) and against pure egalitarianism (Appen-
dix B), and for the principle of ‘a dollar is a dollar’ (Chapter 6). On the
other hand, Part II argues for higher public spending, especially on
research and environmental protection. However, though the two parts
may be opposite in their political philosophy, they are not inconsistent
with each other. For example, the fact that the principle of ‘a dollar is
a dollar’ is consistent with the argument for higher public spending may
be seen easily as the latter is based purely on the efficiency consider-
ations of relative-income effects, environmental disruption effects, the
negative disincentive effects of public spending (the Kaplow principle),
the global public-good nature of research, etc. I may be schizophrenic
in the sense that my heart is leftwing and my head is rightwing.
However, there is no inconsistency in my arguments. In fact, the op-
posite political inclinations of the two parts demonstrate that I follow
the dictates of my logical analysis wherever they take me, without stick-
ing to a political preconception.
While there is no inconsistency, there may be some tension between
different arguments. For example, my allowance for the difference
between preference and welfare due to non-affective altruism (or
malice), ignorance, and irrationality (Chapter 4) and the use of the will-
ingness to pay (Chapter 6) need some explanation. As should be clear
from Chapter 3 and Chapter 4, I view happiness or welfare as the appro-
priate ultimate objective. However, partly because it is more difficult to
obtain information on happiness than on willingness to pay, especially
on economic issues, and partly because of the possible negative effects
of overriding people’s preferences even for their own welfare, I have
135
gone along with the economist’s convention of using preference or

utility instead of welfare in much of the discussion. However, if there
is a big conflict between the two, I would opt for welfare at the ultimate
level. For example, if it is true that economic growth does not increase
happiness or even certain objective indicators of the quality of life much
(Chapter 8), the calculation based on willingness to pay may have to
be adjusted a lot to arrive at something appropriate for the pursuit of
the objective of welfare. Such adjustments may include not only such
factors as relative-income effects, environmental disruption effects, etc.
discussed in Chapter 8, but also ignorance and irrationality if these
prove to be substantial. That there is not much agreement on these
issues reflects mainly our lack of knowledge than our ideological dif-
ferences. (In fact, the so-called ideological differences arise from the lack
of precise knowledge, making each person having then to resort to some
estimates based on one’s experience, education, upbringing, and psy-
chology.) This just shows how much more research is needed.
Appendix A
Electrical Brain Stimulation: a Case
Showing the Importance of Public
Spending in Research
Cutting government funding on research is starving the goose that lays the
golden eggs. In a developed economy like ours, it is research that contributes
most to social welfare and further economic growth. Without sufficient safe-
guard on protecting the environment and without increasing welfare-improving
measures like research, pure increases in income and consumption may well be
welfare-decreasing.
A1 Misconceptions on the costs of public spending

With the perceived needs inflated by comparison to neighbours, friends, class-
mates of children, colleagues, etc. most people still strive hard to get more money
even in our affluent society where most essential needs have been more than sat-
isfied for most people. Politicians respond to this rat race to material abundance
by emphasising private disposable income and consumption. However, after a
certain minimum level of consumption, competition between individuals for
higher income and consumption are largely self-defeating from the social view-
point. More importantly, the belief that taxation causes excess burdens or inef-
ficiency is also questionable. Thus, as argued in Part II in the text, our welfare
can be increased much more by public spending to safeguard the environment,
to acquire and spread knowledge through research and education which also
have positive feedbacks on economic growth.
Imagine a trebling in your income but without access to computers, tele-
vision, phones, modern medical facilities, etc. wouldn’t your welfare be reduced
significantly? We are on the brink of massive scientific–technological break-
throughs that will increase our welfare by quantum leaps. Such leaps in
welfare can be brought forward to be enjoyed by people of this generation if
research activities are significantly increased. Tantalising possibilities abound
in the field of genetic engineering, though caution is needed in such areas.
However, this appendix focuses on the wonderful potential of electrical brain
stimulation.
A2 Direct access to intense pleasure

It has been known for more than four decades that electrical brain stimulation
(EBS) can relieve acute pain, induce intense pleasure, and promote a sense of
137
138 Appendix A
well-being without the undesirable health effects of drug addiction. How-

ever, apart from isolated research experiments and limited therapies (see, e.g.
Devinsky et al., 1993), the enormous potential benefits of EBS have neither been
adequately explored nor widely discussed. Much increased research effort and
eventual widespread use of EBS are called for.
Positive reward associated with EBS leading to voluntary self-stimulation was
discovered by Olds and Milner (1954) when they observed that a rat returned
to the place where it received direct electrical stimulation of certain parts of its
brain. Further research established sites that induce pleasure (medial forebrain
bundle, septal, limbic and hypothalamic areas), pain, and ambiguous or mixed
feelings. Stimulation of the pleasurable sites clearly produces positive reward as
suggested by experiments in which rats were willing to cross a painful shock grid
in order to obtain the stimulation, and as confirmed by human subjects. More-
over, the pleasure induced is so intense that rats prefer EBS to food and sex, and
if not stopped by experimenters, will continuously seek stimulation until exhaus-
tion. In humans, ‘patients who were having emotional or physical pain experi-
enced such intense pleasure with stimulation that the pain was obliterated’
(Heath et al., 1968, p.188).
A3 Enormous benefits
Apart from relieving pain and inducing pleasure, EBS may also be used as a
‘primer’ in improving well-being. For example, Heath (1964, p.236) reported,
‘strong pleasure [from brain stimulation] was associated with sexual feel-
ings, and in most instances the patient experienced spontaneous orgasm. . . .
This patient, now married to her third husband, had never experienced orgasm
before she received . . . stimulation to the brain, but since then has consistently
achieved climax during sexual relations’. Once the right neurons have been
excited, they become excitable more easily. The right neural pathways have been
established.
Among the important social problems of our time are drug addiction, crimes
and (mental) depression. These social problems, and possibly others, seem to be
largely solvable with the widespread use of EBS. In comparison to EBS, the use
of addictive drugs like heroin is a very inefficient and dangerous method of
achieving a ‘high’. If one has easy access to pleasurable sensations by just turning
on the electricity, there seems little reason left to try dangerous alternatives like
heroin. Just as intractable pain may be relieved by EBS, mental depression should
also be largely removable by positive EBS. Since most depressions are caused by
failure to achieve happiness one way or other, the availability of happy sen-
sations by EBS should provide a definite relief. Among others, the amelioration
of stress (Patterson et al., 1994), reduction of stress ulcers (Yadin and Thomas,
1996), improved performance in a maze (Jiang et al., 1997), and the treatment
of alcoholics (Krupitskii et al., 1993) have been reported.
A4 Safe and long-lasting pleasure

Though EBS is not physically addictive, it might be psychologically addictive.
However, in contrast to heroin addiction, EBS addiction is not dangerous to
health. From the quite large amount of evidence we have, the proper use
Appendix A 139
(Patterson and Kesner, 1981) of EBS over a sustained period for a long time (e.g.
a few hours daily over a number of years) has proved to be quite safe. Thus EBS
addiction is only a problem if it leads to the serious disregard of other duties
such as to threaten the welfare of (mainly) other people (especially the future
generations). While the pleasures induced by EBS can be intense, I doubt
that psychological addiction of such a magnitude would occur. Rats choose to
use EBS until exhaustion but humans only for ‘up to half an hour daily’ (Sem-
Jacobsen, reported in Delgado, 1976, p.484). Relative to other pleasures and
objectives, the pleasure of EBS does not seem to be compelling for humans
(Bishop et al., 1964; Valenstein, 1973, p.28). If one believes in creation, perhaps
God made us this way so that we could eventually provide happiness not only
for ourselves but also for animals. In the unlikely event of serious addiction, the
problem could be solved by using legal and/or technical devices restricting the
unlimited use of EBS.
While EBS addiction is unlikely to be so serious as to threaten the survival of
a civilised society, it may be feared that it would significantly reduce mutual
human relationships. If one could obtain pleasure by simply turning on the
electricity, there might be little motivation left for the cultivation of personal
relationships. This is unlikely to happen. Even if one could obtain a variety of
pleasurable sensations by EBS, there would still be the innate need for com-
panionship left. Secondly, the pleasure from EBS to humans does not seem to be
as fulfilling as, say, a full sexual relationship with its simultaneous stimulation
of a number of areas and close personal contact, nor as rewarding as spiritual
fulfilment of the highest order. Thirdly, the provision of pleasure which might
otherwise be unavailable in sufficient amount may in fact create many happy
and easy-to-go-with individuals. This may remove many personal conflicts and
promote better mutual relationships. Fourthly, even if personal relationships
were reduced, the benefits of EBS would still likely to more than compensate for
the loss. For example, the introduction of television probably has significantly
reduced conversation. But that does not necessarily make it a bad thing. Its ben-
efits have to be taken into account as well.
In this connection, the long-lasting nature of pleasure from EBS definitely gives
it a big advantage. Inventions like television usually appear to have enormous
potential benefits around the time of their initial introduction. After prolonged
usage, some of their disadvantages are discovered though some other beneficial
usages may also be found. More importantly, the novelty value has disappeared.
For example, while watching television is very enjoyable for those just getting
access to it, it may become a second best option after its novelty value has dis-
appeared. The benefits of television probably still outweigh its costs by a very
wide margin, but not by as much as it would be if the novelty value could be
maintained. With EBS, the situation would be different. Since EBS is the direct
stimulation of the brain, the pleasure during stimulation does not depend on
any novelty value. Moreover, the intensity of pleasure from EBS does not dimin-
ish with prolonged stimulation (either continuously or daily over a number of
years). Thus the enormous increase in happiness brought about by EBS could be
expected to be maintained largely unabated, and in fact could be greatly
increased through better techniques of stimulation.
EBS may be regarded as unnatural in the sense that it does not occur in the
course of our natural biological survival. But most civilized products, institutions,
medical treatments, etc. are unnatural in this sense. This does not make them
140 Appendix A
bad. To improve our welfare, we have invented many ‘unnatural’ things. EBS is
a recent invention that if properly made use of widely, possesses welfare signifi-
cance surpassing all previous inventions put together.
Many people from the West may find, upon first contact, the culture, tra-
dition, and ways of enjoying life in the East and in some primitive tribes degrad-
ing. The same is true for people from the East on some Western ways of life. But
we have learned from liberalism to be more tolerant towards different cultures
and ways of life as long as they are not harmful. Many liberals would go further
in tolerating individual freedom of action even for those actions which are
harmful to the actors themselves. EBS is about the least harmful way of induc-
ing intense pleasure and should never be regarded as degrading by anyone who
has the slightest adherence to liberalism.
Will God approve EBS? If one does not believe in God, the question does
not arise. If one believes in God, then the answer seems to be affirmative. For
example, the ten commandments do not include: Thou shall not engage in EBS.
Nor do they include: Thou shall not enjoy yourself. Moreover, if God does not
want us to use EBS, why did He create us in a way that EBS can induce intense
positive reward?
If higher funding for research could result in such spectacularly welfare-
improving discoveries and inventions as EBS, the present writer would be pre-
pared to halve his post-tax income to help pay for them.
Appendix B
Pure Egalitarianism: a Critique
Egalitarianism has made much headway. When utilitarianism was first proposed
by classical utilitarians in the 19th century, it was regarded as a radical philos-
ophy used to justify egalitarian policies through such arguments as diminishing
marginal utility. (For a modern statement of this, see Hare, 1997.) Now, most
people writing on related issues (e.g. Rawls, 1971; Dworkin, 1981; Nielsen, 1985;
Sen, 1992 and Roemer, 1996) seem to view utilitarianism as a conservative
philosophy; they prefer much more radical egalitarianism. In this appendix,
it is argued that this is based on questionable reasoning. On the other hand,
there are writers arguing for something much to the right of utilitarianism. For
example, Nozick (1974) argues for liberty and against equality; Posner (1981)
argues for wealth maximisation. Such extreme divergence of views is, at least
partly, based on the confusion of non-ultimate considerations with basic values,
rather than just on the difference in basic value judgements.
For example, as argued in Chapter 6, for whatever degree of equality-efficiency
tradeoff aimed at, it is more efficient to pursue efficiency or wealth maximisa-
tion (treating a dollar as a dollar whomsoever it goes to) on specific issues of
public policy, leaving the equality objective to be pursued through the general
tax/transfer system. However, as the ultimate social objective, wealth maxi-
misation is clearly unacceptable. (Another example of confusing non-ultimate
considerations with basic values is the concern, e.g. Kappel, 1997, for simulta-
neous inequality with life-time equality.)
At the non-ultimate (e.g. political, day-to-day) levels, liberty, efficiency, equal-
ity of incomes (or resources, primary goods, capabilities), etc. are all important
considerations. However, they typically may be in conflict and need be traded-
off with each others. At the ultimate level, the optimal tradeoff should be decided
by the maximisation of aggregate individual welfares, as argued in Section B1.
This is also a form of egalitarianism, equality in the weights attached to the
welfare of all individuals. This is the only appropriate equality to insist on at the
ultimate level (Section B2). At this ultimate level, the arguments of the libertar-
ians (Section B3), Rawls (Section B4), various forms of egalitarianism including
resource and welfare egalitarianism (Section B5) and capability egalitarianism
(Section B6) are shown to be unacceptable.
B1 Why is happiness the only ultimate consideration?

As argued in Section 4.4 in the text, things we want in life like money, position,
relationships, health, etc. are not wanted for their own sake. Only extreme misers
want money for money’s sake. In fact, it may be argued that even misers do not
want money for money’s sake. They just feel happy by looking and knowing that
141
142 Appendix B
they possess a lot of money. We all want money, health, etc. in order to be happy.
But we do not want happiness for the purpose of obtaining anything else. Hap-
piness is good in itself.
We do not have to ask why is happiness valuable or good because we all enjoy
our happy feelings and suffer from unhappy feelings. We are like this because of
biological reasons. We are intelligent beings capable of making choices. (This is
also true for at least some animal species higher up in the phylogenetic scale.)
Due to this capability, God (or evolution) found it necessary to ensure that we
make the ‘right’ choices (from the viewpoint of survival and reproduction, i.e.
biological fitness). Thus we are genetically programmed to want to do the fitness-
enhancing things by being awarded with pleasures when eating fresh and nu-
tritious foods when hungry, when mating with fertile members of the opposite
sex, and penalised with pain when we injure ourselves.
It may also be true that we have an instinct for justice or morality. At least we
have a sense of morality which may be partly inborn and partly learned. As long
as the sense of morality is formed by nature and/or by nurture, we also feel
happy/unhappy in doing/seeing/suffering something morally good/bad. Such
happiness/unhappiness should of course be included in the utilitarian calculus.
However, moral principles are ultimately valuable only because of their contri-
bution to welfare, as argued in Section 4.1.
Like other non-ultimate objectives like freedom, democracy, etc. equality is
a valuable objective if and only if it contributes, directly or indirectly, to the
ultimate objective of happiness or welfare. (Individual welfare is defined to
be the net happiness of the individual.) Generally, equality is a highly valuable
objective as it contributes to welfare in several important ways, including
through the undisputed fact of diminishing marginal utility, through fostering
better social cohesion, and through our sense of justice. Thus, I regard myself as
a utilitarian egalitarian. To the extent that the pursuit of equality does not harm
too much other valuable objectives like liberty and efficiency, it contributes to
welfare.
Consider the hypothetical situation where most people do not like equality in
the distribution of income (or primary goods, etc.). They find equality dull. Each
of them would rather be one of the poor in a society of income inequality than
be one in a society of equal income, even at a higher average income. Then, even
abstracting from disincentive effects, equality itself may make everyone worse
off. Clearly, in this case, equality is not good for the society even if efficiency,
liberty, and other valuable objectives are not diminished. Thus, it is clear that,
ultimately speaking, equality is only of instrumental value; it is not valuable in
itself.
While the above example is hypothetical, it is true in the real world that people
ultimately want happiness, not equality in itself. They do want equality but only
to foster their welfare and perhaps also the welfare of others. Some evidence may
be adduced to support this claim. First, surveys indicate that people view hap-
piness as their ultimate objectives (Veenhoven, 1984 and Ng, 1996a). Secondly,
people from very poor countries of less unequal distribution still eagerly seek to
migrate to countries of higher incomes even if more unequally distributed, and
even if they would become one of the poorest class if they are successful in
migrating. Thirdly, most readers can convince themselves by asking what they
themselves value ultimately. Be persistent and try to go to the truly ultimate. If
Appendix B 143
you think you value equality, ask, ‘Why?’ Answer: ‘equality is good’. Then ask:
‘why is equality good?’ or ‘is equality good if it makes everyone suffer enor-
mously?’, etc.
B2 Equality in the welfare weights

In a sense, virtually everyone is in favour of some form of equality. The impor-
tant question is: ‘equality in what?’ This was realised as early as Aristotle (see
Pojman and Westmoreland, 1997, p.3) and emphasised by Sen (1992). Thus,
Nozick (1974) wants equality of libertarian rights, the utilitarians want equality
in the weights attached to each person’s welfare, Rawls (1971) wants equal liberty
and equality in the distribution of primary goods (unless inequality makes every-
one better off), Dworkin (1981) wants equality of resources, Sen (1992) wants
equality in capability, Arneson (1989), Cohen (1989) and Roemer (1996, 1998)
want equality of opportunities, Nielsen (1985) and Kolm (1996) want equality
in welfare. So, why is equality in welfare weights the right equality at the ulti-
mate level?
The answer is (partly) that resources, income, primary goods, capability, etc.
are not our ultimate objectives, it is welfare that is our ultimate objective. (A
problem faced by most if not all non-welfare objectives is the index number
problem of how to aggregate different bundles of say primary goods into a com-
mensurable measure.) If equality in resources, income, etc. decreases the welfare
level of everyone, then it is clearly undesirable. But then why not the equality
of welfare? The answer is that equality of welfare may also result (in fact it is
almost certainly true in most real-world situations due to individual differences
in endowment and capability and due to the disincentive effects of attempts to
equalise welfare) in very low welfare levels for most people. We want high welfare
levels, not equal welfare levels with others. I hate receiving lower incomes than
others. If the incomes of all others are doubled and mine only increases by 10
per cent, I will probably be made worse off. However, if the welfare levels of all
others are doubled and mine only increases by 10 per cent, I am still 10 per cent
better off! I would love that if it were possible.
We all want higher welfare levels. Thus, for the society, treating everyone’s
welfare on an equal footing and maximising the equally weighted (equivalent to
unweighted) sum of individual welfares is the morally right objective. Moreover,
as outlined and argued in Chapter 5, there are morally compelling reasons in
favour of maximising the unweighted sum of individual welfares.
B3 The libertarians
No doubt, liberty is extremely important. At the political level, it makes sense
most of the times to insists on liberty as if it is sacrosanct. The sacrifice of liberty
on some grounds of expediency may increase welfare at the moment but is likely
to produce undesirable side effects and hence undesirable in the long run even
from the viewpoint of welfare maximisation. Thus, again, I regard myself as a
utilitarian libertarian. However, liberty is not our ultimate objective. If we have
included all the side effects, most people, like me, would rather be happy but
not free than be free but unhappy.
144 Appendix B
Nozick (1974) emphasises the appropriate methods by which properties

are acquired rather than the equality in their distribution. Obviously, this is
extremely important for the orderly functioning of society and the adequate
incentives for productive activities. However, Nozick believes that this liberty
to acquire wealth by appropriate methods should be given unlimited scope. If
someone is endowed by the lottery of birth with a sexy voice that attracts
millions of teenagers and hence can earn millions of dollars by singing a few
songs, he is just enjoying his liberty. Others who are endowed with no income-
earning abilities only have their bad luck to blame. Provided that there is no
unjust method in the acquisition of wealth, the resulting distribution, no matter
how unequal, is just and the government should not try to make the distribu-
tion more equal. Doing so is an infringement on liberty. This is so even if, by a
marginal infringement (e.g. by a mildly progressive income tax/transfer), the
poor could be helped by an enormous amount.
A person may be poor as a result of squandering away his inherited wealth or
spending his time watching TV rather than learning and working. Most people
are not sympathetic towards such a person. On the other hand, a person may
be poor due to sheer bad luck in his endowments and opportunities, despite
his hard work. Similarly, a person may be rich by contributing a lot or by sheer
luck in endowment and opportunity. Inequalities in the distribution of income,
wealth, other primary goods, capabilities, etc. due to unjust methods are abhor-
rent, but those due to sheer luck also need addressing. (For the argument that,
if the fruit of labour is understood to include only ‘that subset of the . . . product
of one’s labour which is due to his choices rather than luck’, redistributive taxes
from the lucky rich to the unlucky poor does not violate self-ownership as
claimed by Nozick, see Michael, 1997.)
B4 Rawls
In contrast to the libertarians, Rawls (1971) does not accept the justice of unequal
endowments. But he goes to the other extreme in believing that primary goods
(i.e. income, wealth, opportunities, the social bases of self-respect, etc.) should
be distributed equally unless inequality makes everyone better off.
Rawls’ (1971, 1982) theory of justice may be summarised into two
principles:
1. Each person has an equal right to a fully adequate scheme of equal basic lib-
erties which is compatible with a similar scheme of liberties for all.
2. Social and economic inequalities are to satisfy two conditions. First, they must
be attached to offices and positions open to all under conditions of fair equal-
ity of opportunity; and secondly, they must be to the greatest benefit of the
least advantaged members of society.
I am prepared to accept the first principle on the following understanding:
• that it is adopted because it promotes the general welfare;
• in circumstances where it is disastrous to the general welfare, it may have to
be suspended;
• in deciding what is the ‘most extensive total system’ and what is ‘compat-
ible . . . for all’, the ultimate criterion is the general welfare.
Appendix B 145
A sex maniac may be in favour of freedom to rape and claim that this is com-
patible with everyone’s freedom to rape.1 It may also happen that the sex maniac
is the person of the lowest welfare level such that freedom to rape for all will
maximise the welfare of the worst off, hence consistent with the spirit of Rawls’
second principle to be discussed below. However, if freedom to rape results in
the reduction of the welfare of those raped and scared of being raped by more
than (in aggregate) the welfare gain of the rapists (even though the former still
have higher welfare levels than the latter group even with freedom to rape), then
freedom to rape should be regarded as not compatible with the freedom of not
being raped. The ‘most extensive total system of basic liberties’ should then not
include the freedom to rape. However, thus interpreted, the first principle is really
a device to promote the general welfare. It is not ultimate. The first part of Rawls’
second principle may be similarly accepted as the first principle.
Now, consider the second part of maximising the benefit of the least advan-
taged, with advantage being judged by the possession of primary goods. This
maximin principle is in direct contrast to the utilitarian principle of maximising
the unweighted sum of welfares. The latter gives equal weights to all while the
former gives an infinite weight to the welfare of the least advantaged. Being the
least advantaged, they probably have very high marginal welfare of income.
Giving equal weights to their welfares already implies valuing a dollar to them
as worth more than to the more advantaged groups. Giving an infinite weight
implies the willingness to sacrifice the welfare levels of all others enormously
(provided that none of them becomes more disadvantaged than the least advan-
taged) for a marginal increase in the welfare of the least advantaged. This is
morally unacceptable. In fact, if I were the least advantaged person, I do not
want any person/group in the society to sacrifice a big welfare loss just to increase
my own welfare a little. (For an interesting argument on the unacceptability of
the Rawlsian version of egalitarianism as it implies: ‘that redistribution ought to
aim to equalize the life expectancy of men and women by making men have
longer and women shorter lives . . . to employ fewer men and more women in
. . . undesirable jobs. Men . . . ought to have shorter working days and longer
vacations than women’, see Kekes, 1997, p.661.)
B5 Pure and welfare egalitarianism

Pure egalitarians insist on the equality in certain things (typically resources) irr-
espective of the consequences. For example, Watson (1977, p.123) argues that
equality of resources is such a transcendent value that, if equal distribution of
food were to result in no one getting enough to eat, we should nevertheless
choose the annihilation of the human race rather than an unequal distribution.
This is clearly an unreasonable position to take. As remarked by Pojman and
Westmoreland (1997, p.5), ‘if equality is a transcendent value . . . [the thorough-
going egalitarian] would have to dumb down the brilliant, infuse the healthy
with disease, and blind the sighted’.
Welfare egalitarians want not just equal weights on the welfare levels of all
individuals but equality in the welfare levels. For example, Ake (1975) regards
that justice in society as a whole ought to be understood as a complete equality
of the overall level of benefits and burdens of each member of that society.
146 Appendix B
Nielsen (1985) is also in favour of radical welfare egalitarianism. Sen (1992) is

not explicitly a welfare egalitarian. However, one may argue that, pursuing his
logic in moving (from some form of resource egalitarianism) to capability egali-
tarianism to its logical conclusion, one will end up in welfare egalitarianism (next
section).
Welfare egalitarianism is patently unacceptable as it may end up making every-
one substantially worse off. Suppose that the relevant utility feasibility frontier
is as illustrated in Figure B1 and the initial point determined by the given endow-
ment (of individual talents, etc.) after Nozick-just acquisitions and exchanges is
at point N. Libertarians insist on the point N even if the welfare of the poor (rep-
resented as Up on the vertical axis) could be increased a lot with the welfare of
the rich (represented as Ur on the horizontal axis) reduced only marginally, taking
everything into account. In contrast, Rawlsians would go all the way to R where
no further increase in Up is possible. Worse, pure and welfare egalitarians would
insist on E even if both the rich and the poor would be significantly worse off
than the initial point N. The equally absurd maximax point M has also received
some support (e.g. Temkin 1987).
Roemer (1986, 1996, Chapter 7) argues that resource egalitarianism implies
welfare egalitarianism. This argument is based on an axiom unacceptable
from the viewpoint of social welfare maximisation with welfare contours allow-
ing some trade off between the individual welfare levels of different persons.
However, since this axiom is compelling from the viewpoint of resource egali-
tarianism, Roemer’s argument remains valid. Nevertheless, the following expla-
nation not only shows how Roemer can reach his rather strong conclusion but
also why both resource and welfare egalitarianism are unacceptable, at least from
the viewpoint of social welfare maximisation with tradeoff. The relevant axiom
of Roemer is Resource Monotonicity which requires that no individual welfare
level should decrease with an increase in resources for the whole society. The
Ur
N
M
E
45°
0 Up
Figure B1
Appendix B 147
Uk
3
A F
1
F
2
F E
W2
W1
45°
0 C UJ
Figure B2
point that this apparently innocent axiom is objectionable and the point that it
(together with other reasonable axioms) leads to welfare egalitarianism may be
illustrated together in Figure B2.
First, suppose that, in situation 1 with a given amount of resources and the
utility functions of the two persons, we have the symmetrical utility feasibility
frontier (UFF) F1. (Points on and within the frontier are feasible to attain; points
above or beyond the frontier are not feasible.) The reasonable axiom of sym-
metry assumed by Roemer then dictates the choice of the equal welfare/utility
point E. The assumed concavity in the individual utility functions without inter-
dependence and the abstraction from production considerations may mean that
the UFF is concave (the feasibility set is convex). However, since no upper limit
is placed on the degree of concavity, we have as the limiting case, F2 as a possi-
bility. The point E still has to be chosen given F2 (from either symmetry or from
the Pareto principle). Now suppose that the amount of resources increases unam-
biguously but in a way that increases the utility possibility of person K only, such
that the new UFF F3 is the curve ABEC. Resource Monotonicity requires that, in
this new situation with F3, no individual utility can be lower in comparison to
the situation at F1 or F2. This means that E still has to be chosen from F3. There
is no upper limit as to how steep the section BE of F3 may be, except that it is
downward sloping. If BE is steep enough, any SWF allowing for some tradeoff
between utility levels of different persons, including those satisfying symmetry,
such as one represented by the welfare contours W1 and W2, will choose a point
B from F3 that is northwest of E. This shows that Resource Monotonity, by insist-
ing on E, is unreasonable.
Consider a concrete example. Suppose that the two persons concerned are my
brother and myself and that no other sentient will be affected. Suppose first that
we are endowed only with food which we enjoy equally with diminishing mar-
148 Appendix B
ginal utility. It makes sense to share the food equally. (The issues of desert and
incentives do not arise by assumption.) Now, suppose that we also have a ball
(assumed not exchangeable into food). I hate ball games while my brother loves
them. It makes sense to let him have the ball. However, he enjoys playing with
the ball only if he has enough to eat; the more he plays with the ball, the hun-
grier he becomes. Thus, his utility can only increase substantially if he has more
than half of the food. Ball games increase his marginal utility of food dramati-
cally. It makes sense for me to agree to let him eat more than I do. The increase
in endowment of a ball should in this case result in an optimal outcome involv-
ing my having less welfare. Is this unfair on me? I genuinely think not. If the
increase in endowment is in the form of a sword with which I like to play very
much and which my brother hates, then I should similarly have more to eat to
be able to enjoy the sword. Provided this will also be done, there is no unfair-
ness. My welfare and that of my brother has the same weight in the social welfare
function of our two-person society. By maximising the unweighted sum of util-
ities, the society maximises the expected utility of each individual under the veil
of ignorance. This is fair and moral.
B6 Sen and capability egalitarianism

Sen (1985, 1992) argues for the equality in capability rather than in objective
resources or subjective welfare. He regards Rawls as making an advancement
in going from just equality in income and wealth to wanting equality in the
distribution of primary goods (unless inequality makes the least advantaged
better off) which also include opportunities and the social basis of self-respect.
However, Sen regards this as insufficient. People who are disadvantaged physi-
cally, mentally, or socially need more resources and opportunities to be able to
perform the various valuable functions. These functionings: ‘can vary from most
elementary ones such as being well-nourished, avoiding escapable morbidity and
premature mortality, etc., to quite complex and sophisticated achievements, such
as having self-respect, being able to take part in the life of the community’ (Sen,
1992, p.5). The disadvantaged should be given more resources and oppor-
tunities to allow them to have equal capability to achieve these functionings
than others.
I regard Sen’s emphasis on capability and functioning and the need to pay
special attention to the disadvantaged as very important at the political level, as
it may be a good practical way to increase overall social welfare. However, if
viewed at the ultimate level, capability egalitarianism is not an acceptable objec-
tive as it violates the Pareto principle. (It should be noted that Sen also empha-
sises other desirable aspect of human conditions including opulence or wealth,
happiness, and agency goals as discussed below.) To see this, consider Figure B1
again, but now using the axes to measure the capability levels of the two groups.
The capability feasibility frontier may look like the UFF curve. To achieve equal-
ity in capability, we may have to tax the advantaged rich so much that the dis-
incentive effects will be so big as to make all groups having a lower capability
level at E. To avoid this Pareto inferior outcome, perhaps the capability egali-
tarians may make a concession like Rawls in going for maximin than for com-
plete equality in capability. This is still unacceptable. If I were the least capable
Appendix B 149
person, I would not think it is morally right for the society to incur huge costs
to improve my capability marginally.
Sen (1992, p.84) argues that: ‘Capability represents freedom, whereas primary
goods tell us only about the means to freedom.’ I will similarly argue that ‘welfare
is the ultimate objective, whereas capability tells us only about the means to
welfare’.
Sen does not go for equality in welfare for two reasons. One is the issue of
responsibility also emphasised by others (e.g. Dworkin, 1981). The society should
compensate the disadvantaged for circumstances for which they are not respon-
sible, not for something for which they are responsible, including inequalities
from the exercise of personal preferences. The other one is Sen’s view that people
are not only concerned with welfare. Let us come to this second point later and
assume now that people are only concerned ultimately with welfare. Then, the
following argument is reasonable. A person is no less disadvantaged and no more
accountable if he is unlucky enough to be born with some handicaps in per-
forming valuable functions than if he is unlucky enough to have lower inborn
earning abilities and/or is less lucky in business. This may justify the movement
from equality in (objective) resources to equality in primary goods and in
capability. However, following this logic to its logical conclusion compels us to
welfare egalitarianism. If people want to maximise their own welfare and, with
equal capability,2 end up with different welfare levels, this must be due either to
different abilities in making appropriate choices (if this ability is included in the
definition of capability and is equalised, only the next reason applies) or to
differences in luck. Differences in ability to make appropriate choices are not
different morally from differences in ability to earn income and in capability
to perform functionings. Luck in obtaining higher welfare levels is no differ-
ent morally from luck in earning higher incomes. Thus, if we should move
from resource egalitarianism to capability egalitarianism, we should also move
from capability egalitarianism to welfare egalitarianism. (Cf. the arguments of
Arneson, 1989/1995.)
Sen will object to the argument of the above paragraph on the ground that
people do not just want to maximise welfare/happiness but have other objec-
tives as well, including opulence (wealth) and agency goals. Even for misers,
wealth or money is not wanted for its own sake. While most people want money
to buy goods now or in the future (for the purposes of consumption, security,
power, rivalry and indication of success), misers derive so much pleasure from
looking at the money that spending money is seriously curtailed. But no one
wants money for its own sake. Some people may enjoy the process or the fact of
making money. Still, it is the enjoyment, the prestige, etc. that counts. These all
increase the happiness for them to be valuable. Thus, opulence is not an ulti-
mate objective.
People also want to do certain things, to accomplish certain achievements, and
prefer to do so themselves. For example, I want the moral compellingness of utili-
tarianism at the fundamental level to be generally accepted. If I myself con-
tribute to this acceptance, all the better. This is what Sen means by agency goals.
This is important enough, but again not at the fundamental level. I like to con-
tribute to this enlightenment myself because it makes me feel good and also
increases the respect of others for me which again makes me feel good and more
likely to be successful in other activities which are in turn pursued for the purpose
150 Appendix B
of increasing welfare (including my own and that of other sentients). Thus, I

see no conflict of rational agency goals (or any other rational objectives) with
welfare, ultimately speaking. (For further arguments in favour of welfare as the
ultimate objective, see Chapters 3 and 4. In particular, welfare-inconsistent
preferences such as myopia, dangerous sexual practices, and achievement for
achievement’s sake are argued to be irrational preferences that may have some
biological and educational explanation, i.e. nature plus nurture. Natural selec-
tion, through our unfeeling genes, programmes us to incline to maximise our
survival fitness which may be inconsistent with our welfare which we, in con-
trast to our genes, should maximise.)
Even with only welfare as the ultimate objective, some people may choose
partly with the welfare of others in mind. Then, with differences in the degree
of genuine altruism (see Section 4.1 for a compelling argument on its existence),
people with the same capability may end up with different welfare levels, even
with no differences in the ability to choose and in luck. Abstracting from unex-
pected outcomes contrary to intention, the more altruistic typically end up with
less welfare levels. On this point, capability egalitarianism seems to be morally
inferior to welfare egalitarianism. However, it has been argued above that both
forms of egalitarianism are unacceptable as the ultimate social objective.
B7 Concluding remark
This appendix has examined several views regarding egalitarianism, including
the libertarian, the Rawlsian, welfare egalitarianism, and capability egalitarian-
ism and has found them to be unacceptable as an ultimate social objective. That
such extremely different views are held by intelligent and reasonable authors
may be explained by the confusion of non-ultimate considerations with basic
values. For example, the reason why Nielsen (1985, p.45, p.53) is in favour of
having ‘the income and wealth . . . be so divided that each person will have a
right to an equal share’ is to avoid giving ‘one person power or control over
another’. Presumably, being subject to the power and control by others is bad
for welfare. Thus, when we reckon at the ultimate level, we may take into account
such negative effects of inequality of wealth and analyse using the ultimate
objective of welfare. Then, the society should not go for equal welfare but for
maximum aggregate welfare as individuals want more welfare for themselves,
not equal welfare with others.
Liberty, equality of resources, of capability, etc. may be desirable aspects usually
conducive to the general welfare when other things are held unchanged.
However, they are not the appropriate ultimate objective for the society. As
argued here and elsewhere in this book, the appropriate ultimate social objec-
tive should be individual happiness or welfare. The maximisation of the
unweighted (equivalent to equal weights for all individuals) sum of these indi-
vidual welfares (i.e. utilitarianism) is the appropriate form of egalitarianism.
In fact, Section 5.2 shows that Rational Individualistic Egalitarianism implies
utilitarianism.
Appendix C
Economic Growth Increases the
Optimal Share of the Public Sector
Joint Author: Siang Ng
There are two very important observations, one well known to all econ-
omists and uncontroversial and one not well known and may be contro-
versial. The first is that government spending as a share of GDP (or GNP)
increases historically and increases with per capita GDP both intertemporarily
and cross-nationally. (See Section 8.1.) The second is that, for rich countries, the
level of happiness does not increase with per capita income. (See Section
8.5.) The main purpose of this appendix is to provide a simple but reason-
able analytical model that simultaneously explains these two apparently
unrelated observations. The main results are summarised into two propositions
below.
Without intending to be exhaustive, some important factors affecting the
welfare of a person are included in the analysis proposed below. First, we accept
the traditional variables focused most closely by economists, personal con-
sumption c (in real terms, as also true for all variables discussed in this appen-
dix) and leisure x. Secondly, the amount of public goods provision g is included.
Obviously, apart from personal consumption, one benefits from the provision of
law and order and other public goods and facilities. This has also been well recog-
nised and analysed by economists. Thirdly, as discussed in Section 8.5 in the text,
we include relative consumption or relative income r. (For our purpose here, it
does not really matter which one is used.) Fourthly, environmental quality E is
included. Again, in the current era of heavy pollution, congestion, deforestation,
and other forms of environmental disruption, the importance of this factor needs
hardly any justification.
Of course, there are other factors affecting happiness; they are not analysed
here. Also, we ignore the possible existence of ignorance, imperfect foresight,
and other possible divergences (non-affective altruism and irrationality) between
preference (represented by a utility function) and happiness (represented by a
personal welfare function). Arguments on the existence and implications of these
divergences have been discussed in Chapter 4.
We thus take the utility or welfare function of an individual as
U = U (c , x, r , g , E ); Uc , Ux , Ur , Ug , UE > 0 (C1)
Where r ∫ y/Y, y is the income of the individual concerned and Y is the average
income of the relevant society, a subscript denotes partial differentiation, e.g. Uc
= dU/dc. An individual maximises (C1) subject to:
151
152 Appendix C
c = (1 - t )y (C2)
y = (1 - x )w (C3)
taking g, E, t, w as given; where t is the proportional (assumed for simplicity)
income tax, and w is the exogenously given earning ability, and the time con-
straint is normalised to be unity. Assuming an interior solution (compelling as
starvation and exhaustion are both unacceptable), the first-order solution is:
Ux = (1 - t )wUc + ( w Y )Ur (C4)
Now consider the effect of personal enrichment. In our model, this is an

increase in one’s own earning rate w. Differentiating (C1) with respect to w,
taking t, g and E as given (as viewed by the individual), we have:
dU dc ˆ dx ˆ dr ˆ
= Uc Ê + Ux Ê + Ur Ê (C5)
dw Ë dw ¯ Ë dw ¯ Ë dw ¯
Substituting in dc/dw and dr/dw from the differentiation of (C2), (C3) and r ∫ y/Y,
taking Y as given, and substituting in (C4) (which makes the term associated
with dx/dw equals zero; intuitively, as the individual starts with an optimal pos-
ition with respect to the choice of leisure x, a marginal change in x has no first-
order effect on utility), we have after multiplying through by w/U to express in
proportionate or elasticity terms:
sUw = hUc + hUr (C6)
where s ∫ (dU/dw)w/U is the proportionate total response of U with respect to
Uw
a change in w (‘total’ in the sense of allowing other relevant variables to change

endogenously; but g, E, Y are not relevant variables here), hab ∫ (da/db)b/a for any
a and b is the proportionate partial response of a with respect to a change in b
(‘partial’ in the sense of holding other variables unchanged). Equation C6 indi-
cates that the degree of importance of enrichment for an individual equals the
sum of the intrinsic consumption effect and the relative-income effect. For the
poor, the intrinsic consumption effect is big though the relative-income effect
may be at least relatively small; for the rich, the relative-income effect is big
though the intrinsic consumption effect may be small. Thus, for almost all
people, making more money seems very important, explaining why most people,
in rich and poor countries alike are engaging in the rat race.
For the whole society, economic growth increases the earning ability not only
of one person but also of most persons. Ignoring distributional changes, we
consider the situation of the representative individual whose earning ability w
increases at the same rate as that of the average earning ability W, i.e. we have
w = W, dw = dW, x = X, Y = (1 - X)W, etc. where capital letter indicates the
average value of the relevant variable. A more complicated formulation in terms
of a continuous distribution of individuals of different earning abilities does not
change the central conclusions. Differentiating (C1) with respect to W, at w =
W, dw = dW, and allowing the aggregate variables Y, g, E to change correspond-
ingly, we have:
dU dc ˆ dx ˆ dg ˆ dE ˆ
= Uc Ê + Ux Ê + Ug Ê + UE Ê (C7)
dW Ë dw ¯ Ë dw ¯ Ë dW ¯ Ë dW ¯
Appendix C 153
Note that Ur does not appear in (C7) as y and Y change by the same proportion,
leaving r unchanged. We should now introduce the determination of govern-
ment spending on public goods g and environmental quality E. In our simple
model, these are:
g = N (1 - a )tY (C8)
E = E( A, Y ); E A > 0, EY < 0 (C9)
where N is the given number of individuals, a is the proportion of tax revenue
used for the abatement (A) of environmental disruption, leaving the proportion
(1 - a) for spending on public goods. That EY < 0 captures the environmental dis-
ruption effect of most production and consumption. We also have:
A = atYN (C10)
Substitute dc/dw, dg/dW, dE/dW from the differentiation of (C2), (C3), (C8), (C9),
taking N, t and a as given, we have, after simplification using (C4) (but we cannot
get rid of all the terms associated with dx/dw this time, as Equation C4 is not
in general optimal from the social viewpoint) and multiplication with W/U to
express in proportionate terms:
x ˆ XW Ur
sUW t,a = hUc + hUg + hUE ( h EA + h EY ) + Ê s [h - hUg - hUE ( h EA + h EY )]
Ë 1- x¯
(C11)
where s ∫ (da/db)b/a and h ∫ (da/db)b/a for any a, b as before, and |t, a| indi-
ab ab
cates that t and a are being held constant. The first three terms in the RHS of
(C11) are the direct effects of an economy-wide increase in earning or produc-
tivity, including an intrinsic consumption effect hUc as the higher productivity
allows higher per capita consumption, a public-good effect hUg as higher national
income allows more spending on public goods, and an environmental disrup-
tion effect hUE(hEA + hEY). For this last effect, the environmental quality effect hUE
has to be multiplied by both the abatement effect hEA and the disruption effect
hEY because an increase in production both increases the disruption and the
abatement (through the higher tax revenue a constant share of which is used for
abatement). The last complex term in (C11) is the indirect effect through sXW,
the (proportionate) effect of earning ability W on leisure X. As everything in
(C11) is put in proportionate terms, this has to be multiplied by x/(1 - x), the
ratio of leisure to working hours. The rest of this complex term, i.e. [hUr - hUg -
hUE(hEA + hEY)], captures the external effects of individual income/leisure choice.
An increased consumption of leisure reduces one’s own income and hence ben-
efits others through the relative-income effect hUr, but harms others through a
reduction in public goods provision (hence minus hUg), and may benefit or harm
others through the environmental effect hUE depending on whether (hEA + hEY) is
negative or positive. It is perhaps not unreasonable to assume that (hEA + hEY) is
negative. Even if a given proportion (provided not excessively large to begin
with) of tax revenue is used for abatement, an increase in production still causes
more disruption to the environment.
The RHS (right-hand side) of (C11), in contrast to that of (C6), is of ambigu-
ous sign. An individual may rationally engage in the rat race for making money.
For the society, an increase in productivity may not be an unmixed blessing even
154 Appendix C
if the higher production finances more public goods provision and more
abatement.
So far, we have assumed that the tax rate t and the proportion of tax revenue
used for abatement a are being held constant. However, as productivity W
increases, the government may wish to change these ratios. Allowing t and a to
change with W, we have, instead of (C11):
t ˆ tw Uc È tw Ê a ˆ aw ˘ Ug
s UW = s UW t , a - Ê s h + s - s h + (s tw + s aw )hUE h EA
Ë 1- t ¯ ÍÎ Ë 1- a ¯ ˚˙
(C12)
where sUW|t,a| is as given in (C11).

Now, suppose that t and a have been both chosen optimally to maximise U,
with W given. Differentiating U in (C1) with respect to t, allowing endogenous
variables c, x, y, g, E to change but taking a as given (one thing at a time; this
does not really matter since, if we allow a to vary, we will just get some ad-
ditional terms which sum to zero when we take account of the first-order con-
dition for the optimal choice of a), and allowing the average values to change
with the individual values (dy = dY, etc.), we have, after substituting in (C4) and
multiplying through by t/U to express in proportionate terms:
t ˆ Uc Ê x ˆ Xt Ur
sUt = hUg + hUE hEA - Ê h + s [ h - hUg - hUE ( hEA + hEY )]
Ë1-t ¯ Ë1-x¯
(C13)
where s and h to the double superscripts are as defined under (C11). Similar to
(C11), the first three terms in the RHS of (C13) are the direct effects and the last
term is the indirect effect. While the indirect effect is completely analogous to
that in (C11) and has the same interpretation, the direct effects (of an increase
in tax rate t) consist in a public goods provision effect hUg, an (environmental
disruption) abatement effect hUEhEA (as a higher tax revenue finances for both
public goods provision and abatement), and a (reduced) consumption effect
Ê t ˆ hUc . The last effect has the term [t/(1 - t)] because of the proportionate
Ë 1- t ¯
nature of all terms in (C13). For this particular effect, it is easier to see its rationale
in non-proportionate terms. If we multiply both sides by U/t to undo the propor-
tionality, we have, showing in the RHS the reduced consumption effect only:
dU ydU
= ... ... (C13¢)
dt dc
In this form, it is clear that an increase in t reduces utility at the rate y through
the consumption effect as an increase in t reduces c at the rate y.
Similarly, differentiating U in (C1) with respect to a, we have:
a ˆ Ug
s Ua = hUE h EA - Ê h (C14)
Ë 1- a ¯
Obviously, the RHS of (C14) indicates that an increase in a entails a benefit of

increasing abatement and a cost of reducing the provision of other public goods.
Reasonably assuming continuity and interior solutions for t and a, optimal
Appendix C 155
choice of t and a entails setting (C13) and (C14) to zero. Substitute the result-
ing equations into (C12), yielding:
t ˆ 1 ˆ tW Ug
sUW t *, a * = ÏÌ1 + Ê (1 - s tW )¸˝hUc + hUE h EY + ÊË s h
Ó Ë1-t¯ ˛ 1- a¯
x ˆ XW hUg ˘
+Ê (s - s xt )ÈÍhUr - hUE h EY - (C15)
Ë 1- x¯ Î 1 - a ˙˚
where |t*,a*| indicates that the tax rate and the proportion of revenue used for
abatement are being optimised. Thus, (C15) gives, in proportionate terms, the
effect of an exogenous increase in productivity on utility, while both the tax rate
and the abatement ratio are being optimised before and after the increase.
Despite this optimisation, the RHS of (C15) is of ambiguous sign. The first term
(consumption effect) is positive (but likely to be small for rich economies), the
second term (environmental disruption effect) is negative (and likely to be
absolutely sizeable for rich economies), the third term is positive/negative if the
optimal tax rate t increases/decreases with W, and the last term is of ambiguous
sign. With likely opposing income and substitution effects, both sXW and sxt are
È hUg ˘
of ambiguous sign. The term Í hUr - hUE h EY - is itself of ambiguous sign, con-
Î 1 - a ˙˚
sisting of the first two positive parts and a negative (and enlarged absolutely, as
1/(1 - a) is larger than one) third part.
If the negative environmental disruption effect is large absolutely and both the
consumption and public goods effects (first and third terms in the RHS of
Equation C15) are relatively small, the first three terms may sum to be negative.
If, in addition, leisure does not respond substantially (as has been the case empiri-
cally; see Pencavel, 1986), the last term fails to make a big difference, leaving the
RHS of (C15) possibly negative. From the contrast between the unambiguously
positive (and likely large for both poor and rich economies alike) RHS of (C6)
and the possibly negative RHS of (C15), we may conclude
Proposition 1: Reasonably assuming the existence of important
environmental disruption effect and relative-income effect, econ-
omic growth may decrease happiness even if the provision of public
goods and disruption abatement are being optimised, and even
though all individuals are rationally engaging in the rat race for
making more money.
Proposition 1 came to us as a surprise, at least with respect to the optimal
choice of the tax rate t. At one stage, we thought that proposition is true only if
t is held constant. If t can be increased at will to shrink the private sector and
expand the provision of disruption abatement and public goods, we thought that
this should ensure the sufficiency of the utility-enhancement effect of pro-
ductivity growth. After ensuring the correctness of our mathematics, we discov-
ered the intuitive reason for the insufficiency. This is related to the fact that we
allow production Y, rather than just private consumption C to enter (negatively)
into the environmental quality function E. This means that the production of
the public sector itself is environmentally disruptive. This is a reasonable assump-
tion. However, this means that, if a higher tax rate does not discourage work sig-
nificantly (sxt not large), the instrument of income tax is powerless in reducing
the higher environmental disruption of higher production (through higher pro-
156 Appendix C
ductivity W) except through the financing of more abatement. However, if abate-

ment is not very effective in reducing disruption or in improving environmen-
tal quality (hEA not large), even optimal abatement may fail to fully offset the
higher disruption effect of more production.1 To avoid this undesirable outcome,
the direct taxation of environmental disruption may be needed.
If the provision of public goods is not environmentally disruptive, we may
revise our model by letting consumption C rather than production Y enter the
function E. Then, similar to the derivation of (C15), we may derive:
t ˆ
s UW t *, a * = ÏÌ1 + Ê (1 - s tW )¸˝( hUc + hUE h EC )
Ó Ë 1- t ¯ ˛
Ê x ˆ tW
Ë 1 - x ¯(
+ s - s )[h - h - hUE ( h EA + h EC )]
xt Ur Ug
(C15¢)
Now, even if hUc + hUEhEC is negative (i.e. the environmental disruption effect
dominates the consumption effect) and the responses of leisure negligible, (C15¢)
may still be positive by having a large enough stW. If the provision of public
goods is not environmentally disruptive, a sufficient increase in tax rate must
be sufficient to check the higher disruption effect of higher productivity and
to provide more abatement and other public goods to ensure higher utility
with growth. Of course, different public goods may have different degrees of
environmental disruption effects. Then, it means that growth may be utility-
enhancing if it finances extra provision of public goods that provide more
benefits than disruption.
The growth in governments

Using the simple model above, it may be shown that under plausible as-
sumptions, the growth in the public sector (relative to the private sector) with
the increase in productivity is required for optimality. The needed assumptions
are:
A. With higher productivity, the importance of intrinsic private consumption
effect decreases, i.e. hUc decreases with W (through a decrease in the former
with c which increases with W).
B. The environmental quality effect hUE increases with higher productivity.
C. The importance of the public-goods effect hUg and the abatement effect hEA
do not significantly decrease with higher productivity.
D. A change in the proportional tax rate t has negligible effects on the working
hours.
Actually, we only need either A or B, with the other remaining unchanged.
Assumption D is justified partly on the conceptual point that the substitution
and the income effects tend to offset each other and partly on the empirical
observation of roughly unchanged working hours for many decades despite big
increases in net wage rates.2 It should be noted that the intrinsic private good
consumption is mainly important at low consumption levels, and that environ-
mental quality and most public goods are likely to be non-inferior goods,
Assumptions A to C are also reasonable.
To analyse how the optimal relative (to total production) size of the public
sector (indicated in our model by the proportional tax rate t), set the RHS of
Appendix C 157
(C13) equal to zero (for optimal t) after substituting in sxt = 0 from Assumption
D. Totally differentiate the resulting equation, yielding:
hUc dt * t dhUc dhUg Ê dh ˆ

EA
Ê dh ˆ
UE
◊ + ◊ = + hUE + h EA (C16)
(1 - t )
2
dW 1 - t dW dW Ë dW ¯ Ë dW ¯
From Assumption A, the second term in the LHS of (C16) is negative. From
Assumption B, the last term in the RHS of (C16) is positive and from Assump-
hUc dt *
tion C, the rest are either negligible or positive. Since 2 is positive,
(1 - t ) dW
must be positive. We thus have
Proposition 2: With economic growth, the optimal size of the public
sector increases relative to the private sector.
Intuitively, with higher productivity/income, environmental quality becomes
relatively more important and the intrinsic private consumption effect less
important. Optimality then requires the use of more resources for environmen-
tal protection and the provision of other public goods.
Appendix D
Non-Affective Altruism: When the
Pareto Principle Is Unacceptable
This appendix argues that, there is a particular set of circumstances where the
Pareto principle in terms of preferences, even in its weak form (dictating a social
preference only when there is a unanimous strong preference by all individuals),
is clearly unacceptable, even in the absence of ignorance and undesirable effects
on the future or the like. The unacceptability of the Pareto principle in the
presence of altruism or the need to exclude ‘external’ preferences in social
choice have been remarked by a number of authors (e.g. Hammond, 1987/1995;
Harsanyi, 1995, p.325), but no distinction has been made between affective
versus non-affective altruism. While non-affective altruism should be excluded
in social choice, affective altruism should not.
D1 The Pareto principle

There has been some ambiguities in the interpretation of the Pareto principle.
For example, some people (e.g. Nath, 1969, p.228) interpret it to mean that a
sufficient condition for a social improvement is for the rich to get richer and the
poor remaining at the same income levels. With such a debatable interpretation,
it is obvious that the principle is not acceptable as the negative relative-income
effects on the poor may make them sufficiently worse off to more than offset
the gain of the rich. However, Pareto (1935, p.1466n) himself spoke of ‘benefit’
and ‘detriment’, not ‘richer’. Moreover, most economists interpret the principle
to refer to preference rather than to income. Thus, I shall follow this orthodoxy.
With this accepted interpretation, it seems that the principle is compelling, as
argued, for example, by Ng (1979/1983, pp.30–2). However, there is a specific set
of circumstances where the principle is unacceptable.
A natural reason for rejecting the Pareto principle is where there is important
ignorance involved in individual preferences. However, I shall ignore ignorance,
imperfect foresight, misinformation, inconsistent preferences, and the like.
Another natural reason for rejecting the Pareto principle is due to unfavourable
effects on the future and on others (including people in other countries, animals,
etc.) For example, if everyone now prefers depletion to conservation (of, say,
some essential resources), one may still not go along with the Pareto principle
for the benefits of future generations. One may argue that in this case the Pareto
principle is not really satisfied, as people in the future would be against de-
pletion now. To avoid these sort of complications, let us assume that there are
no effects on the future or on any other possibly controversial subjects like
animals. Also to avoid getting into the issue of optimal population, let us assume
158
Appendix D 159
that the set of people concerned is fixed. With these assumptions, it seems that
the Pareto principle is compelling. The discovery of a set of circumstances where
the Pareto principle is clearly unacceptable even under these assumptions is thus
remarkable.
It is possible for an individual to prefer x to y even if she will be happier at x
than at y, even without ignorance and irrationality, since the individual may
have a concern for the welfare of others, as argued in Section 4.1. In other words,
an individual may be altruistic in reason, on top of being altruistic in emotion
or affection. In emotional or affective altruism, one’s welfare (i.e. happiness)
is increased by an increase in the welfare levels of others. In ‘rational’ or non-
affective altruism, one’s utility is increased by an increase in the welfare levels
of others over and above the affective altruistic effect through one’s own welfare
level. In the presence of such non-affective altruism, it is possible that the society
should prefer y to x even if every individual prefers x to y, since social preference
should then take account of individual welfare rather than individual utility
levels to avoid multiple counting, as argued in the next section.
D2 When the weak Pareto principle is unacceptable

The existence of non-affective altruism makes the weak Pareto principle unac-
ceptable under a certain set of circumstances even in the absence of ignorance.
Since the existence of affective altruism does not cause a problem here, it is
abstracted away for simplicity. Consider a simple case of three individuals where
person 1 is an egoist with u1 = w1 and persons 2 and 3 both strong non-affective
altruists with respect to the welfare of person 1, with ui = wi + 2/3w1; i = 2, 3, where
u = utility (representing preference) and w = welfare (representing happiness)
and a superscript indicates the person concerned. It should be noted that the
existence of a reasonable social welfare function or the general possibility of a
reasonable social choice presupposes interpersonal comparable cardinal utilities,
as argued in Chapter 2. It is difficult to obtain individual cardinal utilities/
welfares and to compare them interpersonally. However, to concentrate on the
main point, let us abstract from these difficulties and assume the existence of
interpersonal comparable cardinal individual welfares and utilities. The non-
acceptability of the Pareto principle may be shown even under this simplifi-
cation. Consider the choice between social alternatives x and y for a situation
depicted in Table D1, where the welfare and utility figures may be taken as being
in interpersonally comparable cardinal units.
It is clear that every person prefers y to x, i.e. ui(y) > ui(x) for all i. According
to even the weak Pareto principle, the society should prefer y to x. However, I
Table D1
w1 w2 w3 u1 = w1 u2 = w2 + 2/3w u3 = w3 + 2/3w Sw i Sui
x 6 12 18 6 16 22 36 44
y 18 5 12 18 17 24 35 59
160 Appendix D
wish to argue that the weak (and a fortiori the strong) Pareto principle is not
acceptable in such circumstances and that the society should prefer x to y.
We should also abstract away the complications of indirect or long-term effects
by assuming that the individual welfares and utilities figures for x and y in the
above figures capture all relevant outcomes and that there are no other effects.
If we focus attention on individual welfares only, it is clear that x is preferable
to y. Not only is the sum of individual welfares higher in x than in y, all social
welfare functions of individual welfares that is anonymous and increasing
in individual welfares must obviously rank x higher than y. However, should
the social welfare function be a function of individual welfares or of individual
utilities?
If the individual welfare figures in Table D1 already capture all relevant effects
as mentioned above and if the welfare levels of all other sentients are abstracted
away or assumed unchanged, Section 4.1 argues that social choice should be
based on individual welfares than on individual utilities. Here, a more com-
pelling and simple case can be made due to the fact that individual welfares and
utilities differ here because of only a specific reason by construction.
In the example of Table D1, misinformation, ignorance, and all other reasons
that may cause the utility (representing preference) of a person to differ from her
welfare have been abstracted away, except non-affective altruism. Person 1 is
assumed to be not altruistic non-affectively so that his utility does not differ from
his welfare. Persons 2 and 3 are both assumed to be strongly non-affectively altru-
istic towards person 1. (We do not need such a high degree of non-affective altru-
ism to establish the point if we have more individuals.) Each of them has a higher
welfare level at x than at y but prefers y to x because of a non-affective concern
for the welfare of person 1. However, for social choice, since the welfare of person
1 is already fully taken into account as w1 in any reasonable (including the prop-
erty of anonymity) social welfare function, it should not be taken into account
again as components of u2 and u3. Otherwise, multiple counting is involved. To
see this point more clearly, consider a simpler case of just two persons depicted
in Table D2.
For this case, it is obvious that, taking account of the welfare levels of the indi-
viduals, any anonymous social welfare function increasing in individual welfares
ranks x above y. However, if the utility figures are taken as relevant for social
choice instead, any anonymous and non-anti-egalitarian social welfare function
(such as W = Sui) prefers y to x, as y has a more equal utility profile with a bigger
sum than x. Nevertheless, the social preference of y to x is patently unfair to
person 2. The welfare of person 1 is only counted once while that of person 2 is
Table D2
w1 w2 u1 = w1 u2 = w2 + w1 Sw i Sui
x 2 7 2 9 9 11
y 7 1 7 8 8 15
Appendix D 161
counted twice, once as the component w1 in u1, and once as the component w1
in u2, because person 2 has a perfect concern for the welfare of person 1. If she
were to make the choice herself, she would choose in accordance to w2 + w1. If
person 1 had a similar perfect concern for the welfare of person 2, the use of Sui
would rank x above y. Due to the asymmetrical degrees of non-affective altru-
ism, the use of utility figures instead of welfare figures biases social choice in
favour of persons of lower degrees of non-affective altruism, effectively penalis-
ing persons with higher degrees of non-affective altruism. The example of Table
D2 shows clearly that, where the difference between utility and welfare levels is
due only to non-affective altruism, social choice should be based on individual
welfare rather than utility figures.
Coming back to Table D1, the agreement to the use of individual welfare figures
for social choice means that any anonymous and increasing SWF will rank x
above y, despite the fact that every individual prefers y to x. As the society should
prefer x to y even when every individual prefers y to x, the Pareto principle, even
in its weak and hence more compelling version, is not acceptable in this set of
circumstances.
D3 The case of affective altruism

It should be noted that, while non-affective altruism may cause the Pareto
principle to be unacceptable, affective altruism does not. In this case, the
altruists also feel better off by, say, seeing the higher incomes of those they are
altruistically minding for. Thus, the welfares of the altruists are also higher. There
is thus no ground to reject the Pareto principle due to this. To see the point
clearer, consider the case illustrated in Table D3 where mi is the (real) money or
income of person i. Non-affective altruism and ignorance are both assumed
absent.
Here, person 2 is an affective altruist who is happier with her income reduced
from 6 at x to 4 at y as she, say, thinks that person 2 has more need to get more
income and feel good knowing that he gets a higher income of 8 at y. Both
persons prefer y to x and are happier at y than at x. Assuming that Table D3
already includes all relevant effects as should be, there is no ground for the
society not to prefer y to x. The Pareto principle should be respected in this case.
The fact that x has a more equal distribution of income is not a valid reason for
preferring x to y socially. This is so because income is not an ultimate end. Hence,
equal distribution of income is only desirable if it increases our ultimate end –
happiness. In the case of Table D3, the equal distribution at x (in comparison to
Table D3
m1 m2 w1 w2 u1 u2
x 6 6 6 6 6 6
y 8 4 7 7 7 7
162 Appendix D
y) decreases the welfare of all individuals. It may be argued that unequal distri-
bution may have undesirable long-term consequences. If so, that should be
included into the welfare figures in Table D3. Then, we can judge according to
the figures inclusive of all effects.
D4 Should voters be asked to suppress their

non-affective altruism?
It has been shown that even the weak Pareto principle may not be acceptable in
the presence of non-affective altruism which is strongly argued to exist. This
raises some perplexing questions. For one thing, it is generally believed that altru-
ism increases social welfare and improves the social decisions undertaken. This
may certainly be true for decision-making by key persons with altruism. However,
for the processes of revealing individual preferences which are then used as the
information ingredients for making social decisions, non-affective altruism may
be counter-productive. For example, in the case of Table D1, if non-affective
altruism is excluded from individual preferences, the correct social ranking of x
over y is likely to be made. On the other hand, if individual preferences inclu-
sive of non-affective altruism are used, the incorrect social ranking of y over x is
likely. Thus, despite the general beneficial effects of altruism, it may be desirable
to request voters to vote in accordance to their welfare, or their preferences exclu-
sive of non-affective altruism. However, such a practice may decrease the degrees
of altruism in the long run and that may be detrimental to social welfare. Hence,
a kind of a dilemma is present.
Appendix E
The Bergson–Samuelson Tradition
Implies Individualism,
Independence, and Ordinalism
By the ‘Bergson–Samuelson tradition’ we do not mean the general formulation

of a SWF in Bergson (1938) which, in the words of Osborne (1976, p.1003)
‘in some mysterious fashion maps S (social states) directly into social welfare’.
Rather, we refer to the widely used version introduced by Bergson under the label
‘fundamental value propositions of individual preference’. According to that
version, social welfare derived from any social state is a function only of indi-
vidual preferences in that social state:
W ( x ) = W [u1 ( x ),u 2 ( x ), . . . u n ( x )] (E1)
where ui is the utility function representing the ordinal preferences of individ-
ual i, and x is a social state.
While this formulation is in terms of social welfare, it may also be regarded as
a social choice formulation if it is agreed that the social ranking of social states
is to be in accordance with the value of social welfare, i.e. if xRy < = > W(x) ≥
W(y), where R indicates weak social preference.
That the Bergson–Samuelson tradition as represented in (E1) implies indi-
vidualism is obvious, since social welfare depends only on individual utilities.
That it implies independence is also not difficult to see, since W(x) depends only
on individual utilities at the social state x, and this is so for any x.
It cannot be inferred from (E1) alone that the Bergson–Samuelson tradition
relies on ordinalism for the ui could represent cardinal utilities. Rather, ordinal-
ism is imputed from statements made by Bergson and Samuelson. Bergson (1938)
himself writes: ‘In my opinion the [cardinal] utility calculus introduced by the
Cambridge economists [referring to Marshall, Pigou, etc.] is not a useful tool for
welfare economics’ (as reprinted in Arrow and Scitovsky, 1969, p.20). And inter-
preting Bergson, Arrow writes: ‘Bergson considers it possible to establish an order-
ing of social states which is based on indifference maps of individuals, and
Samuelson has agreed’ (Arrow, 1951/1963, p.5, emphasis added). Samuelson
is even more forthright than Bergson. He repeatedly asserts that a Bergson–
Samuelson SWF ‘does exist. . . . No cardinal intensities are ever involved’
(Samuelson, 1977, p.86). If we interpret the ui in (E1) as purely ordinal, the
Bergson–Samuelson tradition thus implies ordinalism as well.
163
Appendix F
Inefficiency in Provision May
Increase Optimal Public Spending
Ignoring the possible interaction between benefit and cost, we may choose g to
maximise net benefit N:
N = B[a( g )] - C( g ) (8.1)
where B = (total) benefit, a = actual or physical amount of public good provided,
g = monetary amount of public spending, C = (total) cost. The use of this general
cost function allows for possible excess burden in financing for g:
a( g ) = Ag (8.2)
where A is an index on the efficiency in the public provision of public goods.
Assuming the satisfaction of the usual conditions (differentiability of the func-
tions, B¢ and C¢ > 0, B≤ < 0 and C≤ > 0), the maximisation of (8.1) with respect
to g, using (8.2) gives:
B¢A = C¢ (8.3)
Totally differentiating (8.3) with respect to A, we have, after using the total
differentiation of (8.2):
dg * dA = ( B¢ + aB¢¢ ) (C ¢¢ - A 2 B¢¢ ) (8.4)
Since the denominator of the RHS is positive, the sign of dg*/dA is the same as
that of the numerator. As (B¢ + aB≤) is positive/negative if and only if B¢ >/< -aB≤
or 1>/< - (dB¢/da)a/B¢, we have Proposition 8.1 in the text. To show that the (pro-
portionate) responsiveness of marginal benefit is the same whether we use the
monetary level of public spending g or the physical level of public good a (given
that the provision efficiency A is independent of g), note that, from the B func-
tion in (8.1) and from (8.2) we have Bg ∫ dB/dg = B¢A from which we have, dBg/dg
= B≤A2, and hence (dBg/dg)g/Bg = (dB¢/da)a/B¢. Thus, the marginal benefit is equally
responsive proportionately either with respect to the monetary amount of public
spending g or to the physical amount of public good a.
164
Notes
Chapter 3
1. A different welfarist reason for including such apparently ‘non-welfarist’ indi-
cators into the short-run SWF is their effects on welfare in the long run. For
example, a less unequal distribution of income may promote social harmony.
2. See also Sen, 1967. I define basicness somewhat differently: ‘a value judge-
ment [is] basic to a person if it is not derived from some other value judge-
ment and he believes in it for its own ethical appeal’ (Ng, 1979/1983, p.19).
Whether this definition differs from Sen’s depends on the interpretation of
‘all conceivable circumstances’ in Sen’s definition.
3. As an example of the importance of this practical consideration, see Creedy
(1996) who argues that the maximisation of the utilitarian objective of the
unweighted sum of individual welfares in the long run may be close to, in
the short run, the pursuit of Rawlsian maximin or the use of target efficiency
in poverty reduction.
4. However, perception and affective feelings probably evolved simultaneously.
As I hope to argue elsewhere, perception without affective feelings also serves
no evolutionary purposes. Here, perception is defined as involving subjective
consciousness, not just in the wider sense ‘reception and interpretation of
signals from the environment’, regarded as evidenced in the bacterium E. coli,
see M. Delbrück, 1986.
5. This leaves a little vague as to whether non-affective malice is rational or ir-
rational. I prefer to regard non-affective malice as irrational. Since both ir-
rationality and non-affective altruism/malice are excluded by Axiom 1, this
vagueness does not affect the argument below.
Chapter 4
1. This chapter is largely based on Ng (1999).
2. In this simple formulation of the utility function, it may appear that some
subtle differences such as whether the higher welfare levels of others are
caused by one’s own actions or not (Sen’s ‘agency’ problem) are not distin-
guished’; however, certain relevant activities such as one’s own contribution
to charity may be allowed to enter one’s welfare function in the text below.
Chapter 5
1. Fleming only lists postulates ensuring the existence of a social ordering. With
continuity, this ensures the existence of a SWF as well. Continuity and the
existence of a SWF is implicitly assumed in Fleming’s derivation. The exis-
tence of a social ordering alone is not sufficient for Fleming’s purpose as a
lexicographic social ordering cannot be represented by a SWF at all.
165
166 Notes
2. Or, rather, if they had an equal chance of being ‘put in the place of’ any indi-
vidual member of the society, with regard not only to his objective social (and
economic) conditions, but also to their subjective attitudes and tastes. In
other words, they ought to judge the utility of another individual’s position
not in terms of his own attitudes and tastes but rather in terms of the atti-
tudes and tastes of the individual actually holding this position (Harsanyi’s
note).
3. This proviso is redundant if we accept Rational Individualism (Axiom 1)
which implies welfarism.
4. This has to be so as non-welfare variables are being held constant; also, non-
affective altruism/malice is to be excluded in social choice from Axiom 1d2.
5. From the continuity of preference, one should really use one to measure the
welfare difference of a just imperceptible indifference, and anything larger
than one will be perceptible preference (Ng, 1975a). However, since a just
imperceptible indifference is continuous with a just perceptible preference,
we will ignore this pedantic point.
6. The difference between individual preferences (used by Fleming) and indi-
vidual welfares is ignored. This does not affect our argument here since
Rational Individualism insists on the use of individual welfares where the two
diverge.
7. If WE and EW always yield the same result, we have WE = W(SjqjUj1, . . . ,
SjqjUjI ) = EW= SjqjW(Uj1, . . . , UjI ). Differentiation with respect to Uji gives
∂W/∂EUi = ∂W/∂Uji at any given situation. Such an equality cannot hold
through for every set of q’s unless W is a sum (unweighted or weighted with
uniform individual weights) of its arguments.
Chapter 6
1. In my 1979/83 treatment, I did not discuss the big differences between will-
ingness to pay and willingness to accept due to the unwillingness of many
people to actually pay for something they believe they are ‘entitled’ to or
other similar problems. (See, for example, Mitchell and Carson, 1989;
Milgrom, 1993.) For such cases, there are added practical difficulties to dis-
cover people’s real willingness to pay as a reflection of their preferences for
the items as such untainted by the objections or ill-feelings associated with
payment. This difficulty of the intertwining of two different types of prefer-
ences/dispreferences is similar to the intrusion of factors such as regret,
anxiety, excitement, etc. into the process or outcomes of choices involving
risk, making the application of the expected utility theory difficult in those
cases. However, in both cases, the principles involved are not invalidated.
2. As real tax liabilities depend on non-income characteristics such as age,
marital status, home-ownership, etc. the evaluation of effective income tax
progressivity may not be simple. However, see Hayes et al. (1995) for an algo-
rithm to evaluate effective income tax progression.
3. Second-best taxation-pricing rules are typically very complicated, even if only
the efficiency consideration is taken into account. For the literature on
optimal taxation, see Mirrlees (1976, 1981) and Sandmo (1976). Conditions
making second-best considerations ineffective are rather stringent, for
example, separability in the utility function (see Atkinson and Stiglitz, 1976;
compare Bergstrom and Cornes, 1983).
Notes 167
4. For example, consider the argument of William Baumol and Dietrich Fischer
(1979) that the use of discrimination in wage rate is much more efficient (less
output foregone) than non-discriminatory taxation in achieving equality.
This is based on the detailed knowledge of individual input supply functions
that they recognise to be unavailable. But they suggest that some rough dis-
crimination between broad groups of income earners such as doctors versus
ditch diggers may yet be feasible. But if a discrimination, such as supple-
menting ‘wage rates for the one and limit[ing] wages for the other’ (p.522),
is feasible, there is little reason to expect that it is not feasible to have higher
income tax rates for doctors and lower for ditch diggers.
5. Borghans and Groot (1998) attribute the discrepancy/inefficiency to some
form of monopolistic power. However, I regard the insignificant losses of the
replacement of the superstars by their runners up as more important, making
the discrepancy/inefficiency non-existent or insignificant for superstars in say
science and technology or the generation of real knowledge in general. Here,
in fact, the superstars earn far less than their marginal contributions due to
the global public-good and long-term nature of knowledge.
Chapter 9
1. Political-economy issues like rent-seeking and tax policy in accordance to
political instead of distortionary costs (on which see, e.g. Brennan, 1984;
Buchanan, 1993; Hettich and Winer, 1997; Holcombe, 1998 and Winer and
Hettich, 1998), while relevant, are not considered here.
2. Batina (1990) in fact allows for heterogeneous individuals and hence ident-
ifies a distributional term but does not relate it to the offsetting of the disin-
centive effects.
3. Combined with this consideration is the point that, while the existence and
size of excess burdens or distortions depend on the net substitution effects,
the marginal cost of fund depends on the gross substitution effect (disincen-
tive effect when related to the income/leisure choice). See Ng, forthcoming
b, (second half of its Appendix A in particular); cf. Browning and Liu (1998).
4. Gross of the income effect, a higher tax may not cause disincentive effects.
Appendix B
1. It is true that Rawls would argue that freedom to rape is not a basic liberty
while the right to non-violation of body is. However, how do we determine
what are basic liberties? Either it is based on the utilitarian (or at least wel-
farist) principle or it is open to the objection of possibly sanctioning a disas-
trous outcome. For the unacceptability of rights-based ethics, see Ng (1990a,
Sec. IIIB).
2. Including holding the degree of altruism/malice the same across individuals;
for differences in this, see the text below.
Appendix C
1. If hEA is large, then the optimality condition for the division between abate-
ment and other public goods (from the equating of the RHS of Equation C14
168 Notes
to zero) dictates a large hUg, making the third term of the RHS of (C15)
dominating.
2. It may also be noted that the term associated with sxt in (C13) contains an
expression (reflecting the indirect effects) of ambiguous sign whose parts are
likely to change in offsetting directions with W. Hence, even if sxt is not
assumed negligible, the qualification to our conclusion below is likely to be
small.
References
Ackerman, F., Kiron, D., Goodwin, N. R., Harris, J. M. and Gallagher, K. (1997),
Human Well-being and Economic Goals (Washington, DC: Island Press).
Adler, Matthew D. and Posner, Eric A. (1999), ‘Rethinking Cost–Benefit Analy-
sis’, Chicago Law & Economics Working Paper No. 72 (2D series).
Aghion, Philippe and Williamson, Jeffrey (1999), Growth, Inequality and Global-
ization: Theory, History and Policy (Cambridge: Cambridge University Press).
Aiken, William and La Follette, Hugh (eds) (1977), World Hunger and Moral
Obligation (Englewood Cliffs, NJ: Prentice-Hall).
Ake, Christopher (1975), ‘Justice as equality’, Philosophy and Public Affairs, 1,
69–89.
Altonji, J. G., Hayashi, F. and Kotlikoff, L. J. (1997), ‘Parental altruism and inter
vivos transfers: theory and evidence’, Journal of Political Economy, 105, 1121–66.
Arneson, Richard J. (1989/1995), ‘Equality and equal opportunity for welfare’,
Philosophical Studies, 56, 77–93; reprinted with a postscript (1995) in Pojman
and Westmoreland (eds), (1999). pp.229–42.
Arrow, Kenneth J. (1951/1963), Social Choice and Individual Values (New York:
John Wiley).
Arrow, Kenneth J. and Intriligator, Michael D. (1986), Handbook of Mathematical
Economics, Vol. III (Amsterdam: North-Holland).
Arrow, Kenneth J. and Scitovsky, Tibor (eds) (1969), Readings in Welfare Econ-
omics (London: Allen & Unwin).
Ashenfelter, Orley C. and Layard, Richard (eds) (1986), Handbook of Labor Econ-
omics, Vol. 1 (Amsterdam: North-Holland).
Atkinson, Anthony B. and Stiglitz, Joseph E. (1976), ‘The design of tax structure:
direct versus indirect taxation’, Journal of Public Economics, 6, 55–75.
Atkinson, Anthony B. and Stiglitz, Joseph E. (1980), Lectures on Public Economics,
(New York: McGraw-Hill).
Auten, Gerald and Carroll, Robert (1994), ‘Taxpayer behavior and the 1986 Tax
Reform Act’, Office of Tax Analysis Working Paper (Washington, DC, US
Department of the Treasury).
Backhouse, Roger E. (1998), ‘If mathematics is informal, then perhaps we should
accept that economics must be informal too’, Economic Journal, 108, 1848–
58.
Bagwell, L. S. and Bernheim, B. D. (1996), ‘Veblen effects in a theory of con-
spicuous consumption’, American Economic Review, 86, 349–73.
Baland, Jean-Marie and Platteau, Jean-Philippe (1997), ‘Wealth inequality and
efficiency in the commons’, Part I, Oxford Economic Papers, 49, 451–82.
Ballard, Charles L. and Medema, Steven G. (1993), ‘The marginal efficiency effect
of taxes and subsidies in the presence of externalities: a computational general
equilibrium approach’, Journal of Public Economics, 52, 199–216.
Ballard, Charles, Shoven, John and Whalley, John (1985), ‘General equilibrium
computations of the marginal welfare cost of taxes in the United States’,
American Economic Review, 75 (1), 128–38.
169
170 References
Barry, Brian M. (1995), Justice as Impartiality (Oxford, England: Clarendon Press;

New York: Oxford University Press).
Batina, Raymond G. (1990), ‘On the interpretation of the modified Samuelson
rule for public goods in static models with heterogeneity’, Journal of Public Econ-
omics, 42, 125–33.
Baumol, William J. (1967), ‘Macroeconomics of unbalanced growth: the anatomy
of urban crisis’, The American Economic Review, 57, 415–26.
Baumol, William J. and Fischer, Dietrich (1979), ‘The output distribution
frontier: alternatives to income taxes and transfers for strong equality goals’,
American Economic Review, Sept., 69, 514–25.
Behrman, J. and Srinivasan, T. N. (eds) (1995), Handbook of Development
Economics, Vol. 3A (Amsterdam: North-Holland).
Bénabou, Roland (1996), ‘Inequality and growth’, NBER Macroeconomics Annual,
No. 1996, pp.11–92.
Bergson (Burk), A. (1938), ‘A reformulation of certain aspects of welfare econ-
omics’, Quarterly Journal of Economics, 52, 310–34.
Bergstrom, Theodore C. and Cornes, Richard C. (1983), ‘Independence of allo-
cative efficiency from distribution in the theory of public goods’, Econometrica,
51, 753–66.
Berridge, K. C. (1996), ‘Food reward: brain substrates of wanting and liking’,
Neuroscience and Biobehavioral Reviews, 20, 1–25.
Binswanger, Hans, Deininger, Klaus and Feder, Gershon (1995), ‘Power, distor-
tions, revolt and reform in agriculture land revolutions’, in Behrman and
Srinivasan, Vol. 3, chapter 42, pp.2659–772.
Birdsall, Nancy, Ross, David and Sabot, Richard (1995), ‘Inequality and growth
reconsidered’, World Bank Economics Review, 9, 477–508.
Bishop, M. P., Elder, S. Thomas and Heath, Robert G. (1964), ‘Attempted control
of operant behaviour in man with intracranial self-stimulation’, in Heath
(1964), pp.55–81.
Blackorby, Charles and Donaldson, David (1990), ‘A review article: the case
against the use of the sum of compensating variations in cost–benefit analy-
sis’, Canadian Journal of Economics, 23 (3), 471–94.
Blaug, Mark (1998), ‘Disturbing currents in modern economics’, Challenge, 41
(3), 11–45. (Followed by an interview).
Boadway, Robin W. (1974), ‘The welfare foundations of cost–benefit analysis’,
Economic Journal, 84, 926–39.
Boadway, Robin W. and Keen, Michael (1993), ‘Public goods, self-selection and
optimal income taxation’, International Economic Review, 34 (3), 463–78.
Bohl, M. T. (1996), ‘Some international evidence on Wagner’s Law’, Public
Finance, 51 (2), 185–200.
Borghans, Lex and Groot, Loek (1998), ‘Superstardom and monopolistic power:
why media stars earn more than their marginal contribution to welfare’, Journal
of Institutional and Theoretical Economics, 154, 546–71.
Bowles, Samuel (1998), ‘Endogenous preferences: the cultural consequences of
markets and other economic institutions’, Journal of Economic Literature, 36,
75–111.
Bowles, Samuel and Gintis, Herbert (1996), ‘Efficient redistribution: new rules for
markets, states and communities’, Politics and Society, 24, 307–42.
Breen, Richard (1997), ‘Inequality, economic growth and social mobility’, British
Journal of Sociology, 48, 429–49.
References 171
Brennan, Geoffrey (1984), ‘Elements of a fiscal politics: public choice and public
finance’, Australian Economic Review, 3rd quarter, 67, 62–72.
Brickman, Phillip, Coates, Dan and Janoff-Bulman, Ronnie (1978), ‘Lottery
winners and accident victims: is happiness relative?’, Journal of Personality and
Social Psychology, 36, 917–27.
Broome, John (1991), Weighing Goods: Equality, Uncertainty and Time (Oxford:
Blackwell).
Brown, Donald and Heal, Geoffrey (1979), ‘Equity, efficiency and increasing
returns’, Review of Economic Studies, 571–86.
Browning, Edgar K. (1987), ‘On the marginal welfare cost of taxation’, American
Economic Review, 77 (1), 11–23.
Browning, Edgar K. and Liu, Ligun (1998), ‘The optimal supply of public goods
and the distortionary cost of taxation: Comment’, National Tax Journal, 51 (1),
103–16.
Buchanan, James A. (1993), ‘The political efficiency of general taxation’, National
Tax Journal, 46 (4), 401–10.
Chang, Ming Chung (1997), ‘Optimal public good provision – cases with
complementarities between the public good and the taxed commodities’,
Typescript.
Chick, Victoria (1998), ‘On knowing one’s place: the role of formalism in econ-
omics’, Economic Journal, 108, 1859–69.
Christiansen, Vidar (1981), ‘Evaluation of public projects under optimal
taxation’, Review of Economic Studies, XLVIII, 447–57.
Clark, J. (1998), ‘Fairness preferences and optimization skills: are they substitutes?
An experimental investigation’, Journal of Economic Behavior and Organization,
34 (4), 541–57.
Cohen, G. A. (1989), ‘On the currency of egalitarian justice’, Ethics, 99, 906–44.
Cole, Harold L., Mailath, George J. and Postlewaite, Andrew (1992), ‘Social
norms, savings behavior, and growth’, Journal of Political Economy, 100 (6),
1092–125.
Congleton, R. D. (1997), ‘Political efficiency and equal protection of the law’,
Kyklos, 50, 485–505.
Corden, W. M. (1974), Trade Policy and Economic Welfare (London: Oxford
University Press).
Creedy, John (1996), ‘Comparing tax and transfer systems: poverty, inequality
and target efficiency’, Economica, 63, S163–S174.
Creedy, John and McDonald, Ian M. (1990), ‘A tax package to reduce the mar-
ginal rate of income tax and wage demands of trade unions’, Economic Review,
66, 195–202.
Cremer, Helmuth and Gahvari, Firouz (1995), ‘Uncertainty, optimal taxation
and the direct versus indirect tax controversy’, Economic Journal, 105, 1165–
79.
Cummins, Robert A. (1998), ‘The second approximation to an international
standard for life satisfaction’, Social Indicators Research, 43, 307–44.
D’Aspremont, Claude and Gevers, Louis (1977), ‘Equity and the informational
basis of collective choice’, Review of Economic Studies, 44, 199–209.
David, P. A. and Reder, M. W. (eds) (1974), Nations and Households in Economic
Growth: Essays in Honour of Moses Abramowitz (New York: Academic Press).
de Bartolome, Charles A. M. (1995), ‘Which tax rate do people use: average or
marginal?’, Journal of Public Economics, 56, 79–96.
172 References
Delbrück, Max (1986), Mind from Matter? An Essay on Evolutionary Epistemology,

edited by Gunther S. Stent, Ernst Peter Fischer, Solomon W. Golomb,
David Presti, Hansjakob Seiler (Palo Alto, California: Blackwell Scientific
Publications).
Delgado, J. M. R. (1976), ‘New orientations in brain stimulation in man’, in
Wauquier and Rolls (1976).
DeMeyer, Frank and Plott, Charles R. (1971), ‘A welfare function using “relative
intensity” of preference’, Quarterly Journal of Economics, 85, 179–86.
Devins, Neal and Douglas, Davison M. (1998), Redefining Equality (New York and
Oxford: Oxford University Press).
Devinsky, O., Berić, A. and Dogali, M. (1993), Electrical and Magnetic Stimulation
of the Brain and Spinal Cord (New York: Raven Press).
Diamond, Peter A. (1967), ‘Cardinal welfare, individualistic ethics, and interper-
sonal comparison of utility: comment’, Journal of Political Economy, 75,
765–6.
Diener, Ed and Suh, Eunkook (1997), ‘Measuring quality of life: economic, social
and subjective indicators’, Social Indicators Research, 40, 189–216.
Dowrick, Steve (1993), ‘Government consumption: its effects on productivity,
growth, and investment’, in Gemmell (1993), pp.136–152.
Duesenberry, James S. (1949), Income, Saving and the Theory of Consumer Behav-
iour (Cambridge, MA: Harvard University Press).
Dworkin, R. (1977), Taking Rights Seriously (Cambridge, Mass.: Harvard Univer-
sity Press).
Dworkin, Ronald (1981), ‘What is equality?’ Philosophy and Public Affairs, 10:
185–246, 283–345.
Easterlin, R. A. (1974), ‘Does economic growth improve the human lot? Some
empirical evidence’, in David and Reder, pp.89–125.
Easterly, William (1997), ‘Life during growth’. World Bank typescript. Revised
(1999) version on web page http//www.worldbank.org/html/prdmg/grthweb/
growth t.htm.
Eaton, B. Curtis and White, William (1991), ‘The distribution of wealth and the
efficiency of institutions’, Economic Inquiry, 29, 336–50.
Eatwell, John, Milgate, Murray and Newman, Peter (eds) (1987), The New Pal-
grave: A Dictionary of Economics, Vol. 1 (London: Macmillan).
Edgeworth, F. Y. (1881), Mathematical Psychics (London: Kegan Paul).
Eisenstein, Louis (1961), The Ideologies of Taxation (New York: Ronald).
Elster, Jon (1998), ‘Emotions and economic theory’, Journal of Economic Litera-
ture, 36, 47–74.
Elster, Jon and Roemer, John E. (eds) (1991), Interpersonal Comparisons of
Well-Being (London: Cambridge University Press).
Eshel, I., Samuelson, L. and Shaked, A. (1998), ‘Altruists, egoists, and hooligans
in a local interaction model’, American Economic Review, 88 (1), 157–79.
Farina, Francesco, Hahn, Frank and Vannucci, Stefano (eds) (1996), Ethics,
Rationality, and Economic Behaviour (New York: Clarendon Press).
Feiwel. G. R. (ed.) (1985), Issues in Contemporary Microeconomics and Welfare
(London: Macmillan).
Feldstein, Martin (1974), ‘Distributional preferences in public expenditure analy-
sis’, in Hochman and Peterson, pp.136–61.
Feldstein, Martin (1995a), ‘The effect of marginal tax rates on taxable income: a
References 173
panel study of the 1986 Tax Reform Act’, Journal of Political Economy, 103 (3),
551–72.
Feldstein, Martin (1995b), ‘Tax avoidance and the deadweight loss of the income
tax’, NBER Working Paper No.5055 (Cambridge, MA: National Bureau of Econ-
omic Research).
Feldstein, Martin (1997), ‘How big should government be?’, National Tax Journal,
50 (2), 197–213.
Feldstein, Martin and Feenberg, Daniel (1996), ‘The effect of increased tax rates
on taxable income and economic efficiency: a preliminary analysis of the 1993
tax rate increases’, in Tax Policy and the Economy, Vol. 10, pp.89–117.
Ferris, Stephen J. and West, Edwin G. (1996), ‘Testing theories of real govern-
ment size: U.S. experience, 1959–89’, Southern Economic Journal, 62 (3), 537–53.
Fleming, Marcus (1952), ‘A cardinal concept of welfare’, Quarterly Journal of
Economics, 66, 366–84.
Fortin, Bernard and Lacroix, Guy (1994), ‘Labour supply, tax evasion and the
marginal cost of public funds: an empirical investigation’, Journal of Public
Frank, Robert H. (1985), Choosing the Right Pond: Human Behaviour and the Quest
for Status (New York: Oxford University Press).
Frank, Robert H. (1987), ‘If homo economicus could choose his own utility func-
tion, would he want one with a conscience?’, American Economic Review, 77,
593–604.
Frank, Robert H. (1999), Luxury Fever: Why Money Fails to Satisfy in an Era of
Excess (New York, NY: Free Press).
Gemmell, Norman (1993), The Growth of the Public Sector: Theories and Inter-
national Evidence (Aldershot, England: Edward Elgar).
Goodin, Robert E. (1995), Utilitarianism as a Public Philosophy (Cambridge, New
York: Cambridge University Press).
Grossman, Herschel (1994), ‘Production, appropriation and land reform’, Amer-
ican Economic Review, 84 (3), 705–12.
Hamilton, W. D. (1964), ‘The genetical evolution of social behaviour. I and II’,
Journal of Theoretical Biology, 7, 1–52.
Hammond, Peter J. (1976), ‘Why ethical measures of inequality need interper-
sonal comparisons’, Theory and Decision, 7, 262–274.
Hammond, Peter J. (1987/1995), ‘Altruism’, in Eatwell et al., Vol. 1, pp.85–7,
reprinted in Stefano Zamagni (1995), pp.165–7.
Hammond, Peter J. (1991), ‘Interpersonal comparisons of utility: why and how
they are and should be made’, in Elster and Roemer, (eds), pp.200–54.
Hammond, Peter J. (1996), ‘Consequentialist decision theory and utilitarian
ethics’, in Farina, Hahn and Vannucci (eds).
Harbaugh, W. T. (1998), ‘What do donations buy? a model of philanthropy based
on prestige and warm glow’, Journal of Public Economics, 67 (2), 269–84.
Hardy, J. D., Wolff, Harold G. and Goodell, Helen (1952), Pain Sensations and
Reactions (Baltimore: The Williams and Wilkins Co.).
Hare, R. M. (1976), ‘Ethical theory and utilitarianism’, in Lewis (ed.), pp.113–31.
Hare, R. M. (1997), ‘Justice and equality’, in Pojman and Westmoreland’s
pp.218–28.
Harris, Richard and Olewiler, Nancy (1979), ‘The welfare economics of ex post
optimality’, Economica, 46, 137–47.
174 References
Harrod, R. F. (1948), Towards a Dynamic Economics (London: Macmillan).

Harsanyi, John C. (1953), ‘Cardinal utility in welfare economics and in the theory
of risk-taking’, Journal of Political Economy, 61, 434–5.
Harsanyi, John C. (1955), ‘Cardinal welfare, individualistic ethics, and inter-
personal comparison of utility’, Journal of Political Economy, 63, 309–21.
Harsanyi, John C. (1995), ‘A theory of prudential values and a rule utilitarian
theory of morality’, Social Choice and Welfare, 12, 319–33.
Harsanyi, John C. (1997), ‘Utilities, preferences, and substantive goods’, Social
Choice and Welfare, 14, 129–45.
Hausman, Jerry A. (ed.) (1993), Contingent Valuation: A Critical Assessment
(Amsterdam: North-Holland).
Haveman, Robert H. (1988), Starting Even: An Equal Opportunity Program to Combat
the Nation’s New Poverty (New York: Simon and Schuster).
Hayes, Kathy J., Lambert, Peter J. and Slottje, Daniel J. (1995), ‘Evaluating effec-
tive income tax progression’, Journal of Public Economics, 56, 461–74.
Headley, Bruce and Wearing, Alexander (1991), ‘Subjective well-being: a stocks
and flows framework’, in Strack, Argyle and Schwarz, pp.49–73.
Heath, R. G. (ed.) (1964), The Role of Pleasure in Behavior (New York: Harper and
Row).
Heath, R. G., John, S. B. and Fontana, C. J. (1968), ‘The pleasure response: studies
by stereotaxic technics in patients’, in Kline and Laska, pp.178–89.
Herrnstein, Richard J. and Prelec, Drazen (1992), ‘Melioration’ in Loewenstein,
George and Elster, Jon, pp.235–63.
Hersoug, Tor (1984), ‘Union wage responses to tax changes’, Oxford Economic
Paper, 36, 37–51.
Hettich, Walter and Winer, Stanley L. (1997), ‘The political economy of tax-
ation’, in Dennis Mueller (1989), pp.481–505.
Hirsch, F. (1976), Social Limits to Economic Growth (Cambridge, MA: Harvard
University Press).
Hochman, Harold M. and Peterson, George E. (eds) (1974), Redistribution Through
Public Choice (New York: Columbia University Press).
Hoff, Karla (1996), ‘Market failures and the distribution of wealth: a perspective
from the economics of information’, Politics & Society, 24, 411–32.
Hoff, Karla and Lyon, Andrew (1995). ‘Non-leaky buckets: optimal redistributive
taxation and agency costs’, Journal of Public Economics, 58, 365–90.
Hoffman, Martin L. (1981), ‘Is altruism part of human nature?’, Journal of Per-
sonality and Social Psychology, 40, 121–37.
Holcombe, Randall G. (1997), ‘Absence of envy does not imply fairness’, South-
ern Economic Journal, 797–802.
Holcombe, Randall G. (1998), ‘Tax policy from a public choice perspective’,
National Tax Journal, 51 (2), 359–71.
Hook, S. (ed.) (1967), Human Values and Economic Policy: a Symposium (New York:
New York University Press).
Ireland, Norman J. (1994), ‘On limiting the market for status signals’, Journal of
Public Economics, 53, 91–110.
Jackson, Peter M. (ed.) (1993), Current Issues in Public Sector Economics (London:
Macmillan).
Jiang, F. Racine, R. and Turnbull, J. (1997), ‘Electrical stimulation of the septal
References 175
region of aged rates improves performance in open-field maze’, Physiology and

Behavior, 62, 1279–82.
Joseph, Stephen and Lewis, Christopher A. (1998), ‘The depression–happiness
scale: reliability and validity of a bipolar self-report scale’, Journal of Clinical
Psychology, 54, 537–44.
Kahneman, Daniel, Wakker, Peter P. and Sarin, Rakesh (1997), ‘Back to Bentham?
explorations of experienced utility’, Quarterly Journal of Economics, May,
375–405.
Kaplow, Louis (1996), ‘The optimal supply of public goods and the distortionary
cost of taxation’, National Tax Journal, 49 (4), 513–33.
Kaplow, Louis (1998), ‘Tax and non-tax distortion’, Journal of Public Economics,
68, 303–6.
Kappel, Klemens (1997), ‘Equality, priority, and time’, Utilitas, 9, 203–25.
Kapteyn, A. and Wansbeek, T. J. (1985), ‘The individual welfare function’, Journal
of Economic Psychology, 6, 333–63.
Karras, Georgios (1996), ‘The optimal government size: further international
evidence on the productivity of government services’, Economic Inquiry, 34,
193–203.
Kau, J. B. and Rubin, P. H. (1981), ‘The size of government’, Public Choice, 37 (2),
261–74.
Kekes, John (1997), ‘A question for egalitarians’, Ethics, 107, 658–69.
Kelsey, David (1988), ‘Policies to achieve a better distribution of income: or is a
dollar a dollar?’, Oxford Economic Papers, 40, 577–83.
Kemp, Murray C. and Ng, Yew-Kwang (1976), ‘On the existence of social welfare
functions, social orderings, and social decisions functions’, Economica, 45,
59–66.
Kenny, Charles (1999), ‘Does growth cause happiness, or does happiness cause
growth?’, Kyklos, 52, 3–26.
King, Mervyn A. (1986), ‘A Pigovian rule for the optimum provision of public
goods’, Journal of Public Economics, 30, 273–91.
Kline N. S. and Laska, E. (eds) (1968), Computer and Electronic Devices in Psy-
chiatry (New York: Grune and Stratton).
Kolm, Serge-Christophe (1993), ‘The impossibility of utilitarianism’, in P.
Koslowski and Y. Shionoya, pp.30–66.
Kolm, Serge-Christophe (1996), Modern Theories of Justice (MIT Press).
Kolm, Serge-Christophe (1998), ‘Chance and justice: social policies and the
Harsanyi–Vickrey–Rawls problem’, European Economic Review, 42, 1393–416.
Konishi, Hideo (1995), ‘A Pareto-improving commodity tax reform under a
smooth nonlinear income tax’, Journal of Public Economics, 56, 413–46.
Koslowski, P. and Shionoya, Y. (eds) (1993), The Good and the Economical: Ethical
Choices in Economics and Management (Berlin: Springer-Verlag).
Krugman, Paul (1998), ‘Two Cheers for Formalism’, Economic Journal, 108,
1829–36.
Krupitskii, E. M., Burakov, A. M., Karandashova, G. F. and Lebedev, V. B. (1993),
‘A method of treating affective disorders in alcoholics’, Journal of Russian and
East European Psychiatry, 26 (3), 26–37.
Laitner, John and Juster, F. T. (1996), ‘New evidence on altruism: a study of
TIAA–CREF retirees’, American Economic Review, 86, 893–908.
176 References
Le Grand, Julian (1997), ‘Knights, knaves or pawns?: human behaviour and social
policy’, Journal of Social Policy, 26, 149–69.
Lee, Woojin and Roemer, John E. (1999), ‘Income distribution, redistributive pol-
itics, and economic growth’, Journal of Economic Growth, 3, 217–40.
Leontief, W. (1966), Essays in Economics: Theories and Theorizing (New York:
Oxford University Press).
Lerner, Abba P. (1944), The Economics of Control (New York: Macmillan).
Levitt, M. S. and Joyce, M. A. S. (1987), The Growth and Efficiency of Public Spend-
ing (Cambridge: Cambridge University Press).
Lewis, H. D. (ed.) (1976), Contemporary British Philosophy (London: Allen and
Unwin).
Linbeck, Assar, Nyberg, Stern and Weibull, Jörgen W. (1999), ‘Social norma and
economic incentives in the welfare state’, Quarterly Journal of Economics, 116,
1–36.
Little, Ian M. D. (1949), ‘The foundations of welfare economics’, Oxford Economic
Papers, 1, 227–46.
Little, Ian M. D. (1951), ‘Direct versus indirect taxes’, Economic Journal, 61,
577–84.
Little, Ian M. D. (1952), ‘Social Choice and Individual Values’, Journal of Political
Economy, 60, 422–32.
Little, Ian M. D. (1957), A Critique of Welfare Economics, 2nd edn (London: Oxford
University Press).
Little, Ian M. D. and Mirrlees, James A. (1974), Profit Appraisal and Planning for
Developing Countries (London: Heinemann).
Livingston, Arthur (ed.) (1935), Mind and Society, Vol. 4 (London: Jonathan
Cape).
Lockwood, Ben and Manning, Alan (1993), ‘Wage setting and the tax system:
theory and evidence for the united kingdom values’, Journal of Public Econ-
omics, 52, 1–29.
Loewenstein, George and Elster, Jon (eds) (1992), Choice Over Time (New York:
Sage).
Lunati, M. Teresa (1997), Ethical Issues in Economics: From Altruism to Cooperation
to Equity (New York: St. Martin’s Press; London: Macmillan Press).
Lybeck, J. A. and Henrekson, M. (1988), Explaining the Growth of Government
Marmola, Elisabetta (1999), ‘A constitutional theory of public goods’, Journal of
Economic Behaviour and Organization, 38, 27–42.
Marr, W. and Raj, B. (eds) (1983), How Economists Explain – A Reader in Method-
ology (Washington: University Press of America).
Marschak, J. (1950), ‘Rational behaviour, uncertain prospects and measurable
utility’, Econometrica, 18, 111–141; also ‘Errata’, 312.
Maskin, Eric (1978), ‘A theorem on utilitarianism’, Review of Economic Studies, 45,
93–6.
Meltzer, A. H. and Richard, Scott F. (1981), ‘A rational theory of the size of gov-
ernment’, Journal of Political Economy, 89, 914–27.
Michael, Mark A. (1997), ‘Redistributive taxation, self-ownership and the fruit of
labour’, Journal of Applied Philosophy, 14, 137–46.
Milgrom, Paul (1993), ‘Is sympathy an economic value?: philosophy, econom-
ics, and the contingent valuation method’, in Hausman, pp.417–35.
References 177
Mirrlees, James A. (1971), ‘An exploration in the theory of optimal income tax-
ation’, Review of Economic Studies, 38, 174–208.
Mirrlees, James A. (1976), ‘Optimal tax theory: a synthesis’, Journal of Public Econ-
omics, 6, 327–58.
Mirrlees, James A. (1981), ‘The theory of optimal taxation’, in Arrow and Intrili-
gator (1986), pp.1197–1249.
Mirrlees, James A. (1982), ‘The economic uses of utilitarianism’, in Sen and
Williams, pp.63–84.
Mitchell, Robert C, and Carson, Richard T. (1989), ‘Using surveys to value public
goods: the contingent valuation method’, Washington, DC: Resources for the
Future (Baltimore: Johns Hopkins University Press).
Mongin, Philippe (1988), ‘A comment on J. Harsanyi’s “Assessing other people’s
utilities” ’, in Munier, (ed.), 139–44.
Mongin, Philippe (1994), ‘Harsanyi’s aggregation theorem: multi-profile version
and unsettled questions’, Social Choice and Welfare, 11, 331–54.
Monroe, Kristen R. (1996), The Heart of Altruism (Princeton University Press).
Mueller, Dennis (1989), Public Choice II (Cambridge: Cambridge University Press).
Mueller, Dennis (ed.) (1997), Perspectives on Public Choice: a Handbook (Cam-
bridge: Cambridge University Press).
Munier, Bertrand R. (ed.) (1988), Risk, Decision and Rationality (Dordrecht: Reidel).
Mulligan, Casey B. (1997), Parental Priorities and Economic Inequality (Chicago and
London: University of Chicago Press).
Murphy, K., Shleifer, A. and Vishny, R. (1989), ‘Income distribution, market size,
and industrialization’, Quarterly Journal of Economics, 537–64.
Myers, David (1996), Social Psychology (New York: Macmillan).
Nagel, Stuart S. (ed.) (1998), Applications of Super-Optimizing Policy Analysis,
Research in Public Policy Analysis and Management, Vol. 8 (Greenwich, Conn.
and London: JAI Press).
Nath, S. K. (1969), A Reappraisal of Welfare Economics (London: Routledge & Kegan
Paul).
Ng, Yew-Kwang (1971), ‘The possibility of a Paretian liberal: impossibility the-
orems and cardinal utility’, Journal of Political Economy, 79 (6), 1397–402.
Ng, Yew-Kwang (1972), ‘Value judgements and economists’ role in policy rec-
ommendation’, Economic Journal, Sept., 1014–18; reprinted in Marr and Raj,
(1983).
Ng, Yew-Kwang (1975a), ‘Bentham or Bergson?: finite sensibility, utility functions
and social welfare functions’, Review of Economic Studies, 42, 545–70.
Ng, Yew-Kwang (1975b), ‘Non-economic activities, indirect externalities and
third-best policies’, Kyklos, 28, 507–25; reprinted in Ng (1990b).
Ng, Yew-Kwang (1977), ‘Towards a theory of third-best’, Public Finance, 1–15.
Ng, Yew-Kwang (1979/1983), Welfare Economics: Introduction and Development of
Basic Concepts (London: Macmillan).
Ng, Yew-Kwang (1980), ‘Toward eudaimonology: notes on a quantitative frame-
work for the study of happiness’, Mathematical Social Sciences, 1, 51–68.
Ng, Yew-Kwang (1981a), ‘Welfarism: a defence against Sen’s attack’, Economic
Journal, 91, June, 527–30.
Ng, Yew-Kwang (1981b), ‘Bentham or Nash?: on the acceptable form of
social welfare functions’, Economic Record, 57 Sept., 238–50; reprinted in Ng
(1990b).
178 References
Ng, Yew-Kwang (1982), ‘Beyond Pareto optimality: the necessity of interpersonal

cardinal utilities in distributional judgements and social choice’, Zeitschrift für
Nationalökonomie (Journal of Economics), 42 (3), 207–33.
Ng, Yew-Kwang (1983), ‘Some broader issues of social choice’, in P. K. Pattanaik
and M. Salles, pp.151–73.
Ng, Yew-Kwang (1984a), ‘Expected subjective utility: is the Neumann–
Morgenstern utility the same as the neoclassical’s?’, Social Choice and Welfare,
177–86; reprinted in Ng (1990b).
Ng, Yew-Kwang (1984b), ‘Quasi-Pareto social improvements’, American Economic
Review, Dec., 74 (5), 1033–50; reprinted in Ng (1990b).
Ng, Yew-Kwang (1985), ‘Some fundamental issues in social welfare’, in Feiwel,
pp.435–69; reprinted in Ng (1990b).
Ng, Yew-Kwang (1987a), ‘Diamonds are a government’s best friend: burden-free
taxes on goods valued for their values’, American Economic Review, 77, 186–91.
Ng, Yew-Kwang (1987b), ‘Relative-income effects and the appropriate level of
public expenditure’, Oxford Economic Papers, 293–300.
Ng, Yew-Kwang (1988), ‘Economic efficiency versus egalitarian rights’, Kyklos,
215–37.
Ng, Yew-Kwang (1989a), ‘What should we do about future generations?: the
impossibility of Parfit’s Theory X’, Economics and Philosophy, 5, 135–253.
Ng, Yew-Kwang (1989b), ‘Individual irrationality and social welfare’, Social Choice
and Welfare, 6, 87–101.
Ng, Yew-Kwang (1990a), ‘Welfarism and utilitarianism: a rehabilitation’, Utilitas,
2 (2), 171–93.
Ng, Yew-Kwang (1990b), Social Welfare and Economic Policy (Hemel Hempstead,
Hertfordshire: Harvester Wheatsheaf ).
Ng, Yew-Kwang (1992a), ‘Do individuals optimize in intertemporal consump-
tion/saving decisions?: a liberal method to encourage savings’, Journal of Econ-
omic Behaviour and Organization, 17, 101–14.
Ng, Yew-Kwang (1992b), ‘Utilitarianism and interpersonal comparison: some
implications of a materialist solution to the world knot’, Social Choice and
Welfare, 9 (1), 1–15.
Ng, Yew-Kwang (1993), ‘Mixed diamond goods and anomalies in consumer
theory: upward-sloping compensated demand curves with unchanged dia-
mondness’, Mathematical Social Sciences, 25, 287–93.
Ng, Yew-Kwang (1995), ‘Towards welfare biology: evolutionary economics of
animal consciousness and suffering’, Biology and Philosophy, 10 (3), 255–85.
Ng, Yew-Kwang (1996a), ‘Happiness surveys: some comparability issues and an
exploratory survey based on just perceivable increments’, Social Indicators
Research, 38 (1), 1–29.
Ng, Yew-Kwang (1996b), ‘Complex niches favour rational species’, Journal of
Theoretical Biology, 179, 303–11.
Ng, Yew-Kwang (1997), ‘A case for happiness, cardinal utility, and interpersonal
comparability’, Economic Journal, 107 (445), 1848–58.
Ng, Yew-Kwang (1999), ‘Utility, informed preference, or happiness?’, Social Choice
and Welfare, 16 (2), 197–216.
Ng, Yew-Kwang (forthcoming a), ‘From separability to unweighted sum: a case
for utilitarianism’, Theory and Decision.
References 179
Ng, Yew-Kwang (forthcoming b), ‘The optimal size of public spending and the
distortionary cost of taxation’, revision under consideration by National Tax
Journal.
Ng, Yew-Kwang and Wang, Jianguo (1993), ‘Relative income, aspiration, en-
vironmental quality, individual and political myopia: why may the rat-race for
material growth be welfare-reducing?’, Mathematical Social Sciences, 26, 3–23.
Nielsen, Kai (1985), Equality and Liberty: a Defence of Radical Egalitarianism
(Totowa, NJ: Rowman & Allanheld).
Nozick, Robert (1974), Anarchy, State, and Utopia (New York: Basic Books).
Okun, Arthur M. (1975), Equality and Efficiency: The Big Tradeoff (Washington:
Brookings Institution).
Olds, J. and Milner, P. (1954), ‘Positive reinforcement produced by electrical stim-
ulation of septal area and other regions of the rat brain’, Journal of Compara-
tive Physiological Psychology, 47, 419–27.
Osborne, D. K. (1976), ‘Irrelevant alternatives and social welfare’, Econometrica,
44, 1001–15.
Oswald, Andrew J. (1997), ‘Happiness and economic performance’, Economic
Journal, 107, 1815–31.
Pareto, V. (1935), The Mind and Society, trans. Livingston (London: Cape).
Parks, Robert P. (1976), ‘An Impossibility theorem for fixed preferences: a dicta-
torial Bergson–Samuelson welfare function’, Review of Economic Studies, 43,
447–50.
Pattanaik, Prasanta K. (1971), Voting and Collective Choice (Cambridge: Cambridge
University Press).
Pattanaik, Prasanta K. and Salles, Maurice (eds) (1983), Social Choice and Welfare
Patterson, M. M. and Kesner, R. P. (1981), Electrical Stimulation Research Techniques
(New York: Academic Press).
Patterson, M., Krupitsky, E., Flood, N. and Baker, D. (1994), ‘Amelioration of
stress in chemical dependency detoxification by transcranial electrostimu-
lation’, Stress Medicine, Apr., 10 (2), 115–26.
Payne, J. E. and Ewing, B. T. (1996), ‘International evidence on Wagner’s hypoth-
esis: a cointegration analysis’, Public Finance, 51 (2), 258–74.
Peltzman, S. (1980), ‘The growth of government’, Journal of Law and Economics,
23, 209–87.
Pencavel, John (1986), ‘Labor supply of man: a survey’, in Ashenfelter and Layard
pp.3–102.
Perotti, Roberto (1996), ‘Growth, income distribution, and democracy: what the
data say’, Journal of Economics Growth, 1, 149–87.
Persson, Torsten and Tabellini, Guido (1994), ‘Is inequality harmful for growth?’,
American Economic Review, 84, 600–21.
Phelps, Edmund S. and Zoega, Gylfi (1998), ‘Natural-rate theory and OECD
unemployment’, Economic Journal, 108, May, 782–801.
Pigou, Arthur C. (1912/1929/1932), Wealth and Welfare. Later editions (1920, 1924,
1929, 1932) assume the title The Economics of Welfare (London: Macmillan).
Pigou, Arthur C. (1928), Public Finance (London: Macmillan).
Pojman, Louis P. and Westmoreland, Robert, (eds) (1997), Equality: Selected Read-
ings (Oxford: Oxford University Press).
180 References
Pollak, Robert A. (1979), ‘Bergson–Samuelson social welfare functions and the

theory of social choice’, Quarterly Journal of Economics, 93, 73–90.
Portney, Paul R. (1990), ‘Economics and the Clean Air Act’, Journal of Economic
Perspectives, 4 (4), 173–81.
Posner, Richard (1981), The Economics of Justice (Cambridge, MA: Harvard Uni-
versity Press).
Putterman, Louis, Roemer, John E. and Silvestre, Joaquim (1998). ‘Does egalitar-
ianism have a future?’, Journal of Economic Literature, 36, 861–902.
Qizilbash, Mazaffar (1998), ‘The concept of well-being’, Economics and
Philosophy, 14, 51–73.
Radford, C. (1988), ‘Utilitarianism and the noble art’, Philosophy, ixiii, 63–81.
Rae, John (1834), New Principles of Political Economy, reprinted as The Sociological
Theory of Capital, edited by Charles W. Mixter (1990), (New York: Macmillan).
Ramsey, F. P. (1928), ‘A mathematical theory of saving’, Economic Journal, 38,
543–59; reprinted in Arrow and Scitovsky (1969).
Rawls, John (1971), A Theory of Justice (Cambridge, MA: Harvard University Press).
Rawls, John (1982), ‘Social unity and primary goods’, in Sen and Williams,
pp.159–85.
Robbins, Lionel (1938), ‘Interpersonal Comparison of Utility: a Comment’,
Economic Journal, 48, 635–41.
Roberts, Kevin W. S. (1980), ‘Social choice theory: the single-profile and multi-
profile approaches’, Review of Economic Studies, 47, 441–50.
Rodrik, Dani (1998), ‘Why do more open economies have bigger governments?’,
Journal of Political Economy, 106, 997–1032.
Roemer, John E. (1986), ‘Equality of resources implies equality of welfare’,
Quarterly Journal of Economics, 101, 751–84.
Roemer, John E. (1996), Theories of Distributive Justice (Cambridge, MA: Harvard
University Press).
Roemer, John E. (1998), Equality of Opportunity (Cambridge, MA and London:
Harvard University Press).
Rose-Ackerman, Susan (1996), ‘Altruism, nonprofits, and economic theory’,
Journal of Economic Literature, 34, 701–28.
Ryan, Richard M., Chirkov, Valery I., Little, Todd D., Sheldon, Kennon M.,
Timoshina, Elena and Deci, Edward L. (forthcoming), ‘The American dream in
Russia: aspirations and well-being in two cultures’, Personality and Social
Psychology Bulletin.
Samuelson, Paul A. (1967), ‘Arrow’s mathematical politics’, in S. Hook.
Samuelson, Paul A. (1977), ‘Reaffirming the existence of reasonable
Bergson–Samuelson social welfare functions’, Economica, 44, 81–8.
Sandmo, Agnar (1976), ‘Optimal taxation: an introduction to the literature’,
Journal of Public Economics, 6, 37–54.
Saunders, Peter (1993), ‘Recent trends in the size and growth of government in
OECD countries’, in Gemmell.
Schmid, G. (1993), ‘Equality and efficiency in the labor market’, Journal of Socio-
Schyns, Peggy (1998), ‘Crossnational differences in happiness’, Social Indicators
Research, 43, 3–26.
Sen, Amartya K. (1967), ‘The nature and classes of prescriptive judgments’, Philo-
sophical Quarterly, 17, 46–62.
References 181
Sen, Amartya K. (1969), ‘Quasi-transitivity, rational choice, and collective

decisions’, Review of Economic Studies, 36, 381–93.
Sen, Amartya K. (1970a), Collective Choice and Social Welfare (Amsterdam: North-
Holland).
Sen, Amartya K. (1970b), ‘The impossibility of a Paretian liberal’, Journal of Polit-
ical Economy, 78, 152–7.
Sen, Amartya K. (1973a), On Economic Inequality (Oxford: Clarendon).
Sen, Amartya K. (1973b), ‘On ignorance and equal distribution’, American Econ-
omic Review, 63, Dec., 1022–4; reprinted in Sen (1982).
Sen, Amartya K. (1979), ‘Personal utilities and public judgments: Or what’s wrong
with welfare economics?’, Economic Journal, 89, 537–58; reprinted in Sen
(1982).
Sen, Amartya K. (1981), ‘A reply to “Welfarism: a defence against Sen’s attack” ’,
Economic Journal, 89, 537–58; reprinted in Sen (1982).
Sen, Amartya K. (1982), Choice, Welfare and Measurement (Oxford: Basil Black-
well).
Sen, Amartya K. (1985), Commodities and Capabilities (Amsterdam: North-
Holland).
Sen, Amartya K. (1986), ‘Social choice theory’, in Arrow and Intriligator,
pp.1073–181.
Sen, Amartya K. (1992), Inequality Reexamined (Oxford: Clarendon).
Sen, Amartya K. (1997), ‘Quality of life and economic evaluation’, Academia
Economic Papers, 25, 269–316.
Sen, A. and Williams, B. (eds) (1982), Utilitarianism and Beyond (Cambridge: Cam-
bridge University Press).
Skolnik, M. L. (1970), ‘A comment on Professor Musgrave’s separation of
distribution from allocation’, Journal of Economic Literature, June, 8, 440–2.
Slemrod, Joel and Yitzhaki, Shlomo (1996), ‘The costs of taxation and the mar-
ginal cost of funds’, International Monetary Fund Staff Papers, 43 (1), 172–98.
Smart, J. J. C. and Williams, B. A. O. (1973), Utilitariansim: For and Against,
(Cambridge: Cambridge University Press).
Snow, Arthur and Warren, Ronald S. Jr. (1996), ‘The marginal welfare cost of
public funds: Theory and estimates’, Journal of Public Economics, 61 (2),
289–305.
Stiglitz, Joseph J. (1988), Economics of the Public Sector (New York: Norton).
Stiglitz, Joseph J. (1999), ‘Public policy for a knowledge economy’, reported in
The Age, 27 Mar. 1999.
Strack, Fritz, Argyle, Michael and Schwarz, Norbert (eds) (1991), Subjective Well-
Being: An Interdisciplinary Perspective (Oxford: Pergamon Press).
Strauss, John and Thomas, Duncan. (1995), ‘Human resources: Empirical mod-
eling of household and family decisions’, in Behrman and Srinivasan
pp.1883–2023.
Stuart, Charles (1984), ‘Welfare cost per dollar of additional tax revenue in the
United States’, American Economic Review, 74 (3), 352–62.
Sugden, Robert and Weale, Albert (1979), ‘A contractual reformulation of certain
aspects of welfare economics’, Economica, 46, 111–23.
Summers, R. and Heston, A. (1991), ‘Penn World Table (Mark 5): an expanded
set of international comparisons, 1950–1988’, Quarterly Journal of Economics,
106, 327–68.
182 References
Sumner, L. W. (1996), Welfare, Happiness, and Ethics (Oxford: Clarendon Press;

New York: Oxford University Press).
Suzumura, Kotaro (1997), ‘Partial welfare judgements as preliminaries for ration-
al social choice’, Institute for Social and Economic Research Working Paper,
Osaka University.
Suzumura, Kotaro (1999), ‘Consequences, opportunities, and procedures’, Social
Choice and Welfare, 16 (1), 17–40.
Temkin, L. S. (1987), ‘Intransitivity and the mere addition paradox’, Philosophy
and Public Affairs, 16 (2), pp.138–87.
Trivers, R. L. (1971), ‘The evolution of reciprocal altruism’, Quarterly Review of
Biology, 46, 35–57.
Turnbull, Geoffrey K. and Chang, Chinkun (1998), ‘The median voter according
to GARP’, Southern Economic Journal, 64, 1001–10.
Valenstein, E. S. (ed.) (1973), Brain Stimulation and Motivation (Glenview, Illinois:
Scott, Foresman).
van Praag, Bernard M. S. (1968), Individual Welfare Functions and Consumer Behav-
iour (Amsterdam: North Holland).
van Praag, Bernard M. S. (1991), ‘Ordinal and cardinal utility: an integration
of the two dimensions of the welfare concept’, Journal of Econometrics, 50,
69–89.
Varian, Hal R. (1993), Intermediate Microeconomics: A Modern Approach (New York:
Norton).
Veblen, Thorstein (1929), The Theory of the Leisure Class, reprinted in 1970, by
(London: Unwin).
Veenhoven, Ruut (1984), Conditions of Happiness (Dordrecht: Kluwer Academic).
Veenhoven, Ruut (1993), Happiness in Nations: Subjective Appreciation of Life in 56
Nations 1946–1992 (Rotterdam: RISBO).
Watson, R. (1977), ‘Reason and Morality in a World of Limited Food’, in Aiken
and LaFollette, pp.115–23.
Wauquier, A. and Rolls, E. T. (eds) (1976), Brain-Stimulation Reward (Amsterdam:
North-Holland).
Weintraub, E. Roy (1998), ‘Controversy: AXIOMATISCHES MIßVERSTÄNDNIS’, Econ-
omic Journal, 108, 1837–47.
West, Edwin G. (1991), ‘Secular cost changes and the size of government’, Journal
of Public Economics, 45, 363–81.
Weymark, J. A. (1991), ‘A reconsideration of the Harsanyi–Sen debate on utili-
tarianism’, in Elster and Roemer, pp.255–320.
Wicksteed, Philip H. (1933), The Common Sense of Political Economy, edited by
Lionel Robbins (London: Routledge).
Wilkinson, Richard G. (1997), ‘Health inequalities: relative or absolute material
standards?’ British Medical Journal, 314, 22 Feb., 591–5.
Wilson, E. O. (1975), Sociobiology: the New Synthesis (Cambridge, MA: Harvard
University Press).
Wilson, John D. (1991), ‘Optimal public good provision with limited lump-sum
taxation’, American Economic Review, 81 (1), 153–66.
Winer, Stanley L. and Hettich, Walter (1998), ‘What is missed if we leave out
collective choice in the analysis of taxation?’, National Tax Journal, 51 (2),
373–89.
References 183
Winkelmann, Liliana and Winkelmann, Rainer (1998), ‘Why are the unemployed
so unhappy?’, Economica, 65, 1–15.
Yadin, E. and Thomas, E. (1996), ‘Stimulation of the lateral septum attenuates
immobilisation-induced stress ulcers’, Physiology and Behavior, 59, 883–6.
Zamagni, Stefano (ed.) (1995), The Economics of Altruism (Aldershot, Brookfield,
Vt, USA: Edward Elgar).
Name Index
Ackerman, F., 98 Chick, V., 11

Adler, M.D., 97 Christiansen, V., 125
Aghion, P., 98 Clark, J., 41
Ake, C., 145 Cohen, G.A., 143
Altonji, J.G., 41 Cole, H.L., 113
Arneson, R.J., 143, 149 Congleton, R.D., 98
Arrow, K.J., 7–8, 72, 82, 163 Corden, W.M., 95
Atkinson, A.B., 107, 124, 166 Cornes, R.C., 166
Auten, G., 125 Creedy, J., 97, 165
Cremer, H., 100
Backhouse, R.E., 11 Cummins, R.A., 116
Bagwell, L.S., 41
Baland, J.-M., 98 D’Aspremont, C., 58
Ballard, C.L., 111, 125, 126 de Bartolome, C.A.M., 94, 107
Barry, B.M., 62 Delbrück, M., 165
Batina, R.G., 124, 127, 167 Delgado, J.M.R., 139
Baumol, W.J., 108, 110, 167 DeMeyer, F., 18
Bénabou, R., 98 Devins, N., 98
Bergson, A., 7, 163 Devinsky, O., 138
Bergstrom, T.C., 166 Diamond, P.A., 76–7
Bernheim, B.D., 41 Diener, E., 116
Berridge, K.C., 43 Donaldson, D., 84
Binswinger, H., 97 Douglas, D.M., 98
Bishop, M.P., 139 Dowrick, S., 107
Birdsall, N., 98 Duesenberry, J.S., 113
Blackorby, C., 84 Dworkin, R., 44, 141, 143, 149
Blaug, M., 11
Boadway, R.W., 125 Easterlin, R.A., 113
Bohl, M.T., 105 Easterly, W., 117, 122
Borghans, L., 98, 167 Eaton, B.C., 98
Bowles, S., 56, 98 Edgeworth, F.Y., 8, 65, 67, 68, 69, 72
Breen, R., 98 Eisenstein, L., 105
Brennan, G., 167 Elster, J., 48
Brickman, P., 50, 118 Eshel, I., 41
Broome, J., 77 Ewing, B.T., 105
Brown, D., 98
Browning, E.K., 125, 126, 167 Feenberg, D., 125
Buchanan, J.A., 167 Feldstein, M., 4, 94, 103, 106, 111,
122, 125–6, 130 –4
Carroll, R., 125 Ferris, S.J., 108
Carson, R.T., 166 Fischer, D., 167
Chang, C., 107 Fleming, M., 58, 59, 61, 66, 165,
Chang, M.C., 124, 127 166
184
Name Index 185
Fortin, B., 106 Karras, G., 107

Frank, R.H., 52, 113, 116 Kau, J.B., 108
Friedman, D., 74 Keen, M., 125
Kekes, J., 145
Gahvari, F., 100 Kelsey, D., 93, 100
Gemmell, N., 110 Kemp, M.C., 8, 22, 33, 82
Gevers, L., 58 Kenny, C., 116
Ginitis, H., 98 Kesner, R.P., 138
Goodin, R.E., 57 Keynes, 49, 118
Groot, L., 98, 167 King, M.A., 124
Grossman, H., 98 Kolm, S.-C., 18, 76, 82, 143
Konishi, H., 125
Hamilton, W.D., 43 Krugman, P., 11
Hammond, P.J., 22, 30, 44, 72, 158 Krupitskii, E.M., 138
Harbaugh, W.T., 41
Hardy, J.D., 73 Lacroix, G., 106
Hare, R.M., 27, 141 Laitner, J., 41
Harris, R., 79 Le Grand, J., 98
Harrod, R.F., 49 Lee, W., 98
Harsanyi, J.C., 3, 34, 35, 36, 37, 38, Leontief, W., 5–6
44, 45–6, 52, 55, 59, 60, 66, 67, Lerner, A.P., 25, 75
68, 69, 70, 76, 158, 166 Levitt, M.S., 108
Haveman, R.H., 98 Lewis, C.A., 18
Hayes, K.J., 166 Linbeck, A., 95
Headley, B., 118 Little, I.M.D., 2, 21–2, 87–8, 89, 97
Heal, G., 98 Liu, L., 167
Heath, R.G., 138 Lockwood, B., 97
Henrekson, M., 108 Lunati, M.T., 41
Herrnstein, R.J., 50 Lybeck, J.A., 108
Hersoug, T., 97 Lyon, A., 97
Heston, A., 116
Hettich, W., 167 Manning, A., 97
Hirsch, F., 113 Marmola, E., 108
Hoff, K., 97 Marschak, J., 60, 78
Hoffman, M.L., 43 Marshall, 163
Holcombe, R.G., 80, 167 Maskin, E., 58
McDonald, I.M., 97
Ireland, N.J., 113 Medema, S.G., 111
Meltzer, A.H., 108
Jackson, P.M., 107 Michael, M.A., 144
Jiang, F., 138 Milgrom, P., 166
Joseph, S., 18 Milner, P., 138
Joyce, M.A.S., 108 Mirrlees, J.A., 73, 74, 97, 125, 166
Juster, F.T., 41 Mitchell, R.C., 166
Mongin, P., 69, 70–1
Kahneman, D., 36 Monroe, K.R., 43
Kaplow, L., 5, 45, 107, 126–33 Mueller, D., 61, 108
Kappel, K., 141 Mulligan, C.B., 41
Kapteyn, A., 113 Munier, B.R., 68
186 Name Index
Murphy, K., 98 Robbins, L., 82

Myers, D., 116, 117 Roberts, K.W.S., 8, 22, 69
Rodrik, D., 108
Nagel, S.S., 98 Roemer, J.E., 11, 67, 68, 98, 141, 143,
Nath, S.K., 158 146
Ng, S., 105, 151 Rose-Ackerman, S., 41
Ng, Y.-K., 2, 7, 8, 10, 17, 20, 22, 24, Rubin, P.H., 108
28–30, 33, 37, 47, 49, 56, 60–1, Ryan, R.M., 116
65, 67, 69, 74, 75, 82–4, 86, 87,
89, 91, 96, 100, 101, 112, 114, Samuelson, P.A., 21–2, 163
115, 118, 123, 129–33, 142, 158, Sandmo, A., 166
165–7 Saunders, P., 105
Nielsen, K., 141, 143, 145, 150 Schmid, G., 98
Nozick, R., 141, 144 Schyns, P., 116
Scitovsky, T., 163
Okun, A.M., 1 Sem-Jacobsen, 139
Olds, J., 138 Sen, A., 5–8, 18, 24–9, 33, 66–7, 69,
Olewiler, N., 79 73–5, 82, 141, 143, 145, 146,
Osborne, D.K., 163 148–9, 165
Oswald, A.J., 116 Skolnik, M.L., 95
Slemrod, J., 125
Pareto, V., 81, 158 Smart, J.J.C., 29
Parish, R., 4 Snow, A., 126
Parks, R.P., 8, 22, 33, 82 Stiglitz, J.E., 107, 123, 124, 166
Pattanaik, P.K., 74 Strauss, J., 97
Patterson, M., 138 Stuart, C., 125
Patterson, M.M., 139 Sugden, R., 58, 59–60, 61
Payne, J.E., 105 Suh, E., 116
Peltzman, S., 108 Summers, R., 116
Perotti, R., 97 Sumner, L.W., 35
Persson, T., 97 Suzumura, K., 1, 29
Phelps, E.S., 107
Pigou, A.C., 1, 5, 49, 124, 163 Tabellini, G., 97
Platteau, J.-P., 98 Temkin, L.S., 146
Plott, C.R., 18 Thomas, D., 97
Pojman, L.P., 143, 145 Thomas, E., 138
Pollak, R.A., 22 Trivers, R.L., 43
Portney, P.R., 114 Turnbull, G.K., 107
Posner, R., 91, 97, 141
Prelec, D., 50 Valenstein, E.S., 139
Putterman, L., 98 van Praag, B.M.S., 8
Varian, H.R., 18
Qizilbash, M., 18 Veblen, T., 113
Veenhoven, R., 8, 115, 116, 142
Radford, C., 30
Rae, J., 113 Wang, J., 115
Ramsey, F.P., 49 Wansbeek, T.J., 113
Rawls, J., 141, 143, 144, 148, 167 Warren, R.S., Jr., 126
Richard, S.F., 108 Watson, R., 145
Name Index 187
Weale, A., 58, 59–60, 61 Wilson, E.O., 43

Wearing, A., 118 Wilson, J.D., 124–5
Weintraub, E.R., 11 Winer, S.L., 167
West, E.G., 108 Winkelmann, L., 117
Westmoreland, R., 143, 145 Winkelmann, R., 117
Weymark, J.A., 69, 70
White, W., 98 Yadin, E., 138
Wicksteed, P.H., 82 Yitzhaki, S., 125
Wilkinson, R.G., 113
Williams, B.A.O., 29 Zamagni, S., 41
Williamson, J., 98 Zoega, G., 107
Subject Index
a dollar is a dollar, 82–100 fairness, 76–81

and Kaplow’s argument 128–30, fallacy of misplaced abstraction, 7
132 formalism, 10–11
absolute income, unimportance of ,
115–19 government, optimal size of (see also
Allais paradox, 71 public spending), 104–34, 151–7
altruism
affective, 37–8, 44–5, 161–2 happiness (see also brain stimulation),
non-affective, 34, 37–45, 158–62 18–21, 32, 54–6, 115–17, 123,
anonymity, 22, 73 141–2
Anonymity Plus Orderings Only Harsanyi, defence against recent
(Kemp & Ng), 22 criticisms, 66–71
basic problems of public policy, 7–8 impossibility theorems

basic problems of social choice, 7–8 intra or single profile framework,
Bergson-Samuelson’s tradition, 163 22
brain, stimulation of, 123, 137–40 individualism, 33, 61–6, 163
burden-free taxes, 111–13 rational, 32–5
ignorance, 48, 94, 118–19
cardinal utility, 1–2, 15–23, 72–6 information, imperfection, 20
capability, 148–50 informed preferences, 34, 36
compensation test, 1–2 interpersonal cardinal utility
consequentialism, 30 paradox of, 82–3
cost–benefit analysis, 82–90, 97 solution, 83–98
interpersonal comparisons of utility,
diamond goods, 111–13 1–2, 15–23, 67–70, 82
direct vs indirect taxes, 87–90 irrational preferences, 45–52, 54
disincentive effects of taxation, 82–
98 just perceptible increments, 72–6
distributional weights, 82–90
reversed weighting, 90–2 libertarians, 143–4
Little’s criterion, 2
economic growth Local Cardinal Individualism, 61
and public spending, 151–7
and quality of life, 117 median-voter model, 107
welfare-reducing, 114–15 moral principles, 29, 50–1, 56
efficiency-equality tradeoff, 1–2 motives, 30
egalitarianism, 11, 62, 141–50
environmental quality, 120, 155 natural selection, 46–52
equality (see also egalitarianism), 2, new welfare economics, 1
84–98, 100, 143 normative statements, 9
expected utility hypothesis, 67–70
expected social welfare, 79 ordinalism, 6, 22, 163
188
Subject Index 189
Pareto principle, 33, 158–62 social welfare functional, 11, 163

political constraints, 92 subjective judgements of fact, 10
political decisions, 101–2
politics, 99–102 tax/transfer system, 82–98
prescriptive statements, 9 ineffectiveness in, 93, 100
process fairness, 76–81 taxation
public goods, 105–7, 120–3 and diamond goods, 111–13
public policy, basic problems of, 7–8 corrective, 111
public spending, optimal size of, direct vs indirect, 87–90
104–34 disincentive effects of, 82–98
and benefit–cost ratio, 124–34 distortionary costs of, 125–6
and distortionary cost of taxation, third best, 87
124–34 transaction costs, 93–4
and relative-income effects, 113–21,
151–7 utilitarianism, 57–81
growth in, 107–8, 151–7 and Harsanyi, 60–71
inefficiency in, 108–10, 121–2, and process fairness, 76–81
164 and Weak MP Criterion, 61–2
implied by rational individualistic
rational species, 118 egalitarianism, 62–6
rationalism, 34
Rawls, a critique of, 144–5 value judgements, basic, 26–7
research, 121
rights, 27–9 Weak Equity Axiom (Sen), 73
Weak Majority Preference Criterion
second best, 87–90, 124 (Ng), 61, 73–6
social choice, basic problems of, 7–8 welfare (see happiness)
social welfare function (see also welfarism, 6, 24–35
utilitarianism), 6, 8 Sen–Ng debate, 24–30
linearity, 58, 60 willingness to pay, 82–6
separability, 58–60
unweighted sum, 58, 60 zero happiness/utility/welfare, 18

Yew-Kwang NG - Efficiency, Equality and Public Policy - With A Case For Higher Public Spending - Palgrave Macmillan (2000)

Uploaded by

Copyright:

Available Formats

Yew-Kwang NG - Efficiency, Equality and Public Policy - With A Case For Higher Public Spending - Palgrave Macmillan (2000)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Yew-Kwang NG - Efficiency, Equality and Public Policy - With A Case For Higher Public Spending - Palgrave Macmillan (2000)

Uploaded by

Copyright:

Available Formats

Efficiency, Equality and

INCREASING RETURNS AND ECONOMIC ANALYSIS (editor with K. Arrow and

First published in the United States of America 2000 by

Printed and bound in Great Britain by

Part I The Foundation of Public Economic Policy 13

4 Utility, Informed Preference, or Happiness 36

5.2 Rational individualistic egalitarianism implies

6 A Dollar is a Dollar: Solution to the Paradox of

7 Economics versus Politics 99

Part II How Much Should the Government Spend? 103

9 The Appropriate Benefit–Cost Ratio for Public Spending 124

10 Concluding Remarks 135

Appendix A Electrical Brain Stimulation: a Case Showing the

A3. Enormous benefits 138

Appendix B Pure Egalitarianism: a Critique 141

Appendix C Economic Growth Increases the Optimal Share of the

Appendix D Non-Affective Altruism: When the Pareto Principle is

Appendix E The Bergson–Samuelson Tradition Implies Individualism,

Appendix F Inefficiency in Provision May Increase Optimal Public

Name Index 184

Subject Index 188

This book addresses some important and fundamental problems of

1.1 A dollar is a dollar: a simple solution to the

As emphasised in the title of Okun’s (1975) book, efficiency versus

1.2 A partial resurrection of the old

The old welfare economists (represented by Pigou, 1912/1932) assumed

1.3 Introductory summary

An economic problem is typically the maximisation of the relevant

The basic argument for ‘a dollar is a dollar’ is this. The objective of

1.4 A case for higher public spending

Part II discusses the appropriate size of government spending, an issue

attempt to reduce or to limit the growth in the size of government, a

1.5 The misguided consensus

As noted by Amartya Sen (1979, p.537), Wassily Leontief has sum-

a general consensus of opinion seems to exist’ in the formal discussion

In the discussion of public economic policies – the normative

Sen then launched a vigorous attack on these four conditions. As usually

on positivity, welfarism implies the Pareto principle. On the other hand,

1.6 Three basic problems of social choice/public policy

If we regard issues like strategic behaviour, incentive compatibility,

I. What types of information are relevant and needed in making

Arrow’s (1951/1963) theorem established the impossibility of a

II. How can we obtain the required information?

III. Which form of social welfare function (SWF)

1.7 Some specific points made

• More research should be done to perfect the methods of electrical

1.8 A methodological note

(General readers may skip this subsection.)

ual utilities/welfares, see Chapter 2. With this convention, it means that

1.9 What this book is not about

2.1 An intuitive explanation using a parent’s choice

Suppose you are a parent having to make certain decisions affecting

H1 = (Jan has $75, Kelvin has $25, Hot weather)

The preference patterns of Jan and Kelvin described above may be

we have the following preference cycle: C1 preferred to H2 (weak

2.2 Economists’ misplaced hostility against cardinal utility

When I first studied Arrow’s impossibility theorem, I quickly attributed

But how do we tell if a person likes one bundle twice as much as

There is in fact an interpretation that is especially compelling. This is