You are on page 1of 210

Cambridge University Press

978-1-316-51026-1 — The Origins of Behavioural Public Policy


Adam Oliver
Frontmatter
More Information

The Origins of Behavioural Public Policy

The use of behavioural science to inform policy is one of the


main developments in the social sciences over the last several
decades. In this book, Adam Oliver offers an accessible introduction
to the development of behavioural public policy, examining
how behavioural economics might be used to inform the design of a
broad spectrum of policy frameworks, from nudges, to bans on certain
individual behaviours, to the regulation of the commercial sector. He
also considers how behavioural economics can explain and predict
phenomena as a challenge to economists’ assumptions around
how people perceive time, utility and money. The book offers an
intellectual foundation for all those concerned with behavioural
public policy, from academics, undergraduate and postgraduate
students with a diverse range of disciplinary perspectives, such as
economics, political science, sociology and anthropology, to policy
makers and practitioners working directly with behavioural public
policy in their everyday working lives.

adam oliver is a Reader in the Department of Social Policy at


the London School of Economics and Political Science. Dr Oliver’s
principal research interests focus upon behavioural economics and its
applications to public and private decision making, on which he has
published extensively. He is a founding editor of the journals Health
Economics, Policy and Law and Behavioural Public Policy.

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Frontmatter
More Information

The Origins of Behavioural


Public Policy
adam oliver
London School of Economics and Political Science

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Frontmatter
More Information

University Printing House, Cambridge CB2 8BS, United Kingdom


One Liberty Plaza, 20th Floor, New York, NY 10006, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
4843/24, 2nd Floor, Ansari Road, Daryaganj, Delhi – 110002, India
79 Anson Road, #06–04/06, Singapore 079906

Cambridge University Press is part of the University of Cambridge.


It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning, and research at the highest international levels of excellence.

www.cambridge.org
Information on this title: www.cambridge.org/9781316510261
DOI: 10.1017/9781108225120
© Adam Oliver 2017
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2017
Printed in the United Kingdom by Clays, St Ives plc
A catalogue record for this publication is available from the British Library.
ISBN 978-1-316-51026-1 Hardback
ISBN 978-1-316-64966-4 Paperback
Cambridge University Press has no responsibility for the persistence or accuracy of
URLs for external or third-party Internet Web sites referred to in this publication
and does not guarantee that any content on such Web sites is, or will remain,
accurate or appropriate.

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Frontmatter
More Information

For Ketevan
Who shoved me into writing this book.

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Frontmatter
More Information

Contents

List of Figures page x


List of Tables xi
Preface xiii
The Target Audience xiv
How to Approach the Book xiv
Notes on Style xv
How the Book is Organised xvi
Acknowledgements xviii

1 Assuming Rationality 1
The Origins of Economic Rationality 2
A Quiet Interlude 5
The Neo-Bernoullian Formulation 7
Utility Elicitation 12

2 Challenging Rationality 16
The Allais Paradox 17
The Ellsberg Paradox 21
Satisficing and Rules of Thumb 25
Classic Preference Reversals 30

3 Describing Risky Behaviours 34


The Markowitz Model 35
Prospect Theory 36
Prospect Theory and Allais’ Conjecture 44
Reflection 46
Regret and Disappointment 50

vii

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Frontmatter
More Information

viii contents

4 About Time 54
The Normative Case for Discounting 55
Present Bias 60
Present Bias and Risk Attitude 64
The Time Trade-Off 67

5 Experiencing and Remembering 71


Empirical Examples of the Gestalts 73
The Gestalts and Respect for Individual Agency 80
Using the Gestalts Descriptively 86

6 Motivational Crowding 88
Self-Determination Theory 90
Empirical Examples of Crowding 92
Personal Financial Incentives 96
Designing Performance Management 101

7 Nudges 108
The Nudge Requirements 111
The British Movement 113
Proposals and Suggestions 117

8 Shoves and Budges 128


Coercive Paternalism 130
Regulating against Harms 136
The Behavioural Public Policy Cube 141

9 Give and Take 146


Other Animals 149
Anthronomics 151
The Ultimatum Game 153
Informing Public Policy with Reciprocity 157

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Frontmatter
More Information

contents ix

10 Summing Up 162
The Burgeoning Interest 163
Embracing Evolution 166
The Policy Approach 168
Human Motivation 171
The Future 174

Note on References 177


Bibliography 179
Index 192

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Frontmatter
More Information

Figures

1.1 Declining marginal utility page 4


1.2 A Marschak-Machina triangle 11
1.3 Constructing a utility curve 13
2.1 The common consequence effect in the
Marschak-Machina triangle 19
3.1 The Markowitz utility curve 37
3.2 The prospect theory value curve 41
3.3 The prospect theory subjective probability curve 42
3.4 Prospect theory and the Marschak-Machina triangle 46
4.1 Discounting 61
4.2 Dynamic inconsistency 62
4.3 Increasing marginal utility of reducing delay 65
4.4 Time trade-off 68
5.1 An optical illusion 72
5.2 A violation of temporal dominance 76
5.3 The James Dean and Alexander Solzhenitsyn effects 77
5.4 The peak/trough effect over QALY profiles 80
6.1 Motivational crowding out 89
7.1 The nudge space 118
7.2a Health improving interventions 1–4 120
7.2b Health improving interventions 5–8 121
7.3 Energy saving interventions 125
8.1 The shove space 131
8.2 Conly’s purported shoves 133
8.3 The budge space 138
8.4 Accidental budges 140
8.5 The behavioural cube 142

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Frontmatter
More Information

Tables

2.1 The common consequence effect page 18


2.2 The Ellsberg paradox 22
3.1 Prospect theory’s reflection effect 47
3.2 Depicting regret 51
6.1 The motivations in self-determination theory 92
6.2 Hospital Quality Incentive Demonstration project 105
6.3 Pros and cons of performance management 106
7.1 The nudge requirements 112
7.2 Mindspace 114
8.1 Characteristics of behavioural public policy
frameworks 144

xi

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Frontmatter
More Information

Preface

In 2013, Cambridge University Press published a collection of essays


that I edited and to which I contributed, titled Behavioural Public
Policy. It will be no surprise to the reader that I still believe that
collection to be a useful resource and a genuine contribution to this
burgeoning but still relatively new approach to public policy analysis,
being comprised of chapters written by many of the leading and
up-and-coming behavioural specialists in the world. However, the
collection lacked a single voice, and thus I felt that there was scope
for me to contribute a companion book to that collection, a book
that you are now reading.
All of my degrees are in economics. My undergraduate degree
was a straight economics degree, I have an MSc in health economics,
and my PhD focussed on challenges to expected utility theory, which
remains the dominant theory of rational choice. Therefore, I consider
myself to be a behavioural economist, and practically all of the
empirical research and much of the conceptual work that I have done
over the years has been, I would argue, behavioural economics, more
often than not applied to health.
Therefore, it is natural I think for me to approach the topic of
behavioural public policy principally from the perspective of
a behavioural economist rather than a behavioural scientist in the
broader sense of the term; indeed, this is the only thing that I can do.
I am not dismissing or downplaying any other discipline because of
this. In fact, quite the contrary. The principal disciplinary inputs in
the development of behavioural economics have been economics and
psychology, and yet my view is that behavioural public policy ought to
develop with contributions from a much broader range of disciplines,
including, but not limited to, economics, psychology, anthropology,
sociology, political science, philosophy, animal behaviouralism,

xiii

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Frontmatter
More Information

xiv preface

history and even literature. I have tried to incorporate some ideas from
most of these disciplines into this book to the extent that I am able,
but as I said, my principal expertise is in behavioural economics and
that will probably show.

the target audience


. . . is everyone. But perhaps that is a little ambitious. When I was
a PhD student, a quite well-known professor told me that a lot of his
economics colleagues set out deliberately to make their writings as
impenetrable as possible in order to get one over on each other, and
another well-known professor informed me that most of the
behavioural economics literature was written by specialists for
specialists. I thought the latter piece of information a shame but
probably understandable, but the former state of affairs struck me as
ridiculous. I have always set out to make my own writings accessible,
even though my readership has remained minuscule, and I have
endeavoured to continue to do that in this book. That said, this is not
a popular science book. Popular science books are often entertaining,
and I suppose they serve their purpose of provoking interest in
behavioural phenomena – as well as bringing riches to their authors –
but if the reader wants a good, thorough understanding of the
behavioural field, with one or two notable exceptions, they tend not to
be very good. I have attempted to limit the use of jargon, or to explain
it in plain English where it is used, and I have tried to pitch the book
somewhere between popular science and a technical academic text.
In terms of accessibility, I had Daniel Kahneman’s Thinking Fast and
Slow in mind, and thus I hope the book will be read by interested
laypersons, students from the undergraduate level upwards – and thus
it can be used as a teaching resource – a broad multidisciplinary range
of academics, and policy makers. So, almost everyone.

how to approach the book


Ideally, I would like readers to keep policy in mind as they read the
book, whether they are reading the theoretical, conceptual or more

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Frontmatter
More Information

preface xv

directly policy-related parts of it. When you know what behavioural


economics entails – when you are learned in the concepts presented in
this book – you will see it at work everywhere, and you will notice that
it is useful for designing, evaluating, critiquing and applying policy in
any sector that interests you. I do not expect anyone to agree with
everything that I have to say, but my intention is not to convince. It is
to offer food for thought, as well as provide a teaching resource.
If I provoke thought, whether or not people ultimately agree with me,
then I would have done my job.
Although I have summarised a rich body of argument and
evidence in the book, I do not pretend to have covered everything that
is relevant to behavioural public policy. There are gaps in the
knowledge presented, both in relation to the topics that are left
uncovered and in relation to an incomplete coverage of many of the
topics that are included. I have, however, provided a foundation on
which readers can build, to fill any gaps that they identify and to
develop some ideas of their own. If the reader absorbs the whole book,
it is, I like to think, a solid intellectual foundation for the development
of informed assertions.

notes on style
My main aim in the book is to pull together many of the ideas that
I have been working on over the past ten years or so in order to
demonstrate the interconnectedness of the issues that, at face value,
might seem quite disparate within the field of behavioural public
policy.
I am English, and therefore I have used British spelling
throughout the book, but I have attached the $ sign to all numerical
examples that require a currency indicator, because that is the
convention in modern behavioural economics. Moreover, I recognise
that women are at least as important as men, but rather than writing
he/she and him/her throughout the text, for brevity and consistency,
when the occasion calls for it, I have used the terms he and him.
She and her would work just as well. I flipped a coin to decide which to

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Frontmatter
More Information

xvi preface

use, and no disrespect is intended. Furthermore, I use the terms lottery


and gamble interchangeably.
Many of my policy examples in the book will lean towards
health and health care, because that is where my knowledge tends to
lie. However, this is a book of concepts and the arguments are
essentially generic; therefore, as stated above, readers can quite readily
think about how these arguments apply to any policy sector that
interests them. I have deliberately chosen, for the most part, not to be
didactic. Rather, I have tried to present a balanced perspective of the
competing views on each topic covered. The only place where I relax
this approach is in the final chapter – Chapter 10 – where I offer my
own opinions and perspective on some of the issues theretofore
considered. This is just to give an indication of where I stand at this
point in time. I encourage – even implore – readers to take me to task.
It is perhaps worth reiterating that the book is not a systematic
literature review. I have not given all of the references for every
argument or piece of empirical evidence that I mention. To do so
would have meant that the list of references would have exceeded the
length of the main text. Enough references are, however, included to
offer a good indicative account of the topics covered. I apologise in
advance to anyone who feels slighted.

how the book is organised


The first thing to note is that the book is intentionally short. This is to
increase the chance that the wide target audience – interested
laypersons, students, multidisciplinary academics, policy makers –
will read the whole thing. It has taken me about ten years to collect my
thoughts on the content of this book. Assuming an average of one hour
per chapter, the reader can attain that knowledge in ten hours, which
strikes me as a good return on investment, assuming, of course, that
I have anything that is worth imparting.
To my mind, one cannot really be a good behavioural public
policy analyst without some knowledge of behavioural economics,
and one cannot understand behavioural economics without some

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Frontmatter
More Information

preface xvii

knowledge of rational choice theory, of which expected utility theory


is an important component. The first three chapters of the book, in
particular, provide that understanding and knowledge and, although
I have attempted to write, as far as is possible, in a non-technical,
accessible style, non-economists in particular will still have to work
reasonably hard to get through those chapters. The effort will be worth
it, as this will give the reader the grounding to speak with authority on
this subject area. That said, the chapters can be read independently
from one another if one wants to skip this important conceptual
material – but reading the whole book will offer the reader a richer
picture of how behavioural public policy emerged, and how the
various developments fit together.
As intimated above, the book starts off with how behavioural
economics – or behavioral economics – originated. There is some
dispute as to the birth date of this discipline. Some contend that the
term – and thus the discipline – surfaced in the 1980s, but I believe that
a discipline can exist without a term to define it. Neolithic man drew
pictures in caves. They didn’t call it art, but it was art nonetheless.
I would argue that modern empirical behavioural economics
was born in the early 1950s. You will learn why when you read the
book. After detailing the birth and development of behavioural
economics in the first three chapters, the next three chapters are
devoted to single issue topics that have each presented a challenge to
standard economic assumptions. These relate to perceptions relating
to time, to utility and to money. Chapters 7 and 8 move on to the
principal conceptual behavioural public policy frameworks that have
been developed mostly by behavioural economists – the so-called
nudge, shove and budge frameworks, all of which lean on the
behavioural economic phenomena that you will have been exposed to
earlier in the book. Chapter 9 considers a human motivational force –
reciprocity – that has not received the attention that it deserves in the
public policy literature, and then Chapter 10 offers an opinionated
summing up of much of what has been up to that point said.

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Frontmatter
More Information

Acknowledgements

I would like to thank Bob Sugden, who, I think in 1989, was the first
person to introduce me to behavioural economics, and has been
a source of inspiration since. I do not always agree with him, but at the
intellectual border of economics and philosophy, there are few who
can match Bob.
When I think back, I have had very few good formal teachers.
Almost everything I have learned, I have taught myself. That said, my
informal teachers – the scholars whose works I have read, including
those who I have referenced in this book – have been the best I could
have hoped for. To all of them, dead or alive, I say thanks. I’ve enjoyed
learning from you.

xviii

© in this web service Cambridge University Press www.cambridge.org


1 Assuming Rationality

Policies inspired by behavioural science have been implemented


throughout the world for many years on an ad-hoc basis, but sub-
stantive efforts to create a broad behavioural public policy approach
is a relatively recent endeavour. The main intellectual catalyst to
these efforts was a series of publications written, in large part, by
some of the world’s leading behavioural economists in the first
decade of the twenty-first century (Camerer et al., 2003; Thaler and
Sunstein, 2003; 2008). Those writings outlined conceptual frame-
works for how behavioural economics can underpin mostly soft
forms of paternalism or, in other words, non-mandatory behaviour
change. Regarding policy influence, the non-mandatory emphasis
within these policy frameworks was important, particularly in the
Anglo-American world, where the political climate was set against
further regulation and enforcement, at least according to govern-
ment rhetoric (Behavioural Insights Team, 2010). Equally important
in an era of austerity was the promise that many of these behavioural
interventions would be financially inexpensive to implement.
In terms of creating a dedicated behavioural public policy unit,
the British were the first to embrace this new approach to policy at the
central government level (Halpern, 2015). In 2010, soon after becoming
the prime minister, David Cameron established the Behavioural
Insights Team with a view to recommending policy proposals informed
by behavioural science. Similar initiatives have now been established,
or at least considered, in several other countries, including Sweden
(McDaid et al., 2014), The Netherlands (National Institute for Public
Health and the Environment, 2011), France (Oullier and Sauneron,
2010) and Denmark (Economist, 2012), as well as in much of the
English-speaking world, most notably in the United States, where the

16:43:53
02
2 the origins of behavioural public policy

Social and Behavioural Sciences Team has tested and proposed a raft of
policies, from promoting retirement security to improving college
access, health coverage, health status and energy efficiency. The
approach has a growing influence in the developing country context
also, with, for example, the World Bank focussing on this topic in its
2015 World Development Report and establishing its own behavioural
insights team, the Global Insights Initiative. Quite apart from the work
undertaken by these new behavioural public policy units, the ad-hoc
implementation of behavioural science-informed policy interventions
around the world continues apace.
The ways in which behavioural economics and, more broadly,
behavioural science might be used to inform policy is a theme that we
will return to later in this book. However, in order to understand
properly what behavioural public policy might have to offer, some
knowledge of the most robust findings of behavioural economics is
warranted. Chapter 2 will describe some of these findings, which were
a response to pre-existing assumptions of rationality in mainstream
economics. A good behavioural public policy analyst ought to have
some knowledge of these origins.

the origins of economic rationality


Economics as a formal field of study did not exist until the latter part
of the eighteenth century. Up until that time, mathematicians
assumed that when faced with a choice between two options, with
each option offering a probability of winning an amount of money,
a rational individual would choose that which offered the greatest
expected value. For example, the expected value of a lottery that offers
a 50% chance of winning $100 and a 50% chance of winning $0 is
$50 (i.e., 0.5*100 + 0.5*0), and the expected value of a lottery that
offers a 25% chance of winning $160 and a 75% chance of winning $0
is $40 (i.e., 0.25*160 + 0.75*0). Most seventeenth-century mathema-
ticians would have therefore assumed that a rational individual would
prefer the former lottery over the latter, if the prices of both lotteries
were reasonable and equal.

16:43:53
02
assuming rationality 3

In 1713, in a correspondence with the mathematician Pierre


Rémond de Montmort, however, Nicolas Bernoulli questioned whether
the assumption of expected value maximisation was always appropriate
(Bernoulli, 1738; Zabell, 1990). Bernoulli devised an ingenious game,
known as the St Petersburg paradox, to illustrate his point. The game
involves the tossing of a fair coin and the participation of an individual
who is informed that he will be paid a prize on the landing of the first
head. Bernoulli used ducats as his currency of choice, but the example
works with all denominations. Let us thus assume that the individual is
informed that he will be paid $2n for his participation in the game, where
n is the number of tosses required for the first head to land. The individual
is then asked how much he is willing to pay to play the game.
According to the principle of expected value maximisation,
the individual should be willing to pay everything he owns, because
the expected value of the game is infinite. To see this, note that the
probability of the first head landing on the first toss of the coin is 0.5, in
which case the individual is paid $21 = $2. The probability of the first
head landing on the second toss is 0.25, in which case the individual is
paid $22 = $4, and if the first head lands on the third toss, which it will
with a probability of 0.125, the payoff is $23 = $8. The expected value of
the game is calculated by summing all of the payoffs, weighted by their
related probabilities of occurrence, associated with the first head landing
on any particular toss of the coin. Numerically, this is given by:

$ð1=2Þ2 þ $ð1=4Þ22 þ $ð1=8Þ23 þ $ð1=16Þ24 þ . . .


¼ $1 þ $1 þ $1 þ $1 þ . . . ¼ $∞:

Bernoulli recognised that people are likely to be willing to pay only


quite modest amounts of money to play the St Petersburg game.
Indeed, Allais (1990) stated that the psychological value of the game
is generally less than $20, and even that may be an overstatement,
with one study showing that most people are unwilling to pay more
than $4 (Schmeidler and Wakker, 1990).
A generation later, Nicolas Bernoulli’s cousin, Daniel Bernoulli,
proposed an alternative to expected value maximisation in order to

16:43:53
02
4 the origins of behavioural public policy

Subjective value/
U
utility
U($200)

U($100)

0 $100 $200
Money

figure 1.1 Declining marginal utility


Note: The declining marginal subjective value, or utility, curve is known
as a concave utility function. It demonstrates that the utility enjoyed from
a relatively large amount will be less than double the utility of an amount
that is half as large. For example, the utility of $200 in Figure 1.1 is less
than double the utility of $100. Similarly, the utility given by an
additional, or marginal increase of, say, $5 on top of $200 will be less than
the utility given by $5 on top of $100. In standard economic theory, the
declining marginal utility curve is also assumed to apply to most goods in
addition to money. A concave utility function implies that an individual
dislikes taking risks, or is risk averse. That is, a person will sacrifice some
of the expected value of a lottery in order to receive an amount of money
for certain. For example, when faced with a fifty-fifty gamble of receiving
$200 or $0, the individual would accept an amount less than $100, the
expected value, in order to avoid facing the gamble.

accommodate the St Petersburg paradox (Bernoulli, 1738). He argued


that the subjective value of money increases at a decreasing rate and
that lotteries, rather than being evaluated in terms of their expected
value, are evaluated in terms of their expected subjective value. This
relationship between subjective value – or what is commonly referred
to in the economic literature as utility – and money is illustrated in
Figure 1.1, and has been a key assumption in the development of
economic theory over the past two centuries.

16:43:53
02
assuming rationality 5

The implication of Daniel Bernoulli’s assertion was that a


rational individual would seek to maximise expected subjective value
rather than expected value, and by conceiving the concept of what is
known as Bernoullian expected utility, he had laid the first cornerstone
for later developments in rational choice theory.

a quiet interlude
Following Bernoullian theory, the subjective value that people place on
a good is meant to reflect their strength of preference for that good. For
example, alluding to Figure 1.1, perhaps some people feel that $200 is not
twice as good as $100, but, say, 1.6 or 1.8 times better. Subjective values
that reflect strength of preference are known as cardinal utilities.
Between the latter part of the eighteenth century and the end of the
nineteenth century, some prominent thinkers, such as Jeremy Bentham
and Francis Ysidro Edgeworth, attempted to develop techniques to mea-
sure cardinal utilities that could be meaningfully compared across dif-
ferent people. Essentially, the utilities were intended to be indicators of
happiness and the measurement device was called a hedonometer, but
they did not get very far in these attempts. The declining marginal utility
curve was a central feature of utilitarianism, of which Bentham was
the founding father. Thus, the Benthamite postulate that one should
attempt to secure the greatest happiness for the greatest number was
given a moral dimension, because the utility curve suggested that this
could best be achieved by focussing society’s efforts upon the relatively
poor. However, by the beginning of the twentieth century, most philo-
sophers and economists had reached the conclusion that an accurate
quantitative measurement of interpersonal cardinal utility was impos-
sible. For example, how could one accurately compare and quantita-
tively measure the enjoyment that one person experiences from
listening to a Beethoven symphony to the pleasure that another person
gets from eating a hamburger and to the pain that yet another person
suffers from stepping on a pin?
Any development towards numerical indicators of utility was
perhaps hindered by a discursive style of political economy dominating

16:43:53
02
6 the origins of behavioural public policy

the early economics discourse. Adam Smith, in the book that he is less
well-known for among economists, had written at length on human
psychology, and made statements pertaining to, for example, present
bias, loss aversion and reciprocity (Smith, 1759). These behavioural
phenomena will be discussed later in this book, but are at odds with
several of the assumptions underlying twentieth-century rational
choice theory and have caused some to label Smith a behavioural econ-
omist (Ashraf et al., 2005). In this regard, however, Smith’s statements
were intuitive rather than empirically quantified, and his successors
over the following century tended to intuit also. For instance, Ricardo
stated that the value of a good is proportional to the cost of the labour
taken to produce it, Malthus believed that population growth would
inevitably lead to famine and John Stuart Mill and Karl Marx warned
that wages would never rise much above subsistence levels, but empiri-
cal analyses to test competing claims were lacking. Consequently, the
perceived usefulness of political economy for informing and assessing
policy was called into question.
Alfred Marshall, one of the founders of neoclassical economics,
set out to make the postulates of the discipline more testable and tested,
and in contrast to the forewarnings of Mill and Marx, demonstrated
empirically that wages were increasing over time as a consequence of
greater productivity necessitated by competition. Although Marshall
introduced mathematical rigour, he felt it important that economic
texts ought to be accessible to the layperson. Perhaps unfortunately,
Marshall’s views on this matter went unheeded, with many economists
keen to transform their discipline into something akin to a natural
science, with the quest increasingly focussed upon developing neat
models of internal consistency that were divorced from the human
experience. Following Marshall, economics was revolutionised, with
often almost impenetrable mathematical and applied empirical eco-
nomics superseding political economy as the dominant force within
the mainstream economics community.
Since most economists had turned away from the attempt to
derive cardinal measures of utility by the time Marshall was writing,

16:43:53
02
assuming rationality 7

welfare economics proceeded to be built instead upon Vilfredo Pareto’s


criterion, which did not require cardinality, a development that Bruni
and Sugden (2007) have called the Paretian turn. Pareto maintained that
an improvement in the economic organisation of society would require
at least one person becoming better off, without necessitating anyone
to become worse off (Pareto, 1971). Therefore, all that was required was
an ordinal measure of value – in other words, an indicator of improve-
ment or deterioration from the existing situation – not a cardinal mea-
sure that specified the strength of those movements. Utility theory was
not immune to the steady introduction of mathematics into econom-
ics. Between the 1920s and 1950s, a number of notable mathematicians
and mathematically minded economists, including Frank Ramsey,
Leonard Savage, John von Neumann, Oskar Morgenstern, Jacob
Marschak and Paul Samuelson, contributed towards developing sys-
tems of formal logic that prescribed how people ought to choose when
they are faced with risk or uncertainty if they want to maximise
expected utility. This movement culminated in the specification of
expected utility theory, or neo-Bernoullian theory, and lent itself to
the development of instruments with which measures of cardinal uti-
lity could, it was proposed, be elicited. The interest in measurable
utility was reborn (Camerer, 1995).

the neo-bernoullian formulation


If Daniel Bernoulli’s assumption that people should aim to maximise
their utility is correct, then people ought to obey the formal axioms –
the logical assumptions – of what is known as expected utility theory.
When we say that people ought to behave in a particular way we are
making a normative statement, as opposed to a descriptive statement,
which relates to how people actually do behave. Expected utility
theory is still today the dominant normative theory of decision mak-
ing when faced with conditions of risk and uncertainty.
At Princeton in the 1940s, the mathematician John von Neumann
and the economist Oskar Morgenstern specified most of the axioms of
expected utility theory as a small part of their seminal work on game

16:43:53
02
8 the origins of behavioural public policy

theory (von Neumann and Morgenstern, 1944). The axioms were almost
immediately recognised as important in the economics community,
not least because it proved easier to judge the intuitive plausibility of
specific axioms than the utility representation – i.e., expected utility
maximisation – that they imply. Expected utility theory was refined in
the years immediately following von Neumann and Morgenstern’s
initial exposition (Marschak, 1950; Samuelson, 1952), and its crucial
axioms are now thought to be ordering, continuity and independence
(Camerer, 1995). Although some contend that there is no broadly
accepted definition of behavioural economics (Heukelom, 2012), the
discipline is commonly thought to focus upon the set of observations
that show that people often systematically, and therefore seemingly
deliberately, violate the assumptions of rational choice theory and the
broader assumptions of standard economic theory, which we will con-
sider later. The challenges to the axioms, particularly the independence
axiom, were the origins of empirical behavioural economics, and thus
a good student of behavioural public policy ought to be familiar
with them.
Before describing the axioms, it is important to note that
although the terms risk and uncertainty are often used interchange-
ably in the popular discourse, in economics they have distinct mean-
ings. Von Neumann and Morgenstern’s axiomatic framework was
developed for decision making under conditions of risk, where prob-
abilities are objectively known. For example, there is a 50 per cent
chance that a fair coin will land heads up. Uncertainty, or ambiguity,
refers to an event where the occurrence of a particular outcome falls
within a range of probabilities. For instance, there might be a 15–30
per cent chance of rain tomorrow. Leonard Savage provided an axio-
matic framework for expected utility theory under conditions of
uncertainty, and developed what is known as subjective expected
utility theory (Savage, 1954). Strictly speaking, subjective expected
utility theory can be applied more broadly than expected utility theory
in real world settings, because specific probabilities of events are
rarely objectively known. However, Savage argued that when faced

16:43:53
02
assuming rationality 9

with mutually exclusive possible states of the world, individuals will


themselves attach a specific probability to the occurrence of each
state. Expected utility and subjective expected utility theory thus
share the same crucial axioms.
To return to the axioms, ordering imposes two requirements on
people; namely, that their preferences should be complete and transi-
tive. Completeness is simply the requirement that people are able to
express a preference between two or more goods. For example, if we
select motor cars as the relevant goods, an individual should be able to
state that he prefers a Mercedes over a BMW, or vice versa, or that
these two types of car are equally preferable to him; that is to say, he is
indifferent to the choice of car. Transitivity implies that if an indivi-
dual prefers a Mercedes over a BMW, but prefers a BMW over a Jaguar,
then he ought also to prefer the Mercedes over the Jaguar. A violation
of transitivity is known as an intransitive cycle, which can have
serious negative economic consequences for the perpetrator. For
example, assume that an individual prefers a particular Mercedes
over his own BMW, prefers the BMW over a Jaguar that he has noticed,
but also prefers the Jaguar over the Mercedes. He would therefore be
willing to swap his BMW plus pay a premium, say, $x, for the
Mercedes. He now owns the Mercedes, but he would be willing to
swap this car plus pay a premium, say, $y, for the Jaguar. He now owns
the Jaguar, but would be willing to swap that car, plus pay a premium,
say, $z, for his original BMW. Therefore, in terms of car ownership he
is back where he started, with the BMW, but has paid out $x + $y + $z
in the process. If he were to repeat this cycle, he may soon find himself
without enough money to buy petrol. In the economic literature, this
is known as a money pump, an economically irrational cycle that can
lead to bankruptcy.
Continuity requires that if an individual is faced with three
goods, there will be a specific, unique probability such that he will
be indifferent between a gamble that offers a chance of the most and
least preferable goods, and the intermediate good for certain. For
example, if money is the good being offered, and, as is likely, an

16:43:53
02
10 the origins of behavioural public policy

individual prefers $20 over $10 and $10 over $5, there will be a unique
probability, say, p, where the individual is indifferent between receiv-
ing $10 for certain, and a gamble offering p chance of $20 and (1−p)
chance of $5. As we will see later, continuity is central to the utility
elicitation instruments, but has not been subjected to much attention
in the behavioural economics literature.
Independence, sometimes call separability (Broome, 1991) or the
sure thing principle (Savage, 1954), is the most controversial axiom of
expected utility theory, and implies that the intrinsic value that an
individual places on any particular outcome will not be influenced by
varying other possible outcomes on offer, or by varying the size of the
probability of the outcome occurring. The implication of the indepen-
dence axiom is that if an individual is asked to choose between two or
more lotteries, then a common outcome that has the same chance of
occurring across the lotteries will be deemed irrelevant to the indivi-
dual when making his choice. The chance of the common outcome
occurring is a sure thing irrespective of what is chosen, and the indi-
vidual ought only to base his choice on the consequences that distin-
guish the options he faces.
Some of the main challenges to the independence axiom will be
detailed in Chapter 2, but an indication that the value that people
attach to different outcomes or goods is often dependent on the other
possibilities that are available in the choice that they are presented
with, even when there is no risk or uncertainty, is given by a phenom-
enon known as asymmetric dominance (Huber et al., 1982). For exam-
ple, a person is more likely to choose to buy a 28-inch television set for
$600 instead of a 24-inch television set for $500 when a 26-inch set for
$650 is also included in the choice set. The 26-inch set serves as a decoy,
affecting the value that people attach to the 28-inch set by making it
appear better value for money than would otherwise be the case.
Lotteries with up to three outcomes can be plotted in a
Marschak-Machina triangle (1989). The axioms of expected utility
theory have implications for the shape of the indifference loci within
the triangle, with the indifference loci depicting the preference

16:43:53
02
assuming rationality 11

p($5m)

D
E

p($1m)

0.33
A

B
0 0.5 0.75 p($0)

figure 1.2 A Marschak-Machina triangle

relationships between all of the possible lotteries. For instance, con-


sider a lottery ticket that offers three possible outcomes, $5 million,
$1 million and $0. Now look at the Marschak-Machina triangle in
Figure 1.2. The probability of winning $5 million is measured on the
vertical axis and the probability of winning $0 is measured on the
horizontal axis. For example, point A represents a lottery that offers
a 33% chance of winning $5 million and a 50% chance of winning $0.
The probability of winning $1 million in this lottery is measured by
the distance between point A and the hypotenuse; i.e. 1 − 0.5 − 0.33 =
0.17. All gambles that involve these three outcomes can be depicted in
the triangle. For instance, point B depicts a lottery that offers no
chance of winning $5 million, a 75% chance of winning $0 and
a 25% chance of winning $1 million.
The dashed lines in Figure 1.2 are the indifference loci. For any
particular individual, these loci would fill the triangle, but only a few
are drawn in the figure for illustrative purposes. If two or more lotteries

16:43:53
02
12 the origins of behavioural public policy

are on the same locus – meaning line or curve – then this indicates that
the individual whose preferences that locus represents will be indiffer-
ent between those lotteries. Thus, Figure 1.2 indicates that the indivi-
dual will equally prefer either lottery C or lottery D. Lotteries that fall
to the right of a particular locus will be less preferred than those on the
locus; for instance, for this individual, C and D will be preferred to
A and B. Lotteries that fall to the left of a locus will be preferred over
those on the locus, and therefore E will be preferred over C and
D. The slope of the loci varies from individual to individual. Those
who are relatively more risk averse will have relatively steeper loci.
However, the axioms of expected utility theory require that the
shape, as opposed to the slope, of the indifference loci across the
triangle be the same for all individuals. Completeness requires that
there be no breaks in any individual locus, transitivity requires the
loci not to intersect within the triangle, continuity results in the loci
being very, very thin and independence implies that the all of the loci
across the triangle will be parallel straight lines. Those drawn in
Figure 1.2 therefore satisfy expected utility theory. In Chapter 2, we
will see that these conditions do not always hold, but let us not jump
ahead of ourselves. As alluded to earlier, the axioms informed the
development of instruments designed to elicit the theretofore elusive
measures of cardinal utility.

utility elicitation
Referring back to the numbers given in Figure 1.1, imagine that an
individual is presented with a lottery that offers a 50 per cent chance of
winning $200 and a 50 per cent chance of winning $0. We then ask the
individual to state the amount of money, guaranteed with certainty,
that would be equally preferable to him as playing out the lottery. This
guaranteed money amount is usually called the individual’s certainty
equivalence of the lottery. If the individual does not like risk, he will
state an amount less than the expected value of the lottery, which is
$100. Assume his certainty equivalence is $90. Now, because $200 is
the best outcome available, and $0 is the worst outcome possible, we

16:43:53
02
assuming rationality 13

Utility

1.0

0.5

0.25

0 $42 $90 $200


Money

figure 1.3 Constructing a utility curve

can assign these two money outcomes utilities of 1 and 0, respec-


tively, on a zero to one utility index. This means that the utility of
$90 is equal to 0.5*1 + 0.5*0 = 0.5. These utility values for $0, $90 and
$200 are depicted in Figure 1.3.
The next step is to repeat the task after changing the lottery to
a 50% chance of winning $90 and a 50% of winning $0. If the indivi-
dual is risk averse, he will state a certainty equivalence of less than the
expected value of $45; let us say $42. For this individual, the utility of
$90 was calculated at 0.5 and thus his utility of $42 is 0.5*0.5 + 0.5*0 =
0.25. These steps could be repeated ad nauseam to create a whole
series of dots in Figure 1.3, which, if joined up, would look something
like Figure 1.1.
There is a theoretically identical alternative to the certainty
equivalence method called the probability equivalence method.
With this method, using outcomes similar to those above, one might
say to an individual that he could either accept, say, $100 for certain or

16:43:53
02
14 the origins of behavioural public policy

play a lottery that offers a probability, p, of winning $200 but that also
presents the possibility, 1−p, of winning $0. The individual is then
asked for the p required for him to be indifferent between the certainty
and the lottery. If the individual states that he would require a 60%
chance of a positive payoff from the lottery, then again normalising
the utilities of $200 and $0 at 1 and 0, respectively, the utility of $100
for this particular individual = 0.6*1 + 0.4*0 = 0.6.
The best and worst outcomes have to be normalised, typically at
one and zero, at the ends of the utility scale, but, of course, these do
not have to be $200 and $0. They can be set at any level that is deemed
appropriate for the investigation that is being conducted. They do not
have to be money outcomes and can, in principle, be outcomes that
suit any policy domain. In health policy, for instance, the outcomes
can be states of health, with the endpoints defined as full health for
a statistical lifetime and immediate death. Indeed, in health economic
evaluation, which is a method used to estimate the value for money of
health programmes and interventions and even the productivity of
whole health care systems, the probability equivalence instrument is
one of the ways in which units of benefit, called quality adjusted life-
years or QALYs, are typically derived. In this context, the probability
equivalence technique is called the standard gamble. These instru-
ments are therefore far more than theoretical curiosities.
For assessing the utility of outcomes for policy purposes, the
probability equivalence method is more convenient than using cer-
tainty equivalences, because the aim is to elicit utilities for predefined
outcomes of interest, rather than to try to elicit an outcome for
a predefined utility. For instance, in health policy the focus of the
analyst will be on deriving the utility of angina, for example, rather
than trying to ascertain which health state has a utility of 0.5.
Nonetheless, as aforementioned, the methods should, in theory, give
identical utilities for any particular outcome, and both are under-
pinned by the axioms of expected utility theory. That is, for the
instruments to work as intended, people have to be able to compare
outcomes, they have to have a fixed idea of the ordering of the

16:43:53
02
assuming rationality 15

outcomes from best to worst, the instruments must be direct applica-


tions of the continuity axiom and the utility that an individual places
on each and every outcome has to be independent of other outcomes
and of the probabilities in the decisions that they face.
Long after the development of expected utility theory, evidence
emerged that the certainty equivalence and probability equivalence
instruments gave substantially and substantively different utilities
for the same outcomes, more so than could be attributed to random
error (Hershey and Schoemaker, 1985; Johnson and Schkade, 1989).
Specifically, the latter of the two methods produces higher utilities
than the former, which, in the behavioural economics literature, is
known as a violation of procedural invariance. The reason for these
discrepancies is that although people ought to adhere to the axioms of
rationality if they are expected utility maximisers, they often do
not do so. This was a fact known for decades by anyone who would
have cared to listen. Indeed, empirical contradictions of the neo-
Bernoullian axioms were published almost before the ink that had
been used to formulate them had been given enough time to dry. With
these violations, empirical behavioural economics was born; it is to
these that we now turn.

16:43:53
02
2 Challenging Rationality

The early 1950s saw the publication of the first violations of expected
utility theory, targeted against the independence axiom and the eco-
nomic assumption of a calculated maximisation of expected utility in
individual decision making. In truth, expected utility theory was origin-
ally postulated as a normative and not a descriptive theory by many
economists. For instance, Friedman and Savage (1948) wrote that they
did not believe that people meticulously assign utilities to each outcome
and then weight those utilities with associated probabilities; rather
the assumption was merely that an individual made decisions as if
they had undertaken these complicated calculations. Howard Raiffa
(1961, pp. 690–691) went even further by stating that expected utility
theory

is not a descriptive or predictive theory of behavior. It is a theory


which purports to advise any one of its believers how he should
behave in complicated situations . . . If most people behaved in a
manner roughly consistent with [the] theory, then the theory would
lose a good deal of its normative importance. We do not have to
teach people what comes naturally. But as it is, we need to do a lot of
teaching.

Nonetheless, perhaps corresponding with the rise of some members of


the Chicago School of Economics in the late 1950s, there was a blur-
ring of the lines between normative and descriptive assumptions in
economics, with most economists believing that any observed viola-
tions of the axioms of expected utility theory could be attributed to
random error (Savage, 1954).
This chapter will detail some of the main objections that were
made to expected utility theory during the first phase of the development

16

22:36:57
03
challenging rationality 17

of empirical behavioural economics up to the mid-1970s. It will


consider the particular dislike – more than can be encapsulated by the
axioms of rational choice theory – that many people feel towards
uncertain situations, and the at-face-value paradoxical finding that peo-
ple appear to sometimes change or reverse their preferences, depending
on whether they are asked to choose directly between goods or whether
they are required to place money values on each good, one at a time.
Related to this, some of the many heuristics, or rules of thumb, that
people employ when making decisions will be outlined, which may be
used because people often do not maximise or optimise utility, but
rather satisfice, or in other words, choose things that they believe are
just about good enough. However, we will begin at the beginning, with
Maurice Allais’ paradox.

the allais paradox


Prior to the development of expected utility theory, the French econ-
omist Maurice Allais had been working on some thought experiments
that challenged formal logic (Allais, 1990). Following the publication
of, and early additions to, von Neumann and Morgenstern’s work,
Allais recognised that his earlier ideas placed a question mark against
the validity of the independence axiom, at least in a descriptive if not a
normative sense. His ideas were seminal (Allais, 1953). To illustrate,
consider the following four options, A, B, A* and B*:

A: $1 million for certain


B: 10% chance of $5 million; 89% chance of $1 million; 1% chance of $0
A*: 11% chance of $1 million; 89% chance of $0
B*: 10% chance of $5 million; 90% chance of $0

Allais argued that if individuals are faced with a choice between A and
B they will show a tendency to choose A. He further argued that if
offered a choice between A* and B*, the tendency would be to opt for
B*. However, this preference pattern – A and B* – known as the
common consequence effect, violates the independence axiom. Look
at Table 2.1 to see why.

22:36:57
03
18 the origins of behavioural public policy

Table 2.1 The common consequence effect

Chance

89% 10% 1%

A $1 m $1 m $1 m
B $1 m $5 m $0
A* $0 $1 m $1 m
B* $0 $5 m $0

Table 2.1 has broken down A, B, A* and B* a little, but these


lotteries remain exactly the same as previously given. For instance, in
the table, B* is an 89% chance of $0, a 10% chance of $5 million and a
1% chance of $0, or, in other words, a 10% chance of $5 million and
a 90% chance of $0, as stated earlier. It is clear to see that A and B offer
a common outcome of an 89% chance of $1 million, and A* and B*
share an 89% chance of $0. If the independence axiom holds, the
chooser should implicitly consider these common consequences as
irrelevant to his decisions. If he does, A is identical to A* and B is
identical to B* and therefore expected utility theory requires the
individual to choose A and A* or B and B*, or to be indifferent in
both choices. Allais’ conjecture that many people will choose A and
B*, or, in other words, that they will tend to prefer security in the
neighbourhood of certainty when considering sums of money that are
large in relation to their capital, has been confirmed in subsequent
empirical analyses (Camerer, 1989; Conlisk, 1989, Slovic and
Tversky, 1974), where it has been suggested that up to 60% of people
will demonstrate these preferences. Note also that far more people
choose A and B* than B and A*. This means that the violations of
independence are systematic – that is, they cannot be attributed to
random error, a feature that applies to many of the most robust find-
ings in empirical behavioural economics.
The common consequence effect is diagrammatically depicted
in Figure 2.1. Since A and A*, and B and B*, are identical to each other

22:36:57
03
challenging rationality 19

p($5m)

p($1m)

B B∗

A
0 A∗
p($0)

figure 2.1 The common consequence effect in the Marschak-Machina


triangle

after the common outcomes are discarded, straight lines drawn


between A and A*, and B and B*, would be parallel lines. If these
lines were the individual’s indifference loci they would satisfy the
axioms of expected utility theory. However, if A is preferred over B,
then B falls to the right of the indifference locus that A sits on, as
illustrated in Figure 2.1. Similarly, if B* is preferred over A*, as the
common consequence effect commands, then A* falls to the right of
the indifference locus on which B* is placed. As can be observed in the
figure, these preferences imply that the indifference loci fan out across
the triangle.
The formulation of the Allais paradox is abstract, but its
demonstration that people dislike risk when certainty is an option,
to an extent far greater than that assumed by standard rational choice
theory, has important implications in terms of public attitudes
towards a host of policy considerations. As a hypothetical example,

22:36:57
03
20 the origins of behavioural public policy

imagine that a government is developing a synthetic biology that


is intended to more effectively combat malaria. The current malaria
strategy, in the absence of the synthetic biology, will save, say,
one million lives over the next year. The synthetic biology, on the
other hand, has a good chance of saving three million lives, but is
associated with a small chance of costing lives such that no one will
be saved. If people demonstrate a large degree of risk aversion when
the perceived certainty of the status quo is an option, there could be
widespread reluctance to implement policies that, while entailing a
small element of risk, will in almost all likelihood facilitate progress.
If certainty is not an option – that is, if every possible strategy pre-
sents, and is perceived to present, an element of risk – the common
consequence effect suggests that a broad reluctance to engage in
riskier but potentially more rewarding strategies will be lessened.
This is not to argue that the riskier strategy is necessarily the best
strategy. It is merely to point out that there may sometimes be a
seemingly strong dislike of risk – which some might believe is
detrimental for public policy and stifles personal decision making –
when certainty, which is often associated with the status quo, is an
option.
More specifically, the instruments, discussed at the end of
Chapter 1, that have been developed to elicit utilities and which are
these days used to inform real policy decisions, ask people to trade off
risk with certainty. These are exactly the circumstances where Allais-
type violations of rational choice theory tend to be most pronounced.
This might be a reason for the discrepancies that have been observed
between the certainty equivalence and probability equivalence ver-
sions of the instruments, and serves to undermine the confidence that
we might otherwise have in the validity of the utilities produced by
these methods.
In addition to the common consequence effect, Allais outlined
another paradox, called the common ratio effect, in his 1953 paper.
Consider the options A, B, A* and B* below. Allais hypothesised that
many people would choose A over B and B* over A*, a conjecture

22:36:57
03
challenging rationality 21

supported by subsequent empirical analyses (Chew and Waller, 1986;


MacCrimmon and Larsson, 1979).

A: $1 million for certain


B: 80% chance of $5 million; 20% chance of $0
A*: 20% chance of $1 million; 80% chance of $0
B*: 16% chance of $5 million; 84% chance of $0

However, A* and B* can be rewritten as:

A*: 20% chance of A; 80% chance of $0


B*: 20% chance of B; 80% chance of $0

A* and B* therefore share the common outcome of an 80% chance of


winning $0, which, if independence holds, ought to be considered
irrelevant to an individual’s decision. If so, the relevant aspects of
A* and B* are an equal chance – that is, a 20% chance – of A or B.
Consequently, if the individual prefers A in the first choice task,
expected utility theory postulates that he should choose A*; if he
chooses B, he should choose B*; or he should be indifferent in both
decisions. The common observation that people often choose A and B*
is another violation of rational choice theory, caused by seemingly
heavy risk aversion in the face of certainty, that again implies a fanning
out of the indifference loci across the Marschak-Machina triangle.
As already noted, Allais’ paradoxes were very important to the
development of behavioural economics, but by the early 1960s it
became clear that he did not have a monopoly on contradicting the
axioms of rational choice.

the ellsberg paradox


In the early 1960s, an article by Daniel Ellsberg was published that
focussed upon a thought experiment that he had presented originally
at a dinner party (Ellsberg, 1961). The first lesson from this is that one
should be very careful when issuing and accepting dinner invitations;
the second lesson is that the independence axiom, through its incar-
nation as the sure thing principle, is as vulnerable to circumstances of

22:36:57
03
22 the origins of behavioural public policy

Table 2.2 The Ellsberg paradox

Black Red Blue

30 balls 60 balls

A $100 $0 $0
B $0 $100 $0
A* $100 $0 $100
B* $0 $100 $100

uncertainty as it is to those of risk. To see why, take a look at


Table 2.2.
Assume that an individual is presented with a black bag that
contains ninety balls, and is told that thirty of the ninety balls are
black, and the remaining sixty are either red or blue, but that we are
unsure exactly how many of them are red and how many are blue.
That is, all of the sixty balls could be red, all could be blue, or there
could be any combination of red and blue balls between those two
extremes. The individual is asked to place their hand in the narrow
opening at the top of the bag, and without seeing any of the balls,
remove one of them from the bag. Before doing this, the individual
chooses either option A or option B outlined in Table 2.2. If he chooses
A and picks a black ball from the bag, he will be paid $100, but he will
get nothing for a red or a blue ball. Option A therefore offers a risky
33% chance of $100. If he chooses B he will get $100 for a red ball and
nothing for a black or a blue ball, and thus B offers an uncertain zero to
66% chance of $100.
In addition to choosing between A and B, the individual is asked
to repeat the process but to choose between A* and B* instead. If he
chooses A* he will receive $100 if a black or blue ball is retrieved from
the bag. The odds that this will occur are an uncertain 33% to 100%.
Alternatively, if B* is chosen, the payoff of $100 is consequent on a red
or blue ball being picked, of which there is a 66% chance. In these two
choices, Ellsberg hypothesised that many people will tend towards

22:36:57
03
challenging rationality 23

choosing A and B*, for which there is now substantial empirical


support (Bernasconi and Loomes, 1992; Curley and Yates, 1989;
Einhorn and Hogarth, 1986).
Now, even a cursory glimpse at Table 2.2 reveals that one would
win nothing if a blue ball is picked, irrespective of whether one chose
A or B. Similarly, a blue ball is associated with a payoff of $100 in both
options in the choice between A* and B*. Therefore, according to the
independence axiom, or sure thing principle, an individual, when
choosing between A and B, and A* and B*, should view the blue ball
consequences as irrelevant. If he does, then A is identical to A* and B is
identical to B*. Thus, subjective expected utility theory requires an
individual to choose A* if he chose A, B* if he chose B, or to be
indifferent in both choices. The Ellsberg preference pattern is not
allowed.
The explanation for the Ellsberg paradox is that people tend not to
like the uncertainty embedded in B and A*. A and B* are risky options
where the chances of winning $100 – 33% and 66%, respectively – are
known objectively. As noted above, the chances of winning $100 in
B are anywhere between zero and 66%, and the chances offered by
A* are between 33% and certainty. People dislike uncertainty. That
is, they are uncertainty, or equivalently ambiguity, averse, more so than
is allowed by rational choice theory. Indeed, people will pay a premium
to avoid ambiguity, which has been estimated at 10–20% of the
expected value of a gamble (Bernasconi and Loomes, 1992; Curley and
Yates, 1989).
Ambiguity aversion may well have profound policy implica-
tions, both when expressed implicitly by policy makers themselves
and when they are forced to respond to ambiguity aversion within the
general population. To take just one example, in 2009 there was a
global outbreak of the H1N1 virus, otherwise known as swine flu. The
effects of the virus were initially highly uncertain, but at least some
governments resourced their responses to the outbreak according to a
worst case scenario, and continued to do so when it became apparent
that the worst case was highly unlikely to occur. That is, they

22:36:57
03
24 the origins of behavioural public policy

appeared to be paying a large premium in order to mitigate the possible


effects of the virus, even though it was unknown and ultimately
unlikely that these effects would be realised. Now, many, perhaps
most, might view such a response as acceptable and even necessary,
but policy makers are in a position where they ought to recognise that
their actions in responding to a perceived threat divert resources away
from other, possibly more beneficial, uses. This is not to argue that
policy action in the face of the swine flu outbreak was definitely
wrong, but it is to point out that assuming the worst case scenario
rather than a more balanced assessment of likely occurrences when
faced with unavoidable uncertainty – which may be an explanation for
the Ellsberg paradox – could lead to poor policy decisions. It is not hard
to imagine other areas of policy that could be susceptible to ambiguity
aversion, from responses to terrorist threats, to immigration policy, to
climate change policy, to decisions to invest in synthetic biology, to
name but four.
The main lessons from the Allais and Ellsberg paradoxes com-
bined are that people tend to dislike risk when certainty is an option,
and to dislike uncertainty when risk is an option, more so than is
allowed by standard rational choice theory, as aforementioned. The
reader may have noticed a further apparent paradox here, in that these
effects might cause policy makers in some situations to not do
enough, if certainty is associated with the status quo, and yet in
other circumstances it might provoke them to do too much, in order
to try to mitigate uncertainty. It is not uncommon, though, for the
various behavioural phenomena that will be discussed throughout
this book to push people, and even a single policy response, in different
and sometimes directly opposing directions.
In the same era that the economists, Allais and Ellsberg, were
publishing their challenges to the expected utility model, which were
incidentally largely disregarded by the mainstream economics com-
munity – the l’Ecole Americaine, according to Allais – for decades, the
polymath, Herbert Simon, attacked rational choice theory from a
slightly different angle. His work was built upon implicitly by a

22:36:57
03
challenging rationality 25

group of psychologists from around the end of the 1960s onwards,


work that ultimately converged with the economists’ earlier concerns
to form the bedrock for later theoretical and empirical progress in
behavioural economics in the late 1970s and thereafter.

satisficing and rules of thumb


In contrast to the assumption that people should and will maximise
utility, which is held by many who adhere to neoclassical economic
theory and the standard model of rational choice, Herbert Simon
contended that people are not optimisers, but satisficers (Simon,
1956). Humans do not have either the capacity to engage in the calcu-
lations or all of the information that is necessary to optimise in all
circumstances. As mentioned earlier, Friedman and Savage (1948) also
noted that people will not undertake all of the calculations that utility
maximisation implies, but that their final decisions will nonetheless
still be consistent with this goal; that is, it would be as if they had
undertaken the calculations. Simon’s approach departs from the ‘as if’
assumption. People face too many decisions each day to be able to
maximise their expected utility. Therefore, humans have developed
heuristics, or in other words, relatively simple rules of thumb to guide
them in reaching their decisions. Simon’s main point was that people
use rules of thumb to reach satisfactory, not optimal, decisions. For
example, a couple on choosing a film to watch on Netflix on a Friday
evening may choose the first movie that both find tolerable, rather
than search through the dozens of options available to try to find a film
that both might expect to enjoy more. Hence, although some may
contend that the search costs outweigh any additional expected uti-
lity, they satisfice.
From the early 1970s, psychologists, in particular Daniel
Kahneman and Amos Tversky, began publishing evidence on the
rules of thumb, or mental short cuts, that people use when reaching
decisions. A great many heuristics are now reported in the literature;
Judgment Under Uncertainty is a good, relatively early reference
source (Kahneman et al., 1982). For our purposes, it suffices to outline

22:36:57
03
26 the origins of behavioural public policy

just some of the most prominent of the short cuts. Most of these were
summarised by Tversky and Kahneman in 1974, although some are of
far older vintage.
Status quo bias, that people are reluctant to move from the
position they are already in, has already been touched upon in this
chapter, and is mentioned again here due to its importance in inform-
ing some aspects of behavioural public policy, which will be dis-
cussed later in this book. Representativeness is the finding that
people may overlook objective probabilities, and resort more to pre-
conceptions or stereotyping, which can be illustrated with an exam-
ple. Imagine that we ask a person to think about a group of one
hundred men, and are told that eighty of these men are engineers
and twenty are librarians. The individual is then told that one of the
men, called Steve, is very shy and withdrawn, invariably helpful but
with little interest in people or the world of reality. Steve is a meek
and tidy soul, with a need for order and structure and a passion for
detail. Our respondent is then asked whether they think Steve is
more likely to be an engineer or a librarian. Tversky and Kahneman
(1974) report that people have a tendency to answer that they think
Steve is a librarian because his profile fits the stereotype of that
profession, and they overlook the fact that, statistically, Steve is
much more likely to be an engineer.
Availability is the tendency for people to assess the probability
of an event by the ease with which similar instances can be brought to
mind. For example, people tend to find it easier to think of words in the
English language that begin with the letter R than those that have R as
their third letter, even though the latter are more numerous. A further
example is that people tend to believe, erroneously, that sharks kill
more people each year than falling coconuts, principally because
deaths by sharks receive more media attention. Indeed, availability
bias suggests that the media can exercise a large and not necessarily
objectively accurate influence on public perceptions, which, in
democracies, could place pressure on governments to distort policy
priorities to detrimental effect.

22:36:57
03
challenging rationality 27

Anchoring, which, as we will see, is an important heuristic in


the development of behavioural economics, is the observation that
people often place a heavy emphasis on particular prominent or salient
features of goods or outcomes, or even on entirely irrelevant prompts,
which can cause great malleability in their valuations of goods. For
instance, Ariely et al. (2003) show that the amount of money that
people are willing to pay for familiar products like bottles of wine can
be influenced substantially by getting them to think about irrelevant
pieces of information, such as digits in their social security numbers.
Anchoring on particular characteristics of goods can cause other
important but less salient characteristics to be unduly overlooked.
Overconfidence bias is self-explanatory, and could cause people
to make rash decisions or reach erroneous conclusions, particularly in
areas where they feel they have a good deal of expertise. Ask a group of
motorists whether they believe that they are better than average
drivers, for example, and their responses will be quite predictable,
even though, in a large enough group, roughly half of those present
cannot be better than average. Unfortunately, overconfidence bias can
have catastrophic effects, particularly if a sector is insufficiently regu-
lated, as witnessed among borrowers and investors in the run up to the
2008 financial crisis.
The final rule of thumb that we will consider here is confirma-
tion bias, where people search for information that supports the opi-
nions that they already hold, rather than strive for an objective view
based on available evidence. Scholars have written of confirmation
bias for many generations. Karl Popper (1957, p. 124), for instance, in
The Poverty of Historicism wrote that

if we are uncritical we shall always find what we want: we shall


look for, and find, confirmations, and we shall look away from, and
not see, whatever might be dangerous to our pet theories. In this
way it is only too easy to obtain what appears to be overwhelming
evidence in favour of a theory which, if approached critically, would
have been refuted.

22:36:57
03
28 the origins of behavioural public policy

The focus upon weapons of mass destruction in Iraq not so very long
ago was perhaps a good example of confirmation bias in action at the
very highest levels of Western political leadership.
Although the mental shortcuts are commonly labelled as biases
in the literature, they have probably evolved for good reasons and may
serve us well much of the time. That they can sometimes cause
mistakes in our decision making is where the error label is perhaps
legitimate, however, and therefore I will stick to that terminology
here. Kahneman and Tversky were, among much else, pioneers in
demonstrating that we often change our preferences when faced
with essentially the same options following a simple alteration in
the description of a choice problem, and the change in preference is
probably driven by the rules of thumb pushing us in different direc-
tions depending on how the choice is described. In standard economics
and rational choice theory, our preferences are assumed to be fixed and
stable, but these changes in preference suggest that they are instead
often constructed according to how the choice is described. Changes
in preferences such as this are commonly called framing effects or,
more formally, violations of descriptive invariance. Countless exam-
ples of this phenomenon can be found in the literature, but for illus-
trative purposes, and since it fits together nicely with our earlier
reference to swine flu, let us consider one of the most famous.
Kahneman and Tversky (1981) asked respondents to imagine
that the United States is preparing itself for the outbreak of a disease,
let us say swine flu, that is expected to kill 600 people. They then
asked their respondents to choose between the following alternative
programmes, A and B, to combat the disease:

A: 200 people are saved


B: 1/3 probability that 600 people are saved, and a 2/3 probability that no one
is saved

The same scenario was described to a different group of respondents,


but they were asked to instead choose between the programmes
A* and B*:

22:36:57
03
challenging rationality 29

A*: 400 people will die


B*: 1/3 probability that no one will die, and a 2/3 probability that 600 people
will die

Now, the first choice problem, A versus B, is identical to the second


choice problem, A* versus B*, the only difference being in the way
that the programmes are described. Specifically, the first choice is
framed in terms of those who will be saved, and the second in terms
of those who will die. Therefore, according to the assumption of
descriptive invariance – that a simple alteration of the description of
the choice problem should not cause preferences to vary – we would
expect roughly the same proportion of respondents to prefer A and A*,
and thus B and B*. Kahneman and Tversky found that large majorities
of their respondents – respectively, 72% and 78% – preferred A and B*.
Public health officials respond in the same way. Thus, when the
choice is framed in terms of gains, or lives saved in this case, respon-
dents have a tendency to prefer the certainty presented by programme
A. That is to say, they tend to be risk averse. However, risk loving or
risk seeking behaviour is far more evident when respondents are faced
with A* and B*, where the choice is framed in terms of losses. There
may be evolutionary reasons for why people are risk averse when they
are faced with gains or abundance, and risk seeking when faced with
losses or scarcity, but a good case can be made to say that it is
irrational for these different risk attitudes to cause a change in prefer-
ences over choices that are essentially identical.
Framing effects, as we have seen, are violations of descriptive
invariance, but, as noted earlier, another type of invariance – procedural
invariance – is also assumed in standard economics. Procedural invar-
iance is the requirement that a person’s preferences remain the same,
irrespective of the procedure that we use to elicit them. For example, if a
person, when buying an apple, chooses a Pink Lady over a Granny
Smith we would expect him to place a higher money value on the Pink
Lady and, in a ranking exercise to place the Pink Lady higher than the
Granny Smith on a scale. At around the same time that Kahneman

22:36:57
03
30 the origins of behavioural public policy

and Tversky started working on the mental shortcuts, close colleagues


of theirs published material that demonstrated that procedural invar-
iance cannot always be taken for granted.

classic preference reversals


Around 1970, the psychologists, Sarah Lichtenstein and Paul Slovic,
among others, published a number of papers demonstrating a violation
of procedural invariance that I shall call classic preference reversals
(Lichtenstein and Slovic, 1971; 1973; Lindman, 1971). In their studies,
Lichtenstein and Slovic offered people a choice between two lotteries,
termed the P-bet and the $-bet. The P-bet offered a high probability of
winning a modest amount and the $-bet offered a modest probability
of winning a relatively large amount. The two bets have similar
expected values. To illustrate, consider the following two bets, taken
from Lichtenstein and Slovic (1971):

P-bet: 35/36 chance of winning $4; 1/36 chance of losing $1


$-bet: 11/36 chance of winning $16; 25/36 chance of losing $1.50.

Lichtenstein and Slovic asked their respondents to choose directly


between the two bets, and also got the respondents to value each of
the bets in terms of how much they would sell them for if they
owned them. It turned out that the majority of respondents – in
some studies as high as 80% – chose the P-bet over the $-bet, but
placed a higher money value on the $-bet. As with the Allais and
Ellsberg paradoxes, it is important to emphasise that preference
reversals are not only substantial but are also systematic. That is,
people tend to choose the P-bet and value the $-bet higher far more
frequently than choose the $-bet and value the P-bet higher. Thus,
this violation of procedural invariance cannot be attributed to ran-
dom error.
So, what have classic preference reversals been attributed to?
One suggestion is that people may have intransitive preferences
(Sugden, 1992). For example, suppose that an individual chooses
the P-bet over the $-bet above, but values the P-bet at $2.50 and the

22:36:57
03
challenging rationality 31

$-bet at $3.50. This implies that in a series of pairwise choices,


he would prefer the $-bet over $3, $3 over the P-bet, and the P-bet
over the $-bet, an intransitive cycle. However, by asking people to
choose between a money amount and the $- and P-bets – for example,
in the above example, by asking them to choose directly between $3
and each of the bets in turn – rather than inferring their preferences in
the direct choices from their answers to the valuation questions,
it has been observed that intransitivity can account for only
10–20% of observed preference reversals (Loomes et al., 1989;
Tversky et al., 1990).
The most plausible explanation for classic preference reversals
is that people use different rules of thumb across choice and valuation
tasks. Slovic and Lichtenstein (1968) noted that choice tasks encou-
rage a greater focus on the probability of winning, which favours the
P-bet, while valuation tasks tend to focus attention on the payoffs,
which favours the $-bet. That the different elicitation procedures
might cause people to focus on different aspects of the bets is known
as contingent weighting (Tversky et al., 1988). Extending this argu-
ment a little further, many have suggested that a likely reason for why
the $-bet tends to be valued higher than the P-bet is because, as a
starting point when valuing the $-bet, people anchor – remember
anchoring? – on its best outcome, but then fail to adjust the overall
value of the gamble downwards sufficiently to take account of the
large chance of a poor outcome (Lichtenstein and Slovic, 1971). This is
compounded by the elicitation mode – i.e., money valuation – causing
a focus upon the aspect of the bet that relates to this mode – i.e., the
money payoff – a phenomenon that has been termed scale compat-
ibility (Slovic and Lichtenstein, 1983). The suggestion is that all of this
causes an overvaluation of the $-bet.
If people rely on anchoring in much of their decision making
behaviour, they will be susceptible to manipulation, whether this be
with good or bad intentions. Furthermore, if, as people, policy
makers overweight the possible benefits of new or unfamiliar pro-
jects that do not have a ready comparator, such as bridges, tunnels

22:36:57
03
32 the origins of behavioural public policy

and the hosting of the Olympic Games, they may find themselves
investing in initiatives that perhaps do not offer the best uses of
their resources. These effects are compounded when the rules of
thumb push in the same direction, as might be the case, for instance,
if policy makers search only for evidence that confirms their
optimism.
However, the mainstream economics community was suspi-
cious of the psychologists’ work on preference reversals, and ques-
tioned whether their studies had been appropriately designed and
incentivised. In 1979, the economists, David Grether and Charles
Plott, reported a study that they had designed specifically with the
intention of disproving the psychologists’ results. They contended
that they had corrected for what they claimed were flaws in the earlier
study designs by, for example, including real financial incentives.
Moreover, by allowing their respondents to express indifference in
the choice tasks, they did not force them to make a choice between
the bets, and they removed the motivation for respondents to hedge
their bets by informing them that at the end of the experiment only
one of the questions, chosen randomly, would be played out for real.
The results showed that preference reversals were evident to an extent
that, if anything, exceeded that which had been observed by the
psychologists.
By the end of the 1970s and through the 1980s – stimulated in
part by Grether and Plott’s findings – a relatively small group of
economists and psychologists produced a body of theoretical and
empirical work that lent growing respectability to behavioural eco-
nomics, even if they were still somewhat marginalised within the
mainstream economics community. Good reviews of this body of
work can be found in Camerer (1995), Camerer and Loewenstein
(2003) and Thaler (1994), and Thaler (2015) has also written an acces-
sible account of the development of behavioural economics in the
United States over that decade. This respectability had not come
easily, and had taken decades to achieve. In 1990, for instance, Allais
wrote that for ‘nearly 40 years the supporters of the neo-Bernoullian

22:36:57
03
challenging rationality 33

formulation have exerted a dogmatic and intolerant, powerful and


tyrannical domination over the academic world; only in very recent
years has a growing reaction began to appear’ (1990, p. 8). An impor-
tant part of this reaction was the development of several alternatives
to expected utility theory as descriptive theories of choice, and it to
those that we now turn.

22:36:57
03
3 Describing Risky Behaviours

The assumptions of expected utility theory have been challenged


empirically for more than sixty years, and therefore we can conclude
with some confidence that the theory does not describe human deci-
sion making particularly well, at least in some circumstances. This
conclusion is consolidated when we remember that these challenges
have tended to be systematic, and therefore cannot easily be attributed
to random error in people’s choices. However, economists, even most
behavioural economists, do not believe that this undermines the
normative status of expected utility theory, even if they are sympa-
thetic to the descriptive challenge. Given that respondents tend to
stick to their choices in Allais and preference reversals decision tasks
after they have been informed about exactly what they have done
(Slovic and Tversky, 1974), and thus appear to have deliberate reasons
for their preferences, one could speculate whether the normative faith
in expected utility theory is entirely well founded.
Be that as it may, several descriptive theories that attempt to
allow for the choice patterns outlined in Chapter 2 were developed
between the late 1970s and early 1990s. These attempts fell into two
broad camps. The first were driven by relaxing the axioms of expected
utility theory, principally the independence axiom, to allow for the
paradoxes and the fanning out of the indifference loci within the
Marschak-Machina triangle (Chew, 1983; Dekel, 1986). There were
three issues with these models. First, their relaxed assumptions were
still violated in experiments (Camerer and Ho, 1994). Second, there is
a danger that if a theory becomes too general, it can allow almost all
preference behaviour, and thus becomes somewhat empty in terms of
making specific predictions. Third, these theories did not really try to
get to grips with how people actually make their choices, which, if we

34

22:37:06
04
describing risky behaviours 35

are genuinely interested in describing human decision making beha-


viour, we ought to try to do.
The second group of theories were built on attempts to under-
stand the reasons why people choose in the ways that they do. That is,
they are driven by an interest in the psychology of choice, and have
proven to be far more influential in the development of behavioural
public policy than those that merely relax the expected utility
axioms. This second group of theories will form the focus of this
chapter. It is important to recognise, however, that these influential
descriptive alternatives to expected utility theory had precedents to
which they must owe some acknowledgement, chief among those
being a model of risk related behaviour laid out by the economist
Harry Markowitz in the 1950s, which perhaps does not get the atten-
tion it deserves.

the markowitz model


In 1952, Markowitz published a paper in which he posed a series of
choices between a number of lotteries and their expected values.
The lotteries offered a 10% chance of money gains or losses and
a 90% chance of gaining or losing nothing, and these gains and losses
ranged from $1 to $10,000,000. For example, three of each of the
questions in the gains and losses domains were:

10% chance of winning $1 or receive 10 cents for sure


10% chance of winning $1,000 or receive $100 for sure
10% chance of winning $1,000,000 or receive $100,000 for sure
10% chance of losing $1 or lose 10 cents for sure
10% chance of losing $1,000 or lose $100 for sure
10% chance of losing $1,000,000 or lose $100,000 for sure

Markowitz presented his questions to a number of his acquaintances.


He discovered that they typically preferred the lottery over the
expected value for small gains. This implied that they were risk seek-
ing in those circumstances. However, for large gains, they tended to
prefer the expected value over the lottery, implying risk aversion.

22:37:06
04
36 the origins of behavioural public policy

Moreover, they typically preferred to lose the expected value for sure
than play the lottery when the losses were small, implying risk aver-
sion, but preferred to accept the risk of the lottery when the losses
were large. Markowitz therefore concluded that people will tend to be
risk averse over large gains and small losses and risk seeking over
small gains and large losses, with the origin – the point dividing losses
and gains – placed at present or customary money holdings. These risk
attitudes imply that the shape of an individual’s utility curve will look
like the one drawn in Figure 3.1, which is worth taking note of in
relation to the theoretical developments in the late 1970s, to be
detailed later.
People often have different attitudes towards risk depending on
the type of choice they face. For example, many people buy insurance
voluntarily for a range of products and services, including appliances
and holidays, which is an expression of risk aversion. Yet the same
individuals often partake in risky games of chance, such as lotteries
and horse racing bets. Standard economic theory cannot account for
these differential risk attitudes. It predicts fixed attitudes to risk and
at most, quite mild risk aversion. Markowitz’s model was an early
attempt to tackle this conundrum by maintaining that risk attitudes
varied with outcome size. He influenced, at least to some degree,
those who later developed slightly more complex descriptive models
of decision making under conditions of risk; enter, once again, Daniel
Kahneman and Amos Tversky.

prospect theory
A quarter of a century after Markowitz proposed his model, Kahneman
and Tversky published one of the most important papers in the history
of behavioural economics (Kahneman and Tversky, 1979); indeed, it is
one of the most cited papers in the history of economics. This paper
introduced prospect theory as a descriptive alternative to expected
utility, the development of which resulted in Kahneman winning
the Nobel Memorial Prize in Economics Sciences twenty-three years
later, which Tversky would have surely shared but for his premature

22:37:06
04
describing risky behaviours 37

U
Positive
utility

Losses Gains

–$2m –$1.5m –$1m Money


Customary money
a holdings

–u1
–u2
b Negative
utility

figure 3.1 The Markowitz utility curve


Note: As was the case in Figure 1.1, the concave section of Markowitz’s
curve in the top right-hand corner implies risk aversion. The shape of the
curve becomes what is known as convex as it approaches the origin in the
domain of gains. A convex utility curve implies risk seeking behaviour.
Similarly, in the domain of losses, the concave part of the utility curve
when losses are low implies risk aversion, but this turns to risk seeking –
convexity – as the losses get larger. To show how this works in relation to
the latter point, imagine that an individual is faced with a lottery that
involves a chance of losing $1 million and a chance of losing $2 million.
The negative utility – or disutility – of losing $1 million and $2 million
can respectively be read from points a and b in Figure 3.1. The expected
disutility of any lottery that offers chances of losing these amounts can be
drawn from the straight line that connects a and b. For example, consider
a lottery that offers a 50% chance of losing $1 million and a 50% chance of
losing $2 million. The expected value of this lottery is −$1.5million, and
the expected disutility, −u1, of this lottery can be drawn from the straight
line. The certainty of losing $1.5 million, on the other hand, can be drawn
from the curve, and is given by −u2. Therefore, the certainty of losing
$1.5 million gives more disutility than a lottery that has an expected value
equal to that amount, which means that the person would rather accept
the gamble than have to pay an amount equivalent to the expected value.
Hence, he will be risk seeking. Other risk attitudes can be read from the
curve in a similar manner.

22:37:06
04
38 the origins of behavioural public policy

death. A parsimonious, behaviourally motivated and universally


accepted alternative to rational choice theory may forever prove elu-
sive, because human decision making is complex and a tolerably sim-
ple theory is likely to be subject to just as many holes as has been
observed with the standard model. Yet it is probably fair to say that
prospect theory is the closest that anyone has yet come to such an
alternative.
In their original work, prospect theory formally allows people to
prefer unambiguously inferior lotteries over unambiguously superior
lotteries, which is known as a violation of dominance. For this, pro-
spect theory received a lot of criticism, because, for most, such
a preference defies logic. Kahneman and Tversky had argued that
people would engage in an initial editing phase before making their
decisions, during which they would reject dominated options.
Nevertheless, there were those who still contended that a decision
model that allows violations of dominance is fatally flawed (Diecidue
and Wakker, 2001). Consequently, in the early 1990s, Kahneman and
Tversky published a development on their original theory that incor-
porated assumptions taken from something called rank dependent
utility theory (Tversky and Kahneman, 1992). This development
assumed that people process probabilities in a slightly different way
to that assumed in original prospect theory, such that formal viola-
tions of dominance are no longer allowed. Kahneman and Tversky
called this development cumulative prospect theory, but our interests
in this book focus upon the characteristics that both versions of pro-
spect theory share, and thus from hereon in the generic term, prospect
theory, will suffice.
Kahneman and Tversky wrote that prospect theory can accom-
modate all of the major violations of expected utility, to which it
essentially makes modifications to assumptions relating to both uti-
lity (or value) and probability. With respect to value, the modifications
resonate with Markowitz’s model, in that rather than the carrier of
value for an individual being what he ends up with – i.e., his final
assets – as is assumed in expected utility theory, prospect theory

22:37:06
04
describing risky behaviours 39

posits that what people really care about are the gains and losses they
experience around a reference point, which is usually assumed to be
the status quo, or what they already have. This is sometimes referred
to, perhaps not very imaginatively, as the reference point effect.
Also, and importantly, prospect theory predicts loss aversion.
That is, the theory assumes that the magnitude of value that people
place upon losses is much greater than equivalently sized gains,
a feature that is not captured by expected utility theory. Loss aversion
suggests, for example, that people will suffer much more pain from
losing $10 than joy from finding $10, and the magnitude of this
phenomenon can be measured by asking people to consider lotteries
that involve chances of both winning and losing. For instance, we
could ask an individual to consider a mixed lottery where there is
a 50% chance of losing $5 and a 50% chance of winning a particular
amount of money, but that we do not know how much. We then ask
the individual for the minimum amount of money he would require
from the gain side of the lottery for him to just start thinking that he
might be willing to accept the gamble. If he states that he would
require $10, then the chance of winning $10 marginally offsets the
chance of losing $5. The gain is double the magnitude of the loss, and
we could thus infer that for this individual, losses count twice as
much as identically sized gains. We would then conclude that the
loss aversion parameter is equal to 2; in their work, Kahneman and
Tversky estimated the parameter to equal 2.25. In gambles where no
outcome is perceived as a loss, a decrease in the smallest gain is fully
compensated for by only a very slightly larger increase in the largest
gain, which shows that the high sensitivity to the worst outcome in
a gamble only applies if the worst outcome is perceived as a loss.
An empirically observed loss aversion parameter of approxi-
mately 2 resonates with the old proverb of a bird in the hand being
worth two in the bush. Loss aversion also explains the endowment
effect, or, in other words, the observation that people place a much
greater value – typically twice as much – on any particular good once
they own it than they did before they owned it, which cannot be

22:37:06
04
40 the origins of behavioural public policy

explained by the curvature of the utility function in standard eco-


nomic theory. An early example of this phenomenon was revealed in
a study by Kahneman et al. (1990), who presented a group of students
with a decorated coffee mug. Some of the students were told that the
mug was theirs and were asked for how much they would be willing to
sell it. Other students were asked how much they would be willing to
pay for the mug. In two experiments, the median value of the mug was
approximately $7 for the sellers and around $3.50 for the buyers.
The endowment effect, driven by loss aversion, was clearly evident.
As with most behavioural phenomena, however, the endowment
effect has perhaps always been intuitively acknowledged. In Ayn
Rand’s The Fountainhead, for instance, the newspaper mogul, Gail
Wynand, ruminates that

I am the most offensively possessive man on earth. I do


something to things. Let me pick up an ash tray from a dime-
store counter, pay for it and put it in my pocket – and it becomes
a special kind of ash tray, unlike any on earth, because it’s mine.
It’s an extra quality in the thing, like a sort of halo. I feel that
about everything I own.
(Rand, 1997, p.564)

In this respect, Rand should have perhaps noted that Wynand is not
atypically offensively possessive at all. The endowment effect has
important policy implications, in that policy makers often find it
difficult to quell public opposition towards their disinvesting in public
sector services, even when the services offer poor value for money.
As we shall see later in this book, loss aversion, in relation to, for
example, motivating behavioural change by emphasising losses rather
than equivalently sized gains, is a finding that has been key to inform-
ing behavioural public policy.
Finally, relating to value, individuals are assumed to be decreas-
ingly sensitive to increasing gains and losses, such that the value
function is concave in the domain of gains, which in itself implies
risk aversion, and, partially in common with Markowitz’s hypothesis,

22:37:06
04
describing risky behaviours 41

V
Positive
value

Losses Gains

Money
Status quo

Negative
value

figure 3.2 The prospect theory value curve


Note: Kahneman and Tversky’s value function is the analogue of the
utility function of standard rational choice theory, drawn in Figure 1.1.
That is, it is a representation of the value that people place on outcomes.
At the origin of the value curve – the status quo – the curve is kinked; it is
steeper in the domain of losses than it is in the domain of gains. This
indicates that people are loss averse – the magnitude of negative value that
people experience from a loss is greater than the positive value they get
from an equivalently sized gain.

convex in the domain of losses, implying a liking for risk. The prospect
theory value function is drawn in Figure 3.2.
Moving away from the value of outcomes and towards the
other main component of a lottery – i.e., the chances of the outcomes
occurring – expected utility theory, and indeed the Markowitz
model, assumes that people process the probabilities embedded in
risky decisions as given, or, in other words, that they weight the
utility or value of outcomes by the mathematical probabilities asso-
ciated with them. However, even before the development of prospect

22:37:06
04
42 the origins of behavioural public policy

1.0
Subjective
probability

0.4

0 0.4 1.0
Objective
probability

figure 3.3 The prospect theory subjective probability curve

theory there had been a long history of challenges to this assumption,


and there is now a considerable literature that demonstrates that
people subjectively transform probabilities (Abdellaoui, 2000;
Bleichrodt and Pinto, 2000; Edwards, 1955; Gonzalez and Wu, 1999;
Preston and Baratta, 1948). Prospect theory incorporates this mod-
ification to expected utility theory, as illustrated in Figure 3.3.
The horizontal axis in the figure measures objective, or mathemati-
cal, probability from zero to one, and the vertical axis measures
subjective, or psychological, probability over the same probability
distribution. The solid, straight diagonal line in the figure is the
assumption embedded in expected utility theory, where subjective
probability equals objective probability over the whole distribution.
The inverse S-shaped dashed line is the prospect theory probability
curve, where people are assumed to overweight small probabilities
and underweight large probabilities. The inflection point, where
subjective probability equals objective probability, was estimated

22:37:06
04
describing risky behaviours 43

by Kahneman and Tversky to be at approximately 0.4, although this


varies somewhat over different studies.
In order to estimate their probability curve, Kahneman and
Tversky presented respondents with a large number of positive and
negative lotteries, with the former being those where no outcome is
perceived as a loss and the latter being those where no outcome is
perceived as a gain (Tversky and Kahneman, 1992). These positive and
negative gambles respectively took a form similar to:

p chance of winning $100 and (1-p) chance of winning $0


p chance of losing $100 and (1-p) chance of losing $0.

Probabilities – p, in the above lotteries – across the entire probability


distribution were used. Kahneman and Tversky elicited the respon-
dents’ certainty equivalences for all of these lotteries. If we term
a certainty equivalence as c, then for the positive gamble given
above, a risk neutral individual – somebody who neither likes nor
dislikes risk – would give c = p*$100; that is, the individual’s cer-
tainty equivalence would be equal to the expected value of the
gamble. For the positive gambles, risk averse individuals would
give a certainty equivalence less than the expected value; that is,
c < p*$100, or, reordering, p > c/$100. Conversely, risk seeking
individuals would give a certainty equivalence greater than the
expected value; that is, c > p*$100, or, reordering, p < c/$100.
In this example, c/$100 can be interpreted as the individual’s sub-
jective probability. Thus, Kahneman and Tversky plotted each prob-
ability, p, used in the lotteries against the average subjective
probability derived from their respondents with the above method.
For example, for positive gambles, when p was set at 0.1, the average
c/$100 over the respondent group was greater than 0.1, implying risk
seeking, and when p was set at 0.9, the average c/$100 came out at
less than 0.9, implying risk aversion. For negative lotteries, the
opposite risk attitudes are predicted from the relationship between
p and c/$100, such that if c/$100 is greater than p, for example, then
risk aversion is implied. Through this procedure, Kahneman and

22:37:06
04
44 the origins of behavioural public policy

Tversky derived their inverse S-shaped probability curve over both


gains and losses.
Like loss aversion, probability weighting is central to not just
prospect theory, but is one of the most robust findings in behavioural
economics more generally and can be used to inform policy design.
The role behavioural economics can play in public policy will be
considered in later chapters, but, to give a taste, if we are trying to
motivate behaviour change such as smoking cessation through the use
of incentives, we might speculate that the use of lotteries that offer,
say, a 1% chance of winning $300 will be more effective than simply
paying people the expected value of that lottery – i.e. $3 – if they quit
smoking. First, people may overweight the small probability of win-
ning, making the lottery appear more attractive than it objectively is,
and second, as we have seen earlier, people often anchor upon rela-
tively large money outcomes. Stepping aside from direct policy con-
cerns for the moment, though, and since Kahneman and Tversky
stated that prospect theory allows for all of the major violations of
expected utility theory, let us briefly consider how it may explain the
first of those violations, Allais’ common consequence effect.

prospect theory and allais’ conjecture


Let us remind ourselves of the choice options offered in Allais’ com-
mon consequence test, where Allais predicted that people would pre-
fer A over B and B* over A*:

A: $1 million for certain


B: 10% chance of $5 million; 89% chance of $1 million; 1% chance of $0
A*: 11% chance of $1 million; 89% chance of $0
B*: 10% chance of $ 5million; 90% chance of $0

Now, it may be the case that loss aversion is particularly strong in the
choice between A and B. This is because individuals have the oppor-
tunity to avoid completely the possibility of winning $0 in this case.
Since A offers $1 million for certain, the $1 million may serve as the
reference point, and thus winning nothing may be perceived as a loss

22:37:06
04
describing risky behaviours 45

that people will be particularly keen to avoid. However, in the choice


between A* and B*, both lotteries involve a high probability of win-
ning nothing. The individual’s reference point may have therefore
fallen, perhaps to $0. Thus, the extent to which winning nothing is
perceived as a loss may have significantly diminished. Compared to
the choice between A and B, this would weaken the influence of loss
aversion on decision making behaviour and prompt many individuals
to base their decision on the size of the best possible outcome rather
than the avoidance of the worst possible outcome, hence the common
preference pattern of A and B*.
The prospect theory probability weighting assumption com-
plements these predictions if people overweight the 1% chance of
winning $0 in B. With probability weighting, it is plausible that the
objectively identical 1% difference in the chance of winning $0
between A and B – i.e. a 0% to 1% chance – and between A* and
B* – i.e. an 89% to 90% chance – will be perceived subjectively as
a bigger difference in the former choice than in the latter choice.
Consequently, probability weighting is a further descriptive psycho-
logical explanation for why people tend systematically to prefer
A and B*.
In terms of the indifference loci in the Marschak-Machina tri-
angle, prospect theory assumes that most of the violations of expected
utility theory will occur around the edges of the triangle where there is
a small probability of the best or the worst, or the best and the worst,
outcomes occurring. An overweighting of a small probability of the
worst outcome would imply that people will be relatively risk averse,
which will cause steepness in the indifference loci, whereas over-
weighting a small probability of the best outcome might be expected
to cause risk seeking behaviour and, therefore, relatively shallow loci.
An approximate representation of the prospect theory indifference
loci is depicted in Figure 3.4.
As already noted, loss aversion and probability weighting are
two distinct features of prospect theory, each one being important in
its own right in terms of its effects on individual decision making

22:37:06
04
46 the origins of behavioural public policy

p($5m)

p($1m)

B B∗

A
0 A∗
p($0)

figure 3.4 Prospect theory and the Marschak-Machina triangle

and its potential for informing policy design. According to Kahneman


and Tversky (1979), however, the value and probability curves in
combination allow for the most distinctive implication of their the-
ory; namely, a fourfold pattern of risk attitudes known as the reflec-
tion effect.

reflection
As earlier noted regarding Figure 3.2, the prospect theory value curve
predicts risk aversion over gains and risk seeking over losses.
However, when we combine the probability curve in Figure 3.3 with
these predictions, we get a slightly more nuanced picture. The under-
weighting of the large probabilities actually serves to compound the
risk attitude predictions inferred from the value function. For exam-
ple, when faced with a large probability of a gain, an underweighting of
the chances of winning will make the gamble appear less attractive,
leading to greater risk aversion, and, conversely, the underweighting

22:37:06
04
describing risky behaviours 47

Table 3.1 Prospect theory’s reflection effect

Gains Losses

High probability Risk aversion Risk seeking


Fear of losing out on gain Hope to avoid loss
Low probability Risk seeking Risk aversion
Hope of gain Fear of loss

Note: The psychological motivations for each of the risk attitudes


are also given in the table. For example, risk aversion when faced
with a high probability of a gain may be driven by the fear of
missing out on a gain, given that the expected value will be close
to the gain in the gamble.

of a large probability of a loss will lead to the perception that the


gamble is less unattractive than it actually is, leading to more pro-
nounced risk seeking. However, if an individual overweights small
probabilities, the risk attitude predictions inferred from the shape of
the value curve could be reversed. For instance, if an individual over-
weights a small probability of a gain the gamble will appear more
attractive than it otherwise would, which could cause risk seeking
behaviour in the domain of gains, whereas overweighting a small
probability of a loss may cause an aversion towards risk taking.
Prospect theory, like the Markowitz model but unlike expected utility
theory, predicts differential risk attitudes across different choices;
more specifically, it allows both insurance and gambling activity by
the same individual in small probability scenarios. The reflection
effect is summarised in Table 3.1.
In a decision relating to a high probability gain – say, for exam-
ple, a choice between a 99% chance of winning $100 and its expected
value of $99 – prospect theory, like expected utility theory, predicts
that people will prefer the $99 for sure. As was indicated in the Allais
paradoxes, people place a high weight upon certainty in the neighbour-
hood of large chances of gains – the certainty effect. It is in this
domain – i.e., high (or at least medium) probability gains – that most

22:37:06
04
48 the origins of behavioural public policy

empirical work in rational choice theory was traditionally underta-


ken. For low probability gains – e.g., the choice between a 1% chance
of $100 and its expected value of $1, prospect theory allows people to
choose the gamble. This is known as the possibility effect, and is
mirrored in circumstances where people play lotteries and engage in
other games of chance. For high probability losses – a 99% of losing
$100 or lose $99 for sure – prospect theory predicts that people will
accept the risk. If people in positions of responsibility and influence,
such as politicians, financial investors and military leaders, engage in
risk seeking behaviours in this context – that is, if they continue to
chase losses – then this can have serious negative consequences for the
people who are affected by their decisions. Choice pertaining to low
probability losses – for instance, a 1% chance of losing $100 or lose $1
for sure – is a classic insurance scenario, where we would usually
expect to observe risk aversion. It can be seen quite clearly in the
table that risk attitude predictions across the domains are
a reflection of each other for high probability gambles; i.e., averse for
gains and seeking for losses. The same can be said for low probability
gambles, and it is from this that the reflection effect gets its name
(Hershey and Schoemaker, 1980; Kahneman and Tversky, 1979;
Wehrung, 1989).
Guthrie (2003) considered reflection, at least in part, in relation
to the economic theory of suit and settlement, and thus took the
argument beyond simple money lotteries. He noted that expected
utility theory is typically used in that area of analysis, with litigants
posited as rational actors who aim to maximise their outcomes.
However, Guthrie maintains that plaintiffs – i.e., those suing – and
defendants operate in the domains of gains and losses, respectively. He
hypothesised that when there is a low probability that the plaintiff
will win if the case goes to decision, he will nonetheless retain the
hope that the judge or jury will find in his favour and will therefore be
risk seeking. The defendant, on the other hand, will be unwilling to
risk a larger loss and will thus be risk averse; settlement would be
expected to favour the plaintiff in these circumstances. In high

22:37:06
04
describing risky behaviours 49

probability litigation where the plaintiff is expected to win, Guthrie


posits that the risk attitudes of plaintiffs and defendants will reverse.
He contends, therefore, that risk attitudes will often be consistent
with the fourfold pattern postulated by prospect theory.
To illustrate, Guthrie presented his law students with
hypothetical non-incentivised, low probability litigation cases,
and asked half of them to play the role of plaintiff and the other
half the role of defendant. Plaintiffs were told that they had to
choose between facing a 1% chance of winning $5,000 at trial and
a $50 payment settlement, which is the expected value of going to
trial; defendants had to choose between facing a 1% chance of hav-
ing to pay $5,000 at trial and paying a $50 settlement. Under
expected utility theory, very mild risk aversion and thus a tendency
towards preferring to settle would be expected from both plaintiffs
and defendants. Guthrie reported that 84% of the defendants but
only 38% of the plaintiffs preferred to settle, a result consistent
with the reflection effect. Guthrie did not undertake an analysis
that presented respondents with cases that involve high probabil-
ities. A task for the enthusiastic reader is to test if the reflection
effect also holds when litigants are presented with scenarios where
there is a high probability that the trial will find in favour of the
plaintiff.
Within the economics profession generally, it is probably fair to
say that prospect theory still lies deep within the shadow of the
standard theory of rational choice, but it has nonetheless received
a lot of attention. It has been the most influential descriptive alter-
native to expected utility theory, partly because Kahneman and
Tversky provided numerical parameters for the main modifications –
i.e., loss aversion and probability weighting – that it makes to standard
theory and thus made the theory reasonably tractable and capable of
offering predictions, and partly because the idea that we assess options
according to what we can gain and lose around what we already have
resonates with many of us. However, there are further alternative
descriptive theories of choice that merit some attention.

22:37:06
04
50 the origins of behavioural public policy

regret and disappointment


In Chapter 2, it was noted that the axiom of expected utility theory
that has been challenged the most over the years, and notably by the
Allais and Ellsberg paradoxes, is independence. However, indepen-
dence is not alone in being scrutinised. Consider the lotteries A,
B and C, below. Loomes et al. (1991) observed that in relation to
these, a significant minority of people prefer A over B, B over C and
C over A, an intransitive cycle. A possible explanation for this eco-
nomically irrational cycle is that some people may simply opt for the
largest probability of a positive outcome so long as they perceive the
outcomes to be reasonably close to each other, and that $4 and $18 are
not perceived as being sufficiently close.

A: 60% chance of $8 and a 40% chance of $0


B: 30% chance of $18 and a 70% chance of $0
C: $4 for certain.

In the early 1980s, prior to the above study, Loomes and Sugden (1982)
had proposed regret theory to allow for violations of transitivity in
individual preferences. In simplified terms, regret theory assumes that
people anticipate feelings of regret associated with the outcomes of
a lottery when they are contemplating the possible outcomes of alter-
native lotteries. These anticipated feelings can serve to modify the
value that an individual places on each outcome. For instance, con-
sider a situation where an individual is asked to pick a ball from a bag
that contains ninety balls, thirty each being red, blue or black.
The individual is asked to choose A, B or C, with the payoffs outlined
in Table 3.2.
Thus, if the individual chooses A, he will win $40 if he picks
a black or a red ball, and $10 if he picks a blue ball. The possible
outcomes from B and C can be read in the same way. Imagine that
the individual is comparing A to B. B may be associated with quite
mild anticipated regret should he pick either a black or a red ball, since
the payoff is a difference of only $10 – i.e., $30 rather than $40.

22:37:06
04
describing risky behaviours 51

Table 3.2 Depicting regret

Black Red Blue


30 balls 30 balls 30 balls

A $40 $40 $10


B $30 $30 $30
C $50 $20 $20

However, A may be associated with quite substantial anticipated


regret in the event of picking a blue ball, since the difference in the
outcomes between A and B is $20. If the anticipated regret from
a difference of $20 is more than double the anticipated regret asso-
ciated with a difference of $10 in this example, then the individual
may well choose B over A. Following similar reasoning, the individual
may choose C over B and A over C; we would have an intransitive
cycle.
As we have seen, regret theory assumes that people modify the
value that they place on outcomes in accordance with the correspond-
ing outcomes across different gambles. A few years after introducing
regret theory, Loomes and Sugden (1986) published a complement to
regret, which they termed disappointment theory. In disappointment
theory, it is assumed that people modify the value that they place on
a gamble in relation to the potential outcomes contained within that
same gamble. For example, assume that the individual is contemplat-
ing option A in Table 3.2. Disappointment theory assumes that he will
place a basic expected value on the whole gamble, which, in this case,
is likely to be pretty close to $40. The possibility of winning only $10,
therefore, will be associated with substantial anticipated disappoint-
ment and the individual’s valuation of that particular outcome will be
modified downwards as a consequence. His assessment of the whole of
option A will therefore be modified accordingly.
The distinction between regret and disappointment in the
Loomes and Sugden sense can be underscored by a circumstance that

22:37:06
04
52 the origins of behavioural public policy

most of us will be familiar with. Imagine that you are contemplating


joining one of two lines at a supermarket checkout. One of the lines is
somewhat longer, but also seems to be moving quicker. You are
thinking of joining the shorter line, but your decision is affected by
the anticipated regret you would feel if, having done so, you observed
that the longer line cleared sooner. Simultaneously, while you are
thinking of joining the shorter line, you have in your mind the expec-
tation that it will clear in ten minutes, but you recognise that there is
a chance that it could take twenty minutes. You anticipate a feeling of
disappointment if, having joined the line, it was to take twenty min-
utes to clear, and this modifies downwards your assessment of that
particular line.
Regret and disappointment can also provide insights when
searching for explanations for some of the classic violations of expected
utility theory. For example, consider one last time the options offered in
the Allais common consequence test:

A: $1 million for certain


B: 10% chance of $5 million; 89% chance of $1 million; 1% chance of $0
A*: 11% chance of $1 million; 89% chance of $0
B*: 10% chance of $5 million; 90% chance of $0.

With respect to A and B, people may anticipate very high regret if they
choose B and win nothing, which may be sufficient reason to push
many in the direction of A. In the choice between A* and B*, on the
other hand, there is no opportunity to avoid winning $0; indeed, in
both choices, the possibility of winning nothing is very high, offering
at least a plausible partial explanation for why many people simply
pick the option that offers the highest possible outcome. Regarding
disappointment, people know they will be successful if they choose A;
their expected utility of A will be the utility they derive from
$1 million. In B, their chances of a good outcome are also excellent;
should they choose this gamble and win nothing they might anticipate
their disappointment to be severe. This anticipated disappointment may
therefore significantly decrease their overall utility of B. However, in

22:37:06
04
describing risky behaviours 53

the choice between A* and B*, they may expect to win nothing whatever
they choose, and will therefore feel relatively little disappointment if
they should win nothing. Even if they do feel disappointment, their
anticipation of this feeling may be very similar when considering A*
and B*. Thus, the effect of anticipated disappointment on their prefer-
ence between A* and B* could be negligible, and they again opt for the
option that offers the highest potential outcome.
As one might imagine, prospect theory, regret theory and dis-
appointment theory have, over the years, had their critics, and specific
aspects of these formulations do not always stand up to empirical
scrutiny, but broadly speaking, all offer important psychological
insights that help to describe choice behaviours. Moreover, the
notions of reference points and loss aversion in particular have been
prominent in the emerging field of behavioural public policy.
A further policy-relevant phenomenon that behavioural economists
have helped to dissect is centred on the human response to time,
which merits a detailed consideration of its own.

22:37:06
04
4 About Time

Time waits for no man. Therefore, a consideration in this book of this


important concept for behavioural economists and public policy ana-
lysts is perhaps already overdue. In the policy discourse, the issue of
whether one ought to discount future benefits – that is, to place less, or
occasionally more, weight on a future benefit than we would if that
same benefit was experienced closer to the present moment – is a topic
of longstanding debate. Many of the great figures in the history of
thought believed that valuing distant outcomes differently to more
proximate ones is a psychological error that ought to play no part in
normative reasoning. David Hume (2004 [1739], Bk.III, pt.ii, sec.vii),
for example, wrote that ‘there is no quality in human nature which
causes more fatal errors in our conduct than that which leads us to
prefer whatever is present to the distant and remote’. Hume’s friend
and contemporary – and, as we have learned, according to some, the
world’s first behavioural economist – Adam Smith (2009 [1759], p.220)
expressed the same sentiment when he wrote that the ‘pleasure which
we are to enjoy ten years hence, interests us so little in comparison
with that which we may enjoy today’ . . . but the impartial ‘spectator
does not feel the solicitations of our present appetites. To him the
pleasure which we are to enjoy a week hence, or a year hence, is just as
interesting as that which we are to enjoy this moment.’ Smith
believed that the impartial spectator – the embodiment of our ration-
ality rather than our emotions – is the voice that we ought to heed.
In the twentieth century, Frank Ramsey (1928, p.543) stated that
discounting benefits ‘is ethically indefensible and arises merely from
the weakness of the imagination’, Arthur Pigou (1932, pp.24–25) wrote
that discounting implies that ‘our telescopic faculty is defective’, and
Robert Solow (1974, p.9) claimed that ‘we ought to act as if the social

54

22:37:06
05
about time 55

rate of time preference were zero’. These are all great figures in the
history of economic thought.
To many, in terms of generating utility or welfare, it is not
immediately obvious why a benefit experienced ten or twenty years
from now should necessarily be valued differently from that same
benefit experienced today, but notwithstanding the sceptics, Paul
Samuelson’s discounted utility model has been accepted as norma-
tively valid in public policy appraisal, including cost-benefit analysis,
for decades (Samuelson, 1937). Samuelson’s model assumes that the
motives that cause people to value benefits differently at different
points in time can be captured in a single parameter – the discount
rate – although it is noteworthy that Samuelson himself had deep
reservations about the normative validity of discounting. An exponen-
tial, or constant, discount rate is typically applied to costs as well as
benefits in public policy analysis.
Like utility maximisation, discounting is an area of economics
where normative and descriptive arguments appear to have become
somewhat muddled over time. Descriptive evidence of time preference –
where people’s preferences indicate that they value the future differ-
ently to the present – is used as a partial justification for why we ought
to apply discounting in public policy analysis. Yet modern empirical
behavioural economists have written that people’s expressed rates of
time preference are driven by emotions and are not of normative
import, concurring with the concerns of the great scholars mentioned
above. However, in an evolutionary sense, there may be good reasons
for these descriptive preferences, and thus they might not be caused by
mistakes per se. We shall consider all of these points in this chapter, but
let us begin by looking in a little more depth at the components that
comprise the normative case for discounting future benefits.

the normative case for discounting


In standard economic theory, it is postulated that a future good should
be assigned a lower value than it is at present for two reasons. The first
reason relies on the assumption that people will be better off in the

22:37:06
05
56 the origins of behavioural public policy

future than they are now, due to technological progress and economic
growth causing a rise in incomes and a decline in product prices.
Looking back at Figure 1.1, we learned that a core assumption in
economics is that people have a declining marginal utility of money;
the assumption extends to other goods also, such that an additional
television set when you already own two of them, or your third ice
cream of the afternoon, will give you less additional utility than the
first of either of these goods brought you. Thus, the assumption is
that if you are better off in the future, then you will have more of
everything – you will have two cars instead of one car, and your second
car will bring you less additional utility than your single car now has
given you. Each individual item that you buy in the future will give
you less additional utility than the same item buys you now, and thus
the utility of those future goods should be discounted to reflect their
current value. The argument suggests that if we do not correct for
declining marginal utility, then we will overinvest in future goods and
goods that reap future benefits, if our objective is one of utility
maximisation.
Now, although people may well have a declining marginal uti-
lity for most material goods, the above argument can be challenged in
relation to the often more intrinsically beneficial outcomes that pub-
lic services provide. Take health, for example, which is the principal
intended outcome of health care and is at least a consideration for
a host of other public services. Even if we experience a declining
marginal utility of health, it is not obvious that improved material
circumstances in the future – even if they are realised – will enable
people to buy more health. In relation to future population health
levels there are a host of other confounding factors, such as rising
obesity rates and greater resistance to antibiotics, that make it at
least plausible that future health levels will be no better than they
are today, and may indeed be worse.
Even if health levels were to improve, there are those who
contest the proposition that people experience a declining marginal
utility for this or any other benefit that affects fundamentally our

22:37:06
05
about time 57

wellbeing. It may in fact be the case that people are able to enjoy an
increasing marginal utility from health improvements if their mate-
rial circumstances simultaneously improve, because with a higher
income, people may experience more opportunities to enjoy their
better health. Whether people will experience more or less utility
from health in the future does therefore seem somewhat ambiguous,
and could go either way, or even neither way. The philosopher Derek
Parfit, for instance, noted that health-related outcomes such as defor-
mities are equally bad whenever they occur and therefore ought to be
associated with constant marginal utility, and John Broome states
that ‘on average, saving one person’s life in one hundred years will
presumably add just as much well-being to the world as saving one
person’s life now’ (Broome, 1994, p.149; Parfit, 1984). Indeed, Broome
explicitly defines health and lives as constant wellbeing goods.
Notwithstanding these challenges, and to reiterate, the first
economic reason for valuing future benefits less than at present is
due to the assumed declining marginal utility curve. The second rea-
son is that of pure time preference – that when you ask them, people
simply prefer to have things sooner rather than later. There have been
numerous studies that have attempted to elicit time preference rates
for various different outcomes, many of which were reviewed by
Frederick et al. (2002). Those authors found that with studies that
have used money, pain, health and life years as outcomes, annual time
preference rates vary widely, from –6 per cent to infinity, although
positive rates predominate. However, Frederick et al. urge caution
against using these rates to help inform an overall discount rate for
use in public policy appraisal. They argue that the rates may be subject
to visceral influences, which we will come back to later in this chap-
ter, such as intemperance, impatience, hunger, desire and cravings.
According to Frederick et al., these are transient fluctuations in tastes
and thus normatively irrelevant when weighing the value of outcomes
across time. Moreover, in preference studies of this type it is intended
that respondents accept that the delayed outcomes will occur with
certainty in order for their answers to measure pure time preference,

22:37:06
05
58 the origins of behavioural public policy

but it is possible, even probable, that they are not able to separate delay
from perceptions of uncertainty. Frederick et al. note that given these
confounding factors, the extent to which the respondents’ answers
measure pure rates of time preference is unclear; it is even possible
that if the confounders could be controlled for, no pure time prefer-
ence would remain. Nonetheless, in standard economic theory, pure
rates of time preference matter and it is possible that if all relevant
factors are controlled for, and if the utility or wellbeing given by
a particular outcome is the same irrespective of when it occurs, people
would still prefer to experience the outcome sooner rather than later.
In addition to the standard economic arguments for discounting,
a couple of other considerations may be viewed by some as norma-
tively important. First, there are real, if hopefully small, possibilities
of catastrophic events such as asteroid impacts, earthquakes and pan-
demics occurring in the foreseeable future, at which point the human
race might even cease to exist. If, as a consequence, nobody or far fewer
people are around post-catastrophe, then it might be deemed sensible
to place a smaller weight on otherwise projected benefits. In a high
profile report on the economics of climate change, for instance,
Nicholas Stern adopted the perspective of a global social planner,
and estimated that if one wishes to account for the uncertainty of
existence, then a rate of 0.1 per cent, which may be added to the
discount rate set for other factors, allows for a 10 per cent chance
that the human race will become extinct from whatever cause over
the next century (Stern, 2007). Second, it could be contended that it is
legitimate for democratically elected governments to place dispropor-
tionate weight on the benefits enjoyed by their current constituents,
since future generations are not currently part of the body politic
(Marglin, 1963). However, others see this latter point as an argument
of political philosophy rather than utility or welfare generation, and
are thus inclined to dismiss it (Broome, 1994).
To give a real world example of how the different normative
arguments summarised above inform the discount rate used in gov-
ernment public policy appraisals, one can consult the Green Book,

22:37:06
05
about time 59

produced by the United Kingdom’s Treasury Department (UK


Treasury, 2003). In order to convert all costs and benefits to present
values, so that they can be compared, the Green Book recommends
a discount rate of 3.5%; 2% to represent the combined effect of declin-
ing marginal utility and the growth in consumption, a pure time
preference rate of 0.5%, and a catastrophic risk premium of 1%.
If we discount at 3.5%, the implications in terms of how much value
we place on an outcome if it occurs in the future compared to it
occurring today can be calculated with the help of the following
simple formula:

wj ¼ ð1 þ rÞðj1Þ

In the formula, wj is the weight that we apply to outcomes that happen


j number of years from now. For example, if we are interested in
the year that occurs ten years from now, j=10 and w10 is the weight
that we give to that year. r is the discount rate; 3.5% in this case. If we
insert these numbers into the formula, we will find that a year ten
years from now is weighted at 0.73 of a year at present value. Through
a similar process, years twenty, forty and one hundred years from now
are respectively weighted at 0.52, 0.26 and 0.03. In other words,
one year now would be worth approximately two years experienced
twenty years from now, four years experienced forty years from now
and about thirty years experienced one hundred years from now.
Discounting at 3.5% will not place public policies that have very
delayed benefits in a good light, which can be problematic for policy
sectors such as public health and climate change, among others, and
presumably is not supported by Broome, Parfit or Stern. However,
3.5% is a fairly conventional rate of discounting, and the normative
reasoning underlying it – and the specific rates attached to each of the
normative components – while subject to much legitimate debate,
is clear.
Although there is nothing in standard economic theory that says
that discounting has to be fixed at a constant rate (Camerer and
Loewenstein, 2003) – i.e., that it has to be exponential – in applied

22:37:06
05
60 the origins of behavioural public policy

economic analysis, this is the general practice. As we have seen, time


preference is also usually taken as exponential; it is broadly perceived
as an exponential component of an exponential rule. The elicited time
preference rate is a descriptive phenomenon, and is only normative if
it is decided that we ought to incorporate it in policy appraisal,
a practice that has provoked widespread opposition, as we have seen.
Descriptively, non-constant time preference rates are often observed.

present bias
Behavioural economists, among others (Kacelnik, 1998), have
observed that there is something about the immediate moment –
the now, or the present – to which people attach a particularly heavy
weight. For time soon after the present moment, people seem to
demonstrate a strong rate of time preference, but this stabilises as
we move further into the future. In fact, this tendency was first
observed in rats (Camerer and Loewenstein, 2003). The heavy weight
upon the immediate moment is commonly called present bias, also
known as the immediacy effect, hyperbolic discounting or plain old
immediate gratification. The phenomenon can be observed by pre-
senting respondents with a choice between early small money
rewards and larger more distant rewards. If the choice is presented
such that the smaller reward is realised now, then there is a tendency
for respondents to opt for the smaller amount. However, if both
rewards are to occur with a delay, but with the same delay between
them as in the original choice, then respondents tend to opt for the
larger prize. For example, if one were to offer a choice between $5
now or $7.50 one week from now, and then a further choice of $5 ten
weeks from now or $7.50 eleven weeks from now, we might expect to
see a tendency for people to prefer the smaller prize in the first choice
but then the larger prize in the second choice. This is a systematic
preference reversal across time that is known as a violation of
dynamic consistency, which can be explained by hyperbolic but
not by exponential rates of time preference. The basic difference
between the implications of exponential discounting and hyperbolic

22:37:06
05
about time 61

Present
value

Exponential
discounting
Hyperbolic
discounting

Time

figure 4.1 Discounting

discounting for the present valuation of any particular good is dia-


grammatically depicted in Figure 4.1.
Read et al. (1999) described a case of dynamic inconsistency in
a study that asked respondents to choose between renting highbrow
and lowbrow films, where the former might be expected to offer
greater long-term rewards and the latter more immediate enjoyment.
The authors found that when their respondents had to pre-commit to
watching a film at a later date, a clear majority expressed a preference
for the highbrow option. This implies that if the respondents were
subject to constant rate discounting over goods, they would also prefer
to watch the highbrow film in the present moment, but this was not
the case. When they were asked to choose a film to watch immedi-
ately, there was a strong tendency for the lowbrow option, an observa-
tion that can be explained by present bias.
Dynamic inconsistency is a difficult concept for standard
economic theory, as it implies either that people’s preferences for

22:37:06
05
62 the origins of behavioural public policy

Present
value of
snacks

Present value of
tasty but unhealthy
snacks

Present value of
healthy but bland
snacks

Now T
Time

figure 4.2 Dynamic inconsistency

future goods are not aligned with the goods that will ultimately
give them the most utility, or that people’s decisions in the
immediate moment are not aligned with their more deliberative
preferences. As we will see later in this book, the latter assumption
has tended to dominate the contemporary behavioural policy dis-
course, since it is typically assumed that immediate preferences are
influenced by the type of visceral factors that Frederick et al. (2002)
alluded to, mentioned earlier, and that people quickly and heavily
underweight the future costs and benefits of current choices.
In short, it tends to be assumed that present preferences are indeed
often biased, which perhaps partly explains a host of public policy
challenges, from obesity to excessive alcohol consumption, to
smoking prevalence and insufficient energy and retirement savings
to the accumulation of too much personal debt, to name just a few.
An example of dynamic inconsistency in relation to food prefer-
ences is depicted in Figure 4.2.

22:37:06
05
about time 63

The figure illustrates findings of how respondents tended to


choose in a study undertaken by Read and van Leeuwen (1998).
It depicts a situation where a person is asked to pre-commit to either
an unhealthy but tasty snack or a healthy but less tasty snack to be
consumed at time period T. At time T, the value, discounted for pure
time preference, is higher for the latter option, and the individual
would therefore commit to the healthy snack. When time period
T arrives, however, we are effectively at the point indicated by the
word Now; i.e. the immediate moment. Since the unhealthy snack,
being enjoyable to eat, is associated with a high level of present bias,
we find that the value of this titbit exceeds that of the healthy snack at
the point of consumption. Thus, when T arrives and becomes the
immediate moment, the individual, if allowed, would reverse his pre-
committed choice and would now opt for the unhealthy option.
Present bias has caused a reversal of preferences over time, which
could not occur if a universal exponential discount rate was used.
In policy contexts, present bias can be to some extent tackled
by an intervention known as a commitment or deposit contract,
although these typically require people to be, in Matthew Rabin’s
words, sophisticated; i.e. able to discern that they suffer from pre-
sent bias and willing to want to try to do something about it (Rabin,
2013). Commitment contracts are sometimes known as Ulysses
contracts, taking their name from the Greek hero who ordered his
men to put wax in their ears and to tie him to the mast of his ship,
so as to avoid being lured to his and their deaths – essentially due to
otherwise unavoidable present bias – by the singing of the sirens.
This ancient myth is another good example of how phenomena that
are central to behavioural economics have been recognised intui-
tively for thousands of years. We will revisit this discussion of
commitment contracts later in the book, but it is perhaps note-
worthy that this was one of the first types of intervention proposed
by the United Kingdom’s Behavioural Insights Team in their early
reports on motivating healthier behaviours (Behavioural Insights
Team, 2010). A proposal in that report was to ask smokers to

22:37:06
05
64 the origins of behavioural public policy

voluntarily pre-commit small amounts of their own money to a


smoking cessation programme, which would be returned to them
if they remained abstinent beyond a set period of time. Although
getting people to pre-commit themselves to behavioural change may
to some extent weaken present bias, the voluntary nature of this
type of programme undermines its potential effectiveness in that
people are allowed to succumb to present bias at any time after the
intervention’s commencement. Moreover, due to a further phenom-
enon known as projection bias, people may typically underestimate
their tendency to succumb to future present bias, and this tendency
could in turn be further strengthened if people are not fully able to
project how they will feel were they to suffer from some future
smoking-related illness (Rabin, 2013). On the flipside, the deposit
contract is potentially strengthened by loss aversion, in that relap-
sing participants lose the money that they have pre-committed.
Present bias also has implications for the risk attitudes that
people express when faced with outcomes that differ in the timing of
their occurrence. To see this clearly, it helps to transform Figure 4.1
a little, to make it resemble Figure 1.1.

present bias and risk attitude


In Figure 4.1, in the context of waiting for something good to happen,
delay is a bad. That is to say, if we were to plot delay against utility,
then maximum delay would be placed at the origin and utility would
increase with each successive reduction in delay. The exponential and
hyperbolic discount functions drawn in Figure 4.1 both suggest that
the utility of reduced delay increases at an increasing rate – that is to
say, they imply convex utility functions. The hyperbolic curve in the
form of the utility of reduced delay is drawn in Figure 4.3. It implies
considerable risk seeking behaviour over delay.
It is noteworthy that behaviour consistent with risk seeking in
the face of delay has been observed in non-human animals, particu-
larly birds (Kacelnik, 1998; Kacelnik and El Mouden, 2013), lending
plausibility to an evolutionary explanation for this phenomenon.

22:37:06
05
about time 65

Subjective value/
utility of $200 at U
various points of
delay

Utility of 50-50
chance of waiting
0 months or 6
months

Utility of waiting
3 months for sure
6 months 3 months Now
from now from now
Reduced delay

figure 4.3 Increasing marginal utility of reducing delay

It seems fair to conclude that practices that have evolved have done so
for a reason and that, in humans, this reason is unlikely to have been to
meet the objective of maximising utility over several decades.
Although animals and humans in their primitive state are likely to
have a variety of what may be called ultimate objectives, including,
among others, reproductive success, survival in the moment and some
form of seasonally driven optimum food storage ability, and although
the relative importance of the different objectives is likely to be
species-dependent, it is at least plausible that they would be more
concerned with their immediate survival than extended optimality,
which in itself lends sense to a focus on the present moment – indeed,
putting aside the possibility that some degree of sacrifice may be
beneficial to the gene pool, one has to survive in order to do anything
else. From this, the psychometric properties that show that humans
and some animals become decreasingly sensitive to increasing
changes in amount – whether this is the amount of money, food,
sound, light, temperature or, indeed, delay – begin to make at least
intuitive sense. From Figure 1.1, we learned that the decreasing sensi-
tivity to increases in wealth is a central component of modern

22:37:06
05
66 the origins of behavioural public policy

economic theory, and the same psychological response tends to be


observed in relation to other physical dimensions. In a primitive state,
even small changes from what we are used to on any of these dimen-
sions may be an indicator that our survival is being threatened or
facilitated, and thus small changes receive disproportionate attention.
Moreover, vis-à-vis loss aversion, it is perhaps telling that we are most
sensitive to changes that are perceived as negative.
Delay for a good does, however, differ from money, food, sound,
light and temperature in one respect, in that delay tends to be per-
ceived as a bad, whereas the other considerations are, up to a point,
goods. As Figure 4.3 shows, for the delay dimension to be charac-
terised as a good – or rather the mitigation or avoidance of a bad – we
need to focus upon reductions in delay, and, as given by present bias,
we are heavily and increasingly sensitive to reductions in delay as we
approach there being no delay. If the objective is survival, present bias
may not be a mistake or bias at all. For example, in the context of the
hypothetical example in Figure 4.3, it may be the case that a debt
ridden and hungry man cannot wait three months to receive $200.
Such a wait would offer little value to him, whereas $200 now could
possibly save his life. Therefore, he may think that opting for the risky
strategy is really his only viable option. As with birds and other
animals that do not even have a concept of risk, he may not intrinsi-
cally like risk, but his circumstances – his increasing sensitivity to
reducing delay as much as he possibly can – steers him in the direction
of the risky opportunity.
As an aside, sensitivity towards minimising delay is, as noted
above, a sensitivity towards reducing a bad or harm, and thus may
place us in the domain of losses. That behaviour consistent with risk
seeking is observed here has parallels with the predictions of prospect
theory, discussed in Chapter 3, at least with respect to moderate to
large probability events, and may have the same evolutionary expla-
nation; that is, the expected loss of a gamble may still be too much to
bear if the objective is survival in a unaccommodating environment,
and thus people may be forced to accept risk instead. It is not too

22:37:06
05
about time 67

fanciful to believe that humans and other animals developed the


tendency towards present bias over hundreds of thousands of years,
and that it is only relatively recently that some human societies have
manufactured the security to enable their inhabitants to plan for many
years or even decades hence. Our brains may not have caught up with
these changing societal circumstances, however, causing many of us
to perhaps focus too much upon the now in our modern circum-
stances. Consequently, some may argue, we are too insensitive to
the longer term negative effects of our unhealthy behaviours, our
profligacy and our insufficient concern for the environment. The
long term was perhaps not the priority for our ancestors when in the
short term, they could all be dead.
Before moving on from the subject of time, it is worth noting
that some analysts, in part motivated by a lack of confidence in the
validity of the probability equivalence method due to the violations of
the axioms of rational choice theory from which the method is
implied, have invented an alternative value elicitation technique
based on the notion of time, for use in public policy evaluation,
specifically in health care. These analysts did not foresee that there
are perhaps as many challenges associated with working with the
human response to time as there are in working with the human
response to risk.

the time trade-off


The utility elicitation method that is based on the response to time
is called the time trade-off. It is a simple method that has been
influential as an input in assessing the value for money of health
care interventions in several countries over many years. For exam-
ple, since 1999 the time trade-off has been recommended for use by
The National Institute for Health and Care Excellence (NICE) to
help assess the cost-effectiveness of interventions in the National
Health Service in England and Wales. Specifically, the time trade-off
is intended to produce numerical values for states of health, so
that all health states across any type of health care intervention

22:37:06
05
68 the origins of behavioural public policy

Health state
value/utility

Full health

Middling
health

Death
T 20 years
Time

figure 4.4 Time trade-off

can be meaningfully compared. Its basic construct is illustrated in


Figure 4.4.
Figure 4.4 depicts a time trade-off that is intended to be used to
elicit the health state value or utility of a condition that is intermediate
to death and full health, called middling health. Respondents are pre-
sented with a fixed period of time that they are asked to imagine living
in middling health, after which they will die. This time period can be
set at any length; in the figure, it is fixed at twenty years. Respondents
are then asked to consider a situation where they can live out their life
in full health instead, and are asked to state the length of time, T, they
would require in full health for them to consider these two alternative
lives for themselves – i.e., middling health for twenty years or full
health for T years – to be equally preferred. The assumption is that
respondents will trade off some length of life for a better quality of
health state. The values of death and full health are respectively nor-
malised as the worst and best points on a zero to one scale. The health
state utility of middling health is then calculated from:

22:37:06
05
about time 69

Utility of full health  T ¼ Utility of middling health  20

Since the utility of full health is normalised at one, the utility of


middling health = T/20. For example, if a respondent was to state
that ten years in full health would for him be equally preferred to
twenty years in the middling health state, then his implied time trade-
off utility of middling health is equal to 0.5; i.e. as half as good as full
health. When time trade-off utilities are used in health economic
evaluation, they are simply multiplied by the time that a patient, or
group of patients, is expected to live in each particular health state, to
derive the quality adjusted life-years – QALYs – mentioned in
Chapter 1. For instance, if a health care treatment is expected to
extend a patient’s life by six years, during which they would live in
the middling health state, and the time trade-off utility for that health
state has been elicited as 0.5, then, assuming for simplicity that no
discounting is applied, the treatment would offer three additional
QALYs for that patient.
For the time trade-off to work as it is intended, it requires
respondents to have zero rates of time preference. It is assumed that
the length of time set for the middling health state – i.e. twenty years
in Figure 4.4 – is immaterial to the utility that the respondent attaches
to the health state. Thus, if, as noted above, the respondent was to
state that ten years in full health would for him be equally preferred to
twenty years in the middling health state, then it would have to be the
case for the respondent that five years in full health would be equally
preferred to ten years in the middling health state in order for his time
trade-off utility to remain at 0.5. If he fails to do this, we can generate
different utilities for the middling health state simply by varying the
time set for the middling health state in the time trade-off method,
and we would not know which utility, if any, is correct.
Unfortunately for the time trade-off method, we know that
Frederick et al. (2002) and others have generally observed positive rates
of time preference, and positive time preference will downwardly bias
the time trade-off utilities (Johannesson et al., 1994). Moreover, the

22:37:06
05
70 the origins of behavioural public policy

reader will by now be aware that in descriptive analyses people often


discount the future not merely exponentially, but hyperbolically. All
this places serious question marks against the validity of the time trade-
off, a method that relies on descriptive stated preferences that is none-
theless used commonly in health care policy evaluation internationally.
To return briefly to the topic of hyperbolic discounting, the
behavioural public policy discussion relates fundamentally to the
issue of respect for human agency. Should policy makers, for instance,
respect people’s choices in the moment, assuming that those beha-
viours do not impose harms on others, which is the approach broadly
embraced by standard economic theory? Or ought they treat present
bias as, indeed, a bias that harms the choosers themselves, and there-
fore as a phenomenon that should be mitigated for their own good?
The study of whether people aim to maximise the utility that they
actually experience, or whether other factors offer legitimate trade-
offs against experienced utility maximisation, is a further aspect of
behavioural science that challenges rational choice theory. These
considerations have gained prominence since the early 1990s, and
like present bias, are heavily concerned with respect for human
agency.

22:37:06
05
5 Experiencing and Remembering

Jeremy Bentham – who was, as noted in Chapter 1, the founding father


of utilitarianism and whose fully clothed preserved skeleton remains
on display at the University College London campus – believed that
mankind was governed by what he referred to as the two sovereign
masters of pain and pleasure, experienced feelings that guide us on
what we ought to do and what we shall do (Bentham, 1988 [1781]).
In other words, we are led by the hedonic experiences of an outcome.
Bentham suggested that we seek to minimise pain and maximise
pleasure, that pain and pleasure are on a continuum and can thus be
compared and that the societal objective ought to be the greatest
happiness for the greatest number. Although the measurement of
interpersonal cardinal measures of happiness had generally become
to be seen as impossible by the beginning of the twentieth century,
modern welfare economics and rational choice theory retained the
assumption that the individual is the best judge of what offers him
utility, and that he will seek to maximise his own experienced utility.
At the beginning of the 1990s, a body of evidence began to
emerge that demonstrated that when people are asked for their retro-
spective and prospective assessment of events, their relative prefer-
ence for different events often fails to coincide with the aggregated
moment-to-moment instant utilities that they actually experience as
they live through each event (Kahneman et al., 1997). These differ-
ences between retrospective/prospective assessments and experi-
enced utility are not random; they are driven systematically by a set
of features that have been termed the gestalt characteristics (Ariely
and Carmon, 2000). In psychology, gestaltism refers to a situation
where the whole is different from the sum of its parts, a line of inquiry
that has been used to explain optical illusions, such as why some

71

22:37:07
06
72 the origins of behavioural public policy

figure 5.1 An optical illusion

people see an old woman looking towards us, and others a young
woman looking away from us, in Figure 5.1. Perhaps similarly, in
the literature on utility the reason why retrospective and prospective
assessments often differ from experienced utility maximisation is
because respondents tend to focus on what they see as the salient
aspects of an episode, and are either unable or unwilling to integrate
sufficiently all of the components of an episode, including, impor-
tantly, its duration, into their assessment.
These salient aspects – the gestalt characteristics – include the
tendency over relatively short episodes for people to prefer worse
outcomes to precede better outcomes rather than vice-versa (Ariely
and Carmon, 2000; Loewenstein and Prelic, 1993), an aversion to

22:37:07
06
experiencing and remembering 73

sudden, steep rates of change in outcomes (Hsee and Abelson, 1991;


Hsee et al., 1991), and the tendency for people to place a heavy empha-
sis on the peak and end moments of an episode (Fredrickson and
Kahneman, 1993; Redelmeier et al., 2003). This last gestalt character-
istic, called peak-end evaluation, has received the most attention in
the literature, and can best be described more fully with reference to
some empirical examples.

empirical examples of the gestalts


Ariely and Loewenstein (2000) review some of the literature that has
shown that people prefer events that improve over time to those that
deteriorate. For example, they ask us to consider four dental treat-
ments spread over a week. In one scenario, the intensity of pain
increases on, say, a ten-point scale from 2, 3, 4 to 5, and in an alter-
native scenario the pain decreases from 5, 4, 3 to 2. Both sequences
produce the same level of total pain, but Ariely and Loewenstein
suggest, with empirical support, that if given the choice, most people
will opt for the latter sequence of declining pain.
As noted above, however, much of the gestalt literature is
focussed upon the significance of end and/or peak moments of an
experience. An early example, which suggests that people’s retrospec-
tive assessments of events sometimes systematically conflict with
experienced utility maximisation, was reported by Kahneman et al.
in 1993, again offering an indication of Daniel Kahneman’s towering
impact on behavioural science. The study’s participants were
informed that they would be exposed to three trials, but unbeknownst
to them the authors intended to conduct only two trials. In the first
trial, the participants were required to immerse one of their hands in
water, with the water temperature carefully controlled at 14°C, for
sixty seconds, which is apparently an unpleasant experience. They
were then allowed to dry their hand with a warm towel. A short time
later, they were required to undertake the second trial, where they had
to immerse their other hand in water for ninety seconds. For the first
sixty seconds of the second trial, the water temperature was again set

22:37:07
06
74 the origins of behavioural public policy

at 14°C, but during the remaining 30 seconds the temperature was


increased to a still unpleasant 15°C. When this was over they were
again allowed to dry their hand on a warm towel. In the third trial,
which, as noted above was not actually undertaken, the respondents
were asked which of the two previous trials they would prefer to
repeat. The rational economic expectation is that the participants
would have remembered that the event that gave them less total
experienced discomfort was, logically, the first trial, since the second
trial was of longer duration and was still universally unpleasant. Thus,
it was hypothesised that the retrospective assessment would drive the
prospective decision, and respondents would choose consistently
with experienced utility maximisation – or rather, in this study,
experienced disutility minimisation – and opt for repeating the first
trial. However, it was observed that a significant majority of the
participants expressed a preference for repeating the second trial, and
similar observations have been reported with respect to loud sounds
(Schreiber and Kahneman, 2000) and to the effort that needs to be
devoted to studying (Finn, 2010). The explanation put forward by the
authors for these preferences is that the participants were driven by
the relatively less unpleasant – but still absolutely unpleasant – end
point of the second trial compared to the first trial, rather than the
integrals of the moments of experienced instant utility in the two
trials, and that the different durations of the two trials was neglected.
Strictly speaking, the participants may not have neglected duration
entirely, but their disproportionate attention to the end moment may
have overwhelmed their concern with the differential duration (Ariely
and Loewenstein, 2000). A focus upon the final moment of an experi-
enced episode is sometimes known as the end effect, and is one
component of peak-end evaluation.
Fredrickson and Kahneman (1993) reported a further study that
demonstrated apparent duration neglect, by exposing participants to
sixteen short film clips. Half the clips were pleasant, such as a film of
a coral reef, and half were unpleasant, for example, films of amputa-
tion procedures. There were two versions of each film, with one

22:37:07
06
experiencing and remembering 75

version lasting for 45 seconds and the other substantially longer at


120 seconds, and each participant was shown some short and some
long clips. Instant utility was elicited while respondents were watch-
ing the clips, by requiring them to turn a knob intended to record the
intensity of pleasure or displeasure that they felt at each viewing
moment; these measures of instant utility could thus be used to
calculate the total utility that each participant experienced while
watching each film. After viewing the films, in order to determine
retrospective assessments, each participant was asked to rank them in
order from the most pleasant to the most unpleasant. Fredrickson and
Kahneman found that the rankings were not related to the different
lengths of the films, but they did correlate quite well with the average
of the peak moments of pleasantness or unpleasantness, and the end
moment of each film – i.e. peak-end evaluation.
In 1996, Redelmeier and Kahneman reported evidence of peak-
end evaluation that they had endeavoured to collect from a real world
setting. The study’s participants were patients undergoing a sigmoi-
doscopy procedure, where a flexible tube is inserted in a patient’s
rectum to examine their colon, or lithotripsy, where kidney stones
are broken into small pieces with ultrasound waves. The duration of
these procedures varies considerably across patients from just a few
minutes to about an hour, and their instant feelings of discomfort – i.e.
their instant moments of disutility – were recorded every 60 seconds
during the procedures. Following their procedure, each patient was
asked to record the total amount of pain they felt they had experienced
on a ten-point scale. Again, duration neglect and peak-end evaluation
were observed.
Redelmeier et al. (2003) followed up on this study by dividing 682
sigmoidoscopy patients into two groups. Without informing the
patients, in one of the groups the flexible tube was left inserted in their
rectums for a few additional moments at the end of the procedure for no
clinical reason, which would have caused some discomfort but, on
average, at a reduced level than when the procedure was ongoing.
Instant disutility and retrospective assessment were recorded similarly

22:37:07
06
76 the origins of behavioural public policy

Pain

Instrument Instrument
removed left inserted Time
immediately

figure 5.2 A violation of temporal dominance


Previously published in Oliver, A. Forthcoming. Distinguishing between
Experienced Utility and Remembered Utility. Public Health Ethics.
Reproduced with permission.

to the study reported above. In addition to peak-end evaluation and


duration neglect, the design of this study was able to highlight another
important effect, in that, in general, the group for whom the instrument
was left inserted tended to remember the procedure as less painful than
those for whom it was removed as soon as possible. The former group
must have experienced greater total experienced disutility, simply as
a consequence of having the procedure duration extended unnecessarily.
That the unambiguously worse experience tended to be remembered as
better than a better experience is a violation of the assumption of tem-
poral monotonicity – in other words, it was a violation of dominance.
Figure 5.2 is a simple diagrammatic depiction of what happened.
The extra period of pain when the instrument was left inserted, repre-
sented under the dashed line, led to greater total pain when the solid

22:37:07
06
experiencing and remembering 77

Age
0 55 60
Happiness
Equally
preferable to
death

Unhappiness
worse than
death
0 25 28
Age

figure 5.3 The James Dean and Alexander Solzhenitsyn effects

and dashed lines are considered together, but is of lower intensity than
the end point of the solid line. Thus, the totality of that under the solid
and dashed lines is remembered as being less bad than when the event
depicted by only the solid line is experienced.
A violation of temporal dominance driven by the psychological
aversion to a poor end has also been demonstrated by the so-called
James Dean effect, named after the Hollywood actor who died in a car
crash in his mid-twenties (Diener et al., 2001). The effect uses lifetime
profiles as opposed to the relatively short episodes that were incorpo-
rated in the studies summarised above, and as such asks respondents
to consider profiles in the abstract, rather than to actually experience
them. Nonetheless, the end effect appears to exercise a strong influ-
ence on people’s choices in these scenarios. The James Dean effect is
in essence the observation that many respondents rate a very happy
life that ends abruptly as better than one that is identical apart from
offering additional less wonderful but still worthwhile years at the
end. The left-hand side of Figure 5.3 diagrammatically depicts the
phenomenon, where the happy profile under the solid line that results

22:37:07
06
78 the origins of behavioural public policy

in an abrupt death at the age of 25 will often be preferred over a profile


that encapsulates both the solid and the dashed lies; i.e., an identical
life up until the age of 25 but then with an additional three years of less
enjoyable but still worthwhile life. Such a preference clearly contra-
dicts utility maximisation.
The converse of the James Dean effect is known as the Alexander
Solzhenitsyn effect, named after the dissident Russian writer who was
at different times imprisoned, persecuted, poisoned, incapacitated by
illness and, for two decades, expelled from his homeland for his views.
He was allowed to return to Russia in 1994, but remained disillu-
sioned with the country. The effect is the observation that when
offered a choice between an awful life – with a level of happiness
that people would consider to be worse than death – and an identical
life except that a few additional years are offered at its end that are not
quite as miserable but that are still worse than death, many people
tend to choose the latter option. That is to say, they choose greater
overall disutility. The right-hand side of Figure 5.3 gives a diagram-
matic depiction of the Alexander Solzhenitsyn effect, with the profile
that people ought to choose if they wish to minimise total disutility
ending at age 55, and the alternative often chosen profile offering an
additional five years of less extreme misery. As with the James Dean
effect, the effect may be explained by a heavy focus upon attaining the
best possible ending to one’s life, which again causes a violation of
temporal dominance.
It could be argued that the James Dean and Alexander Solzhenitsyn
effects are, at best, interesting hypothetical curiosities. However, if
one were to take seriously the possible influence of the gestalt char-
acteristics on preferences over lifetime profiles then they could have
important implications in relation to the assumptions underlying the
evaluation of particular aspects of public policy. For instance, in asses-
sing the value for money of health care interventions, public autho-
rities often assume that the appropriate objective is to maximise the
number of quality adjusted life-years – i.e., the QALYs that have been
discussed earlier in this book – from available health care resources.

22:37:07
06
experiencing and remembering 79

This goal is very similar to that of experienced utility maximisation,


where each QALY is analogous to a (albeit year-long) moment of
instant utility, and where the benefit of treatment is simply the inte-
gral of all of the QALYs gained. As with instant utility, QALYs are
assumed to be additively separable – i.e., no import is attached to how
the QALYs in the profile fit together. Other than discounting over
future life years, no account is taken of the pattern of health over
a temporally extended profile. Whether the QALYs improve or dete-
riorate over time, the speed at which health may change, whether the
profile has a number of health peaks and troughs and how the profile
ends are factors that are assumed to be of no consequence.
A few years ago, I reported a study in which I tested the effect of
several of the gestalt characteristics on people’s preferences over
extended QALY profiles (Oliver, 2008). Considerable evidence of the
effects of the gestalts on people’s choices was observed. For example,
Figure 5.4 illustrates in a simplified form a test undertaken on the
peak/trough effect. The profile represented by the solid line is steady
but comprised of fewer lifetime QALYs than the profile given by the
dashed line, which has a number of highs and lows. In a choice
between the two, a majority of the study’s respondents preferred the
stable profile, even though this offered fewer lifetime QALYs, a direct
contradiction of the QALY maximisation rule and an indication of the
aversion that many people feel towards volatility.
This section has offered a flavour of the now reasonably rich
literature on the influence of the gestalt characteristics, with a focus
on the peak and end effects, in the retrospective and, to some degree,
prospective evaluation of short experienced episodes and long abstract
profiles. Clearly, the gestalts can cause people to choose options that
conflict with the assumption of experienced utility maximisation;
moreover, they can drive duration neglect and violations of temporal
dominance. This begs the question of whether the people who suc-
cumb to the gestalts are making errors of judgment, or whether these
characteristics are legitimate influences on their preferences that
ought to be respected.

22:37:07
06
80 the origins of behavioural public policy

QALY
weight

0 75
Age

figure 5.4 The peak/trough effect over QALY profiles


Previously published in Oliver, A. Forthcoming. Distinguishing between
Experienced Utility and Remembered Utility: A Reflection on Daniel
Hausman’s Valuing Health. Public Health Ethics. Reproduced with
permission.

the gestalts and respect for individual agency


In 2015, the philosopher Dan Hausman wrote that ‘a good life is not
a sum of the net goodness of its moments . . . The same sum of
momentary experiences can add up to a wonderful life or an incoher-
ent and mediocre one, depending on how the experiences are ordered
and what overall narrative they sustain’ (p.114). Such thoughts are not
new. David Hume, for instance, emphasised the importance of con-
sidering the whole of life rather than assessing its individual parts,
drawing the analogy of how a building cannot be valued by summing
independently the values that one would place upon, for example, its
roof, windows, doorway and portico without considering how these
components fit together (Hume, 1777). Even further back, Aristotle
(350BC) noted that happiness cannot be assessed by a feeling or

22:37:07
06
experiencing and remembering 81

sensation in the moment, but can only be seen through the quality of
a whole life, implying again that one cannot judge a life by integrating
the value of its individual parts.
Following this line of argument, the reflective or remembering
self places value on the whole of an experience, on how a temporally
extended period fits together, and this may be normatively important,
but will be overlooked by aggregating moment-to-moment assess-
ments of instant utility (Kelman, 2005). As with the present bias
that was discussed in Chapter 4, there may be evolutionary reasons
for why people focus on particular moments in an episode, such as the
peaks, troughs and endings. For instance, if the normative goal is
survival rather than individual welfare optimisation, it could be that
the peaks and troughs give an indication of what is required to cope
with a particular episode, and the end moment reveals the peak/
trough (Fredrickson, 2000). Moreover, and alluding to Figure 5.4, it is
not obvious that those who want to trade-off some total happiness,
wellbeing or health-related utility in order to avoid substantial vola-
tility over an extended event are suffering from errors in their decision
making. For many, stability across an episode may yield a value quite
apart from the integral of the instant utilities of which the episode
comprises.
Others do not accept that there is a normative justification for
decisions that are guided by the gestalt characteristics, holding the view
that behaviour that leads to duration neglect – or, at least, to insuffi-
cient attention being paid to duration – and violations of temporal
dominance are unacceptable. That is, it is irrational to prefer a tempo-
rally extended event over an alternative that unambiguously yields
greater experienced utility. Those who see the gestalts as errors there-
fore distrust retrospective and prospective preferences for different
profiles, believing them to be unreliable indicators of experienced uti-
lity (Kahneman et al., 1997). Indeed, they distrust preferences full stop,
and in health economic evaluation, for example, do not accept that
QALYs – which are elicited from preferences expressed with the use
of instruments such as the standard gamble and the time trade-off – are

22:37:07
06
82 the origins of behavioural public policy

appropriate proxies of instant utility (Dolan and Kahneman, 2008).


People with this perspective thus maintain that the value of an
episode or event ought to be equated with the subjective value of the
momentary states – the instant utilities – that comprise the event, as
those moments are lived. They propose a variety of methods by which
to measure experienced utility, from the sliding scales and knobs
that people can move to record their continuous mood, which are
sometimes used in experiments over short episodes such as that
which exposed participants to short film clips, reviewed above; to
asking respondents to record their mood every minute or so, used in
the sigmoidoscopy studies; to the so-called day reconstruction method,
where participants are asked to record their current mood and
their activities periodically during the day, which is perhaps the
most practicable method when assessing temporally extended events.
This approach mirrors Benthamite utilitarianism, and its advocates
argue that we ought to seek to direct our activities, including our public
policies, such that experienced utility is maximised, a proposition that
has drawn serious interest from several governments (Hausman, 2015).
Richard Layard is a chief among the neo-Benthamites. He is an
early advocate of the new economics of happiness, which essentially
embraces the experienced utility approach. A principle motivation
for the neo-Benthamites has been the observation that although
much of the world has become substantially better off in material
terms since the mid-twentieth century, survey evidence suggests
that the populations of most countries have become no happier
(Layard, 2005). Layard, like Bentham before him, expresses the
view that in order to avoid inconsistency across actions, all laws
and all rules of morality must be based on a single underlying
principle, which he believes ought to be to secure that greatest
overall happiness within any population. He believes that happiness
is a self-evident good; an intrinsic good rather than a good that
merely offers instrumental worth. That is to say, we want to be
happy for its own sake, as an end in itself, and not as a means of
reaching some other goal.

22:37:07
06
experiencing and remembering 83

Layard further argues that developments in neuroscience, where


self-reported measures of happiness and exposure to pictures depicting
various pleasing or distressing images are correlated with particular
movements in the forebrain, offer promise that more accurate measures
of interpersonal cardinal utility may be possible. More specifically,
using magnetic resonance imaging, activity in the left side of the fore-
brain has been found to correlate with exposure to enjoyable stimuli,
and the same goes for the right side of the forebrain and unenjoyable
stimuli, which some claim confirms the objective nature of happiness
and pain. However, others contend that the use of brain imaging to
guide public policy, if pursued at all, should proceed with great caution,
for while these scientific developments might offer an indication of
movements in ordinal happiness or utility within a single individual, it
is not clear that they point at utility that is either interpersonal or
cardinal. For instance, it is not proven that two persons who experience
the same activity in their left forebrain are experiencing the same
degree of happiness. Consequently, the neuroscience field is not suffi-
ciently developed to enable us to calculate the results of trade-offs
between the happiness and unhappiness experienced by different indi-
viduals and so derive an accurate measure of net happiness.
Perhaps more worryingly, it is possible that neuroscience will
reveal that some people are genetically more predisposed towards
increasing their happiness than others. That is, some may have
a relatively highly active left side forebrain, and others a relatively
active right side, and it is plausible that the differences could have
evolved in order to improve the chances of species survival – i.e.,
differential sensitivities to pleasure and pain across individuals within
a group may be beneficial to group survival. It is also plausible that the
capacity for happiness differs by biologically determined group dis-
tinctions, such as those given by gender or race. If this proves to be the
case, then the greatest happiness principle would appear to prescribe
the systematic prioritisation of some groups over others in public
policy decision making. Prioritisation on the basis of genetic predis-
position would presumably be objectionable to most.

22:37:07
06
84 the origins of behavioural public policy

However, many of the advocates of experienced utility max-


imisation as a guide to public policy do not tend to dwell extensively
on the potential usefulness, or otherwise, of brain imaging. Rather,
they attempt to identify, with tools such as the day reconstruction
method mentioned earlier, activities, aspects of life and public
policies that appear to give people greater experienced happiness
than they might otherwise enjoy. To them, the influences of the
gestalts on preferences are errors that carry no normative force, and
allowing people to act upon their own retrospective and prospective
evaluations – i.e. respecting individual agency with regards to their
expressed preferences – is therefore irresponsible. They hold that
paternalistically steering policy towards securing the greatest
experienced happiness is the right path to pursue; the view is that
this is what any right-minded person would want.
Some scholars, while sceptical of the notion that preferences
satisfy the rational deliberation and complete knowledge assumptions
embedded in standard economic theory, reject the view that the value
of public policy is its impact on happiness. For instance, Hausman
(2015) suggests that the evidential connection between how good one
feels and how well one’s life is going in relation to one’s health or
educational achievements or financial security etc., is fragile. For
example, individual or population adaptation to a poor state of affairs
implies that the poor state does not impact substantially on happiness,
and yet if the poor state falls within the purview of public policy, then
policy makers would presumably still wish to do something about it.
The view that adaptation undermines subjective assessment as a
means to inform public policy is associated with Amartya Sen’s
happy slave conjecture; i.e., few would deliberatively choose to be
a slave even if it was known and accepted that one would adapt well
to that state, because the opportunity to flourish as oneself sees fit
would be substantively curtailed as a consequence (Sen, 1999).
The converse of underestimating one’s plight is of course exag-
gerating it, which can happen, according to Hausman, when transitory
states seem more important to people in the moment than following

22:37:07
06
experiencing and remembering 85

a more dispassionate consideration of their longer term effect on well-


being. Following this line of argument, retrospective and prospective
assessments might be expected to give a more accurate evaluation of
the impact of a particular experience on the whole of one’s life than
the instant utilities of the event while it was running its course,
because we can step out of the event and offer a better view of how
good or bad it really was in terms of its positive or negative contribu-
tion to a flourishing and/or fulfilling life. For example, perhaps people
tend to remember the peak moments of rearing children, because an
integral of all of the moments of instant utility might make the
proposition of repeating the experience somewhat less attractive.
Hausman (2015) himself writes that he hardly remembers the tedious
moments of rearing his children, but does vividly remember the joy of
reading to them.
Overall then, there are normative arguments – reasoned opinions –
for and against accepting the influence of the gestalt characteristics as
aspects of human agency that policy makers might consider. Those who
are convinced by the evolutionary arguments and believe that these
remain relevant in modern societies, and those who hold that retro-
spective and prospective assessments pick up important considerations
that a moment-to-moment integration of instant utilities cannot, will be
more inclined to attribute normative status to the gestalts than those
who think that all that really matters are people’s felt experiences during
an event or episode. Others might take a more contextualised view of the
available evidence. For instance, while they may be convinced that it is
legitimate for people to want to trade-off some total experienced utility
in order to avoid volatility, they may at the same time be circumspect of,
for example, the James Dean effect, believing it to be the result of
insufficient reasoning of everything that the hypothetical profiles
would, in reality, imply (Ariely and Carmon, 2000). For example, with
reference to the left-hand side of Figure 5.3, deliberation might suggest
that most people aged, say, 23 or 24 years, would want to live on to an age
of 28 years, even if happiness diminished in the final three years, if the
only alternative is an abrupt death at the age of 25 years.

22:37:07
06
86 the origins of behavioural public policy

The side of the fence on which one falls – if one wants to get off
the fence at all – is a matter for the individual, but what does once
again seem clear is that the postulates of standard economic theory –
that people ought to maximise experienced utility and this will be
represented in their preferences – often fails to hold. The experiencing
and the remembering selves sometimes differ. The gestalts may be of
use irrespective of any consideration of their normative status, how-
ever. If one has knowledge of them, then they have potential to help us
steer behaviours, including indeed our own behaviours, in directions
that we would like them to go.

using the gestalts descriptively


To reiterate, using a descriptive approach to help inform personal and
public decision making does not prescribe that people ought to focus
on, say, peaks and ends in their retrospective and prospective evalua-
tions. However, we can use our knowledge that people often do assess
decisions in this way to help us to pursue any goals that we deem
desirable, whether this be to satisfy the postulates of Benthamite
utilitarianism, or indeed anything else, excepting perhaps those of
pure libertarianism, where we might be reluctant to trick ourselves
or others into doing anything other than what comes naturally.
Assuming that we can ascertain which behaviours maximise welfare,
we could use the gestalt knowledge paternalistically to steer people in
those directions; indeed, this kind of approach – a behavioural welfare
economic approach – has been embraced in relation to the other
findings of behavioural science by libertarian paternalists – or nudge
theorists. This will be discussed in Chapter 7.
Along these lines, Dan Ariely (2008) recounts an incident where
knowledge of peak-end evaluation could have been put to good use in
his own life, which perhaps offers a lesson for how some people’s lives
could be improved in the future. A serious burns victim in his youth,
Ariely notes that the practice, common among nurses, of removing
bandages as quickly as possible so as to limit the duration of discom-
fort to the patient – and distress to the nurses themselves – can cause

22:37:07
06
experiencing and remembering 87

pain so intense that it provokes significant mental trauma. Ariely


advocates for a slower, more careful removal of bandages from burns
patients by nurses in such circumstances, which may prolong the
experience in the moment (remember, duration is often underweighed
in people’s memories), but would perhaps help to ensure greater longer
term wellbeing.
Ariely also reports his personal history of contracting hepatitis
C via a blood transfusion, where treatment can make patients so
violently ill that many fail to complete the required course. Ariely,
a film lover, promised himself some movies that he had never seen
before after each round of treatment, to serve as an incentive to con-
tinue. This self-imposed enjoyable end moment made treatment more
bearable; he completed his course of treatment and was cured as
a consequence. The addition of an enjoyable end moment, or at least
a less unenjoyable end moment, brings to mind the study on sigmoi-
doscopy patients discussed earlier, and may offer lessons on how to
increase the percentage of patients who return for necessary follow-up
appointments; I always commit myself to buying my wife a nice
present after each medical test that I am required to take. It may also
offer some clues on how we might improve our own personal lifestyle
behaviours, in relation to beneficial actions that are painful to under-
take in the moment. For example, adding an enjoyable experience at
the end of each exercise session might reduce the prevalence of unused
gym memberships. Using the gestalts in these sorts of ways essen-
tially creates a type of behavioural-informed incentive; we will con-
sider other forms of behavioural-informed incentives in the chapters
that follow.
All in all, the gestalt literature offers something of a challenge to
one of the main assumptions of standard economic theory; namely,
that retrospective and prospective evaluations will align with experi-
ences. We will now move on to a different assumption of the theory –
i.e., the relative price mechanism or, in other words, you get what you
pay for – and the challenges to that.

22:37:07
06
6 Motivational Crowding

At face value, the relative price mechanism – i.e. if you pay or penalise
someone for doing something, they are more or less likely to do it –
appears so obvious that it hardly needs stating. Historically, econo-
mists did not question it; it is embedded in standard economic theory
and underpins economic rationality. However, in the early 1970s,
a non-economist published what subsequently became a classic text
in social policy, in which he asserted that, in some contexts, paying
people could have the opposite effects to those intended. Richard
Titmuss set out in The Gift Relationship to compare the practice of
blood donation in the United States and the United Kingdom, where
in the former country people tended to be paid for this activity,
whereas the latter country relied on voluntary donations (Titmuss,
1970). Contrary to the relative price mechanism, Titmuss, rightly or
wrongly, argued that the United Kingdom was less susceptible to blood
shortages than the United States, was less at risk of blood contaminated
by disease and spent far less on collecting this vital resource.
Building on seminal work by the anthropologist Marcel Mauss,
Titmuss (1970) made a distinction between what may be classified as
social goods that rely on institutions that create integration and
economic transactions that are more atomised. The donation of
blood to help others, according to Titmuss, relies on human qualities
such as altruism and reciprocity, rather than the pure self-interest
that is embedded in market transactions. Notwithstanding his occa-
sional confusing of the concepts of pure altruism and reciprocity –
which I will return to in Chapter 9 – Titmuss was essentially arguing
that since the best-intentioned people undertake social actions for
altruistic purposes, introducing payment in these contexts can
crowd out those who are motivated as such – who might then seek

88

22:37:11
07
motivational crowding 89

Price of
blood

p∗

pz

py

0 z x y
Quantity

figure 6.1 Motivational crowding out

alternative arenas to exercise their altruistic urges – without suffi-


ciently or appropriately crowding in those who are selfishly moti-
vated. In short, attempts at crowding in selfish motivations might be
counterproductive.
In simple terms, Figure 6.1 depicts what may happen in terms of
the supply of blood as the price offered to donors increases. At zero
price – i.e. where donations are entirely voluntarily – supply is at x,
which may increase slightly to y if a small financial incentive is
offered, partly because the altruistically motivated may feel further
validated as a consequence and partly because some who are purely
financially motivated may now decide to donate. However, if the
incentive rises above py, up to pz, we may see a fall in donors as
many of those who undertook this task for altruistic purposes might
no longer see it as an altruistic act, and may thus seek fulfilment
elsewhere. If an incentive higher than pz is offered, supply once again
rises steadily beyond z as the now substantial fee crowds in more of
the selfishly motivated. To reach the blood supply that was donated

22:37:11
07
90 the origins of behavioural public policy

voluntarily, however, public authorities now have to pay a fee, p*, to


each donor and at the same time would perhaps need to exercise
greater caution to ensure that the donated blood is of good quality.
Although Titmuss’ work gained a high profile and is heavily
cited, it was not until the 1990s, following the publication of numerous
empirical studies at around that time, most associated with the Swiss
economist Bruno Frey, that motivational crowding was afforded some
respect within the mainstream economics community. A few of these
empirical studies will be summarised later in this chapter, but when-
ever motivational crowding is discussed, one particular study is invari-
ably mentioned. Therefore, since this book would otherwise seem
incomplete without it, let us briefly consider what that study found.
Gneezy and Rustichini (2000) undertook a field study on a group
of Israeli child day care centres. The authors had noticed that parents
had got into the habit of arriving late to collect their children, to the
considerable inconvenience of the teachers. Therefore, from the expec-
tation that a financial penalty would lessen this activity, Gneezy
and Rustichini introduced a fine for tardy parents. Unfortunately, as
a consequence the incidence of lateness among the parents increased
significantly, and remained at those higher levels after the fine was
removed. The explanation for this economically counterintuitive result
was that the parents interpreted the fine as a child minding fee, and felt
that it increased the legitimacy of their late arrival.
Psychologists had actually developed a theory of motivation
that is pertinent to motivational crowding prior to the economists’
interest in this topic, and thus this theory is of course useful when
trying to interpret the empirical results. Therefore, before we go
further, a consideration of these theoretical propositions is warranted.

self-determination theory
Edward Deci and Richard Ryan, with refinements by others, devel-
oped self-determination theory in a series of publications spread over
two decades (Deci and Ryan, 1985; 2000). Importantly, the theory
distinguishes between autonomous actions and behaviours that are

22:37:11
07
motivational crowding 91

perceived to be controlled or influenced by factors external to the self.


Deci and Ryan subdivide autonomously driven actions into two types.
First, there are those that a person finds intrinsically interesting,
enjoyable and/or important to do, and these actions are associated
with a high level of personal satisfaction. Second, there are behaviours
that are associated with a societal standard or value – for example,
cleaning up after one’s dog has done his business in a public park – that
the person identifies with as being an important norm and internalises
it as part of their own morality. The person regards the behaviour as
part of their own identity – i.e. as someone who does not leave their
dog’s mess lying around for a child to fall into, for example – and is not
motivated by what others think of him if he fails to act.
Deci and Ryan also distinguish between two kinds of motiva-
tion that are externally controlled. First, there is the heavy control
of external regulation, where behaviour is influenced directly by
rewards and penalties, financial or otherwise. Second, there is the
type of control where people feel pride or shame if they act, or fail to
act, in a particular way, due to the approval or disapproval of others,
including friends, family members, pets, colleagues or members of
the general public. In self-determination theory, this second type of
externally driven action is known as introjected motivation. If we
take our dog mess example, an individual will be introjectedly
motivated to clean up if he is doing it out of a feeling of guilt
relating to what others may otherwise think of him, which differs
from the identified motivation discussed above where people under-
take the act because it complies with their own sense of what is
right and wrong, regardless of what others think. Table 6.1 sum-
marises the different motivations postulated by self-determination
theory.
Motivations can change over time; for instance, from controlled
to identified. For example, some smokers may have initially complied
with the ban on smoking in public places as a consequence of both
regulatory and introjected control, but over time, learned to identify
with the policy. Disney et al. (2013) note that if one wants to crowd

22:37:11
07
92 the origins of behavioural public policy

Table 6.1 The motivations in self-determination theory

Autonomous Controlled

1 Intrinsically interesting and Heavy regulatory control. Driven


enjoyable. Driven by personal by rewards and punishment.
satisfaction.
2 Morally motivated. Driven by Introjection of external pressure.
one’s own identity. Driven by pride and guilt.

in a particular type of behaviour with a short term stimulus, such as


a financial incentive, then such a shift in motivations towards the
identified kind is crucial if the behaviour change is to be sustained
once the external stimulus is withdrawn, assuming, of course, that
any social pressure dissipates with the removal of the incentive. This
may seem like an obvious point, but it is important in relation to the
use of short term financial incentives intended to motivate healthier
behaviours that we will touch upon later in this chapter.
Thus, as may be deduced from the above, Deci and Ryan suggest
that the evidence indicates that the satisfaction from and the effective-
ness of behaviour change interventions are both greater when people
feel that they are acting with a degree of autonomy – i.e. when people
enjoy or identify with the behaviour that is being promoted, when it has
a meaningful rationale and is perceived as being supportive – than when
they believe that they are being controlled. These concepts offer psycho-
logical insight into the empirical literature that purports to demonstrate
the respective crowding out and crowding in of altruistic and self-
interested behaviours.

empirical examples of crowding


There is now quite a rich literature on motivational crowding out,
with evidence available from both the laboratory and the real world
(Frey, 2001). One or two examples are summarised here to further
illustrate the point. For example, in a study not quite as famous as

22:37:11
07
motivational crowding 93

that which they undertook on Israeli day care centres, Gneezy and
Rustichini (2000b) studied the effect of financial incentives on the
efforts expended by high school students engaged in visiting houses in
order to solicit charitable donations. The authors found that when
students were paid as an incentive 10 per cent of the sum they col-
lected they actually gathered less money than when they performed
the task on an entirely voluntary basis. Gneezy and Rustichini attrib-
uted this outcome to the crowding out of intrinsic motivation,
although it is possible that other explanations carry weight. For exam-
ple, it might have been the case that the financial incentive could have
made the students too eager to persuade people to donate, which
might have paradoxically discouraged potential donors.
Frey (2001) lends support to the Gneezy and Rustichini findings
by reporting the results of an evaluation in Switzerland on the effect of
financial rewards on intrinsic motivation among volunteers. While
rewards were found to induce people to provide more voluntary work
hours, the payments were found to reduce work effort such that,
overall, substantially less work was actually done. A more complete
test might have offered a financial incentive for both hours and effort
combined, because the workers might have felt that the additional
hours compensated for any reduced effort, and it may even be the
case that, physically, effort has to decline if hours are increased.
Nonetheless, the evaluation offers some weak support for motiva-
tional crowding out.
Deci et al. (1999) undertook a meta-analysis of 128 studies that
had studied the issue of motivational crowding – meta-analyses essen-
tially combine the results of previous studies to reach an overall
conclusion. We ought to be cautious of the effect of publication bias,
i.e., studies that show a positive effect of motivational crowding are
perhaps more likely to be accepted for publication than those that
show no effect, but Deci et al. found that tangible rewards undermine
motivation for interesting tasks, i.e., tasks that were intrinsically
valued by those doing them. The result was particularly strong when
money rather than symbolic rewards such as awards – e.g. employee of

22:37:11
07
94 the origins of behavioural public policy

the month – were used as the incentive, perhaps because money was
perceived to be controlling rather than supportive, and the negative
relationship between money and motivation was stronger when the
tasks were relatively simple to do.
Informed by a series of experiments, Ariely et al. (2009) discuss
how pay for performance incentives may be detrimental to jobs that
involve insight and creativity, which may be linked to the evidence
showing that incentives undermine performance in jobs that are
undertaken for their intrinsic worth. That is, these incentives may
cause people to focus upon only those dimensions of the work where
the incentives apply, which turns them into somewhat atomised
automated processes that destroys the creativity, and thus joy, that
is encouraged and generated implicitly by a more holistic approach.
One of the most famous studies in the motivational crowding
literature was undertaken by Frey and Oberholzer-Gee (1997), and is
focused upon the public acceptance of broadly socially worthwhile
but often locally unwanted projects, such as airport, motorway and
nuclear repository construction, a policy problem known as Not
In My Back Yard, or NIMBYism. Frey and Oberholzer-Gee surveyed
respondents to ask if they would accept voluntarily the construc-
tion of a radioactive waste repository in their local community via a
procedure that was identical to that used in practice in Switzerland
to gain public permission for such projects, where the federal com-
munity, the developer and a local town hall meeting all have to
issue consent. Perhaps surprisingly, given the fear associated with
nuclear waste, a majority of the respondents – if only a marginal
majority of 50.8% – supported the construction of the repository.
However, when offered a series of substantial money compensa-
tions for accepting the radioactive repository, only 24.6% of the
respondents continued to give consent. The introduction of sub-
stantial compensation may have alerted many of the respondents
to the possibility that the project might present some danger, but
Frey and Oberholzer-Gee contended that the results suggested that
the introduction of money compensation simply crowded out

22:37:11
07
motivational crowding 95

around 50% of the previously observed intrinsic motivation and


sense of civic duty.
Overall, while the interpretation of the results of at least some of
the literature on motivational crowding is open to debate, it seems
reasonable to conclude that financial incentives may often be viewed
as a form of control when introduced in circumstances that rely quite
heavily on intrinsic motivation, as postulated by self-determination
theory. Thus, these money incentives can consequently crowd out
intrinsic motivation and even civic virtue, leading to a result that
contrasts with that expected by the operation of the relative price
mechanism. This is a relevant finding for public policy, where many
sectors presumably rely on an intrinsically motivated workforce to
secure good performance.
Thus, it is perhaps advisable to search for policy areas where
financial incentives are not perceived as controlling, but rather as
supporting of autonomous actions. The use of incentives in this way
might not only serve to motivate those who already enjoy undertaking
these actions, but may also provide a societal signal that these are good
things for people to do, and could then turn into introjected motiva-
tions for at least some people. Moreover, as was argued earlier, intro-
jected behaviours can in time transform into actions with which
people identify, which in terms of securing behavioural change is the
strongest motivation in self-determination theory.
One policy intervention for which this process may have hap-
pened is the introduction of small charges to discourage supermarket
plastic bag use with the intention of protecting the environment.
Those previously intrinsically motivated to not use plastic bags do
not gain or lose anything from the intervention, and yet they can see
that their values are being supported. For others, the small charge
highlights a widely perceived societal problem and thus some may
initially feel heightened pride or shame from not using or using the
bags following the introduction of the charge. Eventually, the non-use
of plastic bags may become a broadly held individual value and social
norm. A simple google search will reveal to the reader that several

22:37:11
07
96 the origins of behavioural public policy

countries have introduced plastic bag charges of only a few pennies or


cents and have seen apparently remarkable reductions in plastic bag
use as a consequence. One might therefore reasonably ask whether
small incentives can have a comparable impact on perhaps even more
important personal lifestyle behaviours.

personal financial incentives


As noted in Chapter 4, the proposition to use small financial incentives
and disincentives – called here personal financial incentives – as
a means to motivate behaviour change has occupied quite a prominent
position in the recent behavioural public policy discourse, and the
policy sector that has probably received the most attention in this
respect is health. There is actually quite a substantial amount of evi-
dence published in the literature on the effectiveness of personal finan-
cial incentives to improve medical care uptake – e.g. vaccinations – and
adherence, and to discourage unhealthy behaviours, such as smoking
and some dietary habits. With respect to behaviours that only require
those targeted to act once, or at most a limited number of times, the
effectiveness of personal financial incentives is often quite positive.
For example, Seal et al. (2003) did a randomised controlled trial
on a group of socially marginalised intravenous drug users in San
Francisco. They gave all of their participants the first of three
required hepatitis B vaccine doses and then divided them into two
subgroups. The third vaccine dose was to be offered six months after
the first dose. One of the subgroups was assigned weekly contact
with an outreach worker, whose role was to offer guidance, encour-
agement and support. The other subgroup was not offered an out-
reach contact, but instead received a monthly $20 incentive if they
remained in the vaccination programme. At the end of the pro-
gramme, it was observed that 69% of the respondents who were
offered the money incentives received all three doses of the vaccine,
versus only 23% of those who had outreach support. Moreover, the
incentives were cheaper than outreach, at $220 versus $590 per per-
son, respectively.

22:37:11
07
motivational crowding 97

In order to encourage the uptake of mammography, Slater et al.


(2005) tried out two types of mail-based interventions on low-income
women. Both interventions offered a free mammogram if the respon-
dents rang a toll-free number, with one intervention also offering
a small financial incentive of $10 if they went for the mammogram
within one year. More than four times as many calls were received for
the mail plus incentive intervention than the mail only intervention,
and the subsequent mammography rate was significantly higher for
the former than the latter intervention, which in itself produced
a significantly higher mammography rate than doing nothing.
Not all of the evidence points to a positive effect of small finan-
cial incentives on medical uptake and adherence. For example,
Malotte et al. (2004) reported that a $20 incentive did not significantly
increase return rates for retesting after treatment for gonorrhea and
chlamydia, but, overall, there is sufficient research to indicate that
financial incentives show promise in some specific contexts in the
health care domain, particularly those that do not require a long term
sustained effort on behalf of the target group, where motivations of
either the introjected or identified kind are probably not required.
The same cannot be said for small financial incentives to moti-
vate more sustained behavioural change, of the sort necessary in rela-
tion to broader lifestyle behaviours. For example, in a systematic
review of articles focussed upon the use of financial incentives to
encourage people to quit smoking, Cahill and Perera (2008) identified
seventeen studies, only one of which demonstrated significantly
higher cessation rates for those offered incentives compared to those
who were not, beyond six months from the start of the programme.
Any early success with incentives appears to dissipate when they are
no longer offered, which could be because the programmes that they
reviewed were generally too short for the participants to identify with
non-smoking. After Cahill and Perera’s review was published, Volpp
et al. (2009) did report some success with this type of intervention.
Eighteen months after the initiation of their trial, Volpp et al. found
that 9.4% of those who were offered money incentives were still not

22:37:11
07
98 the origins of behavioural public policy

smoking, versus 3.6% who were not. However, Volpp et al.’s incen-
tives were quite large – each participant could earn $750 – and there-
fore their effectiveness, while interesting and potentially useful at
a broader societal level, are nothing more than would be expected
from the relative price mechanism of standard economic theory.
A similar story of general non-effectiveness can be told in rela-
tion to another outcome that requires sustained action – weight loss.
In a meta-analysis of randomised trials on personal financial incen-
tives aimed at reducing obesity rates, Paul-Ebhohimhen and Avenell
(2007) identified nine studies, which ranged in the level of incentives
offered from 0.2% to 10.2% of personal disposable income. Paul-
Ebhohimhen and Avenell estimated that compared to offering no
financial incentives, an incentive of less than 1.2% of personal dis-
posable income is associated with no weight loss at all. Moreover,
incentives greater than that were associated with a weight loss at
eighteen months of only 0.7 kg, about 1.5 lbs – or virtually nothing.
I noted above that Volpp et al.’s study, which used large finan-
cial incentives to encourage smoking cessation, was not really moti-
vated strongly by behavioural science, and is really an application of
standard economic theory. Some may question whether this litera-
ture on – large or small – personal financial incentives contains any
behavioural economics full stop. At face value, it does appear legit-
imate to contend that personal financial incentives are simply a tool
of the relative price mechanism. The counterargument could be that
small financial incentives may be more a means by which to offset
present bias rather than offering a genuine material benefit, in that
they perhaps subconsciously remind people that their immediate
action – e.g. smoking, doughnut eating – although enjoyable, is
potentially harmful. Even if one accepts that contention, the margins
are a little blurred, however, in that we would not know the mone-
tary threshold at which people alter their behaviour for strictly
financial reasons, rather than because they are being reminded impli-
citly that they are engaging or not engaging in potentially harmful or
beneficial acts.

22:37:11
07
motivational crowding 99

Notwithstanding these concerns, it can be contended that the


design of a financial incentive can be informed by what we know from
behavioural economics in order to try to improve the effectiveness of
the instrument. In Chapter 2, we learned that people tend to anchor on
outcomes when they are asked to value lotteries, and in Chapter 3 it
was noted that they also often overweight small probabilities.
The combined effect of these two affects is that we might expect
people to respond more to the promise of being allowed to play
a lottery – e.g. a 1 per cent chance of winning $500 – than they
would to the promise of receiving the expected value of that lottery –
i.e. $5 – for sure. In Chapter 4, deposit contracts – the so-called
commitment or Ulysses contracts – were mentioned, where people
pre-commit an amount of money that is returned to them only if they
meet a prior agreed behavioural change goal, as a means of tackling
present bias and utilising loss aversion.
Volpp et al. (2008) designed a study, albeit one that used a very
small sample, to evaluate specifically whether behavioural economic-
informed lotteries and deposit contracts are more effective than the more
standard practice of offering a small financial incentive when attempting
to motivate people to lose weight. The authors divided their participants
into three groups. The first group were placed in a weight monitoring
programme that required only a monthly weigh-in. The second group
were exposed to the weight monitoring programme plus a deposit con-
tract, where at the beginning of each month participants could deposit
between 1 cent and $3 per day, with the deposited amount matched by
the investigators in addition to a fixed payment of $3 per day, and with
all money refundable to any participant who met the targeted weight
loss at the end of each month. In the third group, participants received
the weight-monitoring program plus a lottery incentive, where if they
met their weight loss target each day, they were allowed to play a lottery
that had an expected value of about $3, with some of the lotteries having
a large payoff with small probability, but most having a small payoff with
larger probability. The end-point target weight loss for all participants
was 16 lbs – 7.25 kg – at sixteen weeks.

22:37:11
07
100 the origins of behavioural public policy

Volpp et al. (2008) found that the average weight losses at six-
teen weeks in the control group, the deposit contract group and the
lottery incentive group were 3.9, 14 and 13.1 lbs, respectively, and that
the proportions of those in each group achieving the weight loss target
of 16 lbs were 10.5%, 47.4% and 52.6%. So far so good for the beha-
vioural economics-informed incentive groups. Unfortunately, seven
months after the initiation of the study, the average losses across the
three groups had narrowed to 4.4, 6.2 and 9.2 lbs, a statistically insig-
nificant difference.
There have been previous studies that used deposit contracts
and lottery incentive mechanisms in the broad domain of health,
without recognising them as, or even realising that they are, beha-
vioural science-informed as such. They have shown mixed results. For
instance, Cahill and Perera (2008) commented that for smoking cessa-
tion, studies that used deposit contracts showed no greater effect than
those that used the conventional incentives, while Kane et al. (2004)
found that of ten studies that had used a lottery payment mechanism,
six had reported a significant effect. We may have to brace ourselves
for the possibility that even behaviourally informed personal financial
incentives will show limited effectiveness in sustaining change in
personal lifestyle behaviours, where the pre-existing activity – such
as smoking, drinking alcohol and eating fast food – offer considerable
enjoyment in the moment. Resistance to perceived control – cf. self-
determination theory – may also play a role. For instance, although
engagement with these incentive mechanisms is usually voluntary,
Green et al. (1986) argue that a social pressure to enrol and comply in
the short run – an introjected motivation – may be resisted by the
targeted population later, when they have formulated their arguments
for refusing to comply. Others have expressed the concern that these
external rewards will crowd out people’s intrinsic motivation to
behave in healthier ways (Shaw, 2007).
Thus, overall, personal financial incentives appear to offer more
promise for aspects of medical uptake and adherence that do not
require sustained behaviour change than as attempts to curtail

22:37:11
07
motivational crowding 101

personal lifestyle behaviours that are harmful in the long term. Giving
up things such as excessive plastic bag use does not impose the limita-
tions on a person’s pleasure to the same extent as does the curtailment
of cigarettes and alcohol, and therefore it is less difficult to achieve.
Moreover, whereas the use of plastic bags is generally framed as an
environmental concern and can thus be considered as a harm to
others, personal lifestyle behaviours in relation to one’s own guilty
or not-so-guilty pleasures are, for the most part, harms imposed on
oneself. Therefore, the tendency towards identifying with actions that
may be perceived as almost entirely personal on moral grounds is
perhaps less strong than when broader social implications have to be
considered, and there are lessons in this for where such incentives
might prove most successful in public policy. All of this does not bode
well for the use of personal financial incentives to discourage health
harming lifestyle behaviours over the long term, although one should
note that there is still a paucity of behavioural science-informed
evaluations in this sphere and thus it is perhaps right that the beha-
vioural public policy community is not yet ready to give up on this
type of intervention, as we shall see in Chapter 7.
Personal financial incentives are not the only type of incentive
that might usefully be informed by the findings of behavioural eco-
nomics, however. The design of incentive mechanisms targeted at
public sector professionals – to perhaps an even greater extent –
might also be beneficially informed as such.

designing performance management


Health care is also a good sector to look at when considering incen-
tives for public sector professionals, principally because there is
copious literature on the topic in this domain. However, the argu-
ments presented below can for the most part be transferred to other
public policy considerations, which has been done to some extent by
Bevan and Fasolo (2013) in relation to education policy. It is probably
fair to say that the principal interest in introducing performance-
related incentives in health care originated in the United States at

22:37:11
07
102 the origins of behavioural public policy

around the turn of the century, in part as a response to two reports,


To Err is Human and Crossing the Quality Chasm (Institute of
Medicine, 1999; 2001), that expressed serious concerns about the
quality of the health care provided in the United States. These reports
threw the spotlight on concerns about patient safety and led to
a heightened demand for professional accountability.
Influenced by the standard economic expectation that you get
what you pay for, some large employers and insurance companies in the
United States introduced so-called pay for performance instruments –
i.e. instruments that rewarded or punished health care providers finan-
cially on the basis of meeting certain performance standards. As a
consequence, it was expected, or perhaps hoped, that quality would
improve. In addition to the pure economic incentive to increase earning
power for health care professionals, some believed that pay for perfor-
mance would also motivate professionals via their intrinsic desire to
comply with best practice. That is to say that the professionals would
identify with the goals of the instrument. Pay for performance methods
are now used in many countries around the world in health care and
beyond.
Unfortunately, the evidence on the effectiveness of what might
be classified as pure pay for performance interventions – i.e. simple
and direct financial rewards for improved performance – appears to be
mixed, at best (Tanenbaum, 2009). For instance, Rosenthal and Frank
(2006), in a review of pay for performance in and outside health care,
found little evidence that the mechanism can improve quality, and
concluded that small scale bonuses are not generally effective.
However, if financial incentives seem large enough, and thus are
firmly informed by standard economic theory rather than behavioural
theory, they are more likely to have an effect, at least on behaviour if
not necessarily on quality, a powerful illustration of which can be
found in the United Kingdom’s National Health Service.
In 2004, the United Kingdom government introduced a new
contract for general practitioners – i.e., primary care doctors – making
part of their income dependent on their performance against a number

22:37:11
07
motivational crowding 103

of clinical and patient experience indicators. Each primary care doctor


typically increased their income by about a third – or £23,000 – as
a direct consequence of this performance framework (Doran et al.,
2006). Thus, the doctors certainly responded to the incentives,
although it is difficult to gauge the overall effect of the incentives on
quality. For instance, there is evidence that suggests that quality was
increasing before the introduction of the performance mechanism
and, at least in some clinical areas, such as coronary heart disease,
the pay for performance instrument had no discernable effect on this
trend (Campbell et al., 2009). What does seem to be clear is that in
order to have confidence that pure financial incentives will work,
a necessary but insufficient requirement is that they have to be quite
substantial, and that, of course, will have significant resource impli-
cations for public sector budgets.
Enter behavioural economics, which, as with personal financial
incentives, could be used to inform the design of performance manage-
ment initiatives in an attempt to strengthen their effect. Reference
points and loss aversion, discussed in Chapter 3, could have an impor-
tant role to play in this regard. Consider, for instance, the concept of
judging whether or not performance is acceptable, with acceptability
given by some predefined threshold. The threshold is the reference
point, and performance that is worse than this is, in some sense, a loss.
If those being assessed face the threat of being exposed publicly as poor
performers against the threshold, we may expect them to be highly
motivated to avoid this potential loss, which may in turn drive per-
formance improvements – at least to the extent that it might be
expected to ensure a decline in poor performance. This type of
mechanism is often called naming and shaming, and institutes as
a motivator the threat of reputational damage for poor performance
(Besley et al., 2009). We will see in Chapter 9 that a concern for
reputation is a key human characteristic that is important to the
concept of reciprocity.
A good example of a naming and shaming mechanism was also
applied in the National Health Service at the beginning of this

22:37:11
07
104 the origins of behavioural public policy

century, and was called the hospital star rating system. Through this
system, hospitals in England were assessed annually on a number of
indicators, including, most importantly, targets against waiting
times. Following assessment, hospitals were each awarded from zero
to three stars, with more stars indicating better performance. For very
poor performance, hospital management teams could be dismissed.
Implicitly, a two-star performance may have been taken as the refer-
ence point by many management teams; zero or one stars would be
seen as a loss against acceptable performance. The fact that perfor-
mance was widely publicised and that there was a threat of dismissal
demonstrates that the star rating system relied on naming and sham-
ing with clear potential consequences in order to motivate perfor-
mance improvements. In terms of reducing waiting times, it worked:
the National Health Service witnessed substantial reduced waiting
times during the period that the star rating system was in operation
(Besley et al., 2009). However, over the longer term a system that relies
on such tight control – bordering on terror – has the potential to
damage the identity that those who work within any organisation
feel towards that organisation, as self-determination theory predicts.
This poses a potential problem for public sector organisations in
particular, where workers are relied upon to have an intrinsic motiva-
tion to perform well.
There is a third type of performance management mechanism
that retains implicitly the notion of reference points and loss aversion,
but dampens the control and terror somewhat. This is a mechanism
where, often, relatively small financial incentives are combined with
the ranking of the relative performance of those who are subject to
assessment. An example of such a mechanism was introduced by the
Center for Medicare and Medicaid Services – the federal body that
oversees the publicly financed health care services for the elderly and
the relatively poor in the United States.
This mechanism was initially trialled for three years beginning in
2003, and was called the Hospital Quality Incentive Demonstration
project. Hospitals were assessed on a number of performance indicators

22:37:11
07
motivational crowding 105

Table 6.2 Hospital Quality Incentive Demonstration project

Percentage of patients experiencing the


quality indicators

Inception End of year 3

Acute myocardial infarction 87.5 96.1


Coronary artery bypass 84.8 97.4
grafting
Heart failure 64.5 88.7
Pneumonia 69.3 90.5
Hip and knee replacement 84.6 96.9

Source: Department of Health and Human Services (2008)


Previously published in Oliver, A., and L. Brown. 2011. Incentivising
Professionals and Patients: A Consideration in the Context of the United
Kingdom and the United States. Journal of Health Politics, Policy and
Law 36: 59–87. Reproduced with permission.

relating to acute myocardial infarction, coronary artery bypass grafting,


heart failure, pneumonia and hip and knee replacement. The hospitals
were ranked annually according to their performance, and those in the
top decile – i.e. the top 10% – received a 2% bonus on their standard
payments, while those in the next decile received a 1% bonus. Thus,
any particular hospital’s ranking in one year plausibly served as the
reference point on which they would want to improve, and certainly
not to worsen. Over the three years, the mechanism appeared to moti-
vate an overall average improvement of 15.8% (Department of Health
and Human Services, 2008). Table 6.2 summarises the performance
improvements by clinical category.
A further example of combining small financial incentives
with a ranking exercise was introduced in the Veterans Health
Administration – the body in charge of the health care system for
veterans of the United States armed forces – in the mid-1990s. In
this mechanism, health care managers were held accountable for
the performance of their hospitals against quite simple quality

22:37:11
07
106 the origins of behavioural public policy

Table 6.3 Pros and cons of performance management

Pay for Naming and Ranking


performance shaming

Motivation Relative price Reputational. Loss League table


mechanism. aversion. competition and
loss aversion.
Pros Can be effective Can be effective in May be effective
with large short term. May over longer term.
incentives. be cheap. Cheap.
Cons Expensive. May Undermines May demotivate
crowd out identity and some. Could
important demotivates crowd out also.
actions. over time.

indicators – e.g. some cancer screening rates, patient cholesterol


levels and the like. Senior managers were eligible to receive bonuses,
which in many cases had the effect of them putting pressure on their
clinical teams to improve performance, and the details of the relative
performance of each hospital, in the form of a sort of league table,
were disseminated throughout the whole system each year. As with
the example given in the previous paragraph, it is plausible that
a hospital’s position in the league table in any particular year served
as their reference point for the next year. The veteran’s system
traditionally had a reputation for poor quality, but within five years
of introducing the performance mechanism it demonstrated remark-
able improvements in quality, and by 2005 was outperforming all of
the other sectors of the United States health care system on almost
all of the quality criteria over which a comparison was possible.
Table 6.3 summarises the motivations for, pros and cons of the
three types of performance management mechanism discussed above.
There is an extensive literature on objections to, and possible pro-
blems with, performance management initiatives. The detractors of
this type of policy intervention make good points that should not be

22:37:11
07
motivational crowding 107

dismissed lightly. For example, it is regularly argued that performance


incentives can be highly detrimental to public sector services, if good,
non-incentivised practices are crowded out as a consequence, and if
workers are moved away from undertaking actions for their intrinsic
worth, expecting an additional payment for every good thing that they
do. Moreover, gaming activities – that is, underhand practices to
secure payments for performance improvements that have not been
realised – are a significant risk. However, the purpose of this section is
not to review that literature by detailing all of the pros and cons of
performance management. Rather, it is to highlight that if one is to
pursue performance management, incentives that are specifically
behaviourally informed are worthy of consideration.
In this chapter, compared to previous chapters, the emphasis
shifted somewhat towards paying more attention to how behavioural
economics can inform policy. Chapter 7 continues this trend.

22:37:11
07
7 Nudges

As noted at the beginning of Chapter 1, although empirical beha-


vioural economics – and behavioural science more broadly – now has
quite a long history, a concerted singular effort to inform the design of
public policy with the findings of this discipline has only gained
currency in relatively recent years. In the early 2000s, several leading
behavioural economists, with the help of some others, published con-
ceptual frameworks on how behavioural economics might be applied
to policy (Camerer et al., 2003; Thaler and Sunstein, 2003; 2008).
These early frameworks are actually similar to each other, in
that they are soft forms of paternalism, intended to encourage – not
mandate – consumers or citizens to behave in ways that they them-
selves would ideally prefer over their current actions. Moreover, the
frameworks are all modifications of welfare economics, in that they
retain wellbeing as the normative goal, but detach the achievement of
wellbeing from individual choice. However, Richard Thaler and Cass
Sunstein’s 2008 book, Nudge: Improving Decisions About Health,
Wealth and Happiness, proved to be much the more influential, and
in part led to Thaler being appointed as an advisor to the Behavioural
Insights Team in the United Kingdom, and Sunstein to the position of
head of the Office of Information and Regulatory Affairs in the United
States.
Part of the reason why Thaler and Sunstein’s book transcended
the academic debate is probably attributable to it being written as
a popular science book for the mass market, using a catchy buzzword –
nudges, the policy applications of their theory – to boot. However, this
of course is not the only explanation. In the same year that Thaler and
Sunstein published their book, much of the world experienced the
most serious financial crisis since the 1930s, which was caused

108

10:14:19
08
nudges 109

substantively by insufficient regulation of the banks due to too much


faith being placed in neoclassical economics and rational choice the-
ory. Some began to look to behavioural economics for answers.
Moreover, liberal-minded politicians, from the left and the right of
the political spectrum, frustrated with too much red tape presumably
in the non-banking sector of the economy, were searching for ways in
which they could motivate people to change their self and society-
harming behaviours, without imposing further regulations or bans on
either the general public or commercial organisations. Additionally,
given the new financial circumstances and the consequent pressures
on public finances, they were trying to identify policies that would be
inexpensive to implement. Thaler and Sunstein appeared to be offer-
ing some answers, or at least, suggestions.
Nudge policies and interventions are applications of a conceptual
framework called libertarian paternalism, an at face value oxymoronic
term until one learns that Thaler and Sunstein used the word libertar-
ian to modify the word paternalism in order to signify that their
approach is liberty-preserving. They contend that retaining the freedom
to choose is the best safeguard against a misguided policy intervention.
The approach is paternalistic in the sense of motivating behaviour
change that aligns with the target population’s deliberative preferences,
as implied above. Thus, libertarian paternalism relies on the assump-
tion that of the great many decisions that each human being makes
automatically and almost unthinkingly each day by following some
innate rules of thumb – see Chapter 2 – a few will lead us in directions
that if we really thought about it, we would prefer not to take. For
example, present bias may lead us to skip the gym and the overconfi-
dence heuristic might cause us to believe that we do not need the gym
in the first place, although much of the time the rules of thumb are
probably advantageous as they typically lead to satisfactory decision
making without us becoming incapacitated by too much thought.
This distinction between automatic and deliberative decision
making – thinking fast and slow – lies at the heart of Daniel
Kahneman’s (2011) intellectual autobiography, and is known in

10:14:19
08
110 the origins of behavioural public policy

psychology as dual process theory. The distinction between thinking


fast and slow has been categorised in a variety of ways. Thaler and
Sunstein defined it as human versus econ ways of thinking; Stanovich
and West (2000) had earlier called it system 1 versus system 2.
However, the distinction was recognised much earlier still, and
resonates, for example, with Adam Smith’s notion in The Theory of
Moral Sentiments (2010 [1759]) of the impartial spectator – the man
within in the breast – casting judgment on our passion-led actions.
Still further back, and although not exactly the same, the concept
brings to mind Plato’s allegory in Phaedrus, where the charioteer –
our reason – attempts to drive and is driven by two winged steeds, the
representation of the parts of our soul that are good and unruly. It is
noteworthy, though, that none of these categorisations of reflexive
versus reflective decision making imply that different parts of the
brain are operating entirely separately from each other, but that the
two ways of thinking work together in complex ways to influence our
choices and behaviour.
The core of the nudge approach is that knowledge of behavioural
economic phenomena, such as present bias, loss aversion, probability
weighting, peak-end evaluation and motivational crowding, among
others – that is, knowledge of how people often do behave and decide,
rather than how they ought to behave according to the assumptions of
rational choice theory – can be used to reshape the environment, or
choice architecture, that people find themselves in, so that their
automatic decisions are more likely to align with their deliberative
preferences. From the above discussion, some of the requirements for
an intervention to be defined appropriately as a nudge are mentioned,
but I think it is important for behavioural public policy analysts to be
very clear on the distinctions of the ways in which behavioural eco-
nomics can inform policy, because different behavioural economists,
to name just one subgroup of the population, have different views on
which approaches are the most appropriate. Libertarian paternalism is
one of the ways in which behavioural economics can inform policy, so
let us briefly consider everything that a nudge entails.

10:14:19
08
nudges 111

the nudge requirements


A close reading of Thaler and Sunstein (2003; 2008) reveals that
nudges are intended to work with the grain of how people already
make their decisions, and thus are trying to work with automatic
decision making processes. That is to say, they are not intended to
change people’s opinions about their behaviours in an overt, explicit
way through information or persuasion campaigns and education
programmes, such that people come to a deliberative decision to
change their behaviours. As noted earlier, nudges are meant to pre-
serve liberty, and do not involve the regulation of activities or outright
bans. There should be no burden placed on those who choose their pre-
existing behaviours rationally and thus wish to continue with those
behaviours.
A further feature of nudges is that they are not meant to
significantly change the economic incentives of those towards
whom they are targeted. Small economic incentives are allowed,
presumably under the reasoning, mentioned in Chapter 6, that they
do not drive material concerns, and rather just serve as almost
unconscious supportive – not controlling – reminders that the
incentivised action is beneficial to the individual. Moreover, large
incentives face the charge of being coercive, thus undermining
liberty.
Further, it goes without saying by now that nudges should be
informed by behavioural economics, or behavioural science. Finally,
libertarian paternalism is, as its name suggests, a form of paternal-
ism, and thus the emphasis is on changing behaviour to benefit those
who are nudged, not to reduce the harms that these behaviours may
impose on other people. Reducing harms to others is a different
sphere of policy; in economics, these harms are known as negative
externalities – i.e., they are external to the individual who is acting.
The harms that the individual imposes on himself are known as
negative internalities, and paternalistic policies are intended to ame-
liorate those.

10:14:19
08
112 the origins of behavioural public policy

Table 7.1 The nudge requirements

The intervention:
relies on Automatic decision making processes.
is Liberty-preserving.
does not use large financial Incentives.
is informed by Behavioural economics.
targets Internalities.

The nudge requirements can be remembered more readily with


the use of the mnemonic ALIBI, summarised in Table 7.1. Checking
any purported nudge against this table will thus help to ensure that no
offence has been committed.
This is not to imply that, for example, information campaigns,
regulation and large incentives are necessarily bad interventions; it is
merely to illustrate that these are not nudges.
The private sector, including supermarkets, financial services
and magazine subscription companies, to name just a few, make
subtle changes to the choice architecture in order to maximise sales
or profits all of the time, of course, and the marketing industry has
implicitly demonstrated substantial expertise in using behavioural
science towards these ends for decades. Libertarian paternalists con-
tend that it is legitimate for the public sector to counter these manip-
ulations with nudges, so as to steer people towards making decisions
that better serve their long term deliberative goals, and thus, in con-
sequence, improve their welfare. Moreover, the advocates for nudging
maintain that it is inevitable that we all face some form of choice
architecture – the default position, if you will – and therefore it is
legitimate to allow governments to shape the default so that it best
serves the good of the people they serve. As noted in Chapter 1, two
years after the publication of Thaler and Sunstein’s book, the then
new Conservative-Liberal Democrat coalition government in the
United Kingdom became the first in the world to embrace nudge
formally at the central policy making level.

10:14:19
08
nudges 113

the british movement


David Halpern, the head of the Behavioural Insights Team, details its
development in his 2015 book, Inside the Nudge Unit. Initially, the
Team was given a two-year remit – subsequently extended – to meet
a number of objectives, or face closure. The objectives were to trans-
form at least two major areas of policy, to spread understanding of
behavioural approaches across the various government departments,
and to achieve a tenfold return on its costs. From reading Halpern’s
book, one gets the impression that much thinking on feet was done,
but planning for what the Team would look like and the tasks that it
would do did have a period of gestation, if not a very long one.
A couple of months prior to the coalition government’s election
to office, a number of those who would go on to work for the
Behavioural Insights Team in various capacities published the intel-
lectual foundation for the approach that was subsequently adopted.
The document that they produced is known as the Mindspace Report
(Cabinet Office and Institute for Government, 2010). Given the impor-
tance of the report to the foundation of the world’s first dedicated
central government behavioural unit, which has subsequently served
as something of a blueprint for the establishment of similar units
elsewhere, the principal conceptual content of the Mindspace
Report is worth considering briefly.
The word Mindspace is a mnemonic that the authors of the
report comprised to highlight what they felt were important charac-
teristics to consider when designing behavioural interventions and
policies. These characteristics are summarised in Table 7.2.
The characteristics are, hopefully, explained sufficiently within
the table, but just to flesh them out a little bit, Messenger contests
that we are more likely to act on information given by those whom we
trust, and the predictable mental shortcuts under Incentives refers to
those we have considered earlier in this book – which are arguably
the most important empirical findings in behavioural economics –
including present bias, loss aversion and the various rules of thumb.

10:14:19
08
114 the origins of behavioural public policy

Table 7.2 Mindspace

Messenger: We are heavily influenced by who communicates


information.
Incentives: Our responses to incentives are shaped by predictable mental
shortcuts.
Norms: We are strongly influenced by what others do.
Defaults: We go with the flow of pre-set options.
Salience: Our attention is drawn to what is novel and seems relevant to us.
Priming: Our acts are often influenced by sub-conscious cues.
Affect: Our emotional associations can powerfully shape our actions.
Commitments: We seek to be consistent with our public promises, and
reciprocate acts.
Ego: We act in ways that make us feel better about ourselves.

Previously published in Cabinet Office and Institute for Government.


2010. MINDSPACE: Influencing Behaviour through Public Policy.
London: Cabinet Office and Institute for Government. Reproduced with
permission.

Norms relates to one of the most substantive findings in the broader


psychology literature, where it has been shown that people very often
follow the actions of their peers, rather than risk being an outlier. This
is sometimes referred to as the herd effect. We want to fit in, we want
to show that we are good co-operators and we want to learn. Defaults
exploit status quo bias – the observation that human beings are very
good at not acting at all – and Salience relates to the anchoring heur-
istic, where we do not attach appropriate weight to all relevant infor-
mation when reaching our decisions. Priming is also associated with
anchoring, and is an important finding in the psychology literature.
The observation by Ariely et al. (2003) that the amount of money that
people are willing to pay for, say, wine can be influenced by asking
them to think about digits in their social security numbers, men-
tioned in Chapter 2, is an example of priming. Affect contests that
our emotions can impact on our decisions; for example, transient
fluctuations in optimism and pessimism can affect our judgments

10:14:19
08
nudges 115

substantively. We have talked about commitment contracts and,


related to this, there is some evidence that merely stating our inten-
tions to others will increase the chances that we will achieve our goals.
Commitments may feed off our concern for our own reputations – that
we are seen as people who keep to our word – which in turn feeds into
reciprocity, as we will see in Chapter 9. Finally, Ego may also relate to
our concern for a good reputation, where we act to achieve and empha-
sise the sense of a positive self-image, possibly in the hope that others
will concur with our self-esteem.
Taken one at a time on their own terms, it is difficult to disagree
that each of the characteristics summarised in Table 7.2 is an important
consideration in behavioural public policy. Some, however, may won-
der whether the mnemonic preceded the list, and that the chosen
characteristics were in some sense a little forced. This may in itself
not be an entirely bad thing, as the catchy mnemonic helps people to
memorise many important features, but as is reasonably clear from the
preceding paragraph, many of the characteristics are highly intercon-
nected and could have perhaps been boiled down somewhat. At the risk
of belabouring the point, Messenger and Norms both seem to be driven
in part by trust, and Norms, Defaults, Salience and Priming are asso-
ciated with giving people clear reference points or anchors. Norms,
Commitments and Ego are all associated with reputation and recipro-
city, and Commitments and Priming may gain their effect from, and be
used to tackle, the mental shortcuts, such as loss aversion and present
bias, that the Mindspace Report groups under Incentives. By focussing
in on a fewer number of characteristics – for example, the robust find-
ings of present bias, loss aversion, peak-end evaluation and some
others – the report may have offered nascent behavioural public policy
analysts a more manageable and tangible set of findings to coalesce
around. This being said, the reader may legitimately disagree, and I do
not wish to criticise the Mindspace Report too much because it is
nonetheless a useful and important document.
The authors of the Mindspace Report stated explicitly that they
were interested in the soft touch of policy rather than regulations or

10:14:19
08
116 the origins of behavioural public policy

bans, and that they intended for their interventions to go with the
grain of, rather than fight, human nature. There are clear similarities
here with at least some of the requirements of libertarian paternalism,
which would be carried over later to the Behavioural Insights Team
with Richard Thaler serving as an advisor. Yet the report did not
correspond with all of the nudge requirements, proposing, for
instance, that its approach be used to address negative internalities
and externalities, which took it beyond a pure paternalistic remit.
Politically, this is understandable, in that attempts by public policy
makers to curtail private behaviours that are harmful to others may
gain broader general public support than efforts to alter personal life-
style behaviours that present no external harms. It is, after all, the
latter type of policy action that falls foul of cries that governments are
introducing a nanny state. However, the Behavioural Insights Team
that subsequently adopted the Mindspace Report became known col-
loquially as the nudge unit, which introduces a degree of confusion in
the research and policy community as to what a nudge actually is.
With reference to the original nudge requirements, we may wonder
about the extent to which the Behavioural Insights Team is engaged in
nudging at all, a point to which we will later return.
The Mindspace authors wrote that whenever the effectiveness
of behavioural public policy interventions relies on triggering people’s
automatic system – their system 1 way of thinking – governments will
inevitably leave themselves open to charges of manipulation. Thus,
they contended that public permission for the nudge approach is
required, although they also maintained that the extent to which the
Mindspace approach preserves substantive freedoms is likely to come
down to political judgment. There may be a dilemma here, however,
in that behavioural public policy covers a large range of qualitatively
different interventions, from the profound and, for many, unaccepta-
bly intrusive – such as defaulting people into particular actions, such
as organ donation, that they would rather not do but are not fully
cognisant that they are now enrolled into – to the seemingly innoc-
uous, such as moving healthy and unhealthy food items around on

10:14:19
08
nudges 117

canteen shelves to alter their salience. One might therefore question


whether translating public permission for the policy approach at the
very general level to the level of each individual intervention is legit-
imate. The philosopher Luc Bovens implies that it is legitimate, by
arguing that ethical acceptability does not require governments to
explain that a particular intervention has been implemented, particu-
larly if full transparency limits the effectiveness of the intervention,
so long as those being nudged have the ability to discern its imple-
mentation (House of Lords, 2011). Bovens’ view is contestable, but
perhaps it is now time to take a look at how the Behavioural Insights
Team applied the Mindspace framework.

proposals and suggestions


As noted above, the Behavioural Insights Team was established initi-
ally under the condition that it would achieve a lot in a short space of
time. It therefore wasted no time in publishing its first applied report,
which focussed upon proposed interventions to encourage personal
health improving behaviours (Behavioural Insights Team, 2010). The
Team received a lot of attention, and were held widely as pioneers in
the use of nudges in public policy, becoming known colloquially as
the nudge unit, as aforementioned. To help deduce whether this
nomenclature was justified, let us consider a diagrammatic depiction
of libertarian paternalism.
If we look once again at the alibi mnemonic in Table 7.1, we can of
course conclude that automatic thinking is associated with the beha-
vioural economic phenomena, and that financial incentives are a tool of
either behavioural economics or standard economic theory, depending
on their size. Thus, if we fold the automatic and incentives requirements
of the mnemonic into the behavioural requirement, we are left with
three core features of libertarian paternalism; namely, that it addresses
internalities rather than externalities, that it preserves liberty and is
therefore antiregulatory and that its applications are informed by beha-
vioural economics rather than the standard model of rational choice.
These features are each represented by an axis in Figure 7.1.

10:14:19
08
118 the origins of behavioural public policy

Regulatory

Classic nudge
Liberty

Behavioural
Internalities Externalities

Rational

figure 7.1 The nudge space

In the figure, moving towards the origin on the vertical axis


indicates that a policy is increasingly liberty-preserving and antiregu-
latory. Moving towards the origin on the horizontal axis indicates that
a policy is increasingly addressing negative internalities rather than
negative externalities, and moving towards the origin on the diagonal
axis indicates that a policy is increasingly informed by behavioural
economics rather than standard economic theory or rational choice
theory. Consequently, a classic nudge – preserving liberty, addressing
negative internalities and informed by behavioural economics – must
lie at the point where the axes intersect, as indicated in the figure.
An example of such an intervention that is often highlighted in the
nudge literature is to place fruit rather than chocolate bars at super-
market checkout counters – in such circumstances, a shopper is not
required to buy fruit – liberty – but might, on reflection, prefer to buy
fruit rather than chocolate – internalities – and is more likely to now

10:14:19
08
nudges 119

buy fruit because the presence of fruit near to the checkout has
invaded his immediate and automatic decision making mental appa-
ratus – behavioural economics. Similarly, in some countries council
authorities have painted green footprints that lead up to rubbish bins
in order to encourage people to discard their litter appropriately,
which is almost a classic nudge if we assume that appropriately dis-
carding their litter is something that casual litterbugs would delibera-
tively want to do. The policy of introducing a small charge for plastic
bag use in supermarkets discussed in Chapter 6 is not, by these cri-
teria, a nudge, however, because – without cheating – people have no
way of not paying the charge while still receiving a bag, and thus their
liberty in this respect has not been preserved.
The three-dimensional space depicted in Figure 7.1 can thus
be used to give a visual representation of the extent to which the
intended health improving interventions proposed by the Behavioural
Insights Team align with the requirements of libertarian paternalism.
Figures 7.2a and 7.2b depict these interventions in the nudge space.
The top left corner of each box within the spaces is assigned a number
that aligns with the numbered paragraphs that immediately follow the
figures, with each paragraph describing an intervention proposed by the
Behavioural Insights Team. To be sure, the distances from the origin on
each dimension in the figures rely on my judgement, but a balanced
judgement nevertheless offers an indication of the extent to which the
approach that was embraced so enthusiastically in the UK govern-
ment’s rhetoric – and elsewhere – has informed the interventions pro-
posed by the unit that was formed on the basis of that rhetoric.

1. The Behavioural Insights Team proposed experimentation with the


commitment or deposit contracts that were considered earlier in this book,
in order to incentivise smoking cessation. Smokers would commit
voluntarily small amounts of their own money to a smoking cessation
programme, to be returned to them if they remain abstinent. Getting poor
people to pre-commit even small amounts of money may be particularly
difficult, and it is in these groups that smoking tends to be the most
prevalent. Nonetheless, the intention is that loss aversion will cause

10:14:19
08
120 the origins of behavioural public policy

Regulatory

Externalities
3

Rational

figure 7.2a Health improving interventions 1–4


Previously published in Oliver, A. 2015. Nudging, Shoving and Budging:
Behavioural Economic-Informed Policy. Public Administration 93:
700–714. Reproduced with permission.

participants to be particularly keen to meet the programme’s objectives.


The intervention is therefore behaviourally motivated, is non-compulsory
and is focussed on addressing negative internalities, although a concern
with saving money for the National Health Service – an externality
consideration – was also given as a justification for the intervention.
Enrolment into commitment devices appears to require more conscious
deliberation than a simple alteration of the choice architecture entails, but
in terms of the three-dimensional space, it appears to conform reasonably
closely to the characteristics of a nudge.
2. In order to increase the number of people on the organ donor register, the
Behavioural Insights Team proposed piloting a move from an opt-in system
to one of prompted choice, where people have to state whether or not they
wish to be a donor at the time of applying for or renewing their driving
licence. This policy was subsequently introduced in England; in Wales, the
policy went one step further, to that of presumed consent, also known as an
opt-out system, where people have to register to refuse to be a donor.
In countries that have an opt-out system, such as France and Portugal,

10:14:19
08
nudges 121

Regulatory

8
5

Externalities
7
6

Rational

figure 7.2b Health improving interventions 5–8


Previously published in Oliver, A. 2015. Nudging, Shoving and Budging:
Behavioural Economic-Informed Policy. Public Administration 93:
700–714. Reproduced with permission.

almost the whole of the adult population is listed as potential organ donors,
compared to roughly 20 per cent in opt-in countries, and the UK
government is also using this type of default intervention to increase the
uptake of personal pensions. The change in the default position to
prompted choice has had some success in increasing the number of people
on the organ donor register in other countries, and is behaviourally
motivated. At face value, prompted choice preserves autonomy in that
people are free to refrain from being on the register if they so wish, although
it is possible that some might feel that they are being coerced by the use of
such a mechanism. Moreover, the introduction of prompted choice
requires regulation. The policy appears to be driven by externality
considerations in that the principle motivation is that others may benefit if
one registers, although considerations of reciprocity can play a role – i.e., if
more people register, we all may benefit.
3. The Behavioural Insights Team suggested that teenagers mentor toddlers
in an attempt to reduce the rate of teenage pregnancy. If the targeted
teenagers are allowed to opt out of the programme if they so wish, then the

10:14:19
08
122 the origins of behavioural public policy

intervention is liberty-preserving. The intervention is driven to some


extent by externality concerns, but does also address negative internalities
if the teenagers, on reflection, genuinely would rather not have a child.
However, the intervention seems only tenuously linked to behavioural
economics, with a possible association with George Akerlof and Rachel
Kranton’s (2010) work on identity economics, where a closer identification
with a particular situation might be expected to lead to a change in
individual behaviour, which is in turn associated with self-determination
theory, discussed in Chapter 6. That is, a greater identification with all that
is required of a mother may motivate some teenagers to more carefully
avoid pregnancy. That said, this particular intervention may equally be
informed by rational choice theory, where more knowledge and
information might lead to different decisions.
4. Due to some evidence that students think that their peers drink more
alcohol than they actually do, the Behavioural Insights Team proposed to
communicate accurate drinking levels to university students. Although
the inaccurate perceptions of alcohol consumption imply a misaligned
reference point, the provision of better information also appears to appeal
to rational decision making. Moreover, although there is some concern
with negative internalities, the thrust of the approach focuses on
addressing negative externalities, in particular costs to the National Health
Service. The intervention leaves students at liberty to drink what they
wish and thus is placed quite far down the vertical axis in Figure 7.2a.
5. It was proposed that markings be placed on supermarket trolleys to define
areas where fruit and vegetables should ideally be placed, as a means of
encouraging increased purchasing of these products. It could reasonably be
argued that this intervention manipulates a reference point, but the
Behavioural Insight Team’s motivation for encouraging healthier eating
habits is again focused upon saving health services resources, this time in
relation to obesity. Consumers would remain free to purchase whatever
amount of fruit and vegetables they wish, and thus their liberty is preserved.
6. The use of small incentives to attempt to change behaviours in children is
proposed for more than one intervention by the Behavioural Insights Team.
For instance, they announced a partnership with LazyTown, an initiative
that has been operating nationally in Iceland since the mid-1990s, in which
children sign contracts with their parents that rewards them for eating
healthily, going to bed early and being active. A further initiative allows

10:14:19
08
nudges 123

diabetic children to claim points, exchangeable for computer games, for


consenting to blood sugar measurements. As was noted in Chapter 6, it can
be contended with some justification that such incentives are tools of
rational economic theory, although, as also previously pointed out, if the
incentives are very small then their objective may primarily be to engage
children in the immediate moment, feeding principally on present bias.
Depending on the strictness of the parents, these methods may preserve
liberty to a considerable extent.
7. The Behavioural Insights Team also proposed the use of non-material
incentives, in the form of so-called piano stairs. These are stairs in, for
example, train stations that are embedded with sound emitting sensors, to
encourage people to avoid escalators and lifts. One could argue that they are
associated with present bias in that they make a task that many people
would otherwise avoid more enjoyable in the moment. People are not
forced to use the stairs, and thus liberty is preserved, although the
Behavioural Insight Team’s stated motivation was again to save money for
the National Health Service – apparently a powerful argument in the
context of the United Kingdom – and is thus targeted to a considerable
extent at addressing externalities.
8. Finally, the Behavioural Insights Team highlighted the United Kingdom’s
National Food Hygiene Rating Scheme as an example of good practice,
where food outlets post voluntarily on their entrances their hygiene rating
in the form of a standardised sticker, from zero points to indicate that
urgent improvement is necessary to five points if very good hygiene
standards have been met. Similar schemes have been used with some
success in other countries (Thaler and Sunstein, 2008). One could argue
that behavioural change is motivated by this scheme in a manner similar to
that observed with the hospital star rating system discussed in Chapter 6,
in that food outlets may take a middling score of, say, three to four points,
as their reference point of acceptability, and will be keen to avoid the
psychological loss associated with scores worse than that. It is important to
note that substantial improvements in performance were observed with
the similarly constructed hospital star rating system when patient/
consumer choice was absent, and therefore performance improvements are
not merely rational demand-induced responses. The food rating scheme is,
however, yet another intervention that is focussed on addressing negative
externalities, this time imposed on consumers by unhygienic food outlets.

10:14:19
08
124 the origins of behavioural public policy

Figures 7.2a and 7.2b give the impression that the type of interven-
tions proposed initially by the Behavioural Insights Team do not fit
too tightly with the requirements of nudge theory. More specifically,
while the proposed health improving interventions do not, for the
most part, rely on regulation, their use of behavioural economics is
sometimes not entirely obvious and they often appear to focus upon
addressing externalities rather than internalities. As noted earlier, the
tendency towards externalities is perhaps understandable due to the
reluctance shown by many politicians to face accusations of nanny-
ing, and thus the Behavioural Insights Team, which was after all
established to advise politicians, may not have wanted to give the
impression that they wished to interfere in personal lifestyle choices
too much, unless, of course, those choices appear to impose harms on
others. Moreover, when we remember that an important part of their
remit was to save public resources, it is more understandable still that
the Behavioural Insights Team often sought to emphasise that their
proposals would reduce the burden on the National Health Service.
The drift away from nudge policy was not health-specific. For
instance, another of the Behavioural Insights Team’s early reports
focussed upon energy – electricity, gas – use (Behavioural Insights
Team, 2011). The proposals within that report are targeted at improv-
ing environmental consciousness and reducing carbon emissions, and
are thus inevitably focused principally on externalities. That said,
some consideration of internalities is also present because individuals
can save themselves money by acting upon at least some of the
proposed initiatives, and, moreover, an increasing number of people
are internally motivated – cf. identification in self-determination the-
ory – to reduce their negative externalities in this regard. Figure 7.3
depicts the proposals in the form of the nudge space. Again, the top left
corner of each box within the space is assigned a number that aligns
with the numbers given to the paragraphs that follow the figure.

1. The Behavioural Insights Team proposed the introduction of upfront


incentives for environmentally beneficial behaviour, which may appeal to

10:14:19
08
nudges 125

Regulatory

2 Externalities

Rational

figure 7.3 Energy saving interventions


Previously published in Oliver, A. 2015. Nudging, Shoving and Budging:
Behavioural Economic-Informed Policy. Public Administration 93:
700–714. Reproduced with permission.

notions of present bias by offsetting the immediate costs associated with


being more energy conscious. For example, a one-month holiday from
council tax payments – the tax that residents pay to their local government
authorities for the provision of services, which varies across geographical
areas but is typically in the region of £1,000–2,000 per year – was proposed
for those who agree to insulate their lofts. These incentives cannot
therefore be classified as small. As with the health-related incentives, the
interventions do not seem to be targeting automatic decision making, and
some may contend that they introduce at least a small element of coercion,
particularly with respect to poorer groups.
2. The Behavioural Insights Team emphasise the potential usefulness of
smart electricity meters that offer a reading of the average energy
consumption of similar households in addition to one’s own electricity
use. This additional information can of course be ignored and therefore
may not impinge upon autonomy, but it could reasonably be contended

10:14:19
08
126 the origins of behavioural public policy

that the smart meters are merely providing additional information and are
hence forms of persuasion rather than nudges. However, it could equally be
argued that the average consumption level of similar households offers
a reference point against which consumers might perceive higher energy
consumption as a loss, motivating them to use less electricity.
3. The Behavioural Insights Team highlight as a good example the
government’s attempt to reduce electricity use within its departments by
experimenting with changing default settings in heating and cooling
systems. Changing this default requires an element of regulation, but if the
employees within the departments have the capacity to override the
default settings and face no internal or external sanction for doing so, then
their autonomy is for the most part protected. As we have learned in this
book, default manipulation is behaviourally motivated.
4. Paralleling the discussion in the latter part of Chapter 6, the Behavioural
Insights Team note that the United Kingdom government publishes league
tables that show the extent to which each department is demonstrating
progress towards a pan-government energy saving target. The autonomy of
each department is compromised by this initiative in that their
performance is involuntary recorded for all to see, but the intervention is
behaviourally informed if, as seems reasonable, each department perceives
average performance as the reference point. Consequently, any
performance worse than this, via an aversion to a loss of reputation, may
serve as a particularly strong motivator to improve.

As with the health improving proposals, the energy saving


initiatives are generally far from the origin within the nudge space.
All in all, the so-called nudge unit has not, for the most part, been
advocating nudges at all. One could take the view that this matters,
because it draws something of a mist over the intellectual clarity of
applied nudge policy, and, moreover, it is important to try to hold
governments accountable for their rhetoric, in the United Kingdom
and elsewhere.
Having said all this, it is also important not to castigate the work
of the Behavioural Insights Team. As has been noted, they had targets
to meet that were both strict and broad over a short time frame in an
area of public policy that they were doing much to progress. Moreover,

10:14:19
08
nudges 127

given the paucity of evidence of effectiveness, even today, in relation


to most aspects of behavioural public policy, including many of the
interventions summarised in this section, the Behavioural Insights
Team’s recommendation for experimentation and piloting before
a broader roll-out of any of the initiatives was surely the sensible
path to pursue. Further, although labelled the nudge unit in the pop-
ular and policy discourses, the Behavioural Insights Team has made it
clear, particularly in recent years, that their interest in behavioural
public policy extends beyond the parameters of nudging (and this was
also implicit in the Mindspace Report). Indeed, although potentially
important in many domains of public policy, the original require-
ments of libertarian paternalism limit substantively the full policy
potential of behavioural economics. The Behavioural Insights Team,
and others, has understandably applied alternative behavioural eco-
nomic policy frameworks, and it is to these that we now turn.

10:14:19
08
8 Shoves and Budges

Nudges, as noted, have captured the popular imagination and beha-


vioural economics-informed policy has to some extent been captured
by the nudge advocates, to the extent that many now equate beha-
vioural economic applications with anti-regulatory nudging and
believe that behavioural economics can be used for nothing else.
One ought to remember, however, that behavioural economics is
really nothing more than the inquiry into how people make decisions,
and as such, can be used to inform, and sometimes justify, most public
policy frameworks.
Moreover, libertarian paternalism has attracted criticism on
a number of fronts, including the contention that many nudges,
including, for example, supermarket trolley markings and piano
stairs, while interesting, are mere gimmicks that are ill equipped to
tackle the major problems that contemporary societies face. When
gimmicks are emphasised, presumably to attract attention, and
when the intellectual clarity of the libertarian approach becomes
diluted, there is a danger that people will lose patience with the
approach. For instance, I have heard alarm clocks and road signs
described as nudges, but to argue that they are behavioural eco-
nomic-informed is perhaps a stretch too far. One could retort of
course that nudges are much more than this, and even if they were
just that, small movements towards desirable policy goals should not
be sniffed at, particularly if the interventions that have brought these
improvements about are almost costless. Furthermore, most liber-
tarian paternalists and certainly many leading behavioural econo-
mists contend that nudges are meant to complement rather than
replace, or, as some have charged (Marteau et al., 2011), crowd out,
other forms of policy intervention (Loewenstein et al., 2012),

128

22:37:38
09
shoves and budges 129

although one cannot discount the possibility that an ideological


preference for non-regulatory interventions among policy makers
will encourage officials to exclude consideration of potentially
more important and effective regulatory measures when thinking
about behaviour change. Even some who are sympathetic to the
behavioural economic justification for wanting to bring about
change in personal lifestyle decision making believe that by retain-
ing a respect for autonomy of choice, nudges will never be suffi-
ciently effective.
Nudges have been further criticised for being manipulative, and
that a covert policy intervention cannot possibly preserve liberty. It is
important to note that many of those who share this concern do not
contend that nudges are necessarily physically hidden, but rather that
they are meant, in theory, to go unnoticed; to be part of the furniture,
so to speak. Indeed, this has possibly been the most prominent criti-
cism of the approach, with some questioning whether it is legitimate
to give governments carte blanche to introduce policies that, if not
explicitly hidden, are designed to go unnoticed, particularly policies
that are meant to alter behaviours that impose no harms on others.
As noted in Chapter 7, even the authors of the Mindspace Report
acknowledged this possible problem. Libertarian paternalists might
reply that they wish only to nudge people towards their own delib-
erative preferences – what people themselves, on reflection, say that
they want. However, even if one assumes that the ends might justify
the means in this respect – i.e. that it is legitimate to manipulate
behaviours covertly towards an accepted objective – some would con-
tend that we cannot actually discern what the ends are, because there
is no guarantee that deliberative preferences exist (Sugden, 2009).
Some even believe that libertarian paternalism is at odds with open
deliberative democracy, and that nudges, and indeed harder forms of
paternalism, are inconsistent with many people’s perception of what
it means to be a free society. However, this opposition to covert policy
interventions may be to some extent relaxed when they are intended
to reduce the harms that individual behaviours impose upon others,

22:37:38
09
130 the origins of behavioural public policy

which might be a reason why the Behavioural Insight Team’s propo-


sals have often been targeted at externalities rather than internalities.
Sugden’s concern that there is no guarantee that the limitations
on attention, cognitive ability and self-control that affect the indivi-
dual’s automatic responses are absent from their deliberative decision
making has led him to conclude that behavioural welfare economics,
of which libertarian paternalism is a part, is misguided, as we are
unable to discern what wellbeing is. Thus, rather than retain well-
being as the normative criteria but detach the definition of wellbeing
from individual choice – which is essentially how behavioural welfare
economics modifies welfare economics – Sugden proposes that oppor-
tunity, or the expansion of choice, should be the normative criteria,
without necessarily claiming that increases in opportunity improve
wellbeing. However, Sugden, who has made substantial contributions
to behavioural economics, is something of an outlier in this respect, at
least within the behavioural economics community. This chapter will
focus on alternatives to nudge theory that by and large retain the
assumptions of behavioural welfare economics. Let us begin by con-
sidering a harder form of paternalism.

coercive paternalism
In 2013, the philosopher Sarah Conly published a book in which she
adopted a similar justification for wanting to effect behaviour change
as that outlined by the libertarian paternalists; namely, that beha-
vioural economic phenomena, in particular present bias, can cause
people to make errors in their decision making that should ideally be
corrected. Although the designs of nudge interventions are informed
by behavioural economics, coercive paternalism uses behavioural
economics to inform when hard forms of paternalism are appropriate;
the design of the legislation itself, which focusses on issuing bans,
does not tend to be informed by behavioural economics. Conly
believes that the autonomy-respecting aspect of nudges will render
that type of intervention insufficiently effective. Thus, she calls for
the explicit regulation of citizens’ behaviours where such regulation

22:37:38
09
shoves and budges 131

Liberty

Classic shove
Regulatory

Behavioural
Internalities Externalities

Rational

figure 8.1 The shove space

indisputably brings benefits to those who are targeted that outweigh


the costs of restricting their freedom. Conley’s framework is called
coercive paternalism, and applications of this framework are known
as shoves.
As with libertarian paternalism, the basic requirements of coer-
cive paternalism can be represented in a three-dimensional space,
depicted in Figure 8.1. As with the nudge space in Figure 7.1, moving
towards the origin on the horizontal axis indicates that a policy is
increasingly addressing negative internalities rather than negative
externalities, and moving towards the origin on the diagonal axis
indicates that a policy is increasingly informed by behavioural eco-
nomics rather than standard economic theory or rational choice the-
ory. Unlike Figure 7.1, however, the shove space inverts the vertical
axis, so that moving towards the origin here shows that an interven-
tion is increasingly regulatory and anti-libertarian. A classic shove,

22:37:38
09
132 the origins of behavioural public policy

which would call for an outright ban on a particular self-harming


activity that is informed by behavioural economics, would therefore
lie at the origin in Figure 8.1.
It is important to keep in mind that Conly is not recommending
that great swathes of personal lifestyle behaviours be banned, but
argues quite forcibly that some actions, undertaken due to a lack of
self-control and bounds on human rationality, have substantial poten-
tial to seriously and irrevocably interfere with important, broadly
held, life goals; in short, these actions, if left unchecked, are them-
selves liberty-inhibiting. She maintains that legislating against these
actions is the least costly and most effective way to prevent people
from engaging in them.
Central to coercive paternalism is deciding on what might be
the broadly held life goals. Conly maintains that these are health and
financial security. Consequently, she recommends the introduction
of legislation pertaining to cigarettes, obesity-inducing practices, and
insufficient retirement savings. Similar to the proposed nudges sum-
marised in Chapter 7, however, there is sometimes a tendency for
interventions that are grouped under the shove heading to wander
a little away from the shove requirements, which risks generating
confusion. For example, although Conly’s recommendation to place
a ban on smoking is a fairly clean-cut shove, further recommenda-
tions, such as regulation against mortgages offered at greater than, say,
50 per cent of a person’s income and a ban on trans-fats above max-
imum acceptable limits, while perhaps laudable, are regulations
against potential supply side harms rather than the personal beha-
viours of citizens, and are therefore focused on ameliorating negative
externalities rather than tackling negative internalities. Moreover,
while lenders may use present bias to sell unfavourable mortgages,
the consumption and use of trans-fats do not appear to be behaviou-
rally motivated; rather, their use is economically motivated, due
mainly due to the relatively long time it takes for trans-fats to turn
rancid. The three interventions – bans on smoking, mortgages set at
more than 50 per cent of a person’s income, and trans-fats – are plotted

22:37:38
09
shoves and budges 133

Liberty

2 Externalities

Rational

figure 8.2 Conly’s purported shoves

as interventions 1, 2 and 3, respectively, in Figure 8.2, which illus-


trates that the second and third interventions are not shoves, strictly
speaking.
A policy intervention that relates to both health and financial
security, and which may be informed by present bias in the sense that
it addresses a general reluctance to save now to pay for possible needs
later, would be a compulsory pre-commitment contract such that
a person’s social care needs are paid for by public authorities on the
agreement that the money is recouped from a person’s assets – for
instance, from the sale price of their house – at the end of his life.
A similar intervention was proposed by The Dilnot Report (2011) on
the future of long term care in the United Kingdom, and in the form
expressed in this paragraph, would satisfy all of the requirements of
a shove.
Even if one focusses only on those interventions – such as a
ban on smoking – that appear to satisfy the coercive paternalism

22:37:38
09
134 the origins of behavioural public policy

requirements, objections to shoves do, perhaps expectedly, arise. For


instance, it may be contended that coercive paternalists are too con-
fident in their assessment of what people most value over their life
course. For example, in relation to cigarettes again, it seems feasible
that for at least some smokers the enjoyment of, say, forty years of
cigarette consumption outweighs the higher risk of suffering from
illness and the possible shortening of life when elderly. To inveigh
that smokers impose greater costs to health services than non-
smokers is to create an externality argument, and thus extends beyond
the remit of coercive paternalism, as we saw happening also with
nudge policy in Chapter 7. The libertarian paternalists allow for the
possibility that some people rationally choose to smoke precisely by
preserving autonomy of choice. Cass Sunstein (2013) himself notes
that our short-term goals might be a large part of what makes life
worth living, and thus, for many for much of the time, hyperbolic
discounting may not be a bias at all, echoing some of the points made
in Chapter 4 of this book. Determining the broadly held life goals that
are deemed worthy of protection – which Conly takes as health and
financial security – is not straightforward. If one considers all poten-
tial relevant trade-offs – e.g. more lifetime health versus greater enjoy-
ment in the moment, enhanced financial security versus taking the
risks that are sometimes necessary to live a rich and varied life – there
may not be objective life goals on which most people can agree.
Rather than leaving the benefit and cost considerations of any
particular behaviour to the individual and allowing those individuals
to continue their behaviours if they believe the benefits to outweigh
the costs – and, indeed, even if they believe the costs to outweigh the
benefits – coercive paternalists ban individual behaviours where they
believe the benefits of legislation and regulation to the individuals
that they target outweigh the perceived costs of those regulatory
measures. Some may contest, however, that since coercive paternal-
ists do not tend to quantify the relevant costs and benefits – in fact, it
is probably impossible to quantify all relevant costs and benefits –
there is inevitably a degree of arbitrariness in this approach. For

22:37:38
09
shoves and budges 135

example, Conly targets smoking but states that alcohol consumption


is acceptable because it is entrenched in many cultures, and it would
be difficult to get people to stop drinking it. Well, ditto smoking, it
could reasonably be said.
Sugden’s (2009) scepticism of soft paternalism mentioned earlier –
i.e., that we cannot hope to discern why people behave in the ways
that they do – applies ever more forcibly to hard paternalism. That
many people object to too much interference in their personal lifestyle
behaviours is evidence of this. However, Conly suggests that counter-
evidence that people often do not themselves know what they want
and thus can be helped to take a more beneficial path is provided by
the fact that they often tend to adapt to and accept coercive paternalis-
tic measures. For example, mandatory cleaning up of one’s dog mess,
the wearing of car seatbelts and motorcycle crash helmets, the wide-
spread ban on smoking in public places and even the more recent policy
of charging for plastic bags in supermarkets were all interventions
that met quite strong initial resistance, but that all now attract
broad public support in the countries in which they have been imple-
mented (although, admittedly, some of these interventions are princi-
pally externality-motivated). There are of course strong echoes of
self-determination theory at work here. Generally speaking, it is plau-
sible that these measures have progressed from being perceived as
forms of control to produce actions that are introjected in the sense
that they are associated with feelings of pride and guilt, to eventually
shape behaviours that are a part of most people’s identity. We might
expect that if the intervention is misguided, then broad identification
will not occur and the policy will eventually be retracted.
As a further justification for paternalistic actions, Conly con-
tends that forcing people to perform actions for their own sake is less
of an imposition than forcing them to act for other people’s sake. Yet,
if one accepts the judgments made in Figure 8.2, one sees that two of
the three principal policy recommendations that Conly makes are
driven by externality rather than internality concerns. As we have
seen, behavioural economics can inform various forms of paternalism,

22:37:38
09
136 the origins of behavioural public policy

but it seems clear that its potential usefulness extends further


than that.

regulating against harms


Many object to soft and, particularly, hard forms of paternalism on the
grounds that they interfere with personal lifestyle behaviours that
bear no consequences for others. Although libertarian paternalism,
by requiring that liberty ought to be preserved, dilutes this concern
somewhat, the fact that nudges are meant to result in unreflective
behaviour change leads some to worry that they are, in practice,
unavoidable and consequently coercive. As aforementioned, libertar-
ian paternalists contend that they are only nudging people in direc-
tions that those people state deliberatively that they want to go, but
Sugden’s (2009) concern with the validity of deliberative preferences
has been rehearsed, and, moreover, it is plausible that answers taken
as deliberative preferences are not really preferences at all, but are
rather responses to pressure, or to what individuals think their inter-
viewer wants to hear. If you are a smoker, for instance, you might
inform an interviewer that you would like to stop smoking, without
having any real desire or intention to abstain from smoking at all.
Thus, there are those who believe that the best approximation of what
a person really wants, taking into account all of the perceived costs
and benefits of the options placed in front of them, is given by what
they choose, particularly for everyday choices over which they have
a great deal of experience.
If policy makers are given a licence to manipulate behaviour
covertly towards ends that are justified on less than concrete grounds,
some worry that the stated aim of reducing negative internalities
takes on a somewhat murky hue. In such circumstances, the inter-
ventions may in fact be worsening internalities, which is an argument
in favour of overt rather than covert intervention when attempting to
change lifestyle behaviours and may lend more support to explicit
bans that can be scrutinised by those towards whom they are targeted.
Nonetheless, to reiterate, these still attract strong hostility from those

22:37:38
09
shoves and budges 137

who are against paternalism in any of these forms. However, the


usefulness of behavioural economics is not restricted to paternalistic
intervention; it can also be used as an input in deciding when and how
to regulate against behaviours that are harmful to others.
As was noted earlier, the marketing industry has garnered an
expertise over many decades in what is essentially behavioural science,
and this has long been used by commercial entities – the food, tobacco,
motor vehicle and lending sectors, among many others – to attempt to
persuade consumers to buy more of their products and services. It seems
reasonable to contend that these marketing efforts exploit behavioural
economics – present bias, loss aversion etc. – in ways that is not always
conducive to substantive improvements in population wellbeing, and
may sometimes impose harms on consumers. For example, knowing
that immediacy and salience are powerful forces on human – and not just
human – behaviours, confectionary companies traditionally paid super-
markets substantial amounts of money to display their products at
checkout counters, where people queue. One can imagine that purchas-
ing too many sweets is harmful to health, and thus an awareness of this
use of behavioural economic phenomena to influence purchasing pat-
terns offers policy makers a legitimate justification for regulating against
these practices. That is to say, a knowledge of behavioural economics
might inform policy makers that regulation to attempt to reduce nega-
tive externalities is appropriate in this case. Here, the design of the
regulation is not informed by behavioural economics in the way that
the design of nudges are meant to be informed as such, but, similar to
shoves, a knowledge of the behavioural economic findings is required, in
this case in order to detect when entities are using these findings for their
own harm-imposing ends. In other instances, the design of the regulation
itself can also be informed by behavioural economics so as to attempt
to strengthen its effect. In still other cases, behaviourally informed
regulations may be imposed on people so that they undertake actions
whenever their inertia would otherwise cause harms. Regulatory inter-
ventions that are informed by behavioural economics and are intended
to reduce harms are known – by me, at least – as budges.

22:37:38
09
138 the origins of behavioural public policy

Liberty

Classic budge
Regulatory

Behavioural
Externalities Internalities

Rational

figure 8.3 The budge space

In common with nudges and shoves, budges can be placed


within a three-dimensional diagram. Figure 8.3 depicts the budge
space, which is identical to the shove space drawn in Figure 8.1 except
that the internalities to externalities axis has now been inverted, such
that moving towards the origin on the horizontal axis indicates that
a policy is now increasingly addressing negative externalities rather
than negative internalities. Classic budges, such as requiring that
cigarettes be sold in plain packets, or banning cereal manufacturers
from advertising that high chocolate content breakfast products for
children bring everything that they want in the moment without any
potentially serious longer term consequences – on dental health, for
example – would thus be placed close to the origin.
An example of a budge that that does not just counter a harmful
exploitation of, say, present bias by commercial entities but, accord-
ing to the findings of behavioural economics, is also designed in such

22:37:38
09
shoves and budges 139

a way that one might reasonably expect it to be particularly effective is


the mandatory labelling of food products with traffic light symbols;
namely, red, amber and green signs to indicate high, intermediate and
low energy dense foods, respectively. That is, we might hypothesise
that people will select amber signs as their reference point, and thus be
motivated to avoid those foods with red symbols, judging them to be
losses in some sense. Indeed, traffic light labelling of this sort has been
shown to work (House of Lords, 2011), provided that the labelling is
not too complex. Mandatory traffic light labelling on food products
would appear to lie very close to the origin in Figure 8.3.
Figure 8.3 can also be used to highlight the arguments pre-
sented earlier in relation to some purported nudges and shoves not
being all that they at first seem, or perhaps being more than they at
first seem. For example, the system of prompted choice for organ
donation, which is a behaviourally informed regulation aimed prin-
cipally at helping others than those targeted and which demonstrates
that those targeted in harm reducing policies are not always com-
mercial entities, was discussed in Chapter 7. In such a system, people
no longer have the choice to do nothing at all if they do not wish to
donate their organs, assuming that they want a driving licence.
The United Kingdom government’s energy saving league table,
enforced on each government department, was also considered in
Chapter 7, and a ban on mortgages set at more than 50 per cent of
a person’s income in order to counter the exploitation of present bias
by lenders was discussed earlier in this chapter. These policies are
plotted as interventions 1, 2 and 3, respectively, in Figure 8.4. They
all appear to be budges.
Classifying the three interventions in Figure 8.4 together in the
same conceptual framework relies on the assumption that a failure to
act in order to benefit others – that prompted choice for organ donation
is intended to address, for example – is a qualitatively equivalent harm
to those where people have potential or realised negative conse-
quences imposed upon them due to the self-interested behaviours of
others – e.g. by being duped into buying an attractive sounding

22:37:38
09
140 the origins of behavioural public policy

Liberty

2
3
Internalities

Rational

figure 8.4 Accidental budges

mortgage that one will have great difficulty repaying. The qualitative
distinction between passive and active harms has a long history. For
instance, it resonates with Philippa Foot’s (1967) Trolley Problem,
where, in a series of hypothetical choices people tend to refuse to
push one person onto a trolley/tram line in order to save five others
who are tied to the track, and yet are more willing to pull a lever to
divert a tram that will ultimately produce the same consequences but
where their own involvement is less hands on. That is to say, the
greater the active involvement in causing a direct harm, the greater
the revulsion that people appear to show towards that harm. Budge
policy sidesteps the philosophical and psychological distinction
between what have been termed here passive and active harms, but
the distinction may nonetheless cause a greater degree of general
ambivalence towards some budges than others, and indeed
a reluctance towards forcing people to benefit others, rather than
reducing more direct harms to others, may be what Conly (2013) was

22:37:38
09
shoves and budges 141

alluding to when she wrote that targeting internalities would be easier


than targeting externalities.
Like nudges and, potentially, shoves, budges can extend to all
sectors of public policy. As a further example, short term lenders,
sometimes known as pay day loan companies, have traditionally mar-
keted their services, often towards relatively poor people who struggle
to get credit elsewhere, at invariably barely revealed, very high and
difficult to interpret interest rates. The companies appeal implicitly
to present bias, and in the process, harm many people. Regulations
imposed upon this sector on how they market their services and on
maximum allowable rates of interest fall under the remit of budge
policy. Yet regulations do of course impose their own costs, not least
the costs of monitoring the extent to which compliance is observed,
and, for those targeted, the costs of complying with what many might
see as more unnecessary red tape, which may in itself harm economic
performance. The advocates of budge policy would, however, retort
that their preferred direction does not call for the imposition of yet
more unnecessary and potentially damaging regulation. Rather, the
regulations introduced under the budge rubric would have a strong
intellectual foundation in being informed by the findings of behavioural
economics, such that there would be firm grounds to believe that their
implementation reduces the negative externalities that are consequent
on a range of human actions and non-actions. The arch-nudger, Cass
Sunstein (2013), has defined something similar as smart regulation.
In this chapter and Chapter 7 we have considered inlibertarian
paternalism, coercive paternalism and behavioural-informed regula-
tion against harms, the three principal behavioural economic-
informed policy frameworks. There is a way in which applications of
these frameworks can be considered in a single diagram.

the behavioural public policy cube


With a little rejigging, the nudge, shove and budge spaces drawn in
Figures 7.1, 8.1 and 8.3 can form a behavioural public policy cube, as
illustrated in Figure 8.5. In the cube, as with the nudge space,

22:37:38
09
142 the origins of behavioural public policy

b
c

Regulatory Classic shove ∗


Classic budge

g
f

Classic nudge
Liberty
a
Behavioural d

Internalities # Externalities

Rational
e h

figure 8.5 The behavioural cube

movement towards a on the ab axis indicates that a policy is increas-


ingly liberty-preserving rather than regulatory; movement towards
a on the ae axis indicates that a policy is increasingly informed by
behavioural economics rather than standard economic theory; and
movement towards a on the ad axis indicates that a policy is increas-
ingly addressing internalities rather than externalities. Consequently,
policies that lie on the adhe cube face (i.e., the bottom of the cube) are
entirely liberty-preserving, those on the abcd face (i.e., the back of the
cube) are heavily informed by behavioural economics and those on the
baef face (i.e., the left-hand side of the cube) entirely address internal-
ities. Classic nudges, shoves and budges lie at points a, b and c,
respectively.
Populating the cube with the behavioural public policies that
have been introduced in any particular country will probably further
demonstrate that pure nudges, shoves and budges are relatively rare,

22:37:38
09
shoves and budges 143

and that most policy interventions will be something in between. For


example, the policy of banning people from smoking in public houses
is indicated by the * in the cube, on the assumption that smoking is
substantively motivated by present bias and that the ban was intended
significantly to reduce exposure to second-hand smoke among bar
staff. Or to take possibly the most famous example in Thaler and
Sunstein’s (2008) book, a recommendation that airports voluntarily
etch pictures of flies on restroom urinals in the hope that this will
provide a salient target for men to aim at, given that most of the
benefit of less spillage would presumably fall to other people, would
place at something close to the # symbol. Populating the cube may
prove informative in the sense of giving an indication of how author-
itarian a government is willing to be, the extent to which it is con-
cerned with addressing external harms rather than internalities and
indeed whether it is using behavioural economics as an input into
policy design at all.
Table 8.1 summarises the main features of nudge, shove and
budge policy, which despite – or rather because of – their differences,
can be considered as complementary rather than mutually exclusive
approaches to behavioural public policy.
It is perhaps also worth noting that although these are prob-
ably the principal behavioural economic-informed policy frame-
works, others have developed ideas that are, at the very least,
worth mentioning here. Prominent among these other works is the
call by the political scientist Peter John and his colleagues to urge us
to think (John et al., 2011). John, although a supporter of the nudge
approach, acknowledges that there is a risk that too much reliance
on nudging might crowd out if not people’s ability to learn, then
a recognition that the act of learning is an important part of what
makes us human. Consequently, aside of nudges, he encourages the
use of interventions that are designed to engage people in delibera-
tive debate about appropriate actions and outcomes; he calls these
think policies, which are presumably fairly non-contentious, if
a little consuming of time and effort. Taking these developments

22:37:38
09
144 the origins of behavioural public policy

Table 8.1 Characteristics of behavioural public policy frameworks

Nudge Shove Budge

Principal Liberty- Regulatory, Regulatory,


requirements preserving, behaviourally behaviourally
behaviourally informed & informed &
informed & target target
target internalities. externalities.
internalities.
Meant to Sometimes Yes (in that those Sometimes
counterbalance targeted are
the use of prone to
behavioural present
economics bias etc.)
Design of the Yes No Sometimes
intervention is
informed
directly by
behavioural
economics
Principal Manipulative, Overly intrusive Economically
criticisms insufficiently damaging
effective

a step further, it is not unrealistic to imagine that one could even be


nudged to think.
There has also been quite extensive consideration of beha-
vioural motivations in the literature on public sector institutional
structures, typically taking the form of discussing the relative merits
and flaws of markets, hierarchies and networks. Competitive mar-
kets, according to many, are based upon the type of selfish egoism
that underpins rational choice theory, and hierarchies, if they are
working well, are assumed to be driven by a form of paternalistic
altruism where those who are charged with issuing orders are, in an
ideal world, serving the best interests of everyone. In networks, the

22:37:38
09
shoves and budges 145

relevant actors identify complementary interests and relationships


are built based on trust, loyalty and reciprocity. Networks are based
on horizontal patterns of interactions as opposed to power asymme-
tries (Héritier and Lehmkuhl, 2008). Conflicts are resolved within
networks on the basis of members’ reputational concerns (Lowndes
and Skelcher, 1998), and thus the network mode of governance
appears to be the most preferable if one accepts that reciprocity is
fundamental to human nature. Some have extended the governance
debate a little further. Bell et al. (2010), for instance, argue that
a persuasion mode of governance can be added to the mix.
By persuading people to change their behaviours, Bell et al. maintain
that government bodies influence but do not enforce, and that states
can enhance their power by building relationships with non-state
actors.
Many of these theories of institutional design were developed to
challenge the postulate of selfish egoism embedded in rational choice
theory, but on the whole, the governance literature in this area does
not appear to offer many lessons on how the various contentions can
be used to inform specific policy design. As noted, the literature on
network governance structures alludes quite heavily to reciprocity,
and yet one could say that it is quite weak on emphasising how this
phenomenon is intrinsic to human nature and is vague on specific
policy prescription. The publications that attempt to consider how
human motivations ought to inform policy design tend to focus on the
dichotomy between selfish egoism and pure altruism, but as Robert
Pinker (2006, p.19) has pointed out, ‘a model of human motivation
based on a sharply drawn distinction between the qualities of egoism
and altruism [bears] little or no relationship to what we know about
human nature and the realities of the world in which we live’. It may
be that most of us are not always seeking to take, and perhaps even
fewer of us are driven to forever unconditionally give. Perhaps a little
more realism would not go amiss.

22:37:38
09
9 Give and Take

As alluded to in Chapter 8, the governance literature on how to


motivate performance improvements in public policy has been insub-
stantial on the topic of reciprocity. The debate there has instead
tended to focus on whether public sector workers are pure altruists
or egoistical utility maximisers – or a bit of both. The economist Julian
Le Grand (1997; 2003), for instance, has argued that the post-war
consensus was that those who work within welfare states are public
spirited altruists. Consequently, public sector services, at least in
some countries, tended to be collectivist and without competitive
incentives. He insists that this consensus ought no longer to hold,
and should be replaced by the assumption that public sector profes-
sionals are also often motivated by their own avaricious tendencies. Le
Grand, inspired by Hume’s (1975 [1742]) view that every man ought to
be supposed a self-interested knave, posited that men – and women –
must be governed in the knowledge that they are often knavish so as to
steer their ambitions towards the public good. Le Grand’s position in
this respect bears some resemblance to that held by the libertarian
paternalists, in that both approaches wish to reshape what are essen-
tially forms of the choice architecture, but for Le Grand this is
done to exploit people’s underlying economic rationality, whereas
the libertarian paternalists focus upon taking advantage of their
irrationality.
Le Grand does not dismiss altruistic behaviour – which he calls
knightly – entirely; he recognises that knavish and knightly beha-
viours coexist, but he believes that the tendency to act altruistically
has been exaggerated. He maintains that since knavish behaviour is
common then competitive forces ought to be used within the public
sector. Unless public sectors workers are incentivised as such, Le

146

22:37:16
10
give and take 147

Grand posits, they will become lazy. If they have to compete for
purchasers or contracts, they will, out of self-interest, be concerned
with their ability to attract resources and will consequently be moti-
vated to provide a better quality service. Some worry, however, that
those motivated by greed will attempt to exploit their situation for
their own benefit whenever possible, and there is plausibly much
scope for suppliers of often quite complex services to do so in
a competitive market. Nonetheless, those aligned with the Le Grand
point of view have seen their views embraced within public policy in
many countries over the last quarter of a century, where a belief that
choice and competition can drive service improvements has been
ascendant. It can reasonably be contended, however, that the ego-
istical and altruistic views of human nature are a little caricatured,
and that most people, much of the time, lie somewhere between
these two ends of the motivational continuum. That is, most people
are not always likely to merely take, nor unconditionally give, but
rather are more inclined to give if others give; that is, people tend to
reciprocate.
Deliberations on the importance of reciprocity as an ethical
imperative – a normative rule – are as old as the history of recorded
thought, and are central to many of the world’s major religions. Mauss
(1954), for instance, noted that the Latin do ut des and the Sanskrit
dadami se, dehi me, which both translate as ‘I give in order that you
may give’, are found in Western and Eastern religious texts, and it is
not hard to find statements that signify the importance of reciprocity
in the Old and New Testaments, the Torah and the Quran. In the
quasi-religious Analects of Confucius we are told that what we do not
wish for ourselves we should not wish for others. The so-called golden
rule – treat others how you wish to be treated – is everywhere.
That people do tend to often reciprocate demonstrates that
reciprocity – unlike, say, procedural invariance discussed in
Chapter 2 – is not restricted to the normative. Reciprocal actions are
often what is known as attitudinal; that is, simple acts with immedi-
ate rewards that are common in the animal kingdom, such as cats

22:37:16
10
148 the origins of behavioural public policy

licking each other and humans holding doors open for one another.
Attitudinal reciprocity is perhaps akin to tit for tat – i.e., if you help
me, I will help you, and if you stop helping me, I will stop helping
you – which was modelled as the optimum strategy of human coopera-
tion by the political scientist Robert Axelrod (1984). Others have since
argued that tit for tat is a poor strategy when the information about the
payoffs or behaviour of one’s partner is less than perfect, or when the
cooperating group is large (Henrich and Henrich, 2007). Nonetheless,
attitudinal reciprocity appears to have arisen earlier in the evolution-
ary chain than a more deliberative form of reciprocity, and may have
served as the kernel for cooperation.
Some animals, particularly humans, engage in a sort of memory-
based reciprocity, where kind acts are repaid at some future date,
in which trust, gratitude and felt obligations all play a role. This
memory-based cooperative behaviour is often referred to as reciprocal
altruism, and involves a willingness to incur a cost in the expectation
that it will be repaid in kind. On the flipside, an unkind response to an
unkind act is a case of negative reciprocity. Gintis et al. (2005) define
as a strong reciprocator any individual who is both a reciprocal altruist
and an altruistic punisher – i.e. a person willing to punish others at
a personal cost to themselves. Gintis et al. contend that a high level of
cooperation in groups can be attained when there is a sufficient pro-
portion of strong reciprocators in the population.
It is plausible that the importance of reciprocity to religious
doctrine arose from the notion being fundamental to human nature.
That is, the normative may have been driven by the descriptive,
a proposition supported by none other than Charles Darwin (2004
[1879]). Religions may therefore have been designed, in part, to rein-
force the importance of a behavioural motivation that most humans
intrinsically accept and which strengthens the group collective.
The plausibility of this proposition would be strengthened if there
were evidence that a memory-based form of reciprocity is present in
young infants, or even in other species, where an influence of norma-
tive proclamations would presumably be absent.

22:37:16
10
give and take 149

other animals
In a relatively rare case of apparent reciprocal altruism in non-
primates, the animal behaviouralist Gerald Wilkinson (1984) reported
a study on food sharing among vampire bats. He observed that when
a bat has been successful in obtaining food during a hunt, which does
not happen every night, it will regurgitate blood into the mouth of an
unsuccessful non-related peer, and there will be reciprocation
between the same bats on a future night when their success levels
are reversed.
Most of the research on reciprocal altruism among animals,
however, has focussed on primates, and here the evidence appears to
be mixed, depending partly on the species studied and partly on the
interpretation given to an action by the researcher. Silk (2005) notes
that mutual grooming is common among primates, but the extent to
which it is more than attitudinal reciprocity is not known defini-
tively. Nonetheless, Silk maintains that it may be done to afford
protection, to build coalitions, to receive food and get access to new-
born infants over which many female primates take a strong interest.
The primatologist Frans de Waal (2010) is convinced that apes and
monkeys display behaviour that suggests something stronger than
attitudinal reciprocity. In one experiment, he separated two capuchin
monkeys – new world monkeys that seem to be particularly promising
when it comes to observing reciprocity in animals – either side of
a wire mesh divide. When both monkeys were required to pull on
a counterweighted tray in order for only one of them to reach a cup
of apple slices, the monkey in receipt of the apple pushed more of it
through the wire mesh to the assistant capuchin than when it secured
the apple entirely from its own efforts. If the assistant monkey was not
rewarded as such, it was less likely to help out when the task was
repeated, indicating an attitude that is memory-based.
However, opinion is split on whether non-human primates
demonstrate evidence of strong reciprocity. The developmental psy-
chologist Michael Tomasello, for instance, whose primate research

22:37:16
10
150 the origins of behavioural public policy

admittedly focusses on apes rather than new world monkeys, is less


convinced than de Waal that primate behaviour is demonstrably reci-
procal and takes the view that they are invariably selfish (Tomasello,
2009) – and thus economically rational. Indeed, Tomasello is sceptical
that chimpanzees behave cooperatively at all, arguing that what seem
to be organised hunts by chimps are really just chimps acting indivi-
dually according to what seems to be the best action in the moment
(Tomasello et al., 2005). De Waal (2010) agrees that wild chimps are
often extremely self-regarding, but notes that a male’s chance of
receiving a share of captured prey depends on his role in the hunt
rather than his dominance within the group, which incentivises full
cooperation in hunting expeditions.
Although Tomasello does not believe that our tendency to
behave reciprocally, beyond the mere attitudinal sense, extends
back to the time when we were non-human apes, he does think
that it extends back to our pre-history. He maintains that humans
developed shared intentions first among two or three people when
foraging, then eventually scaled up these actions in hunter-gatherer
societies where the promise of capturing large game on any
single day was far from secure – cf. the vampire bats – and in
response to threats from others (Tomasello, 2009). Victory went to
the most cohesive groups. Tomasello seems to imply that these
early developments resulted in humans possessing the fundamental
mental apparatus that enables reciprocity to thrive. Moreover, he
contends that from a very young age, children start to care about
their reputation. That is, they want to give the impression that they
are a good member of the group to show that they are worthy of
cooperation. A concern for reputation and acts of reciprocity go
hand in hand in that information about reputation is probably
a key factor in initiating and sustaining cooperation between non-
kin, and thus we have an explanation for why the peer group and
herd effects mentioned in Chapter 7 are such a fundamental part of
human psychology. Unsurprisingly, perhaps, reciprocity as a social
norm has also received considerable attention from anthropologists.

22:37:16
10
give and take 151

anthronomics
Marcel Mauss (1954) explored in detail gift giving in primitive cultures
and concluded that although such behaviours often appear voluntary,
they are not acts of pure altruism. Rather, they are given and repaid
under obligation and therefore foster reciprocity. In some groups,
there is an urge to return a gift of at least equal value at a later date
so as to not allow the other giver a feeling of superiority or dominance,
and thus gift giving in primitive societies was clearly underpinned by
something more than attitudinal reciprocity. Mauss further noted
that the Greeks and Romans were the first civilisations to draw the
distinction between the earlier ritual nature of gifts and exchange
driven by the law and economic interest. They thus developed beyond
the broadly held perception of antiquated gift societies, which were
encumbered by personal considerations that may have been incompa-
tible with the development of the market, trade and productivity.
Mauss argued that as societies became larger and more atomised,
and as people moved from clans to communities, they became more
egoistic, echoing the assumptions that underlie standard economic
theory.
Mauss’ view on greater atomisation is plausible, but, as we have
already seen, many contend that the human tendency to reciprocate,
both positively and negatively, remains central to people’s behaviour,
implying that it is difficult for us to escape entirely from this funda-
mental aspect of human behaviour – even if we should want to –
irrespective of the institutional structure of society. Indeed, some
would argue that reciprocity is necessary for the proper functioning
of contemporary social and economic systems. The great intuitive
behavioural economist, Adam Smith (2010 [1759], p.95), alluded
quite strongly to the importance of what are now known as positive
and negative reciprocity when he wrote that

Actions of a beneficent tendency, which proceed from proper


motives, seem alone to require reward; because such alone are the
approved objects of gratitude, or excite the sympathetic gratitude of

22:37:16
10
152 the origins of behavioural public policy

the spectator. Actions of a hurtful tendency, which proceed from


improper motives, seem alone to deserve punishment; because such
alone are the approved objects of resentment, or excite the
sympathetic resentment of the spectator.

It is admittedly true that Smith did not view that all relations – and
perhaps most economic relations – were driven by love and affection.
He wrote, for instance, that ‘Society may subsist among different men,
as among different merchants, from a sense of its utility, without any
mutual love or affection; and though no man in it should owe any
obligation, or be bound in gratitude to any other, it may still be upheld
by a mercenary exchange of good offices, according to an agreed valua-
tion’ (Smith, 2009 [1759], p.104). However, although mutual love,
affection and gratitude can be part of what informs a reciprocal
exchange, they are not essential to such an exchange. Smith’s phrase –
an agreed valuation – is crucial, as it implies that both sides of the
exchange believe it to be fair, which hints at an element of give and
take being part of the transaction. This is presumably essential if both
parties want to continue trading with each other in the future.
Disagreements on the interpretation of Smith’s ideas in The Theory
of Moral Sentiments and, in particular, in his other great work,
The Wealth of Nations (Smith, 1999 [1776]) – where he made similar
points with respect to trade and exchange – are legion, and I cannot
hope to resolve them. The important point to note here is that he
considered reciprocity to be a basic human motivation.
It is also worth noting that Luigino Bruni and Robert Sugden
(2008; 2013) draw on one of Smith’s contemporaries, the economist
Antonio Genovesi (1765–67), to argue that a reciprocal orientation in
market transactions is compatible with market efficiency. According
to Bruni and Sugden (2008), Genovesi did not believe that there is
a difference between market relationships and those governed by civil
society. He maintained that markets are based on the human tendency
towards mutual assistance, that reciprocity is central to economic
exchange and that each party to a market exchange needs to

22:37:16
10
give and take 153

understand and respect what the other party wants. Sugden is not,
however, the only modern behavioural economist to have taken
reciprocity seriously.

the ultimatum game


Contemporary behavioural economists have contributed to the study
of reciprocity in two principal ways; i.e. through empirical observa-
tion of reciprocal behaviours in controlled settings and with theore-
tical amendments to the neoclassical economic model in response to
those empirical observations. The interest here is in the empirical
observations – the evidence on whether people do indeed reciprocate –
and, in relation to this, one of the most famous and robust findings in
behavioural economics has been uncovered with an experiment
known as the ultimatum game. In this experiment, participants are
paired together, with each pair comprising a donor and a recipient.
Donors are given an amount of money and are asked to allocate a share
of that amount to their paired recipient, who will generally remain
anonymous to the donor. If the recipient accepts the share, then both
donor and recipient go home with those respective allocations, but if
the recipient declines then both parties receive nothing. According to
standard economic theory, the donor, being a selfish egoist, should
offer next to nothing because he ought to want to retain as much of the
money as possible and, for the recipient, anything ought to be better
than nothing. Before reading the next sentence, the reader may briefly
consider what he or she would offer if placed in the ultimatum game.
It is not untypical for average offers to exceed 40 per cent, with offers of
less than 30 per cent frequently rejected by recipients (Gintis et al.,
2005; Kahneman et al., 1986). Donors tend to know that recipients
will often reject a derisory share: recipients will often rather pay
a penalty in order to punish the donor than be, from their perspective,
exploited.
In some primitive cultures it has been observed that offers of
more than 50 per cent of the donor’s holdings are rejected not infre-
quently, indicating perhaps, as Mauss argued, that accepting a large

22:37:16
10
154 the origins of behavioural public policy

gift in certain cultures renders a person subordinate, and that at least


some people resist this circumstance (Gintis et al., 2005) – although it
ought to be acknowledged that refusing a gift can be tantamount to an
act of war. This all implies that it might be wise for those who are
considering offering a gift to refrain from being too generous in these
contexts. Indeed, lower than typical offers have been observed in other
primitive groups, such as Machiguenga horticulturists, who in one
study offered on average 26 per cent of the total pot (Henrich, 2000).
That said, Henrich also cites many studies where responses to the
ultimatum game were similar across cultures. Some may contend that
recipients are less likely to reject small proportions when the stakes
are substantial, but research in developing country contexts has
demonstrated that strong reciprocity continues to be observed when
the initial money allocation is as high as three months’ income
(Cameron, 1999). It could further reasonably be contested that the
construct of the ultimatum game is one that would be rarely if ever
encountered in real life; donors are after all given, in a sense, free
money that they can allocate as they wish, and that both their actions
and the recipients’ responses are likely to differ if the donors’ initial
endowments were earned. Moreover, it is noteworthy that a substan-
tial proportion of respondents – generally about a quarter – do tend to
behave in an entirely self-regarding manner. Nonetheless, the ultima-
tum game supports the conjecture that humans are often motivated by
strong reciprocity. According to Gintis et al. (2005), this is further
supported by qualitative evidence showing that, when asked why they
offer a substantial proportion of their initial endowment, donors com-
monly say that they are afraid that respondents will consider low
offers unfair and reject them, and when recipients reject offers, they
frequently say that they want to punish unfair behaviour.
There is another type of experiment, called the dictator game,
that is a little simpler than the ultimatum game. In the dictator game,
the donor again allocates a share of his endowment to the recipient but
here the recipient has no opportunity to accept or reject the offer; both
parties simply leave with the allocation that is decided by the donor.

22:37:16
10
give and take 155

The threat of negative reciprocity is therefore absent, and yet even


here donors offer positive amounts that typically range from 20–60
per cent of the total (Forsythe et al., 1994). Although the share offered
is normally somewhat lower than that observed in the ultimatum
game, the fact that donor offerings remain substantial may suggest
that a hard-wired degree of reciprocal altruism and/or concern about
the threat of negative reciprocity plays an important role. One might
conclude that the dictator game donors are more concerned with
distributional fairness in final outcomes than reciprocity, yet Fong
et al. (2005) report an experiment that paired several dictator donors
with real-life welfare recipients, and found that significantly more
money was allocated to recipients who expressed strong work prefer-
ences than those who expressed weak work preferences, indicating
that the donors preferred giving to those who were, in a sense, more
willing to offer something back.
Behavioural economists have reported further findings that
lend support to the notion of reciprocity being a strong motivator of
human behaviour. For example, Ernst Fehr, who has probably con-
tributed more to the study of reciprocity than any other behavioural
economist, and his colleagues conducted an experiment where
respondents were asked to assume that they were employers or
employees who are involved in deciding on a contract (Fehr et al.,
1997). Employees committed themselves to an effort level and
employers committed themselves to a wage rate. The standard eco-
nomic assumption is that those placing themselves in the position of
employees would be entirely self-regarding, and will therefore
choose a zero-cost effort level in a hypothetical contract irrespective
of the wage offered to them. In the Fehr et al. experiment, neoclassi-
cal economic postulates infer that those assuming the employer
position would anticipate that employees would choose a zero-cost
effort level and would thus offer no more than the minimum wage.
However, Fehr et al. found that the higher the wage offered by
employers, the higher the effort level committed to by employees.
The authors attributed the results to employer recognition that

22:37:16
10
156 the origins of behavioural public policy

employees would be predisposed towards reciprocity, and thus made


quite generous wage offers.
Social psychologists, whose works inform and are informed
by behavioural economics, have also written extensively on reci-
procity. Jonathan Haidt (2012), for instance, maintains that most
evolutionary theorists still believe that anything that looks super-
ficially like group-related adaptation – i.e. that reciprocity and
cooperation within groups have evolved to help any particular
group outcompete other groups, which is known as group selection
theory – is really adaptation to help individuals outcompete neigh-
bours within the same group, mirroring Tomasello’s earlier stated
conjecture regarding chimpanzee hunts. According to Haidt, how-
ever, intense intergroup competition, with intragroup gratitude and
vengeance, will continually strengthen loyalty, sanctity and reci-
procity in successive generations.
Group selection theory implies a mix of motives, not only
with respect to different people being differently motivated, but
also with respect to the same person being differently motivated at
different moments in time. Many people are often reciprocators,
but not all of the time. As earlier noted, even in the ultimatum
game a substantial number of people tend to behave in an entirely
selfish manner, and negative reciprocity can deter other otherwise
selfish actors from pursuing their favoured actions. On the basis of
evidence from controlled experiments, Fehr and Fischbacher (2005)
estimate that 40–50 per cent of people are strong reciprocators,
implying that 50–60 per cent of people are not, although the per-
centage of strong reciprocators may well be higher if the question is
focused on who is a strong reciprocator some but not all of the
time. To varying degrees, it is likely that most people are selfish,
altruistic and – not or – reciprocal, but for the success of the group,
our cooperative and reciprocal tendencies appear key. It thus seems
fairly uncontroversial to want to try to incorporate our understand-
ing of this important human motivation into the design of public
policy.

22:37:16
10
give and take 157

informing public policy with reciprocity


As noted at the beginning of this chapter, over the past twenty-five
years or so a number of countries with governments that – rhetorically,
at least – span the political divide have placed much faith in the
standard economic expectation that greater choice and competition
will improve efficiency in public services. The evidence on whether
or not it has is highly contestable and contested, partly because it is an
issue that is associated with a great deal of ideological conviction on all
sides. That is to say, it is a topic that tends to generate more heat than
light, but given that reciprocity appears to be crucial for group success,
the implications of this policy direction on people’s tendency towards
reciprocal acts merits consideration. For instance, it could plausibly be
the case that demand-led competition, where suppliers – e.g. schools,
hospitals – compete for market share, introduces incentives for them to
use underhand methods that misrepresent their organisation in
a positive light. For competitive markets to work in public services,
we must hope that egoism does not crowd out any natural tendencies
towards reciprocity; some might question whether it is wise to risk
weakening those tendencies.
The literature on reciprocity appears to suggest that this moti-
vational tendency works most effectively when the group is not too
large, and therefore at a fundamental level of public service organisa-
tion and management it might be the case that reciprocity has the best
chance of being sustained if public sector groupings are not too large
and their memberships are not too fluid. It is harder for relatively
small groups to lack transparency and it is easier than in larger groups
to hold its members to account. There is even some controlled experi-
mental evidence that suggests that the impact of strong reciprocity
on cooperation is better manifested when groups are coherent and
permanent (Fehr and Gächter, 2000), and it is plausible that coherence
is inversely correlated with group size. This all points to the sugges-
tion that public sector services might be better managed if local group-
ings are given a considerable degree of autonomy with respect to

22:37:16
10
158 the origins of behavioural public policy

organisational and managerial decision making, rather than being


too beholden to national-level mandates. Moreover, relatively small
groups may be better able to develop innovative cooperative strategies
to enhance their efficiency, and if the collective of groups were orga-
nised appropriately such that cross-group learning is encouraged –
which would perhaps be the responsibility of national-level policy
makers – then this may be optimal for the entire policy sector.
As Sethi and Somanathan (2005, pp.242–243) have stated, members
‘of groups that exhibit efficient norms will enjoy higher material pay-
offs than members of groups that do not, and such norms may there-
fore spread through the population by the imitation of successful
practices found in neighbouring groups . . . [therefore] norms of reci-
procity are an important component of social capital.’
As well as offering broad guidance on the appropriate size of
a group for the purpose of effective governance, the literature on
reciprocity can perhaps also offer some insights to inform policy
makers on what might be the most appropriate more specific perfor-
mance improving interventions. There are those who believe that
sincere caring for others is eroded by the market mechanism, a view
challenged by others. Nonetheless, as intimated above, it is plausible
that the complex nature of public sector services might give those who
are selfishly motivated much scope to act in socially undesirable ways
without fear of punishment, and that these tendencies would be
encouraged by a mechanism – i.e., a competitive market – that could
erode empathy, trust and reciprocity with and towards others
because of the focus placed upon beating one’s rivals. However, non-
competitive trade, of which pay for performance discussed in
Chapter 6 is an example, is a different kettle of fish.
As we saw earlier in the book, pure pay for performance mechan-
isms might prove to be very expensive if they are to work effectively,
and they come with real possibilities of unintended consequences.
However, they do mirror a straightforward payment mechanism –
i.e., a simple reciprocal exchange whereby one party pays to receive
a good or service from another party without either party seeking to

22:37:16
10
give and take 159

receive payment or goods from elsewhere but with both parties


perhaps keen to trade with each other again in the future. In these
circumstances, the providers’ fear, provoked by a competitive market,
that a service rival will unfairly undermine them, and the temptation
for them to unfairly undermine a rival, is largely removed.
Nonetheless, particularly in relation to complex goods and services
where there are substantial information asymmetries – the term that
economists use when one party to an exchange knows more about the
product or service than the other party – between the provider and the
payer, there remains the risk of some payers trying to get more than
what they have paid for and for some providers to give less than what is
fair. The latter possibility may be exacerbated if pay for performance
methods are imposed on public sector providers, in part because the
methods may then be seen as controlling rather than supportive,
which might undermine morale and the professionals’ identity with
the sector in which they work, leading us back to the predictions of
self-determination theory (Deci and Ryan, 1985). If pay for perfor-
mance is to have a chance of working as intended, then it would appear
that all stakeholders have to be involved in determining fair prices and
indicators of quality that are broadly perceived as appropriate.
Creating these conditions for reciprocity may optimise the chance
that the policy will benefit those it is meant to serve.
Even if the above conditions are met, however, many remain
unconvinced of the potential for pay for performance mechanisms to
be beneficial overall, highlighting the possible problems that were
alluded to briefly in Chapter 6. For example, the psychologist and
legal scholar Dan Kahan (2005) believes that performance incentives
may undermine cooperative tendencies by introducing the expecta-
tion that one must get paid for everything that one does, and thus
might consequently erode any motivation to undertake voluntary
beneficial actions. Perhaps informed by the natural inclination
towards negative reciprocity, Kahan intimates that credible penalties
are likely to be more beneficial than performance-linked rewards,
because people who resent fraud, corruption or cheating might see

22:37:16
10
160 the origins of behavioural public policy

penalties against such activities as supportive of their own belief


system, which may strengthen their identity with the organisation
in which they operate and detract from the temptation to perform an
injustice.
Another aspect of performance management discussed in
Chapter 6 – reputational competition or naming and shaming – is
perhaps even more strongly linked to the notion of reciprocity than
pay for performance. If we somewhat conveniently put to one side
the potentially damaging problem that having one’s performance
publicly reported may demotivate professionals, a concern for one’s
reputation is linked intrinsically to the notion of reciprocity if, as
aforementioned with respect to the work of Tomasello, one wants
to be seen as a good person or organisation with whom to coop-
erate. Bevan and Fasolo (2013, p.56) intimate at the power of
potential negative reciprocity by contending that reputation
‘could work out of fear of having betrayed the public’s trust and
provides an urgent reason for acting before the public reacts and
“punishes” this betrayal . . . [therefore] shocks of this kind are an
integral part of generating the high powered incentives necessary
for improvement’. The literature on reciprocity suggests that
a cautious implementation of something akin to the reputation
model of governance is worthy of consideration.
Reciprocity, and perhaps particularly its link to reputational
concerns, might therefore be used to inform the most appropriate
system-wide incentive mechanism if cooperation is deemed desirable,
but the tendency towards reciprocating behaviours may also be of use
when considering the design of much more targeted policy interven-
tions. The Behavioural Insights Team, discussed at some length in
Chapter 7, has recognised this potential by experimenting with the
differential framing of messages according to a number of human
motivations, including reactions to losses, social norm effects and
reciprocity, to encourage people to register as organ donors
(Behavioural Insights Team, 2013). That which emphasised recipro-
city, by suggesting to potential donors that increasing the number of

22:37:16
10
give and take 161

those on the register would potentially benefit everyone – including


those who registered – turned out to be the most effective message.
Although human beings are motivated by a multitude of factors,
including the assumed egoistical self-interest and pure altruism that
tend to dominate the literature on how people can be motivated to
provide better public sector services, it does seem more than plausible
that the tendency towards reciprocity underlies much of the coopera-
tion that occurs within groups. If this is indeed the case, then this
motivational driver ought to receive considerable attention in the
behavioural public policy literature. I have touched upon a couple of
ways in which reciprocity might usefully inform policy. A more
detailed consideration will be given at a different time in another
place.

22:37:16
10
10 Summing Up

I have tried, throughout this book up until this point, to present


a balanced assessment of the various topics that I have covered, with-
out attempting in any overt way to convince the reader to take
a particular stance on any of the points considered. In this final
chapter, although my intention is for readers to challenge my views
whenever they feel the need rather than to agree with them unquestio-
ningly, I am going to present a series of my own opinions appertaining
to the subject of behavioural public policy. I will not be providing an
exhaustive list of all of my views, otherwise I might never finish the
book; rather, I will relay a few thoughts on particular issues that
struck me as I wrote the preceding chapters, in the hope that these
might offer some further food for thought.
The first point that I would like to make is one that might
surprise readers that have made it this far, and that is to acknowledge
the contributions that rational choice theory and neoclassical eco-
nomics have made to the social sciences. They are tremendously
rigorous disciplines, so rigorous in fact that it takes many years of
effort and a non-negligible amount of intellectual ability in order to
gain more than a basic understanding of them. This rigour is probably
one reason why economists have been so influential over the social as
well as the economic policy making worlds. Moreover, economists
can measure things and make specific predictions; policy makers like
that. It is also worth considering the reasonably strong possibility that
without rational choice theory and standard economics, there might
be no rigorous, experimentally derived behavioural economics, and
behavioural public policy would be much the poorer for that.
However, over the twentieth century much of economics devel-
oped in a way that would have disappointed Alfred Marshall, one of

162

22:37:18
11
summing up 163

the founders of neoclassical theory, in that it became essentially


a branch of mathematics, divorced from the subject matter that it
was invented, in large part by Adam Smith, to grapple with – human
behaviour. Economics became, and still is, a club to which the non-
economist was not invited. Few people can really shake the ground on
which economists stand, because, as noted, the intellectual founda-
tions of their discipline are difficult to penetrate. Of course, this is not
to say that mainstream economists will always get it wrong. Indeed,
they often get it right; taxation, a tool of standard economic theory, is
one of the most effective means by which to effect behaviour change.
But the assumptions on which rational choice theory and mainstream
economics are built will mean that economists will sometimes get it
wrong, and badly wrong, when a richer understanding of human beha-
viour would help them to more often get it right. Man was not created
in the rational choice theorist’s image.
Smith, whose work of course predates neoclassical economic
theory and axiomatic rational choice theory, was probably the greatest
intuitive behavioural economist, and thus, in addition to being widely
recognised as a – perhaps the – founding father of economics as a field
of study, he should really be recognised, particularly due to his work in
The Theory of Moral Sentiments (2010 [1759]), as the founding father
of behavioural economics too. The principal contribution of modern
behavioural economists and psychologists engaged in this field from
the 1950s until recent years was to go beyond intuition by observing in
controlled settings the systematic tendencies that lead people to vio-
late the assumptions of the mainstream neoclassical model, in some
circumstances to estimate numerical parameters of specific beha-
vioural economic phenomena, such as probability weighting and
loss aversion, and to build alternative theoretical models based upon
these observations.

the burgeoning interest


Adam Smith also had a substantial influence on government policy.
His biographer John Rae (1895) recounts an incident where Smith

22:37:18
11
164 the origins of behavioural public policy

entered a room to greet, among others, William Pitt, the prime


minister, and William Wilberforce, the leader of the movement to
abolish the slave trade. All of Smith’s guests rose to their feet, but
on asking them to be seated, Pitt refused, stating that they would
sit down when Smith had sat down, for they were all his scholars.
Other great economists since Smith but before the development of
modern behavioural economists also acknowledged the importance
of human tendencies that are far from rational, according to the
conventional economic understanding of the term. Arguably the
most policy influential economist of the twentieth century, John
Maynard Keynes, for example, wrote that ‘the full consequences
[of our decisions] . . . can only be taken as the result of animal
spirits – a spontaneous urge to action rather than inaction, and
not as the outcome of a weighted average of quantitative benefits
multiplied by quantitative probabilities’ (1936, pp.161–162).
Moreover, for a long time now, government departments through-
out the world have implemented policies that relate to the findings
of behavioural economics, without always being acknowledged as
such, in a sort of haphazard way. For example, the administration
of President George W. Bush introduced the policy of automatic
enrolment in pension plans, and, as we have seen, the United
Kingdom’s policy of hospital waiting times targets in the early
2000s probably relied on reference points and loss aversion for its
effectiveness. Thus, behavioural public policy certainly was not
born with the creation of behavioural insights teams at governmen-
tal levels over quite recent years, but it seems fair to say that these
teams are the first to be officially charged with focussing solely on
the potential of the findings of behavioural economics, and beha-
vioural science more broadly, to inform and improve public policy.
In a sense, these teams have formalised a policy movement.
One may wonder – irrespective of the fact that ad-hoc applica-
tions of behavioural public policy have been around for a long time –
why interest in the policy possibilities of behavioural science has
exploded over the past decade. Part of the reason can be attributed to

22:37:18
11
summing up 165

an increased interest in behavioural economics following the 2008


financial crisis, when neo-liberal economics, with its assumption
that borrowers and lenders always act rationally according to their
best interests, came under intense scrutiny. Libertarian paternalism,
with its promise of inexpensive ways by which to tackle important
social problems, set inside an antiregulatory framework, aligned with
the political climate in an era of perceived austerity in several coun-
tries. Moreover, the two architects of libertarian paternalism were –
and are – very well connected in the policy making world, which
would not have hurt their cause, and they managed to reach an audi-
ence that transcended the academic community by producing
a popular science book at exactly the right time. The heavy interest
in the nudge aspect of behavioural public policy was perhaps in part
due to strategy and skill, but there was also, as is always the case with
such things, a heavy dose of luck involved. Possibly an even better
recent example of a mix of luck and skill causing the interest in some
academic ideas to cascade from time to time is Thomas Piketty’s
(2014) best-selling Capital in the Twenty-First Century, an essentially
neo-Marxist tome that I doubt many people have read in its entirety.
In part, Piketty also probably owes his success to the financial crisis,
but the fact that he was in the right place at the right time was
presumably not entirely accidental.
The principal disciplinary inputs into behavioural public policy
have thus far been economics and psychology, which is to be expected
given its behavioural economics foundation. As will now be clear to
the reader, I personally believe that a much broader range of disci-
plines can further usefully enrich this field of policy analysis, and, to
be fair, this does seem to be the direction of travel – anthropologists,
sociologists, political scientists, philosophers, animal behaviouralists
and others are making important contributions to the area, as I
have alluded to throughout this book. That said, interdisciplinary
work is not easy; indeed, there are often fierce internal fights within
disciplines, let alone across disciplines. For example, as we saw in
Chapter 9, some evolutionary theorists believe that groups have

22:37:18
11
166 the origins of behavioural public policy

adapted – i.e., reciprocity and cooperation have evolved – so as to


enable them to out-compete other groups, whereas others maintain
that any behaviour that looks like group adaptation at face value is
merely the result of individuals acting so as to best serve their own
self-interest. However, to my mind, such disagreements and debates
can enrich behavioural public policy, where evolutionary considera-
tions offer much food for thought.

embracing evolution
Not everyone agrees. Richard Thaler (2015), for instance, believes that
evolution was the probable cause behind the behavioural economic
phenomena – a belief supported by the fact that at least some of the
behavioural economic phenomena have been observed in other spe-
cies, and in primitive societies. Moreover, as mentioned in Chapter 4,
if people, and other animals, are preoccupied with surviving over the
short term rather than optimising over an extended period of time,
then findings such as present bias and loss aversion are not difficult to
comprehend. However, Thaler sees little merit in attaching impor-
tance to evolutionary explanations in economic and policy analyses.
He contends that we know that people are loss averse, for example,
and we do not need to know whether this has an evolutionary
explanation.
My opinion differs from Thaler’s in this regard, in that it strikes
me that analysing the possible evolutionary causes of the behavioural
economic findings, although inevitably to some degree speculative,
might lead us to conclude not only that people often fail to behave
according to the axioms of economic rationality, but that in certain
circumstances they ought not necessarily comply with these axioms
either. Moreover, a better understanding of the causes of the beha-
vioural economic phenomena might help us to better generalise their
effects across different policy interventions.
Perhaps Thaler’s ambivalence to a discourse on explanation is
grounded in his faith in the normative weight of expected utility
theory; that is, although he has done more than most to uncover the

22:37:18
11
summing up 167

behavioural economic phenomena, he does not believe that any expla-


nation for them will undermine the notion that people ought to want
to maximise utility, a view shared by most behavioural economists.
However, Paul Slovic and Amos Tversky (1974) reported that
a majority of their respondents who demonstrated Allais paradox
behaviours did not wish to alter their choices after being informed of
their at face value irrationality, and Sarah Lichtenstein observed simi-
larly with respect to preference reversals. If people often knowingly
and deliberately violate the assumptions underlying standard eco-
nomic theory – if they demonstrate deliberate decision making beha-
viours that contrast with those that are required by expected utility
maximisation – can we really conclude with confidence that utility
maximisation is an appropriate all-encompassing normative goal?
We may contend that we can conclude as such, by maintaining that
although the affects that cause the behavioural economic findings
evolved in environments that were highly uncertain, and allowed
our forebears to reach decisions quickly without expending too
much mental effort that, most of the time, served them well, in
modern, more secure societies, where people live for much longer on
average, they ought to plan to optimise their wellbeing over extended
time periods. Or, in short, the way people make decisions has not
evolved sufficiently to match the changing circumstances in which
they live. It is a plausible argument, but equally, perhaps something
important would be lost by assuming that the behavioural affects
necessarily undermine all that is normatively important. For example,
for many people a focus on the present moment may largely be what
makes life worth living. Perhaps some of us do not do that enough.
It is also worth noting that the rules of rational choice theory –
expected value maximisation and then expected utility maximisation –
were developed with reference to games, with money outcomes, that
one can repeat. For example, let us assume that an individual is asked
whether he will accept a game where he faces one hundred tosses of
a fair coin, and on each toss of the coin he will win $1,000 if it lands
heads and will lose $500 if it lands tails. The expected value of this

22:37:18
11
168 the origins of behavioural public policy

scenario is $25,000 and there is a very large chance that the individual
will end up a net winner. It seems reasonable to conclude that it would
be rational for him to accept the gamble. Imagine now that the indivi-
dual is told that he will face the coin toss not one hundred times, but
once. With these probabilities and money amounts, both expected
value maximisation and expected utility maximisation would suggest
that the individual should accept the gamble. Prospect theory, inciden-
tally, would predict that an average individual is likely to be roughly
indifferent between accepting and rejecting this gamble, because, as the
reader knows, losses are assumed to be weighted about twice as much
as equivalent sized gains, and with a 50 per cent chance of winning and
losing, objective and subjective probabilities would be very close to
each other. However, it seems reasonable, at least to me, if the indivi-
dual should want to demonstrate a greater degree of apparent risk
aversion in this one shot game, because he might feel repelled by the
substantial probability of losing, what may be for him, a significant
amount of money. Unlike in repeat games, there is no possibility of him
receiving anything close to the expected value of the game, which
stands here at $250; the individual will either win $1,000 or lose
$500, and a strong aversion to the possibility of losing $500 may be
driven not so much by choice but by necessity. In some real public
policy contexts – say, when considering investing in synthetic biology
to help tackle malaria or to provide a ready source of fuel – this conclu-
sion is amplified, because these interventions, although offering great
potential benefits, are associated with uncertain possibilities of irrever-
sible and catastrophic environmental harms. They are essentially one
shot games, and it may be impossible to claw back in the future any
realised harms now. In short, the strong aversion to losing in one shot
games – which is often attributed to loss aversion – may often be quite
rational in a broad sense of the term.

the policy approach


It is, however, distrust that people’s preferences will propel them
towards maximising their own wellbeing that has served as the

22:37:18
11
summing up 169

intellectual justification for libertarian paternalism, coercive patern-


alism and experienced utility maximisation. The experienced utility
movement, with its links to the new economics of happiness, has
received quite a lot of attention in the academic and policy discourse
over recent years, with several governments expressing strong interest
in the approach, not least because the populations of rich countries
have apparently become no happier over the past several decades
despite large increases in per capita national incomes. This interest,
at face value, seems sensible, because one could argue that a focus on,
say, indices of gross domestic product can mislead governments if
their populations are not feeling the benefit of any increases in aggre-
gate or average incomes. Yet there is much that is questionable about
using happiness as a tool for guiding public policy. Some of these
arguments were rehearsed in Chapter 5. For example, should policy
makers take note of people’s happiness in the moment – the experi-
enced utility approach – or of indicators of how satisfied they are with
how their lives are going in general, which are different propositions.
The former calls for the aggregation of individual moments of pleasure
and pain to give the total experienced utility of an event, episode or
action, without paying attention to how those moments fit together,
a neo-Benthamite approach.
As the reader would probably have guessed by now, the experi-
enced utility approach strikes me as too reductive to be an appropriate
guide for policy action. Bowing to the findings of the gestalt literature,
it seems to me that how the moments of an episode fit together
can have a profound effect on how people value that episode, and in
some – not all – circumstances, I cannot conclude that those prefer-
ences lack legitimacy. At a very broad level, for example, I can sym-
pathise with the view that a life that begins disappointingly but ends
in triumph – the imagined life of an academic – is superior to a life that
begins triumphantly but ends in disappointment – the real life of an
academic – even if the total experienced utility of the latter life is the
same, or even somewhat larger, than that of the former. By insisting
that only experienced utility matters, the advocates for this approach

22:37:18
11
170 the origins of behavioural public policy

are, in essence, saying that only their preferences matter in this


respect, which one could view as a little ironic given that it is a
distrust in preferences that was foundational to their approach.
The experienced utility approach is not objectively wrong, of course.
It is an idea that people will battle over, but with a little more reflec-
tion those who are immediately attracted to the approach might reach
a similar conclusion to that arrived at by John Stuart Mill. Mill (1873),
who was almost uniquely well qualified to comment on Benthamite
utility theory, after being an early adherent, concluded that it could
not hope to capture everything that is important in a human life.
My preferred approach would be for policy makers to work at
offering people opportunities to lead fulfilling lives. This is not the
same as Sugden’s (2009) preferred normative goal of maximising peo-
ple’s choice set, mentioned in Chapter 8, without attaching any
weight to outcomes, because I believe that policy makers should
also keep at least one eye on wellbeing. I am sympathetic to the
view that governments should work to negate negative externalities,
but, by and large, I am a sceptic when it comes to legitimising policy
makers to interfere in individual behaviours that bear no conse-
quences for others. I would prefer policy makers to seek to educate,
rather than indoctrinate, manipulate or legislate, in their attempts to
reduce negative internalities, although admittedly there will be many
instances when nudges – e.g. cheesecake removed from eye level in
canteens – and shoves – e.g. compulsory seatbelt laws – appear per-
fectly reasonable to me, the former perhaps when their implications
for personal lifestyle behaviours are not profound and the latter when
the behaviour that they focus upon are ones with which I, like most
people, identify. However, the overriding framework that I lean
towards – i.e., extending opportunities to lead fulfilling lives – is
close to Amartya Sen’s (1999) capability approach, where he advo-
cates that important aspects to life such as nutrition, shelter and
freedom – things that are necessary for people to flourish – ought to
be equitably distributed, and is consistent with using behavioural
economics to inform where and when it is appropriate to regulate

22:37:18
11
summing up 171

against harms. Other than when it comes to how and what I teach my
students, I am just not that paternalistic by nature. Happiness may of
course often be a by-product of providing opportunities for fulfil-
ment, but, for me, maximising happiness is not the primary goal of
public policy.

human motivation
Humans are social animals and spend much of the time comparing
themselves to others, a practice that also extends to comparing them-
selves to themselves, in relation to their own past and expected future
positions. Rational choice theory and standard economic theory deal
with absolutes and with final position: i.e. final assets are what count.
Yet, as we know from prospect theory detailed in Chapter 3, the
carriers of value often tend to be what we gain and lose relative to
what we already have, and the gestalt literature covered in Chapter 5
shows that people do not treat all of the moments of an event as
independent experiences, such that any particular ending, for exam-
ple, could be perceived quite differently depending on what preceded
it. We respond to reference points all of the time, from focusing in
upon a recommended daily calorie intake, to following our peers in
avoiding eye contact on underground railways. That we incessantly
compare ourselves to others is not, for most, a recipe for happiness and
is an argument for leaving social media platforms; indeed, comparing
one’s position to others is the probable explanation for why popula-
tion happiness indices have not improved in rich countries despite
large increases in income. Anchoring upon reference points, whether
these be money outcomes, natural or artificially imposed prompts or
what our peers are doing, can explain to a significant extent many of
the findings of behavioural science. We also care about fair process.
For example, on purchasing an item we experience joy and pain
from feeling that we have got a good deal or have been ripped off, a
phenomenon that Thaler (2015) labels transaction utility, which goes
beyond the standard economic assumption that what we end up with –
acquisition utility – is the only thing that matters.

22:37:18
11
172 the origins of behavioural public policy

Whether people ought to compare themselves to others and


themselves so much is a moot point, but they do and they probably
always will. If we conclude that only final outcomes matter, we are
ignoring – or paternalistically overriding – an important part of human
psychology. Again, there are plausible evolutionary explanations for
why we compare; for example, keeping an eye on your position within
the group, and on the trajectory of your own position, may act as
a guard against complacency and perhaps serves to increase both the
ability for you to impose your will on the group and the chance that
other group members will continue to cooperate with you. In your
primitive state, your survival may have depended on you maintaining
your position and cooperating with others. One lesson from this for
policy makers is that if they are concerned with general population
wellbeing and with providing the circumstances for people to believe
that they have opportunities to flourish, then it might be wise for
them to try to ensure that everyone benefits proportionately from
their interventions specifically and from their country’s development
more broadly. If large sectors of a population feel as though they are
being left behind even if their position is improving in absolute terms,
overall population wellbeing and social cohesion may suffer as a con-
sequence (cf. once again, population happiness indices). The former
prime minister of the United Kingdom, Tony Blair, once said that he
did not care how wealthy the richest people in society become so long
as the position of the poorest was improving, but admittedly, that was
not his biggest mistake.
Consideration of the social aspect of human motivation has
been somewhat lacking in the literature on how public sector profes-
sionals can be motivated to improve their performance. The discourse
there has tended to focus on the dichotomy between selfish egoism
and pure altruism. Encouraging selfish egoism will probably ulti-
mately damage the ethos on which the provision of good public ser-
vices is based, and relying on pure altruism is, I think, somewhat
naïve. These two aspects of motivation respectively suggest that we
take from, or give to, others. Much of the time, however, we take from

22:37:18
11
summing up 173

and give to others, by acting reciprocally. It is, to my mind, very


important that the human tendency to act reciprocally – which serves
to strengthen cooperation – is not crowded out by policy interven-
tions, such as the introduction of demand-led competition, that
assume that people are predominantly egoistical. Encouraging public
sector providers – who offer a range of complex services that are riven
with information asymmetries – to compete for users incentivises
them to act in ways that are at odds with a fair exchange.
Introducing reputational league table competition in the absence of
demand-side choice, however, is consistent with motivating people
and groups to want to show that they are worth dealing with, and is
a policy instrument that merits continued attention.
Although I do find it convincing that the tendency to act reci-
procally is a key part of human motivation that underpins coopera-
tion, it does appear to be quite easy to crowd this motivation out
simply by isolating people from each other, to everybody’s disadvan-
tage. If we take drivers, for instance, where each person is isolated and
protected within their own car with only their own goal – their own
particular destination – in mind, reciprocity is often lacking and self-
ish egoism snowballs. To sustain and, one might hope, increase reci-
procity, an element of sociability has to be maintained – forcing
friendly eye contact with other road users seems to help. Another
contemporary example relates to social media, which, perhaps para-
doxically, may have served to increase social isolation. Many people
seem to feel less inhibited to express thoughts, ideas and opinions to
and about others that go against the spirit of reciprocity and coopera-
tion when their physical social isolation remains intact. Indeed, social
media can damage cooperation even when people are not physically
isolated from each other, as I have observed in seminars, where I am
fairly sure that at least some students are checking Facebook when
they ought to be engaging in class discussion. Admittedly, those who
know me may think that I am personally not qualified to bemoan the
erosion of sociability since I probably fall in the bottom 10 per cent of
the population in that respect, but Marcel Mauss, mentioned in earlier

22:37:18
11
174 the origins of behavioural public policy

chapters, might well have been right when he wrote that the atomisa-
tion of economic systems produced greater egoism, and with that, a lot
that was good, despite the gains, was lost.

the future
The only thing I really know about the future is that, in the long-run,
I won’t be in it, but as to the future of behavioural public policy, I can
hazard some guesses. I do fear a little that behavioural public policy
will continue to be somewhat nebulous and ill defined, at least in the
policy and popular discourse, in part because there are people who
attempt to contribute with insufficient scholarly training, after expo-
sure to a few popular science books. The area is, after all, informed by
human behaviour, a topic on which a great many people may feel
qualified to comment. In one sense, this is good, because it brings in
more people and thus ideas, but a lack of intellectual rigour of course
has its drawbacks, in that vacuity can confuse and frustrate. For
example, I have heard taxation described as a nudge, when taxation,
as a regulatory measure informed by standard economic theory, often
as a means to address externalities, is actually the antithesis of
a nudge, as originally defined. This kind of sloppy thinking has the
potential to undermine to some extent the credibility of behavioural
public policy as a field of analysis, and therefore it ought to be guarded
against.
I also feel as though the area needs to move a little more beyond
what are often quite quirky, amusing but ultimately arguably rather
trivial proposed interventions. There are obvious advantages of quirky
examples – they attract interest, and taken together their effects may
be more than trivial. However, I fear that people will eventually lose
patience unless the policy examples become more substantive in and
of themselves. As I have tried to show in this book, the use of beha-
vioural economics in policy can be taken beyond nudging – not that
I am suggesting that nudging is necessarily trivial. As noted, beha-
vioural economics is the study of how people make decisions, and this
knowledge can help inform any policy approach, including where and

22:37:18
11
summing up 175

how to regulate. As alluded to in Chapter 8, these broader policy uses


of behavioural economics are already being applied in several coun-
tries, and a wider recognition of the real potential of behavioural
public policy is warranted – despite my above stated fears, I am hope-
ful that this will happen.
Although I lean towards a personal preference for using beha-
vioural economics to help inform regulation against harms, I would
not dismiss entirely the usefulness of libertarian and coercive patern-
alism, and I am respectful of those whose views differ from mine on
this matter. For me, though, given their potential advantages and
disadvantages, the nudge and shove approaches should not be accepted
en masse; rather, the appropriateness, feasibility, plausible unin-
tended consequences and, if possible, effectiveness of each nudge
and shove – and budge – intervention should, to my mind, be consid-
ered and balanced by the policy makers responsible, gauging public
approval where necessary, which is, I suspect, what generally happens
anyway. Ultimately, applied behavioural public policy is likely to
comprise of many interventions that are informed by a variety of
different frameworks.
Will behavioural public policy – principally thus far via nudge
theory and its applications – retain as high a profile in the academic,
policy and, particularly, popular presses as has been observed over the
last several years? I would guess not. Although policy applications of
behavioural theory have been implemented for many years, the spe-
cific focus on the policy promise of behavioural phenomena – almost
to the extent that this became a movement in some countries – was
something new, as were the buzzwords associated with it. That initial
broad interest will be difficult to sustain, but I feel, or hope, that at an
intellectual level and perhaps as a policy consideration, the field will
mature, and a far wider range of disciplines beyond economics and
psychology will influence and enrich it. In this book, I have attempted
to contribute in a small way to this enrichment by adding a little to the
stock of the intellectual rigour in this area. Most readers will probably
disagree with at least some of what I have had to say, but that is

22:37:18
11
176 the origins of behavioural public policy

something that I encourage and applaud. As my great London School


of Economics antecedent, Karl Popper (1963, p.30), once wrote, and as
I always try to remind myself, ‘it might do us good to remember from
time to time that, while differing widely in the various little bits we
know, in our infinite ignorance we are all equal’.

22:37:18
11
Note on References

The practice of citing one’s own work in order to substantiate a point always strikes
me as a little self-reverential and tautological. I have therefore avoided referencing
myself in the main text, save for one article that I published in Medical Decision
Making, which contains some experimental evidence that I found useful to cite,
and to acknowledge the original sources of some tables and figures that I have
reused. However, in order to avoid a charge of self-plagiarism, and in the interests
of full disclosure, the following writings are ones that I have leaned on in writing
this book.

Oliver, A. 2006. Further Evidence of Preference Reversals: Choice, Valuation and


Ranking Over Distributions of Life Expectancy. Journal of Health Economics
25: 803–820.
Oliver, A. 2006. Review of Layard, R. Happiness: Lessons from a New Science.
Economics and Philosophy 22: 299–307.
Oliver, A. 2008. Assessing the Influence of Gestalt-Type Characteristics on
Preferences over Lifetime Health Profiles. Medical Decision Making 28:
723–731.
Oliver, A., and L. Brown. 2011. Incentivising Professionals and Patients:
A Consideration in the Context of the United Kingdom and the United States.
Journal of Health Politics, Policy and Law 36: 59–87, particularly in relation to
Chapter 6 of this book.
Oliver, A. 2012. Markets and Targets in the English National Health Service:
Is There a Role for Behavioral Economics? Journal of Health Politics, Policy
and Law 37: 647–664.
Oliver, A. 2013. A Normative Perspective on Discounting Health Outcomes.
Journal of Health Services Research and Policy 18: 186–189, particularly in
relation to Chapter 4.
Oliver, A. 2013. From Nudging to Budging: Using Behavioural Economics to Inform
Public Sector Policy. Journal of Social Policy 42: 685–700.
Oliver, A. 2013. Introduction. In: Oliver, A. (ed.). Behavioural Public Policy.
Cambridge: Cambridge University Press.

177

Downloaded from https://www.cambridge.org/core. University of Florida, on 21 Dec 2017 at 16:30:09 Core terms
of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781108225120
178 note on references

Oliver, A. 2013. Ambiguity Aversion and the UK Government’s Response to Swine


Flu. In: Oliver, A. (ed.). Behavioural Public Policy. Cambridge: Cambridge
University Press.
Oliver, A. 2015. Nudging, Shoving and Budging: Behavioural Economic-Informed
Policy. Public Administration 93: 700–714, particularly in relation to
Chapters 7 and 8.
Oliver, A. 2016. Behavioural Economics Health Applications in the UK. In:
Scheffler, R (ed.). Global Health Economics and Policy. New Jersey: World
Scientific Publishing, particularly in relation to Chapters 7 and 8.
Oliver, A. Forthcoming. Distinguishing between Experienced Utility and
Remembered Utility. Public Health Ethics, particularly in relation to
Chapter 5.
Oliver, A. Forthcoming. Do Unto Others: On the Importance of Reciprocity in Public
Administration. American Review of Public Administration, particularly in
relation to Chapter 9.

Downloaded from https://www.cambridge.org/core. University of Florida, on 21 Dec 2017 at 16:30:09 Core terms
of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781108225120
Bibliography

Abdellaoui, M. 2000. Parameter-Free Elicitation of Utilities and Probability


Weighting Functions. Management Science 46: 1497–1512.
Akerlof, G. A., and R. E. Kranton. 2010. Identity Economics: How Our Identities
Shape Our Work, Wages, and Well-Being. Princeton: Princeton University
Press.
Allais, M. 1953. Le Comportement de L’Homme Rationnel Devant le Risque:
Critique des Postulats et Axiomes de L’Ecole Américaine. Econometrica 21:
503–546.
Allais, M. 1990. Allais Paradox. In: Eatwell, J., M. Milgate and P. Newman (eds.).
The New Palgrave: Utility and Probability. New York: Macmillan Press,
pp. 3–9.
Ariely, D. 2008. Predictably Irrational: The Hidden Forces That Shape Our
Decisions. London: HarperCollins.
Ariely, D., and Z. Carmon. 2000. Gestalt Characteristics of Experiences:
The Defining Features of Summarized Events. Journal of Behavioral Decision
Making 13: 191–201.
Ariely, D., and G. Loewenstein. 2000. When Does Duration Matter in Judgment and
Decision Making? Journal of Experimental Psychology 129: 508–523.
Ariely, D., G. Loewenstein and D. Prelec. 2003. ‘Coherent Arbitrariness’: Stable
Demand Curves without Stable Preferences. The Quarterly Journal of
Economics 118: 73–106.
Ariely, D., U. Gneezy, G. Loewenstein and A. Mazar. 2009. Large Stakes and Big
Mistakes. Review of Economic Studies 76: 451–469.
Aristotle. 1998 [350BC]. The Nicomachean Ethics. Oxford: Oxford University
Press.
Ashraf, N., C. F. Camerer and G. Loewenstein. 2005. Adam Smith, Behavioral
Economist. Journal of Economic Perspectives 19: 131–145.
Axelrod, R. 1984. The Evolution of Cooperation. New York: Basic Books.
Behavioural Insights Team. 2010. Applying Behavioural Insights to Health.
London: Cabinet Office.
Behavioural Insights Team. 2011. Behaviour Change and Energy Use. London:
Cabinet Office.

179

10:37:57
12
180 bibliography

Behavioural Insights Team. 2013. Applying Behavioural Insight to Organ


Donation: Preliminary Results from a Randomised Controlled Trial. London:
Cabinet Office.
Bell, S., A. Hindmoor and F. Mols. 2010. Persuasion as Governance: A State-Centric
Relational Perspective. Public Administration 88: 851–870.
Bentham, J. 1988 [1781]. The Principles of Morals and Legislation. New York:
Prometheus Books.
Bernasconi, M., and G. Loomes. 1992. Failures of the Reduction Principle in an
Ellsberg-Type Problem. Theory and Decision 32: 77–100.
Bernoulli, D. 1738. Specimen Theoriae Novae de Mensura Sortis. Commentarii
Academiae Scientiarum Imperialis Petropolitanae 5: 175–192. Translated by
Sommer, L. 1954. Exposition of a New Theory on the Measurement of Risk.
Econometrica 22: 23–36.
Besley, T., G. Bevan and K. Burchardi. 2009. Naming and Shaming: The Impacts of
Different Regimes on Hospital Waiting Times in England and Wales. London:
LSE Health Discussion Paper. London School of Economics and Political
Science.
Bevan, G., and B. Fasolo. 2013. Models of Governance of Public Services:
Empirical and Behavioural Analysis of ‘Econs’ and ‘Humans’. In: Oliver, A
(ed.). Behavioural Public Policy. Cambridge: Cambridge University Press,
pp. 38–62.
Bleichrodt, H., and J. L. Pinto. 2000. A Parameter-Free Elicitation of the Probability
Weighting Function in Medical Decision Analysis. Management Science 46:
1485–1496.
Broome, J. 1991. Weighing Goods. Oxford: Basil Blackwell.
Broome, J. 1994. Discounting the Future. Philosophy and Public Affairs 23:
128–156.
Bruni, L., and R. Sugden. 2007. The Road Not Taken: How Psychology Was
Removed from Economics, and How It Might Be Brought Back. The Economic
Journal 117: 146–173.
Bruni, L., and R. Sugden. 2008. Fraternity: Why the Market Need Not Be a Morally
Free Zone. Economics and Philosophy 24: 35–64.
Bruni, L., and R. Sugden. 2013. Reclaiming Virtue Ethics for Economics. Journal of
Economic Perspectives 27: 141–164.
Cabinet Office and Institute for Government. 2010. MINDSPACE: Influencing
Behaviour through Public Policy. London: Cabinet Office and Institute for
Government.
Cahill, K., and R. Perera. 2008. Competitions and Incentives for Smoking
Cessation. The Cochrane Library 2: 1–31.

10:37:57
12
bibliography 181

Camerer, C. F. 1989. An Experimental Test of Several Generalized Utility Theories.


Journal of Risk and Uncertainty 2: 61–104.
Camerer, C. 1995. Individual Decision Making. In: Kagel, J. H., and A. E. Roth (eds.).
The Handbook of Experimental Economics. Princeton: Princeton University
Press, pp. 587–703.
Camerer, C. F., and T. Ho. 1994. Violations of the Betweenness Axiom and
Non-Linearity in Probability. Journal of Risk and Uncertainty 8: 167–196.
Camerer, C. F., and G. Loewenstein. 2003. Behavioral Economics: Past, Present,
Future. In: Camerer, C. F., Loewenstein, G. and Rabin, M. (eds.). Advances in
Behavioral Economics. Princeton: Princeton University Press, pp. 3–51.
Camerer, C., S. Issacharoff, G. Loewenstein, T. O’Donoghue and M. Rabin. 2003.
Regulation for Conservatives: Behavioral Economics and the Case for
‘Asymmetric Paternalism’. University of Pennsylvania Law Review 1151:
1211–1254.
Cameron, L. 1999. Raising the Stakes in the Ultimatum Game: Experimental
Evidence from Indonesia. Economic Inquiry 37: 47–59.
Campbell, S. M., D. Reeves, E. Kontopantelis, B. Sibbald and M. Roland. 2009.
Effects of Pay for Performance on the Quality of Primary Care in England.
New England Journal of Medicine 361: 368–378.
Chew, S. H. 1983. A Generalisation of the Quasilinear Mean with Applications to
the Measurement of Income Inequality and Decision Theory Resolving the
Allais Paradox. Econometrica 51: 1065–1092.
Chew, S. H., and W. S. Waller. 1986. Empirical Tests of Weighted Utility Theory.
Journal of Mathematical Psychology 30: 55–72.
Conlisk, J. 1989. Three Variants on the Allais Example. American Economic
Review 79: 392–407.
Conly, S. 2013. Against Autonomy: Justifying Coercive Paternalism. Cambridge:
Cambridge University Press.
Curley, S. P., and F. J. Yates. 1989. An Empirical Evaluation of Descriptive Models
of Ambiguity Reactions in Choice Situations. Journal of Mathematical
Psychology 33: 397–427.
Darwin, C. 2004 [1879]. The Descent of Man. London: Penguin Classics.
Deci, E. L., and R. M. Ryan. 1985. Intrinsic Motivation and Self-Determination in
Human Behavior. New York: Plenum.
Deci, E. and R. Ryan. 2000. The ‘What’ and ‘Why’ of Goal Pursuits: Human Needs
and the Self-Determination of Behavior. Psychological Enquiry 11: 227–268.
Deci, E. L., R. Koestner and R. M. Ryan. 1999. A Meta-Analytical Review of
Experiments Examining the Effects of Extrinsic Rewards on Intrinsic
Motivation. Psychological Bulletin 125: 627–668.

10:37:57
12
182 bibliography

Dekel, E. 1986. An Axiomatic Characterization of Preferences under Uncertainty:


Weakening the Independence Axiom. Journal of Economic Theory 40: 304–318.
Department of Health and Human Services. 2008. Premier Hospital Quality
Incentive Demonstration: Rewarding Superior Quality Health Care.
Washington DC: HSS.
de Waal, F. 2010. The Age of Empathy: Nature’s Lessons for a Kinder Society.
London: Souvenir Press.
Diecidue, E., and P. P. Wakker. 2001. On the Intuition of Rank-Dependent Utility.
Journal of Risk and Uncertainty 23: 281–298.
Diener, E., D. Wirtz and S. Oishi. 2001. End Effects of Rated Life Quality: The James
Dean Effect. Psychological Science 12: 124–128.
Dilnot Report. 2011. Fairer Care Funding: The Report of the Commission on
Funding of Care and Support. London: Department of Health.
Disney, K., J. Le Grand and G. Atkinson. 2013. From Irresponsible Knaves to
Responsible Knights for Just 5p: Behavioural Public Policy and the
Environment. In: Oliver, A. (ed.). Behavioural Public Policy. Cambridge:
Cambridge University Press, pp. 69–87.
Dolan, P., and D. Kahneman. 2008. Interpretations of Utility and Their
Implications for the Valuation of Health. Economic Journal 118: 215–234.
Doran, T., C. Fullwood, H. Gravelle, D. Reeves, E. Kontopantelis, U. Hiroeh and
M. Roland. 2006. Pay-for-Performance Programs in Family Practices in the
United Kingdom. New England Journal of Medicine 355: 375–384.
Economist. 2012. Nudge Nudge, Think Think. 24 March 24.
Edwards, W. 1955. The Prediction of Decisions among Bets. Journal of
Experimental Psychology 50: 201–214.
Einhorn, H. J., and R. M. Hogarth. 1986. Decision Making under Ambiguity. Journal
of Business 59: S225–S250.
Ellsberg, D. 1961. Risk, Ambiguity and the Savage Axioms. Quarterly Journal of
Economics 75: 643–669.
Fehr, E., and S. Gächter. 2000. Cooperation and Punishment in Public Goods
Experiments. American Economic Review 90: 980–994.
Fehr, E., and U. Fischbacher. 2005. The Economics of Strong Reciprocity. In:
Gintis, H., S. Bowles, R. Boyd and E. Fehr (eds.). Moral Sentiments and
Material Interest: The Foundation of Cooperation in Economic Life.
Cambridge MA: MIT Press, pp. 151–191
Fehr, E., S. Gächter and G. Kirchsteiger. 1997. Reciprocity as a Contract
Enforcement Device: Experimental Evidence. Econometrica 65: 833–860.
Finn, B. 2010. Ending on a High Note: Adding a Better End to Effortful Study. Journal
of Experimental Psychology: Learning, Memory, and Cognition 36: 1548–1553.

10:37:57
12
bibliography 183

Fong, C. M., S. Bowles and H. Gintis. 2005. Reciprocity and the Welfare State. In:
Gintis, H., S. Bowles, R. Boyd, and E. Fehr (eds.). Moral Sentiments and Material
Interest: The Foundation of Cooperation in Economic Life. Cambridge MA:
MIT Press, pp. 277–302.
Foot, P. 1967. The Problem of Abortion and the Doctrine of the Double Effect.
Oxford Review 5: 1–7.
Forsythe, R., J. Horowitz, E. Savin and M. Sefton. 1994. Replicability, Fairness and
Pay in Experiments with Simple Bargaining Games. Games and Economic
Behavior 6: 347–369.
Frederick, S., G. Loewenstein and T. O’Donoghue. 2002. Time Discounting and
Time Preference: A Critical Review. Journal of Economic Literature 40:
351–401.
Fredrickson, B. L. 2000. Extracting Meaning From Past Affective Experiences:
The Importance of Peaks, Ends, and Specific Emotions. Cognition and
Emotion 14: 577–606.
Fredrickson, B. L., and D. Kahneman. 1993. Duration Neglect in Retrospective
Evaluations of Affective Episodes. Journal of Personality and Social
Psychology 65: 44–55.
Frey, B. S. 2001. Inspiring Economics: Human Motivation in Political Economy.
Cheltenham: Edward Elgar.
Frey, B. S., and F. Oberholzer-Gee. 1997. The Cost of Price Incentives: An Empirical
Analysis of Motivation Crowding-Out. American Economic Review 87:
746–55.
Friedman, M., and L. J. Savage. 1948. The Utility of Choices Involving Risk. Journal
of Political Economy 56: 279–304.
Genovesi, A. 2005 [1765–67]. Delle Lezioni di Commercio o sia di Economia
Civile. Napoli: Instituto Italiano per gli Studi Filofící.
Gintis, H., S. Bowles, R. Boyd and E. Fehr (eds.). 2005. Moral Sentiments and
Material Interest: The Foundation of Cooperation in Economic Life.
Cambridge MA: MIT Press.
Gneezy, U., and A. Rustichini. 2000. A Fine is a Price. The Journal of Legal Studies
29: 1–17.
Gneezy, U., and A. Rustichini. 2000b. Pay Enough or Don’t Pay at All. Quarterly
Journal of Economics 115: 791–810.
Gonzalez, R., and G. Wu. 1999. On the Shape of the Probability Weighting
Function. Cognitive Psychology 38: 129–166.
Green, L. W., A. L. Wilson and C. Y. Lavato. 1986. What Changes Can Health
Promotion Achieve and How Long Do These Changes Last? The Trade-Offs
Between Expediency and Durability. Preventive Medicine 15: 508–521.

10:37:57
12
184 bibliography

Grether, D. M., and C. R. Plott. 1979. Economic Theory of Choice and the
Preference Reversal Phenomenon. American Economic Review 69: 623–638.
Guthrie, C. 2003. Prospect Theory, Risk Preference, and the Law. Northwestern
University Law Review 97: 1115–1163.
Haidt, J. 2012. The Righteous Mind: Why Good People are Divided by Politics and
Religion. London: Penguin.
Halpern, D. 2015. Inside the Nudge Unit: How Small Changes Can Make a Big
Difference. London: WH Allen.
Hausman, D. M. 2015. Valuing Health: Well-Being, Freedom and Suffering.
Oxford: Oxford University Press, p. 114.
Henrich, J. 2000. Does Culture Matter in Economic Behaviour: Ultimatum Game
Bargaining Among the Machiguenga of the Peruvian Amazon. American
Economic Review 90: 973–979.
Henrich, N., and J. Henrich. 2007. Why Humans Cooperate: A Cultural and
Evolutionary Explanation. Oxford: Oxford University Press.
Héritier, A., and D. Lehmkuhl. 2008 The Shadow of Hierarchy and New Modes of
Governance. Journal of Public Policy 28: 1–17.
Hershey, J. C., and P. J. H. Schoemaker. 1980. Prospect Theory’s Reflection
Hypothesis: A Critical Examination. Organizational Behavior and Human
Performance 25: 395–418.
Hershey, J. C., and P. J. H. Schoemaker. 1985. Probability Versus Certainty
Equivalence Methods in Utility Measurement: Are They Equivalent?
Management Science 31: 1213–1231.
Heukelom, F. 2012. A Sense of Mission: The Alfred P. Sloan and Russell Sage
Foundations’ Behavioral Economics Program, 1984–1992. Science in Context
25: 263–286.
House of Lords. 2011. Behaviour Change. London: The Stationery Office.
Hsee, C. K., and R. P. Abelson. 1991. Velocity Relation: Satisfaction as a Function of
the First Derivative of Outcome Over Time. Journal of Personality and Social
Psychology 60: 341–347.
Hsee, C. K., R. P. Abelson and P. Salovey. 1991. The Relative Weighting of Position
and Velocity in Satisfaction. Psychological Science 2: 263–266.
Huber, J., J. W. Payne and C. Puto. 1982. Adding Asymmetrically Dominated
Alternatives: Violations of Regularity and the Similarity Hypothesis. Journal
of Consumer Research 9: 90–98.
Hume, D. 2004 [1739]. A Treatise of Human Nature. London: Penguin Classics.
Hume, D. 1975 [1742]. Of the Independency of Parliament. In: Green, T. H., and
T. H. Gross (eds.). Essays, Moral, Political and Literary, Volume 1. London:
Longmans.

10:37:57
12
bibliography 185

Hume, D. 1983 [1777]. An Enquiry Concerning the Principles of Morals.


Indianapolis: Hackett Publishing Company.
Institute of Medicine. 1999. To Err is Human: Building a Safer Health Care System.
Washington DC: National Academy Press.
Institute of Medicine. 2001. Crossing the Quality Chasm: A New Health System
for the 21st Century. Washington DC: National Academy Press.
Johannesson, M., J. S. Pliskin and M. C. Weinstein. 1994. A Note on QALYs, Time
Tradeoff, and Discounting. Medical Decision Making 14: 188–193.
John, P., S. Cotterill, A. Moseley, L. Richardson, G. Smith, G. Stoker and C. Wales.
2011. Nudge, Nudge, Think, Think: Experimenting with Ways to Change Civic
Behaviour. London: Bloomsbury Academic.
Johnson, E. J., and D. A. Schkade. 1989. Bias in Utility Assessments: Further
Evidence and Explanations. Management Science 35: 406–424.
Kacelnik, A. 1998. Normative and Descriptive Models of Decision Making: Time
Discounting and Risk Sensitivity. In: Oaksford, M., and N. Chater (eds.).
Rational Models of Cognition. Oxford: Oxford University Press, pp. 54–70.
Kacelnik, A., and C. El Mouden. 2013. Triumphs and Trials of the Risk Paradigm.
Animal Behaviour 86: 117–129.
Kahan, D. M. 2005. The Logic of Reciprocity: Trust, Collective Action, and Law. In:
Gintis, H., S. Bowles, R. Boyd and E. Fehr (eds.). Moral Sentiments and Material
Interest: The Foundation of Cooperation in Economic Life. Cambridge MA:
MIT Press, pp. 339–378.
Kahneman, D. 2011. Thinking, Fast and Slow. London: Allen Lane.
Kahneman, D., and A. Tversky. 1979. Prospect Theory: An Analysis of Decision
Under Risk. Econometrica 47: 263–292.
Kahneman, D., P. Slovic and A. Tversky (eds.). 1982. Judgement under
Uncertainty: Heuristics and Biases. Cambridge: Cambridge University
Press.
Kahneman, D., J. L. Knetsch and R. H. Thaler. 1986. Fairness and the Assumptions
of Economics. Journal of Business 59: s285–s300.
Kahneman, D., J. L. Knetsch and R. Thaler. 1990. Experimental Tests of the
Endowment Effect and the Coarse Theorem. Journal of Political Economy 98:
1325–1348.
Kahneman, D., B. L. Fredrickson, C. A. Schreiber and D. A. Redelmeier. 1993. When
More Pain is Preferred to Less: Adding a Better End. Psychological Science 4:
401–405.
Kahneman, D., P. P. Wakker and R. Sarin. 1997. Back to Bentham? Explorations of
Expected Utility. The Quarterly Journal of Economics 112: 375–405.

10:37:57
12
186 bibliography

Kane, R. L., P. E. Johnson, R. J. Town and M. Butler. 2004. A Structured Review of


the Effect of Economic Incentives on Consumers’ Preventive Behavior.
American Journal of Preventive Medicine 27: 327–352.
Kelman, M. 2005. Hedonic Psychology and the Ambiguities of ‘Welfare’.
Philosophy and Public Affairs 33: 391–412.
Keynes, J. M. 1936. The General Theory of Employment, Interest and Money.
London: Macmillan, pp. 161–162.
Layard, R. 2005. Happiness: Lessons From a New Science. London: Allen Lane.
Le Grand, J. 1997. Knights, Knaves or Pawns? Human Behaviour and Social Policy.
Journal of Social Policy 26: 149–169.
Le Grand, J. 2003. Motivation, Agency, and Public Policy: Of Knights & Knaves,
Pawns & Queens. Oxford: Oxford University Press.
Lichtenstein, S., and P. Slovic. 1971. Reversals of Preference Between Bids and
Choices in Gambling Decisions. Journal of Experimental Psychology 89:
46–55.
Lichtenstein, S., and P. Slovic. 1973. Response-Induced Reversals of Preference in
Gambling: An Extended Replication in Las Vegas. Journal of Experimental
Psychology 101: 16–20.
Lindman, H. R. 1971. Inconsistent Preferences among Gambles. Journal of
Experimental Psychology 89: 390–397.
Loewenstein, G., and D. Prelec. 1993. Preferences for Sequences of Outcomes.
Psychological Review 100: 91–98.
Loewenstein, G., D. A. Asch, J. Y. Friedman, L. A. Melichar and K. G. Volpp. 2012.
Can Behavioral Economics Make Us Healthier? BMJ 344: e3482.
Loomes, G. C., and R. Sugden. 1982. Regret Theory: An Alternative Theory of
Rational Choice under Uncertainty. Economic Journal 92: 805–824.
Loomes, G. C., and R. Sugden. 1986. Disappointment and Dynamic Consistency in
Choice under Uncertainty. Review of Economic Studies 53: 271–282.
Loomes, G. C., C. Starmer and R. Sugden. 1989. Preference Reversal: Information
Processing Effect or Rational Nontransitive Choice? Economic Journal 99:
140–151.
Loomes, G. C., C. Starmer and R. Sugden. 1991. Observing Violations of
Transitivity by Experimental Methods. Econometrica 59: 425–439.
Lowndes, V., and C. Skelcher. 1998. The Dynamics of Multi-Organizational
Partnerships: An Analysis of Changing Modes of Governance. Public
Administration 76: 313–333.
MacCrimmon, K. R., and S. Larsson. 1979. Utility Theory: Axioms Versus
Paradoxes. In: Allais, M. and O. Hagen (eds.). Expected Utility Hypotheses
and the Allais Paradox. Dordrecht: Reidel, pp. 27–145.

10:37:57
12
bibliography 187

Machina, M. J. 1989. Choice Under Uncertainty: Problems Solved and Unsolved.


In: Hey, J. D. (ed.). Current Issues in Microeconomics. Basingstoke: MacMillan.
Malotte, C. K., R. Ledsky, M. Hogben, M. Larro, S. Middlestadt, J. S. St. Lawrence,
G. Olthoff, R. H. Settlage, N. L. Van Devanter and the GCAP Study Group. 2004.
Comparison of Methods to Increase Repeat Testing in Persons Treated for
Gonorrhea and/or Chlamydia at Public Sexually Transmitted Disease Clinics.
Sexually Transmitted Diseases 31: 637–642.
Marglin, S. A. 1963. The Social Rate of Discount and the Optimal Rate of
Investment. Quarterly Journal of Economics 77: 95–111.
Markowitz, H. 1952. The Utility of Wealth. Journal of Political Economy 60:
151–158.
Marschak, J. 1950. Rational Behaviour, Uncertain Prospects, and Measurable
Utility. Econometrica 18: 111–141.
Marteau, T. M., D. Ogilvie, M. Roland, M. Suhrcke and M. P. Kelly. 2011. Judging
Nudging: Can Nudging Improve Population Health? BMJ 342: d228.
Mauss, M. 1954. The Gift: Forms and Functions of Exchange in Archaic Societies.
Glencoe, IL: The Free Press.
McDaid, D., A. Oliver and S. Merkur. 2014. What Do We Know About the Strengths
and Weakness of Different Policy Mechanisms to Influence Health Behaviour
in the Population? Policy summary, 15. WHO Regional Office for Europe.
Copenhagen: Denmark.
Mill, J. S. 1873. Autobiography. London: Longmans, Green, Reader and Dyer.
National Institute for Public Health and the Environment. 2011. Towards Better
Health. The Hague: Klutworks.
Oullier, O., and S. Sauneron (eds.). 2010. Improving Public Health Prevention with
Behavioural, Cognitive and Neuroscience. Paris: Centre D’Analyse
Strategique.
Pareto, V. 1971. Manual of Political Economy. Translated by Schwier, A. S.
New York: August Kelley.
Parfit, D. 1984. Reasons and Persons. Oxford: Oxford University Press.
Paul-Ebhohimhen, V., and A. Avenell. 2007. Systematic Review of the Use of
Financial Incentives in Treatments for Obesity and Overweight. Obesity
Reviews 9: 355–367.
Pigou, A. C. 1932. The Economics of Welfare. 4th Edition. London: Macmillan.
Pikkety, T. 2014. Capital in the Twenty-First Century. Cambridge MA: Harvard
University Press.
Pinker, R. 2006. From Gift Relationships to Quasi-markets: An Odyssey along the
Policy Paths of Altruism and Egoism. Social Policy & Administration 40:
10–25.

10:37:57
12
188 bibliography

Popper, K. 2002 [1957]. The Poverty of Historicism. New York: Routledge Classics.
Popper, K. R. 1963. Conjectures and Refutations: The Growth of Scientific
Knowledge. New York: Basic Books.
Preston, M. G., and P. Baratta. 1948. An Experimental Study of the Auction-Value
of an Uncertain Outcome. The American Journal of Psychology 61: 183–193.
Rabin, M. 2013. Healthy Habits: Some Thoughts on the Role of Public Policy in
Healthful Eating and Exercise Under Limited Rationality. In: Oliver, A. (ed.).
Behavioural Public Policy. Cambridge: Cambridge University Press, pp. 115–139.
Rae, J. 1895. Life of Adam Smith. London: Macmillan.
Raiffa, H. 1961. Risk, Ambiguity and the Savage Axioms: Comment. The Quarterly
Journal of Economics 75: 690–694.
Ramsey, F. P. 1928. A Mathematical Theory of Saving. The Economic Journal 38:
543–559.
Rand, A. 2007. The Fountainhead. London: Penguin Books.
Read, D., and B. van Leeuwen. 1998. Predicting Hunger: The Effects of Appetite and
Delay on Choice. Organizational Behavior and Human Decision Processes 76:
189–205.
Read, D., G. Loewenstein and S. Kalyanaraman. 1999. Mixing Virtue and Vice:
Combining the Immediacy Effect and the Diversification Heuristic. Journal of
Behavioral Decision Making 12: 257–273.
Redelmeier, D. A., and D. Kahneman. 1996. Patient’s Memories of Painful Medical
Treatments – Real-Time and Retrospective Evaluations of Two Minimally
Invasive Procedures. Pain 66: 3–8.
Redelmeier, D. A., J. Katz and D. Kahneman. 2003. Memories of Colonoscopy:
A Randomized Trial. Pain 104: 187–194.
Rosenthal, M. B., and R. G. Frank. 2006. What Is the Empirical Basis for Paying for
Quality in Health Care? Medical Care Research and Review 63: 135–157.
Samuelson, P. 1937. A Note on Measurement of Utility. Review of Economic
Studies 4: 155–161.
Samuelson, P. 1952. Probability, Utility, and the Independence Axiom.
Econometrica 20: 670–678.
Savage, L. J. 1954. The Foundations of Statistics. New York: Wiley.
Schmeidler, D., and P. Wakker. 1990. Expected Utility and Mathematical
Expectation. In: Eatwell, J., M. Milgate and P. Newman (eds.). The New
Palgrave. Utility and Probability. New York: Macmillan Press, pp. 70–78.
Schreiber, C. A., and D. Kahneman. 2000. Determinants of the Remembered Utility
of Aversive Sounds. Journal of Experimental Psychology: General 129: 27–42.
Seal, K. H., A. H. Kral, J. Lorvick, A. McNees, L. Gee and B. R. Edlin. 2003.
A Randomized Controlled Trial of Monetary Incentives versus Outreach to

10:37:57
12
bibliography 189

Enhance Adherence to the Hepatitis B Vaccine Series among Injection Drug


Users. Drug and Alcohol Dependence 71: 127–131.
Sen, A. 1999. Development as Freedom. New York: Random House.
Sethi, R., and E. Somanathan. 2005. Norm Compliance and Strong Reciprocity. In:
Gintis, H., S. Bowles, R. Boyd and E. Fehr (eds.). Moral Sentiments and Material
Interest: The Foundation of Cooperation in Economic Life. Cambridge MA:
MIT Press, pp. 242–243.
Shaw, J. 2007. Is It Acceptable for People to be Paid to Adhere to Medication? No.
BMJ 335: 233.
Simon, H. A. 1956. Rational Choice and the Structure of the Environment.
Psychological Review 63: 129–138.
Silk, J. B. 2005. The Evolution of Cooperation in Primate Groups. In: Gintis, H.,
S. Bowles, R. Boyd and E. Fehr (eds.). Moral Sentiments and Material Interest:
The Foundation of Cooperation in Economic Life. Cambridge MA: MIT Press,
pp. 43–73.
Slater, J. S., G. A. Henly, C. N. Ha, M. E. Malone, J. A. Nyman, S. Diaz and
P. G. McGovern. 2005. Effect of Direct Mail as a Population-Based Strategy to
Increase Mammography Use among Low-Income Underinsured Women Ages
40 to 64 Years. Cancer Epidemiology, Biomarkers, and Prevention 14:
2346–2352.
Slovic, P., and S. Lichtenstein. 1968. Relative Importance of Probabilities and
Payoffs in Risk Taking. Journal of Experimental Psychology 78: 1–18.
Slovic, P., and S. Lichtenstein. 1983. Preference Reversals: A Broader Perspective.
American Economic Review 73: 596–605.
Slovic, P., and A. Tversky. 1974. Who Accepts Savage’s Axiom? Behavioral Science
19: 368–373.
Smith, A. 2009 [1759]. The Theory of Moral Sentiments. London: Penguin Classics.
Smith, A. 1999 [1776]. The Wealth of Nations, Book 1. London: Penguin Classics.
Solow, R. M. 1974. The Economics of Resources or the Resources of Economics.
American Economic Review 64: 1–14.
Stanovich, K. E., and R. F. West. 2000. Individual Differences in Reasoning:
Implications for the Rationality Debate. Behavioral and Brain Sciences 23:
645–726.
Stern, N. 2007. The Economics of Climate Change. Cambridge: Cambridge
University Press.
Sugden, R. 1992. How People Choose. In: Hargreaves Heap, S., M. Hollis, B. Lyons,
R. Sugden and A. Weale (eds.). The Theory of Choice: A Critical Guide. Oxford:
Blackwell, pp. 36–50.

10:37:57
12
190 bibliography

Sugden, R. 2009. On Nudging. A Review of Nudge: Improving Decisions about


Health, Wealth and Happiness by Richard H. Thaler and Cass R. Sunstein.
International Journal of the Economics of Business 16: 365–373.
Sunstein, C. 2013. Simpler: The Future of Government. London: Simon and
Schuster.
Tanenbaum, S. J. 2009. Pay-for-Performance in Medicare: Evidentiary Irony and the
Politics of Value. Journal of Health Politics, Policy and Law 34: 717–746.
Thaler, R. H. 1994. The Winner’s Curse: Paradoxes and Anomalies of Economic
Life. Princeton: Princeton University Press.
Thaler, R. H. 2015. Misbehaving: The Making of Behavioural Economics.
New York: Penguin Random House.
Thaler, R. H., and C. R. Sunstein. 2003. Libertarian Paternalism. The American
Economic Review 93: 175–179.
Thaler, R. H., and C. R. Sunstein. 2008. Nudge: Improving Decisions about Health,
Wealth and Happiness. New Haven: Yale University Press.
Titmuss, R. 1970. The Gift Relationship: From Human Blood to Social Policy.
London: New Press.
Tomasello, M. 2009. Why We Cooperate. Cambridge MA: MIT Press.
Tomasello, M., M. Carpenter, J. Call, T. Behne and H. Moll. 2005. Understanding
and Sharing Intentions: The Origins of Cultural Cognition. Behavioral and
Brain Sciences 28: 675–691.
Tversky, A., and D. Kahneman. 1974. Judgement under Uncertainty: Heuristics and
Biases. Biases in Judgments Reveal Some Heuristics of Thinking under
Uncertainty. Science 185: 1124–1131.
Tversky, A., and D. Kahneman. 1981. The Framing of Decisions and the Psychology
of Choice. Science 211: 453–458.
Tversky, A., and D. Kahneman. 1992. Advances in Prospect Theory: Cumulative
Representation of Uncertainty. Journal of Risk and Uncertainty 5: 297–323.
Tversky, A., S. Sattath and P. Slovic. 1988. Contingent Weighting in Judgement and
Choice. Psychological Review 95: 371–384.
Tversky, A., P. Slovic and D. Kahneman. 1990. The Causes of Preference Reversals.
American Economic Review 80: 204–217.
UK Treasury. 2003. The Green Book: Appraisal and Evaluation in Central
Government. London: TSO.
Volpp, K. G., A. B. Troxel, M. V. Pauly, H. A. Glick, A. Puig, D. A. Asch, R. Galvin,
J. Zhu, F. Wan, J. DeGuzman, E. Corbett, J. Weiner and J. Audrain-McGovern.
2009. A Randomized, Controlled Trial of Financial Incentives for Smoking
Cessation. New England Journal of Medicine 360: 699–709.

10:37:57
12
bibliography 191

Volpp, K. G., L. K. John, A. B. Troxel, L. Norton, J. Fassbender and G. Loewenstein.


2008. Financial Incentive-Based Approaches for Weight Loss. A Randomized
Trial. Journal of the American Medical Association 300: 2631–2637.
Von Neumann, J., and O. Morgenstern. 1944. Theory of Games and Economic
Behavior. Princeton: Princeton University Press.
Wehrung, D. A. 1989. Risk Taking over Gains and Losses: A Study of Oil
Executives. Annals of Operations Research 19: 115–139.
Wilkinson, G.S. 1984. Reciprocal Food Sharing in the Vampire Bat. Nature 308:
181–184.
World Bank. 2015. World Development Report 2015: Mind, Society and Behavior.
Washington DC: World Bank.
Zabell, S.L. 1990. Daniel Bernoulli. In: Eatwell J., M. Milgate and P. Newman (eds.).
The New Palgrave. Utility and Probability. New York: Macmillan Press, pp.
12–14.

10:37:57
12
Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Index
More Information

Index

acquisition utility, 171 commitment contract, 63, 115, See deposit


alcohol, 122 contract
Alexander Solzhenitsyn effect, 78 common consequence effect, 17, 18, 20, 44, 52
alibi, 112, 117 common ratio effect, 20
Allais paradox, 17, 19, 44, 167 competition, 157
Allais, Maurice, 17, 32 completeness, 9
altruism, 88, 145, 146, 151, 161, 172 concavity, 37, 40
altruistic punishment, 148 confirmation bias, 27
ambiguity, 8 Conly, Sarah, 130, 135, 140
ambiguity aversion, 23 contingent weighting, 31
anchoring, 27, 31, 114, 171 continuity, 8, 9
anticipated disappointment, 51 convexity, 37, 41
anticipated regret, 51 cooperation, 161, 166, 173
asymmetric dominance, 10 crowd out. See motivational crowding
automatic decision making, 109 crowding in, 89, 92, See motivational
availability, 26 crowding
crowding out, 88, 92, 107
Behavioural Insights Team, 1, 63, 113, 116,
119, 120, 122, 123, 124, 126, 160 Darwin, Charles, 148
behavioural public policy cube, 141 day reconstruction method, 82
behavioural regulation, 136, 137, 175 de Waal, Frans, 149
behavioural welfare economics, 86, 130 Deci, Edward, 90, 92
Bentham, Jeremy, 5, 71, 82 declining marginal utility, 56, 59
Bernoulli defaults, 114
Daniel, 3, 7 deliberative decision making, 109
Nicolas, 3 deliberative preferences, 129
Bernoullian expected utility, 5 deposit contract, 63, 64, 99, 119
Blair, Tony, 172 deposit contracts, 99
blood donation, 88 descriptive invariance, 28, 29
Bovens, Luc, 117 descriptive theory, 16, 34
brain imaging, 83 dictator game, 154
budges, 128, 137, 139, 141, 142, 175 disappointment theory, 50, 51, 52
discounted utility model, 55
capuchin monkeys, 149 discounting, 54, 55
cardinal utility, 5, 6, 7, 11 dominance, 38, 76
catastrophic risk premium, 59 dual process theory, 110
certainty effect, 47 duration neglect, 74, 75, 76, 81
certainty equivalence method, 13, 20 dynamic consistency, 60
chimpanzees, 150 dynamic inconsistency, 61, See dynamic
choice, 157 consistency
choice architecture, 146
civic duty, 95 econs, 110
coercive paternalism, 130, 131, 132, 134, 175 egoism, 146, 151, 157, 161, 172, 174

192

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Index
More Information

index 193

Ellsberg paradox, 21, 23, 24 intrinsic motivation, 93, 95, 100


Ellsberg, Daniel, 21 introjected motivation, 91
end effect, 74, 77, See peak-end evaluation
endowment effect, 39, 40 James Dean effect, 77, 85
energy consumption, 124
evolution, 166 Kahneman, Daniel, 25, 28, 29, 36, 43
expected subjective value, 4 Keynes, John Maynard, 164
expected utility theory, 7, 10, 16, 19, 21 knaves, 146
experienced utility, 71, 82, 84, 169 knights, 146
exponential discounting, 60, 64
external regulation, 91 Layard, Richard, 82
externalities, 111, 118, 131, 138, 170 Le Grand, Julian, 146
league table competition, 106, 173
framing, 28, 29 libertarian paternalism, 109, 111, 112, 117,
Frey, Bruno, 90 119, 127, 128, 165, 175
Friedman, Milton, 25 Lichtenstein, Sarah, 30, 31
loss aversion, 39, 40, 41, 45, 103, 139, 166, 168
Genovesi, Antonio, 152 lottery incentives, 99
gestalt characteristics, 71, 79, 169
gifts, 151 Markowitz, Harry, 35, 41
Global Insights Initiative, 2 Marschak-Machina triangle, 10, 19, 45
golden rule, 147 Marshall, Alfred, 6, 162
Grether, David, 32 Mauss, Marcel, 88, 151
group selection theory, 156 mental short cuts, 25
Mill, John Stuart, 170
Halpern, David, 113 Mindspace, 113, 115
happiness, 82, 84, 169 money pump, 9
hard paternalism, 135, See coercive Morgenstern, Oskar, 7
paternalism motivation, 171
harms, 137 motivational crowding, 88, 90, 92, 93
Hausman, Dan, 80
hedonic experiences, 71 naming and shaming, 103, 160
herd effect, 114 National Health Service, 102, 103, 120,
heuristics, 25 123, 124
Hospital Quality Incentive Demonstration negative reciprocity, 155, 159
project, 104 neoclassical economics, 162
hospital star rating system, 104 neuroscience, 83
human agency, 70, 80, 84 normative theory, 16
Hume, David, 54, 146 nudge unit, 116, 117, 127, See Behavioural
hyperbolic discounting, 60, 64, 70, See Insights Team
present bias nudges, 108, 117, 128, 142, 165, 174, 175

identity, 91, 104, 122, 135, 160 obesity, 98, 122


immediacy effect, 60 one shot games, 168
impartial spectator, 54 opting-in, 120
independence, 8, 10, 16, 17, 23 opting-out, 120
instant utility, 71, 75, 82, See experienced ordering, 8, 9
utility ordinal utility, 7
internalities, 111, 117, 131, 132, 136, 138 organ donation, 120, 160
intransitivity, 30, 50, 51 overconfidence, 27

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Index
More Information

194 index

Pareto, Vilfredo, 7 rules of thumb, 25, 28, 31


pay day loans, 141 Ryan, Richard, 90, 92
pay for performance, 102, 158
peak-end evaluation, 73, 75 salience, 114
pensions, 121 Samuelson, Paul, 55
performance incentives, 159 satisficing, 17, 25
performance management, 101, 106, 160 Savage, Leonard, 8
personal financial incentives, 96, 98, 100 scale compatibility, 31
piano stairs, 123 self-determination theory, 90, 91, 122,
Piketty, Thomas, 165 135, 159
plastic bags, 95, 101 Sen, Amartya, 84
Plott, Charles, 32 separability, 10
Popper, Karl, 27, 176 shoves, 128, 131, 132, 133, 142, 175
possibility effect, 48 Simon, Herbert, 24, 25
preference reversals, 30, 31, 32 Slovic, Paul, 30, 31, 167
present bias, 60, 61, 63, 66, 70, 123, 130, 132, smart meters, 125
141, 143 smart regulation, 141
primates, 149 Smith, Adam, 6, 54, 110, 151, 163
priming, 114 smoking cessation, 64, 97, 119, 132,
primitive culture, 153 134, 143
probability equivalence method, 13, 20 Social and Behavioral Sciences Team, 2
probability weighting, 42, 44, 45 social care, 133
procedural invariance, 15, 29, 30 soft paternalism, 108, 135
projection bias, 64 St. Petersburg paradox, 3
prompted choice, 120, 121 standard gamble, 14
prospect theory, 36, 38, 41, 42, 44, 45, 47, status quo bias, 26, 39
168, 171 subjective expected utility theory, 8, 23
cumulative, 38 subjective probability, 43
publication bias, 93 Sugden, Robert, 130, 135
Sunstein, Cass, 108, 110, 111, 141
QALYs, 80, See quality adjusted sure thing principle, 10, 23
life-years swine flu, 23
quality adjusted life-years, 14, 69, 79 synthetic biology, 20
system 1, 116
Raiffa, Howard, 16
Rand, Ayn, 40 temporal dominance, 78, 81, See temporal
rank dependent utility theory, 38 monotonicity
ranking, 104 temporal monotonicity, 76
reciprocal altruism, 148, 149, 155 Thaler, Richard, 108, 110, 111, 116, 166
reciprocity, 88, 121, 146, 147, 150, 151, 153, think policy, 143
154, 155, 156, 157, 160, 166, 173 time, 54
reference point, 39, 103, 171 time preference, 55, 57, 59
reflection effect, 46, 47, 49 time trade-off, 67, 69
regret theory, 50, 51, 52 Titmuss, Richard, 88, 90
relative price mechanism, 88, 95 Tomasello, Michael, 149
repeat games, 168 traffic light symbols, 139
representativeness, 26 transaction utility, 171
reputation, 126, 150, 160 transitivity, 9
reputational competition, 160 Trolley Problem, 140
risk, 8 Tversky, Amos, 25, 28, 29, 36, 43, 167

© in this web service Cambridge University Press www.cambridge.org


Cambridge University Press
978-1-316-51026-1 — The Origins of Behavioural Public Policy
Adam Oliver
Index
More Information

index 195

ultimatum game, 153, 154, 156 von Neumann, John, 7


Ulysses contract, 63, See deposit contract
uncertainty, 8 waiting times, 104
utilitarianism, 5 wellbeing, 57

© in this web service Cambridge University Press www.cambridge.org