The State of Economics, the State of the World
The State of Economics, the State of the World
Edited by Kaushik Basu, David Rosenblatt, and Claudia Sepúlveda
The MIT Press
Cambridge, Massachusetts
London, England
© 2019 International Bank for Reconstruction and Development / The World Bank
This work is available under the Creative Commons Attribution—NonCommercial—
NoDerivatives 3.0 IGO license (CC BY-NC-ND 3.0 IGO) http://creativecommons.org
/licenses/by-nc-nd/3.0/igo.
Some rights reserved
The findings, interpretations, and conclusions expressed in this work are those of the
authors and do not necessarily reflect the views of The World Bank, its Board of Execu-
tive Directors, or the governments they represent. The World Bank does not guarantee the
accuracy, completeness, or currency of the data included in this work and does not assume
responsibility for any errors, omissions, or discrepancies in the information, or liability with
respect to the use of or failure to use the information, methods, processes, or conclusions set
forth. The boundaries, colors, denominations, and other information shown on any map in
this work do not imply any judgment on the part of The World Bank concerning the legal
status of any territory or the endorsement or acceptance of such boundaries.
Nothing herein shall constitute or be construed or considered to be a limitation upon or
waiver of the privileges and immunities of The World Bank, all of which are specifically
reserved.
Attribution—Please cite the work as follows: The World Bank. 2019. The state of economics,
the state of the world / edited by Kaushik Basu, Claudia Sepulveda, and David Rosenblatt.
Published by MIT Press. © World Bank. License: Creative Commons Attribution—
NonCommercial—NoDerivatives 3.0 IGO (CC BY-NC-ND 3.0 IGO).
Third-party content—The World Bank does not necessarily own each component of the
content contained within the work. The World Bank therefore does not warrant that the
use of any third-party-owned individual component or part contained in the work will not
infringe on the rights of those third parties. The risk of claims resulting from such infringe-
ment rests solely with you. If you wish to re-use a component of the work, it is your respon-
sibility to determine whether permission is needed for that re-use and to obtain permission
from the copyright owner. Examples of components can include, but are not limited to,
tables, figures, or images.
All queries on rights and licenses should be addressed to the Publishing and Knowledge Divi-
sion, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625;
e-mail: pubrights@worldbank.org.
This book was set in Stone Serif and Stone Sans by Westchester Publishing Services. Printed
and bound in the United States of America.
Library of Congress Cataloging-in-Publication Data
Names: Basu, Kaushik, editor. | Sepúlveda, Claudia Paz, 1969– editor. |
Rosenblatt, David, editor.
Title: The state of economics, the state of the world / edited by Kaushik
Basu, Claudia Sepulveda, and David Rosenblatt.
Description: Cambridge, MA : MIT Press, [2019] | Includes bibliographical
references and index.
Identifiers: LCCN 2018046336 | ISBN 9780262039994 (hardcover : alk. paper)
Subjects: LCSH: Economic development. | Information technology—Economic
aspects. | Monetary policy. | Social change.
Classification: LCC HD82 .S8223 2019 | DDC 330.1—dc23
LC record available at https://lccn.loc.gov/2018046336
10 9 8 7 6 5 4 3 2 1
Contents
Preface vii
Introduction: The State of Economics, the State of the World 1
Kaushik Basu 1
I Foundations
1 Equilibrium, Welfare, and Information 23
Kenneth Arrow
Comments by Shantayanan Devarajan and Karla Hoff 34
2 Social Choice and Welfare Economics 47
Amartya Sen
Comments by Célestin Monga and James E. Foster 77
3 The Revolution of Information Economics: The Past
and the Future 101
Joseph Stiglitz
Comments by Ravi Kanbur and Hamid Rashid 139
II Macroeconomic Stabilization and Growth
4 From Chronic Inflation to Chronic Deflation: Focusing on
Expectation and Liquidity Disarray since World War II 153
Guillermo Calvo
Comments by Gita Gopinath and Luis Servén 193
5 Global Liquidity and Procyclicality 207
Hyun Song Shin
Kunt and Maurice Obstfeld 233
Comments by Aslı Demirgüç-
vi Contents
6 Growth and Development from a Schumpeterian
Perspective 253
Philippe Aghion
Comments by Francesco Caselli and Aart Kraay 280
III New Areas of Research and Inquiry
7 Climate Change, Development, Poverty, and Economics 295
Sam Fankhauser and Nicholas Stern
Comments by Michael Toman and Gaël Giraud 321
8 Behaviorally Informed 349
Cass R. Sunstein
Comments by Robert Hockett and Varun Gauri 372
9 Morality: Evolutionary Foundations and Policy
Implications 389
Ingela Alger and Jörgen W. Weibull
Comments by Lawrence E. Blume and Xavier Giné 417
10 The Influence of Randomized Controlled Trials on
Development Economics Research and on
Development Policy 439
Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
Comments by David McKenzie and Martin Ravallion 488
Contributors 499
Index 503
Preface
Kaushik Basu, David Rosenblatt, and Claudia Sepúlveda
Origin
We live in troubled times. Over the past decade, the world economy has
been wracked by financial crises, sovereign debt problems, backlash from
political conflict and migrant crises, and, recently, a rise in xenophobia
and protectionism. These issues raise major questions about the state of the
world and also about the ability of economics to take on such challenges.
ups symptoms of
Are these many economic and political crises and flare-
some deeper, underlying issues? Is economics as a discipline failing us at
this time of soul searching? These are the questions that many are asking
and that prompted the conference at the World Bank on which this book is
based. We decided to bring in some of the finest minds in the profession—
to ponder the state of
economists who have shaped modern economics—
the field and the state of the world in a series of papers. The conference
consisted of 2 days of deliberation: The papers were presented, a distin-
guished group of economists commented on the presentations, and a large
audience engaged with them in conversation and debate. This book is the
outcome of these 2 days of deliberation.
In the 1950s through the 1970s, neoclassical economics reached a rea-
sonable consensus in the economics profession, at least in the “West.” The
United States and Western Europe experienced postwar rapid economic
growth. Asia was still a sleeping giant in economic terms, the Soviet Union—
was very much intact, and African
with its particular economic system—
countries were only beginning a wave of independence from colonial rule.
Development economics focused on structural transformation along the
lines laid out by Sir Arthur Lewis, and dependency theories also emerged
viii Preface
that asserted that the global capitalist system was essentially rigged against
the developing world. Despite the neoclassical consensus, some economists
believed that advanced mathematical and engineering techniques could
allow social planners to optimally set the path of economic growth and
development.
Fast forward in history and one sees a very different evolution of the
global economy over the past 25 years. The latest wave of globalization
has led to the intensification of global value chains. Asia is now home to
some of the most advanced economies on earth. It began with Japan, which
was soon followed by Singapore, South Korea, Taiwan, and Hong Kong. By
1980s, China was a growth leader, and in recent times, India and
the mid-
Vietnam are growing at exemplary rates. The Soviet Union no longer exists.
income countries—
Many middle- have
including those in Latin America—
achieved social progress, but dramatic income inequality persists, as do
challenges to compete in the new global context. Africa has emerged from
debt relief to achieve growth and reduced poverty rates, albeit at a variable
and erratic pace. Rapid technological change provides both opportunities
for technological leapfrogging as well as challenges to adapt.
As we see in this volume, the economics profession has adapted to the
changing state of the world by learning from practical experience, challeng-
ing traditional assumptions, and developing new techniques and the use of
big data. A predominant view in Western universities in the 1980s was that
all economies were alike and that all developing countries needed to do was
to “get prices right.” Development economics languished as a field of study.
Since then, as developing economies have gained more prominence on the
global stage, development economics has become one of the most dynamic
particularly in terms of new statistical techniques and
fields in economics—
the ability to blend economic theory with empirical methods.
Despite all these changes and adaptations, the financial crisis that started
in 2008 and caused a protracted recession has left scars on the world econ-
omy that linger even today. These scars show that the economics profession
still faces major intellectual and research challenges. Addressing such chal-
lenges was one of the motivations for our conference.
With time, the societal goals of economics and the normative presump-
tions underlying the profession have also shifted. From a narrow focus on
gross domestic product, economists have come to recognize the need for a
broader conception of human welfare and capability. Even the World Bank
Preface ix
decided to broaden its mission goals from development and poverty reduc-
tion to a more direct targeting of inequality mitigation, which it refers to
as the promotion of “shared prosperity.” This book is an assessment of our
discipline at the crossroad of all these changes.
We regret that Kenneth Arrow did not live to see the publication of this
book. Kenneth Arrow was one of the greatest minds of our time, an econo-
mist who straddled like a colossus the second half of the past century and
the opening years of this one, and who opened the conference with a pre-
sentation of enormous sweep. Ken Arrow passed away at the age of 95 on
February 21, 2017, while he was working on completing his paper for this
volume. He was in touch with us in our capacity as the volume editors until
the last weeks of his life. After some deliberation, we decided to include his
paper, despite it being an unfinished work. We did not want to put words
into his mouth, nor leave out this final statement from him. We are grateful
to Larry Summers for helping us edit the paper lightly; we also worked on it
to make obvious corrections but took care not to change any of the original
meanings. As a result, some parts of the paper are obviously incomplete.
We hope that this chapter from an economist who helped shape so much
of modern economics will be of value to all readers. Indeed, we believe that
the three sweeping essays by Kenneth Arrow, Amartya
part I of this book—
will be viewed as a short summary of the theoreti-
Sen, and Joseph Stiglitz—
cal foundations of modern economics.
Road Map
The Introduction that follows this Preface recounts the intellectual under-
twentieth
pinnings that preceded the neoclassical consensus of the mid-
century. This historical perspective reminds us of the role of theory and
even in the cur-
intuition in guiding our understanding of economics—
rent age of more abundant data and more evolved statistical analysis. The
Introduction makes the case that both theory and empirics are essential to
closing key knowledge gaps and crafting policy that can enhance human
well-being.
Thereafter, the book is organized in three parts. Part I deals with Foun-
century economic theory—
dations. Twentieth- or neoclassical economics—
reached a pinnacle in the middle of the past century based on two pillars:
x Preface
general equilibrium theory and welfare economics. Part I includes chap-
ters by three Nobel laureates. Ken Arrow and Amartya Sen each provide a
recounting of the two pillars of equilibrium and welfare. Chapter 1, Pro-
fessor Arrow’s contribution, is poignant, being published posthumously. It
takes us through the origins of some of the key ideas of economic theory,
going back to John Stuart Mill and Augustin Cournot and to the birth of
the “demand curve,” which would be such a central idea for so much of
economics. Professor Arrow tells the history of economic thought that led
to the formal characterization of general equilibrium and to a proof of
existence and its optimality properties, which are enshrined in the two
fundamental theorems of welfare economics. It was a monumental break-
through for economics when he and Gerard Debreu published their 1954
paper in Econometrica.1 In chapter 1 in this book, Ken Arrow points out
how we need to be careful when jumping from these abstract ideas to
policy decisions. He reminds us that economics is different from a science
like astronomy: In economics, we are ourselves participants in the system
that we are trying to understand. Thus, we are too close to the subject of
our analysis. As a consequence, we might not see the whole picture, or our
views might be biased.
Amartya Sen has done pioneering work on individual choice and social
welfare, with fundamental research that lies at the intersection of economics
and philosophy, a pointed example being his celebrated “liberty paradox,”
which has spawned a large literature in both disciplines. In chapter 2 of
making
this book, Sen provides a history of the theory of rational decision-
and social welfare. He notes that the early theorists of the late eighteenth
century were preoccupied with two concerns: avoiding authoritarianism
and avoiding arbitrariness. Sen’s chapter is a natural sequel to chapter 1
by Arrow. Just as Professor Arrow was a key figure in general equilibrium
theory, he also provided the initial impetus for social choice theory, with
his famous “impossibility theorem.” Professor Sen, arguably the leading
social choice theorist in the world, relates the work of Arrow to research
going back to the work of John Stuart Mill in the nineteenth century. Turn-
ing to welfare economics, Sen starts with Pigou’s classic 1920 book on the
subject.2 Unlike social choice theory, welfare economics’ philosophical ori-
1. Arrow and Debreu (1954).
2. Pigou (1920).
Preface xi
gin lies in Bentham’s utilitarian approach, and consequently, it focuses on
the sum of the utilities of the individuals in the community. Sen notes that
the disregard for the distribution of those utilities reflects “a partial blind-
ness of considerable ethical and political import” and goes on to elaborate
on how this neglect can be remedied. This major recounting of rational
choice and welfare economics will be useful both for students of economics
and philosophy, and for researchers trying to break new ground.
In chapter 3, Joseph Stiglitz summarizes the evolution of the economics
of information and the role of information asymmetry in market failures,
fields in which he himself has made seminal contributions. Much of early
economics was based on the assumption of perfect information. The con-
sumer knew what kind of good she was buying, the creditor knew exactly
what the risks of lending to a person or a firm were, and the employer knew
how good a worker he was hiring and also had full information on what
the worker was doing when on the job. All these assumptions are of course
wrong. But economists persisted with them, often in the belief that they
were innocuous assumptions that made it easier to build models and make
progress, but at times out of cussedness. In a series of papers, Joe Stiglitz
they led to seri-
showed that, first, the assumptions were not innocuous—
and second, with patience and ingenuity, we could
ous policy mistakes—
make room for imperfect and asymmetric information and still build for-
mal models of analysis.
Some critical features of traditional economics (such as wage rigidities,
excess supply of labor, and excess demand for credit), which in the works
of Keynes and Arthur Lewis were assumptions, could now be explained
endogenously. Thanks to Professor Stiglitz’s early publications, this work is
eye
now part of the mainstream, and chapter 3 of this book provides a bird’s-
view of the background for this field.
Part II of the book consists of three chapters that deal with macroeco-
nomic stabilization and growth. Developing countries have suffered mul-
quarters of a century. But
tiple macroeconomic crises over the past three-
2009 global financial crisis that started in the United States, pum-
the 2008–
meled many rich countries, and then swept through developing economies
has resulted in some deep soul searching in the profession of economics.
One issue that has become clear to the economics profession, based on
experience, is the close link between macroeconomic policies and the regu-
lation and evolution of the financial system. In part II, Guillermo Calvo
xii Preface
(chapter 4) and Hyun Song Shin (chapter 5) discuss new thinking on infla-
run stabiliza-
tion and financial stability, respectively. Moving from short-
run growth, theory has evolved beyond the original Solow
tion to long-
model to endogenize Solow’s careful accounting of the role of total factor
productivity. Part II closes with chapter 6 by Philippe Aghion, which brings
the reader up on the latest thinking on endogenous growth theory.
Guillermo Calvo’s focus in chapter 4 is on more chronic but equally
compelling matters. His concern is with two key macroeconomic phe-
nomena that have occurred since the middle of the past century: chronic
inflation and more recently, chronic deflation. From the perspective of the
history of economic thought, Calvo draws on the role of rational expec-
tations in macroeconomic theory and its role in helping us understand
these phenomena. Although some rich countries experienced unusually
high inflation in the 1970s, emerging markets suffered much more severe
inflationary episodes, accompanied by debt crises. Macroeconomists ini-
tially attributed the emerging market crises purely to policy mistakes that
affected the fundamentals for investing in those markets. However, the per-
sistence and systemic nature of these crises led economists to think about
the role of expectations in generating “sudden stops” of access to foreign
capital. Guillermo Calvo, who pioneered this literature, is clearly in a spe-
cial position to review it. In keeping with a recurring theme of this book,
Calvo points to “intellectual inertia,” triggered by traditional models work-
income countries,
ing well to explain macroeconomic performance in high-
as a probable cause of some our discipline’s failings.
including exchange
More than ever, financial market developments—
are impacting the real economy. As Hyun Song Shin puts
rate movements—
it in “Global Liquidity and Procyclicality” (chapter 5), “the financial tail
appears to be wagging the real economy dog.” More specifically, exchange
rates do not seem to adjust in the required direction to help eliminate exter-
nal imbalances in key economies. Global financial markets have become
highly integrated, implying that policy makers everywhere are focused on
the next move of the US Federal Reserve Board. Anomalies in interest rates
across currencies, the rise of the dollar in global transactions, and cyclical
instabilities have been a focus of a lot of our attention, especially since the
2008−2009 financial crisis. Shin, a world authority on international finance,
dissects and analyzes these concerns in a chapter that is of special interest
in today’s world, especially since the financial sector crisis of a decade ago.
Preface xiii
Philippe Aghion has contributed to many areas of economic theory.
One of his works that attracted an enormous amount of attention with
up research is the “Schumpeterian theory of eco-
an abundance of follow-
nomic growth.” Although nations strive to fulfill many different objectives,
growth is a central concern of development economics, if for no other reason
than as an enabler of some of our other aims and objectives. In chapter 6,
Aghion starts from the Solow model, “the true template in growth econom-
ics,” and goes on to use the Schumpeterian growth paradigm to shed light
on a host of topics of contemporary interest. Thus, his chapter analyzes the
led growth; the possible
relationship between competition and innovation-
causes of secular stagnation; and the recent rise in inequality, especially the
rich and the rest.
gap between the super-
Part III of the book is a set of four chapters brought together under the
heading “New Areas of Research and Inquiry.” These four chapters repre-
sent branches of economics that are relatively new. They are based largely
on challenging the traditional assumptions of neoclassical economic the-
ory and traditional approaches to empirical economics, as well as on the
application of economics to emerging global concerns.
Chapter 7 is based on the lecture at the conference given by Nick Stern,
the world’s leading authority on environmental economics and the eco-
nomics of climate change. The chapter provides an overview of the eco-
perhaps the most pressing—
nomics of climate change— and the most
issue of our times, concerning all nations. Written jointly with
fractious—
Sam Fankhauser, chapter 7 provides a thorough overview of the unique
threat to global prosperity that is posed by climate change. The authors
review the history of environmental and natural resource economics. They
then make the case for a “radical deepening of economics analysis” to accom-
modate sustainability concerns and guide the policy response to climate
change. It is unfortunate that development policy traditionally did not focus
on environmental issues, despite work on environment and natural resources
dating back to the eighteenth and nineteenth centuries. The risks posed by
climate change are staggering, and the options that we have are laid out with
care in this chapter.
No stocktaking of modern economics is complete without an account
of behavioral economics. Cass Sunstein is a leading authority on law and
economics and on behavioral economics, with original works in both these
fields. In chapter 8, Professor Sunstein provides an overview of behavioral
xiv Preface
economics, where the traditional approach of a rational Homo economicus is
challenged by our understanding of human psychology and human behav-
ior in the real world. It seems natural to presume that a nation’s economic
being depends on economic policy. It therefore took time for us to
well-
realize that many drivers of an economy lie outside economics, in social
norms, cultural mores, and psychology. Behavioral economics, the subject
of a recent World Development Report of the World Bank,3 sensitized econo-
mists to these important influences that lie outside the discipline but are
key determinants of development. Behavioral economics, including impor-
tant contributions by Cass Sunstein, alert us to the fact that human beings
are often irrational, and more importantly, that these irrationalities are
often systematic. Understanding them can enable us to promote develop-
ment more effectively.
Professor Sunstein’s chapter is followed by another one dealing with a
relatively new field of inquiry, the evolutionary prospects of economies
and societies. Although the origins of evolutionary game theory go back
to the early 1970s, the entry of this discipline into mainstream econom-
ics is more recent. One of the most prominent contributors to this field
of research is Jorgen Weibull. In chapter 9, he and Ingela Alger discuss the
role of morality and the evolutionary foundations of human motivation,
showing that unqualified selfishness may be good for the individual in an
immediate sense, but if acquired by all in a society, it sets that society on a
course toward extinction. Morality, in the sense of Kant, is evolutionarily
interest
stable. That is, if all of us are prepared to forgo a little bit of our self-
to uphold some of our collective interests in the Kantian sense, our society
will be more robust in terms of surviving natural selection. Even apart from
this reasoning, the ideas of evolution, once a preserve of biology, have now
come into economics in a big way. Chapter 9 summarizes some of the most
important ideas in this discipline for the wider community of economists
and students of social science.
One of the most important advances in modern development econom-
ics is the use of randomized control trials (RCTs) to get at causal explana-
tions of various policy interventions and alternative economic outcomes.
Did the election of women as leaders of village councils play a role in the
3. World Bank (2015).
Preface xv
better provision of local public goods in India? Did deworming help school-
children in Kenya attend school more regularly and do better in their stud-
ies? By bringing the method of RCTs from epidemiology to development
economics, we can now hope to answer such questions with a clarity that
we did not have earlier. The RCT has been a source of celebration, criticism,
and controversy, but as a method in the toolkit of development econom-
ics, its value is undeniable. Chapter 10, which closes the book, is by Esther
Duflo, written jointly with Abhijit Banerjee and Michael Kremer. Duflo’s
own research and publications played a critical role in the development of
this field of research. She gives a detailed account of the rise of the field, its
achievements, and some of its pitfalls.
Acknowledgments
We as editors thank a large group of individuals who helped carry out this
9,
megaproject, including the organization of the conference on June 8–
2016, at World Bank headquarters. Over and above the authors of the
chapters, who presented papers at the conference and whose work we com-
mented on above, there was a stellar cast of discussants, who read and com-
mented on those papers and whose comments are included in this volume.
Here is the list of discussants, in alphabetical order (to minimize discontent)
to whom we are extremely grateful: Larry Blume, Francesco Caselli, Shanta
Devarajan, James Foster, Varun Gauri, Xavier Gine, Gäel Giraud, Gita Gopi-
nath, Robert Hockett, Karla Hoff, Ravi Kanbur, Aart Kraay, Aslı Demirgüç-
Kunt, David McKenzie, Célestin Monga, Maurice Obstfeld, Hamid Rashid,
Martin Ravallion, Luis Servén, and Mike Toman.
The chairs of the conference sessions were Augusto Lopez Claros,
Makhtar Diop, Felipe Jaramillo, Ayhan Kose, Bill Maloney, Kyle Peters,
Martin Rama, Ana Revenga, and Augusto de la Torre. We are grateful to
them for conducting the sessions and also for their comments and ideas
during the session.
We also thank Gabriela Calderón for her help in organizing a superb
conference and the Development Economics Communication team for
their help with disseminating information about the conference. We are
also grateful to Gabriela Calderón and Trang Huyen Hoang for preparing
the manuscript for submission to the MIT Press, as well as our references
“police,” Woori Lee and Ruth Llovet Montañes.
xvi Preface
Finally, we take this opportunity to express our gratitude to our editor at
the MIT Press, Emily Taber, for her help and cooperation at every stage and
also her patience, as we crossed over some of our own deadlines in bringing
this large project to a close.
References
Arrow, Kenneth J., and Gerard Debreu. 1954. “Existence of an Equilibrium for a
290.
Competitive Economy.” Econometrica 22 (3): 265–
Pigou, Arthur C. 1920. The Economics of Welfare. London: MacMillan.
World Bank. 2015. World Development Report 2015: Mind, Society, and Behavior.
Washington DC: World Bank.
Introduction: The State of Economics, the State
of the World
Kaushik Basu
1776 and 1860
For the discipline of economics, and for the world at large, these are unusual
times. The shock and awe of the financial crisis that began in the United
from
States in 2008 and the series of economic fault lines it ripped open—
the sovereign debt crisis in the European Union to the massive slowdown
have led
in several emerging economies that we are currently witnessing—
to much soul searching.
The past nearly two and a half centuries, from Adam Smith’s The Wealth
of Nations (1776) to the flourishing of empirical research and big data in
current times, mark the astonishing rise of a discipline. From a broad,
descriptive, and speculative subject, economics has come to acquire a com-
mon methodological foundation, mathematical structure, and a growing
database. It has vastly enhanced our understanding of markets, exchange,
money, finance, and the drivers of economic development.
How did this come to be? Where is economics headed? Will it be up
to the diverse challenges of our times? Will global poverty be eradicated,
or will it be exacerbated under the strain of a deteriorating environment?
These are the questions we grappled with over the 2 days of the conference
that is the basis of this book. The conference brought together some of the
most prominent individuals who have, for better or for for worse (depend-
ing on your love or distaste for economics), played a role in making eco-
nomics what it is today.
This essay is based on the opening remarks made on June 8, 2016, at the conference
titled “The State of Economics, The State of the World,” at the World Bank, Wash-
ington, DC. I am grateful to Alaka Basu, Oliver Masetti, Claudia Paz Sepulveda, and
David Rosenblatt for comments and discussion.
2 The State of Economics, the State of the World
There have been achievements in economics from well before 1776 to
now. But for me, the transformational period of the discipline was the 100-
odd years, starting from the second half of the nineteenth century. If you
like birthdays, I have a date to propose to mark the birth of modern eco-
nomics: February 19, 1860.
Stanley Jevons wrote a celebrated letter to his brother on June 1, 1860,
saying that he had made a stunning discovery in the past few months that
explained the “value” of different goods and gave him insights into “the
true theory of Economy.” He told his brother that so thoroughgoing and
consistent was his theory that “I cannot now read other books on the sub-
ject without indignation” (Collison Black 1973, 410).
When exactly did he hit upon the idea? Historians of economic thought
have drawn our attention1 to a special entry in Jevons’s diary, on Febru-
ary 19, 1860: “At home all day and working chiefly at Economy, arriving I
suppose at a true comprehension of Value.” Birthdays for scientific break-
throughs are always questionable. But if we can have Mother’s Day, Val-
entine’s Day, Administrative Professional’s Day, I see no reason we cannot
have Modern Economics Day, and February 19 would be my pick.
Of course, thinkers were already laying the foundations for Jevons’s
breakthrough. Gossen had worked out quite a lot of this a good one or two
decades before Jevons. Cournot laid some of the substructure in 1838. And
the law of diminishing marginal utility and its significance were described
by Daniel Bernoulli as early as 1738, to solve the St. Petersburg paradox,
which had been discovered in 1713 by Nicolaus Bernoulli. (And, yes, it was
all in the family, Nicolaus being Daniel’s brother.)
It is also important to note that although Stanley Jevons (1871) was clearly
on to the main ideas of general equilibrium and value, he never quite got all
the way there. We needed Léon Walras (1877) to put up the main structure.
And for the full general equilibrium project to be completed, with the exis-
tence of equilibrium proved and its welfare properties spelled out, we needed
to wait another 75 years for the seminal contributions of Kenneth Arrow.
By the time John Hicks, Paul Samuelson, Ken Arrow, Gerard Debreu, Lionel
McKenzie, and others were doing their work,2 modern game theory had
out
been born. Over the next decades, the combination of a fully worked-
1. See La Nauze (1953).
2. See Hicks (1939), Samuelson (1947), Arrow and Debreu (1954), and McKenzie
(1959).
Introduction 3
general equilibrium system, game theory, and a little later, social choice,
ideas of asymmetric information and adverse selection, endogenous price
rigidities, theories of economic growth and development economics, and
the first understandings of the rudiments of monetary policy would trans-
form the landscape of economics.
Few activities in life are as innately joyous as the pursuit (and if one
is lucky, the discovery) of new ideas, the unearthing of patterns in the
abstract space of concepts and numbers or in the world of data and statis-
tics. Frontline researchers must have the space, like artists and composers,
to do what they do as an end in itself. The greatest benefits of research
product of this freedom. But here at the World Bank, our
are usually a by-
preoccupation is much more down to earth and is driven by policy needs.
Hence, what we wanted to take away from the conference was how we can
draw on the best of economics to promote development and sustained,
inclusive growth, and contribute to making the world a better place. The
World Bank’s research and data analyses have been enormously influential,
reaching the desktops of finance ministers and policymakers all over the
world; indeed, a special responsibility comes with this influence.
At the time of this writing, I have been chief economist of the World
Bank for nearly 4 years. This conference and the book are an opportunity
to share some of my concerns and questions with the distinguished gather-
ing at the conference and also with a wider readership. The hope is that the
conference and its proceedings (to wit, the present book) will strengthen
the World Bank’s mission of promoting development.
Because the World Bank’s engagement is primarily with development
economics, it may be worthwhile to point out that development econom-
ics, like economic theory, has had its moments of epiphany. Arthur Lewis
old question of
had been troubled by two problems. First, there was the age-
why industrial products, such as steel, were so much more expensive than
agricultural products. Second, why were some countries persistently poor,
while others were so rich?
In an autobiographical essay, Lewis (1980, 4) writes about his eureka
moment in 1952: “Walking down the road in Bangkok, it came to me sud-
denly that both problems have the same solution. Throw away the neoclas-
sical assumption that the quantity of labor is fixed. An unlimited supply of
labor will keep wages down, producing cheap coffee in the first case and
high profits in the second. The result is a dual national or world economy.”
This epiphany was the genesis of his classic paper on dual economies in the
4 The State of Economics, the State of the World
Manchester School (Lewis 1954), which would play a major role in his being
awarded the Nobel Prize in 19793 and in triggering research on develop-
ment economics.
Intuition and Causality
I turn now, more specifically, to the subject of development policy. For the
project of converting research to good policy, we need three ingredients:
data (and evidence), theory (and deductive reasoning), and intuition (and
common sense).
One of the great achievements of economics in recent decades has been
in the area of empirical analysis. We have good reason to celebrate the rise
of data and our ability to analyze data using different methods: from intel-
ligent bar charts, through simple regression analysis and structural models,
to randomized control trials. This recent success raises the hope of econom-
ics becoming a truly useful science (see Duflo and Kremer 2005; Banerjee
and Duflo 2011).
There is, however, a propensity among some economists to dismiss all
theory as esoteric.4 Among other dangers, we run the risk of making our dis-
cipline inefficient. Suppose we insisted that Pythagoras could only use empir-
ical methods. Would he ever have gotten to his famous theorem? The answer
angled triangles
is: He might have. If he had collected a large number of right-
and measured the squares on their sides, he might have hit on the conjecture
of the two smaller squares adding up to the one on the hypotenuse. But this
approach would be extremely inefficient. Moreover, there would be a lot of
debating and dissent. Some would charge him with using a biased sample of
angled triangles, all from the Mediterranean region. “Would it work in
right-
the Arctic, in the Southern Hemisphere?” they would query.
We must acknowledge that many truths can be discovered more effi-
ciently and more compellingly using pure reason. Further, there is a great
3. This idea, combined with the rise of modern growth theory (see Arrow 1962;
Lucas 1988; Romer 1994; Ray 1988; Aghion and Howitt 2009), has given us insights
into the development process and development policy that were unthinkable even
a few decades ago.
4. For one of the best discourses on the strengths and vulnerabilities of economic
theory, see Rubinstein (2006).
Introduction 5
deal of sloppiness in the way we reason about the use of evidence. For
headed practitioners will often tell you the following: “If we
instance, hard-
do not have any evidence about whether some policy X works, we must
not implement X.” (I was told exactly this fairly recently, in response to a
suggestion I made.)
Let me call this rule in quotes an “axiom.” To see that it is an unreasonable
axiom, observe that if we do not have any evidence about whether X works,
X works. But
then we also do not have any evidence about whether not-
X, the original axiom has to be flawed.
because we have to do either X or not-
For good policy, we need facts and evidence, but we also need deduction
and reasoning. We can go a step further and make a case for using math-
ematics. Although the use of mathematics can be overdone (as has happened
in economics), the immense achievements of Cournot (1838) and Walras
(1877), and of modern economics, would not have happened without it. This
is because mathematics is a disciplining device, even though it is demanding
and clearly not something that is applicable in all situations. As Krugman
(2016, 23), not being able to make up his mind whether a particular argu-
ment of Mervyn King (2016) was right, observes, “words alone can create an
illusion of logical coherence that dissipates when you try to do the math.”
The power of doing a model right, even if it is abstract and uses assump-
tions that may not be real, can be seen from general equilibrium. Take Gerard
Debreu’s (1959) classic The Theory of Value. This book is of great beauty, as
spare as poetry. In some ways, it is comparable to the work of Euclid, for it
brings together in a systematic way an amazing range of ideas. Euclid may
not have been as original as Pythagoras or Archimedes, but in bringing intel-
lectual order to a scattered discipline, he had few peers, and he served an
enormous role in the progress of knowledge. Likewise for Debreu’s slim book.
The pathbreaking general equilibrium model of Walras, Arrow, and
Debreu provided a template that sparked off some of the most original
notably those by Akerlof and Stiglitz—
works in microeconomic theory—
which have to do with modeling the functioning of markets under imper-
fect information.5 These works have greatly enhanced our understanding
of micromarkets; why markets fail; and why prices are often endogenously
rigid, resulting in credit markets with excess demand and labor markets
5. See Arrow (1963), Akerlof (1970), Stiglitz (1975), and Stiglitz and Weiss (1981).
6 The State of Economics, the State of the World
with excess supply. This research also has hopes of improving our macro-
economic analysis, because, as we know, Keynesian macroeconomic analy-
sis, like Arthur Lewis’s dual economy model, makes extensive use of price
rigidities, and neither Keynes nor Lewis had an explanation for these rigidi-
ties. Thanks to the work of Stiglitz and a few others, we now have a formal
understanding of open unemployment and credit markets that do not clear
despite the absence of exogenous restrictions on interest rate movements.
Along with these positive theories, we have seen the rise of normative
economics. Perched between analytical philosophy, mathematical logic,
and the social sciences, this achievement was remarkable. Major contribu-
tions were also made by Samuelson (1947), Bergson (1938), and others, but
the truly astonishing breakthrough was Ken Arrow’s (1951) slim book: Social
Choice and Individual Values. Arrow’s impossibility theorem became the bed-
rock of an enormous research agenda. The leading figure here was Amartya
Sen, whose work, straddling philosophy and economics, demonstrated that
it is possible to bring the finest traditions of theory and mathematical logic
old questions of ethics and normative principles (Sen 1970;
to bear on age-
see also Suzumura 1983). This work brought into the mainstream of rigor-
ous analysis such concepts as rights, which were widely talked about but
seldom subjected to careful scrutiny (Sen 1996). This body of work has been
important for the World Bank, because its mission goals have foundations
specific
in such concepts (World Bank 2015b) and also in related country-
research (Subramanian and Jayaraj 2016).
It is worth digressing for a moment to note that data and statistics belong
to a larger domain of inquiry, which has to do with description. The term
“descriptive social science” is often treated as a pejorative, which is unfor-
tunate. As Amartya Sen (1980) points out in a powerful essay, developing a
good description is not easy, and a huge amount of the progress of science
depends on description. Description, be it in words or data, entails choice.
Description is not regurgitating everything we see around us. We have to
pick what is vital and make that available to others. How we describe and
what we describe shape our understanding of the world. The “describer” is
therefore a pivotal agent.
It is important to be aware that description can take many forms. What
the anthropologist describes often does not take the form of numbers and
data. But the description of what he or she has seen and, more impor-
tantly, experienced is vital for our understanding of the world. The concept
Introduction 7
which we owe to Gilbert Ryle (1968) and Clifford
of “thick description”—
has vastly enhanced
Geertz (1973) and used by umpteen anthropologists—
our understanding of traditional and remote societies. It has enabled us
to intervene more effectively. At times this intervention has been for the
wrong reasons (for instance, to enable colonial domination), but it has also
helped carry the development agenda further by extending the reach of
modern medicine and education.
Historically, we have learned of the motivation and purpose of other
lives, which are distant from ours, by the ardor and work of anthropolo-
gists. These topics are very difficult to learn and comprehend by data and
statistics alone. Living with the subject and acquiring an intuitive under-
standing are often necessities. This knowledge has been put to good and
bad uses, to help the poor living in distant lands and in traditional societies,
and also to exploit people and spread imperialism and colonial control. For
good or for bad, the knowledge has been useful.
The absence of such knowledge can create major handicaps. Consider
terrorism. Because of the dangers associated with observers interacting with
terrorist groups, we do not have studies of the kind anthropologists have pro-
vided for remote societies, resulting in an insurmountable knowledge gap.
The skeptics, from Pyrrho to David Hume and Bertrand Russell, were
right: Neither fact nor deduction can take you all the way to the best policy
to implement. The reason is that causality, regardless of whether it is pres-
ent, can never be demonstrated. In the end, causality lies in the eyes of the
provoking observation on this comes
beholder. For me, the most thought-
from a tribesman from Nepal. The famous National Geographic photog-
rapher, Eric Valli, seeing the tall trees these tribesmen climbed to gather
honey, asked one of them whether they ever fell out of those trees. The
answer he received was: “Yes, you fall when your life is over.”6
Given the impossibility of discovering causality, for good policy, it is not
enough to have the facts; it is not enough to combine facts with theory. I am
convinced we need one more ingredient: common sense and what I have
elsewhere called “reasoned intuition” (Basu 2014).
Researchers refuse to admit it, but it is true that there is no escape from
the use of intuition, and the bulk of what we call “knowledge” that we
6. This quote, as well as the argument on causality, which is more intricate than may
appear from these brief remarks, are taken from Basu (2014, 458).
8 The State of Economics, the State of the World
acquire through life occurs casually, mainly by using common sense. It
would be a mistake to insist that all knowledge has to be rooted in scientific
method, such as controlled experiments. It is quite staggering to consider
the number of things a child learns through nonscientific methods.
As to why such knowledge, acquired through intuition and common
sense, may have value, we have to recognize that our intuitions are what
they are because of evolution. These methods have survived natural selec-
tion, and so their power must not be dismissed out of hand. Evolution has
shaped a lot of what we see in our economic life; this is widely acknowl-
edged, but our understanding of the interface between evolution and eco-
nomics, for which some foundations were laid by Maynard Smith and Price
quite some time ago (see Maynard Smith and Price 1973; Weibull 1995)
remains rudimentary. There is a foray into this topic in this book (see chap-
ter 9) in the context of morality and its origins (see also Alger and Weibull
2013). But it is arguable that such innate knowledge acquision applies to
many other domains. The way people commonly acquire knowledge may
not meet the test of scientific standards, but it cannot be dismissed out of
hand. At the same time, casual empiricism can lead to superstitions, which
we have to guard against. I have argued elsewhere (Basu 2014) that what we
need is “reasoned intuition,” that is, the use of intuition vetted by reason-
ing. This is not a surefire method, but it is the best we can do.
Data, theory, and intuition are the three ingredients for human knowl-
edge and progress. But even with all three in place, skepticism, as philos-
ophers through the ages have reminded us and as Keynes (1936) did in
chapter 12 of General Theory, must be a part of the thinking person’s mind-
set. One problem with scientists who lash out against superstition but do
not question scientific knowledge is the double standard. They fail to rec-
ognize that, when it comes to certainty about the future, scientific wisdom
is as much open to question as many other forms of knowledge.
Knowledge and Caveats
We are heading into uncharted territory and struggling with the world’s
economic problems. Recent problems include United Kingdom’s vote in
favor of exiting the European Union (I suspect this important issue will
persist for some time) and the decline in commodity prices (especially that
of oil), which is creating a lot of stress in commodity exporting nations and
Introduction 9
in corporations that have invested in this sector. Questions are being raised
about the readiness of the discipline of economics to address such issues.
The first thing to recognize, however, is not that economists misread or
underestimated these crises, but how these problems show that there is still
a lot about the economy that we do not know.
Experts in any discipline suffer from the disadvantage of not knowing
exactly what it is they do not know. Take, for instance, medicine. Given
how little we know about the human body and brain, when we consult a
doctor with health problems, in most cases the right answer for the doctor
to give is: “I have no idea.” But we seldom hear this. Doctors almost invari-
ably tell you what your problem is. What should warn you that when doc-
tors say they know what your ailment is, they in fact often do not is that,
even in the eighteenth century, well before the arrival of modern medi-
cine, doctors seldom said they had no idea what ailed the patient. This is
and doctors now
because doctors in the eighteenth century did not know—
what they did not and do not know. It is much the same
do not know—
with economists.
Among the areas of darkness that hamper development policy is our
inability to link the micro and the macro. Suppose a government under-
takes some intervention X in a thousand villages. X can be a conditional
cash transfer, an employment creation program, or provision of a fertilizer
subsidy. How do we evaluate the success of the program in reducing pov-
being of the people
erty? Typically, we do this by collecting data on the well-
in these villages. If we are fussy, we may use all kinds of controls, including
proper randomization. Suppose, through such a study, it is found that pov-
erty has indeed gone down in the villages where X was implemented. Does
this mean X is a good intervention? Not necessarily. Suppose the interven-
tion X in a village has the following effect. It raises food prices a little and
raises wages more. This will indeed lead to lower poverty in the village. But
because a rise in food prices typically cascades across the whole economy,
this intervention could mean that in other villages, which will only feel the
full rise in food prices and a negligible effect on wages, poverty will rise. So
it is entirely possible that the nationwide effect of the intervention will be
no effect on poverty or even an increase in poverty, though poverty falls in
the villages in which the interventions occurs.
These links between micro interventions and macro effects are poorly
understood. We need to invest much more in this kind of research if we are
10 The State of Economics, the State of the World
to succeed in battling nationwide and even global poverty and to combat
inequality.
theoretic areas, such as finance and the psychological
In other micro-
foundations of human behavior, economics has made great strides, as dis-
cussed in this book.7 But open questions still exist. In finance, it is increas-
ingly recognized that there is no such thing as an ideal regulation. This
is because financial products are amenable to endless innovation. Banks
and financial organizations will keep developing new products, just as the
pharmaceutical industry keeps discovering new drugs. And with each such
financial innovation, we may need to modify and make our regulatory
regime more sophisticated. Hence, this is one area where we have to reject
the language of optimal regulation, which has a static connotation, and to
create regulatory bodies that are flexible and ready themselves to innovate.
This effort is complicated by the fact that when selecting financial prod-
ucts, people are often not rational and instead give into emotions, hyper-
bolic discounting, and framing delusions, as pointed out repeatedly in the
recent behavioral economics literature.
One possibility is to label certain financial products as “prescription
goods” and create the equivalent of doctors in finance, who have to sign off
before a person is allowed to buy a financial product. We could, for instance,
decide to allow balloon mortgages, but before a consumer can commit to
one, he or she has to get a “finance doctor” to sign off on the financial
viability of taking on such a contract. This cannot be done by mechanically
following practices in medicine, but a case can be made for giving serious
thought to such an architecture.
The interface between economics and psychology, and, more specifi-
cally, behavioral economics, has witnessed great strides; and we at the
World Bank have tried recently to bring this progress to bear on the agenda
of development policy with our World Development Report on Mind, Society,
and Behavior (see World Bank 2015a). By drawing on evidence from labora-
tory experiments and field observations from around the world, behavioral
7. For an elegant example of how economic theory can be brought to bear on a com-
pelling idea in finance and financial crisis (namely, the phenomenon of infection,
which has been widely noted, whereby one economy, seemingly unconnected to
another, infects it with financial panic), see Morris and Shin (1998).
Introduction 11
economics teaches us a lot about how and where we should intervene.8
However, this discipline might risk becoming a catalog of findings. I call
this a risk because of a propensity to think of the findings as set in stone,
not realizing that they may be true in some societies at certain stages of
development and might differ with place and time.
What is also needed is an effort to marry these findings more effectively
with the concept of equilibrium (Akerlof and Shiller 2015). Then we would
be able to leverage these findings to get much more out of them and also be
able to predict better how the findings are likely to change from one society
to another and to evolve over time. To my mind, one of the great contribu-
tions of traditional economics is the idea of equilibrium, which has many
manifestations, from the general competitive equilibrium to Nash. We need
to broaden the description of individuals from the narrow Homo economicus
to that of more realistic individuals (with quirks, irrationalities, and social
norms) and to use the idea of equilibrium in conjunction with this more
realistic description.9 What makes this effort intellectually challenging is
that for most real phenomena, which seemingly rely on human irrational-
ity or adherence to social norms, it is possible, with analytical ingenuity, to
accurately model the same behavior using perfectly rational individuals.10
In the end, better modeling calls for the use of judgement and intuition
when deciding what assumptions we should rely on.
The World Bank has been increasingly engaged in this difficult area.
Given the current drift of global concerns, we do not have a choice. These
concerns naturally lead to another related field beyond the narrow confines
of economics, that is, institutions and governance.11 Our World Develop-
ment Report on Governance and the Law (see World Bank 2017) takes on this
8. See Kahneman (2000), Thaler and Sunstein (2008), and Hoff and Stiglitz (2016).
9. For a very interesting paper that that attempts this, see Hoff and Stiglitz (2016).
Earlier, Gintis (2009, chapter 10) provided an elegant model of bringing together
the idea of human sociality and economic equilibrium in a unified game- theoretic
discourse.
10. For an ingenuous exercise in this type of modeling, see Myerson (2004). Behav-
ior, which at first sight seems so obviously driven by an irrational adherence to
norms, can be explained as rational behavior in a more complex setting.
11. The importance of this field is stressed by Bourguignon (2015) in analyzing the
African experience. As he stresses, this analysis is much more than an academic exer-
cise. It is germane to the design of successful policy interventions.
12 The State of Economics, the State of the World
challenging task.12 One important area of policy making is the control
of corruption, a big task faced by those at the helm of policy. Traditional
economics treated an act of corruption (e.g., whether to pay a bribe to
get an illegal electricity connection) on par with any other purchasing
that is, as an exercise in narrow
decision (e.g., whether to buy an apple)—
benefit analysis (see Bardhan 1997; Mishra 2006). It is not surprising
cost-
that we have been so singularly unsuccessful in controlling corruption. To
understand this phenomenon, it is important to bring in psychology and
political institutions. Development policy cannot be built on economics
alone.13
Finally, one area in which we have knowledge gaps but not as much as
conservative commentators make out, is the connection between climate
change and development. If we proceed the way we have done thus far, it
is a journey headlong into disaster. This is unfortunate, because awareness
of the connection between environmental resources and economic devel-
opment came early, as evidenced in the works of Thomas Malthus, David
Ricardo, Knut Wicksell, and others, even though we have been tardy in
terms of action and policy. In recent times, the importance of this connec-
tion has been stressed by several authors, notably by Stern (2007, 2015).
12. The challenge of this task is captured well in the short essay by Green (2016),
which points to the necessity of delving into this arena if we want to do economic
policy right, and to how hard it is to do, because it ruffles feathers and is intellec-
tually such treacherous terrain. Academic research that addresses governance and
political institutions with the sharp scalpel of analysis is still relatively rare, but see
Dixit (2009) and Acemoglu and Robinson (2012).
13. Here I give the example of corruption to illustrate the need for multiple disci-
plines, but the need is quite ubiquitous in today’s world of strife and conflict. An
excellent example is the Middle East. It is difficult to explain what is happening there
purely in terms of economic indices, from gross domestic product through poverty to
various measures of inequality and polarization. As Devarajan and Mottaghi (2015)
argue, what is happening, in essence, is a breakdown of a social contract, which, like
plumbing, goes unnoticed when it functions well but is always important. One can
go further and look at areas that seem squarely situated in the domain of economic
problems, such as the subject of poverty and inequality mitigation, which is central
to the World Bank’s work. Is it enough to rely on market forces and natural economic
growth? Careful econometric studies of countries that have been most successful
in this, such as Brazil, show that we have to go beyond these phenomena. Ferreira,
Ravallion, and Leite (2010), for instance, find hard evidence that changing social
security practices and increasing social assistance expenditure by the federal govern-
ment was critical, and in fact happened because of the 1988 Constitution.
Introduction 13
Now with the Paris Agreement of 2015, there is a platform to relate what
we know on the subject with action on the ground, which is not easy,
country coordination. It is worth stressing
because it entails some cross-
here that this engagement should be viewed very much as part of shared
prosperity, because it entails intergenerational sharing of resources and
well-being.
Money and the Person of Influence
The previous section discussed some gaps in our knowledge. One big gap
is in the area of monetary policy. Although economics has made some dra-
matic breakthroughs in some practical areas (such as how to design auc-
tions and how to micromanage demand and supply in sectors), its grasp
of the impact of macroeconomic and especially monetary policy interven-
tions is rudimentary. It is true that we have learned to manage hyperinfla-
tion, and we can hope never to see again, at least in advanced economies
with sophisticated central banks, the kind of runaway inflation seen in, for
instance, Hungary in 1946 and Germany in 1923. But as the global financial
and growth crisis that began in 2008 continues unabated, and governments
and central banks flail at this with different policies, it is evident that large
gaps exist in our understanding of the impact of macroeconomic policies,
and the linkage between the financial and real worlds (Stiglitz 2011). This
is something I learned by fire, during my nearly 3 years as a policy maker in
India (from 2009 to 2012). Although monetary policy was not my charge,
it became clear during this time that much of our interventions were based
on imitating policies followed by central banks in advanced economies,
unmindful of the fact that their contexts differed.14
One reason for this deficiency is that we do not understand the func-
tioning and role of money in a market economy the way we understand,
for instance, the Walrasian general equilibrium system for real goods and
services. Money in general equilibrium was part of a big research agenda in
the 1980s, but that agenda has remained incomplete. One reason is that it
14. I discuss this in my recent book (Basu 2015), where I also argue for the need
to make more experimental policy interventions in emerging economies, which
would allow them to collect their own data and use these to develop their own, more
context-specific policies.
14 The State of Economics, the State of the World
is mathematically a very hard problem. But it must not be abandoned for
that reason. In the rush to solve the next morning’s problem, often these
deep questions take a back seat. But as the world struggles to cope with the
slowdown, and the widespread use of negative interest rates does not seem
to work (and in fact has a negative backlash from which no country is able
to individually break out of), it is important for economists to keep work-
ing on some of this fundamental research.15 If the full general equilibrium
from Jevons and Walras to Arrow and Debreu—
model took some 75 years—
and the study of money in equilibrium started in earnest in the 1970s and
1980s, we have little reason to abandon the problem as unsolvable.
To see the mystifying nature of money, one can look at a very differ-
the power of peddlers of influence. With the US presidential
ent problem—
election in the offing, there was a lot of writing about lobbying, influence
peddling, and corruption. In my youth in India, I remember talk about
“persons of influence,” referred to those days as “men of influence.” I recall
being baffled by one particular person and wondered why he was so well
off. He had no special skill, no resources. He was just the man of influence
(let me call him “M”). In those days, it took a wait of 6 years to get a phone
connection. If you needed it sooner, you could try calling M and requesting
his help. He would call up the relevant person in government; and more
often than not, the favor would be done. If someone needed to get a child
into a good school, she could ask M, and if M agreed, he would request the
school principal to make an exception and take in this kid out of turn. It
struck me much later what he was doing and I wrote it up as a model of
the man of influence (Basu 1986). M was a person with a mental ledger of
favors done. If i needed something from j, whom she did not know, she
could ask M to ask j. Then j would do the favor, not because j cared for i or
ever expected to need a special favor from i, but because j knew that some-
day he would need a favor from k and would need M to make a request of k.
It is M that no one wanted to offend, because M was a clearinghouse with a
memory. This is what made M a man of influence. In some sense, a person
of influence is like money or a blockchain. It is a record of information and
works only because everybody thinks it will work.
This description and even the model is straightforward enough. But its
integration into a full general equilibrium model is extremely hard and
15. Some of the fundamental questions in this area are raised in Calvo (1996).
Introduction 15
remains an open agenda, thereby handicapping policy makers greatly and
forcing them to rely more on intuition and guesswork than hopefully will
be necessary in the future.
Politics and Economics
When discussing development policy, I have been stressing the role of eco-
in brief, input from professional,
nomic theory and empirical economics—
scientific analysis. The lack of this input dooms many a developing econ-
omy. But it is not always easy to marry scientific analysis with the ground
realities of politics. Maybe because I moved so abruptly from academe to
policy making, I cannot be unmindful of the importance of the role of
how one engages with politics and politicians. When I moved from Cor-
nell University to the Indian government at the end of 2009, I quickly
became aware of the potential conflict between the prescription coming
from theoretical economics and political compulsions. One quickly learned
that when a politician tells an economist, “You are so good at theory,” it is
meant to be a devastating criticism.
I have recounted in Basu (2015) how, at one of my first meetings in my
new job with the prime minister and some of his advisers, I was discussing
how to control food inflation, which was then at double digits. I spoke at
some length on changing the manner in which food reserves are released
in India to get the maximum dampening effect on prices. I basically drew
some policy lessons from the logic of Cournot equilibrium. I was delighted
that my suggestion was accepted, which, I now believe, owes as much to
my not uttering the words “Cournot” or “equilibrium” as to Cournot’s
excellent theorizing.
One gets a fascinating glimpse of the interface between the world of eco-
nomic ideas and political compulsions in developing countries from Arthur
Lewis’s experience as chief economic adviser to the Ghanaian government.
He was invited to take this position by Kwame Nkrumah, the country’s first
prime minister and president. The United Nations and the United States
tried to block this appointment on the grounds that Lewis was “not very
sympathetic to the Bank [the International Bank for Reconstruction and
Development, commonly referred to as the World Bank]” (Tignor 2006,
147). There were also concerns, such as the one expressed by A. W. Snelling,
an official in the British government, that “Lewis is a socialist, but a moder-
ate one” (Tignor 2006, 148).
16 The State of Economics, the State of the World
Lewis’s tenure began extremely well, with Nkrumah personally excited at
the prospect of Lewis steering the Ghanaian economy to a takeoff. On taking
Year
office, Lewis plunged into work, especially related to the second Five-
Plan, with widespread support from others in government. But soon Lewis’s
idea of what constitutes good economics and Nkrumah’s insistence on politi-
cal compulsions came into conflict. Seemingly small differences of opinion—
for instance, whether to spray cocoa trees that had been attacked by capsid
became the
beetles (pardon me for having forgotten who took which side)—
cover for deeper conflict: the professional economist’s insistence on good
economics and the politician’s stubbornness about what is politically good.
Lewis left office at the end of 1958, with Nkrumah’s letter, gracious but
recognizing that they could not work together, in his pocket: “The advice
you have given me, sound though it may be, is essentially from the economic
point of view, and I have told you on many occasions, that I cannot always
follow this advice as I am a politician and must gamble on the future.”16
Interests and Ideas
Some months after I moved from academe to the Indian government, a
reporter asked me: What was the one thing that I had learned in this transi-
tion? Unusually for a question of this kind, I had an answer. The reader may
recall Keynes’s beautiful observation on the power of ideas, which ended
with the following: “I am sure that the power of vested interests is vastly
exaggerated compared with the gradual encroachment of ideas” (Keynes
1936, 283–284).
As an academic, I loved the observation but did not believe in it, view-
serving remark of a professor. It was only after I joined the
ing it as the self-
Indian government and sat in interminable meetings with ministers and
bureaucrats that I came to believe in Keynes’s observation.
Ideas play an unbelievably important role, and so those in the business
of ideas have a special responsibility. As a consequence, I view this confer-
ence and this book not just as an intellectual contribution but as a critical
ingredient for the work that is meant to be done in an organization such as
the World Bank.
16. Nkrumah to Lewis, December 18, 1958, quoted in Tignor (2006, 173).
Introduction 17
References
Acemoglu, Daron, and James A. Robinson. 2012. Why Nations Fail: The Origins of
Power, Prosperity, and Poverty. New York: Crown Publishing.
Aghion, Philippe, and Peter W. Howitt. 2009. The Economics of Growth. Cambridge,
MA: MIT Press.
Akerlof, George A. 1970. “The Market for Lemons: Quality Uncertainty and the
500.
Market Mechanism.” Quarterly Journal of Economics 84 (3): 488–
Akerlof, George A., and Robert J. Shiller. 2015. Phishing for Phools: The Economics of
Manipulation and Deception. Princeton, NJ: Princeton University Press.
Alger, Ingela, and Jörgen W. Weibull. 2013. “Homo Moralis—
Preference Evolution
under Incomplete Information and Assortative Matching.” Econometrica 81 (6):
2269–2302.
Arrow, Kenneth J. 1951. Social Choice and Individual Values. New York: Wiley.
Arrow, Kenneth J. 1962. “The Economic Implications of Learning by Doing.” Review
173.
of Economic Studies 29 (3): 155–
Arrow, Kenneth J. 1963. “Uncertainty and the Welfare Economics of Medical Care.”
973.
American Economic Review 53 (5): 941–
Arrow, Kenneth J., and Gerard Debreu. 1954. “Existence of an Equilibrium for a
290.
Competitive Economy.” Econometrica 22 (3): 265–
Banerjee, Abhijit V., and Esther Duflo. 2011. Poor Economics: A Radical Rethinking of
the Way to Fight Global Poverty. New York: Public Affairs.
Bardhan, Pranab. 1997. “Corruption and Development: A Review of Issues.” Journal
1346.
of Economic Literature 35 (3): 1320–
282.
Basu, Kaushik. 1986. “One Kind of Power.” Oxford Economic Papers 38 (2): 259–
Basu, Kaushik. 2014. “Randomisation, Causality and the Role of Reasoned Intu-
472.
ition.” Oxford Development Studies 42 (4): 455–
Basu, Kaushik. 2015. An Economist in the Real World: The Art of Policymaking in India.
Cambridge: MIT Press.
Bergson, Abram. 1938. “A Reformulation of Certain Aspects of Welfare Economics.”
334.
Quarterly Journal of Economics 52 (2): 310–
Bourguignon, François. 2015. “Thoughts on Development: The African Experience.”
In The Oxford Handbook of Africa and Economics: Volume I, Context and Concepts,
edited by Célestin Monga and Justin Yifu Lin, 247–
270. Oxford: Oxford University
Press.
18 The State of Economics, the State of the World
Calvo, Guillermo A. 1996. Money, Exchange Rates, and Output. Cambridge, MA: MIT
Press.
Collison Black, R. D. 1973. Papers and Correspondence of William Stanley Jevons,
Volume II. London: Macmillan.
Cournot, Antoine. 1838. Recherches sur les Principes Mathématiques de la Théorie des
Richesses. Paris: L. Hachette.
Debreu, Gerard. 1959. The Theory of Value: An Axiomatic Analysis of Economic Equilib-
rium. New Haven, CT: Yale University Press.
Devarajan, Shantayanan, and Lili Mottaghi. 2015. “Towards a New Social Contract.”
Middle East and North Africa Monitor, April. Washington, DC: World Bank.
Dixit, Avinash K. 2009. “Governance Institutions and Economic Activity.” American
24.
Economic Review 99 (1): 5–
Duflo, Esther, and Michael Kremer. 2005. “Use of Randomization in the Evalua-
tion of Development Effectiveness.” In Evaluating Development Effectiveness, edited
by George K. Pitman, Osvaldo N. Feinstein, and Gregory K. Ingram, 205– 231. New
Brunswick, NJ: Transaction Publishers.
Ferreira, Francisco H. G., Martin Ravallion, and Phillipe G. Leite. 2010. “Poverty
Reduction without Economic Growth? Explaining Brazil’s Poverty Dynamics, 1985–
36.
2004.” Journal of Development Economics 93 (1): 20–
Geertz, Clifford. 1973. The Interpretation of Cultures. New York: Basic Books.
Gintis, Herbert. 2009. The Bounds of Reason: Game Theory and the Unification of the
Behavioral Sciences. Princeton, NJ: Princeton University Press.
Green, Duncan. 2016. “The World Bank Is Having a Big Internal Debate about
Power and Governance. Here’s Why It Matters.” Oxfam Blogs, July 26. https:// oxfam
blogs.org/fp2p/the-world-bank-is-having-a-big-internal-debate-about-power-and
-governance-heres-why-it-matters/.
Hicks, John R. 1939. Value and Capital: An Inquiry into Some Fundamentals and Prin-
ciples of Economics. Oxford: Oxford University Press.
Hoff, Karla, and Joseph E. Stiglitz. 2016. “Striving for Balance in Economics: Toward
a Theory of Social Determination of Behavior.” Journal of Economic Behavior and
Organization 126 (June): 25–57.
Jevons, William S. 1871. The Theory of Political Economy. London: Macmillan.
Kahneman, Daniel. 2000. Choices, Values, and Frames. Cambridge: Cambridge Uni-
versity Press.
Keynes, John Maynard. 1936. The General Theory of Employment, Interest, and Money.
London: Macmillan.
Introduction 19
King, Mervyn. 2016. The End of Alchemy: Money, Banking, and the Future of the Global
Economy. New York: W. W. Norton & Company.
Krugman, Paul. 2016. “Money: The Brave New Uncertainty of Mervyn King.” Review
of The End of Alchemy: Money, Banking, and the Future of the Global Economy, by
23.
Mervyn King, New York Review of Books, July 14, 21–
La Nauze, John A. 1953. “The Conception of Jevon’s Utility Theory.” Economica
358.
20 (80): 356–
Lewis, Arthur W. 1954. “Economic Development with Unlimited Supplies of
191.
Labour.” The Manchester School 22 (2): 139–
4.
Lewis, Arthur W. 1980. “Biographical Note.” Social and Economic Studies 29 (4): 1–
Lucas, Robert E. 1988. “On the Mechanics of Economic Development.” Journal of
42.
Monetary Economics 22 (1): 3–
Maynard Smith, J., and George R. Price. 1973. “The Logic of Animal Conflict.”
18.
Nature 246 (5427): 15–
McKenzie, Lionel W. 1959. “On the Existence of General Equilibrium for a Competi-
71.
tive Market.” Econometrica 27 (1): 54–
Mishra, Ajit. 2006. “Corruption, Hierarchies and Bureaucratic Structures.” In Inter-
national Handbook on the Economics of Corruption, edited by Susan Rose-Ackerman,
189–215. Cheltenham, UK: Edward Elgar.
Morris, Stephen, and Hyun Song Shin. 1998. “Unique Equilibrium in a Model of
Fulfilling Currency Attacks.” American Economic Review 88 (3): 587–
Self- 597.
Myerson, Roger B. 2004. “Justice, Institutions, and Multiple Equilibria.” Chicago
107.
Journal of International Law 5 (1): 91–
Ray, Debraj. 1988. Development Economics. Princeton, NJ: Princeton University Press.
Romer, Paul M. 1994. “The Origins of Endogenous Growth.” Journal of Economic
22.
Perspectives 8 (1): 3–
Rubinstein, Ariel. 2006. “Dilemmas of an Economist Theorist.” Econometrica 74 (4):
865–883.
Ryle, Gilbert. 1968. The Thinking of Thoughts. Saskatoon: University of Saskatchewan.
Samuelson, Paul A. 1947. Foundations of Economic Analysis. Cambridge, MA: Harvard
University Press.
Sen, Amartya K. 1970. Collective Choice and Social Welfare. San Francisco: Holden Day.
Sen, Amartya K. 1980. “Description as Choice.” Oxford Economic Papers 32 (3):
353–369.
20 The State of Economics, the State of the World
Sen, Amartya K. 1996. “Rights: Formulation and Consequences.” Analyse & Kritik
170.
18 (1): 153–
Smith, Adam. 1776. An Inquiry into the Wealth of Nations. London: Strahan and Cadell.
Stern, Nicholas H. 2007. The Economics of Climate Change: The Stern Review. Cam-
bridge: Cambridge University Press.
Stern, Nicholas H. 2015. Why Are We Waiting? The Logic, Urgency, and Promise of
Tackling Climate Change. Cambridge, MA: MIT Press.
Stiglitz, Joseph E. 1975. “The Theory of ‘Screening,’ Education, and the Distribution
of Income.” American Economic Review 65 (3): 283– 300.
Stiglitz, Joseph E. 2011. “Rethinking Macroeconomics: What Failed, and How to
645.
Repair It.” Journal of the European Economic Association 9 (4): 591–
Stiglitz, Joseph E., and Andrew Weiss. 1981. “Credit Rationing in Markets with
410.
Imperfect Information.” American Economic Review 71 (3): 393–
Subramanian, Sreenivasan, and Dhairiyarayar Jayaraj. 2016. “The Quintile Income
metric Poverty, and Disequalising Growth in India: 1983 to 2011–
Statistic, Money-
12.” Economic & Political Weekly 51 (5): 73.
Suzumura, Kotaro. 1983. Rational Choice, Collective Decisions, and Social Welfare.
Cambridge: Cambridge University Press.
Thaler, Richard H., and Cass R. Sunstein. 2008. Nudge: Improving Decisions about
Health,Wealth, and Happiness. New Haven, CT: Yale University Press.
Tignor, Robert L. 2006. W. Arthur Lewis and the Birth of Development Economics. Princ-
eton, NJ: Princeton University Press.
Walras, Léon. 1877. Éléments d’économie Politique Pure ou Théorie de la Richesse Sociale.
Lausanne: Corbaz.
Weibull, Jörgen W. 1995. Evolutionary Game Theory. Cambridge: MIT Press.
World Bank. 2015a. World Development Report 2015: Mind, Society, and Behavior.
Washington, DC: World Bank.
World Bank. 2015b. A Measured Approach to Ending Poverty and Boosting Shared Pros-
perity: Concepts, Data and the Twin Goals. Policy Research Report. Washington, DC:
World Bank.
World Bank. 2017. World Development Report 2017: Governance and the Law. Wash-
ington, DC: World Bank.
I Foundations
1 Equilibrium, Welfare, and Information
Kenneth Arrow
Kenneth Arrow passed away on February 21, 2017, before he could complete this
paper, which is based on his lecture at the World Bank conference. In a phone con-
versation on February 17 with one of us, he said he expected to complete the manu-
script within a month, but that was not to be. We are immensely grateful to Larry
Summers, who worked on the transcript of Ken Arrow’s lecture, editing it lightly,
and made this publication possible. We, as editors of the volume, have subsequently
added some minor edits. It was our conscious decision to do minimal work on Ken
Arrow’s transcript, even at the risk of the text reading somewhat colloquially. As a
last statement from him, we expect this paper to be an important document and
were keen to maintain the texture of his voice. It is also clear that during the lecture,
he ran out of time and so in some sense, this paper is not complete. That must have
been the reason that Ken Arrow was keen to work on it before submitting it for pub-
lication. We do not have that choice now. But, as editors, we expect that his fascinat-
ing reflections on how modern economics came to be what it is, and his assessment
of the weaknesses and strengths of modern economics, as well as his views on various
historical figures in economics, will be of wide interest.
—Editors
I was asked to talk today about equilibrium and welfare. The word “infor-
mation” was not in my suggested title, but as I shall argue, issues regarding
information are fundamental to understanding the problem. I won’t go
into technical questions of existence theorems. What I really want to do is
to remark on what exactly the point of equilibrium theory is. What ques-
tion are we asking? How does it contribute to our economic knowledge, to
our understanding of the economy? Inevitably, given the many aspects of
these questions, my remarks will be a bit scattershot.
One of the questions is: Why do people talk in equilibrium terms? What
is the purpose of relying on the notion of equilibrium? Well, knowing about
24 Kenneth Arrow
the economy is a little different from knowing about astronomy, because
it’s part of our daily life. Astronomy is something you have to study. You
have to stop and look at the stars. You have to watch what’s going on.
Whereas we are part of the economy.
It reminds me of the story of an astronomer who used to take summer
vacations hiking. He went to the Pyrenees, France, and ran across a shep-
herd. They decided to walk together for a while and have dinner together.
The astronomer was trying to explain what it was he did. He pointed at
the stars and said, “Well tomorrow they’re going to be in this different
position.” The shepherd listened. “Marvelous,” he said, “I see the point.
Since I follow my sheep and I know where they are, I know if I’m missing
one, he probably went down that valley. So, I can see if you spend enough
time, you’ll begin to know where the stars are. But the one thing I can’t
understand is how do you know their names?” The story captures some of
what we think about when we think of the difference in the positions of
astronomers and economists.
We are part of the economy. For us, the economy is not like the stars are
to the astronomer. The economy is a part of our everyday life; we observe it
from the perspective of a participant. This creates advantages of proximity.
But there is the disadvantage that we are too close in many ways. So, we are
likely to see only one aspect, and even that aspect we do not see in a very
unbiased fashion.
One thing, however, every day observation tells you is that somehow,
I’m provided goods; I don’t really worry that they won’t be there. They’re
usually there when I want to buy them. My house is there, rented or what-
ever. When I go to the store, there’s butter. Or, if you’re up to date, there is
some healthier kind of spread for you. But whatever it is, it’s there.
Early History
Goods and services are available in a straightforward way. I may look at the
price I have to pay, but that’s all I ever have to know. I don’t know how they
make this stuff. I don’t know where it comes from. This aspect of economic
life goes back a long time. In the great days of Athens, the most traveled
and most knowledgeable person about the world was Herodotus. And when
he was writing his history of the Persian wars, he actually went all around
the known world or the eastern Mediterranean, as we’d look at it today. He
Equilibrium, Welfare, and Information 25
writes on the subject of bronze. You make bronze by mixing copper and
tin. Well, copper comes from a lot of places, but tin comes from very few.
In fact, if you look at what we know about the ancient world, tin either
came from Iran or it came from Cornwall. Cornwall is a long way from the
eastern Mediterranean.
Tin from Cornwall, as we now know, was brought to what we now call
Marseille. The Gaelic merchants rafted it down to Rome and sold it. The
at least
Greeks had no idea where it came from. They didn’t even know—
they didn’t even know there was an island now
at the time of Herodotus—
called “Britain.” They didn’t know it existed. All they knew was they paid
their price to the Gaelic merchants and bought their tin, and that was the
tin that was used for making bronzeware. And of course, the modern world
has these transactions multiplied n-fold.
So, we see a relatively smooth operating mechanism. We see it’s regulated
by prices, and prices, for the most part, aren’t arbitrary. Firms, when they
most of the time—
sell things, don’t make 500% profit. They make— some
normal level of profit. So there seem to be some rules, and it’s these obser-
vations that motivate the development of economic theory dating from
the time of Adam Smith or even earlier. In fact, some people ascribe quite a
bit of the development of economic theory to the medieval commentators
who were concerned with the concept of profit and worried about excessive
profit. A vast literature seeks to interpret Smith, but it was this mechanism
and the “normal” level of prices that he had in mind when he famously
spoke of the invisible hand.
This leads naturally to the question of how prices affect behavior, a topic
that really did not come up at the time of Adam Smith or immediately
after. But one thing that was already stressed in Smith, and I suppose, some
of his predecessors, was the importance of competition. The idea that you
really cannot make supernormal profits because somebody will see a profit
opportunity. Now they didn’t spell out how this works. Presumably, if you
have high profits, other people enter, and of course, other people can cut
the price a little bit to take the trade. The implication in Smith is that it’s
more about entry than about firms explicitly moving prices. So, a demand
function must be implicit in the story. Yet you have no explicit notion of a
demand function in Smith, his immediate successors, or Ricardo.
Jevons era that there’s a
It was implicit and became explicit in the post-
circular flow element. Somehow, there are primary factors that enter into
26 Kenneth Arrow
production. Production then goes on, and the goods are delivered, and they
are bought by other producers or consumers. So the prices paid for the
primary factors are the purchasing power. They ultimately determine the
demand functions.
Now, it was the production side rather than the consumption side that
was most emphasized in the Classical period. Returns to scale played and
continue to play a major role in equilibrium theory. One natural assump-
tion is that returns to scale are constant, and so firms can enter an industry
at any scale with equal efficiency. But that poses a constraint, because if the
price of the product is a little too high compared to the prices of the inputs,
then with constant returns to scale, it pays to increase your scale of opera-
tions indefinitely. The question is then: What is it that restricts prices and
the levels of output?
And so the demand function was invented. Cournot certainly uses it
and indeed was an inventor of it. As an observation about how economic
science developed, it is noteworthy that Cournot published his book in
1838, yet the first known review is somewhere around 1877. It was com-
pletely ignored, and it was reviewed because Walras’s book came out, and
people began to go back. And Walras does pay some credit to Cournot, but
Cournot, by this time a rather old man, going blind, was very bitter that
he did not get the credit he deserved. And there was a very famous review
by an astronomer named Bertrand, which is where the concept of Bertrand
Competition is introduced.
But there was another introduction of demand functions besides in
Cournot, and that is in John Stuart Mill. One of Ricardo’s greatest innova-
tions was the idea of comparative advantage as a determining factor in
foreign trade. But without demand functions, you don’t really have an
explanation of quantities, you have theories about prices. So, Ricardo was
taking the prices as cost driven and therefore given. There are a lot of ambi-
guities in that, which I won’t go into now, but that’s the way he saw it. Mill
wanted to know something about quantities. So, he produced the idea of
demand curves. For example, Germany had a demand curve for English cot-
ton. And England had a demand curve for German linen. I think that was
the example he gave. This was Mills’s first paper and probably one of the most
brilliant things he wrote.
The next step in the development of equilibrium theory was the attempt
to provide foundations for thinking about the idea of profit. One of the
Equilibrium, Welfare, and Information 27
questions you get into is: Why are there profits at all? Why aren’t there zero
driven thing, but in the simplest economic
profits? Presumably, it’s a cost-
model, there is just one primary factor, labor, and then everything essen-
tially is priced based on how much labor is embodied in it. That doesn’t
give you any profits at all. This is what Marx, of course, took up. The rates
of profit are equal, but why do they have to equal zero? Nassau Senior,
who was a professor of political economy at Oxford at the time, said, “well,
there’s a cost to waiting.” That’s a subjective cost. That’s not a cost in any
literal sense. If goods are produced, they take time. I’m going to come back
to that as one of my main themes. There were also important contributions
by Gossen, Jevons, and Menger that clarified these matters further.
Externalities
So general equilibrium theory seems to have something to say about a good
part of the economy. Does it say everything? Well, no. We’re now accus-
tomed, I’m sure the World Bank especially, to talk about externalities. We
find that the markets somehow don’t work properly.
And that realization took quite a while. Although you see it recognized:
Walras, for example, has some statements that are pretty clear, not in his
book but in some of his essays, on the subject. Jules Dupuit in 1844 was
concerned with some ideas along these lines: Why the criteria for public
works? When should you build a road? When should you build a railroad?
How do you price railroads? And so forth. He was an inspector of bridges
and highways for the French government.
It was really quite a bit later that Pigou gave us a really clear statement
on externalities. But Pigou’s original formulation was pretty faulty, and it
I don’t know how many have heard of
was reviewed by an economist—
Allyn Young. Allyn Young wrote a book review of the first edition
him—
first called Wealth and Welfare, and the later edi-
of Pigou’s famous work—
tions called The Economics of Welfare. Pigou didn’t get it quite right, but in
the review, Young explained very clearly and correctly what an externality
was. And later there was, in the 1930s (the one I learned it from) a paper
by Jacob Viner, distinguishing pecuniary from technological externalities.
Technological externalities are the ones we think of as the welfare implica-
tions. I don’t want to go into that, because there’s hardly any advanced
country with less than 30 percent of its national income going through
28 Kenneth Arrow
the government. Those are the externalities we attempt to take care of, but
they are not the only ones. Externalities, public goods, whatever you want
to call it. I didn’t want to elaborate, except to mention it now; I’ll come
back to it later.
General equilibrium is useful here. It doesn’t explain the externalities,
it doesn’t explain what’s done to meet the externalities, but it does essen-
tially, at least for many economists, have some effect on real life. When
analyzing policies, we ask: What would general equilibrium say if it were
operative? And that’s the criterion we have. In almost all our analysis, policy
analysis, in situations where externalities govern, we ask: What would gen-
eral equilibrium say, if it even were applicable? And that is in a way the
main theme I want to present in the end. Of course, there’s another aspect,
namely, the failure of effective demand. When I was a graduate student, an
infinite period of time ago, we’d talk about business cycles. That was the big
at least around places like the National Bureau of Economic
macro issue—
Research. I personally took macro from Arthur Frank Burns in the 1940s.
The Influence on Early Econometric Models
The idea of pursuing systematic empirical work (not just collecting numbers
the econometric movement) is the product
but putting them into models—
of the creation of the Econometric Society around 1932. It was kind of a
movement, perhaps a little more European than American, but interna-
tional. One of its first examples in practice was a business cycle model of
the Netherlands by Jan Tinbergen, who subsequently led a much bigger
study sponsored by the League of Nations in Geneva. One thing that Tin-
bergen picked up from general equilibrium theory is the idea of a complete
system. If you’re going to forecast the future, you’ve got to have a complete
system. Or if you’re going to ask what the effect is of a policy, you have to
have a complete system. And we see today at least one tendency is to essen-
tially take a general equilibrium system, say, the prices don’t immediately
move in a right direction (they’re sticky).
So now we have I guess what you would call the “New Keynesian” models.
I don’t know if they do any better, but anyway, they’re complete systems.
And they deal with motivation as to individual relations from the same
basis but put in layers trying to say it’s costly to change your prices all the
time, or something along these lines.
Equilibrium, Welfare, and Information 29
Goods as Complements and Substitutes
A lot of the early literature on the production side assumed fixed coeffi-
cients. In other words, to produce good A, you just need so much of good B.
So you can have intermediate goods, but ultimately, directly or indirectly,
you’re drawing on the primary factors. And the idea that you’re going to
have substitution in alternative kinds of production was elaborated by John
not in his
Bates Clark in the late nineteenth century. Walras in his work—
has production functions. What Walras
first edition but in his later editions—
introduced really was the idea that (and he did this more elaborately, I
think, than Jevons did) the demand for one commodity might depend not
only on the price of that commodity but also on the prices of other com-
modities. Now once you say it’s an allocation problem (and this is certainly
there in Jevons), the idea of demand then becomes more complex, and
we have the standard notion that these commodities are in some sense
substitutes for each other. The fact that they are all competing for a limited
purchasing power means that, in some sense, substitution is bigger than
complementarity. But complementarity is still there: The price of butter
may affect the demand for bread. Once you bring in production functions,
you have a similar idea in production. So the idea that something that hap-
pens in one part of the system can then work its way through and affect
seemingly remote parts of the system is the big lesson to be learned from
general equilibrium. If you think of someone like Alfred Marshall, he clearly
saw this. In fact, his initial review of Jevons wasn’t terribly friendly. He was
angry at Jevons (as he himself said in his memoirs), because Jevons was so
contemptuous of Ricardo. Marshall said in his memoirs that he would write
very angry comments, then cut them out, but they would “reappear” again.
This was a very interesting discussion of the subconscious!
“Complementary Slackness”
Let me make two additional points. The first is an issue that sounds a little
technical, but it really is not. This is what mainly drove the discussion on
existence, which began in the 1930s and was completed in the 1960s. It’s
what the people who have a linear programming background would say:
“complementary slackness.” Menger made this observation. There are some
goods that are free, but they are free only because they are very abundant.
30 Kenneth Arrow
In other words, if they were not so abundant, they wouldn’t be free. What
are the examples? Air is free. In many parts of the world, water is free. In a
region with a lot of rain, water is free: I mean rainwater for agriculture. Of
course, water for drinking has got to be processed. It’s not the water that
is scarce, it is the processing. So the idea that a good is free or not depends
on economic circumstances. Well, this means that supply is not necessarily
equal to demand. Of course, supply can’t be less than demand in equilib-
rium. You can’t meet the demand then. But supply could be greater than
demand, and then the price would be zero. That’s recognized by Menger.
What happened was that several German authors (and two in particu-
lar, Hans Neisser and Heinrich von Stackelberg) in the 1930s had different
I won’t try to reproduce them now—
arguments— as to why the equations
of general equilibrium could be inconsistent. Actually, even though the
arguments were very easy, it would still take a few minutes, and I’m told I
have less than that! A private banker named Karl Schlesinger fled Hungary,
which was then under a communist threat, to Vienna and set himself up
as a private banker there, but he kept his interest in economics as an ama-
teur. (He had earlier received a PhD in economics.) Schlesinger pursued
the existence controversy and grasped the idea that the existence problem
was simply not recognizing complementary slackness. It was insisting that
supply equals demand when you might have supply greater than demand.
Well, he was no mathematician. So he went to Oskar Morgenstern, who
was running a business cycle research institute, financed by Rockefeller. Mor-
genstern had hired a graduate student in mathematics to do some work—
a fellow named Abraham Wald. Wald was
mainly some statistical work—
Romanian. He was actually born in Hungary, but the boundary had been
moved, so he was now a Romanian after World War I. And so there’s Wald,
who, using Schlesinger’s insight about the importance of complementary
slackness, came up with a proof of existence. The assumptions were absurdly
strong. It clearly left an open problem, and I won’t go into the history of that.
The Essential Role of Time
Now, though time is running short, let me turn briefly to the second issue,
that is, the big question that comes up, sort of right at the beginning—
but is usually skated over: Production takes time. It’s not
even in Smith—
for nothing that the word “capitalism” starts with “capital,” which means
Equilibrium, Welfare, and Information 31
production taking time. Well, it can take time in an indirect form. You
buy durable goods, like your plant and your equipment, which last and are
gradually used up in the process. So, one way or another, literally it just
or it may use machines that are durable and so are used
simply takes time—
over time. That means, if I look at a production process, to do it properly,
you put in goods at time zero, you put in more goods at time one, and the
good comes out at time two, or some such process. So, a production process
involves not only different goods but different goods at different times. So,
we can say, okay, no problem, we’ll just think of the same good at different
times as different goods.
The first person, as far as I know, who made this simple observation was
Eric Lindahl, a Swedish economist. It was picked up by Hicks. I got it from
Hicks. To the young theorists of my generation, Hicks was god. His book,
Value and Capital, was the most important thing in the world.
Prior to Hicks, the problem was that you read all this discussion about
capital theory by Frank Knight and other things like that, and it was all
mystical. You didn’t know what they were talking about. Pigou was a little
bit clearer, but he confined himself to simple questions. Hayek was impen-
etrable. But when you read Hicks and then went back to Hayek, you could
see that’s what Hayek was saying. I would never have understood Hayek.
I did read Hayek, Theory of Capital. It was incomprehensible. But as I say,
when you read Hicks, then it’s “oh, now I understand Hayek.” And I think
part of it is that Hicks got something from Hayek. He gives credit in the
footnotes but in a very general sort of way.
So, to this question of time. Now, when Gerard (Debreu) and I wrote, for
example, our proofs weren’t really any different from McKenzie’s, or any-
we carried the Walrasian program
thing like that, but I think we set out—
more thoroughly than anybody else did. That was the advantage of
out—
what we did. And so, we modernized it. We had utility functions, we had
preference orderings, we recognized the ordinalist revolution, things of
that kind, and we stated the need for concavity. It was a modernized ver-
sion of Walras. And we wrote just automatically, but we both thought the
same way without even discussing it. We treated goods at different times as
though they were just different commodities.
But what does that mean? It means we’re talking about a world in which
there are markets for everything. In particular, a market for goods tomor-
row and goods 10 years from now and 20 years from now. Well, you could
32 Kenneth Arrow
wave away a little bit of that, but you need goods markets for everything.
Look at the world. What do we see? There are goods for things tomorrow.
Agricultural goods, minerals, that’s about it. You can’t typically buy a car
in the future. I mean, obviously, if I’m setting up an automobile plant, it’d
be very nice to sell forward the car, a futures market, and credit. Well, the
problem is that I don’t know what the car is going to be like. I do know it’s
going to be different. Something’s going to happen. Maybe nothing impor-
tant, maybe just, you know, different styles or something trivial. But maybe
it will be significantly better in fuel economy or safety or some other way
that is important. So we have this problem. And that’s where general equi-
librium runs into limits. Somehow you can’t carry through the program.
And Hicks knew this, and he said, you have expectations of prices. But he’s
not very good at explaining how you form the expectations.
I’m sorry: I’ll wrap up in a minute.
Expectations and the Role of Information
There had been a literature, in the nascent econometric movement, about
price expectations. What people were really showing was that price expecta-
tions might give rise to trouble. And this is static expectations. Let’s say the
price tomorrow is going to be the same as the price today. And then they
hog” cycle. Well, similar versions of this is when you
had this famous “corn-
plant your crop, you look at the price prevailing and say that’s the price I’m
going to sell it for. In fact, the result is, let’s say, if the price is high, today
you plant a lot, but then the resulting effect is that the price is low tomor-
row. So you can wind up with a cyclical movement in people’s expectations;
of course, they’re already being dashed all the time, and people began to
develop more and more sophisticated kinds of expectations. But this is the
trouble.
The same thing could be extended to uncertainty, but that brings in the
question being asked of why information is key. (I don’t have time to get
to my main theme, but all right). Once you start out on the idea that we’re
ensuring there’s uncertainty, there comes the problem that people know
different things. There’s asymmetric information. And of course, we’ve had
an enormous development of the theory of asymmetric information, but
it tends to be static. You have to laugh at the fact that there’s going to be a
realization.
Equilibrium, Welfare, and Information 33
So let me conclude by saying that where I find general equilibrium the-
ory most used is as a basis for models. Climate change illustrates it. What
have we got there? We have dynamic models, like Nordhaus and others
have developed. That is, you have models of the future, and we make it
clear that they are price clearing models. In fact, they are optimal, so they
clear with full anticipation of what’s going to happen in the future. So these
models are fully specified. They’re used for predictive purposes, and they’re
used for policy formation purposes. And that’s where I think equilibrium
theory is having its biggest use now. Thank you.
Comment: Shantayanan Devarajan
It is an honor to be a discussant for Kenneth Arrow’s presentation. I learned
general equilibrium theory from Gerard Debreu, whom Ken mentioned,
Debreu pair. I
and it’s nice to hear from the other half of the great Arrow-
will focus my remarks on the three nouns in the title of the paper, “Equilib-
rium, Welfare, and Information.”
First, equilibrium. The proof of the existence of a general equilibrium, due
to Arrow and Debreu, is one of the most powerful contributions in econom-
ics. Its power lies not just in its mathematical elegance but in its utility. For
we use general equilibrium reasoning every day, including at the World Bank.
Without the proof that the interaction among sectors through the price
mechanism is a consistent system, we would be spinning tales out of thin air.
The idea of the Dutch disease (Corden and Neary 1982), where a booming
sector (such as oil) increases prices of nontradables and decreases output of
the traditional tradable sector, is not just a random collection of hypotheses.
It is a description of the general equilibrium system that, thanks to Arrow,
we know the conditions under which it exists. I used a general equilibrium
model to estimate the overvaluation of the CFA franc in Africa (Devarajan
1997). That estimate was quite close to the actual devaluation in 1994. Again,
we could not have built the model, much less used it, without a coherent
theory that this way of describing the economy is analytically founded. This
idea of the interdependence of different sectors of the economy, mediated
through prices, is central to development. In recent work on cronyism in
Tunisia, Rijkers, Freund, and Nucifora (2016) looked at monopoly power in
the telecommunications sector, which had been granted because of con-
ruling family. The authors showed that by raising tele-
nections to the then-
coms prices, the monopoly had undermined the competitiveness of Tunisia’s
Equilibrium, Welfare, and Information 35
garments and electronics manufacturing sectors, which is another example
of applied general equilibrium reasoning (and an explanation of why Tuni-
sia’s exports are not growing). Perhaps most importantly, the whole notion
of inequality, which is currently being hotly debated in rich and poor coun-
tries, has to be understood in terms of general equilibrium. This concept is
fundamental, because the distribution of income is a function of both the
uses and sources of income, which in turn are functions of how prices and
rich develop-
quantities adjust in different sectors. For example, in a resource-
ing country like Zambia, a favorable terms of trade shock can lead to greater
inequality, because poor people spend more of their income on nontrad-
a general equilibrium result that may
able goods (Devarajan and Go 2003)—
round, partial equilibrium one. In short—
deviate from a first- and I say this
with some trepidation, because many contributors to this volume have made
the proof of the existence of general
enormous contributions to economics—
equilibrium is one of the most powerful contributions, not just to economics
but also to the welfare of poor people.
This brings me to the second topic of Arrow’s talk, which is welfare. The
two fundamental theorems of welfare economics, which state that, under
certain assumptions, a competitive equilibrium is Pareto optimal and that
any Pareto optimum can be supported by a competitive equilibrium, are,
well, fundamental. But as Arrow points out, they are important because of
what happens when you relax some of the assumptions. For example, when
externalities exist, the competitive equilibrium will not be Pareto optimal.
This is the cornerstone of economic policy: The purpose of economic
policy is, when the assumptions of the first welfare theorem don’t hold,
to get us from the competitive equilibrium to the social optimum. Here
is where I think we have a problem. Although everyone agrees that our
we’ve studied the theorems in
goal should be to maximize social welfare—
some of our behavior does
graduate school and can probably recite them—
not follow suit. Having agreed that our purpose is to increase welfare, we
sometimes develop “special initiatives” that include such goals as universal
primary enrollment, or universal health care, or universal financial access.
To be sure, these are worthy goals, but it is not clear that achieving any one
of them is welfare maximizing. You could likely do better by increasing
access to something at a very low level than spending the marginal dollar
on going from 99 to 100 percent access in one of the other areas. So I think
36 Comments by Shantayanan Devarajan and Karla Hoff
development economists should be vigilant in pursuing the goal of welfare
rather than appealing to constituencies or the latest trends.
Finally, despite its appeal, the general equilibrium model, and general
equilibrium theory, have come under some criticism. One such criticism,
which Arrow alluded to, is that the assumption that you have a complete
set of markets for every contingency is unrealistic. It’s hard to imagine that
everybody knows exactly what they’re going to buy under every possible
state of the world. This is why a whole body of work has developed on
general equilibrium under uncertainty. Joe Stiglitz and other contributors
to this volume have made seminal contributions in this area. A second criti-
cism is that people may not follow the optimizing behavior that is assumed
in standard general equilibrium models. Consumers may not maximize
utility; producers may not maximize profits. People have limited cogni-
tive capacity. This has led to the area of behavioral economics, which my
colleague Karla Hoff will discuss next. Despite the great progress that Karla
and others have made in this field, we have yet to develop a fully specified
theory of general equilibrium where agents are not optimizing, comparable
to the traditional theory of general equilibrium. Such a theory would be a
fitting tribute to the great work of Kenneth Arrow.
References
Industrialisation in
Corden, W. Max, and Peter J. Neary. 1982. “Booming Sector and De-
a Small Open Economy.” Economic Journal 92 (368): 825–848.
Devarajan, Shantayanan. 1997. “Real Exchange Rate Misalignment in the CFA
53.
Zone.” Journal of African Economies 6 (1): 35–
Devarajan, Shantayanan, and Delfin S. Go. 2003. “The 123PRSP Model.” In The
Impact of Economic Policies on Poverty and Income Distribution: Evaluation Techniques
and Tools, edited by François Bourguignon and Luiz A. Pereira da Silva, 277– 300.
New York and Washington, DC: Oxford University Press and World Bank.
Rijkers, Bob, Caroline Freund, and Antonio Nucifora. 2017. “All in the Family: State
59.
Capture in Tunisia.” Journal of Development Economics 124: 41–
Comment: Karla Hoff
Do Social Factors Determine “Who We Are”?
In a conference that asks where economics is headed, it is natural that the
first invited speaker be Kenneth Arrow. As much as anyone else who was
alive in 2016, he had advanced the field of economics. He was the first to
prove Adam Smith’s conjecture that under some conditions, the market
economy attains the ideal of Pareto efficiency (Arrow 1951), and his proof
edged sword: It showed that a market equilibrium is Pareto effi-
was a two-
cient only under conditions so special that they would never be met in
reality, even approximately (Greenwald and Stiglitz 1986). What sets Arrow
apart from every economist before him is that he understood how unre-
alistic the conditions must be for market equilibrium to produce a Pareto
efficient allocation. He also understood that the impersonal price system
supplied a very incomplete description of reality.
Arrow consistently pushed the boundaries of neoclassical economics, in
part by going back to earlier traditions that explored how a society as a whole
functions. He studied peer influences on preferences (Arrow and Dasgupta
2009). He demonstrated that in a competitive economy, the rate of invest-
ment in learning would be too low, since learning benefits future investors
who do not pay for it (Arrow 1962). Although he never left the framework
of rational choice theory, he pushed the boundaries of the emerging field of
behavioral economics, too. Leaders in that field, Richard Thaler and Sendhil
Mullainathan (2008), had defined it as one that introduced psychologically
making into economics. Arrow
more realistic assumptions about decision-
the
(2010, 12) commented: “[T]oday psychology is invading economics—
whole field of behavioral economics. I believe that sociology should play
38 Comments by Shantayanan Devarajan and Karla Hoff
more of a role in economics than it does. The way people behave in econom-
ics is partly influenced by how other people behave. It’s easy to point out
examples, but it’s not so easy to construct a broad theory.”
Behavioral economics has moved in the direction of sociology in the
first century. The research in behavioral economics has two strands.
twenty-
The first focuses on the quasi-rational actor, who is rational when she thinks
“slow” but who much of the time thinks “fast” using heuristic principles
to reduce to simpler operations the complex task of making decisions (for
example, Thaler 2000). The second strand, in which sociology, anthropol-
ogy, and neuroscience play a role, is concerned with a quasi-rational, encul-
turated actor. The cognitive tools she uses to expand her ability to process
information fast are endogenous, not universal. They differ across groups
cultural environments that she
and over time. They are shaped by the socio-
and her ancestors have experienced or been exposed to (Nunn 2012; Hoff
and Stiglitz 2016; Demeritt and Hoff 2018).
Each strand is easy to illustrate. Kahneman and Tversky, pioneers of the
first strand, showed that the mechanisms of cognition (rather than merely
the emotions) produce systematic errors of intuition. For example, when
Kahneman (2011) shows us the box on the next page, we think the middle
symbol is “B.”
But when he shows us the next box, we think the middle symbol is “13.”
In neither case do we think the middle symbol is ambiguous. The example
illustrates that “one does not just see, one sees as” (Bacharach 2003, 63).
Kahneman emphasizes that automatic, not deliberate, thinking is the
“secret author of many of the choices and judgments you make” (Kahne-
man 2011, 13). Automatic thinking entails matching a stimulus to known
patterns and making associations. It does not entail logic or careful reason-
ing. If an individual doesn’t have useful patterns and concepts that are eas-
ily accessible, she won’t make good choices and judgments.
Behavioral economics shows that when people are making choices based
on automatic thinking, interventions can sometimes nudge them to make
choices that leave them better off. “Nudges” have been devised to help peo-
ple in poor countries save enough for medical expenses and health needs
(Dupas and Robinson 2013), buy fertilizer (Duflo, Kremer, and Robinson
2011), treat unclean drinking water regularly with diluted chlorine (Kremer
stage immunization programs to pro-
et al. 2011), and complete multiple-
tect their children from disease (Banerjee et al. 2010).
Equilibrium, Welfare, and Information 39
The second strand of behavioral economics goes beyond nudges. It
considers how to change the repertoire and the accessibility of cognitive
for example, cultural categories and narratives—
tools— that individuals use
to process information. By expanding the repertoire or making some mental
models more accessible, exposure to new social patterns (even in fiction)
run social change.
can induce long-
fourth of the popula-
At the turn of the twentieth century, about one-
tion in Brazil watched a soap opera at 9:15 each weeknight. Globo was the
main producer of soap operas in Brazil. It crafted them with characters who
had few or no children in order to reduce the number of characters in the
stories. Small family size sharply contrasted with the prevailing patterns in
Brazil.
Exposure to the soap operas lowered fertility rates in Brazil! Causal
identification of the impact is possible because of the arguably random year
that different municipalities obtained access to the Globo transmissions.
The fertility rate in a municipality declined after the first year that it had
access to transmissions of these soap operas (La Ferrara, Chong, and Duryea
2012). The decline was greatest for women who were within 4 years of the
age of a leading female character in the soap operas, which is consistent
with a role model effect. The effect was comparable to that of an increase
40 Comments by Shantayanan Devarajan and Karla Hoff
in average education of women by 2 years. Yet the effect was not driven by
a change in assets or skills or prices, but only a change in the kinds of lives
people imagined for themselves.
Changes in markets can also create new prototypes and thereby induce
changes in preferences. A randomized controlled trial by Robert Jensen
(2012) indicates that the proportion of young women in an Indian vil-
lage who have business process outsourcing (BPO) jobs, such as at call cen-
ters, influences the average marriage patterns, education, fertility rates, and
aspirations in the village. To conduct the experiment, Jensen hired eight
call center recruiters and sent them to recruit women in 80 villages ran-
domly chosen from a set of 160 villages about 100 kilometers from Delhi
(too remote for profitable visits from recruiters). His experiment created a
surge in demand in those 80 villages for women in BPO jobs. Before the
experiment, no members of any household in these villages held a BPO
job. As a result of the experiment, there were 11 job matches on aver-
age per village over 3 years. The proportion of young women with BPO
jobs increased from 0 to 5.6 percent in the treatment villages. The surge
in demand changed how women in the treatment villages defined their
lives and how parents perceived and cared for their daughters, as table
1.1 shows.
The change in choice sets would have rationally changed expectations
for women too. But it is plausible that by seeing young women play new
roles, the lives that parents and young women imagined were possible for
them had changed. The increase in the body mass index (BMI) of girls aged
15, shown in table 1.1, is evidence that daughters were better cared for
5–
in treatment than in control villages. It is evidence that a cultural shift
had occurred. Like the study of the effect on fertility rates of Globo soap
operas in Brazil, the randomized controlled trial using call center recruiters
in India shows the kind of social influences that Arrow suggested behav-
ioral economics should take into account.
Social influences can, of course, be bad or good. Just as social experience
and exposure expanded individuals’ sense of “who they were” in the previ-
ous two examples, they can also narrow this sense and make a society rigid.
In a village in which most girls are uneducated, it is possible to sustain a
stereotype of educated women as immoral and a threat to the social order,
which sustains the social pattern of low education for girls.
Equilibrium, Welfare, and Information 41
Table 1.1
Social impacts of hiring female villagers in BPO jobs
Control villages Treatment villages
Women of age 15– 21
Percentage who married during the 0.71 0.66
year period of the experiment
3-
Percentage who gave birth during the 0.43 0.37
year period of the experiment
3-
Number of children that the individual 3.00 2.65
desires
15
Girls of age 5–
z-score of body mass index for age −1.25 −1.01
Source: Based on Jensen (2012).
In interviews throughout India, comments of women demonstrate the
influence of prevailing education levels on attitudes toward educating
girls. When asked why their daughters never went to school, some parents
responded, “We don’t educate girls in our community.” In contrast, when
parents in Kerala, a socially progressive state of South India, were asked
why they send their children to school, “some of them don’t know what to
evident that going to school is what
say simply because they take it as self-
children do” (PROBE Team 1999, 22, 24).
The fact that attitudes and choices about educating girls are widely
shared within a village and vary across villages suggests the existence of
ranked equilibria. Hoff and Stiglitz (2016) for-
multiple stable and Pareto-
malizes this observation in a simple model. It assumes that in each of
the many households in a village, there is a young girl whom the par-
ents have to choose to educate, or not. How they think about the girl’s
education depends on the village stereotype of an educated woman and
determined lifetime earnings, W (call the former
on her expected market-
their “framed utility,” after Kahenman 2011, chapter 34). Consider two
stereotypes of an educated woman, denoted A and P. Under stereotype
A, a woman’s autonomy is held in esteem, and an educated daughter is a
source of pride to her parents. Under stereotype P, an educated woman is
a threat to the patriarchal social order and to her husband’s masculinity,
which means that an educated daughter is difficult to marry off. Parents
do not have fixed preferences over educating their daughter. Instead, their
42 Comments by Shantayanan Devarajan and Karla Hoff
preferences depend on the stereotype that is cued by the environment.
Let U(s) be the weighted sum,
U(s) = ω(s)VA + [1 − ω(s)] VP + W,
where s is the salience of the mental model A, V A is the parents’ intrinsic
valuation of an educated daughter under mental model A, and similarly for
V P. Let s be the fraction of village households that educate their daughters.
The weight ω(s) is increasing in s: If all households educate daughters, ω = 1.
If none do, ω = 0. Figure 1.1A illustrates the function U(s).
For simplicity, assume that having an uneducated daughter would give
parents utility θ that is independent of the fraction of households in the vil-
lage that educate their daughters. Across households, θ varies because some
parents have greater need than others for a young child to tend to another
family member, such as an infant or a sick grandmother. Figure 1.1B assumes
a roughly normal distribution of θ above some fixed, low value.
The evolution of the fraction of educated girls closes the model. A long-
run interior equilibrium is the fraction of daughters who are educated,
s*, at which the marginal parents are indifferent between educating their
daughter or not doing so. In the neighborhood of any value of s* at a stable
equilibrium, parents for whom θ is less than U(s*) would be strictly better
off educating their daughters, and parents for whom θ is more than U(s*)
would be strictly worse off educating their daughters. See figure 1.1C, where
the two graphs are superimposed. There are two stable equilibria (marked
by circles) and one unstable equilibrium between them. In the bad equi-
librium, the village has no educated girls: the patriarchal stereotype P is
so salient that no parents want to educate their daughters. In the good
equilibrium, stereotype A is so salient that most parents have the opposite
preference: most prefer to educate their daughters.
The stereotypes in this model are a linchpin that reflects social patterns
(“normal” girls do, or don’t, get educated) and affects individual behavior
(the parents’ decisions to educate girls) in ways that sustain the stereotypes
and the social pattern in a “cycle of mutual constitution” (Markus and
Kitayama 2010). The social pattern in the village shapes how people think
and the alternatives they can imagine. The social pattern is naturalized,
haps preferable, and prevail
even though other outcomes are possible, per
in other villages. Behavioral development economics, an emerging field in the
first century, sheds light on how dysfunctional social institutions
twenty-
Equilibrium, Welfare, and Information 43
A UƟlity
B θ
θ
VA + W U(s)
VP + W
s F(θ)
0 1 0 1
FracƟon of educated daughters FracƟon of households
C Utility
θ
VA +W U(s)
Good equilibrium
Unstable
equilibrium
Bad equilibrium
VP + W
F(θ), s
0 1
Fraction of educated daughters and fraction of households
Figure 1.1
Role model effects on parents’ decision to educate a daughter
Notes: (A) Parents’ “framed utility” U from an educated daughter. The utility depends
on the salience of a stereotype A, in which an educated girl is a source of pride to her
parents, and a stereotype P, in which an educated girl is perceived to be a threat to
the patriarchal social order. The salience of the stereotype A depends on the fraction
of educated daughters in the village. W is the market- determined lifetime expected
earnings of an educated girl.
(B) Cumulative distribution function of parents’ utility from a daughter who is not
educated.
(C) Multiple equilibria of the proportion of parents who choose to educate their
daughters.
Source: Hoff and Stiglitz (2016).
44 Comments by Shantayanan Devarajan and Karla Hoff
(such as low education for girls) can persist and affect how people think
and what they can imagine. In that sense, social patterns can determine
“who we are.”
In a famous article on medical care and insurance, Arrow (1963) dis-
cusses the problem of asymmetric information. He argued that equilibria in
insurance markets are very far from Pareto efficient. Buying insurance for
the risk of a car accident will reduce the care that the insured party takes.
If she knows she’s a bad driver but the insurance company does not, she
is likely to fully insure. At the high price at which the insurance company
breaks even on bad drivers, good drivers won’t be willing to fully insure.
Sellers and buyers of insurance do not have the same information and,
thus, are not really trading the same things (Rothschild and Stiglitz 1976
show that market equilibrium will thus not be Pareto efficient).
But whatever information decision makers have, neoclassical econom-
ics assumes that they process it objectively. Behavioral economics departs
from that assumption and recognizes the systematic influence of cultural
mental models for subjectively processing information. Perception is selec-
tive. Depending on the activated mental model, an individual sees different
things. Recall the earlier figure that showed that depending on the frame,
a person might be sure that a symbol was “B” or “13.” That is, “one does
not just see, one sees as.” Culture works through the interaction of shared
mental models and the information and context that activate those mental
models to varying degrees (DiMaggio 1997, 264, 274).
Neoclassical Economics Behavioral Economics
Strand 1 Strand 2
Concept of the actor The raƟonal actor The quasi-raƟonal actor The quasi-raƟonal, enculturated actor
Also guided by context in the moment Also guided by experience and exposure,
The drivers of behavior Guided by incenƟves
of decision, e.g. which shape:
Presentation. Mental models
Default options Categories
Language Concepts
Cues Identities
Reminders Narratives
Mental accounting What primes certain behaviors
Figure 1.2
Neoclassical economics and the two strands of behavioral economics
Source: Based on Hoff and Stiglitz (2016).
Equilibrium, Welfare, and Information 45
“Nudges” are based on the the idea that a change in a frame changes
what is seen and may change what one does. Interventions to change expe-
rience or exposure (for example, exposure to new role models) are based on
run, they will change the repertoire or acces-
the idea that in the medium-
sibility of mental models and thereby change the concepetual frames that
one brings to a problem.
Figure 1.2 illustrates the three types of actors assumed in modern work
rational actor; and the quasi-
in economics: the rational actor; the quasi-
rational, enculturated actor. By conceptualizing the last actor, recent work
in behavioral economics has taken up Ken Arrow’s recommendation that
sociology should play more of a role in economics.
References
Arrow, Kenneth J. 1951. “An Extension of the Basic Theorems of Classical Welfare
Economics.” In Proceedings of the Second Berkeley Symposium on Mathematical Statistics
and Probability, edited by Jerzy Neyman, 507– 532. Berkeley: University of California
Press.
Arrow, Kenneth J. 1962. “The Economic Implications of Learning by Doing.” Review
173.
of Economic Studies 29 (3): 155–
Arrow, Kenneth J. 1963. “Uncertainty and the Welfare Economics of Medical Care.”
973.
American Economic Review 53 (5): 941–
Arrow, Kenneth J. 2010. “The Economy of Trust: An Interview with Kenneth Arrow.”
Religion and Liberty 16 (3): 3, 12–13. https://acton.org/pub/religion-liberty/volume
-16-number-3/economy-trust.
Arrow, Kenneth J., and Partha S. Dasgupta. 2009. “Conspicuous Consumption,
F516.
Inconspicuous Leisure.” Economic Journal 119 (541): F497–
Bacharach, Michael. 2003. “Framing and Cognition in Economics: The Bad News
and the Good.” In Cognitive Processes and Economic Behaviour, edited by Marcello
74. New York: Routledge.
Basili, Nicola Dimitri, and Itzhak Gilboa, 63–
Banerjee, Abhijit, Esther Duflo, Rachel Glennerster, and Dhruva Kothari. 2010.
“Improving Immunisation Coverage in Rural India: Clustered Randomised Control
Evaluation of Immunisation Campaigns with and without Incentives.” British Medi-
cal Journal 340 (1): c2220.
Demeritt, Allison, and Karla Hoff. 2018. “The Making of Behavioral Development Eco-
nomics.” History of Political Economy 50 (annual supplement): 303– bit
322. http:// ly
.
/2GwyGUK.
46 Comments by Shantayanan Devarajan and Karla Hoff
287.
DiMaggio, Paul. 1997. “Culture and Cognition.” Annual Review of Sociology 23: 263–
Duflo, Esther, Michael Kremer, and Jonathan Robinson. 2011. “Nudging Farmers to
Use Fertilizer: Theory and Experimental Evidence from Kenya.” American Economic
Review 10 (6): 2350–2390.
Dupas, Pascaline, and Jonathan Robinson. 2013. “Why Don’t the Poor Save More?
Evidence from Health Savings Experiments.” American Economic Review 103 (4):
1138–1171.
Greenwald, Bruce C., and Joseph E. Stiglitz. 1986. “Externalities in Economics with
Imperfect Information and Incomplete Markets.” Quarterly Journal of Economics 101
(2): 229–264.
Hoff, Karla, and Joseph E. Stiglitz. 2016. “Striving for Balance in Economics: Towards
a Theory of the Social Determination of Behavior.” Journal of Economic Behavior and
Organization 126 (Part B, June): 25– 57.
Jensen, Robert. 2012. “Do Labor Market Opportunities Affect Young Women’s Work
and Family Decisions? Experimental Evidence from India.” Quarterly Journal of Eco-
792.
nomics 127 (2): 753–
Kahneman, Daniel. 2011. Thinking Fast and Slow. New York: Farrar, Straus and Giroux.
Kremer, Michael, Edward Miguel, Sendhil Mullainathan, Clair Null, and Alix Peter-
son Zwane. 2011. “Social Engineering: Evidence from a Suite of Take- Up Experi-
ments in Kenya.” Unpublished manuscript, University of California, Berkeley.
La Ferrara, Eliana, Alberto Chong, and Suzanne Duryea. 2012. “Soap Operas and Fer-
31.
tility: Evidence from Brazil.” American Economic Journal: Applied Economics 4 (4): 1–
Markus, Hazel Rose, and Shinobu Kitayama. 2010. “Culture and Selves: A Cycle of
430.
Mutual Constitution.” Perspectives on Psychological Science 5 (4): 420–
Nunn, Nathan. 2012. “Culture and the Historical Process.” Economic History of Devel-
126.
oping Regions 12 (27): 108–
PROBE Team. 1999. Public Report on Basic Education in India. New Delhi: Oxford Uni-
versity Press.
Rothschild, Michael, and Joseph E. Stiglitz. 1976. “Equilibrium in Competitive
Insurance Markets: An Essay on the Economics of Imperfect Information,” Quarterly
649.
Journal of Economics 90 (4): 629–
Thaler, Richard. 2000. Misbehaving: The Making of Behavioral Development Economics.
New York: W. W. Norton.
Thaler, Richard H., and Sendhil Mullainathan, 2008. “Behavioral Economics.” In
37. India-
The Concise Encyclopedia of Economics, edited by David R. Henderson, 34–
napolis, IN: Liberty Fund.
2 Social Choice and Welfare Economics
Amartya Sen
In the making of acceptable social decisions for a group (such as a nation, a
community, a committee, or any other collectivity), the diverse views and
interests of members of the group must receive attention and importance.
This can be an exacting task, because people’s views can differ, and as Horace
pointed out a long time ago, there may be as “many preferences as there
are people.” Choosing actions and policies for a group can be formidably
difficult.
And there are, in addition, difficult issues even in describing what exactly
is happening to a group as a whole. Is it better off or worse? Are its members
happier? Do they have more freedom than before? Is there more poverty
or less than in the past? Has social inequality in the group diminished or
increased? Can the social decisions that emerge be seen as democratic, or
are they, in some important sense, authoritarian? Methods of aggregative
assessment are central to the subject of social choice in general and welfare
economics in particular.
People have speculated on social aggregation throughout human his-
tory. However, social choice theory as a formal discipline first came into its
own around the time of the French Revolution. The subject was pioneered
C.
by French mathematicians in the late eighteenth century, particularly J.-
Borda (1781) and Nicolas de Condorcet (1785). They addressed social choice
problems in rather mathematical terms and initiated the intellectual disci-
pline of social choice theory in terms of voting and related procedures. The
intellectual climate of the period was greatly influenced by the European
Enlightenment, with its interest in reasoned construction of a social order.
Indeed, some of the early social choice theorists, most notably Con-
dorcet, were also among the intellectual leaders of the French Revolution.
Condorcet noted that Anne Robert Jacques Turgot, the pioneering French
48 Amartya Sen
economist (and also the governor of the province of Limoges), whom
Condorcet greatly admired, was the first statesman who “deigned to treat
the people as a society of reasonable beings” (Condorcet 1847, 9, 15, 18).
Condorcet admonished Jacques Necker, an opponent of Turgot, for “exag-
gerating the stupidity of people.” Condorcet took great interest, especially
making in assemblies, including
in his later works, on interactive decision-
“assemblées d’administration,” charged with making decisions about taxa-
tion, public works, militias, the use of public funds, and the management
of public goods.
The motivation for the early social choice theorists included the avoid-
ance of authoritarianism as well as arbitrariness in social choice. Their work
focused on the development of a framework for rational and democratic
decisions for a group, paying adequate attention to the preferences and
interests of its members. However, even the theoretical investigations typi-
cally yielded rather pessimistic results. Condorcet noted, for example, that
majority rule can be caught in an impasse when every alternative is defeated
in voting by some other alternative. To illustrate the “voting paradox,” first
member community in which person 1
spotted by Condorcet, consider a 3-
strictly prefers x to y and that to z; person 2 ranks them in the strict order of
y, z, and x; and person 3 strictly ranks them as z, x, and y. Then x will defeat
y by majority vote, while y defeats z, and z vanquishes x, thereby generat-
ing a “cycle.” More particularly, every alternative is rejected in a majority
vote by some other alternative, and there will be no “Condorcet winner,”
that is, an alternative that wins against (or at least stays undefeated against)
every other alternative.
Even though there is no continuous line of work on social choice theory
following the early lead of French mathematicians, the subject received
sporadic attention in various writings, often from distinguished intellectu-
als, such as Lewis Carroll, the author of Alice in Wonderland (he wrote some
engaging and important papers on group decisions under his real name,
C. L. Dodgson (1876, 1884)).
and fully axiomatized—
However, in its modern— form, modern social
choice theory had to wait until the middle of the twentieth century for
its first rigorous foundation in the work of Kenneth J. Arrow. His famous
“impossibility theorem,” contained in his PhD dissertation, was first
reported in a journal article (Arrow 1950). His thesis was published shortly
thereafter as a monograph (Arrow 1951), which became an instant classic.
Social Choice and Welfare Economics 49
Economists, political theorists, moral and political philosophers, sociolo-
gists, and even the general public took rapid notice of what seemed like—
a devastating result. And in a comparatively short time,
and indeed was—
social choice theory in a modernized and systematically axiomatic form
was firmly established as a discipline with immediate and extensive impli-
cations for economics, philosophy, politics, and the other social sciences.
Very rarely in intellectual history has a young graduate student so pro-
foundly influenced the course of social thought in the world.
Like Condorcet with his “voting paradox,” Arrow was also concerned
with the difficulties of group decisions and the inconsistencies to which
they may lead. Arrow’s “impossibility theorem” (formally, the “general pos-
sibility theorem”) is a result of breathtaking elegance and power. The theo-
rem shows that even some very mild conditions of reasonableness could
not be simultaneously satisfied by any social choice procedure in the wide
family of such procedures that identify a social ordering for any collection
of individual preference orderings over social alternatives.
The fundamental challenge that Arrow considered is that of going from
individual preferences over the different states of affairs to a social prefer-
ence over those states, reflecting something like an “aggregation” of the
points of views of all members of the society. He wanted the social prefer-
ence to be an “ordering” (sometimes called a “complete ordering”). A rank-
one preferred to
ing is an ordering if (1) any two alternatives can be ranked—
the other, or the opposite, or they are indifferent to each other (this is called
the “completeness” of the ranking), and (2) the ranking has a requirement
of coherence that goes by the name of “transitivity” (a flash of grammatical
language in the field of preferences). Transitivity demands that if an alter-
native x is taken to be at least as good as y, and y to be at least as good as z,
then x must be judged to be at least as good as z. Arrow saw these demands
on the social choice as requirements of “collective rationality.”
A social choice procedure that takes us from a cluster (or “profile”) of
individual preference orderings (one ordering per person) to a social pref-
erence ordering is called a “social welfare function,” as defined by Arrow.
Interpreting this in the context of welfare economics, if a state of affairs
x is socially ranked above another state y, then state x yields more “social
welfare” than does y. The impossibility theorem shows that if there are
at least three distinct alternatives and at least two different individuals
(though only a finite number of them), then a set of very mildly demanding
50 Amartya Sen
conditions of reasonableness cannot be satisfied together by any possible
social welfare function.
Consider the following four axioms characterizing a social welfare func-
tion, specifying a social ordering of alternative states of affairs for each pro-
file of individual preference orderings over those states.1
Unrestricted domain (U) claims that a social welfare function must work for
every profile of individual preferences (that is, it must generate a social
ordering for every cluster of individual preferences).
Independence of irrelevant alternatives (I) requires that the social ranking
of any pair of alternatives must depend only on the individual rankings
over just that pair (the “relevant” pair).
The Pareto principle (P) instructs that if everyone strictly prefers some alter-
native x to another alternative y, then social ordering too must place x
strictly above y.
dictatorship (D) demands that there should be no dictator such that
Non-
when that person strictly prefers any x to any y, then society must invari-
ably place x strictly above y.
looking axioms U, I,
Arrow’s impossibility theorem shows that these mild-
P, and D cannot be simultaneously fulfilled by any social aggregation pro-
cedure (or social welfare function).
This is not only an astonishing analytical result, but also one that gener-
ated much despair in the search for rational social choice procedures based
on individuals’ preferences. It also seemed like an antidemocratic result
of profound reach (which, in fact, is not quite the correct interpretation).
One common take on this result was that only a dictatorship would avoid
social inconsistencies, but a dictatorial rule would, of course, involve (1) an
extreme sacrifice of participatory decisions and (2) a gross inability to be
sensitive to the heterogeneous interests of a diverse population.
Two centuries after the flowering of the ambitions of social rationality in
Enlightenment thinking and in the writings of the theorists of the French
Revolution, the subject seemed to be inescapably doomed. Social apprais-
als, economic evaluations, and normative statistics would have to be, it
seemed, inevitably arbitrary or irremediably despotic.
1. This is a somewhat simplified version of the set of conditions that Arrow himself
used (see Sen 1970a).
Social Choice and Welfare Economics 51
The Idea of Social Preference
Arrow’s framework makes substantial use of the idea of social preference,
and Arrovian conditions of “collective rationality” seen in terms of direct
use of maximization based on the binary relation of social preference, or
indirect use of the idea through imposing internal consistency conditions
of choice that has a binary representation. The binary relation can be seen
if social preference.” James Buchanan (1954) has argued power-
as an “as-
fully against the alleged cogency of the idea of social preference, because
evident attribute
society is not an individual and so cannot have any self-
of a “preference.” The objection is particularly relevant in dealing with
political decisions rather than social welfare evaluation, because the lat-
ter demands some notion of a socially acceptable idea of a possibly binary
social welfare ranking relation. But the case for relying on institutional out-
comes rather than on any implicit idea of social preference can be seen to
be strong for political processes.
The possibility of a nonbinary formulation of the social choice has
received considerable attention in the literature of social choice theory
in recent years, led by contributors like Bergt Hansson, Thomas Schwartz,
Peter Fishburn, Donald Campbell, and Charles Plott. In some cases, the
impossibility results of the Arrow type seem resolved, and in others, they
functional framework. The question that
have been revived in the choice-
arises, however, is whether the impossibility results, thus derived, have
been crucially dependent on imposing conditions of internal consistency
of choice, which tend to take us in the direction of a binary representation
of the choice function. However, it turns out (see Sen 1993) that Arrow’s
impossibility theorem can be generalized to hold without any condition
of internal consistency of choice and without imposing any demands of
collective rationality. Through seeing the fuller implications of the rela-
tion between individual preferences and social choice (including seeing
independence of irrelevant alternatives in a more demanding light), the
Arrow impossibility can be shown to resurface without any use of internal
explicit or
consistency in social choice functions and without any idea—
of a social preference.
implicit—
52 Amartya Sen
Voting and Majority Decisions
As far as political decisions are concerned (postponing for the moment
welfare economic investigations), it seems fair to conclude that there is
not going to be any perfect resolution through voting procedures of the
social choice dilemmas of the kind identified by Arrow. This leads to two
different kinds of questions. First, even though there may not be any fault-
less voting procedure, do some of them function much better than others?
Second, is voting a good way at all of trying to resolve social choice prob-
lems of all kinds?
Majority voting has many rather attractive qualities and is considered
by many as a quintessential component of democratic decision making.
and more particularly, of not having
Can the grip of inconsistent choices—
be at least partially subdued? One of the ways of
a “Condorcet winner”—
coping with this challenge that has been much explored in this context is
through
the use of a “restricted domain” of the social welfare function—
that would avoid prob-
limiting the preference profiles that are allowed—
lems of inconsistency in voting results and also avoid the nonexistence
of a “Condorcet winner.” Arrow (1951) himself had initiated, along with
Duncan Black (1948, 1958), the search for adequate restrictions that would
guarantee consistent majority decisions, and he had identified a class of
peaked” preferences) that would work.
preference profiles (“single-
Black identification of sufficiency for consistent
In fact, the Arrow-
peaked preference profiles) can be vastly expanded
majority rule (single-
through using a process of reasoning not dissimilar to Arrow’s own, which
results in a much more general condition: “value restriction” (Sen 1966).
Value restriction demands that in every triple of alternatives (x, y, z), there
is one alternative (say, x) such that everyone agrees that it is either “not
best,” or “not worst,” or “not medium” (the position on which there is such
an agreement can vary from one triple to another).
Going from sufficiency conditions to the demands of necessity, the nec-
essary and sufficient conditions of domain restriction for consistent major-
ity decisions can also be precisely identified (see Sen and Pattanaik 1969). If
that is, they have no indifferences—
individual preferences are strict— then
these rather complex necessary and sufficient conditions boil down simply
to value restriction. However, even though these conditions are much less
restrictive than the earlier conditions that had been identified, they are still
Social Choice and Welfare Economics 53
quite demanding; indeed, it can be shown that they can be easily violated
in many actual situations.
Even though a voting impasse cannot be generally eliminated, it appears
that majority rule is, in fact, far less vulnerable to contradictions than other
procedures of voting. It can be shown that if there is a domain restriction
for which any voting rule other than the majority rule works well, then
so will majority rule (see Maskin 1995, 2014; Dasgupta and Maskin 2008).
Furthermore, for any nonmajority voting rule, there is a class of preference
profiles for which majority rule works well, but the other voting rules do
not. This powerful “dominance result” shows that even though all voting
rules are subject to impasse or contradictions, the method of majority rule,
which has other attractions too, is the least vulnerable among them. The
comparative robustness of majority rule is surely a pointer to its strength
that cannot but be important for many social and political decisions. But
that comfort may not be available for many other types of social choice. For
example, voting rules, including majority rule, may be quite inappropriate
as a basis for welfare economic judgments (on which more presently).
Liberty and Rights
Majority rule can also be severe against minority rights and may also work
against individual liberty. More than a century and a half ago, John Stuart
Mill ([1859] 1959) investigated how a good society should try to guarantee
the liberty of each person. Liberty has many different aspects, including
two rather distinct features:
1) The opportunity aspect: We should be able to achieve what we choose to
achieve in our respective personal domains, for example, in our private
life.
2) The process aspect: We can make our own choices in our personal domains
(no matter whether we achieve what we want).
In social choice theory, the formulation of liberty has been primarily con-
cerned with the former, that is, the opportunity aspect.
Seen in the perspective of the opportunity aspect, liberty demands that
each person should be decisive in safeguarding certain things in his or
her “personal domain,” without interference by others (even if a major-
ity is keen to interfere). J. S. Mill considered various examples of such
54 Amartya Sen
personal domains over which the person involved should be able to prevail,
for example—
including— in the practice of his or her own religion. Note
that the “opportunity aspect” cannot be safeguarded, as it is sometimes
wrongly presumed, by leaving to the person the choices to be made in her
personal domain, as an alleged “process guarantee.” The trouble is that
others can interfere in the practice of this person through their own actions
(for example, a person may be allowed to choose her religious practices,
but others could interfere through making hugely distracting loud noises,
or even by organizing disturbing demonstrations outside her home, mak-
ing life difficult for the person involved). It is the duty of the society, Mill
argued, to make sure that the person’s own choices over a personal domain
prevail (in this case, guaranteeing that the person can perform his or her
private religious actions, without being stopped by others, and also without
being hindered by the actions of others).
It is the conflict of this opportunity aspect of liberty with the Pareto prin-
ciple (given unrestricted domain) that is the subject matter of an impossi-
bility theorem, which is sometimes referred to as “the liberal paradox,” or
“the impossibility of the Paretian liberal” (See Sen 1970a, 1970b). Unlike
the Arrow theorem, this impossibility theorem does not depend on the
independence of irrelevant alternatives (condition I), which is not invoked
at all. Instead, it is shown that unrestricted domain (U) and the Pareto prin-
ciple (P) cannot be combined with “minimal liberty,” demanding only that
at least two persons are each decisive over the choice over one pair each.
There is a huge literature on the subject, including contributions that (1) dis-
pute the result, (2) extend it, (3) attempt to resolve the conflict, and (4) ques-
tion the interpretation of liberty. The theorem shows the impossibility, given
unrestricted domain, of satisfying even a very mild demand for “minimal lib-
erty” when combined with an insistence on Pareto efficiency.
Turning to the process aspect, seeing liberty as a guaranteed process of
leaving people free to do certain things in their own personal sphere is a
requirement that has been particularly pursued by various writers in this
field (led by Robert Nozick (1974), and joined in many distinct ways by
others). In this perspective, what liberty demands is that people remain free
to choose what to do in their personal domain, but it does not really mat-
ter what the actual outcome is (that is, it does not matter as far as liberty is
concerned). I cannot pretend that I find this conclusion particularly persua-
sive, because the opportunity aspect of liberty can also be very important.
Social Choice and Welfare Economics 55
In modern societies in particular, it is hard to give people the agency to
control what happens in all aspects of their lives. My liberty to fly safely is
better guaranteed by leaving many decisions to the pilot, rather than my
taking charge of the agencies in the cockpit. Our lives are saved by better
policing and effective epidemiology, which involve the agencies of many
other people (and not just on what we ourselves do).
However, it is hard to deny that liberty has both opportunity and pro-
cedural aspects. If being free to smoke is an important liberty (there can be
a debate on this), then surely a procedural system that allows anyone to
decide whether to smoke can rightly be seen as a part of liberty. However, if
a person who shuns smoking does not want smoke to be blown in her face,
her liberty to secure this does not depend primarily on what she does, but
mostly on what others do. Leaving her free with her action cannot elimi-
nate this violation of her personal liberty.
based liberty has
In the recent literature, the formulation of process-
been much refined from the simple statements originally made by Nozick
(1974). In particular, the specification of liberty has been given “game-
form” formulations (see Gaertner, Pattanaik, and Suzumura 1992), so that
agency freedoms are judged by the acceptability of combinations of dif-
as a
ferent persons’ actions (e.g., do not smoke if others are present, or—
do not smoke in places where others can be present if
stricter demand—
not deterred by the presence and activities of smokers). This refinement is
surely an important one, but as Gaertner, Pattanaik, and Suzumura explain,
it does not eliminate the impossibility of the Paretian liberal. Its merit lies
elsewhere, in particular, in capturing better the common idea of liberty
with the assignment of individual agency freedoms. It does not, however,
eliminate the relevance of social choice in assessing different game forms
(see Sen 1992; Hammond 1996). Game forms do help the specification and
analysis of liberty, but the motivation behind social choice theory would
continue to apply in the assessment of alternative game forms. And in that
context, we must take note of outcomes as well as processes.
Crisis in Welfare Economics
I turn now to welfare economics. Social choice difficulties apply inter alia
an old subject aimed at judging
to what is called “welfare economics”—
being (and other concerns) of the people,
social states in terms of the well-
56 Amartya Sen
on which A. C. Pigou’s (1920) distinguished book, The Economics of Welfare,
had been something of a classic account. The subject, however, had taken
quite a hard hit in the 1930s, even before Arrow’s impossibility result fur-
or seemed to darken—
ther darkened— the prospects of systematic welfare
economics. The initial crises came because of the economists’ newfound—
conviction that there was something quite
but rather hastily argued—
unsound in making use of interpersonal comparison of individual utilities,
which had been the basis of traditional welfare economics
Welfare economics had been developed by utilitarian economists (such
as Francis T. Edgeworth (1881), Alfred Marshall (1890), and Arthur C. Pigou
oriented social
(1920)) and had taken a very different track from the vote-
choice theory. It took inspiration not from Borda (1781) and Condorcet
(1785), but from their contemporary, Jeremy Bentham (1789). Bentham
had pioneered the use of utilitarian calculus to obtain judgments about
social interest by aggregating the personal interests of the different indi-
viduals in the form of their respective utilities.
and that of utilitarians in general (John Stuart Mill
Bentham’s concern—
was with the total utility of a community. The
was the exception here)—
focus, which has problems of its own, was on the total sum of utilities,
irrespective of the distribution of that total, and in this, we can see a par-
tial blindness of considerable ethical and political import. For example, in
the utilitarian best world of maximizing utility, a person who is unlucky
enough to have a uniformly lower capability to generate enjoyment and
utility out of income (say, because of a physical or mental handicap) would
be given even a lower share of a fixed total income, because of her lower
ability to generate utility out of income. This is a consequence of utilitari-
minded pursuit of maximizing the sum-
anism’s single- no
total of utilities—
matter how unequally distributed. However, the utilitarian interest in taking
comparative note of the gains and losses of different people is not in itself
a negligible concern. And this concern makes utilitarian welfare economics
in the form of compari-
deeply interested in using a class of information—
with which Condorcet
son of utility gains and losses of different persons—
and Borda had not been directly involved.
Utilitarianism has been very influential in shaping welfare economics,
which was dominated for a long time by an almost unquestioning adher-
ence to utilitarian calculus. But by the 1930s, utilitarian welfare economics
came under severe fire. It would have been quite natural to question (as
Social Choice and Welfare Economics 57
Rawls (1971) would do masterfully in formulating his theory of justice)
the utilitarian neglect of distributional issues and its concentration only
totals (in a distribution-
on utility sum- blind way). But that was not the
utilitarian critiques went in the 1930s and in
direction in which the anti-
the decades that followed. Rather, economists came to be persuaded by
arguments presented by Lionel Robbins and others (who were themselves
fashionable philosophical approach of “logi-
deeply influenced by the then-
cal positivism”) that interpersonal comparisons of utility had no scientific
basis: “Every mind is inscrutable to every other mind and no common
denominator of feelings is possible” (Robbins 1938, 636). Thus, the epis-
temic foundations of utilitarian welfare economics were seen as incurably
defective.
There followed attempts to do welfare economics on the basis of each per-
son’s respective ordering of social states, without any interpersonal compar-
isons of utility gains and losses of different persons. Although utilitarianism
and utilitarian welfare economics are quite indifferent to the distribution
of utilities among different persons (concentrating, as they do, only on the
total of utilities), the new regime, without any interpersonal compari-
sum-
sons in any form, further reduced the informational base on which social
choice could draw. The already limited informational base of Benthamite
calculus was made to shrink further to the narrow electoral plane of Borda
and Condorcet (I should explain that I am referring here to Condorcet as
in that capacity, his
a voting theorist, not as a general social philosopher—
attention was much broader). The use of different persons’ utility rankings
without any interpersonal comparison is analytically quite similar to the
each individual taken separately—
use of voting information— in making
social choice.
Attempted Repairs and Further Crises
Faced with this informational restriction, utilitarian welfare economics gave
way, from the 1940s on, to what came to be called—hugely overambitiously–
“new welfare economics,” which used only one basic criterion of social
improvement: the “Pareto comparison.” The Pareto criterion for social
improvement only asserts that a situation can be seen as definitely better
than another if the change would increase the utility of every one (or at
least increase the utility of someone without reducing the utility of anyone
58 Amartya Sen
else). A good deal of subsequent welfare economics restricted attention to
“Pareto efficiency” only (that is, only to making sure that no further Pareto
improvements are possible). This criterion takes no interest whatsoever in
distributional issues, which would tend to involve conflicts of interests of
different persons). So if one person gains while everyone else loses (no matter
and by how much), we were not allowed to declare this change
how many—
to be a deterioration, if we seek only Pareto efficiency.
This remarkable reticence, it seems fair to guess, would have appealed
to Emperor Nero, who evidently enjoyed playing his music while Rome
burned and all other Romans were plunged into misery. In general, the
Pareto efficiency of a state of affairs would not be disturbed even if many
people are forced into terribly famished lives, while some others lead lives
of extreme luxury, provided the misery of the destitute cannot be reduced
rich.
without cutting into the lives of the super-
beyond Pareto efficiency—
Some further criterion— is clearly needed for
making social welfare judgments with a greater reach, and this was insight-
fully explored by Abram Bergson (1938) and Paul A. Samuelson (1947). This
search led directly to Arrow’s (1950, 1951) pioneering formulation of social
choice theory, relating social preference (or decisions) to the set of individ-
ual preferences, that is, to the search for what Arrow called a “social welfare
function.” It was in the framework of social welfare functions that Arrow
(1951, 1963) established his powerful impossibility theorem, showing the
looking conditions (discussed earlier),
incompatibility of some very mild-
including Pareto efficiency, nondictatorship, independence of irrelevant
alternatives, and unrestricted domain. This generated further gloom in an
already gloomy assessment of the possibility of having a reasoned and sat-
isfactory welfare economics.
To escape the impossibility result, different ways of modifying Arrow’s
requirements were tried out in the literature that followed, but other diffi-
culties continued to emerge. The force and widespread presence of impossi-
bility results generated a consolidated sense of pessimism, and this became
a dominant theme in welfare economics and social choice theory in gen-
eral. By the middle 1960s, William Baumol, a distinguished contributor
to economics in general and welfare economics in particular, judiciously
remarked that “statements about the significance of welfare economics” had
concealed resemblance to obituary notices” (Baumol
started having “an ill-
1965, 2). This was certainly the right reading of the prevailing views.
Social Choice and Welfare Economics 59
Welfare Economics and Voting Information
It can be argued that the “obiturial” climate of welfare economics in its
postutilitarian phase was related largely to the epistemic penury of welfare
like inputs.
economics based on confining informational inflow to voting-
based procedures are entirely natural for some kinds of social choice
Voting-
problems, such as elections, referendums, or committee decisions. They are,
however, altogether unsuitable for many other problems of social choice.
For example, when we want to get some kind of an aggregative assessment
of social welfare, we cannot rely on such procedures for at least three distinct
reasons.
First, there are some serious problems in the correspondence between
actual preferences and the votes cast, which must take note of the pos-
sibility of strategic voting, aimed at manipulating the voting outcomes.
proof voting procedures has been well
The impossibility of having strategy-
established.2 The subject occupies a huge literature.
Second, voting requires active participation, and if some groups tend
not to exercise their voting rights (perhaps due to cultural conditioning or
because of procedural barriers that making voting difficult and expensive),
then the preferences of those groups tend to have quite inadequate repre-
sentation in social decisions. Because of lower participation, the interests
for example, of African Americans in the United
of substantial groups—
can have a quite limited influence on national politics.
States—
Third, even with the active involvement of everyone in voting exercises,
we will still be short of important information needed for welfare economic
evaluation. It is absurd to think that social welfare judgments can be made
without some understanding of issues of inequality and disparities that
characterize one society or another. Voting information, taken on its own,
its takes no direct note of how
turns a blind eye to such comparisons—
deprived different voters may be, nor of the extent to which their prefer-
ence reflects large differences or small ones. These limitations are related
being, on the impos-
to the eschewing of interpersonal comparison of well-
sibility of which for several decades, professional economists remained pre-
maturely convinced.
2. See Gibbard (1973), Satterthwaite (1975), and also Pattanaik (1973, 1978), Maskin
(1985) and Maskin and Sjöström (2002).
60 Amartya Sen
There was also the exclusion of what economists call “cardinal util-
ity,” which takes us beyond relying merely on the ranking of alternatives
the so-
in terms of being better or worse (or indifferent)— called ordinal
to giving us some idea of the relative gaps between the utility val-
utility—
ues of different alternatives. Utilitarian welfare economics uses cardinality
of utilities as well as interpersonal comparison of these utilities, and the
new orthodoxy that emerged in the 1930s disputed the scientific status of
both cardinality and of interpersonal comparison of utilities of different
persons.
Informational Penury as a Cause of Social Choice Problems
and influenced by
It is also worth recollecting that utilitarian philosophy—
had huge informational restric-
it, traditional welfare economics as well—
utility
tions of their own. It was not allowed to make any basic use of non-
information, because everything had to be judged ultimately by utility
totals in consequent states of affairs. To this informational exclusion
sum-
was now added the further exclusion of interpersonal comparisons of utili-
totals
ties, along with cardinal utility, which disabled the idea of utility sum-
utility information. This barren
without removing the exclusion of non-
informational landscape makes it hard to arrive at any systematic judgment
of social welfare, based on informed reasoning. Arrow’s theorem can be
interpreted, in this context, as a demonstration that even some very weak
in this case, Arrow’s axioms—
conditions— relating individual preferences
to social welfare judgments cannot be simultaneously satisfied in a world
of such informational privation (see Sen 1977b, 1979).
The problem is not just one of impossibility. Given Arrow axioms
U (unrestricted domain), I (independence of irrelevant alternatives), and
P (Pareto principle), the relation between the profile of individual prefer-
ences and the social ranking emerging from it has to forgo taking any note
of the nature of the alternatives (that is, the social states). The relation must
simply go by the individual preferences over the alternatives, no matter
for
what they are. If person 1 is decisive in the choice over any pair (a, b)—
then that person would be decisive in the social prefer-
whatever reason—
ence over every other pair of alternatives (x, y) as well, even though the
nature of the choice involved may radically differ because of the nature of
the social alternatives involved.
Social Choice and Welfare Economics 61
This requirement is sometimes called “neutrality” (a usage that had the
I very much hope only half-
support— of Arrow (1963) himself).
hearted—
It is, in fact, a peculiarly kind term for what is after all a sanctification of
blindness to all information other than utility information. Perhaps the
alternative term used for it (namely, “welfarism”) is more helpful, in that
it focuses on the limitation imposed by forbidding any direct use of any
information about the states of affairs other than the individual welfares
and that again only in the form of utilities.
they respectively generate—
Adding to that the further requirement that the utility information used
must not involve any cardinality, or any interpersonal comparison of utili-
ties, amounts to insisting that social choices must be made with extremely
little information indeed.
called neutrality tends to play havoc with the discipline
The demand of so-
of reasoned social choice. Consider, for example, a cake division problem,
in which everyone prefers to have a larger share of the cake. If, in this cake
division problem, an equal division between two persons in the form (50,
50) is socially preferred to person 1 having 99 percent of it, with the other
having only 1 percent in the form (99, 1), it is clearly being judged that per-
called
son 2’s preference should prevail over person 1’s, in this case. But if so-
neutrality is demanded, then due to the insistence that the nature of the
alternatives should not make any difference to whose preference prevails,
with person 2 having nearly all in the form
an opposite type of inequality—
should be socially preferred to a (50, 50) division, through the
of (1, 99)—
requirement that person 2, decisive over the earlier choice, should be decisive
over all other pairwise conflicts as well. It is hard to escape the thought that
and
something has gone badly wrong in the underlying intellectual system—
that problem arises even before any impossibility result emerges.
What is being presumed here is to insist that welfare judgments must
be based on something like voting data, taking note of who prefers what
but ignoring who is rich and who is poor, and who gains how much from a
change compared with what the losers lose. We must go beyond the class of
voting rules (explored by Borda and Condorcet as well as Arrow) to be able
to address distributional issues, particularly in welfare economics.
Arrow had ruled out the use of interpersonal comparisons, because he
had followed the general consensus that had emerged in the 1930s that
(as Arrow put it) “interpersonal comparison of utilities has no meaning”
(Arrow 1951, 9). The totality of the axiom combination used by Arrow had
62 Amartya Sen
the effect of confining social choice mechanisms to rules that are, broadly
speaking, of the voting type. His impossibility result relates, therefore, to
this class of rules with this informational abstinence.
It should be emphasized that, unlike ruling out the use of interpersonal
comparison of utilities, which Arrow explicitly invoked, the insistence on
restricting social choice procedures only to voting rules is not an assumption
quite star-
that is directly imposed by Arrow. It is, in fact, a combined result—
of the different axioms that Arrow uses. It can be seen as
tling in its own right—
an analytical consequence of a set of apparently reasonable axioms postulated
for social choice. Interpersonal comparison of utilities is, of course, explicitly
excluded, but in the process of proving his impossibility theorem, Arrow also
shows that a set of seemingly plausible assumptions, taken together, logically
entail other features of voting rules as well, in particular something close to
called neutrality (discussed earlier). This requires that no effective note be
so-
taken of the nature of the social states, and that the social decisions must be
and against—
based only on the votes that are respectively cast in favor of—
them. Although the eschewal of interpersonal comparisons of utilities elimi-
nates the possibility of taking note of the inequality of utilities (and also of
called
differences in gains and losses of utilities), the entailed component of so-
neutrality (or welfarism) prevents attention being indirectly paid to distribu-
tional issues through taking explicit note of the nature of the respective social
states (for example, of the incomes or wealth levels of different persons, as in
division example discussed earlier).
the cake-
This also brings out the disanalogy between Condorcet’s voting paradox
and Arrow’s much more general impossibility theorem (in contrast to some
common statements in the literature). Condorcet’s analysis begins with the
world of voting rules, whereas Arrow gets there only after establishing a
remarkable analytical theorem showing that the combination of a few very
apparently plausible axioms leaves us no option but to confine our vision
to voting rules. Some of the hard work in establishing Arrow’s theorem
ends where the Condorcet exercise begins.
Incorporating More Information in Social Decisions
To lay a broader foundation for a constructive social choice theory (broader
than the framework Arrow developed), we have to resist the historical con-
sensus against the use of interpersonal comparisons in social choice that was
Social Choice and Welfare Economics 63
dominant when Arrow began his research on social choice. That histori-
cal consensus was based on a rather fragile understanding of epistemology,
lived boom of logical positivism. The case for unquali-
derived from the short-
fied rejection of interpersonal comparisons of mental states is hard to sustain
(quite aside from the fact that these comparisons need not be of mental states
on which more presently).3 Indeed, as has been forcefully argued by
only—
the philosopher Donald Davidson (1986), it is difficult to see how people can
understand anything much about other people’s minds and feelings without
making some comparisons with their own minds and feelings. Such com-
parisons may not be extremely precise, but then again, we know from ana-
lytical investigations that very precise interpersonal comparisons may not be
needed to make systematic use of such comparisons in social choice.
However, aside from doubts about the evidential basis of interpersonal
comparisons, there were also questions about the possibility of a systematic
analytical framework for comparing and using the accounting of different
persons’ welfare magnitudes for social decisions, especially because inter-
personal comparisons can take many different forms. John Harsanyi (1955)
and Patrick Suppes (1966) made some early departures in that direction.
But they were more concerned with using interpersonal comparisons (of
“units” in the case of Harsanyi and of “levels” in the case of Suppes) rather
than with working out a comprehensive analytical framework for interper-
sonal comparisons in general, including the possibilities of specific features
of interpersonal welfare calculus.
Inspired by this challenge, I tried my hand at developing a comprehen-
sive analytical framework for interpersonal comparisons in my book Col-
up contributions
lective Choice and Social Welfare (Sen 1970a) and in follow-
(Sen 1977b, 1982). Happily, the 1970s and 1980s also saw the publication
of major contributions on the subject from a dazzling group of social choice
theorists, including Peter Hammond (1976); Claude d’Aspremont and Louis
Gevers (1977); Eric Maskin (1978, 1979); Louis Gevers (1979); Kevin Roberts
(1980a, 1980b); Kotaro Suzumura (1983, 1997); Charles Blackorby, David
Donaldson, and John Weymark (1984); d’Aspremont (1985); d’Aspremont
and Mongin (1998); and others. Even Kenneth Arrow (1977) joined this
3. On this issue and that of making actual interpersonal comparisons with factual
information, see Daniel Kahneman (1999, 2000), Alan Krueger (2009), and Krueger
and Stone (2014).
64 Amartya Sen
exploration. It is fair to say that we now have a much clearer understanding
of the analytical demands of different kinds and extents of interpersonal
comparisons, and the ways and means of making systematic use of that
information in social choice.
Without going into the technicalities that have emerged in the litera-
ture, it can be said that the extent and reach of different kinds of interper-
sonal comparisons can be explicitly invoked in a fully axiomatized form
(prominent types include full comparability, level comparability, unit com-
scale comparability, and so on; see Sen 1977b). Each kind of
pability, ratio-
comparability imposes its own demands on combining welfare numbers
of different persons. Consider, for example, a case of full comparability, by
being numbers 1, 2, 3 for person 1, respectively, for
beginning with well-
social alternatives x, y, and z, with the corresponding numbers for person
being, we
2 being 2, 3, 1. Because there are no naturally fixed units of well-
being numbers of person 1 for x, y, and z to
can easily enough alter the well-
be 2, 4, 6 instead of 1, 2, 3. Full interpersonal comparability would demand
being numbers by doubling them, then
that if we rescale person 1’s well-
being numbers
we must do the same for person 2, and transform her well-
from 2, 3, 1 to a corresponding set 4, 6, 2. With such tying up (they are
axiomatized through “invariance conditions”) implied by full interpersonal
comparability, it would not make any real difference whether we work with
the original numbers (1, 2, 3 for person 1, and 2, 3, 1 for person 2), or deal
instead with the symmetrically transformed numbers (2, 4, 6 for person 1,
and 4, 6, 2 for person 2). As different types of interpersonal comparability
(such as “level comparability” or “unit comparability”) are considered, we
shall have correspondingly different specifications of the invariance condi-
tions (see Sen 1970a, 1977b; Roberts 1980a).
Through the use of “invariance conditions” in a generalized framework
being numbers, going
that allow the use of interpersonally comparable well-
beyond simple rankings (to different extents, depending on the type of
interpersonal comparability), we get what are called social welfare function-
als, which allow the use of much more information than Arrow’s social
welfare functions permit. Indeed, interpersonal comparisons need not even
or-
be confined to all- none dichotomies. We may be able to make interper-
sonal comparisons to some extent, but not in every comparison, nor of
every type, nor with tremendous exactness. To illustrate, we may invoke
the same example of Nero and the burning of Rome, discussed earlier. It
Social Choice and Welfare Economics 65
seems reasonable to argue that there should be no great difficulty in accept-
ing that Emperor Nero’s welfare gain from the burning of Rome was smaller
total of the welfare loss of all the other Romans put together—
than the sum-
who suffered from the fire. But
perhaps hundreds of thousands of them—
this does not require us to presume that we can put everyone’s welfares in
to-
an exact one- one correspondence with one another. Thus, there may be
room for demanding “partial comparability”—denying both the extremes:
full comparability and no comparability at all.
The different extents of partial comparability can be given mathemati-
cally exact forms (precisely articulating the extent of the variations that
may be permitted). It can also be shown that terribly refined interpersonal
comparisons may not be needed for arriving at definite social decisions.
Quite often, rather limited levels of partial comparability will be adequate
for making social decisions. Thus, the empirical exercise need not be as
ambitious as is sometimes feared.
What Difference Does It Make?
How much of a change in the possibility of social choice is brought about by
systematic use of interpersonal comparisons? Does Arrow’s impossibility the-
orem (and related results) go away with the use of interpersonal comparisons
in social welfare judgments? In brief, the answer is yes. The additional infor-
mational availability allows sufficient discrimination to escape impossibili-
ties of this type. For example, with interpersonal comparability we can use
the Rawlsian distributive principle of maximin (what he calls “the Difference
Principle”), which takes the form of giving priority to the interests of the
off person (or persons).4 And this just demands “level comparability,”
worst-
while the units of different persons’ welfares need not be comparable at all.
There is an interesting contrast here. Although interpersonal comparabil-
ity even without cardinality helps dissolve Arrow’s impossibility theorem,
4. For compatibility with the Pareto principle (as well as for making reasonable
sense), this Rawlsian approach has to be used in what is called a “lexicographic”
form, so that in case where the worst-off persons tie with each other in the compari-
son between two states of affairs, we go by the interests of the second worst-off. And
so on. For the wide reach of Rawls’s criterion and its widespread relevance in public
policy, see Edmund S. Phelps (1973).
66 Amartya Sen
cardinality without interpersonal comparability does nothing of the sort.
In the absence of interpersonal comparability, Arrow’s theorem can, in fact,
be generalized to cover the case of fully cardinal utilities or welfares (see
Sen 1970a, chapter 8). In contrast, the possibility of only “ordinal” inter-
being between differ-
personal comparisons (so that the rankings of well-
ent persons remain invariant) is adequate to end the impossibility, even
without any cardinality. We already know of course that with some types
of interpersonal comparisons demanded in a full form (including cardinal
interpersonal comparability), we can use the classical utilitarian approach.
But it turns out that even weaker forms of comparability would still permit
making consistent social welfare judgments, satisfying all of Arrow’s require-
ments, in addition to being sensitive to distributional concerns (even though
the possible rules may have to be confined to a relatively small class; see
Roberts 1980a, 1980b).
Interpersonal Comparison of What?
Even though the analytical issues in incorporating interpersonal compari-
sons have been fairly well sorted out, there still remains the important prac-
tical matter of finding an adequate approach to the empirical discipline of
making interpersonal comparisons and then using them in practice. The
foremost question to be addressed is: interpersonal comparison of what?
being
Even though the debates about interpersonal comparison of well-
have been, historically, concentrated on the comparison of “utilities” in
which utilitarian philosophers were particularly interested, the subject of
interpersonal comparison in general is much broader than that.5
It must be recognized that the formal structures of social welfare func-
tions are not specific to utility comparisons only, and they can, in fact,
incorporate other types of interpersonal comparisons as well. The princi-
pal conceptual issue is the accounting of individual advantage. This need
5. Along with broadening the coverage of information for a better understanding
of poverty, there is also the important question of making sure that the empiri-
cal connections used in the informational expansion are appropriately tested and
scrutinized. Recently, randomized trials have been skillfully used to make the infor-
mational broadening more sure footed, whenever possible (see particularly Banerjee
and Duflo 2011).
Social Choice and Welfare Economics 67
not take the form of comparisons of mental states of happiness or desires
(which have been exclusively championed by utilitarian philosophers). It
being,
could instead focus on some other way of looking at individual well-
or freedom, or substantive opportunities.
Further, if the aggregation considered is that of individual judgments
(not of individual interests), then the question can also be raised about how
the divergent opinions or valuations of different persons may be combined
(this is a social choice exercise of a rather different kind, on which, see
Sen 1977a). This exercise, with complexities of its own, has also received
some attention (see particularly Christian List and Philip Pettit (2002) and
List (2005)). Furthermore, if utility comparisons are taken to be value judg-
ments themselves, rather than purely observational assessments (this was
the position strongly advocated by Lionel Robbins), then the assignment
of individual utilities for use in social aggregation could itself be seen as
involving aggregation of different individuals’ assessments of people’s utili-
ties (see Roberts 1995).
Capabilities and Primary Goods
The main problem with relying on mental state comparisons may not be
at least their allegedly exclusive rele-
their feasibility but their relevance—
being
vance in social choice. There are many difficulties in judging the well-
of a person by his or her mental state. Utilities may sometimes be very
malleable in response to persistent deprivation. A hopeless destitute, or a
downtrodden laborer living under inescapably exploitative arrangements,
or a subjugated housewife in a society with entrenched gender inequality,
or a tyrannized citizen under brutal authoritarianism may come to terms
with her deprivation. She may take whatever pleasure she can from small
achievements and adjust her desires to take note of feasibility (thereby
helping the fulfillment of her downwardly adjusted desires). But her success
in such adjustments will not make her deprivation go away. The metric of
pleasure or desire may sometimes quite inadequately reflect the extent of a
person’s substantive deprivation.
There may indeed be a case for taking incomes, commodity bundles,
or resources more generally to be of direct interest in judging a person’s
advantage. The interest in incomes or resources can arise for many different
not merely for the mental states that opulence may help generate.
reasons—
68 Amartya Sen
In fact, the Difference principle in Rawls’s (1971) theory of “justice as fair-
ness” is based on judging individual advantage in terms of a person’s com-
purpose
mand over what Rawls calls “primary goods,” which are general-
resources that are useful for anyone to have (no matter what her exact
objectives are).
This procedure can be improved on by taking note not only of the hold-
ings of primary goods and resources, but also of interpersonal differences in
converting them to the capability to live well. Indeed, I have tried to argue
in favor of judging individual advantages in terms of the respective capa-
bilities that the person has reason to value, on which, see Sen (1980, 1985a,
1985b) and Nussbaum (1988, 1992, 2000, 2001, 2011). This approach
focuses on the substantive freedoms that people have rather than only on
the particular outcomes they obtain. For responsible adults, the concentra-
tion on freedom rather than only on achievement has some merit, and it
can provide a general framework for analyzing individual advantage and
deprivation in a contemporary society.
Normative Measurement
The variety of information on which social welfare analysis can draw can be
well illustrated by the study of poverty and the battle against it. The intel-
lectual challenges involved in what Angus Deaton (2013) has called “the
great escape” are as important to the subject of social choice as they are
central to the basic engagements of the social sciences in general.
In the standard measurement literature, poverty is typically seen in terms
of the lowness of incomes, and it has been traditionally measured simply
line income; this is
by counting the number of people below the poverty-
count measure.” A scrutiny of this approach,
sometimes called the “head-
which has been an important part of contemporary social choice literature,
yields two different types of questions. First, is it adequate to see poverty
as equivalent to lowness of income? Second, even if poverty is seen as low
income, is the aggregate poverty of a society best characterized by some
count measure of the number falling below the chosen
index of the head-
cut-off poverty-line income?
I take up these questions in turn. Do we get enough of a diagnosis of
individual poverty by comparing the individual’s income with a socially
Social Choice and Welfare Economics 69
line income? What about the person with an income well
given poverty-
above the poverty line, who suffers from an expensive illness (requiring,
say, kidney dialysis)? Is deprivation not ultimately a lack of opportunity to
lead a minimally acceptable life, which can be influenced by a number of
considerations, including of course personal income but also physical and
environmental characteristics, and other variables, related to, say, epidemi-
ological conditions of a person’s regional location. It has been argued that
poverty can be more sensibly seen as a serious deprivation of certain basic
capabilities. This alternative approach leads to a rather different diagnosis
based analysis can yield.
of poverty from the ones that a purely income-
This is not to deny that lowness of income can be very important in
many contexts, because the opportunities a person enjoys in a market
economy can be severely constrained by her level of real income.6 However,
various contingencies can lead to variations in the “conversion” of income
into the capability to live a minimally acceptable life. And if that is what
we are concerned with, there may be good reasons to look beyond income
poverty (see Sen 1984, 1992; Foster and Sen 1997) without ignoring the
income information. There are at least four different sources of variation:
(1) personal heterogeneities (for example, disability or proneness to illness),
prone or flood-
(2) environmental diversities (for example, living in a storm-
prone area), (3) variations in social climate (for example, the prevalence of
crime or epidemiological challenges), and (4) differences in relative depri-
vation connected with customary patterns of consumption in particular
societies (for example, being relatively impoverished in terms of income in
a rich society can lead to deprivation of the absolute capability to take part
a point that was made with compelling force
in the life of the community—
by Adam Smith (1776)).
I turn now to the second question. The most common and most tradi-
tional measure of poverty had tended to concentrate on head counting.
But it must also make a difference as to how far below the poverty line the
poor individually are, and furthermore, how the deprivation is shared and
distributed among the poor. The social data on the respective deprivations of
the individuals who constitute the poor in a society need to be aggregated
6. These issues have been insightfully scrutinized by Philippe Van Parijs (1995).
70 Amartya Sen
to arrive at informative and usable measures of aggregate poverty. This is a
social choice problem, and axioms can indeed be proposed that attempt to
capture our distributional concerns in this constructive exercise.7
Among the new developments in the field are multidimensional measures
of poverty and inequality, powerfully pursued in different forms by Atkinson
and Bourguignon (1982), Alkire and Foster (2011a, 2011b), and others.8 To
understand poverty and inequality, a strong case can be made for looking
at real deprivation and not merely at mental reactions to that deprivation.
The point has been brought out particularly clearly by recent investigations
of gender inequality that focus not just on happiness or unhappiness but
also on women’s deprivation in terms of undernutrition; clinically diagnosed
morbidity; observed illiteracy; even unexpectedly high mortality (compared
with physiologically justified expectations); and in an anticipatory context,
specific abortion of female fetuses.
sex-
and
Multidimensional interpersonal comparisons can be sensibly—
accommodated in a broad framework of welfare econom-
comfortably—
ics and social choice theory, enhanced by the removal of informational
constraints that are explicitly invoked or implicitly imposed in traditional
welfare economics.
A Closing Remark
Broadening of the informational basis has become a major concern in mod-
ern social choice theory. This applies, first of all, to addressing Arrow’s impos-
sibility result. Second, it is central to being inequality sensitive in welfare
economics. Third, it is relevant to being liberty conscious in politics, law,
and the pursuit of human rights. Fourth, it is especially important for having
being of people.
better informed normative measurement of the well-
7. I will not survey here the huge axiomatic literature on this subject. The measure of
poverty on the income space in Sen (1976) can, in fact, be improved by an important
but simple variation illuminatingly proposed by Anthony F. Shorrocks (1995). I have
to confess favoring the “Sen- Shorrocks measure” over the original “Sen index.” See
also Foster and Sen (1997).
8. See also Kolm (1977), Maasoumi (1986), Alkire et al. (2015), and Maasoumi and
Racine (2016), among many other contributions to the rich literature on multidi-
mensional aggregation in the context of the measurement of inequality and poverty.
Social Choice and Welfare Economics 71
As has been discussed and illustrated in different contexts in this chap-
ter, the reasoned use of appropriate information involves both epistemol-
ogy and ethics. More engagement in each is crucially important for further
progress in social choice and welfare economics.
References
Alkire, Sabina, and James E. Foster. 2011a. “Counting and Multidimensional Poverty
Measurement.” Journal of Public Economics 95 (7– 487.
8): 476–
Alkire, Sabina, and James E. Foster. 2011b. “Understandings and Misunderstand-
ings of Multidimensional Poverty Measurement.” Journal of Economic Inequality
314.
9 (2): 289–
Alkire, Sabina, James E. Foster, Suman Seth, Maria Emma Santos, Jose Manuel Roche,
and Paola Ballon. 2015. Multidimensional Poverty Measurement and Analysis. Oxford:
Oxford University Press.
Arrow, Kenneth J. 1950. “A Difficulty in the Concept of Social Welfare.” Journal of
346.
Political Economy 58 (4): 328–
Arrow, Kenneth J. 1951. Social Choice and Individual Values. New York: Wiley.
Arrow, Kenneth J. 1963. Social Choice and Individual Values, second edition. New
York: Wiley.
Arrow, Kenneth J. 1977. “Extended Sympathy and the Possibility of Social Choice.”
225.
American Economic Review 67 (1): 219–
Atkinson, Anthony B., and François Bourguignon. 1982. “The Comparison of Multi-
Dimensioned Distributions of Economic Status.” Review of Economic Studies 49 (2):
183–201.
Banerjee, Abhijit V., and Esther Duflo. 2011. Poor Economics: A Radical Rethinking of
the Way to Fight Global Poverty. New York: Public Affairs.
Baumol, William J. 1965. Welfare Economics and the Theory of the State, second ed.
Cambridge, MA: Harvard University Press.
Bentham, Jeremy. 1789. An Introduction to the Principles of Morals and Legislation.
London.
Bergson, Abram. 1938. “A Reformulation of Certain Aspects of Welfare Economics.”
334.
Quarterly Journal of Economics 52 (2): 310–
Black, Duncan. 1948. “The Decisions of a Committee Using a Special Majority.”
261.
Econometrica 16 (3): 245–
72 Amartya Sen
Black, Duncan. 1958. The Theory of Committees and Elections. Cambridge: Cambridge
University Press.
Blackorby, Charles, David Donaldson, and John A. Weymark. 1984. “Social Choice
with Interpersonal Utility Comparisons: A Diagrammatic Introduction.” Inter
356.
national Economic Review 25 (2): 327–
Charles de. 1781. Memoire sur les Elections au Scrutin. Memoires de
Borda, Jean-
l’Academie Royal des Sciences. Translated by A. de Grazia. 1953. Isis 44 (1/2):
42–51.
Buchanan, James M. 1954. “Individual Choice in Voting and the Market.” Journal of
343.
Political Economy 62 (4): 334–
Condorcet, Nicolas de. 1785. Essai sur l’Application de l’ Analyse à la Probabilité des
Decisions Rendues à la Pluralité des Voix. Paris: L’ Impremerie Royale.
Condorcet, Nicolas de. 1847. Ouvres de Condorce, edited by A. Condorcet O’Connor
and M. F. Arago, vol X1. Paris.
Dasgupta, Partha, and Eric Maskin. 2008. “On the Robustness of Majority Rule.”
973.
Journal of the European Economic Association 6 (5): 949–
d’Aspremont, Claude. 1985. “Axioms for Social Welfare Orderings.” In Social Goals
and Social Organization: Essays in Memory of Elisha Pazner, edited by Leonid Hurwicz,
David Schmeidler, and Hugo Sonnenschein, 19– 76. Cambridge: Cambridge Univer-
sity Press.
d’Aspremont, Claude, and Louis Gevers. 1977. “Equity and the Informational Basis
209.
of Collective Choice.” Review of Economic Studies 44 (2): 199–
d’Aspremont, Claude, and Philippe Mongin. 1998. “Utility Theory and Ethics.” In
Handbook of Utility Theory, edited by Salvador Barberá, Peter J. Hammond, and Chris-
tian Seidl, Volume 1, 371– 481. Dordrecht: Kluwer Academic.
Davidson, Donald. 1986. “Judging Interpersonal Interests.” In Foundations of Social
211. Cambridge: Cam-
Choice Theory, edited by Jon Elster and Aanund Hylland, 195–
bridge University Press.
Deaton, Angus S. 2013. The Great Escape: Health, Wealth, and the Origins of Inequality.
Princeton, NJ: Princeton University Press.
Dodgson, Charles L. 1876. A Method of Taking Votes on More Than Two Issues. Oxford:
Oxford University Press.
Dodgson, Charles L. 1884. The Principles of Parliamentary Representation. London:
Harrison and Sons.
Edgeworth, Francis T. 1881. Mathematical Psychics. London: Kegan Paul.
Social Choice and Welfare Economics 73
Foster, James E., and Amartya K. Sen. 1997. “On Economic Inequality after a Quarter
Century.” In Economic Inequality by Amartya K. Sen, expanded edition, 107– 219.
Oxford: Oxford University Press.
Gaertner, Wulf, Prasanta K. Pattanaik, and Kotaro Suzumura. 1992. “Individual
177.
Rights Revisited.” Economica 59 (234): 161–
Gevers, Louis. 1979. “On Interpersonal Comparability and Social Welfare Order-
89.
ings.” Econometrica 47 (1): 75–
Gibbard, Allan F. 1973. “Manipulation of Voting Schemes: A General Result.” Econo-
601.
metrica 41 (4): 587–
Hammond, Peter J. 1976. “Equity, Arrow’s Conditions, and Rawls’ Difference Prin-
804.
ciple.” Econometrica 44 (4): 793–
Hammond, Peter J. 1996. “Consequentialism, Structural Rationality and Game
Theory.” In The Rational Foundations of Economic Behaviour. Proceedings of the IEA
Conference held in Turin, Italy, edited by Kenneth J. Arrow, Enrico Colombatto, Mark
Perlman, and Christian Schmidt, 25– 42. Basingstoke, UK: Macmillan
Harsanyi, John C. 1955. “Cardinal Welfare, Individual Ethics, and Interpersonal
321.
Comparison of Utility.” Journal of Political Economy 63 (4): 309–
Kahneman, Daniel. 1999. “Objective Happiness.” In Well-
Being: Foundations of
Hedonic Psychology, edited by Daniel Kahneman, Edward Diener, and Norbert
25. New York: Russell Sage Foundation.
Schwarz, 3–
Kahneman, Daniel. 2000. “Evaluation by Moments: Past and Future.” In Choices,
692. Cam-
Values, and Frames, edited by Daniel Kahneman and Amos Tversky, 673–
bridge: Cambridge University Press.
Kolm, Serge-Christophe. 1977. “Multidimensional Egalitarianism.” Quarterly Journal
3.
of Economics 91(1): 1–
Krueger, Alan B., ed. 2009. Measuring the Subjective Well-
Being of Nations: National
Being. Chicago: University of Chicago Press.
Accounts of Time Use and Well-
Krueger, Alan B., and Arthur A. Stone. 2014. “Progress in Measuring Subjective Well-
43.
Being.” Science 346 (6205): 42–
List, Christian. 2005. “The Probability of Inconsistencies in Complex Collective
32.
Decisions.” Social Choice and Welfare 24 (1): 3–
List, Christian, and Philip Pettit. 2002. “Aggregating Sets of Judgments: An Impos-
110.
sibility Result.” Economics and Philosophy 18 (1): 89–
Maasoumi, Esfandiar. 1986. “The Measurement and Decomposition of Multi-
997.
Dimensional Inequality.” Econometrica 54 (4): 991–
74 Amartya Sen
Maasoumi, Esfandiar, and J. S. Racine. 2016. “A Solution to Aggregation and an
Being’ Frontiers.” Journal of Econometrics 191
Application to Multidimensional ‘Well-
(2): 374–383.
Marshall, Alfred. 1890. Principles of Economics. London: Macmillan.
Maskin, Eric. 1978. “A Theorem on Utilitarianism.” Review of Economic Studies 46 (4):
93–96.
Making under Ignorance with Implications for Social
Maskin, Eric. 1979. “Decision-
337.
Choice.” Theory and Decision 11 (3): 319–
Maskin, Eric. 1985. “The Theory of Implementation in Nash Equilibrium: A Survey.”
In Social Goals and Social Organization: Essays in Memory of Elisha Pazner, edited by
Leonid Hurwicz, David Schmeidler, and Hugo Sonnenschein, 173– 204. Cambridge:
Cambridge University Press.
Maskin, Eric. 1995. “Majority Rule, Social Welfare Functions, and Games Forms.” In
Choice, Welfare, and Development: A Festschrift in Honour of Amartya K. Sen, edited by
Kaushik Basu, Prasanta P. Pattanaik, and Kotaro Suzumura, 100– 109. Oxford: Oxford
University Press.
Maskin, Eric. 2014. “The Arrow Impossibility Theorem: Where Do We Go from
Here?” In The Arrow Impossibility Theorem, edited by Eric Maskin and Amartya K.
55. New York: Columbia University Press.
Sen, 43–
Maskin, Eric, and Tomas Sjöström. 2002. “Implementation Theory.” In Handbook of
Social Choice and Welfare, edited by Kenneth J. Arrow, Amartya K. Sen, and Kotaro
Suzumura, 237– 288. Amsterdam: Elsevier.
Mill, John Stuart. [1859] 1959. On Liberty. New York: Gateway Editions.
Nozick, Robert. 1974. Anarchy, State, and Utopia. New York: Basic Books.
Nussbaum, Martha C. 1988. “Nature, Function, and Capability: Aristotle on Political
184.
Distribution.” Oxford Studies in Ancient Philosophy (Supplementary Volume): 145–
Nussbaum, Martha C. 1992. “Human Functioning and Social Justice: In Defense of
246.
Aristotelian Essentialism.” Political Theory 20 (2): 202–
Nussbaum, Martha C. 2000. Women and Human Development. New York: Cambridge
University Press.
Nussbaum, Martha C. 2001. “Disabled Lives: Who Cares?” New York Review of Books,
37.
January 11, 34–
Nussbaum, Martha C. 2011. Creating Capabilities. Cambridge, MA: Harvard Univer-
sity Press.
Pattanaik, Prasanta K. 1973. “On the Stability of Sincere Voting Situations.” Journal
574.
of Economic Theory 6 (6): 558–
Social Choice and Welfare Economics 75
Holland.
Pattanaik, Prasanta K. 1978. Strategy and Group Choice. Amsterdam: North-
Phelps, Edmund S. 1973. Economic Justice: Selected Readings. Harmondsworth, UK:
Penguin.
Pigou, Arthur C. 1920. The Economics of Welfare. London: Macmillan.
Rawls, John 1971. A Theory of Justice. Cambridge, MA: Harvard University Press.
Robbins, Lionel. 1938. “Interpersonal Comparisons of Utility: A Comment.” Eco-
641.
nomic Journal 48 (192): 635–
Roberts, Kevin W. S. 1980a. “Interpersonal Comparability and Social Choice
439.
Theory.” Review of Economic Studies 47 (2): 421–
Roberts, Kevin W. S. 1980b. “Possibility Theorems with Interpersonally Comparable
420.
Welfare Levels.” Review of Economic Studies 47 (2): 409–
Roberts, Kevin W. S. 1995. “Valued Opinions or Opinionated Values: The Double
Aggregation Problem.” In Choice, Welfare, and Development: A Festschrift in Honour of
Amartya K. Sen, edited by Kaushik Basu, Prasanta P. Pattanaik, and Kotaro Suzumura,
109. Oxford: Oxford University Press.
100–
Samuelson, Paul A. 1947. Foundations of Economic Analysis. Cambridge, MA: Harvard
University Press.
Satterthwaite, Mark A. 1975. “Strategy- Proofness and Arrow’s Conditions: Existence
and Correspondence Theorems for Voting Procedures and Social Welfare Func-
217.
tions.” Journal of Economic Theory 10 (2): 187–
Sen, Amartya K. 1966. “A Possibility Theorem on Majority Decisions.” Econometrica
499.
34 (2): 491–
Sen, Amartya K. 1970a. Collective Choice and Social Welfare. San Francisco: Holden Day.
Sen, Amartya K. 1970b. “The Impossibility of a Paretian Liberal.” Journal of Political
157.
Economy 78 (1): 152–
Sen, Amartya K. 1976. “Poverty: An Ordinal Approach to Measurement.” Economet-
231.
rica, 44 (2): 219–
Sen, Amartya K. 1977a. “On Weights and Measures: Informational Constraints in
1572.
Social Welfare Analysis.” Econometrica 45 (7): 1539–
examination.” Econometrica 45
Sen, Amartya K. 1977b. “Social Choice Theory: A Re-
(1): 53–89.
Sen, Amartya K. 1979. “Personal Utilities and Public Judgements: Or What’s Wrong
with Welfare Economics.” Economic Journal 89 (355): 537–558.
Sen, Amartya K. 1980. “Equality of What?” In The Tanner Lectures on Human Values,
220. Cambridge: Cambridge University Press.
volume I: 197–
76 Amartya Sen
Sen, Amartya K. 1982. Choice, Welfare, and Measurement. Oxford: Blackwell.
Sen, Amartya K. 1984. Resources, Values, and Development. Cambridge, MA: Harvard
University Press.
Holland.
Sen, Amartya K. 1985a. Commodities and Capabilities. Amsterdam: North-
Sen, Amartya K. 1985b. “Well- Being, Agency and Freedom: The Dewey Lectures
221.
1984.” Journal of Philosophy 82 (4): 169–
Sen, Amartya K. 1992. Inequality Reexamined. Cambridge, MA: Harvard University Press.
521.
Sen, Amartya K. 1993. “Internal Consistency of Choice.” Econometrica 61 (3): 495–
Sen, Amartya K., and Prasanta K. Pattanaik. 1969. “Necessary and Sufficient Condi-
tions for Rational Choice under Majority Decision.” Journal of Economic Theory 1 (2):
178–202.
Shorrocks, Anthony F. 1995. “Revisiting the Sen Poverty Index.” Econometrica 63 (5):
1225–1230.
Smith, Adam. 1776. An Inquiry into the Nature and Causes of the Wealth of Nations.
London: George Routledge and Sons.
4):
Suppes, Patrick. 1966. “Some Formal Models of Grading Principles.” Synthese 16 (3–
284–306.
Suzumura, Kotara. 1983. Rational Choice, Collective Decisions, and Social Welfare.
Cambridge: Cambridge University Press.
Suzumura, Kotara. 1997. “Interpersonal Comparisons of the Extended Sympathy
Type and the Possibility of Social Choice.” In Social Choice Re- Examined. Proceedings
of the International Economic Association Conference held at Schloss Hernstein, Berndorf,
Vienna, Austria, edited by Kenneth J. Arrow, Amartya K. Sen, and Kotaro Suzumura,
202–229. London and New York: Palgrave Macmillan.
Van Parijs, Philippe. 1995. Real Freedom for All: What (If Anything) Can Justify Capital-
ism? Oxford: Oxford University Press.
Comment: Célestin Monga
The Economy of Tastes, Feelings, and Opinions
I still remember vividly the strange mix of excitement and bewilderment
that overwhelmed me in my high school years when our professor of
cost analysis. I immedi-
accounting taught us the fundamentals of benefit-
ately went to my dormitory and spent most of the evening trying to apply
this powerful technique, not to assess whether the advantages of a hypo-
thetical investment project were likely to outweigh its drawbacks, but to
cost analysis seemed like a rigor-
evaluate my own life prospects. Benefit-
ous and revealing tool to examine whether my minuscule and uncertain
existence was a “profitable” venture, or at least a worthwhile escapade that
deserved to be continued. Of course, the few friends to whom I confided
costs anal-
this found it a ludicrous idea. They reminded me that a benefit-
ysis is always controversial, even when circumscribed to real investment
decisions or to public policies. They were right: Applying it to one’s life
opened even more unresolved conceptual questions. But so what?
I kept running the numbers. To ascertain the net effect of an imaginary
being, I first had
list of positive and negative changes to come in my well-
to come up with a way of measuring the gains and the losses. The identified
benefits and costs, even though they were expressed in monetary terms,
went well beyond changes in my projected individual income: My well-
being was to be affected positively or negatively by nonmonetary factors,
whether linked to my individual and personal preferences or related to the
being of people around me (social benefits and costs).
well-
I also had to decide how to imagine and estimate the prospective ben-
efits and costs of my entire life to come. Using my own personal value scale,
78 Comments by Célestin Monga and James E. Foster
I calculated the costs as the amount of compensation required to exactly
offset negative consequences of being alive for the 50 years or so of life
expectancy ahead. The compensation required was the monetary amount
that would leave me just as well off as before engaging in this exercise.
Benefits were measured by my willingness to stay alive and enjoy all the
things and emotions that I could reasonably expect for the decades ahead.
Knowing that, in the end, life always results in death, typically following
either an abrupt and tragic event like a car or airplane crash, or a long and
painful illness, I could not find many benefits whose present and expected
value could match and compensate for the pains and disappointments of
cost analysis were not very promising:
the costs. The results of my benefit-
Taking into consideration all current and expected streams of good and bad
news, life did not appear to be a “profitable” investment.
Shocked by the outcomes, I quickly did some sensitivity analyses to
check the robustness of the findings: No matter what discount rates I
chose, the calculations still yielded disappointing numbers to the question
of whether life was a worthwhile venture. This was all the more puzzling,
because I actually loved many aspects of my life. Not knowing what to do
with the analyses, I concluded that one should either doubt the validity of
certain measurement instruments and our ability to use them “objectively,”
or radically give more weight to whatever we define as “positive” outcomes
for our actions or inactions, or accept the very probable hypothesis that
happiness may be an illusion but those who choose to live should learn to
ignore its downsides. I could only forget the outcomes of my own study by
learning to radically change whatever assumptions I used in carrying it out.
“Life is impossible without the ability to forget,” philosopher Emil Cioran
once said. But some memories are just too long lasting to be erased.
cost analysis today, even with the same
Carrying out the same benefit-
elements and discount rates, would obviously yield different results. This
illustrates some of the truly challenging conceptual problems at the heart
being, whether it is approached through the lens of
of the study of well-
welfare, utility, or the standard of living of one individual. The challenges
are even more formidable when one tries to assess not just the perspectives
and preferences of one person but also the social preferences of people in a
group; then one has to aggregate and make sense of the various viewpoints
of all members of the society. The complexities are not just “technical” or
after all, these can be addressed with carefully designed
methodological—
Social Choice and Welfare Economics 79
quantitative frameworks and clearly formulated assumptions; they also
involve ethical and psychological issues that do not fit nicely in any linear
models of aggregative social choice theory.
I should not have been surprised to feel lost trying to determine and assess
the validity of my own present and future welfare. Cioran also warned
about the dangers of loving oneself, which is falling in love with someone
about whom we know nothing. If capturing one’s own utility, welfare, and
standard of living is so challenging, how about doing the same exercise at
the level of a group or society? The instability of my preferences and of my
own subjectivity, the constantly changing moods and mental states, and
the inability to even decide for myself what my objective functions are or
cost analysis was unsatisfac-
should be explain why my schematic benefit-
tory and inconclusive. These problems are compounded when one gets to
the level of social aggregation. How would one confidently compute and
aggregate individual tastes and opinions that are moving targets? What is
making, both at the individual level
the right approach to ethical decision-
and at the social/aggregative level? And what are the appropriate ethical
stances for comparative analyses of such scope?
Central to the general topic of social aggregation is the issue of interper-
being, which has preoccupied economists, social
sonal comparisons of well-
scientists, and philosophers for centuries. At least three types of problems
must be addressed to elaborate intellectual and policy frameworks for mak-
ing socially acceptable decisions. One must obviously start with valid meth-
ods for defining, understanding, capturing, and measuring the notion of
being. Second, these methods should be extended to social
individual well-
groups in ways that make them meaningful and credible. Third, one should
remember that the very purpose for carrying out such an exercise may affect
the answers to the two initial questions posed (Elster and Roemer 1991). All
this supposes that individual preferences can be measured at a satisfactory
level of confidence that the intrinsic subjectivity in such exercises are more
than compensated by objectivity in the methods used.
The various steps that one must go through (from theory to specific
concepts and empirical strategies) are therefore both daunting and exciting.
Not surprisingly, many of the most creative minds in economics have tried
to climb that mountain, a task that requires not only using the traditional
quantitative tools of economics but also taking stock of the relevant find-
ings of philosophy, psychology, and even biology. Amartya Sen’s chapter,
80 Comments by Célestin Monga and James E. Foster
“Social Choice and Welfare Economics,” which builds on several important
previous contributions (most notably Sen 1970), is the latest attempt to
do so. As always with Sen, the reader is taken on an erudite and insightful
journey, intellectually challenging but always rewarding. Before offering a
summary exposition of his bold thesis, let me provide an initial overview
of some of the elements of the puzzle that he heroically tries to assemble.
My comment offers a brief reassessment of the elements of the debate.
Section 1 summarizes the intellectual progress made by economists in
the search for a valid social choice theory and outlines a few aspects of
Amartya Sen’s new contribution on the topic. Section 2 discusses some of
the remaining ethical questions and urges economists to be more attuned
to the research findings in the other social sciences and the humanities.
Section 3 offers a few concluding remarks.
Beyond Utilitarian Calculus: Amartya Sen’s Bold Thesis
How to assess and report our own pleasures, utility, state of mind, and opin-
ions? How to make individual and collective choices? How to prioritize
and rank them? And how to compare and aggregate our selections with
those of other people in a credible and legitimate social welfare function?
How should we make collective decisions that reflect optimally the prefer-
so that they can all live,
ences and welfare of everyone in a social group—
if not happily, at least with the feeling that the decisions are made in ways
that are acceptable to everyone? Underpinning these questions of social
aggregation of utility, tastes, and opinions is the issue of interpersonal com-
being, which has preoccupied economists, social scientists,
parisons of well-
and philosophers for centuries. Various waves of research on the topic
have basically identified several types of problems that must be addressed
to elaborate an intellectual framework for making socially acceptable deci-
sions. Such frameworks obviously start with valid methods for defining,
understanding, capturing, and measuring individual preferences and then
extend them to social groups, with a satisfactory level of confidence that
subjectivity is more than compensated by objectivity.
Capturing one’s feelings and converting them into indicators of welfare
or utility, measuring them and aggregating opinions from groups of people
have long been challenging questions for researchers. In an introduction to
one of his books, Jevons ([1871] 1970, 85) warned that:
Social Choice and Welfare Economics 81
The reader will find again, that there is never, in any single instance, an attempt
to compare the amount of feeling in one mind with that in another. I see no
means by which such comparisons can be accomplished. … Every mind is thus
inscrutable to every other mind, and no common denominator of feeling seems
to be possible.
Economists followed suit and showed a strong reluctance to carry out
interpersonal comparisons of utility that were forcefully promoted by
logical positivists. The economists justified their position by arguing that
ethical statements were always unverifiable and therefore lacked scientific
see Ayer ([1936] 1971).
foundations—
Utilitarian economists were particularly adamant in their opposition to
interpersonal comparisons of utility, arguing that it is unsound to make use of
interpersonal comparisons of individual utilities. Jeremy Bentham, the lead-
ing proponent of such utilitarian calculus, was concerned only with maxi-
mizing the total utility of a community, irrespective of its distribution. Even
the early critics of utilitarianism thought that interpersonal comparisons of
utility had no scientific basis: “Every mind is inscrutable to every other mind
and no common denominator of feelings is possible” (Robbins 1938, 636).
Such views were rooted in logical positivism, also called logical empiricism, a
philosophical movement that emerged in Vienna in the 1920s and consid-
ered scientific knowledge to be the only kind of factual knowledge.
The general reluctance of researchers to move to that terrain led to major
intellectual impasses in both social choice theory and welfare economics.
Although positive economics could be carried out without interpersonal
comparisons of utility, social choice theory without interpersonal compari-
sons of utility could not go very far: The scope of normative economics and
welfare economics was basically limited to theoretical developments con-
cerning the identification of Pareto efficient outcomes or Pareto improve-
ments to existing economic situations. “Traditional comparisons of utility
have to be made if there is to be any satisfactory escape from Arrow’s Impos-
sibility theorem,” notes Hammond (1991, 235). But the lingering funda-
mental question raised by the logical positivists had to be answered: How
can one rigorously construct an interpersonally comparable utility function?
Starting in the 1950s, economists, mathematicians, and philosophers
took up the task. Alternative methods of making different forms of interper-
sonal comparisons of utility were offered by several researchers, with various
degrees of complexity and success. The really exciting intellectual journey
82 Comments by Célestin Monga and James E. Foster
in the quest for a more convincing social welfare function was launched by
Arrow ([1951] 1963), who put social choice theory in its modern, fully axi-
omatized form. He tried to identify the most valid procedures for deriving a
collective or “social ordering” of the alternatives (from better to worse) from
people’s preferences. His search for a “general possibility” theorem, as he
no single
called it, led to the conclusion that it was in fact an impossibility—
procedure could satisfy a few straightforward assumptions concerning the
autonomy of the agents and the rationality of their preferences.
Several generations of researchers subsequently attempted to modify
Arrow’s requirements and come up with a solution to the impossibility
theorem (see Maskin and Sen 2014). Generally these solutions led to other
difficulties. This research quickly became a journey into the dilemmas and
challenges of normative ethics and how economics has struggled with them.
It strongly focused on discussions of utilitarianism, understood in its generic
definition as the view that the morally right actions are those that generate
the most good, with the implication that the social good is the sum of the
assuming that the latter are interperson-
welfares of individuals in a group—
ally comparable. Harsanyi (1953, 1955, 1977) provided the most debated axi-
omatic arguments in support of utilitarianism. His work set the stage for the
issues of utility and preferences as seen by economists and mathematicians,
and it suggested a framework for modeling moral value judgments.
Harsanyi’s main insight has been to imagine an impartial observer who
can determine a social ordering of the existing alternatives faced by all
members of a given group or society. Although detached from the group,
the observer in question is also sympathetic to its concerns, and he imag-
ines how he would determine a social ordering of the available alternatives
based on an impartial attitude toward the interests of all members of the
group. The neutral observer imagines how he would assess the various alter-
natives if he were in the shoes of, say, individual i, with i’s objective circum-
stances, tastes, and opinions. Harsanyi makes two additional and important
suppositions: The impartial observer has preferences about these hypotheti-
cal alternatives that satisfy the expected utility axioms,1 and these prefer-
1. The von Neumann– Morgenstern axioms of the expected utility theory that define
a rational decision-maker are as follows: completeness, which assumes that an indi-
vidual has a set of well-defined preferences and can always decide between any two
alternatives; transitivity, which assumes consistency in the decision-making of the
Social Choice and Welfare Economics 83
Morgenstern utility function. It is
ences are represented by a von Neumann–
also assumed that the observer (who plays the role of and seeks the interests
of society as a whole) respects the orderings of social alternatives by the
individuals. With the adoption of the impartial perspective, the resulting
judgments computed from the observer’s utility can be considered moral
judgments, as they give equal consideration to the interests of each person
in the group.2 Harsanyi used this framework to elaborate aggregation and
impartial individual theorems with strong assumptions: the existence of a
single profile of individual preference orderings and of a single social prefer-
ence ordering of a set of social alternatives (consisting of all lotteries that
can be generated from a finite set of alternatives).
Harsanyi’s approach is based on the notion of “impersonality,” which
posits that it is possible for an ethical observer of any situation to free him-
self from selfish perspectives when weighting moral issues by pretending to
be entirely uncertain about which individual the oberver will become after
the issue has been decided. In sum, one should be willing and capable of
becoming somebody else completely: This is a clever device, comparable
to Hare’s (1963) principle of “universalizability” and Rawls’s (1971) notion
of the “veil of ignorance.” These ideas paved the way for other influen-
tial approaches, which recommended inferring interpersonal comparisons
from different aspects of the behavior of individuals. Yet in the end, such
behaviorist empirical methods were often found to be unsatisfactory, as
they typically required ethical judgments and also led to normative state-
ments that could not be made from empirical observations alone.
Then came Amartya Sen, the most daring theorist among those who
have studied the issues surrounding the rationality of economic agents
from various angles. In this chapter, he revisits the theme but approaches
it obliquely and offers a comprehensive analytical framework for interper-
sonal comparisons. One obvious and striking feature of the chapter is its
individual; independence, which assumes that two lotteries mixed up with an irrele-
vant third one will maintain the same order of preference as when the two initial lot-
teries are presented independently of the third one; and continuity, which assumes
that when there are three lotteries (1, 2, and 3) and the individual prefers 1 to 2 and 2
to 3, then there should be a possible combination of 1 and 3 in which the individual
is indifferent between this particular mix and lottery 2.
2. See Weymark (1991) for an excellent discussion.
84 Comments by Célestin Monga and James E. Foster
style: Sen’s prose is always very precise, soft, and elegant. It constantly keeps
the reader in focus, even when the issues discussed are technically demand-
ing. Sen is also a master at challenging erroneous ideas without ruffling
feathers. It can be said about him what is often said about former US sena-
tor Joseph Lieberman: “He is so elegant in his criticism of his opponents
that even if he tells you to go to Hell, you would actually enjoy the ride!”
Sen begins with a reexamination of some old questions in the theory of
making, which he traces back to Jean-
collective decision- Charles de Borda
(1781) and de Condorcet (1785). Sen’s deconstruction of the problem at
hand starts as follows: Suppose a group of people is facing some alterna-
tives to choose among (such as candidates in an election, policy options,
projects and programs, and distribution of income). How does one make
acceptable social decisions for a group (such as a nation, or a community,
or any other collectivity) in a way that the diverse views and interests of
members of the group all receive attention and importance? How does one
go from individual preferences over different states of affairs to a social pref-
erence over those states, reflecting an “aggregation” of the points of views
of all members of the society?
In fact, Sen had attempted to answer these questions in many previous
works. He gracefully fired multiple salvos to some of the earlier theories
of and approaches to social welfare (Sen 1970, 1977, 1986). Building on
Arrow’s work, Sen did not hesitate to question it, but with elegance and
he always did it in homeopathic doses, relaxing assumptions
admiration—
here, delicately challenging the rigidity of the impossibility theorem there,
or taking the tangent whenever he believed that his predecessors’ frame-
works were erroneous. Sen’s analyses have brought new hope to the search
for rational social choice procedures based on individuals’ own preferences.
Sen begins the chapter with the acknowledgment that there is not going
to be any perfect resolution of the social choice dilemmas of the kind iden-
tified by Arrow through voting procedures. He rejects the notion that they
based procedures are entirely natural
can be used in all situations: “Voting-
for some kinds of social choice problems, such as elections, referendums,
or committee decisions. They are, however, altogether unsuitable for many
other problems of social choice.”
Sen’s reasoning is logical: If it is true that there are no faultless voting pro-
cedure out there to be found, the next logical question is whether some of
them could yield better results than others. And by the way, is voting itself
Social Choice and Welfare Economics 85
a good method to resolve social choice problems of all kinds? Didn’t Win-
ston Churchill famously say that “The best argument against democracy is
minute conversation with the average voter?” (Priest 2017, 3). Sen is
a five-
an optimistic economist: He is skeptical of the traditional welfare econom-
ics developed by the utilitarian researchers. He is very confident that inter-
personal utility can be measured satisfactorily. He challenges the historical
consensus against the use of interpersonal comparisons in social choice.
Sen’s recommendation is bold and hopeful: One must go beyond the
class of voting rules (studied by Borda, Condorcet, and Arrow) to address
distributional issues, particularly in welfare economics. The decision to
reject the philosophical basis of logical positivism and to believe instead,
like philosopher Donald Davidson, that people can understand and relate
to other people’s minds and feelings only by making some comparisons
with their own minds and feelings, allows new ways of thinking about
social choice. Then Arrow’s impossibility theorem and its related results
just go away when different kinds of interpersonal comparisons are used in
social welfare judgments.
Sen observes that each kind of comparability requires a particular way of
combining welfare numbers of different people in a group. Of course, such
comparisons need not be very precise before they can be used systemati-
cally in social choice. He writes:
We may be able to make interpersonal comparisons to some extent, but not in
every comparison, nor of every type, nor with tremendous exactness. … It can also
be shown that terribly refined interpersonal comparisons may not be needed for
arriving at definite social decisions. Quite often, rather limited levels of partial
comparability will be adequate for making social decisions.
A very clever way of using minimalism to achieve maximum intellectual
impact, indeed.
Beyond Aggregation Techniques: Some Ethical Challenges
one that
Developing a legitimate framework for making social decisions—
accounts “democratically” for the preferences and interests of the members
is likely to remain an elusive
of the group or society under consideration—
quest. It requires much more than an intellectual consensus on the mea-
surement and aggregation techniques that game theory and mathematics
have so far offered. It is indeed impossible to carry out any social choice
86 Comments by Célestin Monga and James E. Foster
theory without acknowledging the underlying question that is the basic
problem of moral philosophy: “What should I do?” Issues of individual
and group preferences or interests are likely to collide in ways that can-
not be fully captured by the rigid laws of averages, which underpin most
aggregative theories. Group decisions are also mired in ethical dilemmas
and conceptual inconsistencies that economics is not equipped to handle.
The impossibility theorem, which Sen describes as a result of breathtak-
ing elegance and power, is a very useful tool for assessing which outcome is
“right” when thinking about social choices. Each of its axioms is reasonable
and compelling, but taken together, they are overwhelming. I agree with Sen
that Arrow may have overstated the negative case by insisting that each rule
under consideration satisfies all the axioms no matter what people’s rank-
ings of their preferences and choices turn out to be.3 I also agree that to lay a
broader foundation for a constructive social choice theory, we have to reject
the historical consensus against the use of interpersonal comparisons that
was prevalent in the first part of the twentieth century and became conven-
tional wisdom. Sen argues that we should resist such historical consensus,
because it “was based on a rather fragile understanding of epistemology.” I
would suggest that we explore new frameworks for different levels of interper-
sonal comparisons of utility but remain mindful of the intrinsic limitations
of such analytical tools, which clearly rely on rigid and sometimes simplistic
assumptions, and that lessons from various disciplines be considered.
Sen believes that the search for a social welfare function may not even
need to be very precise. This valid point also leaves open many questions
about the “appropriate,” acceptable standards of comparability of welfare
numbers of different persons. Even in situations of full comparability of
reported well-
self- being numbers (which Sen would use to justify full inter-
personal comparability), one obvious question is how much faith should be
assessments. How much trust should be given to self-
given to self- reported
welfare numbers? The legitimacy of someone judging her own welfare
and giving a metric to characterize it doesn’t solve the problem of being
assessment. As Cioran reminded us, among the many
“wrong” in that self-
reasons for invalidating narcissism is the fact that it is based on profound
3. See Maskin (2009) and Sen and Maskin (2017) for new and interesting ways of
approaching voting measures.
Social Choice and Welfare Economics 87
uncertainty and randomness, because it is basically an exercise in which we
fall in love with someone we know very little about.
Fortunately, Sen also believes that rigorous interpersonal comparisons
need not be of mental states only. He is right in his benign neglect of the
evaluation of mental states in interpersonal utility compari-
validity of self-
sons. Can we trust ourselves to know what we actually go through in each
particular life situation, how we actually feel, what we actually believe in
each situation, and how we actually convey it to ourselves and to others?
And does what we believe and how we feel matter if our behavior, actions,
objective welfare, and standards of living are not really impacted by such
perceptions? If the answers to such questions are positive, what are the
implications for the analytical frameworks for interpersonal comparisons
reported indicators of welfare?
that rely on self-
reported welfare and happiness numbers may be too subjective to
Self-
be relied on. The problem goes beyond narcissism. Recent work on the eco-
nomics of “motivated” belief distortions, both individual and social, shows
images
how agents often try even unwittingly to maintain positive self-
and identities (Bénabou 2015). It has been shown, for instance, that most
people believe they are more likely than others to experience favorable life
events and less likely to suffer adverse ones, such as unemployment, acci-
dents, divorce, or major illness (Weinstein 1980).4 “We also commonly
see ourselves as better drivers, better citizens, less biased and more attractive
than others. Some widely held beliefs are just plainly implausible or demon-
strably false, given publicly available knowledge” (Bénabou 2015, 3). Such
departures from objective cognition may have subjective or objective value.
Still, the prevalence of overoptimism and the reality of overconfidence has
heavy economic and social costs. An illustration of the problem is the fact
income countries who could afford
that large numbers of people in high-
life insurance (given the risks they face) choose not to buy it.
being of a person by
“There are many difficulties in judging the well-
his or her mental state,” Sen rightly points out. “The metric of pleasure or
desire may sometimes quite inadequately reflect the extent of a person’s
substantive deprivation.” True. Hence, his recommendations that such
variables as incomes, commodities bundles, or resources more generally be
4. For a more nuanced analysis, see Harris and Hahn (2011).
88 Comments by Célestin Monga and James E. Foster
“of direct interest in judging a person’s advantage.” Perhaps. But this pre-
scription raises several uncomfortable obvious questions. If mental states
reported) are insufficient or even invalid as metrics of personal util-
(as self-
ity, who has the legitimacy to select the more “relevant” additional or sub-
stitute variables to carry out interpersonal comparisons of utilities? Who
gives us the right to judge anyone’s mental states and to even decide that
some “objective” variables of their welfare should be given consideration?
Who decides that another person is living “well” or “poorly”?5
A sequence in Sergio Leone’s epic movie The Good, the Bad, and the Ugly
shows the main character Tuco (a bandit) is being lectured by his brother
Pablo, who is a priest. “Outside of evil, what else have you managed to do?”
Pablo asks him. Tuco listens patiently to his sermons and reprimands and
then responds vehemently:
You think you’re better than I am. Where we came from, if one did not want to
die of poverty … one became a priest or a bandit! You chose your way, I chose
mine. Mine was harder. You talk of our mother and father. You remember when
you left to become a priest. I stayed behind! I must have been ten, twelve. I don’t
remember which, but I stayed. I tried, but it was no good. Now I am going to tell
you something. You became a priest because you were … too much of a coward
to do what I do!
In some ways, Tuco emerges from that scene as more than the cartoonish
bandit character that he appears to be in the first half of the movie. He also
is revealed to be a humble and thoughtful man who simply faced impos-
sible choices in his life and made those that seemed to him to be the most
courageous and even “ethical.” When Pablo chose to abandon the family
to pursue (selfishly) his calling as a priest, Tuco was left to take care of their
parents. He tried to the best of his abilities and presumably in the most
ethical way but failed. The only other option left for his own survival was
to become an outlaw.
the notion that
This is more than the often derided “situational ethics”—
when assessing human responsibility, one should keep in mind that the
“right” or “wrong” thing to do depends on the situation,6 because there
5. See Monga (2015a, 2015b, 2017) for further discussion.
6. Situation ethics (Fletcher 1967) may have its flaws. But one should remember that
even John Dewey held views that rejected moral universality: such a stance “would
assume the existence of final and unquestionable knowledge upon which we can fall
Social Choice and Welfare Economics 89
are no universal moral rules or rights that apply everywhere and always.
Tuco’s apparently shocking discourse can still be viewed as rational and
his willingness as a minor child to
deeply rooted in moral philosophy—
stay home and take care of his parents when his older brother selfishly left
the family home to (egoistically) pursue his personal calling. Tuco may be
the worse bandit the West has ever produced, but he would argue that his
making is still profoundly moral not just descriptively (in terms of
decision-
the codes of conduct put forward by his society) but also normatively (the
necessary behavior and actions that, given specified conditions, would be
put forward by all rational persons). In sum, Tuco is actually a moral agent
in the Kantian sense, who simply finds himself expressing what he saw as
a “categorical imperative.”7 If Tuco and other comparable characters are
indeed justified in their “perverse” moral stance, perhaps one should con-
clude that rationality cannot be defined at the moral philosophy level in a
way that allows for interpersonal comparisons. This would be another real
impossibility theorem.
In fact, rationality assumptions (more precisely, some conceptions of
rationality) are everywhere in the reasoning and modeling of the social
choice procedures offered by all social choice theorists. Without such
assumptions, no valid ordering of social preferences can take place, because
any ranking must be based on preferred alternatives by people who are
supposed somehow to be rational agents. Sen’s very sophisticated and
extremely elegant framework for interpersonal comparison also shows a
lot of faith in some generic level of Rationality (with a capital “R”), which
back in order to settle automatically every moral problem. It would involve the com-
mitment to a dogmatic theory of morals” (Dewey and Tufts 1908, 488). However,
Dewey’s skepticism of moral universality mainly reflects his skepticism about one
method (the method of abstract moral reasoning) in favor of another (what he calls
the “experimental” or the “method of democracy”). His proposed method
implies that reflective morality demands observation of particular situations,
rather than fixed adherence to a priori principles; that free inquiry and freedom
of publication and discussion must be encouraged and not merely grudgingly
tolerated; that opportunity at different times and places must be given for trying
different measures so that their effects may be capable of observation and com-
parison with one another.
See Dewey and Tufts (1908, chapter XVI (1)) on “Morals and Social Problems.”
7. See Kant ([1797] 1993).
90 Comments by Célestin Monga and James E. Foster
presumes that people always have reasons for their actions. Even when
people offer reasons for their actions, such reasons may not necessarily
need to be validated identically. Infinitely many explanations exist for why
people are (or are not) motivated to do the “right” thing.
Economists should be cautious in their faith in rationality, regardless of
its scope and use. Some cognitive scientists have conjectured that reason
a
may be an evolutionary attribute to human beings, just like bipedalism—
trait that occurred only over time. Mercier and Sperber (2017) suggest that
reason initially emerged in the savannas of Africa when human beings real-
ized that they needed to cooperate among themselves. In their view, reason,
which has become the ultimate and unique characteristic of the human
race, developed mainly to allow the resolution of problems posed by liv-
ing in collaborative groups. Reason had a purely utilitarian genesis as “an
adaptation to the hypersocial niche humans have evolved for themselves”
(Mercier and Sperber 2017, 330). Reason emerged not to help people solve
abstract problems but rather to fill their trust deficit, which was the critical
criterion for improved living conditions and for survival.
Reason is therefore a constantly changing human trait, a unique fac-
ulty that is also moving target. It is therefore an enigma. If one agrees to
link human reasoning to evolutionary processes, such as natural selection,
then it is understandable that the dynamics of social change always cre-
ates distortions between phenomena that human brains can grasp, study,
even though most phenomena that humans
and debate, and real life—
which sometimes occurs at a much
can grasp may be a part of reality—
more rapid pace. The Neanderthal man didn’t have to worry about cyber
attacks or the ideal curriculum for training a good economist. His life pre-
scription did not include the need to see a dentist twice a year. He lived
gatherers, and his reasoning could be used to
in small groups of hunter-
focus only on the key elements of such an existence. Today, few people
live like Neanderthals and have to confront and solve the problems similar
to those from 25,000 years ago. Some wealthy people live in spectacular
houses or skyscrapers and mainly worry about finding the time to enjoy all
the many comfortable features in their lives, or about what is said about
them on Facebook. Other individuals live in poverty and permanently face
the burden of social exclusion, stigma, and the destruction of their human
dignity. In sum, the differential of pace between social change challenges
and the adaptation of human reason to them would explain why many
Social Choice and Welfare Economics 91
and why rea-
economic agents who seem reasonable often act foolishly—
son often fails us.
A good illustration of this differential in pace is the discrepancies often
observed in the way societies that strive for morality also seem to toler-
ate for an inordinate time laws, regulations, and norms of behavior that
are subsequently viewed as violating and even damaging their own moral
philosophies. Appiah (2010) has examined moral revolutions and campaigns
against repugnant practices, and he concludes that appeals to reason, moral-
ity, or religion aren’t enough to spur fundamental changes in ethical stan-
dards. Objectionable practices seem to be eradicated only when they come
into conflict with the prevailing conception of honor. Appiah’s work con-
vincingly demonstrates how moral codes evolve across space and time, and
why we should be skeptical of any form of immanent rationality. Genera-
tions of historians have wondered how Thomas Jefferson, the intellectual
and visionary humanist who wrote in 1776 the words “all men are created
equal,” could have been the proud owner of a 5,000 acre working plantation
and owned 607 slaves over the course of his life (Thompson 2017). Jefferson,
the third president of the United States, was the father of six children of
one of his slaves, Sally Hemings. Was he simply another cynical hypocrite?
Not necessarily. Simply, perhaps, just another human being going through
the tragic contradictions and mysteries of life. One can safely guess that
there have always been millions of Thomas Jeffersons and Sally Hemings out
report their well-
there, who would have struggled to define and self- being,
utility, or welfare metrics. If that is the case, then any social choice theory
that places too much faith in any conception of rationality runs the risk of
being at some level, a non sequitur.
Sen has carefully avoided falling into that trap by making his compara-
tive utility framework broad and flexible enough to accommodate many
of the conceptual challenges faced by social choice theorists. His remark-
able insights certainly open up interesting new avenues for solving Arrow’s
impossibility theorem. He also provides valid arguments for ignoring the
skepticism of the likes of Lionel Robbins. He emboldens researchers who
struggle with the complex issues of social aggregation to rethink util-
ity comparisons at levels that may not require the types of rigid condi-
tions imposed by Arrow. Sen’s more relaxed approach makes possible the
design of consistent analytical frameworks to assess and measure interper-
sonal welfare. But one can only take his proposed intellectual route at the
92 Comments by Célestin Monga and James E. Foster
(somewhat heavy) cost of accepting the big assumptions that such exercises
should be done at several different levels and that the exclusive reliance on
mental state comparisons may not be relevant in social choice. These are
elegant but big assumptions.
Conclusion
In the end, we should perhaps acknowledge that there are situations in
which one simply cannot win. Francis Blanche, the late French comic,
often said in one of his sketches: “I was married twice, two catastrophes:
The first time my wife left; the second time, she stayed!” He never won-
selection skills, or with him
dered whether the problem was with his wife-
more generally. But would it matter? The more serious points are our innate
centered natures, our shift-
inability to look beyond our intrinsically self-
ing egos and psyches, and our unstable preferences; our inability to con-
sistently define our own tastes, feelings, and opinions; and the structural
limitations of any attempt to consistently capture and aggregate the criteria
being.
for common well-
Such a perspective alters one’s view of rationalities. It also allows me
to regard rather favorably the various attempts by economists and other
social scientists to free their disciplines from the tyrannies of rationality. In
this critical endeavor, Sen’s contribution in particular, has been salient and
spectacular. I still am hopeful that, one day, perhaps using Sen’s analytics,
cost analysis of my life and
I will be able to carry out a rigorous benefit-
find out whether it had enough meaning to look like a “profitable” invest-
ment. But the constantly shifting values of time, discount rates, and ethical
criteria for interpersonal welfare comparisons may render my intellectual
journey irrational and foolish.
References
Appiah, Kwame Anthony. 2010. The Honor Code: How Moral Revolutions Happen. New
York: W. W. Norton & Company.
Arrow, Kenneth J. [1951] 1963. Social Choice and Individual Values, second edition.
New York: Wiley.
Ayer, A. J. [1936] 1971. Language, Truth, and Logic. Harmondsworth, UK: Penguin
Books.
Social Choice and Welfare Economics 93
Bénabou, Roland. 2015.”The Economics of Motivated Beliefs.” Jean- Jacques Laffont
Lecture, delivered at the Congress of the French Economic Association, June 2014.
685.
Forthcoming in Revue d’Economie Politique 2015/5 Vol. 125, 665–
Charles de. 1781. “Mémoire sur les élections au scrutin.” Mémoires de
Borda, Jean-
665. Translated by A. de Grazia. 1953. Isis 44 (1/2):
l’Académie Royal des Sciences, 657–
42–51.
Condorcet, Jean- Antoine- Nicolas de Caritat. 1785. Essai sur l’Application de l’Analyse
à la Probabilité des Décisions Rendues à la Pluralité des Voix. Paris: L’Imprimerie Royale.
Dewey, John, and James H. Tufts. 1908. Ethics. New York: Henry Holt & Co.
Elster, Jon, and John E. Roemer. 1991. “Introduction.” In Interpersonal Comparisons
16. New York: Cambridge
of Well-Being, edited by Jon Elster and John E. Roemer, 1–
University Press.
Fletcher, Joseph. 1967. Moral Responsibility: Situation Ethics at Work. Philadelphia:
Westminster Press.
Hammond, Peter J. 1991. “Interpersonal Comparisons of Utility: Why and How
Being, edited by
They Are and Should Be Made.” In Interpersonal Comparisons of Well-
254. New York: Cambridge University Press.
John Elster and John E. Roemer, 200–
Hare, Richard M. 1963. Freedom and Reason. Oxford: Clarendon Press.
Harris, Adam J. L., and Ulrike Hahn. 2011. “Unrealistic Optimism about Future Life
154.
Events: A Cautionary Note.” Psychological Review 118 (1): 135–
Harsanyi, John C. 1953. “Cardinal Utility in Welfare Economics and in the Theory
435.
of Risk-Taking.” Journal of Political Economy 61 (5): 434–
Harsanyi, John C. 1955. “Cardinal Welfare, Individualistic Ethics, and Interpersonal
321.
Comparisons of Utility.” Journal of Political Economy 63 (4): 309–
Harsanyi, John C. 1977. Rational Behaviour and Bargaining Equilibrium in Games and
Social Situations. Cambridge: Cambridge University Press.
Jevons, William Stanley. [1871] 1970. The Theory of Political Economy. London:
Macmillan.
Kant, Immanuel. [1797] 1993. Groundwork of the Metaphysics of Morals: With On a
Supposed Right to Lie Because of Philanthropic Concerns, third edition, translated by
J. Ellington. Indianapolis, IN: Hackett.
Maskin, Eric. 2009. “The Arrow Impossibility Theorem: Where Do We Go from
Here?” Arrow Lecture presented at Columbia University, New York, December 11.
Maskin, Eric, and Amartya K. Sen. 2014. The Arrow Impossibility Theorem. New York:
Columbia University Press.
94 Comments by Célestin Monga and James E. Foster
Maskin, Eric, and Amartya K. Sen. 2017. “The Rules of the Game: A New Elec-
10.
toral System.” New York Review of Books, 64(1): 8–
Mercier, Hugo, and Dan Sperber. 2017. The Enigma of Reason. Cambridge, MA:
Harvard University Press.
Monga, Célestin. 2015a. “Principles of Economics: African Counter- narratives.” In
The Oxford Handbook of Africa and Economics. Volume 1: Context and Concepts, edited
by Célestin Monga and Justin Y. Lin, 303–
333. New York: Oxford University Press.
Monga, Célestin. 2015b. “Measuring Democracy: An Economic Approach.” In The
Oxford Handbook of Africa and Economics. Volume 1: Context and Concepts, edited by
Célestin Monga and Justin Y. Lin, 427–
452. New York: Oxford University Press.
Monga, Célestin. 2017. “The Macroeconomics of Marginal Gains: Africa’s Lessons
to Social Theorists.” In The Political Economy of Everyday Life in Africa: Beyond the
Margins, edited by Wale Adebanwi, 115– 132. Woodbridge, Suffolk; Rochester, New
York: Boydell and Brewer.
Priest, Kyle. 2017. “Churchill’s Argument Against Democracy: The Average Voter.
Information Levels and News Sources among Americans.” Political Science Senior
Thesis. Bemidji State University.
Rawls, John. 1971. A Theory of Justice. Cambridge, MA: Harvard University Press.
Robbins, Lionel. 1938. “Interpersonal Comparisons of Utility: A Comment.” Eco-
641.
nomic Journal 8 (192): 635–
Day.
Sen, Amartya K. 1970. Collective Choice and Social Welfare. San Francisco: Holden-
Sen, Amartya K. 1977. “On Weights and Measures: Informational Constraints in
1572.
Social Welfare Analysis.” Econometrica 45 (7): 1539–
Sen, Amartya K. 1986. “Social Choice Theory.” In Handbook of Mathematical Econom-
1181.
ics, volume 3, edited by Kenneth J. Arrow and Michael D. Intriligator, 1073–
Amsterdam: North-Holland.
Thompson, Krissah. 2017. “For Decades They Hid Jefferson’s Relationship with Her.
Now Monticello Is Making Room for Sally Hemings.” Washington Post, February 19.
Weinstein, Neil D. 1980. “Unrealistic Optimism about Future Life Events.” Journal of
820.
Personality and Social Psychology 39 (5): 806–
Weymark, John A. 1991. “A Reconsideration of the Harsanyi– Sen debate on Utili-
Being, edited by John Elster and John
tarianism.” In Interpersonal Comparisons of Well-
E. Roemer, 254– 320. New York: Cambridge University Press.
Comment: James E. Foster
Measurement as Social Choice
I am terribly biased when it comes to Professor Sen and hence feel obli-
gated, in the spirit of full disclosure, to let you know why. I first met
not
A. K. Sen in a welfare economics class at New College, Florida, in 1976—
in person, but through his book Collective Choice and Social Welfare. The
starred chapters captured me and wouldn’t let go until I had extended his
liberal paradox to a world where groups, rather than persons, were decisive.
I sent a draft to Professor Sen, and he responded with guidance on how to
revise the paper and where to publish it, which happened soon after. Thus
began my journey from mathematics to economics via social choice theory,
guided at a distance by Professor Sen.
We met in person a few years later, when I was a graduate student at
at-
Cornell. As Andrew D. White Professor- Large at Cornell, Professor Sen
encouraged me to consider research in poverty measurement, which led
to my work with Joel Greer and Erik Thorbecke. He also provided a list of
problems on partial orderings to explore along with my thesis advisor (and
alas—
his coauthor), Mukul Majumdar, which— we never jointly pursued.
In 1982, there was a wild ride from London to Oxford in a yellow Alfasud,
during which Professor Sen explained how, despite Thatcher’s cutbacks, he
was able to conduct a research project on gender discrimination in Indian
villages by diverting funds from his telephone budget. In 1993 we began a
project to expand his classic On Economic Inequality, which led to many late
taught Econom-
nights, as my wife remembers well. Then in 2008, we co-
ics 2054, Social Choice and Welfare Economics, at Harvard. Now an expanded
the
edition of Collective Choice and Social Welfare has been published—
book that began the process some 40 years ago. Professor Sen has been an
96 Comments by Célestin Monga and James E. Foster
inspiration to generations of researchers. I have received a full measure of
his generosity, for which I am most grateful.
The present chapter is a prime example of why we love to read Sen:
remarkably clear summaries of difficult literatures, woven together with
entertaining quotes and remarkably apt phrases. On one hand, it is a lucid
exposition of the key results from social choice, including Condorcet’s vot-
ing paradox, Arrow’s pathbreaking general possibility theorem, Gibbard’s
Black theorem on single-
equivalent result on strategic voting, the Arrow-
peaked preferences and majority voting, and Sen’s result on the impossibil-
ity of a Paretian liberal. On the other hand, it is a masterful exposition of
the downs and ups of welfare economics: including Bentham’s utilitari-
anism, the Robbins “logical positivist” revolution and its progeny, “new
welfare economics,” which privileged “Pareto efficiency” and its “remark-
able reticence” to discuss distributional issues. Then the paper moves on
Samuelson social welfare functions and back to Arrovian social
to Bergson-
welfare and its accompanying informational privations, with no cardinal
or interpersonal comparisons allowed and in a world of purely welfarist
information. The final section breaks free from the tyranny of impossibil-
ity and narrow informational bases, through rigorous definitions of partial
comparability and an expansion of the informational basis of comparisons
to human capability and freedom. It concludes with a discussion of poverty
both monetary and multidimensional.
measurement—
All right, you say, this is a fine exercise in the history of economic thought.
But what practical lessons does social choice and welfare economics have for
the World Bank, or for that matter, policymaking in general? My answer
focuses on metrics and measurement, a topic of particular interest to me,
and the foundation of policy analysis, wherein data are identified and aggre-
gated in meaningful ways to inform social decisions. Let us examine a few
of the messages that are especially pertinent to the process of measurement.
Broadening the informational basis. Sen attributes the impossibility in
Arrow’s theorem to the paucity of information contained in its preference
extension rule also illustrates
profiles. His characterization of the Pareto-
how restricting consideration to interpersonally noncomparable profiles
makers unable to address distri-
of individual orderings leaves decision-
butional issues.1 Broader bases of information are necessary to overcome
1. See Sen (2017, Theorem 5*3).
Social Choice and Welfare Economics 97
these challenges. The capability approach, which operates in the space of
“functionings” and considers achievements as well as “capability sets” of
achievements (containing both chosen and unchosen alternatives), is one
answer. The approach has become a generally accepted way of conceptu-
being, opportunity, and empowerment, and it is the notion
alizing well-
of progress underlying Sen’s (1999) masterpiece Development as Freedom.
The approach also leads to measures that are multidimensional and linked
across dimensions at the individual level, such as the multidimensional
poverty measures of Alkire and Foster (2011). However, it presents chal-
lenges to empirical researchers, as traditional datasets and measurement
methodologies may not be applicable.
The measurement properties of variables. Broader information brings with it
the need to use that information appropriately. After data have been identi-
fied, the next important task is to understand the measurement properties
of the data’s underlying variables and apply a methodology that is suit-
able. For example, ordinal variables are commonly used in measurement,
reports (such as self-
whether as part of self- reported health or life satisfac-
tion) or due to the inherently qualitative characteristics of the indicators
(such as the quality of floors or sanitation facilities). In addition, issues
of noncomparability or partial comparability can easily arise across per-
sons or dimensions. The variables cannot simply be treated as if they were
fully cardinal and fully comparable across different individu-
monetary—
als. An intuitive way of thinking about this issue is to view it as a form of
robustness. If many rescalings of the data are possible, or if many ways of
relating the data across persons (or across dimensions) could be used, would
each of the possibilities yield the same results? The results for a single cardi-
nalization or one way of one of linking data across persons or dimensions
are not enough. Meaningful interpretation of the data requires agreement
across the full range of possibilities.2
Axioms as policy. A third message pertains to the centrality of axioms and
the axiomatic approach in this literature.3 Although it is not always appar-
basic requirements or qualities
ent, axioms are in essence chunks of policy—
that an object must exhibit if it is to be seen as functioning appropriately.
For an Arrovian social welfare function, axioms can ensure that it is broadly
2. See Alkire et al. (2015, section 2.3).
3. See Foster and Sen (1997, 119).
98 Comments by Célestin Monga and James E. Foster
applicable, is appropriately oriented when preferences are in agreement,
ignores irrelevant information, or rules out unambiguously problematic
methods. For measurement, axioms ensure that a measure is capturing the
desired phenomenon. The main axioms come in three varieties: invariance
axioms (like anonymity), which identify the sorts of information a measure
should ignore; subgroup axioms (like decomposability), which specify how
local and national measures are to be linked; and dominance axioms (like
the transfer principle), which require the measure to move in a specific
direction in the presence of an unambiguous change in the data.4 Axioms
help define what the measure should be measuring.
axioms”
Desiderata. Some authors also include a list of desiderata or “proto-
to help guide the construction of measurement methodologies.5 A com-
mon desideratum is that the measure should be understandable and easy
a requirement that can trump formal axioms when commu-
to describe—
nication is important. This property might explain: the prevalence of the
headcount ratio in poverty measurement despite its axiomatic failings; how
the traditional Human Development Index (based on the arithmetic mean)
2010 Index (based on the geometric mean);
might be preferable to the post-
the measure underlying the
and why the mean of the bottom 40 percent—
was selected instead of an Atkinson
World Bank’s shared prosperity goal—
“equally distributed equivalent” income function or the Sen welfare mea-
sure. There is a clear tension between this key desideratum and the more
nuanced policy aims embodied in axioms.
The use of partial orderings. Partial orderings are central to Sen’s presentation
of social choice theory and also are at the core of measurement.6 To deter-
mine whether the income distribution has taken an unambiguous turn for
the worse, the Lorenz criterion or the various orders of stochastic dominance
can be consulted. Likewise, poverty orderings point out when poverty has
fallen for an entire range of poverty lines (or measures). In multidimensional
analysis, dashboards of dimensional achievements provide a partial order for
being when there is little guidance on how to value dimensions.
assessing well-
4. See Alkire et al. (2015, section 2.5).
5. See, for example, Székely (2005), who gives the list used in setting the Mexican
income poverty methodology.
6. See Sen (2017, xxix– 121).
xxxi) and Foster and Sen (1997, 120–
Social Choice and Welfare Economics 99
Partial orderings identify unambiguous (or unanimous) changes; however,
they are also incomplete and unable to decide between certain pairs of
options. Axioms and desiderata can help narrow options and reduce the
incompleteness. But policy discussion typically demands a headline measure
that is real valued as well as complete, facilitates discussion, and encourages
policy analysis. Once again, there may be tension between communication
and other policy objectives. In some circumstances, however, a partial order-
ing can actually facilitate the selection of a specific measure. For example,
the choice of a specific monetary poverty line seems less problematic when
a poverty ordering is available to test robustness for a range of poverty lines.
Measurement as choice. The process of measurement, like that of descrip-
possibly difficult—
tion, “involves the exercise— of selection” across the
many ways of viewing a phenomenon.7 Over time, the justification for
the choices underlying measurement tends to become “this is how it has
always been done.” Institutions like the World Bank are the repositories of
the art of measurement, and they have the responsibility of being transpar-
evaluating their methods. With the estab-
ent and, from time to time, re-
lishment of its Commission on Global Poverty, the World Bank is working
toward fulfilling this goal for the flagship monetary poverty measure and
may consider a multidimensional approach to poverty as outlined in the
present chapter and other writings. In any event, Professor Sen’s many con-
tributions to measurement will undoubtedly prove useful in guiding this
and other related efforts.
References
Alkire, Sabina, and James E. Foster. 2011. “Counting and Multidimensional Poverty
487.
Measurement.” Journal of Public Economics 95 (7): 476–
Alkire, Sabina, James E. Foster, Suman Seth, Maria E. Santos, Jose M. Roche, and
Paola Ballon. 2015. Multidimensional Poverty Measurement and Analysis. Oxford:
Oxford University Press.
Foster, James E., and Amartya K. Sen. 1997. “On Economic Inequality after a Quar-
ter Century.” In Economic Inequality, Amartya K. Sen, expanded edition, 107–219.
Oxford: Oxford University Press.
7. Sen (1980, 353).
100 Comments by Célestin Monga and James E. Foster
Foster, James E., Joel Greer, and Erik Thorbecke. 1984. “A Class of Decomposable
766.
Poverty Measures.” Econometrica 52 (3): 761–
369.
Sen, Amartya K. 1980. “Description as Choice.” Oxford Economic Papers 32 (3): 353–
Sen, Amartya K. 1999. Development as Freedom. New York: Anchor.
Sen, Amartya K. 2017. Collective Choice and Social Welfare, expanded edition. London:
Penguin.
Székely, Miguel, ed. 2005. Números que Mueven al Mundo: La Medición de la Pobreza en
México. City: Miguel Ángel Porrúa.
3 The Revolution of Information Economics:
The Past and the Future
Joseph Stiglitz
The economics of information has constituted a revolution in economics,
upsetting longstanding presumptions, including the presumption of mar-
ket efficiency, with profound implications for economic policy. The central
models of information economics, developed almost a half century ago
but greatly elaborated on in the intervening years, have proven remarkably
robust. At the same time, these advances in the economics of information
have shown the lack of robustness of the standard competitive paradigm.
The models have provided a deeper understanding of other ways in which
actual markets differ from the perfect markets paradigm. For instance, the
sharing are two features that matter
imperfections of competition and risk-
a great deal, and the economics of information provided new insights into
both of these.
Early work in the economics of information also showed how it would
help us understand better the role of institutions and the form that institu-
tions take; work since then has confirmed the promise. So, too, the eco-
nomics of information has provided new intellectual underpinnings to
branches of the subject that seemed devoid of a theoretical framework,
such as accounting, finance, and corporate governance, and has helped us
understand better why work in these subfields is so important.
Elaborations of the early models and the adaptation of these models to
different market contexts have occupied much of the economics profes-
sion’s attention in the decades since the first models were presented.
Paper presented at the conference “The State of Economics, The State of the World,”
World Bank, Washington, DC, June 9, 2016. I acknowledge research assistance from
Andrew Kosenko and editorial assistance from Debarati Ghosh.
102 Joseph Stiglitz
Not surprisingly, the policies derived from the new paradigm are often
markedly different from those derived on the basis of the standard model.
Most importantly, as I emphasize below, there is no presumption that mar-
kets are efficient; quite the contrary, the presumption is that markets are
not efficient. And in those sectors where information and its imperfec-
tions play a particularly important role, there is an even greater presump-
tion of the need for public policy. The financial sector is, above all else,
about gathering and processing information, on the basis of which capi-
tal resources can be efficiently allocated. Information is central. And that
centrality is at least part of the reason that financial sector regulation is so
important.
Markets where information is imperfect are also typically far from per-
fectly competitive (as that concept is understood, say, in the models of
Arrow and Debreu).1 In markets with some— competition,
but imperfect—
firms strive to increase their market power and to increase the extraction
of rents from existing market power, giving rise to widespread distortions.
In such circumstances, institutions and the rules of the game matter. Public
policy is critical in setting the rules of the game. Distributive effects of alter-
native rules may outweigh any efficiency gains.
Undoing the adverse distributive effects created by these market
imperfections may be very costly, again, largely because of information
imperfections.2
Many recent changes in the rules may have had both adverse efficiency
and distributive effects. The economics of information has explained why
distributive effects themselves may have efficiency consequences, espe-
cially in the presence of macroeconomic externalities.
Looking forward, changes in the structure of demand (that is, as a coun-
try gets richer, the mix of goods purchased changes) and in technology may
lead to an increased role for information and increased consequences of
1. The market failures referred to in the previous paragraph arise even when firms
and households are price takers. I am now describing an important second set of
market failures typically arising in markets with imperfect information.
2. In standard economics, the second welfare theorem explains how any Pareto
efficient allocation can be achieved simply through the redistribution of initial
endowments. When there is imperfect information, the second welfare theorem is in
general not true. For an exposition, see Stiglitz (1994).
The Revolution of Information Economics 103
information imperfections, decreased competition, and increasing inequal-
ity. Many key battles will be about information and knowledge (implicitly
and the governance of information. Already, big debates are
or explicitly)—
going on about privacy (the rights of individuals to keep their own infor-
mation) and transparency (requirements that government and corpora-
tions, for instance, reveal critical information about what they are doing).
In many sectors, most especially, the financial sector, there are ongoing
obligations on the part of individuals or firms
debates about disclosure—
to reveal certain things about their products. Many of these issues can be
who owns the right to certain pieces of
framed in terms of property rights—
information. But these property rights issues are different from and more
complex than those concerning conventional property rights, where it is
usually assumed the stronger the better. Here, the ambiguities in the assign-
called strong (intellectual)
ment of property rights are apparent, and so-
property rights may lead to poorer economic performance.
Globalization has heightened all the associated controversies because
now, how the rules are set affects not only distribution among individuals
within countries but also the distribution of income between countries.
Many in the former colonial world see the attempt by some in the advanced
countries to impose their set of rules as not just an attempt to enrich their
corporations but also to entrench old inequities.
How we handle these issues will affect inequality, economic perfor-
mance, and the nature of our polity and society for decades to come.
This paper is divided into seven sections. In the first, we lay out some of
the key insights of the New Information Economics, contrasting it with the
old paradigm, which assumed perfect information. The central result of the
new paradigm is that markets are not, in general, efficient: There is a need
for government intervention. Adam Smith’s invisible hand failed, simply
because it wasn’t there. The second section describes several failed but
to show that the market was in fact
still important attempts to respond—
efficient, if not always, at least in relevant cases. The third then describes
some of the policy corollaries, and the ongoing policy battles over informa-
tion. The fourth section sets the Information Revolution in the context
of the longstanding battle of how to understand the persistent inequality
is it exploitation (as Marx suggested) or just rewards in
under capitalism—
response to differences in social contribution? We suggest that although
Marx had the wrong model of the economy, there is more than a little grain
104 Joseph Stiglitz
of truth in his exploitation theories. The fifth section describes the role of
the information revolution in promoting broader changes in the economic
to the implications of the new para-
paradigm. The sixth looks forward—
first century. I end
digm for the economy that is evolving in the twenty-
with a few concluding remarks.
The Information Revolution
Economists had, of course, long recognized the importance of imperfect infor-
mation. Indeed, some economic discussions actually trumpeted the informa-
arguing that efficiency can be achieved in a
tional efficiency of the market—
decentralized price system, so there is no need for a central planner. All the
information that a firm or a household needed to know to make its decisions
was to be found in the prices. Prices coordinated all economic activity. Yet
these statements were made without any formal models of the economy as
and-
an information processor. Resource allocations were once- all deci-
for-
sions. Moreover, the kinds of information imperfections were limited. There
was no uncertainty about the quality of a worker or a product.
other
By and large, formal models made no mention of information—
than to assume that there was perfect information. The hope was that anal-
yses assuming perfect information would still be relevant so long as infor-
mation was not too imperfect.
Some Chicago school economists thought that one could develop an
based on the analysis of the supply and
“economics of information”—
demand for information (much like the “economics of agriculture”) and
focusing on the particular characteristics of the demand for and production
of information (just like agriculture economics focuses on the particular
characteristics of the demand for and supply of food). But it should have
been clear, even before the formal development of the field described below,
that such a development was unlikely. Information (knowledge) is funda-
mentally different from steel, corn, or the other goods on which ordinary
economics focuses. Information is a public good3—
indeed, more broadly,
3. In the sense defined by Samuelson, as a good characterized by nonrivalrous con-
sumption (the enjoyment of a pure public good by one individual does not detract
from its enjoyment by others). Pure public goods are also typically characterized
by the impossibility (or at least difficulty) of appropriation. As we discuss below,
The Revolution of Information Economics 105
knowledge is a global public good (Stiglitz 1999), and markets on their own
typically are not efficient in the provision of such goods.
Arrow and Debreu provided the key benchmark model describing the
behavior of a competitive economy with perfect information through a
model of competitive general equilibrium in which all firms were price
takers. Most importantly, Arrow and Debreu provided conditions under
which Smith’s “invisible hand” conjecture was correct, not just the first
welfare theorem (showing that market economies were Pareto efficient)
but also the second fundamental theorem. The latter showed that every
Pareto-efficient outcome could be obtained through a market mechanism,
provided that there was an appropriate initial (lump sum) redistribution of
wealth. Arrow and Debreu focused on the technical conditions that were
such as convexity of production sets (making use of the key eco-
required—
as well on as the economic
nomic assumption of diminishing returns)—
conditions: perfect competition, a full set of risk markets (subsequently
Debreu “AD” securities), and the absence of externalities.
called Arrow-
They had provided sufficient conditions for the efficiency of the market.
The question was: Would results still be true under more general condi-
tions? Were the sufficient conditions necessary, or almost necessary? After
several decades of research, it became clear that Arrow and Debreu had
essentially discovered the necessary and sufficient conditions.4
Most of the limitations on which Arrow and Debreu had focused had
in some sense been widely recognized well before their work. They had
put these longstanding understandings on sound footings. And there were
developed public policies in response: environmental regulation or
well-
corrective taxes, for instance, to deal with environmental externalities,
trust policies to deal with imperfect competition. The existence
and anti-
intellectual property rights are an attempt to enable the partial appropriation of the
returns to the production of knowledge. Inherently, such attempts have a social cost,
because the usage of the information or knowledge is restricted, though there is no
marginal cost associated with usage.
4. There were a few other sets of uninteresting conditions—conditions that, remark-
ably, came to play a central role in a particular branch of macroeconomics. The
economy would be efficient even in the absence of a complete set of risk markets
if all individuals were identical— precisely because when they are identical, there
would be no insurance. There would be no one else to whom someone could transfer
the risk he faces.
106 Joseph Stiglitz
of a natural monopoly required either strong regulation or government
ownership.
Absence of a Complete Set of Risk Markets
The one “new” market failure to which Arrow and Debreu called attention
was the absence of a complete set of risk markets. It was obvious that indi-
viduals and firms could not buy insurance against many of the risks that
workers couldn’t buy unemployment insurance, firms couldn’t
they faced—
buy insurance against the risk that the demand for their products declined.
But economists had not realized the importance of this failure. For Arrow
and Debreu to establish the Pareto efficiency of the economy required the
securities
existence of a full set of what came to be called “AD securities”—
delivering a specific amount of some commodity in a particular state at a
particular date, in effect, a complete set of insurance markets. It was obvious
that this was more than a matter of mere technicalities; there were many
important risks for which households and firms simply couldn’t obtain
insurance at all. One could think of public provision of social protection as
having arisen to partially “correct” this market failure.
Presumption That Markets Are Not Efficient
Arrow and Debreu had, however, shunted aside the key question of infor-
mation in all of its dimensions. Earlier, I described how market advocates
viewed the informational efficiency of the economy as one of its triumphs.
These advocates especially celebrated how much one could achieve without
anyone knowing anything about any other firm or household: All relevant
information was conveyed by prices.
But this model made extraordinarily strong assumptions that were not
even stated: Products were homogeneous, and any individual could tell
costlessly any deviation of the product from the “specified” characteristics.
Cheating on quality was impossible. Everyone knew fully the “true” prob-
ability distribution of returns of every asset. There were no asymmetries of
informed individual could take advantage of a
information, where a well-
less informed one.
In the real world, these quality differences are critical. Workers are not
homogeneous. A great deal of effort goes into finding workers who are well
matched for the job. Insurance firms worry about the risk profile of those
they insure. The entire financial industry is focused on identifying “under-
priced” assets.
The Revolution of Information Economics 107
Obviously, these information problems are important to all market
participants. The early literature showed that information asymmetries—
presented a
where one agent had information not available to another—
special set of problems. Attempts to extract that information or to exploit
the informational advantages gave rise to multiple distortions. A great
deal of activity is concerned with addressing these information problems
(both the lack of information and asymmetries in information), improving
information and reducing asymmetries, if not eliminating them. At the
same time, some market participants realize that opportunities for profit
can be enhanced by increasing information asymmetries. They devote
their efforts to ensuring the existence and persistence of these information
asymmetries, as costly as these asymmetries may be to the economy as a
whole.5
Some two decades after Arrow and Debreu’s work, Greenwald and Sti-
glitz (1986, 1988) showed that information market failures were much more
pervasive and consequential. Whenever there was imperfect and asymmetric
that is, essentially always—
information or incomplete risk markets— the econ-
omy was not (constrained) Pareto efficient, taking into account the limita-
tions of information. There were always interventions in the market that
could make some individuals better off without making anyone else worse
off.6 (For brevity, in the discussion below, I refer to this result as the “GS theo-
rem.”) Correcting these market failures is not so easy: They are not isolated,7
they are diffuse, and they are an integral part of the market economy. In
the presence of asymmetries of information and incomplete markets, there
are pervasive pecuniary externalities that matter: What one firm or individ-
ual does has consequences for others, and that is true even when it is only
through the price system. Price changes are more than purely redistributive.8
5. With perfect competition there are no pure profits, and firms realize (as already
noted) that markets where information is imperfect are likely to be less than per-
fectly competitive. This principle holds in other contexts, as we discuss below: Man-
agers may take actions that result in greater information asymmetries to entrench
themselves.
6. Geanakoplos and Polemarchakis (1986) provided an alternative proof of the inef-
ficiency of market equilibria when there is an incomplete set of markets.
7. This stands in marked contrast to pollution externalities, where at least in prin-
ciple, one could ascertain the emissions of pollutants and impose a charge.
8. Greenwald and Stiglitz’s proof of market inefficiency focused on these pecuniary
externalities, showing that in markets with imperfect information or incomplete risk
108 Joseph Stiglitz
Consider a group of seemingly similar people buying health insurance
in a world in which smoking is not observable. Should one person smoke,
it will increase the risk of disease, driving up the health insurance premi-
ums of everyone. There is a real cost to this externality, which the smoker
does not take into account. The market response is to limit the amount of
insurance that an individual can obtain, so that she has some incentive to
averse indi-
behave well. But a real cost results from this restraint; with risk-
viduals, restricting the purchase of insurance lowers expected utility.
Information market failures obviously affect resources devoted to col-
lecting, processing, and disseminating information. Information is a public
good, with no marginal cost associated with the use of an idea by someone
else, so normally one would expect an underinvestment in information.
Thus, an idea that had some popularity for a while was that markets were
informationally efficient, that is, they transmitted through prices all infor-
mation from the informed to the uninformed. But in a sense, that idea
(popularized by Fama (1970, 1991) but totally discredited by Shiller (1990)
as well as Grossman and Shiller (1981)), was intellectually incoherent, as
Grossman and Stiglitz (1976, 1980) pointed out: If the market fully trans-
mitted information, no one would devote any resources to its collection.
Moreover, private returns to information often can exceed social returns:
If I can prove that I am more able than someone else with whom I would
otherwise have been grouped (in the absence of information), my wages
will go up, but his wages will go down. My gains are at his expense. Much
of the returns to information are thus distributive.9
In addition, firms will attempt to create barriers to the dissemination of
politically, they try to create property rights (called “intellec-
information—
tual property rights”). These rights are costly to enforce and seldom enable
markets, their effects are markedly different than in the standard model, where such
price effects cancel, with the gains of one individual being offset by the losses of
others. Arnott, Greenwald, and Stiglitz (1994) explicitly show how changes in prices
affect the self- order effects. Similar results hold for
selection constraints with first-
price effects on incentive compatibility or collateral constraints. The analysis of these
effects has been at the center of the macro- externalities literature discussed below.
9. See Hirshleifer (1971) and Stiglitz (1975). While Hirshleifer identified the distribu-
tive effects of information, Stiglitz succeeded in analyzing the market equilibria. He
showed that there can be multiple equilibria, with a pooling equilibrium (where the
two groups are not differentiated) Pareto dominating the “separating” equilibrium
(where the two groups are differentiated).
The Revolution of Information Economics 109
those investing in information to appropriate all the social returns from
their information. However, to the extent that they are successful, these
rights create a static market inefficiency: Because information, once cre-
ated, is a public good, any barrier to its free dissemination introduces a
distortion in the economy. In practice, the static costs are often increased,
because these restrictions create barriers to entry, supporting a less competi-
tive market environment, and yet the incentives provided for the creation
of knowledge may be limited. Indeed, because the most important input
into the production of knowledge is knowledge, by restricting the use of
knowledge, these rights may actually impede innovation itself. More gener-
ally, the dynamic benefits are markedly less than the supporters of strong
intellectual property rights suggest.10
differing from worlds
Thus, the key insight of information economics—
in which there is perfect information where social and private returns are
is that social returns to information expenditures typically
normally the same—
differ from private returns, in some cases they are greater, in other cases less.
This insight has many implications, including that privately profitable
transactions may not be socially desirable. The subsequent literature has
exposed a huge number of distortions in specific contexts. They include
marginal inefficiencies, where a Pigouvian corrective tax might induce
market participants to do more of the things that they are doing too lit-
tle of and less of the things that they are doing too much of; and struc-
tural inefficiencies, associated with multiple equilibria, with the economy
sometimes being in a Pareto dominated equilibrium (Stiglitz 1972, 1975).
Sometimes, limited government actions can ensure that the economy is
in the “good” equilibrium.11
Information asymmetries can be endogenous Moreover, households
and firms have incentives for creating information imperfections
they may gain from a lack of transparency. So can manag-
(asymmetries)—
it can enhance their “market power” by creating an entry barrier to
ers—
competitive managerial teams (see Edlin and Stiglitz 1995).
Complexity is one way that financial firms in particular introduce opac-
ity. Many financial transactions seem designed more to increase complexity
10. See Stiglitz (2008), Stiglitz (2014a) and Baker, Jayadev, and Stiglitz (2017).
11. For instance, discrimination laws can prevent an equilibrium in which some
groups are treated worse than others (Stiglitz 1973, 1974b).
110 Joseph Stiglitz
and the associated market power than to solve societal problems. Recent
research has shown how complexity increases uncertainty even about sys-
temic stability and the effects of regulatory policy. Although society would
like a better functioning, more stable financial system, market participants
are simply concerned with maximizing profits. The GS theorem empha-
sizes the disparity between private returns and social returns arising from
information asymmetries and incomplete markets. But this recent work has
noted other aspects of the market failures in the financial sector: By becom-
ing too big to fail, too interlinked to fail, or to correlated to fail, financial
institutions can ensure a bailout, in effect a transfer of resources from the
public to themselves. Firms thus have incentives to become too big, too
interlinked, too correlated to fail: There is a systemic problem.
out, they can engage in excessive risk
With a high probability of a bail-
taking, in which they realize the upside (the profits), and the public bears
the downside (the losses). Moreover, with financial institutions that are too
big to fail, too interconnected to fail, or too correlated to fail, success may
not be based on relative efficiency but on relative size and linkages. And the
huge excessive complexity that they have brought to the financial system
makes the consequences of regulations more uncertain. If, as a result, regu-
for instance,
lators are discouraged from undertaking necessary regulations—
regulation—
relying on self- this provides an opportunity for those in the sec-
tor to increase further their profits.
These problems would simply not exist if there were perfect information, in
which case private contractual arrangements would internalize these infor-
mation-related externalities. These market failures clearly provide a ratio-
nale for government intervention. Much of the intervention has focused on
behavior (e.g., restricting excessive risk taking and actions that enhance the
risk of conflicts of interest). But this analysis has suggested that government
needs to go beyond this focus, for example, to regulate the size of banks (to
reduce the risk of being too big to fail), linkages among banks (to reduce the
risk of being too interconnected to fail), and contractual arrangements (to
reduce the risk of excessive complexity).12 Recent research has also noted
that (in part because government cannot monitor the actions of individual
banks) what matters is the entire “ecology,” that is, the diversity (and inter-
connectedness) of financial institutions. Regulating this ecology (by, for
12. See Battiston et al. (2013, 2016a) and Roukny, Battiston, and Stiglitz (2016).
The Revolution of Information Economics 111
instance, preventing the creation of universal banks) mitigates the dangers of
“too correlated to fail,” and provides part of the rationale for structural regu-
Steagall Act, which separated commercial and invest-
lations (e.g., the Glass-
ment banks).
Production and information are interlinked But the inefficiencies of the
market economy go deeper, because production of knowledge and informa-
tion is intertwined with other activities. Thus, the presumption is that the
market is not only inefficient in the production of information/knowledge
but also in the production of goods. For instance, knowledge or informa-
product of the production of goods; if this informa-
tion is produced as a by-
tion leaks out to others, then the value of this information won’t be fully
internalized in the determination of the levels of production (Stiglitz and
Greenwald 2014).
Macro consequences of informational externalities Keynes provided an
explanation of the Great Depression and other deep downturns that had
afflicted capitalism from its beginning. But in the 1970s, dissatisfaction
grew over the disparity between macroeconomics, as it had developed fol-
lowing Keynes, and standard microeconomics. Information economics
provided the necessary underpinnings to reconcile the two. It explained,
for instance, why credit and equity rationing occurred,13 why this led to
averse behavior on the part of firms (Greenwald and Stiglitz 1990), and
risk-
(See
why wages might not adjust even when unemployment is significant.
Shapiro and Stiglitz 1984 and other variants of efficiency wage theory
[Stiglitz 1987c].) These “financial frictions,” as they came to be called, gave
rise to a financial accelerator, whereby small shocks to the net worth of a
firm could give rise to large shifts in both the aggregate demand and sup-
ply curves.14 The effects of a shock could persist—
the restoration of balance
sheets and thus the recovery of the economy to full employment could take
a long time. Moreover, the decentralized adjustment of wages and prices
meant that in response to a shock, the economy might not instantaneously
move to the new equilibrium set of wages and prices consistent with, say,
persistent full employment. Indeed, the economy could persist with wages
and prices each adjusting, but real wages and unemployment remaining
13. See Greenwald, Stiglitz, and Weiss (1984) and Stiglitz and Greenwald (2003) and
the extensive lists of references cited there.
14. See Greenwald and Stiglitz (1993a) and Bernanke and Gertler (1990).
112 Joseph Stiglitz
relatively unchanged (Solow and Stiglitz 1968), or even worse, the adjust-
ments might lead to even higher unemployment (Stiglitz 2016).15
As already mentioned, Greenwald and Stiglitz (1986) noted that one
could describe the market failures associated with adverse selection and
moral hazard as giving rise to pecuniary externalities that matter. These
microeconomic pecuniary externalities have their macroeconomic mani-
festation, which have been the center of much recent work in macroeco-
nomics. For instance, the market equilibrium may be characterized by
excessive foreign-denominated indebtedness (Jeanne and Korinek 2010).
More generally, borrowers may not take fully into account the effects of
their decisions on prices in the future, say, if they were forced to liquidate
their assets. Each small borrower takes the price distribution as given; but of
course, if they all borrow more, then if a crisis occurs, next period prices of
certain assets will fall as they all are forced to liquidate more of their assets.
One of the implications of the theory is that it may be (in general will
be) optimal to treat differently things that are observably different. Thus,
contrary to prevailing attitudes, taxes and regulations affecting foreign cap-
ital and financial institutions should differ from those affecting domestic
capital. The “nondiscrimination” provisions of some trade agreements can-
not be justified in the context of a model with imperfect information.
Theory of second best Long ago, Meade (1955) and Lipsey and Lancaster
(1956) warned the profession about the theory of second best. Just because
an economy is inefficient doesn’t mean that moving the economy closer to
a perfect model will improve welfare. In the presence of multiple distortions,
removing one may worsen economic welfare. Newbery and Stiglitz (1984)
demonstrated this idea in the context of a longstanding presumption by
economists in favor of free trade. So long as there are imperfect risk markets,
trade integration may lower welfare for everyone. But we will never have full
information or a complete set of markets, so we are always in a second best
15. This line of work emphasized a quite different aspect of Keynes than that which
has been the center of much recent work in macroeconomics, highlighting the conse-
quences of wage and price rigidities. Here, it is price adjustments that give rise to prob-
lems (consistent with much of the recent policy concerns over deflation). It can be
deflation theories (1933). Information economics also
viewed as reviving Fisher’s debt-
provided an alternative explanation of the slow pace of wage and price adjustments,
associated with differential risk (Greenwald and Stiglitz 1989) and of adjustments in
employment (Greenwald and Stiglitz 1995). The contrast between the alternative
approaches to macroeconomics is discussed in Greenwald and Stiglitz (1987, 1993b).
The Revolution of Information Economics 113
world. Hence, we need to tread carefully when using the perfect markets
paradigm as a guide to policy reform. Often it gives misleading advice.
One example concerns the absence of a complete set of risk markets.
The question is: Will creating new financial instruments/markets increase
welfare? The advocates of structured finance seem to have suggested that it
will. The answer is far from clear. What is clear is that these new financial
products give rise to at least three distinct problems.
The first one we have already noted: the increased complexity of the
financial system results in financial fragility and reduces the ability of the
regulator to effectively regulate the financial system. Financial interlink-
with the possibility of
ages may lead to an increase in intrinsic uncertainty—
there being multiple equilibria (even with rational expectations.)16
The second problem is that differences in beliefs give rise to gambling
(risk trading) opportunities. In such cases, both sides of the gamble (which
sum) overestimate the probability of gain and react as if their actual
is zero-
wealth has increased. This gives rise to what Guzman and Stiglitz (2016a,
wealth, the wealth that only exists in the imagination
2016d) call pseudo-
wealth can give rise to macroeco-
of the gamblers. Changes in pseudo-
nomic fluctuations. Guzman and Stiglitz suggest that some of the observed
increased volatility may be due to these new structured products, which
open up new gambling opportunities.
The third problem is that the interlinkage of finance undermines the
decentralizability of the economy, one of the main virtues of the market
economy. To know the financial position of any firm requires knowing the
financial position of all creditors, which requires knowing the financial
positions of all creditors of creditors.17
Financial architecture matters In short, different architectures affect the
extent of externalities and the nature of information requirements. There
driven architectures are efficient: Because of the
is no evidence that market-
disparity between private and social incentives, one would not expect effi-
cient outcomes. The design of the architecture can affect the magnitude and
16. Indeed, complex derivatives may even result in the nonexistence of equilibria.
That is, without coordination, market participants can sign a set of mutually incon-
sistent contracts.
17. Requiring trading to go through adequately capitalized clearing houses—
adamantly opposed by the financial sector— would go a long way toward resolving
this problem.
114 Joseph Stiglitz
consequences of the disparity of private and social incentives. Many of the
new financial products giving rise to greater complexity may result in more
“distorted” architectures, which increase the risk of financial fragility.18
Structured finance was thus not (as it claimed) really about match-
ing risk.19 Significant moral hazard can also be associated with increased
determined
indebtedness, but there is no presumption that the market-
contractual bankruptcy provisions are efficient. Indeed, the presumption
is to the contrary, as each firm tries to signal that it is better than others.
This is one of the reasons bankruptcy laws are necessary. (Advocates of the
contractual approach to sovereign debt restructuring seem not to under-
stand this.)20
Information and Other Market Failures
Imperfect competition One of the important insights of the economics
of information is that in the absence of good information, typically com-
petition will be imperfect. And with imperfect competition, there is the
possibility (likelihood) of firms exploiting market power, and indeed, with
imperfect and costly information, of undertaking actions that enhance
their market power.
Information is a fixed cost, introducing a natural “nonconvexity” into
production. Convexity played a key role in the proofs of Arrow and Debreu.
But these mathematical properties have economic implications. The law of
diminishing returns long played a central role in economic analysis; but
this “law” will not be satisfied when information is endogenous.21
With fixed search costs, no matter how small, it pays any firm to raise its
until the monopoly price is
price above that of others by a small amount—
reached, so the only possible equilibrium is the monopoly price (Diamond
1971, Stiglitz 1985). But then it is worthwhile for firms to engage in nonlinear
18. Recent research on credit networks (Battiston et al. 2016a) highlights inefficien-
cies associated with particular architectures, for example, bankruptcy cascades and
increased systemic risk with large/correlated shocks (following on earlier work by
Allen and Gale (2000) and Stiglitz and Greenwald (2003)). For analogous results for
border financial linkages, see Stiglitz (2010c, 2010d).
cross-
19. The information that was collected was markedly different from that which would
be needed if markets were engaged in “matching.” For example, see Stiglitz (1982).
20. See Brooks et al. (2015) and Guzman and Stiglitz (2016b, 2016e).
21. See, for example, Radner and Stiglitz (1984) and Arnott and Stiglitz (1988).
The Revolution of Information Economics 115
to the
pricing, which extracts some of the remaining consumer surplus—
point that there exists no market equilibrium (see Stiglitz 2013 and the refer-
ences cited there).
Indeed, the major distortion of monopoly is in fact associated with its
trying to extract information to enable it to extract more surplus from con-
sumers (Stiglitz 1977). With perfect information, monopoly extracts all the
consumer surplus, and it can do so (in theory) in a nondistortionary way.
Distortions arise because the monopolist cannot easily differentiate those
who enjoy different levels of surplus from its products: Marketing strate-
gies, which are distortionary, are designed to maximize its ability to extract
this surplus from its customers (Salop and Stiglitz 1977).
and expenditures on information are
More generally, small sunk costs—
can give rise to persistent monopoly rents with Bertrand
always sunk costs—
competition (Stiglitz 1987b).
Not only does imperfect information lead to imperfect competition, but
also firms’ attempts to manage information imperfections reduce compe-
tition. Efficient management of adverse selection/moral hazard involves
contracts extending over multiple periods, where,
intertemporal linkages—
say, payments in one period are dependent on events/performance in ear-
lier periods (Stiglitz and Weiss 1983). This limits the scope for the usual
where contracts are short term, and the threat
competitive mechanisms—
and enhances scope
of leaving acts as an important discipline device—
for monopolistic exploitation. It also gives rise to institutions (like banks)
responding by internalizing some of the information externalities.
Explanation of some key market failures The Arrow and Debreu analysis
also gave rise to another question: How do we explain key market failures,
such as the lack of a complete set of securities markets or limitations in cap-
ital markets? Information economics (adverse selection and moral hazard)
provides at least part of the answer: Almost surely, the firm knows more
about its profits prospects than do possible insurers, and so it would not be
expected to buy insurance against a risk of low profit levels unless the terms
terms that would make it unprofitable for the insurer.22
were favorable—
22. In the absence of risk aversion, there obviously would be no trade in such secu-
rities. This is the implication of the Akerlof (1970) lemons model and the no- trade
theorems of Grossman and Stiglitz (1980) and Milgrom and Stokey (1982). See also
Stiglitz (1982).
116 Joseph Stiglitz
Information economics also provides one of the explanations for why
Coasian bargaining would not resolve problems posed by externalities.
Coase suggested that through bargaining, an efficient outcome could be
achieved only if there were clear property rights. However, bargaining
with information asymmetries typically is not efficient, as parties engage
in costly actions to convey information about the value of the externality
imposed on them.
Responding to Market Failures: The Possibility of Dysfunctional
Social Institutions
related externalities are not only pervasive, they are also dif-
Information-
fuse, making it difficult to address them with corrective taxation, though
corrective taxation should be part of the policy response (see Arnott and
Stiglitz 1986).
Sometimes the appropriate response is the public provision of informa-
tion (or restrictions on withholding information). Thus, when designing
systems for leasing oil in different tracts, auctions will suffer greatly if some
firm is known to have more information than the others. This provides a
rationale for exploratory drilling to be done by the government.
Sometimes the consequences of these market failures are so obvious and
severe that society responds through the creation of social institutions.
The absence of life insurance led to the creation of burial societies to help
families meet the unexpected costs of an untimely death. Such societies,
mentioned as early as Ancient Rome, were widespread in Victorian Eng-
no
land and still exist today. There was no moral hazard problem here—
and
one would die just to have his or her family collect burial insurance—
the problem of adverse selection was slight. Perhaps the simplest explana-
tion of this “market failure” is that the transactions costs were high. As a
result, it may be more efficient to provide such social protection through
the government.
More generally, society responds to market failures by developing insti-
tutions and contracts. But there is no presumption that these institutional
solutions lead to Pareto efficiency. Indeed, Arnott and Stiglitz (1991) show
that institutional interventions may actually be dysfunctional. Imperfect
“family” insurance (imperfect because risk is shared only among a few
individuals) displaces (“crowds out”) more efficient (but limited) market
insurance.
The Revolution of Information Economics 117
Further Key Insights of the Information Paradigm
Robustness of the standard model As information economics devel-
oped, a key question was: How robust is the standard model, which had
ignored information imperfections? The answer was: not very, with even
slight imperfections of information leading to marked changes in results
(e.g., concerning the nature, optimality, and even existence of equilibrium
(Rothschild and Stiglitz 1976)). Many of the key characterization results also
changed, once information imperfections were recognized. For instance,
markets might not clear even in equilibrium, and the Law of Single Price
was repealed. Markets could be characterized by a price distribution, even
when no source of exogenous noise was present.
Robustness of the new paradigm It was natural, at this point, to ask:
How robust are these new models? The key information problems and
modes of analysis that were identified early (adverse selection, moral haz-
ard) have remained the central foci of research for almost a half century.
At the same time, the precise characterization of the equilibrium turned
out to be dependent on details of markets and, in particular, on assump-
tions about information. The early literature differentiated between a
price equilibrium (in which sellers of, say, insurance had no information
about the characteristics of the buyers or their actions, such as how much
insurance they purchased),23 as characterized by Akerlof (1970), and the
quantity constrained equilibrium (in which insurance firms had such
information, with in effect each buyer buying exclusively from one firm).
More recently, Stiglitz, Yun, and Kosenko (2017) have shown that if indi-
viduals/firms can decide whether to hide or disclose information, then
Stiglitz/quantity equilibrium can
neither Akerlof/price nor Rothschild-
Stiglitz),
be sustained. An equilibrium always exists (unlike Rothschild-
and the unique equilibrium is a disclosed pooling contract (the one most
risk individuals) supplemented by an undisclosed price
favored by low-
risk individual’s odds purchased only by high-
contract at the high- risk
individuals.
In the presence of adverse selection and moral hazard, a pooling quan-
tity equilibrium may exist (Stiglitz and Yun 2013), something that could
not occur if there were only adverse selection.
23. Or, correspondingly, the buyers of cars had no information about the sellers.
118 Joseph Stiglitz
One of the significant contributions of information economics was to
show the importance of, and to analyze the forms of, contracts (Stiglitz
1974a) and institutions, like banks. Loans are not made through auctions
but through institutions like banks, which gather and process informa-
tion. Information economics also led to a new focus on enforcement and
commitment (time consistency). A key issue in contract enforcement, for
instance, is verifiability and thus relates to information.
Debreu framework,
All of this stood in marked contrast to the Arrow-
where not only was the information structure exogenous, with a complete
set of markets, but there were also no problems with enforcement and no
issues of commitment.
Second fundamental theorem also reversed As noted earlier, Greenwald
and Stiglitz (1986) showed that when there was asymmetric information,
markets were not efficient, thus undoing the first fundamental welfare the-
orem of economics. Rather than the presumption being that markets are
efficient, now there is a presumption that they are not.
But what about the second fundamental theorem, which asserts that any
feasible Pareto efficient distribution of income could be attained through
a market mechanism, with the correct initial redistribution of assets? This
theorem was enormously important, because it enabled the separation of
issues of efficiency from those of distribution. Economists should focus on
efficiency, leaving distribution to politics, or so it was argued.
The new paradigm, however, shows that the distribution of wealth
(assets) matters, and distributional effects cannot be undone through (lump
partly because the information required to achieve
sum) redistributions—
those lump sum distributions is not available, and the only feasible redis-
tributive taxes are distortionary.24
Key question: What is the critical market failure? Much of the early litera-
ture on imperfect information focused on information asymmetries, with
some discussions of imperfect information going so far as to suggest that
virtually all distortions associated with imperfect information arise from
these information asymmetries. But the real issue is not so much asymme-
try of information as the endogeneity of information. For instance, the life
insurance firm may know far more about the statistics of life expectancies
24. See Mirrlees (1971), Shapiro and Stiglitz (1984), Stiglitz (1987a), and Brito et al.
(1990).
The Revolution of Information Economics 119
than those they are insuring. The individual may not know whether he or
she is a high-risk or low-risk individual. The life insurance company may
selection
still engage in costly screening activities (including the use of self-
mechanisms) to identify individuals who have characteristics that are sys-
tematically associated with longer life expectancy (see Stiglitz 2002).
Not only is information endogenous but so also are asymmetries of infor-
mation (in contrast, most of the earlier literature simply assumed that the
asymmetries are given exogenously). As already noted, firms and individu-
als have large incentives to create and enhance market power and to maxi-
mize rent extraction through the creation of information asymmetries.
Information and Delegation
Imperfect information implies that the standard analysis of efficient decen-
tralization, based on the AD model with perfect information, is not correct.
But it is the costs of collecting and disseminating information that make
decentralization necessary and give rise to delegation, with profound implica-
tions for economic organization. Delegation means, for instance, that there is
a separation of ownership and control: This separation undermines the stan-
dard theory of the firm and gives rise to problems of corporate governance.
Among the important market failures are those associated with corporate
governance. Managers do not necessarily do what is in the interests of share-
holders. Even larger differences arise between social returns and managerial
returns, implying that the market solution cannot be presumed to be efficient.
There are imperfections in all control mechanisms (e.g., takeovers). That is
the laws governing corporate governance—
why the rules of the game—
matter.25 These issues are particularly relevant in the financial sector.
Economics of Knowledge
Most of the results I have just described have applicability beyond infor-
mation economics narrowly defined, to the economics of knowledge.26
Indeed, knowledge can be thought of as a particular form of information.
Knowledge is, of course, at the center of the theory of innovation. With
a modern economy often characterized as a knowledge or an innovation
economy, it is clear that understanding the economics of knowledge is
25. Stiglitz (2015).
26. The ideas in this section are developed more fully in Stiglitz and Greenwald
(2014).
120 Joseph Stiglitz
key. Knowledge, like information, is different from an ordinary commod-
ity. The tools and insights of standard economics, developed for thinking
about the demand and supply of pins, steel, oil, and other conventional
products, are of only limited relevance to understanding a knowledge
economy.
As I have suggested, knowledge is a form of information with many or
most of the latter’s key properties. Most importantly, knowledge is a quasi-
with, as already noted, no marginal cost associated with the
public good—
use of an idea by someone else. Hence, there is always an inefficiency asso-
ciated with restricting usage, such as through intellectual property rights.
Like many public goods, the appropriation of returns is also difficult. There
are typically large spillovers from an important innovation, such as the
laser or the transistor, with the innovators typically capturing a small frac-
tion of the social benefits.
The implication is that the insights that we have gleaned from the study
of the economics of information apply to innovation and the produc-
tion of knowledge. Markets on their own are not likely to be efficient, and
competition is likely to be imperfect. This runs contrary to a longstanding
view that the real strength of a market economy is the drive for innovation
through Schumpeterian competition.
to Recover Previous Results
Early Attempts to Broaden Perspective—
Failed
on Market Efficiency—
Arrow and Debreu had provided sufficient conditions for the efficiency of
the economy, but not necessary ones. A search ensued for weaker conditions
under which the market was still efficient.
known example was that of Diamond (1967), who established
The best-
the (constrained) efficiency of an economy with a stock market. Even with
the highly restricted notion of optimality and highly restrictive assump-
tions about risk (each firm fell within a risk class and couldn’t change the
probability distribution of returns; it could only change the scale of pro-
duction), the result turned out not to be general. With just two commodi-
ties, or with bankruptcy costs, or with decisions that affect the pattern of
risk distribution, the result was not true: The market was not (constrained)
efficient.
The Revolution of Information Economics 121
As already noted, this quest for weaker conditions under which mar-
Stiglitz (1986) theorem, which
kets are efficient ended with the Greenwald-
showed that markets were generically inefficient; they would be efficient
only in special cases. For instance, the absence of risk markets would make
no difference in an economy with a single individual, because there is no
one with whom the individual could share or trade risk.27
how markets dealt (imperfectly) with the
But there was a second issue—
consequences of imperfect information, including the absence of state-
contingent commodities. Contracts (with payments dependent on observ-
able state outcomes) provided a way of simultaneously sharing risk and
providing incentives (Ross 1973; Stiglitz 1974a).
A huge literature ensued, exploring optimal contract design. One
interesting result is that the predicted complexity28 was far greater than
what was observed. For instance, because common shocks are among the
unobservable variables, optimal contracts should make compensation
dependent on others’ outcomes: The predicted forms of contracts thus are
typically different from those which are observed (see Nalebuff and Stiglitz
1983a, 1983b).
New Institutional Economics
Although the contracts that were observed differed markedly from those
that were predicted, the information paradigm more generally helped
explain many aspects of observed institutions. For instance, sharecropping
with half or more of
has long been criticized as attenuating incentives—
the (marginal) returns going to the landlord. But Stiglitz (1974a) explained
a “reasonable”
sharecropping as balancing out incentives and risk sharing—
contract, given the limitations of information and risk markets.
Although many aspects of contract design are consistent with what the-
ory predicts, the hope that these institutions would lead to Pareto efficiency
failed; as already noted, they could even worsen welfare.
27. As already noted, the failure of markets to be efficient can be simply explained:
with imperfect information, the key constraints— incentive compatibility constraints,
selection constraints, and collateral constraints—
self- are all affected by what other indi-
viduals do; each individual fails to take into account how his or her actions affect these
order importance. These externalities matter.
constraints. And these effects are of first-
28. Except under special and easily rejected specifications of utility functions.
122 Joseph Stiglitz
Policy Corollaries
There are many policy corollaries to the ideas that I have just discussed. In
particular, Washington Consensus/neoliberal policies were predicated on
the Smithian presumption that markets are efficient and the presumption
that moving toward a perfect market would be welfare-enhancing, ignoring
second best economics. As already noted, it is wrong to presume that mov-
ing the economy toward first best economy is welfare-enhancing. But even
if this were not the case, there would be winners and losers, the adverse
distributive effects could outweigh any gains, and the cost of undoing dis-
tributive effects could be large.
Frequency Trading
Policy Battles over Information: High-
Today, a new set of battles has emerged, many directly related to informa-
tion. It is in this arena that social and private returns are most likely to be
large, and therefore the insights of this chapter are most likely to be relevant.
frequency trading. It
Consider, for instance, the development of high-
uncovering prices to enable the
was often justified by “price discovery”—
efficient allocation of resources.29 But this was a self-
serving justification of
the financial sector: No evidence has ever been presented of its importance;
no evidence suggests that having slightly more accurate prices a nanosecond
earlier than otherwise has led to higher growth or more efficient resource
those
allocations. The reality is that it may be a new form of front-running—
who get information about bids and offers or trades before others can make
a profit. Indeed, by extracting some of the rents that would have gone to
frequency trading reduces the overall
those who actually do research, high-
efficiency of the economy à la Grossman-
Stiglitz (see Stiglitz 2014b).
Other New Policy Insights: Structured Finance
The new theory changes views about a variety of government policies. For
instance, I have already noted how creating additional risk instruments
may actually increase risk. So, too, welfare may be increased by requiring
frequency trading is also justified by “liquidity”—
29. High- enabling individuals to
easily move into or out of assets, enhancing willingness to make real investments.
serving argument of the financial sector: The
But this also seems largely to be a self-
evidence is that liquidity dries up when it’s needed.
The Revolution of Information Economics 123
market equilibrium disclosures do not suffice. And welfare
disclosures—
may be increased by requiring trading to occur in markets (through clearing
houses), as long as they are adequately capitalized,30 because that improves
the decentralizability of the economy.
Securitization The information paradigm helps us understand what went
2008 financial crisis,
wrong with the securitization market. Before the 2007–
there was enormous enthusiasm about securitization because it allowed the
dispersion of risk throughout the economy. But securitization entailed the
delegation of different aspects of information gathering and analysis to dif-
ferent entities. For securitization to work well required complex contracts
(with put backs and warranties). It failed, partly because of massive fraud31
but also because of extensive problems in contract enforcement: Mortgage
originators and even seemingly reputable investment banks simply refused
to honor their contracts. This behavior highlights the issues of contracts
and enforcement noted earlier and the important role of government in
preventing fraud in information markets (Greenwald and Stiglitz 1992).
These failures of securitization (capital markets) should not come as a
surprise. What is a surprise is the failure of both markets and government
regulators to understand and anticipate the limitations of capital markets
and securitization, including the limitations on informational efficiency
of markets (Grossman and Stiglitz 1980) associated with the difficulties of
appropriating returns.32
30. Which can be accomplished by requiring joint and several liability among mar-
ket participants.
31. That is, the information provided to those who bought the mortgages and mort-
gage products was massively incorrect— with relatively clear evidence that the sellers
did so at least partially intentionally.
32. The credit rating agencies not only were massively wrong in their evaluations of
the probability of default of different tranches of the structured products (for which
they were paid handsomely); again, there is also evidence of fraudulent behavior.
I was privy to the evidence on fraud and the failure to comply with contract provi-
sions as an expert witness in several cases against the rating agencies, the invest-
ment banks, and other financial institutions. But the federal government and state
governments have brought cases in which some of this evidence has been publicly
disclosed. The Final Report of the National Commission on the Causes of the Financial
and Economic Crisis in the United States (2011) identifies the behavior of the credit rat-
ing agencies and the structured financial products as two of the main causes of the
financial crisis of 2008– 2009. See also Stiglitz (2010b, 2010d).
124 Joseph Stiglitz
Banks can be viewed as the alternative institutional solution to these
informational problems.33 It is noteworthy that a decade after the collapse
of the mortgage securitization market in the United States, it has not been
in spite of their belief in free markets—
restored. Evidently, the banks— want
a structure that entails unacceptable levels of public risk bearing.
Other aspects of financial sector regulation Much of the profits arising
from financial activity is associated with market exploitation (much of
which would not arise in the presence of perfect information), including
creating and exploiting asymmetries of information and market manipula-
tion. In their book Phishing for Phools, Akerlof and Shiller (2015) describe the
incentives for exploiting “ignorance,” irrationalities, and market power.34
Predatory lending and abusive credit card practices are only the most obvi-
ous examples.
and
I have also noted banks’ incentives for increasing complexity—
the disparity between social and private returns in increasing complexity.
to-
Increased complexity even gives rise to new opportunities for hard-
detect fraud. Banks availed themselves of these opportunities. High legal
costs, statutes of limitations, and political capture all make it difficult to
prosecute.
The financial sector has developed new ways of increasing its rents and
new justifications for its exploitive activity that have sometimes prevailed
in courts. Changes in technology and knowledge (e.g., about individual
irrationalities and how to exploit them) and legal frameworks may have
also enhanced the ability of the financial sector to exploit others.
Competing Theories for Describing Market
Reconciling Two Long-
Equilibrium and Explaining Inequalities
For more than 200 years, there have been two basic strands of economic
theory. One emphasizes the role of competition (competitive equilibrium
theory); the other, market power (exploitation).
33. Advocates of securitization never explained why one could not obtain adequate
risk diversification through diversified ownership of banks.
34. Here I am focusing on the consequences of imperfections in information. The
financial sector also enjoyed enormous rents from exploiting other sources of market
power, for example, from running payment systems (credit and debit cards).
The Revolution of Information Economics 125
In recent decades, the former theory has dominated in the West. Of
course, some constraints are always placed on the exercise of market power,
some competition exists. But the standard (price-taking) competitive model
describes few markets. Many tests of competition are only tests of the pres-
ence of some competitive constraints, not tests of how close the economy
approximates a perfect competition model.
The imperfect information/imperfect competition model is fundamen-
tally different from either polar case of perfect or no competition. I believe
the real world is best described by this mixed model. In an economy that is
perfectly competitive, there are, of course, no rents. In an economy where a
monopoly exists in each sector, there are no battles over rents: The monop-
olist simply gets them. In reality, the key battle is over grabbing or limiting
rents, over the structuring of markets and the rules of the game, which
affect the magnitude and distribution of rents.
markets do not exist in a vacuum. Differ-
The rules of the game matter—
being of different groups; each tries to restrain the
ent rules affect the well-
feasible set of contracts and actions of others in ways that benefit them-
selves, and more generally, change the rules to enrich their interests at the
expense of others. The public interest, of course, is to create institutional
frameworks for corporate and public governance that benefit ordinary citi-
zens and society as a whole. This is why the presumption that markets are
basically competitive is a poor starting point for policy analysis, because it
shunts aside all issues associated with the grabbing of rents. Governance is
who makes the decisions, and the rules under which the decisions
crucial—
each firm
are made. In the AD model, there is no real governance issue—
simply maximizes its market value, and all shareholders agree that that is
what it should do. With imperfect information and imperfect risk markets,
it matters whose judgments are decisive, and how different judgments are
“aggregated.” Different individuals will have different views about what the
firm should do (Grossman and Stiglitz 1977).
Economists have long recognized that governance matters in the public
sector and that there is no simple way of aggregating preferences. That was
the essential insight of Arrow (1951). For example, monetary policy made
by those representing workers, focusing on unemployment, will be mark-
edly different from that made by those representing bond holders, focusing
on inflation. Information economics has made it clear that this is true in
the private as well as in the public sector.
126 Joseph Stiglitz
Indeed, the rules of the game matter in every aspect of the economy—
corporate governance, financial sector, monetary policy, bankruptcy,
trust, and labor. Workers will do better with rules that facilitate the
anti-
formation of unions, encourage union membership, and strengthen their
collective bargaining rights, recognizing the “public good” they provide
(all workers benefit when wages are increased). All consumers benefit with
trust policy that recognizes that when there is market power,
a strong anti-
prices increase, and an increase in prices lowers standards of living of
ordinary citizens just as a decrease in wages would. Even bankruptcy law
can have important effects: Laws giving derivatives first priority in bank-
ruptcy, even over workers, encourage derivatives and impose greater risks
on workers. Laws saying that student loans cannot be discharged, even in
bankruptcy, encourage predatory student lending, lead to the immisera-
tion of those at the bottom, discourage investments in education, and
increase inequality overall.
Broader Theoretical Impacts of Information Economics
The information revolution played a critical role in some broader changes
in economics, beyond those just described, including giving rise to new
subfields like contract theory. As noted in the Introduction, it provided for
the first time intellectual foundations for fields like accounting. In finance,
it created tensions between two branches, one focusing on the benefits
of risk diversification, the other on the collection, processing, and dis-
semination of information. As noted, these branches are often in tension:
securitization and structured financial products allegedly led to better risk
diversification and matching of risk profiles with individuals’ preferences
and situations, but they also reduced the incentives for the collection and
processing of information. The financial crisis demonstrated that the latter
effect dominated the former.
But among the greatest legacies of information economics is its con-
tribution to the growth of behavioral economics. Although models with
imperfect and asymmetric information were able to explain many previ-
ously unexplained phenomena, models with rational behavior with imper-
fect information still could not explain some of what was going on (e.g.,
in financial markets). This provided the impetus for the development of
behavioral economics.
The Revolution of Information Economics 127
The original work (e.g., Kahneman and Tversky 1979; Tversky and
Kahneman 1974, 1981) incorporated insights from psychology. Individual
making, especially when decisions were made quickly, involved a
decision-
myriad of biases, such as confirmatory bias, where individuals weight more
heavily evidence that is consistent with their priors (Kahneman 2011).
More recent work, focusing on endogenous preferences and beliefs,
and emphasizing the role of “mental models” (the lens through which
we see the world), has incorporated insights from sociology and social
psychology. Both fields have helped provide insights into societal rigidi-
ties and social change (Hoff and Stiglitz 2010, 2016). They have provided
new instruments for policy, especially in the context of development, as
illustrated by the World Development Report, Mind, Society, and Behavior
(World Bank 2015).
A Look Forward
At one time, it was hoped that advances in technology, including the Inter-
net, would increase competition by lowering search costs. This is true in
specified commodities and
some areas, which have homogeneous or well-
manufactured goods. But new technology has also increased the ability to
increasing asymmetries of information and market power of those
exploit—
who have differential access to information.
in technology, in
More broadly, some of the changes in our economy—
have exacerbated
demand structure, and in our regulatory framework—
the disparity between private and social returns to information (knowl-
edge) and enhanced rent seeking and the capacity for rent extraction.
These changes in underlying fundamentals will require changes in policy
to prevent increasing market power and inequality. There is a risk that
the move to the “information economy” may give market power to those
who dominate in grabbing information (such as Google and Facebook),
distorting both the markets for goods and services (increasing the abil-
ity to price discriminate)35 and innovation. Innovation will be encour-
aged in areas with high potential for grabbing rents based on information,
thereby moving scarce research resources away from areas where social
35. Recall our earlier discussion that imperfections in information have fundamen-
tal effects on production.
128 Joseph Stiglitz
benefits would be higher. The extent to which this occurs will be deter-
mined by the rules of the game, for instance, about privacy, transparency,
ownership rights of information (data) transmitted over a platform, and
constraints on the ability of individuals to give up their rights. This is an
area rife with externalities and other market imperfections, so govern-
ment cannot shy away from taking a role; it cannot just “leave it to the
market.”
Moreover, partly because of the network externalities, it is hard to dis-
place incumbents or change structures: Decisions today will have long-
lasting effects, with the market characterized by having one or at most a
few dominant firms whose dominance persists for long periods.
New Technology
The new technologies of the past two decades have played a particularly
important role in forcing these issues on us. They are responsible for the
creation of the information economy. Network effects and the increasing
role of knowledge may naturally lead to more scale economies. When net-
work effects are strong, there is a natural monopoly. The classical literature
on natural monopolies states that they either have to be closely regulated
or nationalized. Until recently, these new natural monopolies have man-
aged to fend off even the recognition of their market power, and there-
fore of any serious attempt at regulation. As Europe has taken a closer look
at their practices and found them anticompetitive, the United States has
American position.
complained about the European Union taking an anti-
trust authorities are doing what they should,
This is wrong. European anti-
trying to ensure that market power is not abused. It is partially because of
monopolies that the United
the political influence of these American near-
States has not taken actions.
The abuse of their market power is especially likely and troublesome. I
noted earlier that the real distortion associated with monopoly arose from
the attempt to differentiate among customers, to extract more of each
individual’s consumer surplus for the monopoly itself. An understanding
of behavioral economics and the theory of discrimination (based on the
economics of asymmetric information) plus access to enormous amounts
of new data enhance their ability to exploit their market power. Even
more troublesome is that their access to and ability to exploit data on
The Revolution of Information Economics 129
individuals raises deep questions about rights to privacy and the nature
of our society.
Schumpeter argued that we should not be much worried about monop-
olies. One monopoly will be succeeded by another, and competition to
be that monopolist incentivizes innovation. Those ideas have now been
discredited.36 But the special features of these new technologies, with their
access to large amounts of data that cannot be replicated, may have enhanced
the ability of incumbents to persist, in spite of some instances of disruptive
technology.
The Changing Structure of the Economy
Other changes in the economy may have changed the role of information—
again in ways that make the economy less competitive. It is widely noted
that we are moving from a manufacturing economy to a service economy.
Manufactured goods are produced and sold globally. Thus, it is relatively
easy to obtain and transmit information about these products.
By contrast, many of the services that will constitute an increasing frac-
tion of gross domestic product are produced and provided locally. Consum-
ers care about the quality of the services provided, and therefore information
about quality is key and reputation effects are critical. But all of this gives rise
to local market power.
Interplay between Increased Market Power and Politics
Increased economic inequality arising from the natural market forces I have
which in turn leads
just described leads to increased political inequality—
to restructuring the rules of the game (e.g., rules governing privacy and
transparency) to enhance market power and increase inequality. But as the
rules of the game are shaped to enhance the incomes of those with market
power, not only is inequality increased but also economic performance is
likely weakened.
36. Dasgupta and Stiglitz (1980) showed that incumbents have the power and incen-
tive to persist, and Fudenberg et al. (1983) showed that they could persist with a low
level of expenditures on research, and thus a low level of innovation. For a more gen-
eral and updated discussion, see Stiglitz and Greenwald (2014), especially chapters 5
and 6 of the 2015 revision.
130 Joseph Stiglitz
Concluding Comments
Information economics has had a transformative effect on economics and
branches of economics,
economic policy, directly giving rise to new sub-
such as contract theory, which have developed enormous literatures of
their own.
It has provided explanations of phenomena that previously had been
unexplained. A century ago, there was a conflict between institutional
economics and “theoretical” economics, derived from the work of Smith,
Ricardo, Walras, and Cournot. Information economics has, in a sense,
united these two schools by highlighting the importance of institutions,
at the same time that it has demonstrated the limits of markets. In many
cases, it has been able to explain not only the existence of certain institu-
tions but also their structure.
It was also noted that some phenomena could not be explained in a
framework of rational individuals making decisions with imperfect infor-
mation. These “failures” were important in encouraging the development
of behavioral economics.
Information economics, together with other work derived from advances
in game theory, has strongly suggested that the economy is best viewed
through models that highlight market imperfections rather than through
the lens of the competitive equilibrium model. These imperfections include
imperfect and asymmetric information and the other market failures to
which they give rise: incomplete risk markets, market power, and the pos-
sibilities for enhanced rent seeking and exploitation.
and in
Most importantly, information economics has questioned—
longstanding presumptions of economic policy. The
many cases reversed—
presumption is that market economies are not efficient. In the case of perva-
sive market power, there are interventions that can simultaneously increase
efficiency and equity.
These ideas are particularly important for an institution like the World
Bank, attempting to promote development in some of the poorest countries
existent, and
of the world. In these countries, markets are often weak or non
the institutions that promote the gathering, production, and dissemination
of information are particularly weak. For a long time, the Bank predicated
its advice on an economic model that ignored the role of imperfect infor-
mation. Fortunately, for the past two decades, the Bank has been at the
The Revolution of Information Economics 131
forefront in raising questions about that model and enhancing our under-
like those discussed
standing of the implications of alternative frameworks—
for development policy.37
here—
References
Akerlof, George A. 1970. “The Market for ‘Lemons’: Quality Uncertainty and the
500.
Market Mechanism.” Quarterly Journal of Economics 84 (3): 488–
Akerlof, George A., and Robert J. Shiller. 2015. Phishing for Phools: The Economics of
Manipulation and Deception. Princeton, NJ: Princeton University Press.
Allen, Franklin, and Douglas Gale. 2000. “Financial Contagion.” Journal of Political
33.
Economy 108 (1): 1–
Angelides, Phil, et al. 2011. “Final Report of the National Commission on the Causes
of the Financial and Economic Crisis in the United States.” Financial Crisis Inquiry
Commission, U.S. Government Printing Office, Washington, DC, January.
Arnott, Richard, and Joseph E. Stiglitz. 1986. “Moral Hazard and Optimal Commod-
24.
ity Taxation.” Journal of Public Economics 29 (1): 1–
Arnott, Richard, and Joseph E. Stiglitz. 1988. “The Basic Analytics of Moral Hazard.”
Scandinavian Journal of Economics 90 (3): 383–413.
market Institu-
Arnott, Richard, and Joseph E. Stiglitz. 1991. “Moral Hazard and Non-
tions: Dysfunctional Crowding Out or Peer Monitoring.” American Economic Review
190.
81 (1): 179–
Arnott, Richard, Bruce Greenwald, and Joseph E. Stiglitz. 1994. “Information and
88.
Economic Efficiency.” Information Economics and Policy 6 (1): 77–
Arrow, Kenneth J. 1951a. Social Choice and Individual Values. New York: Wiley.
Arrow, Kenneth J. 1951b. “An Extension of the Basic Theorems of Classical Welfare
Economics,” Proceedings of the Second Berkeley Symposium on Mathematical Statistics and
Probability, edited by J. Neyman, 507–532. Berkeley: University of California Press.
Arrow, Kenneth J. 1964. “The Role of Securities in the Optimal Allocation of Risk-
96.
Bearing.” The Review of Economic Studies 31(2): 91–
Arrow, Kenneth J., and G. Debreu. 1954. “Existence of an Equilibrium for a Com-
290.
petitive Economy.” Econometrica 22 (3): 265–
37. See, for example, the 1998 World Development Report, Knowledge for Develop-
ment (World Bank 1998), and the 2015 World Development Report already cited
(World Bank 2015).
132 Joseph Stiglitz
Baker, Dean, Arjun Jayadev, and Joseph E. Stiglitz. 2017. “Innovation, Intellectual
Property, and Development: A Better Set of Approaches for the 21st Century.”
AccessIBSA, Innovation and Access to Medicines in India, Brazil, and South Africa,
Bangalore, India.
Pierre Georg, Robert M. May, and Joseph E.
Battiston, Stefano, Guido Caldarelli, Co-
125.
Stiglitz. 2013. “Complex Derivatives.” Nature Physics 9 (1): 123–
Battiston, Stefano, Guido Caldarelli, Robert M. May, Tarik Roukny, and Joseph E.
Stiglitz. 2016a. “The Price of Complexity in Financial Networks.” Proceedings of the
10036.
National Academy of Sciences of the United States of America 113 (36): 10031–
Battiston, Stefano, J. Doyne Farmer, Andreas Flache, Diego Garlaschelli, Andrew G.
Haldane, Hans Heesterbeek, Cars Hommes, Carlo Jaeger, Robert M. May, and Marten
Scheffer. 2016b. “Complexity Theory and Financial Regulation.” Science 351 (6275):
818–819.
Bernanke, Ben, and Mark Gertler. 1990. “Financial Fragility and Economic Perfor-
114.
mance.” Quarterly Journal of Economics 105 (1): 87–
Brito, Dagobert L., Jonathan H. Hamilton, Steven M. Slutsky, and Joseph E. Stiglitz.
77.
1990. “Pareto Efficient Tax Structures.” Oxford Economic Papers 42: 61–
Brooks, Skylar, Martin Guzman, Domenico Lombardi, and Joseph E. Stiglitz. 2015.
“Identifying and Resolving Inter-Creditor and Debtor- Creditor Equity Issues in
Sovereign Debt Restructuring.” CIGI Policy Brief 53, Centre for International Gover-
nance Innovation, Waterloo, ON.
Dasgupta, Partha, and Joseph E. Stiglitz. 1980. “Uncertainty, Industrial Structure,
28.
and the Speed of R&D.” Bell Journal of Economics 11 (1): 1–
Debreu, G. 1959. The Theory of Value. New Haven, CT: Yale University Press.
Diamond, Peter A. 1967. “The Role of a Stock Market in a General Equilibrium
776.
Model with Technological Uncertainty.” American Economic Review 57 (4): 759–
Diamond, Peter A. 1971. “A Model of Price Adjustment.” Journal of Economic Theory
158.
3 (2): 156–
Edlin, Aaron S., and Joseph E. Stiglitz. 1995. “Discouraging Rivals: Managerial Rent-
Seeking and Economic Inefficiencies.” American Economic Review 85 (5): 1301– 1312.
Fama, Eugene F. 1970. “Efficient Capital Markets: A Review of Theory and Empirical
417.
Work.” Journal of Finance 25 (2): 383–
Fama, Eugene F. 1991. “Efficient Capital Markets: II.” Journal of Finance 46 (5):
1575–1617.
Deflation Theory of Great Depressions.” Economet-
Fisher, Irving. 1933. “The Debt-
357.
rica 1 (4): 337–
The Revolution of Information Economics 133
Fudenberg, Drew, Richard Gilbert, Joseph E. Stiglitz, and Jean Tirole. 1983. “Preemp-
tion, Leapfrogging, and Competition in Patent Races.” European Economic Review
22 (1): 3–31.
Geanakoplos, John D., and Heraklis M. Polemarchakis. 1986. “Existence, Regular-
ity, and Constrained Suboptimality of Competitive Allocations When the Asset
Market Is Incomplete.” In Uncertainty, Information, and Communication: Essays in
Honor of Kenneth J. Arrow, volume III, edited by Walter P. Heller, Ross M. Starr, and
96. Cambridge: Cambridge University Press.
David A. Starrett, 65–
Greenwald, Bruce, and Joseph E. Stiglitz. 1986. “Externalities in Economics with
Imperfect Information and Incomplete Markets.” Quarterly Journal of Economics
264.
100 (2): 229–
Greenwald, Bruce, and Joseph E. Stiglitz. 1987a. “Keynesian, New Keynesian and
132.
New Classical Economics.” Oxford Economis Papers 39 (1): 119–
Greenwald, Bruce, and Joseph E. Stiglitz. 1987b. “Examining Alternative Macroeco-
270.
nomic Theories.” Brookings Papers on Economic Activity, 1, 1988, pp. 207–
Greenwald, Bruce, and Joseph E. Stiglitz. 1988. “Pareto Inefficiency of Market Econo-
mies: Search and Efficiency Wage Models.” American Economic Review 78 (2): 351– 355.
Greenwald, Bruce, and Joseph E. Stiglitz. 1989. “Toward a Theory of Rigidities.” Ameri-
369.
can Economic Review 79 (2): 364–
Greenwald, Bruce, and Joseph E. Stiglitz. 1990. “Asymmetric Information and the
New Theory of the Firm: Financial Constraints and Risk and Behavior.” American
165.
Economic Review 80 (2): 160–
Greenwald, Bruce, and Joseph E. Stiglitz. 1992. “Information, Finance, and Markets:
The Architecture of Allocative Mechanisms.” Industrial and Corporate Change 1 (1):
37–68.
Greenwald, Bruce, and Joseph E. Stiglitz. 1993a. “Financial Market Imperfections
114.
and Business Cycles.” Quarterly Journal of Economics 108 (1): 77–
Greenwald, Bruce, and Joseph E. Stiglitz. 1993b. “New and Old Keynesians.” Journal
44.
of Economic Perspectives 7 (1): 23–
Greenwald, Bruce, and Joseph E. Stiglitz. 1995. “Labor Market Adjustments and the
Persistence of Unemployment.” American Economic Review 85 (2): 219–225.
Greenwald, Bruce, Joseph E. Stiglitz, and Andrew Weiss. 1984. “Informational
Imperfections in the Capital Markets and Macroeconomic Fluctuations.” American
199.
Economic Review 74 (2): 194–
Grossman, Sanford J., and Robert J. Shiller. 1981. “The Determinants of the Vari-
227.
ability of Stock Market Prices.” American Economic Review 71 (2): 222–
134 Joseph Stiglitz
Grossman, Sanford J., and Joseph E. Stiglitz. 1976. “Information and Competitive
253.
Price Systems.” American Economic Review 66 (2): 246–
Grossman, Sanford J., and Joseph E. Stiglitz. 1977. “On Value Maximization and
402.
Alternative Objectives of the Firm.” Journal of Finance 32 (1): 389–
Grossman, Sanford J., and Joseph E. Stiglitz. 1980. “On the Impossibility of Informa-
408.
tionally Efficient Markets.” American Economic Review 70 (3): 393–
Guzman, Martin, and Joseph E. Stiglitz. 2016a. “A Theory of Pseudo- Wealth.” In Con-
temporary Issues in Macroeconomics, edited by Joseph E. Stiglitz and Martin Guzman,
International Economic Association Conference Volume 155- 33. Basingstoke,
II, 21–
UK: Palgrave Macmillan.
Guzman, Martin, and Joseph E. Stiglitz. 2016b. “Creating a Framework for Sovereign
Debt Restructuring That Works.“ In Too Little, Too Late: The Quest to Resolve Sovereign
Debt Crises, edited by Martin Guzman, José Antonio Ocampo, and Joseph E. Stiglitz,
32. New York: Columbia University Press.
3–
Wealth and Consumption
Guzman, Martin, and Joseph E. Stiglitz. 2016c. “Pseudo-
Fluctuations.” NBER Working Paper 22838, National Bureau of Economic Research,
Cambridge, MA.
Guzman, Martin, and Joseph E. Stiglitz. 2016d. “A Soft Law Mechanism for Sov-
ereign Debt Restructuring Based on the UN Principles.” FES International Policy
Analysis Paper, Friedrich Ebert Stiftung, Berlin.
Haldane, Andrew G., and Robert M. May. 2011. “Systemic Risk in Banking Ecosys-
355.
tems.” Nature 469 (7330): 351–
Henry, Claude, and Joseph E. Stiglitz. 2010. “Intellectual Property, Dissemination of
Innovation, and Sustainable Development.” Global Policy 1 (3): 237– 251.
Hirshleifer, Jack. 1971. “The Private and Social Value of Information and the Reward
to Inventive Activity.” American Economic Review 61 (4): 561– 574.
Hoff, Karla, and Joseph E. Stiglitz. 2010. “Equilibrium Fictions: A Cognitive Approach
to Societal Rigidity.” American Economic Review 100 (2): 141– 146.
Hoff, Karla, and Joseph E. Stiglitz. 2016. “Striving for Balance in Economics: Towards
a Theory of the Social Determination of Behavior.” Journal of Economic Behavior and
Organization 126 (Part B): 25–57.
Jeanne, Olivier, and Anton Korinek. 2010. “Excessive Volatility in Capital Flows: A
Pigouvian Taxation Approach.” American Economic Review 100 (2): 403– 407.
Kahneman, Daniel. 2011. Thinking, Fast and Slow. New York: Farrar, Straus and Giroux.
Kahneman, Daniel, and Amos Tversky. 1979. “Prospect Theory: An Analysis of Deci-
291.
sions under Risk.” Econometrica 47 (2): 263–
The Revolution of Information Economics 135
Lipsey, Richard G., and Kelvin Lancaster. 1956. “The General Theory of Second
32.
Best.” Review of Economic Studies 24 (1): 11–
May, Robert M. 2014. “Stability and Complexity in Financial Ecosystems.” presen-
tation to British Ecological Society and London School of Economics Joint Annual
Symposium: Eco2: Ecology × Economics=Eco2 , BMA House, London, September 8– 10.
Meade, James E. 1955. Trade and Welfare. London: Oxford University Press.
Milgrom, Paul, and Nancy Stokey. 1982. “Information, Trade and Common Knowl-
27.
edge.” Journal of Economic Theory 26 (1): 17–
Mirrlees, James A. 1971. “An Exploration in the Theory of Optimum Income Taxa-
208.
tion.” Review of Economic Studies 38 (2): 175–
Nalebuff, Barry J., and Joseph E. Stiglitz. 1983a. “Information, Competition, and
Markets.” American Economic Review 73 (2): 278–283.
Nalebuff, Barry J., and Joseph E. Stiglitz. 1983b. “Prices and Incentives: Towards
a General Theory of Compensation and Competition.” Bell Journal of Economics
43.
14 (1): 21–
Newbery, David M. G., and Joseph E. Stiglitz. 1984. “Pareto Inferior Trade.” Review of
12.
Economic Studies 51 (1): 1–
Radner, Roy, and Joseph E. Stiglitz. 1984. “A Nonconcavity in the Value of Informa-
tion.” In Bayesian Models of Economic Theory, edited by Marcel Boyer and Richard E.
Kihlstrom, 33–52. Amsterdam: Elsevier.
Ross, Stephen A. 1973. “The Economic Theory of Agency: The Principal’s Problem.”
139.
American Economic Review 63 (2): 134–
Rothschild, Michael, and Joseph E. Stiglitz. 1976. “Equilibrium in Competitive
Insurance Markets: An Essay on the Economics of Imperfect Information.” Quarterly
649.
Journal of Economics 90 (4): 629–
Roukny, Tarik, Stefano Battiston, and Joseph E. Stiglitz. 2016. “Interconnectedness as
106.
a Source of Uncertainty in Systemic Risk.” Journal of Financial Stability 35 (2): 93–
Salop, Steven, and Joseph E. Stiglitz. 1977. “Bargains and Ripoffs: A Model of Monop-
510.
olistically Competitive Price Dispersion.” Review of Economic Studies 44 (3): 493–
Shapiro, Carl, and Joseph E. Stiglitz. 1984. “Equilibrium Unemployment as a Worker
Discipline Device.” American Economic Review 74 (3): 433– 444.
Shiller, Robert J. 1990. “Market Volatility and Investor Behavior.” American Economic
Review 80 (2): 58– 62.
Solow, Robert M., and Joseph E. Stiglitz. 1968. “Output, Employment and Wages in
560.
the Short Run.” Quarterly Journal of Economics 82 (4): 537–
136 Joseph Stiglitz
Spence, Michael. 1973. “Job Market Signaling.” Quarterly Journal of Economics 87 (3):
355–374.
Stiglitz, Joseph E. 1972. “On the Optimality of the Stock Market Allocation of Invest-
60.
ment.” Quarterly Journal of Economics 86 (1): 25–
Stiglitz, Joseph E. 1973. “Approaches to the Economics of Discrimination.” American
Economic Review 62 (2): 287–295.
Stiglitz, Joseph E. 1974a. “Incentives and Risk Sharing in Sharecropping.” Review of
Economic Studies 41 (2): 219–255.
Stiglitz, Joseph E. 1974b. “Theories of Discrimination and Economic Policy.” In
Patterns of Racial Discrimination, edited by George M. von Furstenberg, Bennett Har-
rison, and Ann R. Horowitz, 5– 26. Lexington, MA: Lexington Books.
Stiglitz, Joseph E. 1975. “The Theory of Screening, Education and the Distribution of
Income.” American Economic Review 65 (3): 283– 300.
linear Pricing and Imperfect Information:
Stiglitz, Joseph E. 1977. “Monopoly, Non-
430.
The Insurance Market.” Review of Economic Studies 44 (3): 407–
Stiglitz, Joseph E. 1982. “Information and Capital Markets.” In Financial Economics:
Essays in Honor of Paul Cootner, edited by William F. Sharpe and Cathryn M. Coot-
ner, 118– 158. Upper Saddle River, NJ: Prentice Hall.
Stiglitz, Joseph E. 1985. “Equilibrium Wage Distributions.” Economic Journal 95
(379): 595–618.
Stiglitz, Joseph E. 1987a. “Pareto Efficient and Optimal Taxation and the New Wel-
fare Economics.” In Handbook on Public Economics, edited by Alan J. Auerbach and
Martin S. Feldstein, 991–1042. Amsterdam: Elsevier Science.
Stiglitz, Joseph E. 1987b. “Technological Change, Sunk Costs and Competition.”
947.
Brookings Papers on Economic Activity 3 (1): 883–
Stiglitz, Joseph E. 1987c. “The Causes and Consequences of the Dependence of
48.
Quality on Prices.” Journal of Economic Literature 25 (1): 1–
Stiglitz, Joseph E. 1994. Whither Socialism? Cambridge, MA: MIT Press.
Stiglitz, Joseph E. 1999. “Knowledge as a Global Public Good.” In Global Public Goods:
International Cooperation in the 21st Century, edited by Inge Kaul, Isabelle Grunberg,
and Marc A. Stern, 308– 325. New York: Oxford University Press.
Stiglitz, Joseph E. 2002. “Information and the Change in the Paradigm in Econom-
501.
ics.” American Economic Review 92 (3): 460–
Stiglitz, Joseph E. 2008. “Economic Foundations of Intellectual Property Rights.”
1724.
Duke Law Journal 57 (6): 1693–
The Revolution of Information Economics 137
2008 and Its Macroeconomic
Stiglitz, Joseph E. 2010a. “The Financial Crisis of 2007–
Consequences.” In Time for a Visible Hand: Lessons from the 2008 World Financial
Jones, José Antonio Ocampo, and Joseph E. Sti-
Crisis, edited by Stephany Griffith-
glitz, 19–49. Oxford: Oxford University Press.
Stiglitz, Joseph E. 2010b. “Responding to the Crisis.” In Time for a Visible Hand:
Jones, José
Lessons from the 2008 World Financial Crisis, edited by Stephany Griffith-
Antonio Ocampo, and Joseph E. Stiglitz, 76– 100. Oxford: Oxford University Press.
Stiglitz, Joseph E. 2010c. “Risk and Global Economic Architecture: Why Full Finan-
392.
cial Integration May Be Undesirable.” American Economic Review 100 (2): 388–
Stiglitz, Joseph E. 2010d. “Contagion, Liberalization, and the Optimal Structure of
Globalization,” Journal of Globalization and Development 1(2), Article 2, 45 pages.
Stiglitz, Joseph E. 2013. The Selected Works of Joseph E. Stiglitz, Volume II: Information
and Economic Analysis: Applications to Capital, Labor, and Product Markets. Oxford:
Oxford University Press.
Stiglitz, Joseph E. 2014a. “Intellectual Property Rights, the Pool of Knowledge, and
Innovation.” NBER Working Paper 20014, National Bureau of Economic Research,
Cambridge, MA.
Stiglitz, Joseph E. 2014b. “Tapping the Brakes: Are Less Active Markets Safer and
Better for the Economy?” Paper presented at the Federal Reserve Bank of Atlanta,
Atlanta, GA, April 15.
Stiglitz, Joseph E. 2015. Rewriting the Rules of the American Economy: An Agenda for
Growth and Shared Prosperity. New York: W. W. Norton.
Stiglitz, Joseph E. 2016. Towards a General Theory of Deep Downturns. (Presidential
address to the 7th World Congress of the International Economic Association, Dead
Sea, Jordan, June 6– VI. Houndmills, UK, and
10, 2014), IEA Conference volume 155-
New York: Palgrave Macmillan.
Stiglitz, Joseph E., and Bruce Greenwald. 2003. Towards a New Paradigm for Monetary
Policy. London: Cambridge University Press.
Stiglitz, Joseph E., and Bruce Greenwald. 2014. Creating a Learning Society: A New
Approach to Growth, Development, and Social Progress. New York: Columbia University
Press.
Stiglitz, Joseph E., and Bruce Greenwald. 2015. Creating a Learning Society: A New
Approach to Growth, Development, and Social Progress. Readers Edition. New York:
Columbia University Press.
Stiglitz, Joseph E., and Andrew Weiss. 1983. “Incentive Effects of Terminations: Appli-
cations to the Credit and Labor Markets.” American Economic Review 73 (5): 912– 927.
138 Joseph Stiglitz
Stiglitz, Joseph E., and Jungyoll Yun. 2013. “Optimality and Equilibrium in a Com-
petitive Insurance Market under Adverse Selection and Moral Hazard.” NBER Work-
ing Paper 19317, National Bureau of Economic Research, Cambridge, MA.
Stiglitz, Joseph E., Jungyoll Yun, and Andrew Kosenko. 2017. “Equilibrium in a
Competitive Insurance Market under Adverse Selection with Endogenous Informa-
tion.” NBER Working Paper 23556, National Bureau of Economic Research, Cam-
bridge, MA.
Tversky, Amos, and Daniel Kahneman. 1974. “Judgment under Uncertainty: Heuris-
1131.
tics and Biases.” Science 185 (4157): 1124–
Tversky, Amos, and Daniel Kahneman. 1981. “The Framing of Decisions and the
458.
Psychology of Choice.” Science 211 (4481): 453–
World Bank. 1998. World Development Report 1998/1999: Knowledge for Development.
New York: Oxford University Press.
World Bank. 2015. World Development Report 2015: Mind, Society, and Behavior. Wash-
ington, DC: World Bank.
Comment: Ravi Kanbur
Left Field Observations on the Information Revolution in Economics
There is no question that an information revolution has occurred in eco-
nomics. And there is no question that Joe Stiglitz is a revolutionary leader.
The classic papers in this literature bear the names of Stiglitz, Rothschild-
Stiglitz, Stiglitz-Weiss, Shapiro-Stiglitz, Grossman-Stiglitz, Greenwald-Stiglitz,
Stiglitz, and so on.
Newbery-
And there is no question that development economics is closely entwined
with the information revolution. The development context provided the
spur for the theorizing and conceptualizing of Stiglitz, Akerlof, and oth-
ers. The information revolution in turn has implications for development
economics and development policy, including, for example: (1) share crop-
ping and agrarian relations; (2) credit rationing, moneylenders, and micro-
finance; (3) asymmetric information and efficiency wages; (4) migration
models; (5) commodity price stabilization; and (6) free trade and uncer-
tainty, and many other topics.
So what can you say after Joe Stiglitz has given his account of the infor-
mation revolution in economics? It is a bit like critiquing Fidel Castro’s
account of the Cuban Revolution, or taking issue with Dwight Eisenhower’s
Day landings. Commentary is particularly difficult when
narration of the D-
you agree with the revolution and the revolutionary on almost everything,
and consider yourself to have been a foot soldier, having fought in the
“risk taking and inequality” detachment of the revolutionary brigades.1 So,
1. See, for example, Kanbur (1979).
140 Comments by Ravi Kanbur and Hamid Rashid
what to do? To make the commentary somewhat interesting, I will come at
the revolution from left field and pose some methodological questions for
myself, for Joe, and for all of us to ponder.
Expected Utility Analysis
The core analytical tool in the information revolution armory has been
expected utility (EU) analysis. As we all know, questions have been raised
about the independence axiom that undergirds the EU representation of
preference orderings. It is this axiom that allows the representation to be
separable in a specific way between the utility of an outcome with cer-
tainty and the probability of that outcome. But individuals do not appear
to behave according to this axiom, with research on this going back at least
as far as the Allais paradox.
At one level, it is remarkable that so many features of the real world, like
credit rationing or insurance market failures, can be explained with models
in which agents are assumed to behave in a manner that they do not actu-
ally behave like in practice. And it may not matter methodologically, so
long as the predictions of the models are not falsified by observations. But it
does raise the question: How exactly would the iconic results of the classic
models in the revolution survive without EU?
known exercises that establish the iconic results of the
In all of the well-
imperfect information revolution, we use EU. For example, in the classic
Rothschild and Stiglitz (1976) paper on insurance, when we show that a
pooling equilibrium can be broken by a separating insurance contract, and
a separating equilibrium can be broken by a pooling contract, we use EU
comparisons. In another classic (Stiglitz and Weiss 1981), when we show
that credit rationing is an equilibrium for lenders, we use EU. And so on.
Could we construct these equilibria, or show nonexistence of equilib-
rium, if agents did not behave according to EU? My instinct is that we
EU preferences might
could. In the insurance context, for example, non-
allow a wider range of contract offers, which could break an existing equi-
librium. But the twist is that the candidate equilibrium would first have to
EU frame. This is an open and interesting area for
be described in a non-
research. And note that it is not enough to argue, as Machina (1982) does
many of the
in a famous paper, that EU works locally as a linearization—
results require global comparisons.
The Revolution of Information Economics 141
Radical Uncertainty and Behavioral Economics
EU analysis, the foundation of Stiglitzian imperfect information analysis, is
also confined to risk, where probabilities of outcomes are well defined and
known, as opposed to uncertainty, where this is not the case (also known
as Knightian uncertainty). Such radical uncertainty was well described by
214), in an article that introduced the conceptual foun-
Keynes (1937, 213–
dations of the General Theory to American audiences:
By “uncertain” knowledge, let me explain, I do not mean merely to distinguish
what is known for certain from what is only probable. The game of roulette is not
…
subject, in this sense, to uncertainty. The sense in which I am using the term
is that in which the prospect of a European war is uncertain, or the price of cop-
per and the rate of interest twenty years hence. … About these matters there is no
scientific basis on which to form any calculable probability whatever. We simply
do not know.
215) then goes on to develop the argument further,
Keynes (1937, 214–
especially the implications of such radical uncertainty for behavior. Sum-
marizing somewhat:
How do we manage in such circumstances to behave in a manner which saves
our faces as rational, economic men? We have devised for the purpose a variety of
techniques, of which much the most important are the three following :(1) … …
(2). … . (3) Knowing that our own individual judgment is worthless, we endeavor
to fall back on the judgment of the rest of the world which is perhaps better
informed. … Now a practical theory of the future based on these three principles
has certain marked characteristics. In particular, being based on so flimsy a foun-
dation, it is subject to sudden and violent changes. … At all times the vague panic
fears and equally vague and unreasoned hopes are not really lulled, and lie but a
little way below the surface.
These “behavioral considerations,” as they would now be called, are not
Stiglitz, Stiglitz-
present in Rothschild- Stiglitz, and so
Weiss, Grossman-
forth. In all of those models, agents are rational choice EU maximizers with
risk rather than uncertainty. This leads to a set of questions.
Does it matter that the models that describe so well outcomes in actual
markets have models of individual behavior that are so far removed from
reality? How different would the outcomes of those models be if agents in
them followed the precepts of recent developments in behavioral econom-
ics rather than rational choice EU analysis? And would it matter for policy?
I believe these are open questions for research and debate.
142 Comments by Ravi Kanbur and Hamid Rashid
Keynesian Interventionism or Burkean Conservatism?
Does imperfect information, particularly of the radical uncertainty variety
(“We simply do not know”), make one tend towards Keynesian interven-
tionism or Burkean conservatism? Keynes himself was greatly influenced
by Edmund Burke. In an as yet unpublished2 undergraduate essay (Keynes,
15), he lauds Burke’s conservatism in considerations of war and
1904, 4–
other momentous decisions:
Burke ever held, and held rightly, that it can seldom be right … to sacrifice a pres-
ent benefit for a doubtful advantage in the future … ; we should be very chary
of sacrificing large numbers of people for the sake of a contingent end, how-
ever advantageous that may appear. … We can never know enough to make the
chance worth taking.
The direct descendant of this line of thinking is Keynes’s famous 1923 state-
ment from his Tract on Monetary Reform: “But this long run is a misleading
guide to current affairs. In the long run we are all dead.” (Keynes, quoted in
Skidelsky, 2013).
Skidelsky (2013) argues that “Keynes would have rejected the claim of
term pain, in the form of budget
today’s austerity champions that short-
term economic growth. The pain
cuts, is the price we need to pay for long-
is real, he would say, while the benefit is conjecture.”
So far, so good. Radical uncertainty appears to favor such progressive
positions as caution in launching wars and austerity programs. But from
Burke’s prudence principle also flowed an institutional conservatism, as
made clear in a famous passage (quoted in Edlin 2017, 50) from Burke’s
Reflections on the Revolution in France:
You see, Sir, that in this enlightened age I am bold enough to confess that we [the
English] … instead of casting away all our old prejudices, we cherish them … and,
to take more shame on ourselves, we cherish them because they are prejudices;
and the longer they have lasted, and the more generally they have prevailed, the
more we cherish them.
A modern version of this argument for conservatism is provided by Edlin
(2017, 49):
Decision makers suffer from switcher’s curse if they forget the reason that they
maintained incumbent policies in the past and if they naively compare rival and
2. Brief extracts from it are published in Skidelsky (2016).
The Revolution of Information Economics 143
incumbent policies with no bias for incumbent policies. I find that conservatism
emerges as a heuristic to avoid switcher’s curse. The longer a process or policy has
been in place, the more conservative one should be. On the other hand, the more
conservative were past decision makers, the more progressive one should be today.
Keynes (1904, 15) interpreted the Burkean recoil from revolution in his
1904 undergraduate essay: “We can never know enough to make the chance
worth taking, and the fact that cataclysms in the past have sometimes inau-
gurated lasting benefits is no argument for cataclysms in general. These
fellows, says Burke, have ‘glorified in making a Revolution, as if revolutions
were good things in themselves’.”
This is not the place to develop the argument, and others have devel-
oped it as well, that an institutional conservatism was also deeply ingrained
in Keynes, who wanted to save capitalism, not end it. Actually, what Keynes
really wanted was to save the world of late Victorian and Edwardian Eng-
land, which came to an end in 1914.
Conclusion
So, imperfect information in the form of radical uncertainty, and its conse-
quent undermining of EU analysis, opens up a wide area of research, asking
whether the classic Stiglitzian propositions will still hold in this brave new
world.
Further, radical uncertainty can be the basis for either Keynesian inter-
ventionism or Burkean conservatism or, in Keynes’s mind, both! In any
event, so far as Joe Stiglitz is concerned, to paraphrase Keynes on Burke,
“This fellow has glorified in making a revolution, as if a revolution was a
good thing in itself.” And there is no question that the information revolu-
tion has indeed been a good thing in itself. As a foot soldier in the informa-
tion revolution, I salute our leader!
References
Edlin, Aaron. 2017. “Conservatism and Switcher’s Curse.” American Law and Eco-
95.
nomics Review 19 (1): 49–
Kanbur, Ravi. 1979. “Of Risk Taking and the Personal Distribution of Income.” Jour-
797.
nal of Political Economy 87 (4): 769–
Keynes, John Maynard. 1904. The Political Doctrines of Edmund Burke. Keynes papers,
KP: UA/20/315, Kings College Archives, Cambridge.
144 Comments by Ravi Kanbur and Hamid Rashid
Keynes, John Maynard. 1937. “The General Theory of Employment.” Quarterly Jour-
223.
nal of Economics 51 (2): 209–
Machina, Mark J. 1982. “‘Expected Utility’ Analysis without the Independence
323.
Axiom.” Econometrica 50 (2): 277–
Rothschild, Michael, and Joseph E. Stiglitz. 1976. “Equilibrium in Competitive
Insurance Markets: An Essay on the Economics of Imperfect Information.” Quarterly
649.
Journal of Economics 90 (4): 629–
Skidelsky, Robert. 2013. “True, Keynes Cared Little about the Long Run. But That
Wasn’t Because He Was Gay.” Op- Ed, Washington Post, May 9.
Skidelsky, Robert. 2016. The Essential Keynes. London: Penguin Classics.
Stiglitz, Joseph E., and Andrew Weiss. 1981. “Credit Rationing in Markets with
410.
Imperfect Information.” American Economic Review 71 (3): 393–
Comment: Hamid Rashid*
Information Asymmetry, Conflicts of Interest, and the Financial Crisis:
Lessons Learned and the Way Forward
Information asymmetry is often the main cause of market failures, as Joe
explains earlier in the chapter. Firms, especially financial firms, have incen-
tives to exploit information asymmetries, hiding critical information about
with
their incentives, behavior, and performance. Conflicts of interests—
can allow financial firms
information asymmetry hiding their existence—
report risks, triggering devastating market
to ignore, misprice, and under-
failures. This is what we saw in the run up to the financial crisis, the most
significant market failure of our lifetime. One lesson from the crisis is that
regulators, rating agencies, and investors largely failed to detect widespread
m
conflicts of interest in the mortgage market, when some large banks co
ingled appraisal, origination, servicing, securitization, underwriting, and
even rating functions. A bank originating a mortgage typically relies on
an independent third party to appraise the value of the property and thus
avoid potential conflicts of interest in the valuation. But this practice
changed during the boom years before the financial crisis. If a bank stood to
earning hefty commis-
gain more from a higher valuation of the property—
it would use a com-
sions and fees, as we saw during the mortgage boom—
plicit appraiser willing to inflate the property value. By 2006, 90 percent of
often by the originating bank or its
the property appraisers felt pressured—
to inflate home values.1 The independent appraiser was supposed
agents—
*The views expressed here do not reflect the views of the United Nations or its Mem-
ber States.
1. Financial Crisis Inquiry Commission (2011).
146 Comments by Ravi Kanbur and Hamid Rashid
to protect the lender (and by extension, the banks’ depositors) against the
risk of a mortgage default. But during the mortgage boom, the appraiser
a clear conflict of interest—
and mortgage originator worked together— to
inflate property values and originate as many mortgages as quickly as pos-
sible, which exacerbated the risk of a crisis.
Conflicts of interest were also pervasive in transactions between the orig-
inator and the mortgage securitizer. Both often worked for the same bank,
and the originator knew that the securitizer would buy whatever mortgages
she would originate, without raising any question about the quality of the
mortgages. In addition, the securitizer knew that he would be able to pack-
rated securities and sell them to the investor
age any mortgage into AAA-
clients of the same bank, then neither the originator nor the securitizer had
an incentive to assess underlying risks accurately and price the mortgage-
backed securities correctly. With all transactions taking place among related
parties and no consequences for ignoring conflicts, due diligence became a
waste of time for our banks.
During the mortgage boom, our banks routinely hid conflicts of interest
and originated trillions of dollars of subprime mortgages that did not meet
with
minimum underwriting standards. In a “issuer pays” rating model—
more than 80 percent of subprime mortgage-
manifest conflicts of interest—
possible AAA ratings,2 making many
backed securities received the highest-
investment-
below- grade securities highly attractive to investors. Had the
and
investors been fully aware of the extent of the conflicts of interests—
backed
how these conflicts contributed to the mispricing of mortgage-
the mortgage bubble that precipitated a global financial crisis
securities—
might have been avoided.
It is surprising that the pervasive conflicts of interest that led us to
the crisis did not attract the attention of our regulators, given that only
7 years earlier, the Enron scandal exposed widespread and harmful con-
flicts of interest in corporate America. Drawing on the Enron lessons,
Oxley Act in 2002, with the stated
the US Congress passed the Sarbanes-
objective: “to protect investors by improving the accuracy and reliabil-
ity of corporate disclosures.” Title V of the Act deals with conflicts of
interest, requiring a clear separation between the securities analysts and
Pinkham, and Vickery (2010).
2. Ashcraft, Goldsmith-
The Revolution of Information Economics 147
underwriting functions of a financial firm. Large banks blatantly disre-
garded the separation and exploited conflicts of interest in securitization
Oxley Act,
deals. Yet no banker was charged for violating the Sarbanes-
although it contained provisions for holding senior management person-
ally responsible for a breach.
label
As the issuers of billions of dollars of Alt-A and subprime private-
backed securities, our largest banks were fully aware of the qual-
mortgage-
ity of underlying assets that backed the securities and yet hid that informa-
tion from their investors. The banks put their own interests ahead of the
interests of their investors to make a quick profit on risky bets. The sheer
financial supermarkets—
size and complexity of these banks— that com-
bined mortgage, retail, and investment banking activities, allowed them
to exploit conflicts of interest with impunity. Their status as “too big to
supervise” allowed them to evade regulatory oversight, while being “too big
to fail” meant they faced no consequences of a devastating financial crisis.
Aiming to address the root causes of the financial crisis, the US Congress
Frank Wall Street Reform and Consumer Protection Act in
passed the Dodd-
2010. The Act was intended to mitigate, among other issues, the inherent
conflicts of interests in securitization. Section 621 of the Act, for example,
prohibits any transaction that could create a conflict of interest with an
investor in a securitization transaction. The subsequent rule issued by the
Securities and Exchange Commission (SEC) included a negative list of con-
flicts of interest in securitization that is laden with exceptions and loopholes.
For example, the rule provided that a securitization transaction would not
making, or for pro-
represent a conflict of interest if it is for hedging, market-
viding liquidity. This leaves room for subjective interpretation, requiring the
regulator to differentiate ex ante between hedging and speculation. There is
a growing recognition that it is hard, if not impossible, to detect conflicts of
interest in securitization, especially when it involves many parts of a large
and complex financial firm.
Frank Act, even if implemented fully, is unlikely to mitigate
The Dodd-
conflicts of interest in securitization, largely because of its reliance on a
narrow set of rules and a long list of exceptions. Instead of prohibiting
Frank Act needed to effectively
a limited number of activities, the Dodd-
address the structural causes of the crisis, such as the “too big to fail” cri-
option based executive compensations, which incentivize
terion or stock-
banks to hide conflicts of interests and take excessive risks. Conflicts of
148 Comments by Ravi Kanbur and Hamid Rashid
material, perceived, or potential—
interest— are often unobservable until
their adverse effects become apparent. But organizational structures, such
as bank size and compensation packages of senior executives, are clearly
observable. The regulators need to target and regulate the observables
instead of trying to regulate unobservable behavior. The Federal Reserve
Board, for example, recently imposed a limit on the growth of the assets of
a large bank that engaged in inappropriate behavior.3 This is clearly a bold
step in the right direction.
Steagall Act managed to keep conflicts
For nearly 70 years, the Glass-
of interest under control by enforcing a clear and structural separation
between commercial and investment banking activities and making sure—
that banks were not too big to supervise and regulate.
albeit indirectly—
Frank Act, it incorporated specific measures to address
Unlike the Dodd-
the problems of information asymmetry and conflicts of interest in the
Frank recognizes the “too big to fail” prob-
financial sector. Although Dodd-
lem, it has not prevented the growth of our largest banks. The large banks
have since become even larger. In fact, the market share of the top 10 or 15
largest banks has increased relative to the pre-crisis level (figure 3.1). The
largest bank in the United States was 57 percent larger in 2014 than it was
in 2007.
Frank also does not adequately address the problems of incentive
Dodd-
option based compensation schemes
structures in large banks. The stock-
create a conflict of interest, as they encourage managers to act more like
term
investors or speculators and to take excessive risks that boost short-
stock price of the firm, even if doing so undermines the financial stability
up to the crisis, large financial firms
and interests of the firm. In the run-
offered significant amounts of stock options to their senior managers,
based compensation also
ostensibly to incentivize best performance. Stock-
contributes to the “too big to fail” problem, encouraging top managers to
Frank intro-
aggressively increase size and market share. Although Dodd-
limits, and claw-
duces certain prohibitions, time- back provisions, stock-
based compensation remains as pervasive as it was before the crisis. If this
practice continues unabated, financial firms will continue to find ways to
3. See https://www.reuters.com/article/us-usa-wells-fargo-fed/fed-orders-wells-fargo
-to-halt-growth-over-compliance-issues-idUSKBN1FM2V9.
The Revolution of Information Economics 149
0.7
0.6 Market share of the largest
bank
0.5 Market share of the 10
largest banks
0.4
Market share of 5 largest
%
0.3 banks
0.2
0.1
0
2003
2008
2001
2002
2004
2005
2006
2007
2009
2010
2011
2012
2013
2014
Figure 3.1
Market share of large US banks: 2001– 2014
Percentage of total assets of banks with more than $300 million in assets
Source: Author’s compilation of data from https://www.federalreserve.gov/releases
/lbr/.
term profits and market valuation. This
make risky bets and boost short-
also perhaps explains the spectacular growth of the market valuation of US
financial firms since the crisis, increasing from $2.8 trillion in 2008 to $7.3
trillion in 2015 (figure 3.2).
The financial crisis is a sad testimony to the failure of the revolution in
information economics that Joe spearheaded, which should have fostered
and enabled effective regulation of our financial sector, where information
asymmetry matters the most. The advances in our thinking and under-
standing of how information shapes market behavior and the scope and
intensity of financial regulations have moved in the opposite direction dur-
ing the past few decades. We now see a starker, and more disconcerting,
disconnect between the lessons of information economics and the state of
financial regulation. Financial regulation of the past few decades has relied
on the imaginary narrative of perfectly competitive financial markets with
Frank Act is no exception. The revolution
perfect information. The Dodd-
150 Comments by Ravi Kanbur and Hamid Rashid
8000
7000
6000
5000
Market value of
4000
financial corporate
3000 equity ($ billion)
2000
1000
0
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
2014
Figure 3.2
Market value of US financial corporate equity ($ billion)
Source: US financial accounts, https://www.federalreserve.gov/releases/Z1/Current
/data.htm.
in information economics will remain incomplete until the economics of
information guides and shapes financial regulation. Unless we bridge the
gap between what we know and how we regulate financial markets, another
financial crisis is just around the corner.
References
Ashcraft, Adam, Paul Goldsmith-Pinkham, and James Vickery. 2010. “MBS Rating
and the Mortgage Credit Boom.” Staff Report 449, Federal Reserve Bank of New York.
Financial Crisis Inquiry Commission. 2011. The Financial Crisis Inquiry Report. Wash-
ington, DC: US Government Publishing Office.
II Macroeconomic Stabilization and Growth
4 From Chronic Inflation to Chronic Deflation: Focusing on
Expectation and Liquidity Disarray since World War II
Guillermo Calvo
The organizers of this conference have asked me to distill in a few pages
my experience with macroeconomics, focusing on issues that are relevant
for policy making. After several false starts, I concluded that I could bet-
ter serve the objective if I identified a few theoretical topics that helped
in the discussion of critical policy issues during the period covered. Ratio-
nal expectations (RE) stands up, given its role in the flourishing of macro-
economics since the 1970s. Whether or not one endorses its relevance for
positive theory, RE has proven to be immensely useful to sort out analytical
Miller
issues and offer useful insights on applications. Like the Modigliani-
theorem or Ricardian equivalence, the RE insights provide benchmarks that
shed light even on cases in which RE does not hold.
Macroeconomics is a very rich and varied field. To keep this chapter
within reasonable bounds, I confine the discussion to two grand themes,
namely, chronic inflation and chronic deflation, and associated issues.
Chronic inflation took center stage in developed market economies (DMs)
in the 1970s (a period called the “Great Inflation”), and in emerging market
economies (EMs) during much of the twentieth century after World War
II. The Great Inflation has been subject to a good number of studies (for a
recent discussion, see Bordo and Orphanides (2013) and McKinnon (2013)).
Therefore I will focus on EM episodes. Simple rules for stopping inflation,
inspired by available theory, failed to work and, in several instances, gave
This is an abridged version of a paper, under the same title, prepared for the World
Bank conference entitled The State of Economics, The State of the World, held in Wash-
ington, DC, June 8 and 9, 2016. I am thankful to Edmar Bacha, Sara Calvo, Fab-
rizio Coricelli, Roque Fernandez, Arvid Lukauscas, and Pablo Ottonello for valuable
comments.
154 Guillermo Calvo
rise to serious distortions and costly crises. However, chronic deflation is
galvanizing world attention since the dramatic financial crisis episodes in
EMs and, more recently, the ongoing Great Recession that started in 2007.
Research on EM chronic inflation focused mostly on local or domestic
factors and, as a general rule, assumed that DMs were stable and provided
the services of deep capital markets. This view started to be challenged by
the rise of EM financial crises in which external factors have a significant,
partly
if not necessarily dominant, role (e.g., the debt crisis in the 1980s—
and Mexico’s “Tequila”
triggered by Volcker’s stabilization program— crisis
in 1994/1995, which followed on the heels of a more modest but still
important rise in US interest rates). These crises involved a host of financial
factors, but the conventional wisdom tended to attribute them to EM weak
domestic institutions and domestic policy mistakes. Global capital markets
might have played a role, but they were not seen as the main culprit. This
1998,
view proved harder to defend after the Asia/Russia crises in 1997–
because some of epicenter economies had followed the Washington Con-
sensus. At any rate, the succession of these crises gave a strong impetus to
research that pointed in a sharply different direction. For example, toward
driven crunch in international capital
sudden stop (i.e., a severe supply-
oiled financial markets. Moreover,
flows), a phenomenon alien to well-
given that the abovementioned crises involved several economies outside
the crisis epicenter, research focused on systemic sudden stop. This set off a
search for factors that may turn a regular contraction in international capi-
tal flows into systemic sudden stop (e.g., Calvo 1998; Cavallo and Frenkel
2008; Calvo, Izquierdo, and Mejía 2016).
These crises raised the suspicion that the explanation went beyond stan-
dard fundamentals and that liquidity phenomena were at work. “Liquidity”
is a slippery word. For my purposes here, it will suffice to define liquidity
services as the services provided by assets or, more generally, arrangements
that may facilitate market transactions. Assets that provide those services
will be called “liquid assets.” This does not imply that they are mostly
employed as means of exchange. Liquid assets can be easily transformed
into means of exchange but can be held as store of value or employed as
credit collateral, for example. It is important to notice, though, that liquid-
ity services depend on implicit compacts in which the equilibrium value
of, say, a liquid asset is a function of the compacts themselves. Therefore,
From Chronic Inflation to Chronic Deflation 155
liquidity is inherently illusory. Its value can collapse on the spur of the
moment, giving rise to what is usually called a “liquidity crunch.” More-
over, the latter can occur in the absence of real shocks. In fact, real and
liquidity shocks are seldom independent of each other. The main point is
that liquidity shocks can be rationalized without appealing to other kinds
of shocks (e.g., total factor productivity shocks). In fact, as argued below,
liquidity shocks can give rise to sudden stops, and to issues associated with
liquidity traps and price deflation.
In a nutshell, this chapter will be divided into two parts, the motivation
evident as we proceed. Expectations, spiced up
of which will become self-
with chronic inflation issues, will be the theme of the first part of the chap-
ter; while liquidity, spiced up with recent capital market episodes, will be
the theme of the second part. Context and more details follow.
Setting the Stage and Overview
Most people would likely agree that Keynes’s (1936) General Theory (GT)
played a pivotal role in establishing macroeconomics as a field different
from, but not incompatible with, microeconomics. The GT was born dur-
ing the Great Depression and was greatly influenced by issues that have
become once again relevant during the Great Recession (e.g., the liquidity
trap). The GT downplayed the relevance of monetary policy for the recov-
ery phase and gave rise to the view that “money does not matter.” The
appeal of this view, however, started to fade in the wake of World War II,
when inflation spiked and the world economy recovered from its initial
slump and started to grow at relatively high rates, despite the large con-
traction of public expenditure after the war. As a result, the liquidity trap
became a bogeyman of the past, and the view that “money matters” came
back with renewed vigor. Friedman and Schwartz (1963), for instance, argue
that the Fed caused the Great Depression by ignoring the harmful effects
money stance.
of price collapse and failing to adopt a more aggressive easy-
The relevance of monetary policy got further support from the 1970s Great
Inflation episode in DMs (see Bordo and Orphanides 2013; McKinnon
2013), and chronic inflation in EMs (see Calvo and Végh 1995).
First attempts to accommodate inflation in a Keynesian context involved
off
sticking a Phillips curve (an empirical regularity that suggests a trade-
156 Guillermo Calvo
between inflation and unemployment) in Hicks’s (1937) IS/LM model, sub-
ject to little microeconomic backing (or microfoundations). This approach
gave rise to a highly fruitful literature around the question of whether the
off could be used to lower unemployment by raising the rate of infla-
trade-
tion. This literature is very well known and need not be discussed in great
detail here (see Gordon 2011). However, I think it is worth pointing out
that the Phillips curve literature brought “expectations” to center stage and
helped establish the view that in the long run, inflation is ineffective for
lowering unemployment and could even make it worse (see Phelps 1972;
Friedman 1977). This view got further support from the RE literature, in
which context it can be shown that inflation ineffectiveness could also
hold in the short run (Lucas 1972) and, more fundamentally, that empiri-
cal regularities like the Phillips curve could be misleading for policy making
(Lucas 1976; Sargent and Wallace 1981).
Moreover, the RE literature illustrated the possibility that frank and
intentioned policy makers could throw the economy into a destruc-
well-
tive black hole, given that in the RE context, policy making is subject to
a serious birth defect: time inconsistency. Time inconsistency arises when
policy makers renege from earlier policy announcements or commitments.
It is a birth defect, because policy makers have incentives to engage in time
inconsistency, even though cheating is not in their DNA, their foremost
objective is to maximize social welfare, and (not a minor detail) RE implies
that individuals cannot be easily fooled (see Kydland and Prescott 1977;
Calvo 1978). The time inconsistency literature offers support for the adop-
tion of rules rather than discretion, and central bank independence is a
natural corollary. All these insights are in the toolkit of modern macroecon-
omists, and several have already been incorporated in governments’ macro-
economic models around the globe.
The RE approach allows analyzing policy credibility issues in isolation
from other, perhaps important but disparate, issues like the public’s imper-
fect information about the relevant model. RE does not answer all relevant
questions concerning policy credibility but signifies a major step forward
compared to the case in which expectations are assumed to be backward
looking (e.g., adaptive expectations). I will illustrate this by discussing some
key policy roadblocks faced by EMs subject to chronic inflation problems
in the next section.
From Chronic Inflation to Chronic Deflation 157
1990s,
As pointed out in the beginning of this chapter, since the mid-
the world economy has been buffeted by crises in which the role of finan-
cial dysfunction has become increasingly evident. Moreover, these crises
are severe and bear an eerie resemblance to the Great Depression. Expres-
sions like “liquidity trap” and “price deflation,” popular in the 1930s, have
become part of the daily lingo. This prompted the economics profession
to look back to the 1930s and brush up on the rich menu of new financial
instruments that have been created since the 1990s (see Eichengreen 2015;
Ohanian 2016). Prior to that, a macroeconomist could get her paper pub-
ranked journal by assuming, say, that debt contracts took
lished in a top-
contingent bonds, free from default risks. Moreover, she
the form of state-
would not have faced major referee’s objections if the paper assumed that
liquidity was confined to an object called “money,” which did not interfere
in a major way with the workings of the capital market. Issues in which
indebtedness and default are the order of the day could
unplanned over-
and the long time to recov-
not be accommodated in that type of model—
ery that we were experiencing until recently, accompanied by unrelenting
deflationary forces (particularly in the Eurozone and Japan) even less so.
These issues are very troubling, and policy makers are clamoring for a rapid
analytical response.
What to do? Compared to the tame “reality” prior to the Great Reces-
sion, the new reality looks extremely complex. Thus it is easy to give in to
the temptation of increasing models’ complexity. This could be a serious
mistake. Taking that route might make macroeconomics look like a feather
driven by the flow but unable to change the direction of the
in the wind—
wind. For macroeconomic policy to have a chance to make a difference,
theory has to identify a few key factors that could have a major impact on
the direction of the wind. As mentioned at the start of the chapter, I think
liquidity is one of them, and I will argue that one can get useful insight
tidbits (“intuition pumps,” as Krugman (2011) calls them) by setting liquid-
ity at the center of the macro universe. This will be fleshed out in the third
section in this chapter.
Much of the literature that I refer to is available in print (especially that
in the next section) and, therefore, I thought that it would be more use-
ful if I focus on the flow of ideas and leave out technicalities, unless they
are necessary to clarify the argument. I should note, incidentally, that I
158 Guillermo Calvo
will confine the discussion to narrow economic models and will have to
apologize for not covering attendant and highly relevant political economy
issues.
Chronic Inflation: Theory and Practice in EMs
that is, high inflation or stop-
Chronic inflation— go high inflation epi-
and-
has been the nemesis of several
sodes that occur over an extended period—
large EMs during the twentieth century (see, e.g., Dornbusch and Simon-
sen 1983; Bruno et al. 1988, 1991). Many stabilization programs employed
the exchange rate as a nominal anchor. This choice was prompted by the
existence of shallow domestic capital markets that made interest rates inef-
fective monetary policy instruments, and the growing evidence that mon-
especially
etary aggregates have a weak and volatile link with inflation—
rate-
when inflation rates are high. In the 1970s, exchange- based stabiliza-
tion programs were expected to produce quick results. This view was based
on the belief that purchasing power parity will bite and force domestic
prices to grow at about the same rate as international prices plus the rate
of devaluation. In general, this was not to happen. Domestic prices contin-
ued unabated and caused unwanted (and, I must say, unexpected for many
trained economists) major real currency appreciation. Moreover, many
well-
of these programs started with a consumption boom that increased fiscal
a common feature
revenue and gave the impression that fiscal imbalance—
inflation economies—
in high- was going away without additional sacrifice.
These optimistic expectations were hard to change, because, of course, pol-
icy makers (and international financial institutions, especially those that
endorsed these stabilization and reform programs) became enthusiastic
cheerleaders. Besides, as I argue below, some of the popular monetary models
before the 1970s were unsuitable for discussing certain critical issues, like
imperfect policy credibility.
Imperfect Credibility and Excessive Inflation
To motivate this section, I start by referring to a provocative paper by Milton
Friedman (1971) that, abstracting from credibility issues, concludes that
dependent economies was excessive, in the
inflation in several seigniorage-
sense that a lower rate of inflation would collect higher seigniorage. This
From Chronic Inflation to Chronic Deflation 159
looks puzzling. However, the puzzle is a result of focusing on a restricted
set of policy options. Friedman (1971) focuses on permanent or steady state
inflation paths and thus rules out inflation spikes. If the public is taken by
surprise, for example, it can easily be shown that inflation spikes could be
effective in further increasing revenue from inflation.
To illustrate, consider a standard model in which the demand for money
is a decreasing function of the expected rate of inflation. Suppose inflation
is set to maximize seigniorage à la Friedman (1971), and consider an unex-
and-
pected once- all spike in the rate of inflation, coupled with a credible
for-
policy announcement that future inflation will remain unchanged. The infla-
tion spike lowers the stock of real money, but it does not affect the demand
for money, because expected inflation would stay the same. Thus, the public
state demand for
will be willing to spend extra resources to restore the steady-
money, which results in seigniorage higher than what would be attained if
maximizing inflation rate.
authorities stuck to Friedman’s seigniorage-
Repeated use of surprise inflation is unlikely to be successful in increas-
ing seigniorage, because the public will start to expect a rate of inflation
state revenue from inflation.
larger than the one that optimizes steady-
inflation terri-
Thus, eventually the economy may land on the excessive-
tory highlighted in Friedman (1971). However, this is not due to an ele-
mentary economics error on the part of the central bank, as Friedman’s
results might lead us to conclude. An inflation spike is, in the short run,
one of the cheapest and most expeditious methods for securing additional
fiscal revenue. Moreover, this “carrot” is always there. As noted, though,
a problem arises if the government repeatedly reaches out for the carrot.
But even in this case, the evidence presented in Friedman (1971) does
not prove that authorities were making an error. To assess that, one needs
information about how quickly the public catches up with the inflation-
spike strategy.
The central lesson from the above example is that there are harmful
incentives that lead policy makers to implement inflation levels that they
may eventually come to regret. These incentives are no rarity; they are very
common in economies that do not have the instruments to reach a first best
equilibrium. Moreover, these incentives cannot be ruled out even under RE.
This is shown in the time inconsistency literature (see, e.g., Kydland and
Prescott 1977; Calvo 1978). However, there is room for policy. In the above
160 Guillermo Calvo
example, one could try to neutralize these harmful incentives if the central
bank is banned from extending loans to the fiscal authority.1
Inflation surprise is effective for liquidating the real value of financial
powered money. Important examples are public debt
assets other than high-
obligations denominated in nominal terms (e.g., principal or coupon not
indexed to the price level). Thus, in designing public debt instruments, pol-
icy makers should take these seigniorage incentives into account, especially
if the fiscal authority is constrained to have small fiscal latitude. Calvo and
Guidotti (1990) address these issues and discuss public debt configurations
in terms of maturity and indexation. Price indexation, for example, would
remove incentives for surprise inflation; however, it may make public debt
service too rigid in the face of real shocks (more on this in the next subsec-
maturity nominal debt may also remove incentives
tion). Moreover, short-
for surprise inflation if fiscal cost grows exponentially with the rate of infla-
change surprise much higher if it takes
tion (e.g., making the cost of a price-
place, say, in a day rather than in a month). However, the government gives
term debt.2
up the resilience provided by long-
Remark 1. An embarrassing error and a warning. These insights were not
common knowledge at the time of Friedman (1971), partly because the pro-
looking expec-
fession did not have the instruments for modeling forward-
looking scheme, was
tations. At the time, adaptive expectations, a backward-
in vogue. It was employed to model inflation expectations. Thus, inflation
expectations at time t were assumed to be a function of the path of infla-
tion prior to t, weighted by a factor that declined geometrically with the
distance between time t and the time of the inflation realization. The rate
of decline was determined by a parameter that I denote by γ > 0, such that
the larger is γ, the steeper the decline of the weighting factor will be. Cagan
(1956) showed, in the context of a simple monetary model, that there is a
critical γ = γ , such that if γ > γ , the system becomes unstable. This implies,
1. However, this is not a foolproof solution to the excessive inflation problem. See
Calvo (1986a) for a discussion of an episode in which the central bank of Argentina
was banned from lending to the treasury and, hence, private banks took that role.
When the treasury went bankrupt, though, the central bank bailed out private banks,
which was equivalent to taking a long and tortuous route to lending to the treasury.
2. These ideas were developed at the International Monetary Fund (IMF) and helped
make debt indexation and maturity part of the IMF’s program design. See Guidotti
and Kumar (1991) and Calvo (1991).
From Chronic Inflation to Chronic Deflation 161
for example, that if the economy starts off in steady state, it is possible for
the model to generate hyperinflation even though money supply is con-
stant over time! This counterfactual implication led to the conclusion that
the RE approach was incompatible with realistic monetary models, because
RE was identified with the case in which γ → ∞. This is, of course, wrong,
because no matter how large the weight given to very recent observations,
it does not make adaptive expectations rational: They are doomed to be
backward looking! It is interesting to note, though, that it took around 15
years and the RE revolution to get rid of this error (see Sargent and Wallace
1973).3 This episode should send a warning to the profession, because it
shows emphatically that formal models can be dangerously misleading if
they are not disciplined by a good dosage of common sense.
Inflation Stabilization and Incredible Reforms
rate-
In the 1980s, several EM exchange- based stabilization programs failed
to achieve their objectives (see Little et al. 1993; Kiguel and Liviatan 1994).
An unwanted side effect was a large real currency appreciation accompa-
nied by a consumption boom and large current account deficits. This took
and the profession at large—
policy makers— by surprise, because according
much of it based on DM
to the (then) prevalent conventional wisdom—
inflation stabilization is associated with a slump in economic
experience—
activity. The opposite happened. The disconnect between conventional
wisdom and practice was dramatic and, as happens on these occasions,
brought to the surface a myriad of lightweight and even opportunistic
comments. Neoclassical theory and “monetarism” were easy targets, but
an answer from the beleaguered camp did not take long to come. It relied
on the assumption that these stabilization programs were likely imperfectly
credible. The analysis is very simple, thanks to the RE revolution. Calvo
(1986b) shows, for example, that if the public expects that the stabilization
program will eventually be abandoned and high inflation stages a come-
back, it might be rational for the public to anticipate consumption. This
anticipation obviously enlarges the current account deficit and, under nor-
mal circumstances, lowers the real exchange rate (i.e., the relative price of
tradable goods with respect to nontradable goods). The model assumes that
3. This does not invalidate the relevance of adaptive expectations. In fact, it can be
a useful complement to RE.
162 Guillermo Calvo
the total cost of consumption includes the purchase price plus the cost of
holding money in advance to carry out the transaction (i.e., Clower 1967).
The latter is an increasing function of the nominal interest rate, which rises
with expected inflation, and causes the expectation that the total cost of
consumption will be higher after the program is abandoned. Intertempo-
ral substitution trivially follows and gives a rationale for the consumption
boom. For a recent version of the model, which can accommodate the usu-
ally sizable consumption booms, see Buffie and Atolia (2012).4 The argu-
ment would also go through if inflation increased the cost of credit as a
result of high price volatility, for example.5
This model can also be employed to study the impact of temporary trade
liberalization (see Papageorgiou, Michaely, and Choksi 1991). Consider the
case in which the government announces that trade tariffs will be perma-
nently eliminated, but the public believes that they will eventually be rees-
tablished. As in the monetary example, this amounts, in the opinion of the
private sector, to making tradable goods cheaper today relative to tomor-
row. Calvo (1986b), for example, shows that this brings about a current
account deficit that would not take place if the government’s announce-
ment was fully credible. Moreover, the implied intertemporal substitution
is Pareto inefficient, because it is based on an intertemporal distortion. Even
if the government does not intend to abandon trade liberalization, lack of
credibility brings about the same deleterious effects. The government could
disappoint expectations by never reestablishing trade barriers, but that will
not undo the damage! This is, thus, a glaring example of the power of cred-
ibility for the success or failure of economic reform, a phenomenon that
in-
I coined in the (tongue- cheek) phrase “Incredible Reforms” (see Calvo
1989).
An implication of these models that policy makers should take into
run effects that
account is that lack of credibility could give rise to short-
might give the impression that policies are highly successful. For example,
4. Calvo and Drazen (1998) extend the basic model to account for uncertainty about
the duration of announced policies.
5. Sargent (1982) is closely linked to this literature and makes a strong case for credible
stabilization programs. However, the paper focuses on short- lived astronomic inflation
episodes that could hardly be called “chronic.” Moreover, it seems unlikely that indi-
viduals believe in the sustainability of hyperinflation, which would tend to enhance
the credibility of any reasonable stabilization program and, thus, its effectiveness.
From Chronic Inflation to Chronic Deflation 163
the consumption boom that follows the announcement of an exchange-
based stabilization program brings about an increase in the demand for
rate-
money, which gives rise to larger international reserves. If the program is
prompted by high inflation, these developments are likely to be interpreted
as stemming from greater trust that those in charge are serious and able to
carry out the necessary reforms.
It is worth noting that the deleterious effects of lack of credibility high-
lighted here depend on the existence of intertemporal trade (e.g., credit).
Without this channel, the economy would not benefit from intertemporal
trade geared to the fundamentals stressed by conventional trade theory;
however, the economy would be free from credibility distortions. Thus,
these types of models are especially relevant for EMs that have access to
financial markets but have not succeeded in developing resilient market-
friendly institutions. Depending on the circumstances, the model may
justify imposing controls on capital mobility, for instance. But a major con-
tribution of this literature is to highlight the relevance of expectation man-
agement and, above all, ensuring policy credibility.6
Expectations Dominance
Chronic inflation is typically associated with fiscal dominance (i.e., a situ-
ation in which the central bank loses control of money supply because it is
forced to finance the fiscal deficit by issuing domestic money, as in the pre-
vious subsection). The phenomenon is especially relevant when the central
bank faces a recalcitrant fiscal authority that, say, for political reasons, is
not willing to lower the fiscal deficit. But (what appears to be) fiscal domi-
nance can also arise in an analytically much more interesting situation in
which the fiscal authority is fully committed to support the inflation stabi-
lization program, as announced.
This is illustrated in Calvo (1998), which was motivated by trying to
understand why Brazil struggled to stop high inflation when public debt
and the primary deficit were not grossly out of line. Let b,π, and π e denote
period forward-
real public debt, one- looking inflation, and expected
6. The consumption boom phenomenon associated with stabilization programs has
received a lot of attention. Some outstanding alternative explanations do not rely on
imperfect credibility but on a combination of lower nominal interest rates, as a result
of lower inflation expectations and sticky prices. For example, see Rodriguez (1982).
164 Guillermo Calvo
inflation, respectively. For simplicity, I assume that, at RE equilibrium, the
period interest rate is equal to zero. Thus, under risk neutrality,
real one-
period inflation, π e,
the equilibrium interest rate will equal expected one-
period debt service bill (including amortization) in
in which case, the next-
real terms equals
1+ πe
b . (1)
1+ π
Therefore, given the rate of inflation, the larger expected inflation is, the
larger will be the real debt service burden. For simplicity, let us assume
that the government is bound to service debt in its totality at the end of
next period and that the central bank is obliged to rebate seigniorage to
the private sector in the form of a lump sum subsidy (so that seignior-
age net of rebate equals zero). The government is assumed to manage
the rate of inflation, π , by manipulating the rate of devaluation. Thus,
for instance, if output is homogeneous, there are no barriers to trade,
and international prices are constant in foreign exchange, it follows that
inflation equals the rate of devaluation: π = ε, where ε stands for the rate
of devaluation.
Under the above assumptions, expression (1) denotes the real tax rev-
enue necessary for debt service. I assume that the fiscal authority can com-
fortably generate tax revenue to service its debt if π = π e, but not a cent
more.7 It follows that the government will have to default if it sets π < π e and,
if default is too costly, it will be forced to make π ≥ π e and become hostage to
inflation expectations. For the casual observer, this would be a case of fiscal
dominance but, in essence, the situation is better characterized as a case of
expectations dominance, which becomes effective through the credit chan-
nel. Notice that across RE equilibriums in which π = π e, investors get the
same revenue. Hence, if the economy generates inflation higher than the
government’s target, the solution is Pareto inefficient. This problem holds
even in a world of RE, in which individuals are fully aware that the gov-
ernment’s inflation target is feasible if expectations are equal to the target.
In this case, however, RE depends on beliefs about market expectations. A
single individual has no command over the latter, and rationally aligns her
expectations to the expectations of others, a phenomenon that the GT calls
“expectations of expectations.”
7. In Calvo (1988), government is allowed to collect higher tax revenue.
From Chronic Inflation to Chronic Deflation 165
An interesting implication of the above example is that RE equilibrium
may be validated, not because individuals are rational but because policy
makers are forced to corroborate individuals’ expectations.
Calvo (1988) also shows that the problem would go away if the interest
rate on government bonds were indexed to the rate of inflation. In terms
of the above example, it is clear that if the rate of interest ex post was
set equal to the realized rate of inflation, the government would be able
to implement the target inflation rate independently of market inflation
expectations!8 This rule has been adopted in Chile through the Unidad de
Fomento (a unit of account) and may have helped to support inflation tar-
geting. Moreover, there seems to be wide consensus that eliminating infla-
tion uncertainty in financial contracts has helped financial deepening and
the development of the mortgage market (Fontaine 1996; Shiller 1998). In
other instances (e.g., the 1989 Bonex plan in Argentina), expectations dom-
inance led to denominating financial contracts in terms of US dollars. In
the simple model developed here, US dollar indexation gives similar results,
but this would not be the case if one allows for the existence of nontradable
goods, for example.
Expectations dominance can also have a deleterious effect on the private
sector. For example, if the economy comes from high inflation and people
have structured their contracts on the expectation that inflation will con-
tinue unabated, a cold turkey stabilization program, which stops inflation
in its tracks, will cause the same kinds of problems highlighted above. At
one point in the 1980s, for example, Brazil inflation was about 30 percent
per month. Imagine the impact of lowering inflation to single digits, annu-
ally! Several stabilization programs had to be abandoned, because keep-
high ex post real interest rates that would wreak
ing the course meant sky-
chaos in the financial sector and the payments system. This phenomenon
has been recently discussed in Lara Resende (2016). It bears some resem-
blance to Irving Fisher’s (1933) debt deflation theory. The latter, inspired by
the Great Depression, is a case in which the real value of debt skyrockets as
a result of a sharp and unexpected fall in the price level (during the Great
Depression, wholesale prices fell by more than 30 percent). In contrast, the
8. In practice, inflation indexation is applied with a lag. This may make indexation
less effective for shielding investors from inflation risk, especially during periods of
high and accelerating inflation. Moreover, financial indexation may lower policy
makers’ incentives for price stability.
166 Guillermo Calvo
harmful effects of cold turkey stabilization highlighted here would arise
even though prices do not fall and may continue rising, albeit at a sharply
lower rate than expected.
These problems are akin to what is called the “peso problem,” an expres-
sion popularized in the 1970s and 1980s as Mexico’s interest rates exceeded
the rate of devaluation by a wide margin (Lewis 2016). An explanation that,
in a way, foreshadowed RE was that the phenomenon was triggered by the
devaluation. This
expectation that Mexico’s peso would exhibit a maxi-
type of devaluation involves isolated jumps in the exchange rate. Thus,
interest rates will look “too large” during stretches in which the exchange
rate is constant. The peso problem is indeed closely related to the example
discussed above. However, in Calvo (1988), the authorities are forced to
validate devaluation expectations, despite the existence of another, more
benign, RE equilibrium. The latter has important policy implications,
because, for instance, it highlights the relevance of indexation for stop-
ping high inflation, even though policy makers are fully credible. Notice
that these implications would be missed in models displaying equilibrium
oriented macro models tend to favor.9
uniqueness, a feature that policy-
So far, the discussion has abstracted from debt default. A government
that is adamant on stabilizing inflation but is facing high inflation expecta-
tions may entertain the idea of default. This case is analyzed in Calvo (1988)
and further developed by Corsetti and Dedola (2016). A sketch follows.
Debt default can be analyzed in the context of a nonmonetary economy
employing the framework developed above. I will reinterpret inflation,
π , and inflation expectations, π e, as rate of default and expected rate of
default, respectively. In this instance, the expectation that the government
will default would force the government to default. In contrast to the infla-
tion example, solving this problem is likely to be more difficult. In the
inflation example, the problem would go away by adopting new types of
contracts (i.e., indexation). This approach is less likely to work if default is in
the cards, because the private sector may be less predisposed to believe the
government will honor its contracts. Therefore, to improve the situation,
9. The literature also abounds with backward- looking “wage indexation” as a factor
preventing speedy price stabilization. Although this could be reinterpreted as a case
of backward- looking expectations, I will refrain for discussing this issue here, given
this chapter’s emphasis on RE.
From Chronic Inflation to Chronic Deflation 167
it may be necessary to bring in independent parties that are willing and
capable to credibly insure investors against sovereign default. This is not
easy, given the legal privileges enjoyed by sovereign states. But it seems
to have worked in the Eurozone. Worried about the high interest rate pre-
mium in satellite Eurozone economies, reflecting investors’ concerns about
the solvency of those economies, Mario Draghi, president of the European
Central Bank (ECB), gave a speech on July 26, 2012, pledging to “do what-
ever it takes” to lower those interest rates. This was read by the market as an
ECB commitment to purchase as much of those sovereign debt instruments
free levels. It resulted
as necessary to squash their risk premium to default-
in an astonishing fall in those rates of interest, as predicted by the model.
Why the ECB can muster such impressive muscle is an important issue. A
common conjecture is that Germany is the actual credible lender of last
resort, in view of Germany’s strong fundamentals. But another conjecture
that cannot be dismissed is that the ECB can print credible liquidity. I will
revisit that issue in the next section.
contingent financial
Once again, intertemporal trade and nonstate-
contracts are at the heart of these problems. Fortunately, there is room for
policy, as illustrated by the Chile and ECB experiences mentioned above.
Remark 2. Staggered prices. Calvo and Végh (1993) extend the credibility
discussion to the case in which prices are set in advance in a staggered and
uncoordinated manner à la Calvo (1983). Results are in line with the above
analysis, but the richer environment helps show that, for instance, a non-
credible inflation stabilization program faces an additional powerful chal-
lenge. If agents fail to be persuaded that authorities have the determination
and public support to carry out the program, prices may continue rising at
a high rate despite tight monetary policy.
The results in Calvo and Végh (1993), taken at face value, imply that
controlling inflation might become easier if prices/wages were flexible.
However, this conclusion, which enjoys widespread appeal among policy
makers, would be hasty. In the next section, I argue that staggered prices
could play a fundamental role in a monetary economy. They could pro-
vide a stable output anchor to fiat monies and units of account, without
which a monetary economy may become unstable, unless the currency is
credibly anchored (but not necessarily pegged) to a resilient foreign cur-
rency, for example, the US dollar. This is common practice in EMs (see,
168 Guillermo Calvo
e.g., Calvo and Reinhart 2002), but credibility usually calls for large and
costly holdings of international reserves (see Calvo, Izquierdo, and Loo-
Kung 2013).
Sudden Stop, Chronic Deflation, and Sluggish Recovery:
Liquidity Explanations
The discussion in the previous section was framed in terms of conventional
macro theory under the assumption of RE. Until recently, the correspond-
ing models were taken with a high degree of confidence by policy makers.
However, amid that placid scenario, the Great Recession rose with shat-
tering force, putting into question everything, from RE to the feasibility
of capitalism. Minsky’s (2008) nightmares could no longer be discounted!
In this section, I start to explore the new issues by giving “liquidity”
a more central role than it had in mainstream macro theory prior to the
Great Recession. Otherwise, however, the models stick to the assumption
of RE and other assumptions of traditional economic theory. This smoothes
out the transition from the previous section, but the reader must be pre-
pared for a sharp turn, because the new vistas that the liquidity approach
conveys are anything but ordinary.
Liquidity is an issue that only recently has been given serious attention in
the literature (see, e.g., Holmström and Tirole 2011; Calvo 2016). This situa-
tion may be partly because mainstream models appeared to be adequate for
monetary policy before the Great Recession, at least for DMs. But I would
not discard the possibility that model builders were reluctant to focus on
liquidity issues because they cannot be easily accommodated in canonical
general equilibrium models. In other words: intellectual inertia was at work.
This section argues that liquidity offers promising insights, but we have to
make sure that we are treading on firm ground. Although liquidity has become
a ubiquitous word, “fashion over substance” seems to dominate. For example,
several observers claim that the Lehman 2008 crisis involved a phenomenal
backed
liquidity crunch on financial assets backed up by real assets (e.g., asset-
securities (ABS)). And they seem undisturbed to say, in the same breath, this
shock was accompanied by a flight to quality involving the US dollar, a fiat
money. Something is amiss here and forces us to delve into the reasons for
fiat money to hold positive value in terms of output, a characteristic of fiat
money that conventional macroeconomics tends to take for granted.
From Chronic Inflation to Chronic Deflation 169
The next subsection considers Frank Hahn’s (1965) fundamental obser-
vation that, as a general rule, conventional general equilibrium monetary
models cannot rule out the existence of barter equilibria. This result makes
to-
the flight- dollars phenomenon even more puzzling and enhances the
US-
relevance of finding plausible explanations for the resilience of money.
to-
The flight- money phenomenon was a central issue in Keynes’s GT (and
it is associated with what was elsewhere called the “liquidity trap”).10 In
an isolated and wholly ignored paragraph, the GT puts forward a simple,
but in my opinion insightful, conjecture that I labeled the price theory
of money (PTM).11 The PTM claims that money derives its liquidity and
positive purchasing power from the existence of staggered prices. Staggered
prices provide an output backing to money that, as a general rule, govern-
ments fail to give. Notice that this output backstop does not extend to
other liquid assets with flexible nominal prices.
Although staggered prices give a real platform for liquidity of money that
helps explain its resilience during episodes of financial crisis, this does not
because money’s output
rule out liquidity fragility or liquidity shortage—
backstop is anything but ironclad. This leads naturally (in the second sub-
section below) to considering a world with multiple monies and a variety of
backed securities, EM US dollar-
nominal liabilities (e.g., asset- denominated
bonds). Under these conditions, resilient and fragile liquid assets live next
to each other. Since, by definition, liquid assets are transactions facilitators,
a liquidity crunch of a subset of liquid assets generates a sudden decelera-
tion of transaction flows that rely on those assets. In practice, this takes the
a large and largely unexpected fall in credit
form of a credit sudden stop—
that could become systemic, given that liquidity is in the eye of the
flows—
beholder.12 These insights can also be employed as a guide for monetary
market operations
policy. It can be shown, for instance, that standard open-
and that the latter may
could be ineffective for restoring potential output—
10. I conducted a search in a Kindle edition of Keynes’s General Theory and could not
find the expression “liquidity trap.”
11. See Calvo (2012, 2016).
12. As noted at the start of the chapter, “sudden stop” is an expression introduced
to refer to severe contraction in international capital flows. The phenomenon has
also been observed in Europe during the Great Recession (see Merler and Pisani- Ferry
2012). Nowadays it has been extended to credit flows. To avoid confusion, I choose
to dub the latter “credit sudden stops.”
170 Guillermo Calvo
be better served by unconventional monetary policy instruments, which do
not call for lowering the central bank’s policy interest rate.
The third subsection below focuses on the case in which the official sec-
tor is unable to increase the stock of real liquidity. This could be the result
of having increased liquid public debt far beyond its output backing. I show
that this situation may generate chronic deflation. Finally, the fourth sub-
section below argues that liquidity shortage can also help rationalize “slug-
gish recovery” (also known as “secular stagnation”).
Hahn’s Problem, the Price Theory of Money, and Fear of Floating
The typical mainstream macro model assumes that there exists an object
usually denoted M—
called “money”— that provides liquidity services. A
in-
popular assumption in the literature is “cash- advance,” according to
which, to conduct market transactions, agents have to bring to market a
quantity of M proportional to the monetary (or nominal) market value of
planned purchases (e.g., the Clower (1967) constraint). In simple models,
the proportionality coefficient is assumed to be constant. Despite its sim-
in-
plicity, the cash- advance assumption dramatizes an important fact that
is easily ignored in nonmonetary economics, namely, that liquidity services
are essential for trade. In this setup, if M = 0, no trade is possible!
Let planned purchases be denoted by c (in terms of homogeneous real
output), and the real (or output) price of money (i.e., the inverse of the
price level) by Γ. Then, setting the factor of proportionality = 1, the cash-
in-
advance condition can be expressed as:
MΓ = c. (2)
Thus, as pointed out above, in equilibrium, if M = 0, then c =
0, and there
cannot be trade. But, what if Γ = 0? Clearly, the result is the same: Agents
will be doomed to operate under full autarky. Is Γ = 0 a possible equilibrium
outcome? Hahn (1965) shows that it is. The proof is trivial if M has no intrin-
sic market value, because in that case, money cannot buy output and the
situation is equivalent to bringing no money to the market.13 This is a deep
observation that does not apply to regular goods: If the price of bread is zero
in terms of other goods, say, there is likely to be excess demand for bread.
13. Notice that if holding M were a minor nuisance, its demand would be nil, causing
excess supply in the money market. However, by Walras’s Law, that does not gener-
ate excess demand in the rest of the economy, because the real price of money Γ = 0.
From Chronic Inflation to Chronic Deflation 171
output
There have been attempts to show conditions under which zero-
value of money can be ruled out; for example, assuming that real monetary
balances (i.e., MΓ) enter utility functions that satisfy Inada-
type condi-
tions.14 These conditions sound somewhat artificial in this case; moreover,
I do not think they are enough to rule out Γ = 0 If the latter holds, then
MΓ = 0, independently of how large M is. No matter how valuable monetary
balances would be for individual agents, there is nothing single individuals
can do to make MΓ > 0. In fact, as noted in a note in the previous paragraph,
if holding worthless M involves just a minor nuisance, agents would dump
M even though they are starving for MΓ > 0!
The GT offers a conjecture for why Γ >
0. In short, the conjecture is
that Γ > 0 because agents employ nominal prices to communicate to the
market the quantity of units of account (money, in this case) at which
they are ready to sell their staples. Moreover, they are prepared to keep
those prices “live” for some interval of time. Hence, nominal prices come
in-
first: We are in the world of “prices- advance.” For an individual agent
to have incentives to set her price in advance, it helps that a substantial
number of other agents have already posted their prices in similar fash-
ion, and that most of those prices can be taken for granted by present
price setters. So this is also a world of “staggered prices.” In this world,
individual price setters have a clear reference when setting their prices in
terms of money, because at time t, say, Γt is (essentially) predetermined
and positive.15 Moreover, keeping their price quotations live for a period
of time does not involve great risks of price misalignment if the expected
rate of inflation is low.16
The PTM can be criticized for being no more than a tautology: Γ > 0
because Γ > 0. But the case is subtler than this. The PTM states: Γt > 0 because
Γt –1 > 0, and just a few agents can or will change their prices at t. This mecha-
nism is incentive compatible: Price setters at t will have no incentives to set
their money prices = ∞ (which is equivalent to refusing to quote their prices
14. See Obstfeld and Rogoff (1983, 1986).
15. However, this does not necessarily imply inflation in advance. Thus, the output
backup of money will also be a function of inflation expectations, and the issues
raised in the second section of this chapter still apply.
16. However, the risks of setting prices in advance could be large in periods in which,
say, the economy is buffeted by large swings in its terms of trade, which involve
prices set outside the domestic economy.
172 Guillermo Calvo
in-
in terms of money). Compare this with canonical models like the cash-
advance model, or models in which real monetary balances are an argument
and prices are perfectly flexible. Even if
in utility or production functions—
Γt−1 > 0, in these canonical models, individuals have no incentives that would
rule out Γt = 0! Notice that the PTM does not rely on the existence of physical
money. It is a theory that applies equally well to a cashless economy with a
unit of account in terms of which prices are set in a staggered manner (see
Woodford 2003). To be sure, it would be interesting to explore the process
by which units of account are established, but that does not make canonical
models superior to the PTM, because models that are anchored on M instead
of Γ also need a rationale for the choice of a particular unit of account.
The PTM helps rule out Γ = 0 but does not guarantee that Γ will be stable
in realistic situations, because not all prices are set in terms of the same unit
of account.17 To wit, the world displays many units of account subject to
variable bilateral exchange rates. Interestingly, though, there is more sta-
bility in bilateral exchange rates than the existence of multiple currencies
would lead one to expect. For example, Calvo and Reinhart (2002) show that
called reserve currencies, a phenom-
EMs tend to peg their currencies to so-
enon called “fear of floating.” Reserve currencies are units of accounts that
are employed as invoice currencies in a wide variety of international trade
and financial transactions (see Gopinath 2016). Consequently, pegging to
a reserve currency strengthens EM currencies output backing, making them
more reliable as stores of value, which, in turn, enhances the liquidity of
reserve currencies. The US dollar is the king among reserve currencies and
has shown its muscle during the Lehman crisis, as the dollar appreciated
relative to other currencies, even though the US economy was at the epi-
center of the crisis. The US dollar privilege is rooted in considerations that
fall outside the scope of the present chapter, and I will not discuss them
here. However, it is worth pointing out that, especially in small EMs, the
realm of their national units of account is very limited. Thus, unless their
currencies are pegged to a reserve currency, their currencies’ output back-
ing would be very narrow, which could make them easy targets of currency
17. The PTM does not ensure uniqueness of the Γ path even if there exists a unique
unit of account. Uniqueness may require rules like the Taylor rule, a central topic in
New Keynesian literature. See Woodford (2003) and also Calvo (2016) for a skeptical
assessment of the relevance of New Keynesian models in that respect.
From Chronic Inflation to Chronic Deflation 173
runs’ episodes, and large currency devaluations or appreciations (recall the
sharp and surprising appreciation of the Swiss franc in January 2015).
To make the previous statements more intuitive, it is useful to think of
account with the stock of money on the liability
currencies in terms of a T-
side and a pot of goods (output) on the asset side. The pot of goods stands
for the currency’s output backing. This is similar to a bank’s balance sheet
hand side and illiquid loans on the left-
with deposits on the right- hand
side. In the present case, the pot of goods stands for the goods and services
that money holders can grab in exchange for money if they wish. The pot
of goods is likely to be smaller than the output value of money, ΓM. Hence,
as in banking models, there may exist multiple equilibriums (see, e.g., Dia-
mond and Dybvig 1983). In a “good” equilibrium, ΓM could far exceed the
pot of goods; in a “bad” one, ΓM would just be equal to the pot of goods.18
Accumulating international reserves in terms of reserve currencies increases
the pot of goods. It is intuitive that pegging, especially if accompanied by
reserve accumulation, is likely to diminish the probability of currency runs
and thus lowers the need for trade to rely on derivative markets, which
sized enterprises.
are costly and not easily available to small and medium-
This helps give a rationale to “fear of floating” and international reserve
accumulation.
It should be noted that fear of floating is not unique to EMs. During the
Lehman crisis, for example, the Fed signed a large currency swap agreement
with the ECB to prevent a wave of massive bankruptcies in the Eurozone
(with possible spillover effects on the United States), given that the Euro-
zone was undergoing a severe shortage of US dollars. Thus, despite the large
menu of national currencies, the world economy appears to be groping
like scheme with the US dollar as the nominal
toward a Bretton Woods–
(and hence, real) anchor.
A Larger Set of Liquid Assets: Sudden Stop
In practice, national currencies’ own rates of interest are nil. Thus, unless
monies. This
price deflation is rampant, there are incentives to create quasi-
process goes back to at least medieval banking (see Cipolla 1989) and ran
at full steam prior to the Great Recession. The phenomenon has already
18. Equilibria could be Pareto ranked by ΓM in models in which ΓM is an argument
in utility or production functions (or both) and exhibits positive partial derivatives.
174 Guillermo Calvo
been covered in multiple sources (e.g., Brunnermeier 2009), so here I just
highlight some salient features that relate to the discussion in the previous
monies take the form of
subsection. A common characteristic is that quasi-
income obligations denominated in terms of a unit of account. The
fixed-
old example is bank deposits backed up by a credible lender of last
age-
resort (typically, a central bank able to print currency or public liabilities
denominated in the bank deposits’ unit of account). A more recent example
backed securities (MBS), which are large pools of mortgage con-
is mortgage-
tracts denominated in terms of a unit of account. Barring systemic shocks,
pooling allows MBS to take advantage of the law of large numbers, reduc-
ing the need for information about individual contracts and exhibiting low
return volatility in terms of the corresponding unit of account. As a result,
bearing money.
securitized assets like MBS can come to resemble interest-
monies does not stop there.
The similarity between money and quasi-
monies, because they are subject to
Hahn’s problem also applies to quasi-
runs that are akin to those discussed in the banking literature (see Diamond
and Dybvig (1983) and the notes about national monies in previous sec-
tion). In those models, bank deposits provide liquidity services, but unless
there is a credible lender of last resort, other equilibriums exist in which a
sizable share of depositors tries to get their money out of the bank at the
same time, the bank goes bankrupt, and the liquidity services of the associ-
monies can occur even though their
ated deposits evaporate. Runs on quasi-
fundamentals show no fissure prior to the run, similar to the phenomenon
referred to under Hahn’s problem. Except for bank deposits fully ensured by
a lender of last resort, most other liquid assets have flexible prices in terms
of the unit of account. Hence, if the market refuses to take them as a means
of exchange, their price may plunge. Prices may not go to zero because, say,
MBS involve obligations that will eventually be at least partially honored,
but the price fall of these securities may still be significant.
monies play an important role as credit collateral (e.g., repurchase
Quasi-
agreements or repos). They do not circulate as fiat money or bank depos-
its, but they are important transaction facilitators for intertemporal trade
monies fall under the category of liquid assets
transactions. Therefore, quasi-
as defined here. Positive welfare effects generated by stable liquid assets are
bound to be very large, given that credit is essential for trade in modern capi-
talist economies. Without liquid assets, it would be hard to realize gains from
trade. A major problem, though, is that these assets are subject to liquidity
From Chronic Inflation to Chronic Deflation 175
crunch without warning and can cause major interruption of credit flows.
There is still no good understanding of how liquidity crunch takes place,
which leaves the credit market at the mercy of large shocks that are hard or
impossible to insure against. This problem is exacerbated by the fact that,
given that liquidity is only partially linked to standard fundamentals, a credit
crunch triggered by a liquidity crunch in one corner of the market can eas-
ily spread to the rest of the economy. Thus, a local liquidity crunch episode
could become systemic, a situation for which insurance markets are ineffec-
2008 sub-
tive. This phenomenon was clear in the 1998 Russian and 2007–
prime crises (see Calvo 2016). As noted above, a large interruption of credit
flows under these circumstances is called “sudden stop” and typically causes
(1) large capital loss in the financial sector and, more importantly, (2) casts
serious doubts on the reliability of liquid assets. The latter, in particular, con-
tributes to making these crises highly persistent (see Reinhart and Reinhart
2010; Calvo 2016, chapter 6). The Great Recession is a telling example.
The above observations were not central to the DM policy discussion
prior to the Great Recession. Instead, the opposite view prevailed. There was
wide consensus that DM financial systems ran like clockwork driven by the
hand of sophisticated operators (see Andrews 2008). And, moreover, if cri-
currency central banks could rapidly
sis erupted, the view was that reserve-
stabilize the situation by lowering their interest rates by a few basis points.
This view was partly based on the highly influential conjecture by Friedman
and Schwartz (1963) that the Great Depression would have been a regular
US recession if the Fed had kept the price level from plunging (e.g., in the
Great Depression, the Wholesale Price Index fell by more than 30 percent,
peak to trough). Unfortunately, the Great Recession put a question mark
Schwartz conjecture. The Fed and other reserve-
on the Friedman- currency
level deflation was avoided.
central banks followed the advice, and price-
lasting recession. In the
But these actions did not prevent a deep and long-
Eurozone, for example, GDP recovered its level prior to the Lehman cri-
sis only in 2016. To be sure, the evidence suggests that monetary expan-
sion was helpful, perhaps because it partially prevented a replay of I. Fisher
(1933) debt deflation,19 but the results are much worse than expected. What
is missing? The above discussion offers a clue: Central banks’ liquidity does
19. However, the Fed did not prevent debt deflation in the housing market, where
dollar prices fell by about 30 percent.
176 Guillermo Calvo
not necessarily solve liquidity problems triggered by the liquidity crunch,
unless such liquidity is directed to restore the market for liquid assets hit by
crisis (see Calvo 2012). Without that directed restoration policy, credit flows
stop and can cause major damage. Liquid assets are not born equal, indeed!
DM central banks became aware that something was seriously amiss
when they hit the zero lower bound, and they adopted policies aimed at
unclogging the credit channel in a more direct fashion. It took the form of
quantitative easing (QE), such as central bank purchases of MBS and mea-
sures that directly stimulate credit to the private sector. The ECB, for exam-
ple, announced a modus operandi on March 7, 2016, that among other
things, expands the scope of a liquidity window for some corporate bonds,
and de facto subsidizes loans to the private sector. All of these actions are
consistent with the view that the liquidity crunch calls for heterodox cen-
tral bank policy (which, incidentally, is dangerously close to being cata-
logued as a surreptitious form of fiscal policy).
Remark 3. Some microfoundations. To clarify the discussion, let us consider
backed security, which underlying
a simple case in which there is an asset-
asset I identify as “land.” Land, denoted by k, is in fixed supply and is subject
to no maintenance costs. Output is a function of land as a standard factor of
production, but in addition, land is a transactions facilitator for firms; land’s
liquidity (measured in terms of output) also has a positive effect on output.
Hence, I assume that output is given by f(θqk), where q and θ are the output
price of land and a liquidity coefficient, respectively; θ is between 0 and 1. Let
interest rate on output) be denoted by r.
the real interest rate (i.e., the own-
Then, at a steady state in which q is expected to be constant over time, profit
maximization at k > 0 implies the following first-
order condition with respect
to k: f'(θqk)θ = r. One can show that if function f is Cobb-
Douglas, the price
of land q rises with the liquidity coefficient θ. Hence, a liquidity crunch on
land could bring about a collapse in the relative price of land with respect to
output. In this simple setup, money supply has no role to play. Therefore, if
indebtedness, stan-
the price of land causes side effects like unplanned over-
dard monetary policy cannot help. One needs instruments that can have an
impact on q. The unconventional purchase of toxic assets, as in the Fed’s ini-
tial quantitative easing program, is a possible, albeit not foolproof, example.20
20. These issues are discussed in greater detail in Calvo (2012) and Calvo (2016,
chapters 3 and 5).
From Chronic Inflation to Chronic Deflation 177
As noted above, liquidity crunch is no DM monopoly. The systemic EM
crises in the 1990s can also be characterized in the same way. But there
are important differences. Consider the 1997/1998 Asian/Russian crises,
which involved a run against EM bonds floated in the international capital
market. First and foremost, unlike in DMs, those bonds were denominated
in US dollars or other reserve currencies, not EM domestic monies. The
meltdown could have been prevented by a massive purchase by EM bonds
using international reserves, or drawing on credit lines from an interna-
tional lender of last resort (e.g., the IMF). But the latter was not available,
and EMs had neither the resources (i.e., international reserves) nor the abil-
ity to launch a coordinated counteroffensive. Therefore, this gave rise to
a sudden stop episode that, employing the metaphor in an earlier subsec-
tion, lowered the pot of goods backing up domestic EM money and trig-
in sharp contrast with the
gered currency devaluation, not appreciation—
United States during the Lehman crisis. Furthermore, currency devaluation
currency-
weakened EM balance sheets, because foreign- denominated debt
is partly employed to fund projects denominated in domestic currency.
a hallmark of EM sudden stops—
Thus, large devaluation— brought about
harmful effects that are akin to I. Fisher debt deflation, as the value of debt
obligations skyrocketed relative to the flow of domestic currency revenue,
exacerbating the depth of the financial crisis. Clearly, high initial debt and
low levels of international reserves enhance the severity of the crisis. These
conditions prevailed prior to the Russian crisis, because, in my opinion, few
investors and policy makers foresaw the massive systemic meltdown that
occured in the Russian crisis.
Interestingly, after the Asian/Russian crises, favorable circumstances that
gave rise to improving current account balances and large accumulations
of international reserves in several Asian and Latin American economies
placed those economies on a stronger footing to face the 2008 Lehman cri-
sis (see International Monetary Fund 2010, chapter 2). The shock was felt,
but recovery was fast and was followed by a string of relatively high growth
rates, which suggests that the size of the “pot of goods” makes a difference.
This idea is also borne out by empirical research (see Calvo, Izquierdo, and
Kung 2013; Calvo 2016; Calvo, Izquierdo, and Mejía 2016).
Loo-
As argued in an earlier subsection, fear of floating could be traced back
to an attempt by EMs to anchor their currencies on reserve currencies. This
works for regular shocks but it is probably too costly to prevent currency
178 Guillermo Calvo
runs in a sudden stop episode. Still, sizable international reserves could
help contain runaway inflation. The reason is simple: Employing the meta-
phor in an earlier subsection, devaluation increases the nominal value of
the asset side of the balance sheet (the “pot of goods”) without, in prin-
ciple, changing the supply of money. Therefore, money’s output backup
becomes stronger and gives the central bank more ammunition to stop
inflation from spiraling out of control. However, it is easy to show that if
the central bank intervenes and stops devaluation in its tracks, money’s
output backup would weaken, in the normal situation in which monetary
domestic liabilities exceed international reserves. This helps explain why,
during the recent sizable contraction of capital flows to EMs, many coun-
tries in Latin America decided to meet the shock with large devaluations
and only modest sacrifices of international reserves. Spiraling inflation, the
nemesis of these economies in the 1980s, has not been a major problem
(see International Monetary Fund 2016, chapter 2).
Remark 4. Endogenous liquidity: Currency substitution. Liquid assets have
a long history in which tyrants and wars play a major role. But liquid assets
also owe their existence to much more friendly technical change and run-
the-
of- mill incentives. EMs are a rich laboratory that illustrates that high
inflation, for instance, can give rise to the creation of local liquid assets in
the form of foreign currencies, a phenomenon labeled “currency substitu-
Végh discussion in Calvo 1996). The foreign curren-
tion” (see the Calvo-
cies in question are typically reserve currencies, but they need some help
from domestic agents to become liquid at the local level. Incentives for
the creation of liquid assets or arrangements can also take very different
forms. Gorton and Metrick (2012), for instance, claim that shadow banks
were partly prompted by attempting to offer more reliable deposit insur-
ance arrangements for large depositors, such as pension funds.
The topic of endogenous liquidity is still in its infancy. The currency
substitution literature calls attention to some constraints that the phenom-
enon imparts to monetary policy, but I feel that the literature has scarcely
scratched the surface. Taking an approach similar to that of the micro bank-
ing literature (e.g., Diamond and Dybvig 1983), for instance, suggests the
existence of sharp discontinuities or nonlinearities that I do not think have
been fully exploited in the currency substitution literature. Moreover, a better
understanding of endogenous liquidity could help establish a more solid
currency interest rates, a highly
grasp on the implications of low reserve-
From Chronic Inflation to Chronic Deflation 179
topical issue. For instance, this type of theory may help rationalize the com-
monly heard statement that low international interest rates are spawning
EM fragile liquid assets that are subject to costly runs.
The Deflation Cycle: Chronic Deflation
Price deflation has pushed chronic inflation from center stage, and issues
from the distant past, like liquidity traps, have come back with a ven-
geance. Thus, momentarily at least, the voluminous inflation literature will
fashioned deflation papers and a few essays by eco-
be swapped for old-
nomic historians of the Great Depression. It is worrisome, though, that past
deflation episodes occurred under very different circumstances and data are
scant. Moreover, although chronic deflation could be partly explained by
indebtedness and balance sheet problems (e.g., Koo 2009), these prob-
over-
lems could well have arisen in a hyperinflationary context, as highlighted
in Sargent (1982). This motivated me to try alternative explanations.
In this subsection, I explore a tentative road inspired by the PTM. The
basic idea is straightforward. Consider an economy in which (fiat) money
is the only liquid asset. Money enjoys some output backup thanks to the
existence of sticky prices. In that context, doubling the stock of money
but it does not necessarily
supply doubles real monetary balances— double
money’s output backup. If money’s output backup stays constant, for
instance, the expected purchasing power of money may less than double.
In Calvo (2016), I call this effect “liquidity deflation.” It is tantamount to
a pecuniary externality for atomistic agents. The initial doubling of the
money supply may make people feel that their monetary wealth has dou-
bled in real terms, but they will soon be disabused of this notion as they
realize that they would have to share money’s output backup with the rest
of the agents, even if prices are sticky.
It is interesting to compare the above situation to the conventional one
in monetary theory, in which individuals assess money’s liquidity services
by their individual holdings of real monetary balances. Suppose, for sim-
plicity, that prices are flexible and the demand for the liquidity of real mon-
etary balances is constant. Hence, in the conventional model, doubling
money supply, will double the equilibrium price level. In contrast, if liquid-
ity deflation is at work, prices may less than double. Therefore, liquidity
deflation gives a rationale for the difficulties central banks may find in stop-
ping deflationary forces by expanding their balance sheets. This reasoning
180 Guillermo Calvo
applies with special force to reserve currencies, for which it is difficult to
find more reliable alternative liquid assets. Formal details follow.
To stay on familiar ground, I will start focusing on the Pigou effect, a
pivotal concept for the classical (as defined in the GT) argument against
the relevance of the liquidity trap, according to which wage and price flex-
ibility could help restore full employment. Formally, the argument is that
the liquidity of real monetary balances, MΓ, rises without bound as the
price level falls (i.e., as Γ rises). Under normal circumstances, the associ-
ated wealth effect will lift aggregate demand (this is the Pigou effect), a
process that will not stop until full employment is restored. This argument
ignores I. Fisher’s (1933) debt deflation, but I will not let this distract us,
because the main point is to show that the argument could be fallacious
nonetheless.
The Pigou effect relies on the assumption that economic agents will take
MΓ as a highly reliable yardstick of how much output can be fetched in the
market by exchanging MΓ for output, even in cases where aggregate MΓ
exceeds total nonmonetary wealth by a large margin. This assumption is
consistent with individual rationality under the assumption that there is no
run against money. The latter may not sound like a strong assumption for
monies,
the US dollar, but runs cannot be discounted if M contains quasi-
even if the latter are indexed to the US dollar (as illustrated by ABS’s melt-
down in the Lehman crisis; see Gorton and Metrick 2012). Thus, if runs are
in the cards, it is plausible to argue that, beyond a certain point, an increase
in MΓ may be equivalent to less output in case of a run, as individuals rush
to exchange money for output and take advantage of price stickiness while
it lasts (recall the metaphor in an earlier subsection). Therefore, agents that
take runs into consideration will attach a liquidity coefficient to MΓ that is
less than unity. This corresponds to the liquidity deflation effect mentioned
above. Following these lines, I assume that the liquidity of money for a
single individual is given by the expression:
(3)
MΓ + Z (( MΓ )e ), Z ′ < 0,
where (MΓ)e stands for equilibrium aggregate real monetary balances, and
the function Z captures liquidity deflation. This is equivalent to assum-
ing that it is rational for a single individual in an atomistic environment
to take her own MΓ as real wealth but adjusts liquidity services of money
downward as a function of aggregate MΓ. Liquidity deflation opens the
From Chronic Inflation to Chronic Deflation 181
possibility that the expansionary effect of a larger stock of real monetary
balances fizzles out as monetary balances become large.
in-
To couch the discussion in more familiar terms, consider the cash-
hand side the new definition of
advance equation (2), and stick on the left-
liquidity services from equation (3). Because (MΓ)e = MΓ in an RE equilib-
rium with a representative individual, we get
MΓ + Z(MΓ) = c. (4)
Clearly, it is now conceivable that the Pigou effect is nil, because the
wealth effect is offset by the negative liquidity effect. Hence, a fall in the
price level, or an increase in money supply, given the price level, could have
no effect on aggregate demand. Suppose, for example, that real liquidity
hits the upper bound and the associated aggregate demand is below full
capacity output. This would tend to depress the price level, which exacer-
lowering money’s output backup and eventually
bates liquidity deflation—
triggering a run against M that destroys money’s liquidity. Notice that the
and the resulting liquidity trap—
failure of the Pigou effect— highlighted
side considerations. I will call it the “supply-
here is due to supply- side
liquidity trap.” This is radically different from the GT rationale, which
relies on the assumption that the demand for money is infinitely elastic
with respect to “the” interest rate. It is worth noting, though, that GT
liquidity traps and liquidity deflations are complementary rationales for
situations in which increasing money supply has a hard time stimulating
output.
Remark 5. ECB puzzle. At the end of a previous subsection, I referred to
the highly successful ECB strategy for lowering risk premiums on some
Eurozone sovereign bonds, which consisted of announcing that the bank
“would do whatever it takes” to achieve this objective. Given the small
ECB capital relative to the stock of sovereign bonds from vulnerable econ-
omies (e.g., those of Italy and Spain), a popular and plausible conjecture
is that success of the strategy stems from the expectation that Germany
would bail out the ECB if necessary. This conjecture is in accord with the
above discussion, because Germany would be providing the “pot of goods”
behind the ECB liabilities. It is interesting, though, that in 2007/2008,
when the Great Recession reached a boiling point, the actual lender of last
resort happened to be the Fed! The Fed’s comparative advantage over Ger-
many under those circumstances was its capacity to print US dollars, an
182 Guillermo Calvo
asset toward which the whole world was running for safety. This suggests
that even though the ECB was very successful in lowering risk premiums
in the Eurozone, it may again need the support of the Fed if, for instance,
the federal funds rate rises faster than expected. Thus, it would be a mis-
free, simply because the ECB was able to
take to think that the euro is run-
lower risk premiums. This observation implies that the assumption behind
liquidity deflation above is not vacuous, even in the case of a reserve cur-
rency like the euro.
side liquidity trap. The above results may
Remark 6. More on the supply-
look confusing to those familiar with the standard approach in monetary
theory (see, for instance, Patinkin 1965, where individuals internalize the
pecuniary externalities introduced in expression (3)). Thus, if one follows
in-
the standard approach, the cash- advance constraint would take the
form of equation (4) above. Let MΓ denote the value of real monetary bal-
ances that maximize MΓ + Z(MΓ). If MΓ is not large enough to generate full
capacity utilization, then the situation would be one of real money short-
age. But it would not correspond to a liquidity trap, because an increase in
money supply will paradoxically generate excess supply of money and, if
nominal prices are upwardly flexible, it would result in a fall in Γ (i.e., an
increase in the price level) that pushes real monetary balances back to MΓ.
trained economists,
This would validate the view, popular among well-
that an increase in the supply of money raises nominal prices, unless the
GT liquidity trap holds and the demand for money is infinitely interest
elastic.
In contrast, if the pecuniary externality is not internalized, as assumed
in expression (4), increasing M when MΓ = MΓ , given Γ, implies of course
that MΓ > MΓ . The larger stock of real monetary balances MΓ yields lower,
not higher, liquidity services, because MΓ + Z(MΓ) is maximized at MΓ , and
not less, as implied
individuals will vie for more real monetary balances—
in the standard approach. This situation, if anything, will put downward
pressure on the price level, raising MΓ even further and driving the system
into a vicious chronic deflation cycle.
An interesting extension of the model that can also help to make the
new results more intuitive is to assume that (MΓ)e runs behind MΓ. Con-
sider the following example:
( MΓ )te+1 = MΓ t , (5)
From Chronic Inflation to Chronic Deflation 183
which, taking equations (3) and (5) into account, implies
MΓt + Z(MΓt−1) = ct. (6)
Hence, an increase in money supply will succeed in stimulating aggregate
demand at time t, but money stock will have to continue rising to prevent
liquidity deflation from catching up.
In this example, even if initially MΓ = MΓ (recall remark 6), the cen-
tral bank would be able to generate full capacity utilization by helicopter
money, say, but it will have to continue doing so to prevent renewed reces-
sionary pressures and possibly price deflation. This scenario is interesting,
because it is an example in which deflation is a persistent threat requiring
an endless expansion of money supply: Pigou meets Sisyphus!
An interesting twist is to replace equation (4) by
MΓ + Z ( MΓ ) = L(i − i m , y ), Li −im < 0, Ly > 0, (7)
where L is the standard textbook liquidity preference function, and im
stands for the interest rate on money. The latter is a shortcut of the Calvo
and Végh (1995) model in which money is a mix of cash and treasury
bills, and im can be interpreted as the interest rate controlled by the central
bank (e.g., the federal funds rate in the United States).21 To put equation
(7) through its paces, note that in the IS/LM apparatus, equation (7) cor-
responds to the LM curve. Thus, a rise in im will increase the demand for
money (i.e., will shift up the LM curve) and generate output contraction.
Note that contraction holds even in the case in which QE is ineffective.
This helps rationalize the opinion, popular in current debate, that QE is
no longer effective, but a rise in the Fed’s rate can deepen the extent of
recession.
However, the impact of increasing im could have the opposite effect.
For the sake of the exposition, I will assume equation (5). Suppose that
money (including other safe assets) has a role as a medium of exchange
for firms’ transactions. This can be captured by assuming that real mon-
etary balances, MΓ, enter the production function. Let the latter be denoted
by F(MΓt + Z(MΓt−1)), where the function F is strictly concave and satisfies
21. Technical note. The absence of the Liquidity Deflation term Z from the demand
individual model
side in equation (7) holds if derived from a standard representative-
in which MΓ + Z((MΓ)e) is an argument in the utility function. However, this would
not hold true if the Z function multiplies MΓ.
184 Guillermo Calvo
Inada conditions around 0. The representative firm’s profit (in real terms)
is given by:
F(MtΓt + Z(Mt−1Γt−1))−(i−im)MtΓt. (8)
order condition with respect to Mt is
Thus, the first-
F ′( M t Γ t + Z ( M t −1Γ t −1 )) = i − i m . (9)
Hence, lowering the central bank interest rate, im, leads to a fall in output
(and the zero lower bound is a nonissue), because it increases the oppor-
tunity cost of money holdings. The negative output effect from lower im
would also hold if money had a role as credit collateral. I find it curious that
the literature and policy debate systematically assumes that “easy money”
is expansionary, despite the popularity of the literature that highlights col-
lateral assets (e.g., Kiyotaki and Moore 1997) and the central role of collat-
eral meltdown in the Lehman crisis (see Gorton and Metrick 2012).22 Notice
state output is achieved at
that under these assumptions, maximum steady-
MΓ = MΓ . If this output level is thought to be too low, interest rate policy
alone could not help take the economy out of that rut. As in the previous
case, the central bank will be doomed to rely on unconventional monetary
policy in aeternum.
In sum, liquidity deflation could generate chronic deflation. Standard
and unconventional monetary policy may fail to generate the liquidity
necessary to restore full employment. Moreover, as deflation proves to be
much more resilient than expected and output is dragged down by lack of
aggregate demand, the private sector may start considering money as an
attractive investment vehicle, exacerbating price deflation. These effects
will be less acute if the economy operates below MΓ, but they may start
to be felt, leading policy makers to turn their attention to alternatives like
fiscal policy. This may be the right way to go. However, given credit mar-
ket difficulties, it would be misleading to analyze the effects of fiscal policy
while ignoring financial constraints. Liquidity shortage could have a major
impact on the size of the Keynesian multiplier. Ilzetzki, Mendoza, and Végh
(2013), for instance, found that the multiplier is negative in highly indebted
economies.
22. For further discussion on this topic, see Calvo (2016).
From Chronic Inflation to Chronic Deflation 185
Remark 7. Spillover effects. Liquidity shortage and deflation in DMs
could spill over to EMs, generating new liquid assets centered on EM lia-
bilities (Gorton 2017; Calvo 2016). For EMs that display large international
reserves, this situation may enhance the liquidity of public sector obliga-
through coefficients and making
tions, for example, leading to lower pass-
inflation targets easier to achieve. This is, in principle, good news for EMs
but as usual, there is also a dark side: Liquidity of EM liabilities is likely to
be sensitive to DM interest rates.
side liquidity trap
In closing, it is worth pointing out that the supply-
phenomenon discussed here is a close relative to the burgeoning safe-
asset shortage literature (see Caballero, Farhi, and Gourinchas 2016).
Both emphasize difficulties in stimulating aggregate demand or output
related factors. The value-
supply due to supply- added of the approach in
this chapter is that these factors are linked to liquidity, traced to the large
loss of liquidity (in e.g., the inception of the Great Recession) and the dif-
ficulty of increasing liquidity by pumping in reserve currency public sec-
denominated)
tor liabilities, or a fall in the international (e.g., US dollar-
side liquid-
price level. Moreover, the discussion suggests that the supply-
ity trap for reserve currencies is linked to collateral trouble in the credit
channel that lowers the output backstop of liquid assets, a topic addressed
next.
Sluggish Recovery
Empirical evidence shows that economies may take long to recover from
severe financial crises (e.g., Reinhart and Reinhart 2010). The Great Reces-
sion is a striking example. In 2016 the European Union was still strug-
gling to recover its output peak in 2008. The United States has been more
successful, but output is still now below trend. This phenomenon has
been attributed to the credit boom prior to the crisis and resulting over-
indebtedness (e.g., Koo 2009; Reinhart and Reinhart 2010; Taylor 2015).
heory has put financial frictions and imperfections at center
Naturally, t
although, it should be noted, more as amplifiers than as main trig-
stage—
gering factors (see, e.g., Queraltó 2013). Less attention has been paid to
liquidity fragility, a birth defect of the financial sector. I am afraid that this
hanging fruits” that
bias may result in losing sight of some valuable “low-
help explain not only sluggish recovery but also other central features of
186 Guillermo Calvo
systemic financial crises (e.g., nominal price deflation). A model displaying
those features is discussed in Calvo (2016, chapter 5). I will sketch it out in
what follows.
economy, representative-
Consider a closed- agent model under perfect
to-
price flexibility. Output can be allocated on a one- one basis to consump-
in-
tion or raw materials, and households are subject to a cash- advance con-
straint, similar to equation (2) above, where now M stands for fiat money.
in-
The representative firm is also subject to a liquidity- advance constraint
for its raw material purchases. Moreover and realistically, I assume that the
firm can hold both fiat money and highly liquid securities, say, ABS. The
return on ABS, including liquidity services, is also a function of its liquidity
≤ θ ≤
coefficient, indicated by θ in the formal model (0 1). Clearly, if θ = 0,
ABS cannot be employed to satisfy the firm’s liquidity constraint, and the
firm will hold liquidity entirely in the form of fiat money. In contrast, if
θ =
1, ABS would be perfect substitutes for fiat money and, under normal
dominate the latter. Thus, I assume that if θ = 1,
circumstances, will return-
firms would prefer to hold their entire liquidity portfolio in ABS. The for-
mal model considers intermediate cases, but the two limit cases are enough
for illustration.
Liquidity crunch is defined as a sudden exogenous fall in the parameter
θ. For motivation, this can be thought of as a run on ABS along Diamond-
Dybvig (1983) lines. Consider the case in which, initially, θ = 1, and as a result
of the liquidity crunch, θ goes all the way down to 0. Because the return on
ABS prior to the crisis is higher than the return on fiat money, the return
on the liquid portfolio that the firm is constrained to hold in advance will
be lower after the liquidity crunch. This increases the cost of raw materials
and, if the production function satisfies Inada’s conditions, induces a fall
in output. If consumers were the only holders of fiat money and money
supply was given, the slump would cause a rise in the price level, because
output contraction would bring about a fall in the demand for fiat money.
But in this model, an additional effect points in the opposite direction,
because as noted, the liquidity crunch provokes a massive switch in firms’
liquid portfolio from ABS to fiat money. This switch can offset the fall in
the demand for money from households and cause price deflation. Thus,
the model can rationalize price deflation, even though the cards were
stacked against it by the assumption that households are subject to a cash-
in-advance constraint.
From Chronic Inflation to Chronic Deflation 187
The model can be extended to a growth context in which the liquidity-
advance constraint applies to investment. In a model in which output
in-
is proportional to the stock of capital, one can show that the rate of capital
accumulation is a negative function of the opportunity cost of liquidity.
Thus, for instance, a liquidity crunch would bring about a fall in growth
in-
(i.e., sluggish recovery). Moreover, if liquidity- advance also applies to the
purchase of raw materials, the liquidity crunch will bring about output con-
traction on impact, possibly accompanied by price deflation (as discussed
in the previous paragraph).
Some policy experiments in terms of this model are conducted in Calvo
(2016, chapter 5). Here I just note that, despite its simplicity, the model cap-
tures several realistic features associated with liquidity crunch. This suggests
that policies that aim at restoring the economy’s vitality after a liquidity
crunch should pay special attention to factors that caused the crunch and
moderate its effects. Actually, some popular policies that do not address those
issues may fail to work. For instance, an increase in money supply or govern-
ment expenditure would be totally ineffective, unless they help to restore
the liquidity of ABS without simultaneously provoking a large drop in their
pure rates of return (i.e., rates of return that do not include liquidity services).
References
Andrews, Edmund L. 2008. “Greenspan Concedes Error on Regulation.” New York
Times, October 23.
Bordo, Michael D., and Athanasios Orphanides, eds. 2013. The Great Inflation: The
Rebirth of Modern Central Banking. Chicago: University of Chicago Press.
Brunnermeier, Markus K. 2009. “Deciphering the Liquidity and Credit Crunch
100.
2007–2008.” Journal of Economic Perspectives 23 (1): 77–
Bruno, Michael, Guido Di Tella, Rudiger Dornbusch, and Stanley Fischer. 1988.
Inflation Stabilization: The Experience of Israel, Argentina, Brazil, Bolivia and Mexico.
Cambridge, MA: MIT Press.
Bruno, Michael, Stanley Fischer, Elhanan Helpman, and Nissan Liviatan, with Leora
(Rubin) Meridor. 1991. Lessons of Economic Stabilization and Its Aftermath. Cam-
bridge, MA: MIT Press.
Buffie, Edward F., and Manoj Atolia. 2012. “Resurrecting the Weak Credibility
Hypothesis in Models of Exchange- Based Stabilization.” European Economic
Rate-
372.
Review 56 (3): 361–
188 Guillermo Calvo
Olivier Gourinchas. 2016. “Safe
Caballero, Ricardo J., Emmanuel Farhi, and Pierre-
518.
Asset Scarcity and Aggregated Demand.” American Economic Review 106 (5): 513–
Cagan, Phillip. 1956. “The Monetary Dynamics of Hyperinflation.” In Studies in the
117. Chicago: University
Quantity Theory of Money, edited by Milton Friedman, 25–
of Chicago Press.
Calvo, Guillermo A. 1978. “On the Time Consistency of Optimal Policy in a Mon-
1428.
etary Economy.” Econometrica 46 (6): 1411–
Calvo, Guillermo A. 1983. “Staggered Prices in a Utility Maximizing Framework.”
398.
Journal of Monetary Economics 12 (3): 383–
Calvo, Guillermo A. 1986a. “Fractured Liberalism: Argentina under Martínez de
533.
Hoz.” Economic Development and Cultural Change 34 (3): 511–
Calvo, Guillermo A. 1986b. “Temporary Stabilization: Predetermined Exchange
1329.
Rates.” Journal of Political Economy 94 (6): 1319–
Calvo, Guillermo A. 1988. “Servicing the Public Debt: The Role of Expectations.”
671.
American Economic Review 78 (4): 647–
Calvo, Guillermo A. 1989. “Incredible Reforms.” In Debt, Stabilization and Develop-
ment, edited by Guillermo A. Calvo, Robert Findlay, Pentti Kouri, and Jorge Braga de
Macedo, 217– 234. New York: Basil Blackwell.
926.
Calvo, Guillermo A. 1991. “The Perils of Sterilization.” IMF Staff Papers 38 (4): 921–
Calvo, Guillermo A. 1996. Money, Exchange Rates, and Output. Cambridge, MA: MIT
Press.
Calvo, Guillermo A. 1998. “Capital Flows and Capital- Market Crises: The Simple
54.
Economics of Sudden Stops.” Journal of Applied Economics 1 (1): 35–
Run Per-
Calvo, Guillermo A. 2012. “Financial Crises and Liquidity Shocks: A Bank-
326.
spective.” European Economic Review 56 (3): 317–
Calvo, Guillermo A. 2016. Macroeconomics in Times of Liquidity Crises: Searching for
Economic Essentials. Cambridge, MA: MIT Press.
Calvo, Guillermo A., and Allan Drazen. 1998. “Uncertain Duration of Reform:
455.
Dynamic Implications.” Macroeconomic Dynamics 2 (4): 443–
Calvo, Guillermo A., and Pablo Guidotti. 1990. “Indexation and Maturity of Gov-
ernment Bonds: An Exploratory Model.” In Capital Markets and Debt Management,
edited by Rudiger Dornbusch and Mario Draghi, 52–82. New York: Cambridge Uni-
versity Press.
Calvo, Guillermo A., and Carmen M. Reinhart. 2002. “Fear of Floating.” Quarterly
408.
Journal of Economics 117 (2): 379–
From Chronic Inflation to Chronic Deflation 189
Calvo, Guillermo A., and Carlos A. Végh. 1993. “Exchange- Based Stabilisation
Rate-
under Imperfect Credibility.” In Proceedings of the IEA Conference on Open-
Economy
Macroeconomics, edited by H. Frisch and A. Worgötter, 3– 28. Hampshire, UK: Mac-
millan. Reprinted in Calvo (1996, chapter 18).
Calvo, Guillermo A., and Carlos A. Végh. 1995. “Fighting Inflation with High Inter-
Open-
est Rates: The Small- Economy under Flexible Prices.” Journal of Money, Credit,
66.
and Banking 27 (1): 49–
Calvo, Guillermo A., Alejandro Izquierdo, and Rudy Loo- Kung. 2013. “Optimal
Insurance against Sudden Stop.” Monetaria
Holdings of International Reserves: Self-
35.
35 (1): 1–
Calvo, Guillermo A., Alejandro Izquierdo, and Luis- Fernando Mejía. 2016. “Systemic
Sudden Stop: The Relevance of Balance- Sheet Effects and Financial Integration.” In
Macroeconomics in Times of Liquidity Crises: Searching for Economic Essentials, edited by
Guillermo A. Calvo, 143–200. Cambridge, MA: MIT Press.
Cavallo, Eduardo A., and Jeffrey A. Frenkel. 2008. “Does Openness to Trade Make
Countries More Vulnerable to Sudden Stops, or Less? Using Gravity to Establish
1452.
Causality.” Journal of International Money and Finance 27 (8): 1430–
Century Florence. Berkeley: University of
Cipolla, Carlo M. 1989. Money in Sixteenth-
California Press.
Clower, Robert. 1967. “A Reconsideration of the Microfoundations of Monetary
9.
Theory.” Western Economic Journal 6 (1): 1–
Corsetti, Giancarlo, and Luca Dedola. 2016. “The ‘Mystery of the Printing Press’:
Monetary Policy and Self-Fulfilling Debt Crises.” CEPR DP11089, Centre for Eco-
nomic Policy Research, London.
Diamond, Douglas W., and Philip H. Dybvig. 1983. “Bank Runs, Deposit Insurance
419.
and Liquidity.” Journal of Political Economy 91 (3): 401–
Dornbusch, Rudiger, and Mario Henrique Simonsen. 1983. Inflation, Debt, and Index-
ation. Cambridge, MA: MIT Press.
Eichengreen, Barry. 2015. Hall of Mirrors: The Great Depression, the Great Recession,
and the Uses—
and Misuses—of History. Oxford: Oxford University Press.
Deflation Theory of Great Depressions.” Economet-
Fisher, Irving. 1933. “The Debt-
357.
rica 1 (4): 337–
Fontaine, Juan Andres. 1996. “La Construcción de un Mercado de Capitales. El caso
de Chile.” EDI Learning Resources Series. Washington, DC: World Bank.
Friedman, Milton. 1971. “Government Revenue from Inflation.” Journal of Political
856.
Economy 79 (4): 846–
190 Guillermo Calvo
Friedman, Milton. 1977. “Nobel Lecture: Inflation and Unemployment.” Journal of
472.
Political Economy 85 (3): 451–
Friedman, Milton, and Anna J. Schwartz. 1963. A Monetary History of the United
States, 1867–1960. Princeton, NJ: Princeton University Press.
Gopinath, Gita. 2016. “The International Price System.” In Inflation, Dynamics and
Monetary Policy, Proceedings of the Federal Reserve Bank of Kansas City, August 27–
29,
2015, Jackson Hole, WY, 71–150.
Gordon, Robert J. 2011. “The History of the Phillips Curve: Consensus and Bifurca-
50.
tion.” Economica 78 (309): 10–
Gorton, Gary. 2017. “The History and Economics of Safe Assets.” Annual Review of
586.
Economics 9 (1): 547–
Gorton, Gary, and Andrew Metrick. 2012. “Securitized Banking and the Run on
451.
Repo.” Journal of Financial Economics 104 (3): 425–
Guidotti, Pablo E., and Manmohan S. Kumar. 1991. “Domestic Public Debt of Exter-
nally Indebted Countries.” IMF Occasional Paper 80, International Monetary Fund,
Washington, DC.
Hahn, F. H. 1965. “On Some Problems of Proving the Existence of an Equilibrium
in a Monetary Economy.” In The Theory of Interest Rates, edited by F. H. Hahn and
135. London: Macmillan.
F. P. R. Brechling, 125–
Hicks, John R. 1937. “Mr. Keynes and the ‘Classics’; A Suggested Interpretation.”
159.
Econometrica 5 (2): 147–
Holmström, Bengt, and Jean Tirole. 2011. Inside and Outside Liquidity. Cambridge,
MA: MIT Press.
Ilzetzki, Ethan, Enrique G. Mendoza, and Carlos A. Végh. 2013. “How Big (Small?)
254.
Are Fiscal Multipliers?” Journal of Monetary Economics 60 (2): 239–
International Monetary Fund. 2010. World Economic Outlook (WEO): Recovery, Risk
and Rebalancing. Washington, DC.
International Monetary Fund. 2016. World Economic Outlook (WEO): Too Slow for Too
Long. Washington, DC.
Keynes, John M. [1936] 1961. The General Theory of Employment, Interest and Money.
London: Macmillan & Co.
Kiguel, Miguel A., and Nissan Liviatan. 1994. “Exchange- Rate Based Stabilizations
in Argentina and Chile: A Fresh Look.” In Frameworks for Monetary Stability, edited
by Tomás J. T. Baliño and Carlo Cottarelli, 162–
185. Washington, DC: International
Monetary Fund.
From Chronic Inflation to Chronic Deflation 191
Kiyotaki, Nobuhiro, and John Moore. 1997. “Credit Cycles.” Journal of Political
248.
Economy 105 (2): 211–
Koo, Richard C. 2009. The Holy Grail of Macroeconomics: Lessons from Japan’s Great
Recession. Hoboken, NJ: John Wiley and Sons.
.
Krugman, Paul. 2011. “Mr. Keynes and the Moderns.” Voxeuorg, June 21.
Kydland, Finn, and Edward C. Prescott. 1977. “Rules Rather Than Discretion: The
493.
Inconsistency of Optimal Plans.” Journal of Political Economy 85 (3): 473–
Lara Resende, André. 2016. “A Teoria da Política Monetária: Reflexōes sobre um
caminho sinuoso e inconclusivo.” In A Crise Fiscal e Monetária Brasileira: Ensaios em
homenagem a Fábio O. Barbosa, edited by E. Bacha, 483– 506. Rio de Janeiro: Editora
Civilização Brasileira. English version online: http://iepecdg.com.br/wp-content
/uploads/2016/03/The-Theory-of-Monetary-Policy5.pdf.
Lewis, Karen K. 2008. “Peso Problem.” In The New Palgrave Dictionary of Econom-
ics, second edition, edited by Steven N. Durlauf and Lawrence E. Blume. New York:
Palgrave-Macmillan.
Little, I. M. D., Richard N. Cooper, W. Max Corden, and Sarath Rajapatirana. 1993.
Boom, Crisis, and Adjustment: The Macroeconomic Experience of Developing Countries.
New York: Oxford University Press.
Lucas, Robert E., Jr. 1972. “Expectations and the Neutrality of Money.” Journal of
Economic Theory 4 (2): 103–124.
Lucas, Robert E., Jr. 1976. “Econometric Policy Evaluation: A Critique.” Carnegie-
46.
Rochester Conference Series on Public Policy 1: 19–
McKinnon, Ronald I. 2013. The Unloved Dollar Standard: From Bretton Woods to the
Rise of China. Oxford: Oxford University Press.
Ferry. 2012. “Sudden Stops in the Euro Area.” Bruegel
Merler, Silvia, and Jean Pisani-
bruegel
Policy Contribution 2012/6, Bruegel, Brussels; http:// org.
.
Minsky, Hyman P. 2008. Stabilizing an Unstable Economy. ebook, McGraw Hill.
Obstfeld, Maurice, and Kenneth Rogoff. 1983. “Speculative Hyperinflations in Maxi-
687.
mizing Models: Can We Rule Them Out?” Journal of Political Economy 91 (4): 675–
Obstfeld, Maurice, and Kenneth Rogoff. 1986. “Ruling Out Divergent Speculative
362.
Bubbles.” Journal of Monetary Economics 17 (3): 349–
Ohanian, Lee E. 2016, “The Great Recession in the Shadow of the Great Depression:
A Review Essay on ‘Hall of Mirrors: The Great Depression, The Great Recession and
the Uses and Misuses of History’.” NBER Working Paper 22239, National Bureau of
Economic Research, Cambridge, MA.
192 Guillermo Calvo
Papageorgiou, Demetris, Michael Michaely, and Armeane M. Choksi. 1991. Liberal-
izing Foreign Trade. Oxford: Basil Blackwell.
Patinkin, Don. 1965. Money, Interest, and Prices. New York: Harper and Row.
Phelps, Edmund S. 1972. Inflation Policy and Unemployment Theory. New York: W. W.
Norton.
Queraltó, Albert. 2013. “A Model of Slow Recoveries from Financial Crises.” FRB Inter-
national Finance Discussion Paper 1097, Federal Reserve Board, Washington, DC.
Reinhart, Carmen M., and Vincent R. Reinhart. 2010. “After the Fall.” NBER Work-
ing Paper 16334, National Bureau of Economic Research, Cambridge, MA.
Rodríguez, Carlos Alfredo. 1982. “The Argentine Stabilization Plan of December
20th.” World Development 10 (9): 801–811.
Sargent, Thomas J. 1982. “The Ends of Four Big Inflations.” In Inflation: Causes and
98. Chicago: University of Chicago Press.
Effects, edited by Robert E. Hall, 41–
Sargent, Thomas J. 1983. “Stopping Moderate Inflations: The Methods of Poincaré
and Thatcher.” In Inflation, Debt, and Indexation, edited by Rudiger Dornbusch and
Mario Henrique Simonsen, 54– 98. Cambridge, MA: MIT Press.
Sargent, Thomas J., and Neil Wallace. 1973. “The Stability of Models of Money and
1048.
Perfect Foresight.” Econometrica 41 (6): 1043–
Sargent, Thomas J., and Neil Wallace. 1981. “Some Unpleasant Monetarist Arithme-
17, Federal Reserve Bank of Minneapolis.
tic.” Quarterly Review 5 (3): 1–
Shiller, Robert J. 1998. “Indexed Units of Account: Theory and Assessment of Histor-
ical Experience.” NBER Working Paper 6356, National Bureau of Economic Research,
Cambridge, MA.
Taylor, Alan M. 2015. “Credit, Financial Stability, and the Macroeconomy.” Annual
339.
Review of Economics 7: 309–
Woodford, Michael. 2003. Interest and Prices: Foundations of a Theory of Monetary
Policy. Princeton, NJ: Princeton University Press.
Comment: Gita Gopinath
Without a doubt, the 2008−2009 global financial crisis and its lingering
effects have changed the lens through which macroeconomists view the
world. Previously, developed markets had been characterized as having
frictionless and benign financial markets, but the 2008−2009 financial
crisis that originated in the developed world dispelled all notions of this.
As a consequence, it is now nearly impossible to discuss macroeconomics
without explicitly describing the interactions of economic agents and the
imperfect world of finance.
However, it is important to note that for international macroeconomists,
of which Guillermo Calvo is most prominent, the failures of financial mar-
kets has been at the center of understanding the economies of emerging
markets, economies that have routinely been buffeted by financial and debt
crises. This gives Guillermo, who is one of the leading experts on emerg-
ing market crisis, an edge over other macroeconomists in analyzing this
crisis and pointing out to us lessons for the future of macroeconomics. This
is why his contribution to this volume is so valuable, and I thoroughly
enjoyed reading his chapter.
Guillermo makes several important points, of which I will highlight a
few, but I encourage the reader to delve into the many other contributions
in the chapter. Guillermo flags two major blind spots that policy makers
have ignored at the peril of their economies. The first is the power of expec-
fulfilling crises when policy makers suffer the original
tations to drive self-
called expectations dominance. The second
sin of not being able to commit so-
lasting effects on
is that liquidity scarcity can arise rapidly and have long-
the economy, and conventional monetary policy can fail to rescue the
economy. As Guillermo goes on to describe, two important policy recom-
mendations that follow from these observations. The first is the need to
194 Comments by Gita Gopinath and Luis Servén
ensure sufficient supply of safe and liquid assets. The second is that the
world benefits from a global coordination of policies, so that expectations
are coordinated on the good equilibria. I will reinforce both these points.
As Guillermo highlights, the inability of even social welfare maximizing
central bankers to commit to policy was a major factor in the hyperinfla-
tions of the past. This is tied to the time inconsistency problem, where
monetary authorities would like to commit to not inflating ex ante but
then ex post have every incentive to general surprise inflation so as to stim-
ulate the economy, increase seignorage, and lower the real value of nominal
looking private agents of course expect this behavior and
debt. Forward-
raise prices in anticipation, thereby raising equilibrium inflation.
Similarly, expectations can generate temporary booms that eventu-
ally go bad, and governments can misinterpret the cause of the boom. As
an example, Guillermo points to the consumption booms that followed
rate-
exchange- based stabilizations in emerging markets. He argues that it is
expectations of the failure of the stabilization reform measures that gener-
load purchases of goods
ate a temporary consumption boom as agents front-
in anticipation of a return to high inflation in the future.
fulfilling crises was
The importance of expectations dominance and self-
evident in the 2012 debt crisis in the Eurozone. As yields on government
debt rose rapidly in Greece and spilled over to Ireland, Portugal, Spain, and
Italy, the European Central Bank’s (ECB) president Mario Draghi promised
to do whatever it takes to save the euro, including possibly buying stressed
government debt. The mere promise of this brought yields down rapidly,
even in the absence of any purchases by the ECB (figure 4.1). This event
not only highlights the role of expectations dominance in generating cri-
sis, it also importantly points out the errors of the framers of the common
currency area who restricted the ECB from being the lender of last resort.
fulfilling crisis in monetary unions and the
Aguiar et al. (2015) describe self-
contingent manner
important role of central banks to intervene in a state-
to alleviate such crises.
Aguiar et al. (2013) also describe how the ability to inflate does not nec-
fulfilling crises. In the midst of the
essarily reduce the potential for self-
Greek crisis, it was argued by several leading economists that the problem
arose because Greek debt was real, as the Greek’s did not control the supply
of the currency in which the debt was denominated, and required fiscal
surpluses to pay it down. In contrast, if the debt had been denominated
From Chronic Inflation to Chronic Deflation 195
10%
Draghi Put
9%
8% Lehman
Collapse
7% Greek
Crisis
6%
5%
4%
3%
Spain
2% France
Italy
1%
Germany
0%
38…
38…
38…
38…
38…
38…
38…
39…
39…
39…
39…
39…
39…
39…
39…
39…
39…
40…
40…
40…
40…
40…
40…
40…
40…
40…
41…
41…
41…
41…
Figure 4.1
year yields
Sovereign 10-
in a currency over which the country has direct control, as in the case of
U.S. and Japanese debt, then governments also have the option of inflating
some of the debt so as to make it easier to repay. This argument is flawed to
the extent that it ignores the role of expectations. When debt is in nominal
terms and lenders expect the use of inflation to reduce the real value of
debt, then this expectation gets priced into nominal interest rates. Conse-
quently, there is no additional gain from being able to control the currency
in which the debt is denominated.
A second theme in Guillermo’s chapter is about liquidity and its fragil-
ity. Clearly it can be tricky to describe what a liquid asset is, something
Guillermo gets into at some length. But the point that sudden collapse can
occur in liquidity relative to the demand for it, which in turn can have
important negative and lasting consequences for the economy, is a point
that has been emphasized recently by many economists. In the “safe assets”
literature, Caballero and Farhi (2014) point to the collapse in safe assets
following the 2008−2009 financial crisis as important for understanding
the decline in real interest rates, the drop in output, and the increase in
risk premia in equity markets. The excess demand for safe assets also calls
196 Comments by Gita Gopinath and Luis Servén
for unconventional monetary interventions, such as the purchase by the
central bank of toxic assets as opposed to the more conventional purchase
of safe treasuries.
Whether there continues to be a liquidity crisis is up for debate. How-
ever, there is little doubt that the world needs to be prepared for future
financial crises that may arise from China. With China’s debt exceeding
240 percent of its GDP and the ongoing credit boom there is sufficient
cause for concern.
A theme in the chapter that I have spent little time discussing is the price
theory of money. Guillermo argues that the reason people hold certain cur-
rencies is because prices denominated in that currency tend to be preset and
staggered and therefore deliver predictable output. This is an appealing argu-
ment, but of course one could ask what comes first. The reason prices are
sticky in a currency is plausibly because of the faith in the monetary authori-
ties that manage the currency to keep inflation low.
I look forward to reading future work by Guillermo on this and related
issues so as to gain important insights into the functioning of the world
economy, something Guillermo has delivered in spades over many years.
References
Aguiar, Mark, Manual Amador, Emmanuel Farhi, and Gita Gopinath. 2013. “Crisis
and Commitment: Inflation Credibility and the Vulnerability to Sovereign Debt
Crises.” NBER Working Paper 19516, National Bureau of Economic Research, Cam-
bridge, MA.
Aguiar, Mark, Manuel Amador, Emmanuel Farhi, and Gita Gopinath. 2015. “Coor-
dination and Crisis in Monetary Unions.” Quarterly Journal of Economics 130 (4):
1727–1779.
Caballero, Ricardo, and Emmanuel Farhi. 2014. “The Safety Trap.” NBER Working
Paper 19927, National Bureau of Economic Research, Cambridge, MA.
Comment: Luis Servén
2009 global financial crisis and the ensuing Great Recession have
The 2008–
prompted a critical reassessment of mainstream macroeconomic models.
Among their key weaknesses, many observers have singled out the virtual
neglect of the financial system. Indeed, the description of the financial side
crisis mainstream macro model was pretty much limited to a
in the pre-
defined concept of “money.”
demand function, presumed stable, for a well-
Financial frictions and amplification mechanisms, two ingredients widely
seen at core of the financial crisis and its propagation, were altogether absent.
These themes at the nexus of macroeconomics and finance have long
attracted Guillermo Calvo’s attention. His chapter brings together a broad
array of big macrofinancial issues that reflects the wide range of his contri-
butions to macroeconomic thinking, and it showcases his mastery at draw-
ing insights from highly stylized analytical settings. The common threads
that tie the chapter together are expectations and, especially, liquidity and
its role in past and recent crises. This is the focus of my comments below.
Needless to say, it has been a longstanding concern for Guillermo Calvo, as
proven, for example, by his seminal work on sudden stops.
The chapter argues that liquidity should take center stage in macro-
crisis
economics and places it at the root of the global crisis and the post-
slump. The central role of liquidity reflects two key facts. The first is that
liquid assets are essential to the operation of modern economies. They facil-
itate market transactions, can (almost) always be transformed into means
of exchange at full face value, and exchanged for goods and services or
other assets. In particular, their widespread use as collateral in financial
functioning credit market. But
transactions makes them essential to a well-
especially
the second key fact is that liquidity is also fragile: Liquid assets—
are
those privately produced, with bank deposits as the classic example—
198 Comments by Gita Gopinath and Luis Servén
fulfilling runs. This puts the spotlight on the role of
vulnerable to self-
expectations and coordination mechanisms in triggering sudden shifts in
the valuation of liquid assets.
liquid assets as the key to the credit mechanism in the
These two issues—
context of financial frictions and the vulnerability of those assets to shifts
have been explored by an ample literature, which could
in expectations—
have featured more prominently in the chapter. Recent examples that come
to mind are those of Martin and Ventura (2012), who show how bubbles
can unlock credit and growth, and Gorton and Ordoñez (2013), who ana-
lyze the endogenous nature of financial fragility.
The Backing of Money
As the chapter reminds us, fragility is a fundamental feature of fiat money,
the ultimate liquid asset. Fiat money is an intrinsically worthless asset, valu-
able only to the extent that it is (or is expected to be) valued by others. Thus,
it fits the standard definition of a bubble. This in turn opens the door to
fulfilling equilibria in monetary economies—
the existence of multiple self-
including barter equilibria, in which the price of money is zero.
The question of why money is valuable attracted considerable attention
from monetary theory during the 1960s and 1970s. Yet flight away from
money has been a rare occurrence in modern times. It has not been a fea-
ture of recent crises; indeed, if anything, the opposite has been the case.
The chapter sets out a “price theory of money” (or PTM for short) to explain
this resilience of money: What anchors the value of money is nominal price
stickiness. In a world of staggered price setting, the positive value of money
is just a result of hysteresis: Money is valuable today, because it was valuable
yesterday. Because only a limited number of individual prices may have
changed in the interim, the general price level (the inverse of the value
of money) cannot have moved much. By the same reasoning, if money is
valuable today, it can be anticipated to remain valuable tomorrow. Thus,
staggered price setting provides an output backing for money.
This approach casts nominal rigidities in an unusual light. In the macro-
economic literature they routinely get the blame for hampering adjustment
to shocks, but the PTM holds instead that they also deserve credit for pre-
hyped price flexibility is
serving monetary stability. Put differently, much-
not an unmixed blessing after all, as it may come with increased monetary
fragility.
From Chronic Inflation to Chronic Deflation 199
But there is some circularity underlying the PTM. Pricing arrangements
are themselves not invariant to perceptions about monetary and aggre-
gate price stability. For example, if (for whatever reason) the price level
is expected to rise quickly, more agents are likely to revise upward their
individual prices, and by larger amounts, than if they expect the overall
price level to rise slowly (Burstein 2006). Thus, as a result of the combined
actions of individual agents, the degree of price level stickiness, and hence
its contribution to the backing of money, in effect depends on expectations.
Ultimately, this suggests that the power of the PTM to explain the backing
that is, the degree of monetary and price stability—
of money— may itself
depend on the perceived degree of monetary and price stability. In other
words, the PTM may not take us too far in resolving the indeterminacy sur-
rounding the value of money.
Liquidity and Fragility
In modern economies, other assets beyond fiat money provide liquidity
services. Much of the recent literature (e.g., Gorton and Ordoñez 2013;
Caballero and Fahri 2018) refers to them as “safe assets.” They include pub-
lic debt backed by the government’s taxation capacity, as well as privately
of-
produced debt backed by either a lender- resort guarantee (as in the
last-
case of insured bank deposits) or by credible collateral (as in the case of
asset-backed securities).
What distinguishes safe assets from the rest is the fact that they can
(almost) always be exchanged at full face value. They retain (much of) their
value in large systemic events. Also, their value is information insensitive—
there is no benefit to producing private information about it. In other
words, they are free from adverse selection, that is, concerns that the coun-
terparty may have superior private information about their value.
label assets help meet the overall demand for liquidity, but their
Private-
use also raises financial fragility. They can be close, but not perfect, sub-
stitutes for safe public debt. Their value is impaired in systemic events. In
term private-
particular, unless fully backed by a lender of last resort, short-
label safe assets are vulnerable to runs, as shown in the global financial
crisis (Brunnermeier 2009; Gorton 2010).
All these issues are touched on, to varying extents, in Guillermo Calvo’s
chapter. But they have important implications for public debt, a missing
label liquidity implies that safe public debt has
theme. The fragility of private-
200 Comments by Gita Gopinath and Luis Servén
a key role to play in protecting the credit mechanism. More specifically, pub-
lic debt is net wealth, to the extent that it allows sustaining credit at times
when privately produced assets cease to be accepted as collateral
of crisis—
(Gorton and Ordoñez 2013). Even if the choice between taxes and debt to
finance government expenditure may be inconsequential in normal times,
Ricardian equivalence still breaks down when financial crises can occur.
Failure to recognize this may result in an undersupply of safe public debt.
Another important policy question is the ability of financial regulation
to mitigate the fragility of privately produced liquid assets. This subject
has focused the attention of financial regulators worldwide after the cri-
sis, although it receives limited attention in the chapter. Yet, as Guillermo
crisis has gone
Calvo notes, the tightening of regulatory requirements post-
in the direction of raising the mandated liquidity holdings of financial
institutions, which will likely have the unintended consequence of increas-
ing the aggregate shortage of safe assets.
Expectations and Fragility
Investor runs are often attributed to “shifts in sentiment.” But the causes
of those shifts remain poorly understood. This echoes the fact that theo-
retical work on models with multiple equilibria typically has little to say
for example, what causes transition
on what prompts jumps across them—
from a bubbly to a bubbleless equilibrium in a model of asset bubbles. In
practice, the factors responsible are often difficult to determine even in ex
post forensic analysis of financial crashes. The Minsky moment that marks
their onset does not usually follow large shocks to fundamentals or major
news about their future path. Instead, it tends to occur after the arrival of
relatively minor, sometimes almost irrelevant, news.
The subprime crash is a case in point. The sharp increase in the default
rate of subprime mortgages in the United States is commonly viewed as the
trigger of the global crisis. But it is hard to see how the souring of a fairly
minor segment of the US mortgage market could have reversed expectations
about the future prices of broad categories of assets so dramatically as to trig-
ger runs on a wide variety of leveraged investors across the financial system.
What makes for this disproportionate effect of seemingly innocuous
news? The literature on amplification mechanisms in financial crises (e.g.,
Brunnermeier and Oehmke 2013) offers some hints. One example can be
From Chronic Inflation to Chronic Deflation 201
found in Guillermo’s own work on the interplay between informed and
uninformed investors (Calvo and Mendoza 2000). The latter investors infer
the state of fundamentals from the actions of the former. In appropriate
conditions, the uninformed investors may stage a run just because informed
investors are redeeming assets to meet their liquidity needs, which uni-
formed investors misinterpret as a sudden worsening of fundamentals.
A related mechanism arises when rational investors hold heterogeneous
expectations due to the presence of private information about the funda-
mentals. Asset prices then reflect average market expectations, and rational
investors have to face Keynes’s “beauty contest” (i.e., they need to form
expectations about the expectations of others). In such settings, noisy pub-
lic signals about the fundamentals drive a wedge between asset prices and
fundamental values (Bacchetta and van Wincoop 2008). In particular, asset
prices may overreact to public signals (Allen, Morris, and Shin 2006) and
experience abrupt shifts in response to nearly irrelevant news.
From this it would seem tempting to conclude that steps aimed at improv-
such as enhanced dis-
ing the reliability and accuracy of public information—
might help reduce asset price volatility
closure rules for leveraged investors—
and stem investor panics. It is doubtful, however, that such measures would
make much of a material contribution to anchor investor expectations and
deter runs. Calvo’s chapter points in a different direction. For example, he
suggests more use of pegs to limit the indeterminacies surrounding flexible
exchange rates or of backward indexation to anchor inflationary expecta-
which are
tions. How, if at all, this could translate to the case of asset prices—
is not discussed, but it seems like a natural
fundamentally forward looking—
up question. For example, should policy make more systematic use of
follow-
floors (or ceilings) to the levels, or the changes, of asset prices?
The Post-Crisis
Almost 10 years after the global crisis, world economic growth remains
sluggish, and advanced economies continue to exhibit deflationary pres-
sures. This disappointing performance has attracted a wide variety of
explanations (see Teulings and Baldwin 2014). They range from those that
crisis as a new normal, driven by slow-
portray the post- moving supply or
demand factors (i.e., the “secular stagnation” view) to others that take more
term perspective and attach a central role to Keynesian aggregate
of a short-
202 Comments by Gita Gopinath and Luis Servén
crisis sluggishness well in
demand deficiencies. Yet others find the post-
accordance with the past history of major financial crashes, which are typi-
cally followed by protracted recessions.
centered view: The crisis was driven by the
The chapter takes a liquidity-
collapse of liquid assets, which brought the financial system to the verge of
collapse. As credit supply dried up, output and employment fell across the
crisis world reflects the continuing liquidity
globe. Low growth in the post-
shortage and malfunction of the credit market.
Few dispute the key role of the liquidity crunch in the onset of the cri-
sis, but there is much less agreement on whether the shortage of credit
remains the main cause of the subsequent sluggish growth. Casual obser-
vation suggests that many firms in the United States and Japan are awash
with liquidity, yet investment has been slow to recover. Empirical tests by
Mian and Sufi (2014) indicate that the credit crunch cannot explain the US
employment collapse. On the whole, the seeming implication is that aggre-
gate demand shortages, actual or anticipated, might also be a major factor
behind the weak growth recovery.
Most observers believe that the powers of monetary policy to reignite
crisis as the economy fell into a
growth have been weakened in the post-
liquidity trap posed by the zero lower bound on interest rates. Although the
chapter shares this perspective, its distinguishing feature is the view that
side liquidity trap—
what is at work is a supply- as distinct from the Keynes-
side liquidity trap. The latter arises from an insatiable demand
ian demand-
for liquidity; the former, according to Calvo, from the inability of monetary
policy to raise the supply of liquidity services.
In this narrative, expansionary monetary policy may be able to raise
real money balances but fail to raise liquidity or even reduce it; such policy
may prompt deflation rather than inflation, as individuals vie for yet more
liquidity. The mechanism responsible for this intriguing result is not fleshed
out, but it appears to rely on agents’ competition for liquidity services in a
setting with pecuniary externalities and anticipated runs on liquid assets.
In a variation on the same idea, the central bank might be able to raise
liquidity, and thereby output, only as long as it keeps expanding the money
supply indefinitely.
side
Strictly speaking, it is not clear if this really qualifies as a supply-
liquidity trap, because the underlying mechanism seems to rely on the
behavior of liquidity users on the demand side. And, on the whole, it seems
From Chronic Inflation to Chronic Deflation 203
doubtful that central banks’ attempts to implement expansionary policies
really belong among the chief factors behind the deflationary pressures in
advanced countries.
Leaving aside these issues, however, Calvo’s perspective on the post-
crisis has a lot in common with the recently proposed “safety trap” view
(e.g., Caballero and Fahri 2018). In that narrative, the market for safe
term increase in demand, largely driven by the
assets witnessed a long-
growing liquidity needs of financial intermediaries, as well as the self-
country governments around the world in
insurance needs of emerging-
the face of global external disturbances (Gourinchas and Jeanne 2012).
The growth of demand far outstripped the available supply of safe public
label (quasi-
debt and led to a boom in the supply of private- )safe assets,
through securitization and similar mechanisms. Indeed, the US evidence
label liquid assets is negatively cor-
confirms that the net supply of private-
related with the supply of government debt (Krishnamurty and Vissing-
Jorgensen 2012).
These assets unraveled in the crisis and brought down with them large
volumes of formerly safe sovereign debt, notably that of European periph-
ery countries struggling to rescue their financial systems. By some esti-
mates, the supply of safe assets relative to global GDP fell by half, opening
à-
up a massive gap vis- vis their demand and pushing down into negative
territory their “natural” rate of return (i.e., that consistent with full employ-
ment; Caballero and Fahri 2018). With the actual rate constrained by the
zero lower bound, the economy fell into a safety trap, and equilibrium in
the safe asset market was restored through an output fall.
This story seems to have a lot in common with that outlined in the
chapter. The safety trap is akin to a liquidity trap, with the added feature
of an endogenous risk premium that shapes the output effects of macro-
in
economic policy. And some policy implications seem broadly similar—
particular, the scope for conventional monetary policy is limited in both
side liquidity trap” perspective
narratives. In truth, however, the “supply-
in the chapter is not developed in sufficient detail to allow the reader to
see how, or why, appropriate policy actions to revive liquidity would differ
side liquidity trap or a safety trap.
from those needed under a demand-
In a safety trap, for example, issuance of (safe) public debt, quantitative
easing through central bank purchases of risky assets, or inflation target
increases are all effective for raising output (see Caballero and Fahri 2018
204 Comments by Gita Gopinath and Luis Servén
for details). In turn, Calvo’s chapter seems skeptical regarding risky asset
purchases. Because such purchases essentially amount to changing the rela-
tive supply of safe and risky assets, one may conclude that (safe) public
in
debt issuance, which is not explicitly discussed, may be ineffective, too—
sharp contrast with the “safety trap” optic. This seems puzzling, although
strictly speaking, both risky asset purchases and public debt issuance should
be expected to be similarly unhelpful in conventional liquidity traps. In
turn, inflation target increases are not contemplated either, although one
would conjecture that they should be of help, as in standard liquidity traps.
What about the international perspective? Many central banks, espe-
cially from emerging markets, hold massive amounts of safe assets at pres-
insurance purposes. This tends to
ent, in most (but not all) cases for self-
pooling arrangements,
worsen the global asset shortage. Improved reserve-
insurance needs,
through the IMF or in other ways, might help reduce self-
as Calvo notes. But these steps may also require higher levels of mutual
trust than currently exist. A more intriguing option, recently proposed by
market reserve holdings to
Rogoff (2016), would partly reallocate emerging-
gold, which is a highly liquid asset whose rate of return is not subject to a
thus potentially helping release the safety trap. In addi-
zero lower bound—
tion, reforms to enhance emerging markets’ ability to supply safe assets,
rather than just demand them, would seem worth considering too, but they
are not discussed in the chapter.
Final Thoughts
Over the past decades, the overall demand for liquid assets has grown
steadily, largely driven by the growing liquidity demand of the global finan-
cial system. Demand has far outpaced the supply of outside liquid assets
(i.e., fiat money and safe public debt), resulting in an increasing resort to
label assets) that is seen by many observers as one
inside assets (i.e., private-
of the key ingredients behind the global crisis and its disappointing after-
math. Much of the policy debate has centered on how to engineer a com-
mensurate increase in asset supply to bridge the gap with demand.
This view prompts two concluding questions. First, because much of
the growth in demand stems from the increasing collateral needs of an
expanding financial system, we may wonder whether such expansion
increasing. In other words, is it possible for financial
really is welfare-
From Chronic Inflation to Chronic Deflation 205
intermediation, and thus its derived collateral needs, to grow “too large”
from a social welfare viewpoint?
In practice, externalities are at work that may easily lead to excessive
financial intermediation in a general equilibrium setting. Eden (2016)
offers an example, based on the fact that, although both fiat money and
monies can be used to facilitate socially efficient transactions, it is
quasi-
cheaper to use fiat money, because it is costless to produce. The private
monies are
incentives for spending resources on the production of quasi-
always greater than the social incentives, as they do not internalize the
equilibrium adjustment of the price level. A similar reasoning applies to
credit: Although it facilitates efficient transactions, its production requires
real resources in the form of monitoring services. Thus, the private incen-
tives to produce credit are likely to be excessive, because they do not inter-
nalize equilibrium price adjustments.
It is easy to think of situations in which financial intermediation grows
too large because of other externalities. A prominent example is that of
intermediation facilitating socially excessive risk taking, driven by the fact
that individual intermediaries do not take into account their contributions
a theme
to systemic risk and hence to the likelihood of adverse scenarios—
prudential literature.
explored by the macro-
Leaving aside the scale of the financial system, the second question con-
cerns the roots of its collateral needs. These ultimately arise from the pres-
ence of frictions, such as asymmetric information, monitoring costs, and
imperfect contract enforcement. The natural question is whether the pri-
mary focus of policy should be just to meet the collateral needs imposed by
these frictions, possibly at the cost of increasing financial fragility. Granted,
it is not likely that frictions can be eliminated altogether. But there prob-
ably is ample room for regulatory and other policies to substantially limit
expanding collateral needs of
their scope, and thereby contain the ever-
financial intermediation.
References
Allen, Franklin, Stephen Morris, and Hyun Song Shin. 2006. “Beauty Contests and
752.
Iterated Expectations in Asset Markets.” Review of Financial Studies 19 (3): 719–
Bacchetta, Philippe, and Eric van Wincoop. 2008. “Higher Order Expectations in
866.
Asset Pricing.” Journal of Money, Credit and Banking 40 (5): 837–
206 Comments by Gita Gopinath and Luis Servén
Brunnermeier, Markus K. 2009. “Deciphering the 2007– 2008 Liquidity and Credit
100.
Crunch.” Journal of Economic Perspectives 23 (1): 77–
Brunnermeier, Markus K., and M. Oehmke. 2013. “Bubbles, Financial Crises, and
Systemic Risk.” In Handbook of the Economics of Finance, volume 2B, edited by George
M. Constantinides, Milton Harris, and Rene M. Stulz, 1221– 1288. Boston: Elsevier.
Dependent Pric-
Burstein, Ariel T. 2006. “Inflation and Output Dynamics with State-
1257.
ing Decisions.” Journal of Monetary Economics 53 (7): 1235–
Caballero, R., and E. Fahri. 2018. “The Safety Trap.” Review of Economic Studies 85 (1):
223–274.
Calvo, Guillermo A., and Enrique Mendoza. 2000. “Rational Contagion and the Glo-
113.
balization of Securities Markets.” Journal of International Economics 51 (1): 79–
Eden, Maya. 2016. “Excessive Financing Costs in a Representative Agent Frame-
237.
work.” American Economic Journal: Macroeconomics 8 (2): 215–
Gorton, Gary. 2010. Slapped by the Invisible Hand: The Panic of 2007. New York:
Oxford University Press.
Gorton, Gary, and Guillermo Ordoñez. 2013. “The Supply and Demand for Safe
Assets.” NBER Working Paper 18732, National Bureau of Economic Research, Cam-
bridge, MA.
Olivier, and Olivier Jeanne. 2012. “Global Safe Assets.” BIS Work-
Gourinchas, Pierre-
ing Papers 399, Bank of International Settlements, Basel.
Krishnamurty, Arvind, and Annette Vissing- Jorgensen. 2012. “The Aggregate
267.
Demand for Treasury Debt.” Journal of Political Economy 120 (2): 233–
Martin, Alberto, and Jaume Ventura. 2012. “Economic Growth with Bubbles.”
3058.
American Economic Review 102 (6): 3033–
2009 Drop in Employ-
Mian, Atif, and Amir Sufi. 2014. “What Explains the 2007–
2223.
ment?” Econometrica 82 (6): 2197–
Rogoff, Kenneth. 2016. “Emerging Markets Should Go for the Gold.” Project Syn-
dicate, May 3. https://www.project-syndicate.org/commentary/gold-as-emerging
-market-reserve-asset-by-kenneth-rogoff-2016-05.
Teulings, Coen, and Richard Baldwin. 2014. Secular Stagnation: Facts, Causes and
Cures. London: CEPR Press.
5 Global Liquidity and Procyclicality
Hyun Song Shin
It is an honor to join this distinguished group and to take part in this event.
I feel especially privileged to have Maurice Obstfeld and Aslı Demirgüç-
Kunt as my discussants. I have learned a lot from Aslı and Maury over the
years and no doubt will learn much from their comments today.
Exchange rates are back in the news. It is a cliché that the world has
become more connected, but the external dimension of monetary policy
has figured more and more prominently in central bankers’ speeches lately.
Financial markets, for their part, appear to be tethered more closely than
ever to global events, and the real economy appears to dance to the tune of
global financial developments rather than the other way round. If you will
excuse a rather extravagant metaphor, the financial tail appears to be wag-
ging the real economy dog. This is not how things are supposed to work.
According to the traditional approach to international finance, financial
flows are no more than the accounting counterparts to savings and invest-
ment decisions. The current account is the borrowing need of the country
as a whole, and exchange rates steer net exports to restore external balance.
When a country experiences an appreciation of its currency, this is presumed
to be contractionary, as net exports fall.
I am grateful to Raphael Auer, Fernando Avalos, Stefan Avdjiev, Morten Bech, Clau-
dio Borio, Michael Chui, Ben Cohen, Dietrich Domanski, Peter Hoerdahl, Krista
Hughes, Jonathan Kearns, Catherine Koch, Bob McCauley, Pat McGuire, Andreas
Schrimpf, Ilhyock Shim, Vlad Sushko, and Philip Turner for comments on earlier
el Berger, Anamaria Illes, Emese Kuruc, Denis Petre, Jeff Slee, and
drafts and to Bat-
Agne Subelyte for excellent research assistance. The views expressed here are my own
and not necessarily those of the Bank for International Settlements.
208 Hyun Song Shin
However, events have not always played out this way, especially in
emerging economies. Rather than dampening economic activity, episodes
of sustained currency appreciation often go hand in hand with buoyant
economic activity on the back of strong capital inflows. The boom may be
accompanied by the buildup of financial vulnerabilities. Think back to the
years before the latest bout of financial turbulence in emerging markets. My
known empirical paper with Pierre-
discussant Maurice Obstfeld has a well-
Olivier Gourinchas (Gourinchas and Obstfeld 2012) that sheds much light
on this phenomenon. The combination of a rapid increase in leverage and
a sharp appreciation of the currency emerges as a strong indicator of finan-
cial vulnerability and of subsequent crises.
There is also a flip side to the argument based on the current account.
If a country is running current account surpluses, the argument goes, then
its currency will tend to appreciate unless the authorities are keeping the
currency artificially low. This is the familiar argument heard around the
G20 table, directed at economies running current account surpluses. By the
same token, the currency of a deficit country should depreciate. However,
2000s, the US cur-
again, events do not always play out this way. In the mid-
rent account deficit widened to historical highs, and many commentators
expected an imminent depreciation of the dollar. In the event, the dollar
went in the opposite direction. It appreciated strongly with the onset of the
crisis, wrong footing many commentators. The appreciation of the dollar
was accompanied by a tightening of global financial conditions.
The wheel has turned full circle, and financial markets are once again
keeping a wary eye on a stronger dollar. Observers are keenly attuned to
every twist and turn in the monetary policy debate in the United States.
Markets rally and the dollar weakens on any temporary reprieve from the
normalization of US interest rates, only to reverse course when monetary
tightening is back on the agenda.
Why are global financial conditions so attuned to the strength of the
dollar? And why is the real economy so sensitive to global financial condi-
tions? These are the two questions addressed in this chapter.
The chapter starts by describing a market anomaly in the currency mar-
ket that is symptomatic of the strains currently being placed on global capi-
tal markets. In spite of the outward tranquillity, tensions lurk beneath the
surface. Market anomalies offer a window on these strains.
Global Liquidity and Procyclicality 209
A Telling Market Anomaly
There is an intriguing market anomaly in the foreign exchange market right
now: the widespread failure of covered interest parity. Covered interest par-
ity (CIP) is the proposition that interest rates implicit in foreign exchange
markets should be consistent with market interest rates.1
Before 2008, CIP held as an empirical regularity with very few exceptions
worth mentioning. As an academic, I used to tell my students that CIP is
about the only relationship that can be relied on in international finance.
I know better than to say this now. Textbooks still say that CIP holds, but
it is no longer true.
Figure 5.1 shows the evidence. A foreign exchange swap (FX swap) is an
arrangement where one party borrows US dollars by pledging another cur-
that is, lending the other currency in exchange for dol-
rency as collateral—
lars. The forward rate is the agreed exchange rate at which repayment takes
place. From the forward rate and the current spot rate, we can calculate the
implied interest rate on the US dollar. The top panels of figure 5.1 plot the
month interest rate on the dollar from forward rates embedded
implied 3-
in FX swaps. Each series shows the particular currency pledged as collateral.
month US dollar LIBOR, the mar-
Figure 5.1 plots the comparison of the 3-
ket interest rate for dollars. When the implied dollar interest rate from FX
swaps is above LIBOR, then the borrower of dollars in the FX swap is paying
more than the rate available in the open market. This has been the case for
the yen, Swiss franc, and euro.
CIP held with barely a blip until the crisis (Akram, Rime, and Sarno
2008). Large deviations from CIP did take place during the 2008−2009
2012. However, these
global financial crisis and the euro area crisis of 2011–
were periods when financial intermediaries came under severe stress (Baba
and Packer 2009; Baba and Shim 2010; Avalos and Moreno 2013). What is
remarkable now is that deviations from CIP have appeared during periods
of relative calm. Recent deviations have been especially large for the yen,
F
1. Formally, CIP is the statement that 1 + rA = (1 + rB ), where rA and rB are the mar-
S
ket interest rates on two currencies A and B, and S and F are the spot and forward
exchange rates, respectively, of A in terms of B.
210 Hyun Song Shin
A B
C D
Figure 5.1
US dollar interest rate implied by FX swaps
month US dollar interest rate implied by FX swaps1
A., B. Three-
C., D. FX swap spread, 3-month2
1. Implied US dollar interest rate in an FX swap involving the indicated currency.
month US dollar LIBOR is plotted for comparison.
The 3-
2. Spread between the 3- month dollar rate implied
month US dollar LIBOR and 3-
by FX swaps.
Source: Bloomberg; Datastream; BIS calculations.
although the Swiss franc also had a large deviation following the surprise
revaluation of the Swiss franc in January 2015.
The bottom two panels of figure 5.1 show the magnitude of the devia-
tion from CIP, where the deviation is measured as US dollar LIBOR minus
implied dollar interest rate. The difference is called the “cross-
the FX swap-
currency
currency basis,” and for the currencies listed in figure 5.1, the cross-
basis is negative, meaning that dollar borrowers in FX swaps pay more than
LIBOR.
Global Liquidity and Procyclicality 211
shocked even—
Traditionalists will be surprised— to discover that CIP
fails. But there it is, in the full glare of daylight. Not only does CIP fail
systematically, the observed deviations from CIP have become more pro-
nounced in the past 18 months or so.2 In textbook settings where someone
could borrow and lend without limit at prevailing market interest rates,
currency basis could not deviate from zero, at least not by much
the cross-
and not for too long. This is because someone could borrow at the cheaper
dollar interest rate and lend out at the higher dollar interest rate. However,
executing such a trade entails a sequence of transactions, often through
taking capacity of
intermediaries. Thus, it makes demands on the risk-
dealer banks as well as on counterparties.3
What is the link between CIP violations and the dollar? One can draw
a parallel with recent strains in emerging markets. At first sight, advanced
economy currency markets seem a million miles away from stresses in
emerging markets, but the common element is that a stronger dollar and
tighter credit conditions go together.
Figure 5.2 plots the value of the US dollar (in light gray), calculated as the
simple average of the exchange rates against six advanced economy curren-
cies as indicated. When the light gray line goes up, the dollar strengthens. On
currency basis. Notice
the same chart, plotted in dark gray, the average cross-
currency basis is the mirror image of the strength of the dollar.
how the cross-
currency basis widens. This is espe-
When the dollar strengthens, the cross-
cially so in the past 18 months or so, reflecting the stronger dollar.
The relationship is even clearer if we plot changes in exchange rates and
currency basis. Figure 5.3 shows this for the bilateral
changes in the cross-
exchange rate of the euro against the dollar. See the reflected symmetry in
the left panel, just like mountains reflected in a lake, where a strengthening
of the dollar is associated with a widening of the deviation from CIP. The
2. The recent evidence is examined by Du, Tepper, and Verdelhan (2016), who find
currency basis is not confined to LIBOR and appears across many mar-
that the cross-
currency
ket interest rates. Borio et al. (2016) show that that the sign of the cross-
basis depends on the net swap position of the banking sector.
3. Gabaix and Maggiori (2015) propose a theory of exchange rate determination
taking
based on intermediary balance sheet constraints. More generally, a bank’s risk-
capacity is limited by its capital, as described in two of my recent speeches (Shin
2016a, 2016b).
212 Hyun Song Shin
Figure 5.2
US dollar exchange rate and the cross- currency basis
1
Simple average of bilateral exchange rate of the dollar against CAD, EUR, GBP, SEK,
CHF, and JPY. Higher values indicate a stronger US dollar.
2
Simple average of the five- currency basis swaps against CAD, EUR, GBP,
year cross-
SEK, CHF and JPY vis- vis the US dollar.
à-
Source: Avdjiev et al. (2016); Bloomberg; BIS calculations.
right panel shows the same information as a scatter chart. The negative
slope is clear to see; a strengthening of the dollar goes hand in hand with a
widening of the deviation from CIP.
The key takeaway is that a stronger dollar is associated with more severe
market anomalies. The amazing thing is that this is true not only for emerg-
ing markets but also for “safe haven” currencies, such as the yen and the
Swiss franc. To understand the nature of this relationship, we need to cast
the net wider and take in the larger picture concerning the role of the dollar
in the global banking system.
The Global Banking System and the US Dollar
The global role of the US dollar is reflected in its preeminent role in the
global banking system. The dollar is the unit of account in debt contracts
in that borrowers borrow in dollars and lenders lend in dollars, irrespective
of whether the borrower or lender is located in the United States.
border bank claims denomi-
Figure 5.4 gives a sense of the size of cross-
nated in US dollars, arranged by region. The size of the arrows represents
the size of the claims. In 2002, the arrow from the United States to Europe
was $462 billion, meaning that banks resident in the United States had
Global Liquidity and Procyclicality 213
A
B
Figure 5.3
Change in euro/US dollar exchange rate and change in cross- currency basis1
1
Changes in quarterly averages.
2
An increase represents an appreciation of the US dollar against the euro.
Source: Avdjiev, Du, Koch, and Shin (2016); Bloomberg; BIS calculations.
claims of $462 billion to borrowers in Europe. This grew to $1.54 trillion
by 2007. The return leg from Europe to the United States went from $856
billion in 2002 to more than $2 trillion in 2007.4
4. McGuire and Tarashev (2007) and McCauley, McGuire, and von Peter (2010) map
border lending.
the geography of cross-
214 Hyun Song Shin
I will return to figure 5.4 when discussing the macro implications. For
now, notice that the US dollar is used widely throughout the global
banking system, even when neither the lender nor the borrower is a US
resident.
Why is the US dollar so important in the global banking system? One
border trans-
answer invokes the dollar’s broad international role in cross-
actions, including its dominant role as an invoicing currency for interna-
tional trade.5 Trade financing or associated hedging activity can account for
denominated bank credit.
some of the US dollar-
A second answer builds on the first. The dollar’s role as an invoicing
currency spills over to the currency denomination of lending that finances
real assets. For export firms, if the invoice is in dollars, it may make sense to
borrow in dollars. Figure 5.4 shows only the bank claims, but an important
denominated
funding source for emerging market firms has been dollar-
bonds. This is especially so for the oil and gas sector. Caruana (2016) and
Chui, Kuruc, and Turner (2016) provide further evidence.
The story does not end there, however. This reasoning has a third level.
The role of the dollar as the funding currency of choice means that the uni-
denominated assets extends beyond the United States. For
verse of dollar-
large institutional investors with a global portfolio of assets, there may be
a currency mismatch between the assets they hold and the commitments
they have to their domestic stakeholders. For instance, pension funds and
life insurance companies have obligations to their beneficiaries and pol-
in
icy holders. These obligations are denominated in domestic currency—
euros, yen, or Swiss francs. However, a large investor will not be limited to
domestic assets and will look abroad to form a diversified portfolio of global
assets, including securities issued in US dollars.
To the extent that investors face currency risk, they will hedge that risk.
We know that investors from emerging economies with large funded pen-
sion systems hedge actively.6 However, institutional investors from rich
economies will face the problem most acutely, as they have the largest port-
folios of global assets. The hedging counterparty is typically a bank, and the
bank lays off its own currency risk by borrowing dollars. That way, dollar
claims are counterbalanced by dollar debts.
5. Goldberg and Tille (2009) and Gopinath (2015).
6. See Avalos and Moreno (2013) for evidence from Chile.
Panel A. 2002
A
Europe
462
47
►
856
◄
20
5► Emerging
9
US ◄4
36
Europe
◄2
238
◄59
69
►
◄
◄7 87
►
LaƟn Asia-
America
◄56
Pacific
15
►
◄6
Africa &
Middle
East
Panel B. 2007
B
Europe
43
1,5
19
56
7►
2,0
◄
57
19 ► Emerging
8
US ◄7
49
Europe
◄10
393
◄82
83
►
◄
◄19 19
0►
LaƟn Asia-
◄142
America Pacific
35
►
6
◄2
Africa &
Middle
East
Figure 5.4
US dollar-denominated cross-border claims (billions of US dollars)
Source: BIS locational banking statistics.
216 Hyun Song Shin
A
B
Figure 5.5
Cross-border US dollar denominated credit, all sectors (trillion US dollars)
1
The break in the series between Q1 2012 and Q2 2012 is due to the Q2 2012
introduction of a more comprehensive reporting of cross- border positions (for more
details, see http://www.bis.org/publ/qtrpdf/r_qt1212v.htm).
Source: BIS locational banking statistics, tables A5 (by residence) and A7 (by nationality).
The upshot is that banks take on liabilities denominated in dollars in
the process of providing hedging services. This is the third level of the argu-
ment. The consequence of the dollar’s international role in transactions is
that the global banking system runs on US dollars.
denominated cross-
Figure 5.5 provides a window on the total dollar-
border bank credit arranged by region. The two panels are plotted using
Panel A. Germany1
A
1
For Germany, long-term debt securities of insurance companies. Transactions indi-
cate acquisitions minus external financing.
Panel B. Japan2
B
2
For Japan, life insurance companies. Positive (negative) transactions indicate a net
purchase (sale) of medium-and long- term bonds.
Figure 5.6
Outward bond investment of insurance companies
Source: Deutsche Bundesbank; Japanese Ministry of Finance; Statistics Sweden; Life
Insurance Association of Japan; BIS calculations.
218 Hyun Song Shin
Panel C. Sweden
C
Figure 5.6
(continued )
data that we have started posting on the BIS website as part of our effort to
make more detailed data available.7
pointing bars indicate assets, and downward-
In both panels, upward-
pointing bars indicate liabilities. The left panel breaks out the total by
residence, and the right panel breaks out the total by nationality, mean-
border
ing the location of the headquarters. So, for instance, the cross-
claims of a German bank office in London would be classified as “UK”
in the left panel, but as “euro area” in the right panel. By comparing the
two panels of figure 5.5, we see that Swiss and euro area banks have been
active in other jurisdictions, especially in the United Kingdom and the
United States.
border dollar liabilities track global
Notice how the undulations in cross-
financial conditions. The totals in figure 5.5 grew strongly up to 2008 but
contracted with the onset of the global financial crisis, and then also with
2012. Interestingly, the most recent period of
the euro area crisis of 2011–
2014 has been associated with a decline in the
dollar strength from mid-
.
7. See BIS locational banking statistics, tables A5 and A7, www .
bis /
orgstatistics
/bankstats.htm.
Global Liquidity and Procyclicality 219
border liabilities. The inference is that banks have been less
aggregate cross-
willing to roll over hedges put in place by institutional investors during the
earlier period of more ample dollar liquidity.
Direct evidence of institutional investor holdings is not very compre-
hensive. However, some evidence from national data from a few coun-
tries indicates that institutional investors have increased their holding of
external bonds. Figure 5.6 gathers some evidence on the outward portfolio
flows of insurance companies from Germany (panel A), Japan (panel B),
and Sweden (panel C). The bars indicate flows, and the line plots outstand-
ing amounts, where available. The outstanding amounts of foreign bond
holdings have fluctuated in recent years, but the general trend has been
upward.
Another source of information on foreign exchange hedging comes
yearly BIS surveys of over-
from the twice- counter foreign exchange
the-
derivatives. Panel A of figure 5.7 shows the outstanding notional amounts
by instrument, and panel B shows the breakdown by counterparty. There
back during the 2008 crisis, but strong growth in its
was a sharp pull-
2014,
aftermath. We see, however, that there has been a decline since end–
currency basis has widened.
coinciding with the period when the cross-
The category consisting of nonreporting financial institutions has seen
the largest decline in notional amounts since the end of 2014. This decline
has come after a period of strong growth and is consistent with the mar-
ket having entered a phase where foreign exchange derivative stocks have
taking in the bank-
declined amid a strengthening dollar and subdued risk-
ing sector more generally.
Thus far, the activities of advanced economy banks and investors have
been described. But a consistent theme also runs through to events in
emerging economies. For this reason, we will broaden the perspective by
considering recent events in emerging economies, especially the activities
of emerging market economy (EME) corporate borrowing in dollars. Swiss
and Japanese life insurance companies could not be more different from
emerging market corporates, but they all have in common their strong
links to the banking system and their exposure to the procyclical tenden-
taking channel” of exchange rate changes. Let us
cies driven by the “risk-
consider this now. It is the core of this chapter.
220 Hyun Song Shin
Panel A. By instrument
A 80
70
60
50
40
30
20
10
0
06-1998
12-1998
06-1999
12-1999
06-2000
12-2000
06-2001
12-2001
06-2002
12-2002
06-2003
12-2003
06-2004
12-2004
06-2005
12-2005
06-2006
12-2006
06-2007
12-2007
06-2008
12-2008
06-2009
12-2009
06-2010
12-2010
06-2011
12-2011
06-2012
12-2012
06-2013
12-2013
06-2014
12-2014
06-2015
12-2015
Forwards and swaps Currency swaps Total Options
Panel B. By sector of counterparty
B 80
70
60
50
40
30
20
10
0
06-1998
12-1998
06-1999
12-1999
06-2000
12-2000
06-2001
12-2001
06-2002
12-2002
06-2003
12-2003
06-2004
12-2004
06-2005
12-2005
06-2006
12-2006
06-2007
12-2007
06-2008
12-2008
06-2009
12-2009
06-2010
12-2010
06-2011
12-2011
06-2012
12-2012
06-2013
12-2013
06-2014
12-2014
06-2015
12-2015
Reporting dealers Nonreporting financial institutions Nonfinancial institutions
Figure 5.7
1
Over-the- counter foreign exchange derivatives— Notional principal
1
year end (end June and end December). Amounts denominated in cur-
At half-
rencies other than the US dollar are converted to US dollars at the exchange rate
prevailing on the reference date.
Source: BIS over- counter derivatives statistics.
the-
Global Liquidity and Procyclicality 221
Taking Channel and the Exchange Rate
The Risk-
In a nutshell, the proposition is this: When an international currency depreci-
ates, there is a tendency for foreigners to borrow more in that currency. Figure
5.8, which is taken from Avdjiev, Koch, and Shin (2016), illustrates the
taking channel for the US dollar. The precise mechanism will depend
risk-
taking channel is that when
on the context, but the key feature of the risk-
the dollar depreciates, banks lend more in US dollars to borrowers outside
the United States. Similarly, when the dollar appreciates, banks lend less or
even shrink outright the lending of dollars. In this sense, the value of the
taking and global credit conditions.
dollar is a barometer of risk-
A weaker dollar is associated with greater lending in dollars, lower vola-
tility, and more risk taking, but a stronger dollar is associated with higher
volatility and a recoiling from risk taking. For instance, a standard carry
taking channel (Menkhoff
trade motive would be consistent with the risk-
et al. 2012).
Panel B of figure 5.8 shows the coefficients of rolling regressions with
quarter sample window. What is notable is that the coefficient has
a 20-
crisis period. Before the 2008–
become more negative in the recent post-
0.2 to –
2009 crisis, the coefficient hovered around – 0.3, but after the crisis,
0.5. In other words, a 1 percent appre-
the coefficient has been around –
ciation of the dollar in terms of the nominal effective exchange rate is
associated with a 0.5 percent decline in the quarterly growth rate of dollar
border credit. In this sense, the value of the dollar is a key barometer
cross-
of global dollar credit conditions.
We saw earlier in figures 5.2 and 5.3 how the deviation from CIP tracked
closely the value of the US dollar. We now have a way of making sense of
this relation. The breakdown of CIP is a symptom of tighter dollar credit
conditions putting a squeeze on accumulated dollar liabilities built up dur-
ing the previous period of easy dollar credit. During the period of dollar
weakness, global banks were able to supply hedging services to institu-
border dollar credit was grow-
tional investors at a reasonable cost, as cross-
ing strongly and was easily obtained. However, as the dollar strengthens,
the banking sector finds it more challenging to roll over the dollar credit
previously supplied.
One way to summarize the finding is that a “triangle” links a stron-
border flows, and a widening of the
ger dollar, more subdued dollar cross-
Panel A. Cross-border bank lending to nonresidents versus
change in nominal effecƟve exchange rate1
A
Panel B. Twenty-quarter rolling window regressions2
B
Figure 5.8
US dollar cross- border bank lending and the dollar exchange rate
1
Plot of quarterly growth rate of cross-border bank lending in US dollars on quarterly
changes in the US dollar nominal effective exchange rate for Q1 2003– Q3 2015. Lend-
ing refers to loans by BIS reporting banks to all (bank and nonbank) borrowers outside
the United States. The line is a fitted regression line. Positive changes indicate an
appreciation of the dollar.
2
quarters window.
Rolling regression coefficient for 20-
Sources: BIS locational banking statistics; BIS effective exchange rate indices; BIS
calculations.
Global Liquidity and Procyclicality 223
currency basis against the dollar. This is the main theme explored in
cross-
Avdjiev et al. (2016). The preeminent role of the US dollar as the global
funding currency means that US monetary policy has an especially impor-
tant place in the determination of global financial conditions.
The euro, after a slow start, is showing signs of joining the dollar as an
international funding currency. Borrowers outside the euro area are bor-
term interest
rowing more in euros, taking advantage of very low long-
rates, just as borrowers outside the United States have been borrowing in
US dollars for some time. To be sure, the sums are still small for the euro.
denominated debt of nonbanks outside the euro area is
The stock of euro-
only about a quarter of the equivalent US dollar amount. But the trajectory
is steep. US companies have been particularly active in borrowing in euros.
This type of borrowing is common enough to have its own name: “reverse
Yankee” borrowing.
taking channel for the euro is starting to
Figure 5.9 shows that the risk-
show the telltale negative relationship between a weaker currency value
border lending in that currency; it was not there
and expanding cross-
before but has emerged since the crisis. The coefficient of the rolling
0.7, the coefficient is even larger in
regression is now negative. At about –
absolute terms than for the dollar. For the Japanese yen, Avdjiev, Koch,
and Shin (2016) find that its role as an international funding currency
has waxed and waned over the decades, but the telltale signs of the risk-
taking channel have reappeared in recent years with monetary easing in
Japan.
As the euro and yen join the dollar in the ranks of international fund-
ing currencies, we are left with a dilemma. With each successive wave of
monetary easing since the financial crisis, greater demands are being made
on international capital markets. One important task that remains is to
investigate how much of the observed market anomalies can be attributed
to exchange rate pressures and changing market dynamics wrought by
monetary spillovers. Spillovers and “spillbacks” have been an important
theme in international finance,8 and it looks to stay that way for the time
being.
8. This theme has been tackled by Caruana (2012), Rajan (2014), Rey (2015), and
Borio (2016).
Panel A. Cross-border bank lending to nonresidents versus
nominal effecƟve exchange rate1
A
Panel B. 20-quarter rolling window regressions2
B
Figure 5.9
Euro- denominated cross- border bank lending
1
Plot of quarterly growth rate of cross-border bank lending in euros on quarterly
changes in the euro nominal effective exchange rate for Q1 2003– Q3 2015. Lending
refers to loans by BIS reporting banks to all (bank and nonbank) borrowers outside
the euro area. Positive changes indicate an appreciation of the euro.
2
quarter window.
Rolling regression coefficient for 20-
Sources: BIS locational banking statistics; BIS effective exchange rate indices; BIS
calculations.
Global Liquidity and Procyclicality 225
Taking Channel
Macro Implications of the Risk-
taking channel has macro implications, too, and may explain
The risk-
why currency appreciation in emerging markets may sometimes be expan-
sionary rather than contractionary. Exchange rate fluctuations influence
the economy through both real and financial channels. The real effects
through the net exports channel are well known and are standard in open
Fleming model.
economy macro models, such as the textbook Mundell-
However, exchange rate fluctuations influence the economy through a
financial amplification channel as well as through net exports.
The financial channel of exchange rates operates when currency appre-
ciation elicits valuation changes on borrower balance sheets. For instance,
if the borrower has local currency assets but has borrowed in dollars, there
is a naked currency mismatch. Even if the assets generate dollar cash flows,
an empirical association may exist between a stronger dollar and weaker
cash flows, as in the case of oil firms. For whatever reason, when the poten-
tial for valuation mismatching arises from exchange rate effects, a weaker
dollar flatters the balance sheets of dollar borrowers, whose liabilities fall
relative to assets. From the standpoint of creditors, the stronger credit posi-
tion of the borrowers creates spare capacity for credit extension even with
at-
a fixed exposure limit, for instance, through a value- risk constraint. The
spare lending capacity is filled through an expansion in the supply of dollar
credit (see Bruno and Shin 2015a, 2015b).
on effects of the risk-
There are knock- taking channel on the govern-
ment’s fiscal position, too. When credit supply expands, so does the set
of investment projects, raising economic activity and improving the fiscal
position (Turner 2014; Chui, Kuruc, and Turner 2016). If corporate dollar
owned enterprises (as is the case for the
borrowing is done through state-
oil and gas sector in many EMEs), then the fiscal impact may be even more
direct through the dividends that are paid into government coffers.
Figure 5.10 shows how sovereign credit default swap (CDS) spreads for a
group of EMEs have moved with shifts in the bilateral exchange rate against
the US dollar. The horizontal axis in each panel is the percentage change in
the bilateral exchange rate against the US dollar from the end of 2012. The
year sovereign CDS
vertical axis gives the change in the local currency 5-
denominated debt
spread. The size of the bubbles indicates the total dollar-
owed by nonbanks in the country.
226 Hyun Song Shin
Panel A. End September 2013 Panel B. End June 2015
A B
Panel C. End September 2015 Panel D. End April 2016
C D
Figure 5.10
taking channel for EMEs: Bilateral US dollar exchange rate and 5-
The risk- year sover-
eign CDS, change from end–2012
BR = Brazil; ID = Indonesia; MX = Mexico; MY = Malaysia; RU = Russia; TR = Tur-
key; ZA = South Africa. The size of the bubbles indicates the size of US dollar-
denominated credit to nonbanks in the respective economies in Q4 2015.
Source: Avdjiev, McCauley, and Shin (2016); Datastream; Markit; national data; BIS;
BIS calculations.
section relation-
We see from figure 5.10 that both time series and cross-
ships exist between the CDS spread and the bilateral dollar exchange rate.
section, the bubbles line up along a downward-
In the cross- sloping line,
indicating that those countries that have depreciated more against the US
dollar tend to have CDS spreads that are higher. Over time, as the US dollar
hand corner of the
appreciates, the bubbles migrate toward the upper left-
graph; in other words, as the domestic currency weakens against the US
dollar, EME sovereign CDS spreads have tended to rise.
Interestingly, these results go away when we consider instead the trade-
weighted effective exchange rate that is unrelated to the US dollar (Hof-
mann, Shim, and Shin 2016). When we consider the component of the
Global Liquidity and Procyclicality 227
effective exchange rate that is unrelated to the US dollar, there is no evi-
dence that a currency appreciation is associated with loosening of financial
conditions. Indeed, we actually find the opposite result for some measures
of financial conditions. Again, the takeaway is that dollar strength is key for
financial conditions in emerging markets.
Beyond the Current Account
Capital flows are traditionally viewed as the financial counterpart to sav-
ings and investment decisions, and exchange rates are the automatic stabi-
lizers. In textbook models, a current account deficit can be remedied when
the exchange rate depreciates, raising net exports and closing the current
account gap.
border
Going back to 2002, figure 5.4 shows a snapshot of the cross-
banking claims denominated in US dollars around the world. Even then,
way flow was quite active between Europe and the United States.
the two-
way flow resulted from the “round-
The two- tripping” of dollars intermedi-
ated by the large European banks. These banks raised wholesale funds by
using their US branches to borrow from US money market funds, shipping
the funds back to headquarters, and then recycling the proceeds back to
the United States by purchasing securities based on mortgages of US house-
holds. A large chunk of US subprime mortgages were financed this way.
In 2002, the arrow from the United States to Europe was $462 billion (see
figure 5.4). This grew to $1.54 trillion by 2007. The return leg of the round
trip went from $856 billion in 2002 to more than $2 trillion in 2007.
The outflows to Europe were matched by the inflows from Europe,
and so the net flows were small compared to the gross flows. The current
account between Europe and the United States remained broadly in bal-
ance, even though the gross capital flows from Europe into the United
States grew enormously. Lending standards, though, are based on the size
of the balance sheet. So, gross flows are what count for lending standards.
Gross flows surged, easing lending standards and fueling the rapid increase
in credit to subprime borrowers. Borio and Disyatat (2011, 2015) give a
detailed account of why current account reasoning led some commentators
astray. My discussant Maury Obstfeld was one of the first to highlight the
importance of gross flows (Obstfeld 2010, 2012).
Why did policy makers miss the surge in subprime funding coming
from Europe? For once, we cannot blame the lack of data. Figure 5.4 was
228 Hyun Song Shin
constructed from the BIS locational banking statistics, but the BIS simply
border posi-
aggregates the data supplied by central banks. In fact, the cross-
tion data between Europe and the United States actually comes from the
central banks in those regions.
If it’s not the lack of data, then why did we miss this? The blind spot is
most likely due to our accounting conventions in international finance.
When we do international finance, we often buy into the “triple coinci-
making unit, and the currency area
dence,” where the GDP area, decision-
are one and the same (Avdjiev, McCauley, and Shin 2016). Textbooks there-
fore start with the assumptions that each GDP area has its own currency
and the use of that currency is largely confined to that economic area. The
Mundell-Fleming model is a classic example of the triple coincidence, but
even in sophisticated macroeconomic models, the triple coincidence is
rarely questioned. Currency appreciation or depreciation then acts on the
economy through changes in net exports.
One reason that triple coincidence reasoning has led researchers astray
comes from another common error that economists were making before
the crisis. As the US current account deficit grew to historically high levels,
triple coincidence reasoning would point to a depreciation of the dollar.
Many commentators wondered aloud whether there would be “sudden
stop” in the capital flows to the United States, just as in emerging market
crises (Summers 2004; Edwards 2005; Obstfeld and Rogoff 2005; Roubini
and Setser 2005; Krugman 2007).
In the event, the US dollar appreciated sharply with the onset of the
2009 global financial crisis. The dollar’s surge was associated with
2008–
a deleveraging of financial market participants outside the United States
term dollar funding to invest in risky long-
that had used short- term dol-
lar assets, with the European banks mentioned above being the most
prominent example. As the crisis erupted, these financial institutions
found themselves short the dollar and overleveraged, and they sought
to reduce their dollar liabilities, bidding up the value of the dollar in the
process.
Looking Back and Looking Ahead
2014 brings us back full circle
The strengthening of the dollar since mid-
to the mechanisms at play today. But meanwhile the protagonists have
Global Liquidity and Procyclicality 229
changed. The dollar borrowers are not European banks, but emerging mar-
ket corporates. And the borrowing is done through corporate bonds rather
than wholesale bank funding.
denominated debt of nonbanks outside the United
The stock of US dollar-
denominated
States currently stands at $9.7 trillion. Of this, the US dollar-
debt of nonbanks in EMEs stands at $3.3 trillion. This overhang of US dollar-
denominated debt has been weighing on macroeconomic conditions in
emerging market economies since the dollar started to strengthen in 2014.
To be sure, there are some mitigating factors. For one thing, much of the
recent increase in dollar debt in EMEs has been in the form of debt securi-
ties issued by emerging market corporates. These debt securities have long
maturities. In addition, many emerging economies hold substantial foreign
exchange reserves, in contrast to their situation in past crises. Demirgüç-Kunt
and Detragiache (1998) is a classic reference on the determinants of banking
crises, and many of the factors identified there do not show up currently.
Nevertheless, we have no room for complacency. Even if the bonds have
long maturities, there are other repercussions on the economy if US dollar-
denominated borrowing begins to unwind. Nonfinancial firms are deeply
embedded in the economy, and their financial activities spill over into the
rest of the economy. Bruno and Shin (2015c) find that dollar borrowing by
emerging market corporates has had the attributes of a “carry trade,” where
for every dollar raised through a bond issue, around a quarter ends up as
cash on the firm’s balance sheet. Here, cash could mean a domestic cur-
rency bank deposit or a claim on the shadow banking system, or indeed a
financial instrument issued by another firm. So, dollar borrowing will spill
over into the rest of the economy in the form of easier credit conditions.
When the dollar borrowing is reversed, these easier domestic financial con-
ditions will be reversed, too.
Furthermore, even if a country has large foreign exchange reserves, the
corporate sector itself may find itself short of financial resources and may
cut investment and curtail operations, resulting in a slowdown of growth.
So, even a central bank that holds a large stock of foreign exchange reserves
may find it difficult to head off a slowing real economy when global finan-
cial conditions tighten. Arguably, such a slowdown is part of what we are
seeing right now in emerging market economies.
All this goes to show that international financial developments have
to be placed in the broader context of past and anticipated central bank
230 Hyun Song Shin
actions. We will undoubtedly have more opportunities to discuss these
issues in policy circles in the months ahead.
References
Akram, Farooq, Dagfinn Rime, and Lucio Sarno. 2008. “Arbitrage in the Foreign
Exchange Market: Turning on the Microscope.” Journal of International Economics 76
(2): 237–253.
Avalos, Fernando, and Ramon Moreno. 2013. “Hedging in Derivatives Markets: The
Experience of Chile.” BIS Quarterly Review (March): 53–64. www.bis.org/publ/qtrpdf
/r_qt1303g.htm.
Avdjiev, Stefan, Catherine Koch, and Hyun Song Shin. 2016. “Exchange Rates and
the Transmission of Global Liquidity.” Paper presented at the Central Bank of the
Republic of Turkey Conference on Global Liquidity, March, Istanbul.
Avdjiev, Stefan, Robert N. McCauley, and Hyun Song Shin. 2016. “Breaking Free of
451.
the Triple Coincidence in International Finance.” Economic Policy 31 (87): 409–
Avdjiev, Stefan, Wenxin Du, Catherine Koch, and Hyun Song Shin. 2016. “The
Dollar, Bank Leverage, and the Deviation from Covered Interest Parity.” BIS Work-
www
ing Paper 592, Bank for International Settlements, Basel. https:// bis
. org
. publ
/
/work592.htm.
Baba, Naohiko, and Frank Packer. 2009. “Interpreting Deviations from Covered
08.” Journal of Banking
Interest Parity during the Financial Market Turmoil of 2007–
and Finance 33 (11): 1953–1962.
Baba, Naohiko, and Ilhyock Shim. 2010. “Policy Responses to Dislocations in the FX
Swap Market: The Experience of Korea.” BIS Quarterly Review (June): 29–39. www.bis
.org/publ/qtrpdf/r_qt1006e.htm.
Borio, Claudio. 2016. “More Pluralism, More Stability?” Speech at the Seventh High-
Level SNB- IMF Conference on the International Monetary System, May 10, Zurich.
www.bis.org/speeches/sp160510.htm.
Borio, Claudio, and Piti Disyatat. 2011. “Global Imbalances and the Financial Crisis:
Link or No Link?” BIS Working Paper 346, Bank for International Settlements, Basel.
www.bis.org/publ/work346.htm.
Borio, Claudio, and Piti Disyatat. 2015. “Capital Flows and the Current Account:
Taking Financing (More) Seriously.” BIS Working Paper 525, Bank for International
Settlements, Basel. www.bis.org/publ/work525.htm.
Borio, Claudio, Robert N. McCauley, Patrick McGuire, and Vladyslav Sushko. 2016.
“Covered Interest Parity Lost: Understanding Cross- Currency Basis.” BIS Quarterly
Review (September): 45–64. https://www.bis.org/publ/qtrpdf/r_qt1609e.htm.
Global Liquidity and Procyclicality 231
Bruno, Valentina, and Hyun Song Shin. 2015a. “Cross- Border Banking and Global
564.
Liquidity.” Review of Economic Studies 82 (2): 535–
Taking
Bruno, Valentina, and Hyun Song Shin. 2015b. “Capital Flows and the Risk-
132.
Channel of Monetary Policy.” Journal of Monetary Economics 71: 119–
Bruno, Valentina, and Hyun Song Shin. 2015c. “Global Dollar Credit and Carry
Trades: A Firm-Level Analysis.” BIS Working Paper 510, Bank for International
Settlements, Basel. www.bis.org/publ/work510.htm.
Caruana, Jaime. 2012. “Policymaking in an Interconnected World.” Speech at Fed-
eral Reserve Bank of Kansas City Economic Policy Symposium, August 31, Jackson
Hole, WY. www.bis.org/speeches/sp120903.htm.
Caruana, Jaime. 2016. “Credit, Commodities and Currencies.” Lecture at the London
School of Economics, February 5. www.bis.org/speeches/sp160205.htm.
Chui, Michael, Emese Kuruc, and Philip Turner. 2016. “A New Dimension to Cur-
rency Mismatches in the Emerging Markets– Non- financial Companies.” BIS Work-
ing Paper 550, Bank for International Settlements, Basel. www.bis.org/publ/work550
.htm.
Demirgüç- Kunt, Aslı, and Enrica Detragiache. 1998. “The Determinants of Banking
Crises: Evidence from Industrial and Developing Countries.” IMF Staff Papers 45 (1):
1–109.
Du, Wenxin, Alexande Tepper, and Adrien Verdelhan. 2016. “Deviations from Cov-
ered Interest Rate Parity.” papers.ssrn.com/sol3/papers.cfm?abstract_id=2768207.
Edwards, Sebastian. 2005. “Is the US Current Account Deficit Sustainable? If Not, How
288.
Costly Is Adjustment Likely to Be?” Brookings Papers on Economic Activity 1: 211–
Gabaix, Xavier, and Matteo Maggiori. 2015. “International Liquidity and Exchange
1420.
Rate Dynamics.” Quarterly Journal of Economics 130 (3): 1369–
Goldberg, Linda, and Cedric Tille. 2009. “Macroeconomic Interdependence and the
1003.
International Role of the Dollar.” Journal of Monetary Economics 56 (7): 990–
Gopinath, Gita. 2016. “The International Price System.” In Inflation, Dynamics and
Monetary Policy, Proceedings of the Federal Reserve Bank of Kansas City, August 27–
29,
2015, Jackson Hole, WY, 71–150.
Olivier, and Maurice Obstfeld. 2012. “Stories of the Twentieth
Gourinchas, Pierre-
Century for the Twenty- First.” American Economic Journal: Macroeconomics 4 (1):
226–265.
Hofmann, Boris, Ilhyock Shim, and Hyun Song Shin. 2016. “Sovereign Yields and
Taking Channel of Currency Appreciation.” BIS Working Paper 538, Bank
the Risk-
for International Settlements, Basel. http://www.bis.org/publ/work538.htm.
467.
Krugman, Paul. 2007. “Will There Be a Dollar Crisis?” Economic Policy 22 (51): 436–
232 Hyun Song Shin
McCauley, Robert, Patrick McGuire, and Goetz von Peter. 2010. “The Architecture of
Global Banking: From International to Multinational?” BIS Quarterly Review (March):
25–37. www.bis.org/publ/qtrpdf/r_qt1003e.htm.
McGuire, Patrick, and Nikola Tarashev. 2007. “International Banking with the Euro.”
BIS Quarterly Review (December): 47–61. www.bis.org/publ/qtrpdf/r_qt0712f.htm.
Menkhoff, Lukas, Luciano Sarno, Maik Schmeling, and Andreas Schrimpf. 2012. “Carry
718.
Trades and Global Foreign Exchange Volatility.” Journal of Finance 67 (2): 681–
Obstfeld, Maurice. 2010. “Expanding Gross Asset Positions and the International
Monetary System.” Panel Remarks at Federal Reserve Bank of Kansas City Economic
Policy Symposium at Jackson Hole, August 27.
Obstfeld, Maurice. 2012. “Does the Current Account Still Matter?” American Eco-
23.
nomic Review 102 (3): 1–
Obstfeld, Maurice, and Kenneth S. Rogoff. 2005. “Global Current Account Imbalances
146.
and Exchange Rate Adjustments.” Brookings Papers on Economic Activity 1: 67–
Rajan, Raghuram. 2014. “Competitive Monetary Easing: Is It Yesterday Once More?”
Remarks at the Brookings Institution, April 10, Washington, DC.
Rey, Hélène. 2015. “International Channels of Transmission of Monetary Policy and
the Mundellian Trilemma.” Mundell- Fleming Lecture presented at the International
Monetary Fund Fifteenth Jacques Polak Annual Research Conference: Cross- Border
Spillovers, November 13– 35.
14, Washington, DC. IMF Economic Review 64 (1): 6–
Roubini, Nouriel, and Brad Setser. 2005. “Our Money, Our Debt, Our Problem.” For-
200.
eign Affairs 84 (4): 194–
Shin, Hyun Song. 2016a. “Bank Capital and Monetary Policy Transmission.” Panel
.
remarks at the ECB and Its Watchers XVII Conference, April 7, Frankfurt. wwwbis
.org/speeches/sp160407.htm.
Shin, Hyun Song. 2016b. “Market Liquidity and Bank Capital.” Remarks at the Per-
spectives 2016: Liquidity Policy and Practice Conference, April 27, London Business
School. www.bis.org/speeches/sp160506.htm.
Summers, Lawrence H. 2004. “The US Current Account Deficit and the Global
Economy.” Per Jacobsson Lecture at the International Monetary Fund, October 3,
Washington, DC.
Turner, Philip. 2014. “The Global Long- Term Interest Rate, Financial Risks and Policy
Choices in EMEs.” BIS Working Paper 441, Bank for International Settlements, Basel.
www.bis.org/publ/work441.htm.
Kunt
Comment: Aslı Demirgüç-
It was a pleasure to read Hyun Shin’s chapter on global liquidity and procy-
clicality. Indeed, any paper that starts with emphasizing the importance of
finance for the real economy is music to my ears, as I have spent a large part
of my professional life arguing that “finance matters for economic develop-
ment,” rather than merely the other way around. In the beginning of the
chapter, Hyun says “the financial tail is wagging the real economy dog.”
For many of us in the finance and development field, finance is the brain
anyway, not the tail, so this is not very surprising from that perspective.1
Hence I like the emphasis on the role of the financial system in the inter-
national economy and how problems in the financial system and the inter-
mediation process might spill over to the rest of the economy. Therefore
I am predisposed to agree with the arguments and the main conclusion of
the chapter.
However, the job of the discussant is to think of ways to sharpen the
arguments and strengthen the chapter, so that is what I will try to do in my
comments. My first observation is that although there is a lot to like in this
chapter, there are also a lot of moving parts. It pulls together a lot of data
and analysis from different pieces of work. Indeed, I would characterize it
as a collection of interesting, provocative hypotheses rather than a fully
developed argument. So, although the data and evidence presented are
compelling, it is not always clear how the links are made, and sometimes
possible alternative explanations are not adequately covered to present a
I am grateful to Sergio Schmukler and Ha Nguyen for helpful comments.
1. See, for example, Levine (2005); Demirgüç-
Kunt and Levine (2008, 2009); Cull,
Demirgüç- Kunt, and Morduch (2011); Ayyagari, Demirgüç- Kunt, and Maksimovic
(2013); and Cihak and Demirgüç- Kunt (2014); among others.
234 Kunt and Maurice Obstfeld
Comments by Aslı Demirgüç-
coherent storyline. This approach leaves the reader with more questions
than answers. Nevertheless, the ideas presented here are very thought pro-
voking, which no doubt will lead to much more research in these areas.
Identification Issues
The chapter starts by asking two main questions. Why are global financial
conditions so attuned to the strength of the US dollar? And why is the
economy so sensitive to global financial conditions? These are important
yet complicated questions, and drawing from my own area of expertise,
they are immediately subject to the identification problem. In other words,
when we try to answer these types of questions looking at equilibrium out-
comes, it is very difficult to figure out the direction of causality. It could
simply be that the global financial conditions are sensitive to the economy,
or what we observe could simply be reflecting other factors at play.
It is not that different here. Take, for example, the centrality of the dol-
lar in the global banking system. Although the dollar certainly plays a large
part in the world economy, transactions in other currencies are also grow-
ing, as the chapter also mentions. Hence the dollar may not be the only
driving factor.
Shin observes that when an international currency depreciates, there is a
tendency for foreigners to borrow more in that currency. Hence, banks lend
more internationally when the dollar is weak. But again, how much of this
trend is a mere reflection of other currencies strengthening? For example,
as emerging markets boom, capital flows in, their currencies appreciate, and
à-
the dollar depreciates vis- vis these currencies. This process is not necessar-
ily driven by the dollar; instead, the dollar exchange rate is just a reflection
of this process.
2009
Another observation made in the chapter is that during the 2008–
global financial crisis, the dollar appreciated strongly with the onset of the
crisis, despite the large US current account deficit. But again, we need to
remember that these developments coincided with a run toward safe assets
(notably, US treasuries), so it is not possible to disentangle how much of
this appreciation was due to dollar per se, which was surely attractive for
other reasons.
Overall, it is not clear that causality goes from the dollar to other mar-
kets; the dollar may not be as central as Shin argues, but may be just a
Global Liquidity and Procyclicality 235
reflection of an entirely different set of factors at play. Indeed, Shin also
mentions that similar patterns are observed with other currencies, like the
yen and Swiss franc.
Limits to Arbitrage, Portfolio and Foreign Direct Investment Flows,
Gross versus Net Flows
Other points would also benefit from a more detailed explanation in the
chapter. First, an interesting market anomaly that is highlighted is the fail-
ure of covered interest parity (CIP). We generally expect market interest
rates and the implied interest rates from forward rates embedded in foreign
exchange swaps to be more or less consistent. But as Shin reports, this has
not been the case in recent years, particularly for periods of a strong dollar.
Unfortunately, there is little explanation of why we observe this phenom-
taking capacity (or
enon. The chapter mentions in passing issues of risk-
limits to arbitrage) and counterparty risk, which could play important roles
in explaining this anomaly. But given that a big part of the story depends
on the inability of financial markets to hedge risk, it seems that this should
deserve more attention than it gets in the chapter. For example, why does
currency basis swap
a dollar appreciation lead to a more negative cross-
spread?
Second, why is the central focus of the chapter on bank flows as opposed
to other flows? Shin focuses mostly on the importance of bank flows, which
are, of course, highly relevant. However, a significant part of the increasing
flows are portfolio and foreign direct investments. And for many countries
around the world, these other two components have grown more quickly
and might now surpass bank flows. The chapter should at least acknowl-
edge this and discuss the implications.
Third, Shin makes a distinction between net versus gross flows, which
is welcome.2 But a significant part of the story is related to net financing.
As home bias diminishes and residents have more wealth to invest, gross
flows will expand as individuals diversify their portfolios internationally
and hold one another’s portfolios. Figure 5.11 illustrates that as countries
2. Shin relies on BIS data for this analysis, but gross flows are also available from
balance of payments data by type of flow. Gross investment, issuance, and portfolio
positions are available, too. See for example, World Bank (2015).
Panel A. High-Income Countries
Panel B. Upper-Middle-Income Countries
Panel C. Lower-Middle-Income Countries
Figure 5.11
Net and gross capital flows
income countries
A. High-
B. Upper-middle-income countries
C. Lower-middle-income countries
Global Liquidity and Procyclicality 237
become richer, we expect gross flows to grow, although the trends in net
flows are much less clear (Broner et al. 2013).
The effects of a shock may play out very differently depending on the
reallocation between foreign and domestic investors as they retrench from
the expansion period. To the extent that gross flows expand, what is impor-
tant is how the asset and liability positions expand. As many emerging mar-
ket economies have accumulated reserves, reduced sovereign borrowing,
and received foreign direct investment and equity inflows, dollar apprecia-
tions and market collapses have been accompanied by a strengthening of
their net foreign positions.
Overborrowing by Emerging Market Corporates?
Looking ahead, the chapter also tries to identify sources of fragility. One
interesting conjecture is whether emerging market corporates will cause
the next crisis. Shin asks whether we are going to see another East Asian
crisis, where corporates were at the heart of the problem. Though the chap-
ter does not devote much space to this discussion, it is nevertheless worth
commenting on. Excessive borrowing to finance risky investments can be
exacerbated by global liquidity conditions and may be a valid source of
concern. However, there are mitigating factors, and some questions need to
be answered to ascertain whether this concern is serious.
First, measuring risk taking in financial markets is difficult, because posi-
tions can be hedged. So an important question is: What proportion of these
positions are open or unhedged? It is also difficult to decide what should
be the benchmark level of indebtedness when discussing whether corpora-
tions are overborrowing.3
Second, as discussed at length in Global Financial Development Report
Term Finance (World Bank 2015), the emerging market
2015/2016 on Long-
corporates that borrow abroad do so through bond issuance in foreign cur-
rency, but this means they also extend their maturity at the same time, as
foreign corporate bond markets are longer than domestic ones. Indeed, as
figure 5.12 shows, in developing countries, maturity of international bond
issues tends to be longer than that of domestic issues, although the reverse
3. See, for example, Alfaro et al. (2016) for a discussion of different benchmarks and
the sensitivity of conclusions to the choice of these benchmarks.
238 Kunt and Maurice Obstfeld
Comments by Aslı Demirgüç-
12
10.8
11
10.1
10
9.2
9
8
Maturity in years
6.8
7
6
5
4
3
2
1
0
Developed countries Developing countries
Domestic bond issuances International bond issuances
Figure 5.12
Maturity of domestic and international issuances, corporate-bond market
Notes: This figure reports the weighted average maturity of domestic and interna-
tional corporate bonds issued by firms from developed and developing countries.
The sample period is 1991– 2014.
Source: Cortina, Didier, and Schmukler (2016).
is true for developed countries (Cortina, Didier, and Schmukler 2016). This
is only briefly mentioned in the chapter.
Third, the important role of reserve accumulation by emerging markets
is also mentioned in passing in the chapter, but it deserves more elabora-
tion. To the extent that governments hold foreign reserves, they benefit
from an appreciation of the US dollar, compensating for the potential losses
that the corporates might suffer. The dollar appreciation may have fiscal
costs (due to a potential bailout), but the government will have additional
resources. Whether this is enough will depend on the size of government
assets versus unhedged corporate liabilities. Otherwise on net, it is not clear
whether the result would be gains or losses from an appreciation of the US
dollar due to funding abroad. At any rate, only a very few of the largest
corporates in emerging markets are able to access international markets.
Fourth, the main concern expressed in the chapter is that firms engage
in carry trade (i.e., they issue bonds at low rates to accumulate cash and
undertake risky financial intermediation activities in their home coun-
tries). But according to Bruno and Shin (2015), at most firms accumulate
23 percent of each dollar raised through bond issues (the estimates vary
substantially and can be as low as 4 percent). This does not seem to be a
Global Liquidity and Procyclicality 239
large enough figure to be concerned about this effect. Clearly, the major-
ity of the finance raised is used to finance growth opportunities through
capital investment, growth in employment, mergers and acquisitions, and
the like, as expected.
Fifth, large firms could indeed be using some of the cash to finance other
firms, such as their suppliers. This intermediation process might channel
funds from large companies to small and medium enterprises that cannot
access capital markets directly because of information asymmetries; as a
result, it could relax their financing constraints. If large companies have
better information and are able to overcome information asymmetries that
these smaller firms often face, this activity may be beneficial.
Finally, the fact that the Bruno and Shin (2015) results are driven by
emerging markets makes the reader wonder what is special about these
countries. Another important question is what the role of financial firms is.
One would think they would be in a better position to engage in carry trade.
Offs and Parallels
Trade-
One implication of the chapter is that although it is potentially an impor-
tant source of economic benefits, financial globalization also has potential
offs that monetary policy faces in navigat-
downsides. It worsens the trade-
ing among multiple domestic objectives. There is the basic one between
inflation and unemployment. But financial stability considerations are also
important. So, for example, optimal monetary policy may have to be pulled
away from the traditional macroeconomic goals of price stability and full
employment to restrain debt buildups, particularly in the absence of effec-
prudential tools.
tive macro-
These problems only become worse in an open economy, because open-
ness to global financial markets will inevitably reduce the effectiveness of
prudential tools that are available. So the trade-
the macro- off between macro
stabilization and financial stability becomes even more difficult. If a bigger
interest rate change is required to bring about a given demand response
prudential problem by
in an open economy, this may worsen the macro-
increasing the fragility of banks and encouraging gross financial flows.
This discussion has important parallels to banking globalization, which
is the topic of the Global Financial Development Report 2017/18 (World Bank
2018). It, too, describes an inherent tension between risk diversification
240 Kunt and Maurice Obstfeld
Comments by Aslı Demirgüç-
return countries and the
and sharing as capital flows from low-to high-
implied necessity for exposing oneself to shocks and trends from abroad.
The benefits are many: In addition to resource mobilization and risk
sharing, importantly, the entry of international banks can increase com-
petition in the domestic banking industry, improving the efficiency of
resource allocation, which is key to promoting economic development.
When entry happens through brick and mortar, foreign banks often bring
new technical knowledge, improve human capital in the industry, generate
demand for improving regulation and supervision, and are generally less
subject to political manipulation. These findings are quite well established
in the literature.
But there are also potential costs. As in the global financial crisis, host
countries might be exposed to external shocks transmitted by international
banks, endangering their stability. It is also true that international banks
might fuel excessive credit booms in host countries that end up in busts,
because domestic financial systems are not capable of handling such flows.
amplified by global liquidity conditions—
Such behavior— might be harm-
ful for the financial stability of home and host countries, ending up in
border contagion risks.
costly boom and bust cycles and cross-
level data from more than
For example, in a recent paper, we use bank-
2010 to study bank lending behavior over the
100 countries during 1999–
business cycle (Bertay, Demirgüç-
Kunt, and Huizinga 2015). Of all the banks
in the sample, lending by foreign banks is the most procyclical, increasing
their lending much more during upswings compared to domestic banks
(figure 5.13). This is potentially because they can access funding from their
international parent firms to take advantage of local lending opportunities
during economic growth periods.
A Research Agenda for Developing Countries
It is useful to frame this discussion in the context of recent trends in bank
internationalization (namely, the dramatic growth of foreign banking in
the 1990s), followed by the retrenchment as a result of the crisis and the
south flows to at least partially compensate this retrench-
increase in south-
ment. Thus viewed, this discussion raises important policy questions for
developing countries and lays out a research agenda. Several questions are
in the minds of policy makers.
Global Liquidity and Procyclicality 241
2.5
2.1***
2
1.5
Percent
1.2***
1
0.5 0.4**
0
Foreign banks DomesƟc private banks State banks
Figure 5.13
Change in bank lending associated with 1 percent growth in GDP per capita, 1999–2010
Note: The figure shows marginal effects from a regression of bank lending on GDP
per capita growth and number of control variables, estimated using a sample of
1,633 banks from 111 countries. Significance level: ** 5 percent, *** 1 percent.
Kunt, and Huizinga (2015).
Source: Bertay, Demirgüç-
First: Are international banks too fickle to be heavily relied on by devel-
oping countries? Especially if they enter through acquisition, is there a risk
that they will hollow out existing banks by substituting for local provision
key functions from foreign headquarters? If so, information technology,
certain aspects of payments capability, and even risk management skills
could be lost or substantially eroded locally if the bank decides to exit the
country. Although this question is age old, it has been receiving increased
policy attention since the global financial crisis, as capital regulations on
many European and US international banks induce them to retrench from
international business (for example, the recent retreat of Barclays from
Africa). So with the retrenchment after the crisis, has our policy advice to
developing countries on foreign banking changed?
south entry, should developing coun-
Second: Given the rise of south-
try authorities be especially cautious in their approach to admitting south-
south international banking activities? For example, Chinese banks may
be beginning to expand into Africa, Southeast Asia, and Latin America.
242 Kunt and Maurice Obstfeld
Comments by Aslı Demirgüç-
Should one worry about the lack of experience and perhaps insufficient
CFT supervision in some south-
home country prudential and AML- south
specific knowledge give these banks
cases? Or does the cost base and region-
a better potential to provide banking services on a solid basis in the host
countries?
Third: What is the development impact of international banking, partic-
ularly when it comes to access and inclusion? Does allowing foreign banks
a larger share risk reducing the access and increasing the price of banking
services to small and medium enterprises (SMEs) and lower income house-
holds? This is an old question, but not as much work has been devoted to it
as has been to analyzing efficiency and stability concerns. Yet it is still one
of the big policy questions.
And finally: What is the future going to look like? How do we expect
tech to modify global banking? How would
technological advances and fin-
border and brick and mortar banking change
potential blurring of cross-
our answers to the questions above? What should financial regulation and
supervision look like in a world in which international banking is much
larger?
Overall financial globalization, including banking globalization, can
offs. The challenge of policy will be to maximize
lead to important trade-
the benefits of bank internationalization while minimizing the costs. It is
an exciting agenda, which we will be working on over the coming years.
References
Alfaro, Laura, Gonzalo Asis, Anusha Chari, and Ugo Panizza. 2016. “Lessons Learned?
Corporate Debt in Emerging Markets.” NBER Working Paper 3407, National Bureau
of Economic Research, Cambridge, MA.
Ayyagari, Meghana, Aslı Demirgüç-Kunt, and Vojislav Maksimovic. 2013. “Financ-
ing in Developing Countries.” In Handbook of the Economics of Finance, volume 2B,
edited by George Constantinides, Milton Harris, and Rene Stulz, 683–757. Boston:
Elsevier.
Bertay, Ata, Aslı Demirgüç-Kunt, and Harry Huizinga. 2015. “Bank Ownership and
Credit over the Business Cycle: Is Lending by State Banks Less Procyclical?” Journal of
Banking and Finance 50 (C): 326– 339.
Broner, Fernando, Tatiana Didier, Aitor Erce, and Sergio L. Schmukler. 2013. “Gross
133.
Capital Flows: Dynamics and Crises.” Journal of Monetary Economics 60 (1): 113–
Global Liquidity and Procyclicality 243
Bruno, Valentina, and Hyun Song Shin. 2015. “Global Dollar Credit and Carry
Trades: A Firm-Level Analysis.” BIS Working Paper 510, Bank for International
Settlements, Basel.
Cihak, Martin, and Aslı Demirgüç-
Kunt. 2014. “Revisiting the State’s Role in Finance
and Development.” In Handbook of Banking, second edition, edited by Allen N.
Berger, Philip Molyneux, and John O. S. Wilson, 777– 806. New York: Oxford Uni-
versity Press.
Cortina, Juan Jose, Tatiana Didier, and Sergio L. Schmukler. 2016. “How Long Is the
Maturity of Corporate Borrowing? Evidence from Bond and Loan Issuances across
Markets.” Policy Research Working Paper 7815, World Bank, Washington, DC.
Cull, Robert, Aslı Demirgüç-
Kunt, and Jonathan Morduch. 2011. “Microfinance
Tradeoffs: Regulation, Competition, and Financing.” In Handbook of Microfinance,
edited by Beatriz Armendariz and Marc Labie, 141– 157. Hackensack, NJ: World
Scientific.
Demirgüç-Kunt, Aslı, and Ross Levine. 2008. “Finance, Financial Sector Policies, and
Long Run Growth.” M. Spence Growth Commission Background Paper 11, Wash-
ington, DC.
Demirgüç-Kunt, Aslı, and Ross Levine. 2009. “Finance and Inequality: Theory and
318.
Evidence.” Annual Review of Financial Economics 1: 287–
Levine, Ross. 2005. “Finance and Growth: Theory and Evidence.” In Handbook
of Economic Growth, volume 1A, edited by Philippe Aghion and Steven Durlauf,
865–
Holland.
934. Amsterdam, Boston: Elsevier, North-
Term Finance.
World Bank. 2015. Global Financial Development Report 2015/16: Long-
Washington, DC: World Bank.
World Bank. 2018. Global Financial Development Report 2017/18: Banker without
Borders. Washington, DC: World Bank.
Comment: Maurice Obstfeld
Thank you, Kaushik, for inviting me to comment on this chapter by Hyun
Song Shin. It’s always a pleasure to come across 19th Street to the World
Bank. This is a very nice chapter that summarizes and brings together mate-
rial about global liquidity and credit that Hyun Song Shin and the Bank for
International Settlements (BIS), more generally, have been calling to our
attention for a while. There are two main themes. One is that US financial
conditions drive global conditions. The second is that the US dollar’s value
is a key barometer of global liquidity conditions and hence of risk taking.4
In establishing propositions one and two above, the chapter looks at a
number of pieces of evidence, such as CIP deviations, US dollar denomi-
nated bank lending data, and sovereign CDS spreads. The underlying driver
put forth to explain the facts is that the US dollar has a unique role as an
international currency: as an invoice currency, as a funding currency, as a
vehicle currency, and as a reserve currency.
My comments will be based on four observations, some of which are
macro comments and some are finance comments. Before covering these,
let me flag the chapters’s important observation with respect to CIP devi-
ations. My interest in this should not be a surprise, given that my text-
book (Krugman, Obstfeld, and Melitz 2017) is one that commits the sin of
claiming that CIP holds (more precisely, held quite closely for about three
2009 global financial crisis). We will have to be
decades up until the 2008–
I gratefully acknowledge helpful assistance from and discussions with Eugenio
Cerutti. All opinions and errors are mine alone.
4. One subtheme in the chapter that was not emphasized is that the euro and yen
may be growing in international importance. I won’t have time to go into this issue,
but I’m a bit skeptical, given the challenges that those economies currently face.
Global Liquidity and Procyclicality 245
sure in the next edition of the book to acknowledge more fully the seeming
arbitrage opportunities that have persisted long after the end of the crisis;
more on these below.
My first observation concerns the relation between the exchange rate
and the current account. Even in theory, a current account deficit does
not necessarily signal future depreciation over any specific time frame.
Even in the simplest model with perfect substitution among assets, and
where portfolio effects therefore are not important, the relationship is not
straightforward. A current account deficit could arise because of a fall in
foreign demand or a rise in domestic demand, and these two events will
have completely opposite effects on the exchange rate and output in the
short run. The point is that the exchange rate movements are going to be
endogenous, so we cannot really speak of an exchange rate change leading
to a contractionary effect. This really depends on what is driving it. Now, if
we go to the kind of world that Hyun is talking about, where there are also
way gross capital flows and a rich array of different assets and liabili-
two-
ties traded, then indeed, life is going to become much more complicated.
We can think about portfolio shifts between asset classes, possibly due to
changes in preferences, policy liquidity conditions, and the like. But here
again, currency appreciation need not be contractionary, as a more tra-
ditional approach to international economics might indicate through its
exclusive focus on the net export effect. For example, one very important
channel that Hyun and others have stressed arises from the presence of
dollar liabilities, such that domestic net worth can increase when the cur-
rency appreciates. Any resulting easing of binding credit constraints will be
expansionary. More recently, Olivier Blanchard and coauthors (Blanchard
et al. 2016) have suggested a different channel. They look at nonbond
inflows and show that these can be expansionary. So, more generally, I see
here a very interesting research agenda that looks more deeply to under-
stand the complex links among the current account, the exchange rate,
and the macroeconomic conjuncture.
Let me turn to my second comment, which is about CIP violations. This
is a fascinating anomaly. In perspective, there are many other asset market
anomalies that have arisen since 2008, some of which do not obviously
have much to do with the international economy specifically, but likely
have to do with liquidity and asset markets in general. Part of the rethink-
ing we’ve been doing since the global financial crisis centers on figuring
246 Kunt and Maurice Obstfeld
Comments by Aslı Demirgüç-
Figure 5.14
US dollar exchange rate and the cross- currency basis
1
Simple average of bilateral exchange rate of the dollar against CAD, EUR, GBP, SEK,
CHF and JPY. Higher values indicate stronger US dollar.
2
year cross-
Simple average of the five- currency basis swaps against CAD, EUR, GBP,
SEK, CHF, and JPY vis- vis the US dollar.
à-
Sources: Bloomberg; BIS calculations. This chart is from S. Avdjiev, W. Du, C. Koch,
and H. S. Shin, 2016. Exchange rates, currency hedging and the cross- currency basis.
out how things that we thought were true and obvious seem not so true or
obvious anymore. But CIP is a particularly fascinating case, because, since
Keynes (1923) first explained covered interest parity in 1923, it has been an
article of faith (despite deviations over long stretches, when currency mar-
kets and international arbitrage were restricted). But what you see in figure
5.14, which is a repeat of figure 5.2 from the Shin chapter, is that since the
global financial crisis, CIP no longer works very well. The upper line graphs
an average exchange rate against the dollar, and when it rises, the US dollar
appreciates. The lower line is the swap basis, which as Hyun explains, is the
+ iUS,
difference between the gross LIBOR interest rate, which is denoted 1
and the covered foreign gross interest rate. This gap has generally been
negative and substantial in absolute magnitude since the financial crisis.
and
Why? Hyun argues that the gap shrinks when the dollar is weaker—
owing to the easier global
presumably when Fed policy is relatively easy—
liquidity conditions that result. My guess, however, is that different factors
are of greater or lesser importance over different periods.5
5. For an exploration of the changing factors driving CIP deviations over time, see
Cerutti, Obstfeld, and Zhou (2019).
Global Liquidity and Procyclicality 247
For example, we see a big widening of the swap basis in the period of
the euro crisis. During that period, the dollar is actually somewhat weak,
compared to its period average, because this is also the period before the
temper tantrum unwinds. So it is likely that the story is more complex than
other things may be going on. One very interesting
in Hyun’s account—
theory, one that focuses on the euro crisis, is told by Ivashina, Scharfstein,
and Stein (2015). Interestingly, it is based on a structural factor that is very
central to Hyun’s story: the large extent of dollar financial intermediation
in the world economy. Ivashina and coauthors point out that European
banks have a structural deficit of US dollar funding in the sense that they
want to lend a lot of dollars, but their natural (explicitly and implicitly
insured) deposit base, which therefore is somewhat cheaper to tap, is in
euros. Please look at figure 5.15, based on a paper out of the IMF Research
Department by Eugenio Cerutti and coauthors. As you can see, there is a
lot of bank lending to emerging markets, euro area banks play a key role,
and they lend predominantly in US dollars. This snapshot is very consistent
with the story that Hyun is telling.
Figure 5.15
border bank lending to emerging markets
Cross-
Sources: BIS Banking Statistics; and Cerutti, Claessens, and Ratnovski (2017). The
sample of emerging markets includes 49 large emerging markets.
248 Kunt and Maurice Obstfeld
Comments by Aslı Demirgüç-
So, what do these banks do when they have a deposit base in euros, but
they want to lend US dollars? They borrow euros and swap them into dol-
lars, and then they can keep rolling over those swaps. This imbalance, how-
ever, gives rise to a structural excess supply of forward dollars, and thus,
the pattern of CIP deviations that Hyun has shown us. Why does classical
arbitrage not eliminate these gaps? Given even small repayment frictions,
but in a much different environment since the global financial crisis, limits
to arbitrage (which can be due to liquidity, limited capital, market struc-
ture, etc.) allow CIP gaps to persist. In the Ivashina, Scharfstein, and Stein
(2015) work, when euro area banks become more stressed, as they certainly
seem to be now, they may find that the comparative advantage of euro over
US dollar funding rises, which will induce them to do more synthetic US
dollar borrowing through the swap market. The result of what is basically a
demand effect will push up the cost of such funding.
Hyun’s chapter does not go into a lot of detail here, but my reading is
that he puts more emphasis on the suppliers of these swaps, which are
likely to be other banks. These banks also face limited capital and other
impediments to arbitrage, impediments that recede when US monetary
demand and supply—
policy is easier. So, both forces— are going to be in
play. The big central banks have recently changed the architecture of some
of these markets quite substantially through the introduction of stand-
ing swap lines among themselves, but it is unclear in the very short run
whether disruptions could occur nonetheless. I would join Hyun in the
plea for more research on this general topic, and more work on developing
equilibrium picture.
a general-
I would also observe, putting a macro hat back on, that there could be
a real channel that works against Hyun’s hypothesized mechanism. When
the US dollar strengthens due to tighter Fed policy, the euro weakens, which
has positive effects on the real euro area economy and thereby helps its banks.
So a range of complex macro and financial effects are in play. One interest-
ing question that Hyun does not address is evident from the figures in the
chapter: some currencies (like sterling) have pretty small basis deviations, but
for others they are quite large. For the Swiss franc, we see some huge spikes,
because it is now a safe haven and the Swiss National Bank’s interventions in
currency markets have been associated with considerable turbulence. What is
going on across currencies? We have no good sense of that, but the fact that
euro area banks appear especially challenged should not surprise us.
Global Liquidity and Procyclicality 249
My comment number three is also a macro comment: Is the US Federal
powerful? There are powerful global forces at work, but
Reserve really all-
they also lie behind the global level of the natural real interest rate, and
one can argue that the latter is driving monetary policies worldwide. Sure,
the US dollar’s role is important; but is it really the central fact here? I think
2000s, when Alan Greenspan was lamenting the conun-
back to the mid-
term dollar interest rates, with little apparent impact
drum of raising short-
term dollar interest rates. At the same time, there was widespread
on long-
discussion of global saving gluts and global imbalances, and the limits of
US monetary policy in the face of those global flows, which were held to
have depressed real interest rates worldwide. In light of current debates over
the role of the Fed in the global economy, it is useful to recall those debates
of the past decade.
In a related vein, Hyun mentions some work by his colleague Claudio
Borio, and Hyun himself has also done some work along the same lines.
financial analysis, of which I think Hyun’s chapter is
A strand of macro-
representative, downplays the role of the Wicksellian natural or neutral real
interest rate in favor of the primacy of financing conditions. That approach
does help make sense of issues like US dollar funding and liquidity, is criti-
term market dynamics, and certainly illuminates problems
cal for short-
that macroeconomists missed before the global financial crisis. But the old
conventional macroeconomic issues still remain important. For example,
beyond other measures of financial conditions, we have seen that global
mostly driven by nominal interest rates—
real interest rates— have shown a
powerful downward trend since at least the 1990s, as shown in figure 5.16,
and macroeconomic flow factors seem likely to be key drivers.
When I think about the interaction of complex financing and macro
issues, I find it helpful to remember Tobin’s work, which was very influen-
tial at the time but at some level never became totally mainstream. Tobin’s
research program aimed to reconcile stock and flow equilibrium phenom-
ena in models with a rich menu of assets. (See, for example, Tobin 1981.)
Taking Tobin seriously, one would acknowledge the mutual consistency of
stock equilibrium and flow equilibrium, as well as their tendency to interact
over time and thereby determine the economy’s dynamic path. The condi-
tions of stock equilibrium matter, because changes there (for example, a rise
in the portfolio demand for safe assets) change asset prices and affect flows
of saving and investment, with effects that alter the entire future path of
250 Kunt and Maurice Obstfeld
Comments by Aslı Demirgüç-
Figure 5.16
Global 10-year real interest rates
Notes: Calculated as nominal 10- year-
year bond yields minus 10- ahead CPI inflation
forecast (consensus forecast). Sample includes Australia, Canada, France, Germany,
Italy, Japan, Netherlands, New Zealand, Norway, Spain, Sweden, Switzerland,
United Kingdom, and United States.
Sources: IMF, Global Data Source; Bloomberg L.P.; and Consensus forecasts.
the economy. In contrast, the fundamental Wicksellian natural rate, which
is established in the flow equilibrium of global saving and investment, is
the foundation for the whole array of risky rates of return that the econo-
my’s available assets offer. If Tobin were here, he would certainly endorse
adding realistic financial constraints and financing frictions to the models
we use, as those also feed into the flow equilibrium. When I call for devel-
equilibrium models, I am calling for a reconciliation of the
oping general-
stock and flow points of view, because I do not think they are contradictory.
However, it is important to recognize that in some situations, shocks to the
flow equilibrium will dominate. For example, China’s entry into the world
economy is a stock story, but it also represents a big flow shock. It was a
flow shock in the first instance, because China started out not being inte-
grated into world markets. Now that China is rebalancing and is somewhat
more integrated financially, we are seeing stock shocks galore emanating
some through direct Chinese financial relationships; most
from China—
through expectation effects in foreign asset markets.
My fourth and final comment is on the implications of Hyun’s findings for
emerging market monetary independence. I will keep an open mind about
world pertinence of the following argument, but it is one implication
the real-
of thinking hard about the questions on CIP that Hyun is raising. Assume
Global Liquidity and Procyclicality 251
of costs being lower when borrowing in
the pattern that Hyun describes—
the US currency market versus borrowing dollars by borrowing foreign cur-
rency, buying dollars with it, and using forward transactions to offset cur-
F
rency risk. Then 1 + iUS < (1 + i * ), where F is the forward dollar price of the
S
foreign currency, S is the spot dollar price of foreign currency, and 1 + i* is
the gross foreign- currency interest rate. But in that case also, you get another
⎛ F −1 ⎞
inequality: ⎜ −1 ⎟ (1 + iUS ) < 1 + i * . This expression states that if you reside in
⎝S ⎠
an emerging market and Hyun’s pattern of forward rates, spot rates, and
interest rates holds for emerging market currencies, then it is going to be
cheaper to borrow US dollars and swap into local currency than to borrow
local currency. Importantly, however, this will be true not because the dol-
lar borrowing rate is low, but because domestic financial frictions make the
effective domestic-currency borrowing rate high. This idea is also consis-
covered foreign bor-
tent with other research on the prevalence of swap-
rowing in some emerging markets (for example, Munro and Wooldridge
2009). Clearly further research is needed, but one implication concerns
the transmission to emerging markets of changes in US monetary policy.
Imagine that the US raises interest rates: iUS goes up, and the emerging mar-
ket central bank raises its short term interest rate to match that. If Hyun’s
the swap basis rises when US monetary policy
empirical regularity holds—
then the basis gap will widen, making it relatively more attrac-
tightens—
tive to borrow US dollars and swap into domestic currency. In turn, this
widening has the effect of cushioning the impact of the domestic interest
rate rise on domestic financial conditions. Is this correct? If so, is it likely to
be important? In truth, I have no idea. But the possibility illustrates that in
this world where CIP does not hold, the process through which US dollar
particularly to emerg-
liquidity conditions are transmitted across borders—
is likely to be complex and subtle and involve transmission
ing markets—
mechanisms that we do not yet fully understand.
In sum, Shin’s chapter is very useful and thought provoking, one that
will surely help to encourage much future research.
References
Blanchard, Olivier, Jonathan D. Ostry, Atish R. Ghosh, and Marcos Chamon. 2016.
“Capital Flows: Expansionary or Contractionary?” American Economic Review, Papers
569.
and Proceedings 106 (5): 565–
252 Kunt and Maurice Obstfeld
Comments by Aslı Demirgüç-
Cerutti, Eugenio, Stijn Claessens, and Lev Ratnovski. 2017. “Global Liquidity and
Border Banking Flows.” Economic Policy 89 (32): 81–
Cross- 125.
Cerutti, Eugenio, Maurice Obstfeld, and Haonan Zhou. 2019. “Covered Interest Parity
Deviations: Macrofinancial Determinants.” IMF Working Paper 19–14, January.
Ivashina, Victoria, David S. Scharfstein, and Jeremy C. Stein. 2015. “Dollar Funding
and the Lending Behavior of Global Banks.” Quarterly Journal of Economics 130 (3):
1241–1281.
Keynes, John Maynard. 1923. A Tract on Monetary Reform. London: Macmillan.
Krugman, Paul, Maurice Obstfeld, and Marc Melitz. 2017. International Economics:
Theory and Policy, eleventh edition. New York: Pearson.
Munro, Anella, and Philip Wooldridge. 2009. “Motivations for Swap- Covered
Pacific Bond
Foreign Borrowing.” BIS Symposium on Internationalisation of Asia-
Markets, Paper One.
Tobin, James. 1981. “Money and Finance in the Macroeconomic Process.” Nobel
Memorial Lecture, December, Stockholm. http://www.nobelprize.org/nobel_prizes
/economic-sciences/laureates/1981/tobin-lecture.pdf.
6 Growth and Development from a
Schumpeterian Perspective
Philippe Aghion
Thirty years ago, Peter Howitt and I elaborated a new theory, now known
as the “Schumpeterian theory,” of economic growth. Why did we need a
new theory of economic growth? What did we find unsatisfactory with the
dominant theory at the time, both theoretically and empirically?
In this chapter, we shall revisit some current debates about the growth
and development process and about growth policy design, using the lenses
of the Schumpeterian growth paradigm.
Thus, in the first part of this chapter, I touch on four open questions
on which the Schumpeterian approach sheds new light: the relationship
led growth, the debate on secular
between competition and innovation-
stagnation, the recent rise in top income inequality, and firm dynamics.
In the second part of the chapter, I argue that the Schumpeterian growth
paradigm can be used to further bridge the existing gap between growth and
development economics.
And finally, in a third part, I will show how the paradigm can be used to
think about (or rethink) growth policy design.
Why Elaborate a New Theory of Economic Growth?
During my student years, the dominant paradigm in growth economics
was the neoclassical growth model, which would be taught first under the
assumption of a constant savings rate (the Solow model) and then in the
context of an economy where a representative consumer decides about
consumption, savings, and investment by maximizing her intertemporal
Cass-
utility (the Ramsey- Koopmans model).
The Solow model is the true template in growth economics, just as
Miller is the benchmark in corporate finance. This is first due
Modigliani-
254 Philippe Aghion
to it being a model of elegance and parsimony: The whole dynamics of
the economy is described in two equations. The second reason is that the
run growth without
model shows very clearly why there can be no long-
technical progress. The model was published in 1956 (I was born that same
year) and was rewarded by a Nobel Prize to its author in 1987.
No need to go into the details of this model, which economists all know
too well. But in a nutshell, the model describes an economy where final
output is produced using capital as input, and where therefore it is the accu-
mulation of capital that generates output growth. This corresponds to the
first equation of the model. Then the question is: Where does capital accu-
mulation come from? This in turn is answered with the second equation
of the model: from savings (aggregate savings equal aggregate investment
in equilibrium), and savings in the Solow model are a constant fraction of
final output (i.e., of aggregate GDP).
You might think that everything should go well in such an economy:
More capital stock financed by savings will produce more final output,
which will translate into more savings (as savings are proportional to final
output) and therefore in still more capital stock, and so on.
The problem is that we run into decreasing returns when trying to
increase output by increasing the capital stock: The higher the existing
stock of capital (number of machines) is, the lower will be the marginal
increase in output from increasing the stock of capital by one unit (i.e.,
from adding one more machine). Thus, the lower the increase in savings
and therefore the lower the induced increase in capital stock will be.
At some moment, the process of capital accumulation runs out of steam
(it stops when capital depreciation catches up with marginal savings), at
term
which point the economy stops growing. To generate sustained long-
economic growth, there must be continuous technical progress to increase
the quality (productivity) of machines. But Solow does not tell us where
technical progress is coming from.
In addition, if the model predicts conditional convergence, it does not
give us the tools to understand why the distribution of per capita income
has kept spreading out over time, why some countries converge to the stan-
dards of living (per capita GDP) of developed countries whereas other coun-
tries do not converge, or why some countries start converging and then
stop at midway. It does not explain why some countries with lower capital
stocks grow less rapidly than other countries with higher capital stocks,
A Schumpeterian Perspective 255
or why capital does not necessarily flow from rich to poor countries (the
so-called Lucas Paradox).
Moreover, the model does not look at growth from the point of view of
firms and entrepreneurs: How does growth relate to the size distribution of
firms, to the creation and destruction of firms and jobs, to firm dynamics
more generally? It does not provide keys to understand how institutions or
policies affect growth by affecting innovation and entrepreneurship.
These shortcomings motivated Peter Howitt and I to elaborate a new
paradigm.
The Schumpeterian Paradigm
The paradigm Howitt and I formalized in the fall of 1987 revolved around
three important ideas laid out by the Austrian economist Joseph Schumpeter.1
run growth is primarily generated by innovations (this
First idea: Long-
run growth
is the natural counterpart of Solow’s conclusion that no long-
can be expected without sustained technological progress).
Second idea: Innovations result from entrepreneurial investments (R&D,
training, computer purchase, and so forth), and entrepreneurs respond to
the economic incentives (positive or negative) that result from economic
based growth typi-
policies and economic institutions. Thus, innovation-
cally will be discouraged in environments with poor property right protec-
tion or with hyperinflation, as these conditions will damage the profitability
based growth is a social pro-
from innovation. In other words, innovation-
cess, and we can talk about policies of growth and institutions of growth.
Third idea: creative destruction. New innovations replace old technolo-
gies, and Schumpeterian growth is a conflictual process between the old
and the new: It tells the story of all these incumbent firms and interests that
permanently try to prevent or delay the entry of new competitors in their
sector. Hence there is something called “the political economy of growth.”
Thus, a first distinctive prediction of the Schumpeterian growth model is
that firm or job turnover should be positively correlated with productivity
growth. Another distinctive implication of the model is that innovation-
faire. Growth is excessive (resp.
led growth may be excessive under laissez-
1. See Aghion and Howitt (1992).
256 Philippe Aghion
faire when the business-
insufficient) under laissez- stealing effect associated
with creative destruction dominates (resp. is dominated by) the intertem-
poral knowledge spillovers from current to future innovators.
Four Growth Enigmas
In this section, I show how the Schumpeterian paradigm can be used to
shed light on four important enigmas associated with the growth process:
led growth, (2)
(1) the relationship between competition and innovation-
the debate on secular stagnation, (3) the dynamics of income inequality,
and (4) firm dynamics.
Led Growth
Competition and Innovation-
Our original model predicted that more competition should be detrimental
to growth by reducing monopoly rents from innovation and thus entre-
preneurs’ incentives to invest in innovation in the first place (incidentally,
this latter argument has been used by Bill Gates when facing antitrust
action). However, Blundell, Griffith, and Van Reenen (1995, 1999) used UK
level data to regress firm-
firm- level innovation intensity and/or produc-
tivity growth on the degree of product market competition in the firm’s
sector. And they found a positive correlation between competition and
innovation/growth.
How could we reconcile theory and evidence? Should we just dismiss the
Schumpeterian paradigm and start again from scratch? Should we simply
ignore the empirical evidence? I went for a third way: to look more closely
at the model and try to identify the assumption or assumptions that gener-
ate this counterfactual prediction of a negative relationship between com-
petition and growth.2
Having tried several alternative stories,3 we finally identified the main
culprit: In our initial model, only currently inactive firms innovate, not the
currently active firms (i.e., not the current technological leaders). Thus, an
innovation)
innovating firm in our model would move from zero profit (pre-
innovation). Then, not surprisingly, competition
to a positive profit (post-
2. See Aghion, Harris, and Vickers (1997) and Aghion et al. (2001).
3. For example, see Aghion, Dewatripont, and Rey (1999).
A Schumpeterian Perspective 257
innovation
would discourage innovation: Competition reduces the post-
profit, which here is equal to the net profit from innovation.
However, in practice we find at least two types of firms in most sectors of
the economy, and these two types of firms do not react in the same way to
increased competition. You first have what we call “frontier firms,” that is,
firms that are close to the current technological frontier in their sector. These
firms are currently active, and they make substantial profits even before
innovating this period. Second, you have what we call the “laggard firms,”
which are firms far below the current technological frontier. These firms
make low profits and try to catch up with the current technology frontier.
To try to understand why these two types of firms react differently to
competition, imagine for a moment that what you are looking at are not
firms but students in a classroom. And among them, you have the top stu-
dents and the bottom of the class. And suppose that you are opening the
class to an additional student, who turns out to be a very good student.
This is how I represent an increase in competition in this context. How will
the students react to this new student joining the classroom? The answer
(here I refer to important work by Caroline Hoxby, who studied precisely
this scenario) is that letting the new student in will encourage the other top
students to work harder to remain the best, whereas it will further discour-
age students at the bottom of the class, as they will find it even harder to
catch up.
Quite strikingly, firms react like classroom students: Faced with a higher
degree of competition in their sector, firms that are close to the technology
frontier will innovate more to escape competition, whereas firms that are
far from the technological frontier and try to catch up will be discouraged
by the higher degree of competition, and as a result innovate less: the latter
firms behave like those in the basic Schumpeterian model.
Overall, the effect of competition on innovation and productivity
growth is an inverted U, which synthetizes the positive escape competition
effect and the negative discouragement effect. The prediction of opposite
reactions of frontier versus nonfrontier firms to competition, and of an
inverted U overall, were tested and confirmed in joint work with Richard
Blundell, Nick Bloom, and Rachel Griffith (see Aghion et al. 2005) using the
level data as in the empirical studies I mentioned above.
same kind of firm-
To reconcile theory with evidence, we extended our basic Schumpet-
by-
erian model by allowing for step- step innovation in the Schumpeterian
258 Philippe Aghion
growth model.4 Namely, a firm that is currently behind the technological
leader in the same sector or industry must catch up with the leader before
by-
becoming a leader itself. This step- step assumption implies that firms in
and-
some sectors will be neck- neck. In turn, in such sectors, increased prod-
and-
uct market competition, by making life more difficult for neck- neck
firms, will encourage them to innovate to acquire a lead over their rival in
the sector. This we refer to as the “escape competition effect.” In contrast,
and-
in unleveled sectors where firms are not neck- neck, increased product
market competition will tend to discourage innovation by laggard firms,
run extra profit from catching up with the leader.
as it decreases the short-
state fraction
This we call the “Schumpeterian effect.” Finally, the steady-
and-
of neck- neck sectors will itself depend on the innovation intensities in
and-
neck- neck versus unleveled sectors. This we refer to as the “composi-
tion effect.”
This extended model predicts that in the aggregate, the relationship
U pattern.
between competition and innovation should follow an inverted-
Intuitively, when competition is low, innovation intensity is low in neck-
neck sectors; therefore most sectors in the economy are neck-
and- neck
and-
(the composition effect). But it is in precisely those sectors that the escape
competition effect dominates. Thus overall aggregate innovation increases
with competition at low levels of competition. When competition is high,
and-
innovation intensity is high in neck- neck sectors. Therefore most sec-
tors in the economy are unleveled sectors, so that the Schumpeterian effect
U prediction is confirmed by Aghion et al.
dominates overall. This inverted-
(2005), using panel data on UK firms.
The prediction that more intense competition enhances innovation
in frontier firms but may discourage it in nonfrontier firms was tested by
Aghion et al. (2009a), again using panel data on UK firms.
Another prediction from our extended model is that there is comple-
mentarity between patent protection and product market competition in
fostering innovation. Intuitively, competition reduces the profit flow of
innovating neck-
non- neck firms, whereas patent protection is likely
and-
and-
to enhance the profit flow of an innovating neck- neck firm. Both con-
and-
tribute to raising the net profit gain of an innovating neck- neck firm;
4. See Aghion, Harris, and Vickers (1997) and Aghion et al. (2001).
A Schumpeterian Perspective 259
in other words, both types of policies tend to enhance the escape competi-
tion effect.
That competition and patent protection should be complementary in
enhancing growth rather than mutually exclusive is at odds both with our
first model and with Romer (1990), where competition is always detrimental
to innovation and growth (as we discussed above) for exactly the same rea-
son that intellectual property rights in the form of patent protection are good
innovation rents, whereas
for innovation: Namely, competition reduces post-
patent protection increases these rents. But it is also at odds with Boldrin
and Levine (2008), who hold that patent protection is always detrimental to
innovation and growth in their model where competition is good for growth.
Our prediction of a complementarity between competition and patent
protection was tested by Aghion, Howitt, and Prantl (2013) using OECD
industry panel data.
country-
The Debate on Secular Stagnation
In 1938, economist Alvin Hansen explained in his presidential address
before the American Economic Association5 that in his opinion, the
United States faced inexorable weak growth in the long term. The nation
was just emerging from the Great Depression, and Hansen did not antici-
pate another world war that would stimulate a rebound in public spending
and thereby of aggregate demand.
Since then, we have experienced another major financial crisis, the
2008 crisis, which led Larry Summers (2013) and others to revive the
2007–
expression “secular stagnation” to characterize a situation that they assimi-
lated to the one described by Hansen in 1938. Summers’s argument is that
investment demand was so weak that negative interest rates were necessary
for a return to full employment.
Robert Gordon (2012), however, believes that the risk of secular stagna-
tion reflects a supply problem. Gordon proposes that the age of great inno-
hanging fruit
vations is past. He uses the metaphor of a fruit tree: The low-
is the best; after that, the fruit is harder to pick and less juicy.
Schumpeterian economists are more optimistic about the future
than Summers and Gordon are. A first argument (Jorgenson) is that the
5. See Hansen (1939).
260 Philippe Aghion
revolution in information and communications technologies (ICT) has rad-
producing technology; meanwhile, global-
ically and durably improved IT-
ization (which was concomitant with the ICT revolution) has substantially
the scale effect—
increased the potential returns on innovation— as well as
the competition effect. A second
the potential downside of not innovating—
argument against the secular stagnation view is that we have witnessed an
acceleration in innovation over the past several decades, which has not
been fully reflected by measured productivity growth.
In particular, Aghion et al. (2017) argue that innovation involving cre-
ative destruction is not properly taken into account by current measures of
factor productivity (TFP) growth. Whenever old products in the pro-
total-
ducer price index are replaced by new entrants, statistical offices typically
resort to imputation. For each product category in the economy, imputa-
adjusted price growth for a set of surviving
tion uses the rate of quality-
products in that category (i.e., products that were not subject to creative
destruction) to compute the inflation rate for the whole product category.
Using the Schumpeterian growth paradigm, together with the assump-
tion that the statistical office cannot observe the innovation coming from
adjusted
creative destruction and instead computes the aggregate quality-
price growth for the entire economy as being equal to the average price
growth over existing products that are not subject to creative destruction,
Aghion et al. (2017) provide an explicit expression for economywide miss-
ing growth from creative destruction. Then they use this expression to
quantify missing growth based on two different approaches. In the first
exercise, they use micro data from the US Census on the employment
shares of incumbents, entrants, and exiters in all nonfarm business sec-
tors. In the second exercise, they use data on the flow and quality of pat-
ents (exploiting information from patent citations) to directly estimate the
arrival rates and step sizes of the various kinds of innovations and from
there calculate the missing productivity growth from imputation. These
two exercises yield missing growth of comparable magnitudes, of about 0.5
percentage points on average per year over the past 30 years.
My third and last argument for optimism regarding future growth pros-
pects is also based on the observation that many countries have taken only
belated and incomplete advantage of technological advances (e.g., because
of structural rigidities or inappropriate economic policies).
A Schumpeterian Perspective 261
run technological waves, with
We do not question the existence of long-
their acceleration and slowdown phases. These waves are typically associ-
ated with the diffusion of new general purpose technologies, defined as
generic technologies that affect most sectors of the economy.6 Obvious
nineteenth century,
examples include steam energy in the early and mid-
electricity and chemistry in the early twentieth century, and the informa-
tion and communication technology revolution in the 1980s.
2012 on labor
And indeed, using annual and quarterly data for 1890–
productivity and TFP for 13 advanced countries (the G7 plus Spain, the
Netherlands, Finland, Australia, Sweden, and Norway) plus the reconsti-
tuted euro area, Bergeaud, Cette, and Lecat (2014) show the existence of two
big productivity growth waves during this period. The first wave culminates
in 1941, the second culminates in 2001. The first wave corresponds to the
second industrial revolution: that of electricity, internal combustion, and
chemistry. The second wave is the ICT wave.
However, Cette and Lopez (2012) show that the euro area and Japan
experienced the waves with a lag compared to the United States. Thus the
first wave fully diffused to the current euro area, Japan, and the United
World War II. As for the second productivity wave, so
Kingdom only post–
far it has not shown up in the Euro area or in Japan. Moreover, through
an econometric analysis, Cette and Lopez show that this lag of ICT diffu-
sion in Europe and Japan, compared to the United States, is explained by
institutional aspects: a lower educational level, on average, of the working-
age population and more regulations on labor and product markets. This
in turn suggests that by implementing structural reforms, these countries
up to the
could benefit from a productivity acceleration linked to a catch-
US ICT diffusion level. The lower quality of research and higher education
in the euro area and Japan compared to the United States also appears to
matter for explaining the diffusion lag.
One can contrast the evolution of TFP in Sweden versus Japan over the
past decades. In particular, there has been a positive break in TFP growth in
Sweden after 1990, in contrast with the case of Japan, where we see no such
break but instead decelerating TFP growth since 1980. Our explanation is
that Sweden implemented sweeping structural reforms in the early 1990s:
6. See Bresnahan and Trajtenberg (1995).
262 Philippe Aghion
in particular, a reform of the public spending system to reduce public defi-
cits and a tax reform to encourage labor supply and entrepreneurship. No
significant reform has taken place in Japan over the past 30 years.
To conclude this discussion on secular stagnation, although we do not
run technological waves, what leads us to be
question the existence of long-
somewhat more optimistic than Gordon is that (1) the ICT revolution has
improved the technology to produce ideas, and globalization has increased
the potential rents to successful innovators; (2) measured TFP growth does
not properly take into account innovation involving creative destruction;
and (3) some developed countries, particularly in Europe, have not yet
implemented the structural reforms that would allow them to fully take
advantage of the most recent wave of innovation.
Innovation, Inequality, and Social Mobility
Over recent decades, developed nations have experienced an accelerated
increase in income inequality, especially at the top tier, with the top 1 per-
cent capturing a rapidly growing share of total income.7 What explains
this evolution?
Figure 6.1 compares the evolution of innovation in the United States
since 1960 (as measured by the number of patents registered annually with
the United States Patent and Trademark Office), with extreme inequality (as
measured by the share of income attributed to the top 1 percent of earners).
The similarity in the two curves (innovation and the top 1 percent’s share
of income) is striking.
A new study by Antonin Bergeaud, Richard Blundell, Ufuk Akcigit, David
Hemous, and myself8 shows that this strong correlation reflects a causal
link between innovation and extreme inequality: Income from innovation
contributes significantly to the increase in the share of income going to the
top 1 percent.
The observation that the observed increase in the top 1 percent results
in part from innovation, and not solely from returns from real estate and
speculation, provides an important insight, because innovation has virtues
that the other sources of high income do not necessarily share.
7. See Atkinson, Piketty, and Saez (2011) and Piketty (2013).
8. See Aghion et al. (2015a).
A Schumpeterian Perspective 263
Figure 6.1
Evolution of top income share and patents per capita in the United States
Source: Aghion et al. (2015b).
First, as previously mentioned, innovation is the main motor of growth
in developed economies. Second, although in the short term innovation
benefits those who generated or enabled the innovation, in the long term
its returns are dissipated due to imitation and creative destruction. In other
words, the inequality induced by innovation is temporary. Third, because
of the link between innovation and creative destruction, innovation gener-
ates social mobility: It enables new talent to enter the market and to displace
(partially or totally) the firms in place. Thus in the United States, California
(currently the most innovative state in the union) far outpaces Alabama
(which is among the least innovating states) both in terms of the inequality
of income going to the top 1 percent and in terms of social mobility.
The two figures below are especially eloquent. Figure 6.2 describes the
relationship between innovation and social mobility by comparing Amer-
ican municipalities. Social mobility is defined as the probability that an
individual from a modest background (i.e., one whose parents were in the
lowest quintile in the earnings scale between 1996 and 2000) will reach
the highest quintile in 2010 on reaching adulthood (based on the work
of Chetty et al. (2014)). Innovation is measured by the number of patents
filed with the United States Patent and Trademark Office per resident in
the municipality. The resulting graph shows a strong positive correlation
between innovation and social mobility.
264 Philippe Aghion
Figure 6.2
Relationship between innovation and social mobility across municipalities in the
United States
Source: Aghion et al. (2015b).
Figure 6.3
No correlation between innovation and the Gini measure of inequality
Source: Aghion et al. (2015b).
Figure 6.3 shows that there is no correlation between innovation and the
broader measures of inequality, such as the Gini coefficient, which mea-
sures the deviation between the actual distribution of income in an econ-
omy and a perfectly equal distribution.
By taking into account all pieces of the puzzle, we can respond to the
question of whether we should object to innovation on the grounds that it
A Schumpeterian Perspective 265
contributes to income inequality. The response is no, because innovation
generates growth. It does not increase inequality in broader terms; instead,
it stimulates social mobility. As a corollary to this discussion, tax policy
must differentiate between innovation and other sources of top income.
Put differently, we must distinguish between a Steve Jobs and a Carlos Slim.
Tax policy that discourages innovation would not only inhibit growth but
also reduce social mobility, whereas innovation does not increase inequal-
ity measured broadly.
Firm Dynamics and Economic Development
The empirical literature has documented various stylized facts about firm
level data. In particu-
size distribution and firm dynamics using micro firm-
lar: (1) the firm size distribution is highly skewed; (2) firm size and firm age
are highly correlated; and (3) small firms exit more frequently, but the ones
that survive tend to grow faster than the average growth rate.
Schumpeterian growth models cannot
These are all facts that non-
account for. In particular, the first four facts listed require a new firm to
enter, expand, then shrink over time, and eventually be replaced by new
entrants: These and the last fact on the importance of reallocation are all
embodied in the Schumpeterian idea of creative destruction.
The Schumpeterian model by Klette and Kortum (2004) can account
for these facts. This model adds two elements to the baseline model: First,
innovations come from both entrants and incumbents; and second, firms
are defined as a collection of production units where successful innovations
by incumbents will allow them to expand in product space (see figure 6.4).
This model allows us to explain the above stylized facts:
Prediction 1: The size distribution of firms is highly skewed.
Recall that in this model, firm size is summarized by the number of prod-
uct lines of a firm. Hence, to become large, a firm needs to have succeeded
in many of its attempts to innovate in new lines and at the same to have
survived many attempts by potential entrants and other incumbents at tak-
ing over its existing lines. This is turn explains why there are so few very
state equilibrium (i.e., why firm size distribution is
large firms in steady-
highly skewed), as shown in a vast empirical literature.
Prediction 2: Firm size and firm age are positively correlated.
In the model, firms are born with a size of 1. Subsequent successes are
required for firms to grow in size, which naturally produces a positive
266 Philippe Aghion
correlation between size and age. This regularity has been documented
extensively in the literature.
Prediction 3: Small firms exit more frequently. The ones that survive tend
to grow faster than average.
In the above model, it takes only one successful entry to make a one-
product firm exit, whereas it takes two successful innovations by potential
product firm exit. The facts that small firms exit
entrants to make a two-
more frequently and grow faster conditional on survival have been widely
documented in the literature.
Various versions of this framework have been estimated using micro-
level data by Lentz and Mortensen (2008), Acemoglu et al. (2013), and
Akcigit and Kerr (2010).9
In more recent work, Acemoglu et al. (2013) analyze the effects of vari-
ous industrial policies on equilibrium productivity growth, including entry
subsidy and incumbent R&D subsidy, in an enriched version of the above
framework. Their extended framework also sheds new light on whether or
how one should conduct industrial policy. In particular, allowing for high-
ability innovators, they argue that subsidizing incumbent firms
and low-
has a detrimental effect on aggregate innovation and productivity growth
ability) incumbents at the expense of
by inducing a bias in favor of (low-
high-ability entrants.
Growth Meets Development
Michael Kremer, Abhijit Banerjee, and Esther Duflo have revolutionized
development economics by introducing experimental random methods of
analysis drawn from pharmaceutical science to evaluate the effectiveness of
new medicines and vaccines.10 In particular, their work has enabled us to
understand better the behavior of individuals and households in extreme
poverty and to see how they react to different policies of aid and assistance.
However, this line of research suffers from two main limitations. First,
firms and firm dynamics play little role in these analyses of the develop-
ment process. Second, the link between micro and macro development is
9. See Aghion, Akcigit, and Howitt (2014) and Akcigit and Kerr (2010) for more
references.
10. See Banerjee and Duflo (2012).
A Schumpeterian Perspective 267
not fully spelled out. However, my own view is that one cannot disregard
macroeconomic and systemic factors, or the effects of firm dynamics and
resource reallocation, when the goal is to eradicate poverty at a national or
regional level.
To see why macroeconomics matters, consider the following example.
The rate of poverty in urban zones of India (the fraction of the population
1988 to 12
living on less than $1 per day) fell from 39 percent in 1987–
2000. Over the same period, growth took off: From less
percent in 1999–
1980s, it climbed to 3.2 percent in the 1990s.
than 0.8 percent in the mid-
This upswing in growth in India resulted less from local actions than from
systemic reforms, such as the liberalization of trade and of the market for
goods and services, with the suppression of the “raj license.”11
But looking at the systemic and macroeconomic aspects of a problem
by no means implies that we should ignore the microeconomic aspects, in
particular, at the level of the firm or sector. Specifically, our discussion of
growth enigmas in the previous section has implications for how Schum-
peterian growth theory can help bridge the gap between growth and devel-
enhancing
opment economics: first, by capturing the idea that growth-
policies or institutions vary with a country’s level of technological develop-
ment; and second, by analyzing how institutional development (or the lack
of it) affects firm size distribution and firm dynamics.
Appropriate Institutions and the Transition Trap
In 1890, Argentina enjoyed a GDP per capita approximately 40 percent that
income country. This level
of the United States, which made it a middle-
was three times the GDP per capita of Brazil and Colombia and equivalent
to that of Japan at the time. Argentina sustained this level of 40 percent
of the GDP per capita of the United States through the 1930s. To be pre-
cise, Chow’s test (a statistical test) shows a break around 1938, after which
Argentina’s productivity declines relative to American productivity by
off?
approximately 21 percent per year. What explains this drop-
Schumpeterian growth theory offers the following explanation. Coun-
tries like Argentina either had institutions or had implemented policies
(in particular, import substitution) that fostered growth by accumulation
up. They did not, however, adapt their
of capital and economic catch-
11. See Aghion et al. (2008).
268 Philippe Aghion
institutions to enable them to become innovating economies. As demon-
strated in joint work with Daron Acemoglu and Fabrizio Zilibotti,12 the
greater the level of development is in a country (i.e., the closer it gets to
the technology frontier), the greater the role of cutting edge innovation
becomes as the motor of growth, replacing accumulation and technologi-
cal catch-up.
This phenomenon also exists in Asia. Japan, where the state has always
tightly controlled competition, is another example: Japan’s Ministry of
Economy, Trade and Industry caps the number of import permits, and the
financial consortia known
state subsidizes investment by the big industrial-
as keiretsu. It is thus not surprising that from an extremely high level
the envy of other developed countries—
between 1945 and 1985— Japan’s
growth has fallen to a very low level since 1985.
In the previous subsection, I discussed the prediction that competi-
tion and free entry should be more growth enhancing in more frontier
firms, which implies that they should be more growth enhancing in more
advanced countries, because such countries have a larger proportion of fron-
country panel of more than 100 countries
tier firms. Similarly, using a cross-
2000 period, Acemoglu, Aghion, and Zilibotti (2006) test the
over the 1960–
following predictions from the Schumpeterian prediction between imita-
driven growth:
tion and innovation-
Prediction 1: Average growth should decrease more rapidly as a country
approaches the world frontier when openness is low.
Acemoglu, Aghion, and Zilibotti (2006) repeat the same exercise using
entry costs faced by new firms instead of openness. They show:
Prediction 2: High entry barriers become increasingly detrimental to
growth as the country approaches the frontier.
These two empirical exercises point to the importance of interacting
institutions or policies with technological variables in growth regressions:
Openness is particularly growth enhancing in countries that are closer to
the technological frontier; entry is more growth enhancing in countries or
sectors that are closer to the technological frontier.
Next, to the extent that frontier innovation makes greater use of research
education than imitation, the prediction is:
12. See Acemoglu, Aghion, and Zilibotti (2006).
A Schumpeterian Perspective 269
Prediction 3: The more frontier an economy is, the more growth in this
economy will rely on research education.
type education is
And indeed, Aghion et al. (2009b) show that research-
always more growth enhancing in US states that are more frontier, whereas
year colleges is more growth-
a bigger emphasis on 2- enhancing in US states
that are farther below the productivity frontier. Similarly, using cross-
country panel data, Vandenbussche, Aghion, and Meghir (2006) show that
tertiary education is more positively correlated with productivity growth in
countries that are closer to the world technology frontier.
In the same spirit, one can look at the relationship between technologi-
cal development, democracy, and growth. An important channel is Schum-
peterian, namely, democracy reduces the scope for expropriating successful
innovators or for incumbents to prevent new entry by using political pres-
sure or bribes. In other words, democracy facilitates creative destruction
and thereby encourages innovation.13
To the extent that innovation matters more for growth in more frontier
economies, the prediction is:
Prediction 4: The correlation between democracy and innovation/growth
is more positive and significant in economies that are closer to the frontier.
This prediction is confirmed by Aghion, Alesina, and Trebbi (2007) using
employment and productivity data at industry level across countries and
over time.
Innovation, Institutions, and Firm Dynamics in Developing Countries
Tai Hsieh and Peter Klenow
The two figures below, from the work of Chang-
(2009), illustrate the importance of firm dynamics and firm size distribu-
tion in the process of economic development. Figure 6.4 compares the dis-
tribution of Indian firms by productivity with that of American firms. Note
that many more firms have low productivity in India than in the United
States. Figure 6.5 represents the evolution of the average size of a company
as a function of its age in India, Mexico, and the United States. It shows
that US firms continue to grow, whereas the growth of Indian firms drops
13. Acemoglu and Robinson (2006) formalize another reason, also Schumpeterian,
as to why democracy matters for innovation, namely, new innovations do not only
destroy the economic rents of incumbent producers, they also threaten the power of
incumbent political leaders.
270 Philippe Aghion
.3 INDIA
.2
.1
0
1/256 1/64 1/16 1/4 1 4
.3
USA
.2
.1
0
1/256 1/64 1/16 1/4 1 4
Figure 6.4
Distribution of firm productivity, India and the United States
Source: Hsieh and Klenow (2009).
8
Standardized number of employees
4
USA
MEXICO
2
INDIA
1
1/2
<5 5–9 10–14 15–19 20–24 25–29 30–34 35–39 >=40
Firm age
Figure 6.5
Link between the age and size of firms
Source: Hsieh and Klenow (2009).
A Schumpeterian Perspective 271
off. In fact, Hsieh and Klenow show that although US establishments grow
five times relative to their entry size by the age of 30, Indian counterparts
barely show any growth.
Both these figures look at microeconomic characteristics. Yet when
placed side by side, they tell a story that has consequences for the Indian
economy as a whole: The inability of Indian firms, even the most innova-
tive and productive ones, to grow beyond a certain size enables firms with
low productivity to survive. But in the aggregate, innovation, and thereby
the growth of the Indian economy overall, suffers.
To explain these two figures, we must consider the systemic character-
istics of the Indian economy. Why do establishments not grow in India?
Bloom et al. (2013) show that lack of trust and the weak rule of law are
major obstacles to firm growth.
Kortum
More recently, Akcigit, Alp, and Peters (2014) extend the Klette-
model of firm dynamics discussed in the previous section by adding two
major ingredients: (1) production requires managers, as owners’ time is lim-
ited, and therefore owners face an overload constraint; (2) firm owners can
ability owners are more creative and
be of high or low ability, where high-
ability
therefore have the potential to expand much faster than can low-
owners (but this potential for expansion materializes more when the scope
for delegation is higher).
Their model generates the following predictions:
Prediction 1: The expected number of outside managers is (1) increasing
in firm size and (2) increasing in the rule of law.
Larger firms involve a higher degree of overload for firm owners, which
in turn increases the returns from hiring outside managers. Finally, stronger
rule of law implies higher net return to delegation. Akcigit, Alp, and Peters
(2014) provide empirical support for these predictions using Indian manu-
facturing establishments.
Prediction 2: The average firm size increases with the rule of law.
Firm value is increasing in owner time, and therefore the firms are will-
ing to innovate and expand more when firm value is higher. The empirical
support for this prediction is provided by Bloom et al. (2013). The positive
link between firm size and the rule of law has been extensively documented
in the literature (see, for instance, Bloom, Sadun, and Van Reenen (2012)
for a detailed discussion). Finally, Akcigit, Alp, and Peters (2014) show that
272 Philippe Aghion
trust regions in
the link between firm size and family size is weaker in high-
India.
Prediction 3: Firm growth decreases with firm size, and the more so the
weaker the rule of law.
Indeed in larger firms, the span of control is larger, and therefore the
owner has less time to allocate to each product line. This in turn implies
that any constraint limiting the scope for delegation will have more dra-
matic effects on large firms. In particular, the weaker the rule of law is, the
lower the larger firms’ incentive to grow will be, which in turn implies that
the difference in growth incentives between large and small firms will be
higher in countries with weaker rule of law. Akcigit, Alp, and Peters (2014)
trust regions in India.
show that growth decreases faster in firm size in low-
Prediction 4: Everything else being equal, creative destruction and real-
location among firms will be higher in economies where the rule of law is
stronger.
Clearly this last prediction is in line with the main findings of Hsieh
and Klenow’s work, which showed the missing growth and reallocation in
developing countries. Understanding the reasons behind the lack of reallo-
cation and creative destruction is essential when designing the right devel-
opment policies. The Schumpeterian growth framework provides a useful
framework to conduct counterfactual policy exercises, which can shed light
on this important debate.
I see this approach as potentially quite fruitful. For example, one could
look at the extent to which characteristics (such as the quality of education,
infrastructure, or labor market regulations) also affect firm dynamics and
the ability of better performing firms to grow faster. More generally, a bet-
ter understanding of the process of growth of firms and the reallocation of
resources among firms or sectors would undoubtedly provide new keys to
understand the relationship between growth and development and to find
lasting remedies for underdevelopment and poverty in the world.
Rethinking Growth Policy
Economists have responded in different ways to the question of whether to
get involved in economic policy debates or to stay out of the debates and
concentrate on basic research. My work lies between these two attitudes.
A Schumpeterian Perspective 273
Although I am first and foremost a researcher and a teacher, I find economic
policy debates compelling for two reasons. First, as a strictly scientific mat-
ter, analyzing public policy and action enables us to better understand the
mechanisms of growth. Second, theoretical and empirical economic analy-
sis combats “false good ideas” by clarifying the terms of the policy debate,
and it helps suggest guidelines for growth policy design.
The Growth Diagnostics Approach
In an influential paper titled “Growth Diagnostics,” Hausmann, Rodrik,
and Velasco (2005), henceforth HRV, have proposed an attractively simple
enhancing policy. In this section, I first
methodology to design growth-
summarize the methodology, point out some of its potential limitations,
and then propose an alternative approach based on growth regressions that
are themselves suggested by the theory, particularly the Schumpeterian
paradigm outlined above.
enhancing policies
HRV start from the relevant observation that growth-
should vary from one country or region to another. For example, growth in
the United States and other industrialized countries over the past 10 years
appears to have benefited from market deregulations and privatizations.
However, in Asian countries (including China) high growth rates have been
promoted under limited competition or limited privatizations. The next
question then is: Can one use existing new growth theory to provide a flex-
country vari-
ible guide to growth policy making, one that fully takes cross-
ability into account? HRV provide a positive and attractively simple answer
to this question: namely, to use price comparisons to infer the importance
of each potential constraint to growth. To illustrate their methodology, HRV
consider a few Latin American examples, including Brazil and El Salvador.
In Brazil, returns to capital are high (with a net interest margin equal
to 11.5 in 2001). This leads HRV to point to the low level of local savings
(with very negative public savings) and the high tax rates as the main con-
straints on growth (the importance of the former is further supported by
the positive and significant correlation between the interest rate and the
current account deficit over time). The rate of return on education is also
high in Brazil, which suggests that the rate of return on capital, and thereby
growth, could be further increased by investing more on education. How-
ever, the argument goes, the already high rate of return on capital suggests
that investing in education may not be a priority in Brazil.
274 Philippe Aghion
In El Salvador, interest rates are low (a net interest margin equal to 3.7 in
2001), but so is the tax rate on capital. Is the lack of education responsible
for the rate of return on capital? The HRV answer is no, given that the rate
of return on education in El Salvador is low. Nor is there a lack of contrac-
tual enforcement that would reduce profitability. Lack of savings cannot be
the binding constraint either, otherwise the interest margin would be high.
Having failed to identify true obstacles to growth in El Salvador, HRV men-
tion the “absence of profitable investment opportunities” as yet another
potential suspect to consider.
Now suppose we used the same growth diagnostic approach to deal with
the slow EU growth problem. The return to education is lower in the Euro-
pean Union than it is in the United States, which HRV would interpret
as an indication that education is the most binding constraint to growth.
Instead, they would presumably point to the high European tax rates as the
main suspect, and thereby advocate lower tax rates as the primary cure to
the growth problem in the European Union.
The simple and ingenious approach proposed by HRV raises at least
two concerns. First, equilibrium prices do not necessarily reflect a con-
straint on growth. Consider interest rates. A low interest rate does not
mean that the local credit market is not constrained. In fact, low interest
rates may reflect a high degree of credit rationing, as shown by Aghion
and Bolton (1997). Indeed, the more restricted the access to credit is (that
is, the more individuals are barred from undertaking their own projects),
the more supply of loanable funds there will be in the economy, as all
rationed individuals will end up lending to a few entrepreneurs.
credit-
But this in turn should result in a lower domestic equilibrium interest
rate. Next, consider the rates of return on labor, which are measured by
called Mincerian wages, that is, by the forgone wage income of 1
the so-
more year in education at different levels of education. Mincerian wages
of course provide some useful indication on the marginal value of private
investments in education in different fields and at different levels of edu-
cation. However, a big shortcoming of the Mincerian approach is that the
Mincerian wage does not account for externalities. In particular, it does
not account for the intertemporal knowledge externalities that lie behind
the positive relationship between education and growth. That intertem-
poral externalities matter is evidenced by the large effects of education on
growth.
A Schumpeterian Perspective 275
More generally, current prices reflect a current state of the economy.
They do not inform directly about the growth dynamics that would result
for various types of policies.
A second concern with the HRV approach is that it cannot lead to
growth prescriptions that would affect simultaneously the demand side
and the supply side of markets. Thus, for example, HRV would never rec-
ommend that a country invest in education (thereby increasing the supply
of research labor) and at the same time invest in structural reforms that
increase the profitability of innovations (thereby fostering the demand for
R&D labor by firms).14
An alternative to the above methodology is to use theory to construct
growth regressions that are meant to inform us directly about the impact of
different institutions or policies on growth.
Led Growth
Pillars of Innovation-
led growth and thereby avoid the middle-
To enhance innovation- income
trap, the Schumpeterian paradigm and our discussion in the previous two
sections suggest policy priorities such as:
1. Liberalize entry and increase competition among existing firms. This
policy favors creative destruction and also encourages incumbent firms
to innovate to escape competition from their rivals.
2. Liberalize labor markets to make it easier for labor to reallocate from
old to new activities. This policy in turn requires active labor policies
that combine unemployment support with retraining programs. This
approach is quite intuitive: The more advanced a country is, the more
productivity growth will rely on frontier innovation. But frontier inno-
vation in turn entails more creative destruction, and thus more job turn-
up.
over, than does technological catch-
funded and autonomous universities to promote frontier
3. Invest in well-
research and innovation-led growth. Indeed, frontier innovation requires
frontier researchers and therefore good universities and research centers,
whereas good undergraduate education is sufficient for imitation.
14. Incidentally, HRV would never recommend more active competition policies
whose effect in the simple growth paradigm they consider is simply to reduce the
rate of return on capital.
276 Philippe Aghion
based financial system enhances productivity growth more for
4. If a bank-
based financial system enhances
less advanced countries, a more market-
productivity growth more in more frontier countries where growth is
driven by frontier innovation. Intuitively, frontier innovation, which
breaks new ground, entails a higher level of risk than imitation activi-
ties, which are already well defined. But this in turn implies that outside
financiers involved in frontier innovation will ask for a higher share
of upside revenues and also for higher control rights: hence the role of
equity in financing frontier innovation.
To enhance productivity growth based on imitation or adaptation in
less developed (catching up) countries, the examples of China, India, or
the Asian Tigers suggest that reallocation and technology transfers are key.
These properties in turn appear to benefit from good basic education sys-
access to infrastructure, access to
tems and from institutional features—
that favor factor mobility
(bank) finance, and labor market flexibility—
and the creation and growth of new business activities. Thus Aghion et al.
(2008) showed that the delicensing reforms in India spurred productiv-
ity growth particularly in provinces with higher degrees of labor market
flexibility.
Conclusion
In this chapter, we have seen how Schumpeterian growth theory can shed
light on key growth enigmas: in particular, the relationship between compe-
led growth; the existence of transition traps; secular
tition and innovation-
stagnation; the relationship between growth and inequality; and the rela-
tionship between growth and firm dynamics. We also discussed how growth
theory can guide growth policy design. Finally, I argued that the theory can
further contribute to reconciling growth with development economics: first,
by bringing out the notion of appropriate growth institutions and policies;
and second, by looking at how institutional development shapes the rela-
tionship among firm size distribution, reallocation, and growth.
Numerous paths have yet to be explored to better understand the enig-
mas of growth, the relationship between growth and innovation, and the
role of institutions and economic policy in the process of development.
Understanding this process will benefit not only science but also society as
a whole, because we are less fearful of what we understand.
A Schumpeterian Perspective 277
References
Acemoglu, Daron, and James A. Robinson. 2006. “Economic Backwardness in Politi-
131.
cal Perspective.” American Political Science Review 100 (1): 115–
Acemoglu, Daron, Philippe Aghion, and Fabrizio Zilibotti. 2006. “Distance to Fron-
tier, Selection, and Economic Growth.” Journal of the European Economic Association
4 (1): 37–74.
Acemoglu, Daron, Ufuk Akcigit, Nicholas Bloom, and William R. Kerr. 2013. “Inno-
vation, Reallocation and Growth.” NBER Working Paper 18993, National Bureau of
Economic Research, Cambridge, MA.
Down Growth and
Aghion, Philippe, and Patrick Bolton. 1997. “A Theory of Trickle-
172.
Development.” Review of Economic Studies 64 (2): 151–
Aghion, Philippe, and Peter Howitt. 1992. “A Model of Growth Through Creative
351.
Destruction.” Econometrica 60 (2): 323–
Aghion, Philippe, Ufuk Akcigit, and Peter Howitt. 2014. “What Do We Learn from
Schumpeterian Growth Theory?” In Handbook of Economic Growth, volume 2, edited
563. Amsterdam: Elsevier.
by Philippe Aghion and Steven N. Durlauf, 515–
Aghion, Philippe, Alberto Alesina, and Francesco Trebbi. 2007. “Democracy, Tech-
nology, and Growth.” NBER Working Paper 13180, National Bureau of Economic
Research, Cambridge, MA.
Aghion, Phillipe, Mathias Dewatripont, and Patrick Rey. 1999. “Competition, Finan-
cial Discipline, and Growth.” Review of Economics and Statistics 66(4): 825–852.
Aghion, Philippe, Christopher Harris, and John Vickers. 1997. “Competition and
Growth with Step- Step Innovation: An Example.” European Economic Review 41
by-
(3–5): 771–782.
Aghion, Philippe, Peter Howitt, and Susanne Prantl. 2013. “Patent Rights, Product
Market Reforms, and Innovation.” NBER Working Paper 18854, National Bureau of
Economic Research, Cambridge, MA.
Aghion, Philippe, Robin Burgess, Stephen J. Redding, and Fabrizio Zilibotti. 2008.
“The Unequal Effects of Liberalization: Evidence from Dismantling the Raj License
1412.
in India.” American Economic Review 98 (4): 1397–
Aghion, Philippe, Christopher Harris, Peter Howitt, and John Vickers. 2001. “Com-
petition, Imitation and Growth with Step- Step Innovation.” Review of Economic
by-
492.
Studies 68 (3): 467–
Aghion, Philippe, Nick Bloom, Richard Blundell, Rachel Griffith, and Peter Howitt.
2005. “Competition and Innovation: An Inverted- U Relationship.” Quarterly Journal
728.
of Economics 120 (2): 701–
278 Philippe Aghion
Aghion, Philippe, Richard Blundell, Rachel Griffith, Peter Howitt, and Susanne
Prantl. 2009a. “The Effects of Entry on Incumbent Innovation and Productivity.”
32.
Review of Economics and Statistics 91 (1): 20–
Aghion, Philippe, Leah Boustan, Caroline Hoxby, and Jerome Vandenbussche.
2009b. “The Causal Impact of Education on Economic Growth: Evidence from the
U.S.” Unpublished manuscript, Harvard University, Cambridge, MA.
Aghion, Philippe, Ufuk Akcigit, Antonin Bergeaud, Richard Blundell, and David
Hemous. 2015a. “Innovation and Top Income Inequality.” NBER Working Paper
21247, National Bureau of Economic Research, Cambridge, MA.
Aghion, Philippe, Ufuk Akcigit, Antonin Bergeaud, Richard Blundell, and David
Hemous. 2015b. “Innovation, Income Inequality, and Social Mobility.” https://
voxeu.org/article/innovation-income-inequality-and-social-mobility.
Aghion, Philippe, Antonin Bergeaud, Timo Boppart, Peter J. Klenow, and Huiyu Li.
2017. “Missing Growth from Creative Destruction.” NBER Working Paper 24023,
National Bureau of Economic Research, Cambridge, MA.
Akcigit, Ufuk, and William R. Kerr. 2010. “Growth Through Heterogeneous Innova-
tions.” NBER Working Paper 16443, National Bureau of Economic Research, Cam-
bridge, MA.
Akcigit, Ufuk, Harun Alp, and Michael Peters. 2014. “Lack of Selection and Limits
to Delegation: Firm Dynamics in Developing Countries.” Unpublished manuscript,
University of Pennsylvania, Philadelphia.
Atkinson, Anthony B., Thomas Piketty, and Emmanuel Saez. 2011. “Top Incomes in
Run History.” Journal of Economic Literature 49 (1): 3–
the Long- 71.
Banerjee, Abhijit V., and Esther Duflo. 2012. Repenser la Pauvreté. Paris: Seuil.
Bergeaud, Antonin, Gilbert Cette, and Remy Lecat. 2014. “Productivity Trends from
1890 to 2012 in Advanced Countries.” Working Paper 475, Banque de France, Paris.
Bloom, Nicholas, Raffaella Sadun, and John Van Reenen. 2012. “The Organization
1705.
of Firms across Countries.” Quarterly Journal of Economics 127 (4): 1663–
Bloom, Nicholas, Benn Eifert, Aprajit Mahajan, David McKenzie, and John Roberts.
2013. “Does Management Matter? Evidence from India.” Quarterly Journal of Econom-
51.
ics 128 (1): 1–
Blundell, Richard, Rachel Griffith, and John Van Reenen. 1995. “Dynamic Count
344.
Data Models of Technological Innovation.” Economic Journal 105 (429): 333–
Blundell, Richard, Rachel Griffith, and John Van Reenen. 1999. “Market Share,
Market Value and Innovation in a Panel of British Manufacturing Firms.” Review of
554.
Economic Studies 66 (3): 529–
A Schumpeterian Perspective 279
Boldrin, Michele, and David K. Levine. 2008. Against Intellectual Monopoly. Cam-
bridge: Cambridge University Press.
Bresnahan, Timothy F., and M. Trajtenberg. 2005. “General Purpose Technologies:
108.
‘Engines of Growth’?” Journal of Econometrics 65 (1): 83–
Cette, Gilbert, and Jimmy Lopez. 2012. “ICT Demand Behavior: An International
410.
Comparison.” Economics of Innovation and New Technology 21 (4): 397–
Chetty, Raj, Nathaniel Hendren, Patrick Kline, and Emmanuel Saez. 2014. “Where
Is the Land of Opportunity? The Geography of Intergenerational Mobility in the
1623.
United States.” Quarterly Journal of Economics 129 (4): 1553–
Gordon, Robert J. 2012. “Is U.S. Economic Growth Over? Faltering Innovation
Confronts the Six Headwinds.” NBER Working Paper 18315, National Bureau of
Economic Research, Cambridge, MA.
Hansen, Alvin H. 1939. “Economic Progress and Declining Population Growth.”
15.
American Economic Review 29 (1): 1–
Hausmann, Ricardo, Dani Rodrik, and Andrés Velasco. 2005. “Growth Diagnostics.”
Working Paper, John F. Kennedy School of Government, Harvard University, Cam-
bridge, MA.
Hsieh, Chang-Tai, and Peter J. Klenow. 2009. “Misallocation and Manufacturing TFP
1448.
in China and India.” Quarterly Journal of Economics 124 (4): 1403–
Klette, Tor Jakob, and Samuel Kortum. 2004. “Innovating Firms and Aggregate Inno-
1018.
vation. Journal of Political Economy 112 (5): 986–
Lentz, Rasmus, and Dale T. Mortensen. 2008. “An Empirical Model of Growth
1373.
Through Product Innovation.” Econometrica 76 (6): 1317–
Piketty, Thomas. 2013. Le Capital au XXIe Siècle. Paris: Seuil.
Romer, Paul M. 1990. “Endogenous Technical Change.” Journal of Political Economy
S102.
98 (5): S71–
Summers, Lawrence H. 2013. “Why Stagnation Might Prove to Be the New Normal.”
Financial Times, December 15.
Vandenbussche, Jérôme, Philippe Aghion, and Costas Meghir. 2006. “Growth, Dis-
tance to Frontier, and Composition of Human Capital.” Journal of Economic Growth
127.
11 (2): 97–
Comment: Francesco Caselli
Philippe Aghion’s paper is a wonderful introduction to some of his contribu-
tions in the area of growth theory and empirics over the past 30 years. Every-
body knows, of course, that Philippe is a major figure in this field. But even
those who have followed his work fairly closely cannot fail to be inspired
anew by the ambition, cohesiveness, and ultimate success of his efforts as
they emerge from the account he gives in the preceding pages. Indeed, the
feeling of awe is only magnified by the knowledge that there are many other
important contributions that Philippe has chosen not to discuss here.
The chapter lays out an understanding of the growth process that encom-
passes a broad set of phenomena, including the role of competition in foster-
ing (or, in some cases, discouraging) innovation; the need for institutions to
evolve to keep the growth process going; the implications of innovation and
growth for income inequality and for the evolution of the size distribution
of firms, and so forth. It is one of the hallmarks of Philippe’s work to raise
the bar for the set of regularities that a theory of growth should be required
to address. In that spirit, while Phillippe’s paper rightly and convincingly
celebrates the successes of the modern growth agenda, and of his many con-
tributions to it, my brief remarks will try to look at the work ahead. Which
challenges, or “enigmas” in Phillipe’s parlance, still await growth theory?
Think of it as my wish list for Phillipe’s next 30 years of work.
With some huge oversimplification, it is possible to create a taxonomy
of growth experiences. All parts of the world, as far as we know, experienced
Malthusian growth. In the Malthu-
many centuries of Malthusian, or near-
sian era, most increases in aggregate output translated into larger popula-
tions, resulting in relatively modest (if any) increases in living standards
over very long horizons. A plot of GDP per capita against time looks nearly
flat. The taxonomy of growth experiences I alluded to before is based on
A Schumpeterian Perspective 281
when and whether countries exited the Malthusian regime, and what hap-
pened after they left it. It features four groups.
The first category is composed of a small number of economies that
experienced an industrial revolution at the beginning of the nineteenth
century. After the industrial revolution, the rate of growth in per capita
income picked up very markedly and has remained roughly steady, when
year periods, ever since. In a plot of (log)
averaged over decades or twenty-
income per capita against time, we see something of a “kink” at the time
sloping line
of the industrial revolution, and a remarkably straight upward-
ever since. We might call these countries the “pioneers.”
A second group is made up of countries that underwent a similar indus-
trial revolution at later dates, and subsequently (broadly) converged to the
pioneers. The date at which these later industrializations took place varied
nineteenth century for some European coun-
enormously, from the mid-
twentieth century for some East Asian ones. However, the
tries to the mid-
common feature is that subsequent to their industrial revolution, these
countries experienced a sustained period in which they grew faster than
the pioneers, so that eventually living standards became quite similar for
the first and second groups. In a plot against time, (the log of) per capita
sloping, concave trajectory,
income shows a kink, followed by an upward-
asymptoting to a straight line that runs close to the straight line of the pio-
neers. I’ll call these countries the “convergents.”
In the third bin are countries that, similar to the convergents, expe-
rienced industrialization followed by a period of growth exceeding the
growth of the leaders, but whose convergence process, unlike the conver-
gents, aborted prematurely. That is, these countries were able to make up
some of the income gap with the pioneers, but then their growth rate sta-
bilized at a rate similar to those of the pioneers well before living standards
became similar. The plot of their (log) income per capita against time is
similar to that of the convergents, except that the linear section is well
below the linear path followed by the pioneers. We may refer to this group
income trapped.”1
as the “middle-
income trap” in terms of income relative to
1. I am implicitly defining the “middle-
the pioneers, not in terms of absolute income. The hypothesis of a middle- absolute-
income trap does not withstand even a modest amount of scrutiny: Most of the
income trap (e.g., Brazil, South Africa)
countries cited as poster boys for the middle-
282 Comments by Francesco Caselli and Aart Kraay
The fourth and final group includes those countries that never under-
went a proper industrialization stage. As a result, these countries did not
income
even exhibit the temporary phase of convergence that the middle-
trapped experienced. I do not mean to suggest that most of these countries
are still stuck in the Malthusian regime. Some structural transformation
from agriculture to services is happening, as is a fair amount of urbaniza-
tion. Nevertheless, these changes do not seem able to ignite a sustained
up process, and as a result, countries in this group tend, at best, to
catching-
maintain their relative position and, at worst, to diverge from the leaders.
Figuratively if not technically, the plots for these countries’ GDP per capita
never shows the kink. We can call the countries in this group the “poverty
trapped.”
How well does modern growth theory explain these four types of expe-
riences, individually and collectively? My view is that it gets full marks in
some areas, passing marks in some others, and bad fails in a few.
The biggest achievement of modern growth theory is its ability to ratio-
nalize the pattern of sustained, steady growth in the pioneers (and in the
up phase)—
convergents after the completion of the catch- the linear growth
in the plots. Steady per capita income growth over the span (in some cases)
of two centuries is an astonishing fact, and the success of growth theory
in illustrating the mechanisms that make this possible is one of the great-
est achievements in macroeconomics. The theory centers on innovation
and clarifies the roles of R&D, property rights protection, competition, and
many other elements. Needless to say, Philippe has been a key force in
developing this body of work.
Another part of the growth landscape that growth theory does pretty well
with is the tendency of later industrializers to convergence to the pioneers.
The modern understanding still largely builds on the old tradition centered
on the “advantage of backwardness,” which has been successfully incorpo-
rated in contemporary growth models (much as theories of growth at the
frontier incorporate older arguments based on creative destruction). In this
view, which seems hard to refute, later industrializers can grow faster, because
they can imitate and adopt technologies already invented by the leaders.
have reasonably steady growth rates when averaged over long periods. What they are
income countries.
failing to do is to close the gap with the high-
A Schumpeterian Perspective 283
Overall, the postindustrialization experiences of the pioneers and the
convergents are the areas where modern growth theory deserves full marks.
This may be because these are the success stories, and economics often
seems to do better at explaining successes than failures.
up experi-
With the question of why some initially promising catching-
income trap, we have some ideas but
ences petered out, ending in a middle-
no consensus and little evidence. Philippe does offer a hypothesis, and oth-
ers have advanced their own. The good news is that this is an active area of
research, and I am optimistic that a greater understanding will emerge in
the not-too-distant future.
I would offer a similar assessment on the question of why the industrial
revolution happened in the first place (the first “kink”). This is of course
a classic question both in macro and in economic history, and count-
less books and articles have been written about it. Although a consensus
account still eludes us, I do sense considerable progress in recent years. The
emerging vision features some combination of intellectual developments
(scientific discoveries combined with the enlightenment mindset), political
developments (particularly in terms of the power relations between landed
aristocracy and emerging urban bourgeoisie), and possibly evolutionary
forces.
income trap and on the rea-
In sum, both on the question of the middle-
sons for the transition from the Malthusian to the modern growth regime,
for effort if not for success.
I’d give growth theory a passing mark—
And now for the bad fail: the later “kinks” and, in particular, the lack
thereof. Why did some countries manage to properly industrialize and oth-
ers never do so? What is the difference between the convergents and the
income trapped, on one hand, and the poverty-
(eventually) middle- trapped
on the other? I hope I am not too harsh in saying that not only do we not
know, but we are not even trying to know. Perhaps, as I mentioned, modern
macroeconomics is so bad at explaining failures that we are not even will-
ing to try. But the question of the failure to industrialize is too important to
give up on. Our friends on the microeconomic development side know this
trapped
and are working hard on finding out how individuals in poverty-
countries can do a little bit better. We macroeconomists should join them
and try to figure out ways in which the whole country can get that “kink.”
Comment: Aart Kraay
Philippe Aghion’s paper for this conference provides a concise and insight-
ful summary of his fundamental contributions to the theory and empirics
of economic growth over the past 25 years. It is challenging for Philippe to
do justice to such an impressive body of work in the short space of a confer-
ence presentation, but he succeeds remarkably in doing so. It is even more
challenging to follow Philippe as a discussant, so my expectations for this
discussion are appropriately modest.
We have known since the fundamental work of Solow and Swan in
the 1950s that, in the presence of diminishing returns, sustained growth
in output in the long run requires sustained growth in technology. But it
took another 30 years for the profession to begin to articulate theories that
spelled out mechanisms through which improvements in technology came
about. Philippe Aghion, in work with Peter Howitt, was at the forefront
of this movement in the 1980s, formalizing earlier insights from Joseph
articulated
Schumpeter about the process of creative destruction into well-
and elegant models of innovation and growth.
By spelling out the incentives for innovation, these models not only
provided a theoretical basis for technology growth, but they also gener-
ated a rich set of insights for policy makers contemplating changes to laws
and regulations affecting property rights, competition, and firm entry and
exit. As discussed in Philippe’s paper, the insights of Schumpeterian growth
theory also have implications for current debates about secular stagnation,
trends in inequality, and much more.
I focus my discussion on the fundamental underlying theme in Philippe’s
the importance for growth of technology growth and the impor-
paper—
tance of Schumpeterian creative destruction in generating the innovations
A Schumpeterian Perspective 285
that lead to growth in technology. Given my professional background as
a World Bank economist, I want to reflect particularly on the relevance of
these themes for policy makers in developing countries. I organize my dis-
country and
cussion around three questions: (1) How important are cross-
time differences in technology? (2) What is “inside” the differences in
over-
technology that we can isolate at the aggregate level?, and (3) What do the
answers to these questions imply for development policy?
How Important Are Differences in “A”?
A basic premise in Philippe’s work is the importance of understanding the
forces that drive differences across countries and changes over time in the
level of technology, conventionally referred to as A in a neoclassical produc-
tion function Y = AF(K, H), where K and H represent physical and human
capital, respectively. Various recent accounting exercises have contributed
country differences in A are large (for example, see
to the view that cross-
Caselli (2005), whose notation I follow here). These typically are based on
a decomposition of income differences between rich and poor countries
along the following lines:
YRICH A RICH F( K RICH , H RICH )
= × .
YPOOR A POOR F( K POOR , H POOR )
Depending on how “rich” and “poor” are defined, one can easily confront
fold differences in
the task in such a decomposition of explaining up to 40-
YRICH
incomes between rich and poor countries, i.e. ≈ 40. Baseline assump-
YPOOR
Douglas, (b) physical capi-
tions that (a) the production function is Cobb-
tal stocks are related to the accumulation of observable past investments,
and (c) human capital stocks are some straightforward linear aggregate of
workers with productivity differences adjusted for some observed measure
country dif-
of schooling,can be used to evaluate the contribution of cross-
ferences in factors of production to these differences.Under these baseline
assumptions, it is typically possible to generate something in the range of
fold cross-
5-to 8- country differences in the contribution of factors of pro-
F( K RICH , H RICH )
duction to cross-country income differences,i.e. ≈ 5 − 8 This
F( K POOR , H POOR )
country differences in the level of technology A must
in turn implies that cross-
286 Comments by Francesco Caselli and Aart Kraay
country
also be in the five-to eightfold range to account for observed cross-
differences in output.
Thus taking the data at face value suggests a very large role for cross-
country differences in technology, and therefore a comparably great
importance for imposing structure on these differences through theories
of innovation that lead to differences in technology levels across countries.
Although I do not spell it out in detail here, one can of course perform
similar decompositions in countries across time, leading to measures of the
growth rate of A in countries over time. Such growth (as opposed to devel-
country differ-
opment) accounting exercises often reveal very large cross-
ences in measured growth rates of technology.
However, as with many things, the devil is in the details, and one does
not have to go very far into the literature to find careful consideration of
measurement issues that, when properly addressed, suggest that we should
country and over-
take a more nuanced view of the importance of cross-
known example comes from
time differences in A. One early and very well-
Alwyn Young’s meticulous growth accounting exercises for rapidly growing
East Asian economies, which suggested that once increases in factors of
production were more comprehensively measured, the productivity growth
underlying the extraordinary output growth in these countries was actually
quite ordinary (Young 1995). Perhaps the starkest case is that of Singapore
year period 1966–
over the 25- 1990 studied by Young: Although output
grew at nearly 9 percent per year, productivity growth was indistinguish-
able from zero, once such factors as increasing labor force participation,
increased human capital, and a more efficient allocation of resources across
sectors were taken into account.
Turning to more recent examples, Jones (2014) and Manuelli and Ses-
hadri (2014) tackle in different ways the question of the contribution of
human capital to differences in output per capita across countries. Jones
(2014) emphasizes the consequences of considering alternatives to the stan-
dard linear human capital aggregator. The standard aggregator plausibly
assumes that skilled workers are X times more productive than unskilled
woirkers, but it implausibly assumes that skilled and unskilled workers are
perfectly substitutable after this rescaling by productivity levels is taken into
a skilled task can be accomplished by one skilled worker or by X
account—
unskilled workers. It does not take much introspection to realize the implau-
sibility of this benchmark assumption, and Jones (2014) spells out a variety
A Schumpeterian Perspective 287
of more realistic human capital aggregators that recognize the complemen-
tarity between different skill types. These in turn lead to much greater dif-
ferences in aggregate human capital across countries, which in turn imply
country differences in factors of production and a
a greater role for cross-
country differences in productivity.
commensurately smaller role for cross-
In a related paper, Manuelli and Seshadri (2014) take seriously incentives
to invest in human capital. Although their paper is much richer than this,
if individuals rationally take into account the
the basic insight is simple—
quality of human capital formed through investments in education, then
low observed investments in education signal not just that the level of
human capital is low but also that the quality of human capital is low. Cali-
country data suggests a much larger
brating their model seriously to cross-
role for human capital differences to per capita output differences across
countries, and therefore again a smaller role for productivity differences.
country or over-
All of this is not to say that cross- time differences in
productivity are unimportant. Rather, it emphasizes that (1) careful,
consistent measurement of factors of production is important,
theory-
and (2) understanding the forces that create incentives for investments in
physical and human capital is at least as important from a policy perspec-
tive as is understanding better the incentives for innovation that lead to
increases in A.
What Is “Inside” A?
country differ-
As noted above, careful measurement suggests that cross-
ences in A may not be quite as large as a naïve first look at the data might
suggest. However, even after careful measurement,they likely are nontrivial
and therefore worth understanding more deeply. The literature on Schum-
peterian innovation that Philippe has made seminal contributions to has
based view of these differences.But cross-
offered an innovation- country dif-
ferences in the abilities of society to allow innovation to take place and
bear fruit are not the only reason why A may be different. These alternative
explanations are worth taking seriously, because they may suggest alterna-
tive policy levers to promote sustained growth.
A first set of explanations that has attracted considerable empirical atten-
tion over the past decade hinge on misallocation of resources across firms
or sectors of the economy, particularly in response to policies that favor
288 Comments by Francesco Caselli and Aart Kraay
some firms or sectors over others. To the extent that such policies prevent
marginal products of factors from being equalized across alternative uses,
country differences in A even when measured
they can contribute to cross-
aggregate factors of production, such as K and H , are the same. In one
of the seminal contributions to this literature, Hsieh and Klenow (2009)
document differences in marginal products of capital across manufacturing
firms in narrowly defined industries. Their results suggest that a country
such as China could effectively double its level of aggregate productivity in
manufacturing simply by reducing its level of resource misallocation to that
observed in the United States.
Another set of explanations for what might contribute to low values of
A revolves around managerial incompetence rather than lack of access to
the best technology or dulled incentives to innovate at the technological
frontier. Bloom et al. (2013) document extremes of mismanagement in a set
of Indian firms, such as basic failures to manage inventories and materials,
or failures to maintain minimal standards of cleanliness and safety in and
around factories. Bloom et al. (2013) go on to show that an experimental
intervention that provided management training to firms resulted in a sig-
nificant improvement in productivity in these firms.
In fairness, misallocation and mismanagement are probably not fully
separate causal factors in driving the low levels of A, and indeed, one might
argue that they are in part a manifestation of the same lack of competitive
pressures that also contribute to low innovation. In an environment with
weak competition, the incentives to ensure that resources are efficiently
deployed in and between firms may also be weak. However, this is a some-
what different mechanism than the effect of competition on incentives to
innovate that is stressed in the Schumpeterian approach.
Finally, although it is perhaps not so surprising that a lack of Schumpe-
terian innovation may not be the main reason behind low productivity in
a developing country, it seems more plausible that it is an important factor
Macia, Hsieh, and
in advanced economies. Yet in a recent paper, Garcia-
Klenow (2016) study the dynamics of innovation at the firm level in the
United States and document some patterns that seem at odds with Schum-
peterian dynamics. For example, contrary to the Schumpeterian view of
“creative destruction,” where innovative new firms replace existing firms
that fail to innovate, they document that most of growth in the United
States seems to come from growth in incumbent firms rather than from
A Schumpeterian Perspective 289
new firms replacing old ones. They also document that much of innovation
seems to take the form of improvements in existing products rather than
creation of new products. Both of these observations suggest that a more
nuanced interpretation of the Schumpeterian emphasis on innovation and
creative destruction is in order.
Implications for Development Policy?
Philippe’s paper concludes with a set of policy prescriptions designed to
unleash Schumpeterian growth. The list is short, sound, and sensible:
markets,
(1) liberalize entry and encourage competition, (2) liberalize labor
(3) promote institutions such as autonomous universities that foster
research, and (4) develop a policy framework to encourage equity finance
of risky investments in R&D in richer countries near the technology fron-
tier. One does not have to squint very hard at this list to see key elements of
traditional policy advice included in the “Washington Consensus,” nor is it
very hard to provide a Schumpeterian interpretation of key ingredients in
the Washington Consensus. For example, classic elements on John William-
but not on Philippe’s list—
son’s list— such as competitive exchange rates,
trade liberalization, and deregulation, can all be thought of as fostering
competitive pressures that drive Schumpeterian innovation and growth.
In fact, this raises the question of whether the four policy prescriptions in
Philippe’s paper are uniquely Schumpeterian, or whether they are just plain
sensible. For example, liberalization of entry and deregulation of labor mar-
kets arguably have direct effects on resource misallocation, which through
this channel may raise productivity, even if they do not directly promote
competition. Conversely, the emphasis on property rights protection in the
Washington Consensus can be interpreted as a key factor in promoting
Schumpeterian innovation (because innovators require assurance of their
property rights over the new ideas they develop). But at the same time, it
there are many
is hardly a uniquely Schumpeterian policy prescription—
other channels through which the protection of property rights promotes
economic growth that do not operate through the channel of innovation.
Another issue raised by Philippe’s list is the question of prioritization,
particularly when one considers developing countries, and especially those
very far below the frontier, who face much more primordial challenges
than the lack of innovation. Prescriptions to foster autonomous universi-
ties are probably sensible advice for advanced economies and a handful of
290 Comments by Francesco Caselli and Aart Kraay
emerging economies near the frontier, but they are unlikely to be priorities
in the many developing countries that struggle to provide even minimal
education and health care to kids.
A final difficult question that merits serious consideration when turning
Schumpeterian insights into development policy advice concerns the polit-
ical feasibility of this advice. Recall that the fundamental Schumpeterian
insight is that when firms face competitive pressures, they are forced to
innovate to escape these competitive pressures, unleashing a virtuous circle
of innovation, competition, and further innovation that raises growth. But
the reality, particularly in many developing countries facing governance
connected firms have at their disposal tools other
challenges, is that well-
than innovation to escape competitive pressures, and these tools lead to
less virtuous outcomes. There are many such possibilities, but a particularly
vivid example comes from recent work by Rijkers, Freund, and Nucifora
induced bar-
(2014). They meticulously document the incidence of policy-
riers to entry across different sectors in Tunisia and then go on to show
that the presence of these barriers is strongly associated with the presence
President Ben Ali. More generally,
of firms connected to the family of then-
how to implement procompetitive Schumpeterian growth policies in envi-
ronments in which politically powerful incumbents are precisely the ones
benefiting from the absence of competition remains a deeply challenging
question for development policy makers.
References
Bloom, Nicholas, Benn Eifert, Aprajit Mahajan, David McKenzie, and John Roberts.
2013. “Does Management Matter? Evidence from India.” Quarterly Journal of Econom-
51.
ics 128 (1): 1–
Country Income Differences.” In
Caselli, Francesco. 2005. “Accounting for Cross-
Handbook of Economic Growth, volume 1A, edited by Phillipe Aghion and Steven Dur-
lauf, 679–741. Amsterdam: Elsevier.
Macia, Daniel, Chang-
Garcia- Tai Hsieh, and Peter J. Klenow. 2016. “How Destruc-
tive Is Innovation?” Unpublished manuscript, University of Chicago.
Hsieh, Chang-Tai, and Peter J. Klenow. 2009. “Misallocation and Manufacturing TFP
1448.
in China and India.” Quarterly Journal of Economics 124 (4): 1403–
Jones, Benjamin F. 2014. “The Human Capital Stock: A Generalized Approach.”
3777.
American Economic Review 104 (11): 3752–
A Schumpeterian Perspective 291
Manuelli, Rodolfo, and Ananth Seshadri. 2014. “Human Capital and the Wealth of
2762.
Nations.” American Economic Review 104 (8): 2736–
Rijkers, Bob, Caroline Freund, and Antonio Nucifora. 2014. “All in the Family: State
Capture in Tunisia.” World Bank Working Paper 6810, World Bank, Washington,
DC.
Young, Alwyn. 1995. “The Tyranny of Numbers: Confronting the Statistical Reali-
ties of the East Asian Growth Experience.” Quarterly Journal of Economics 110 (3):
641–680.
III New Areas of Research and Inquiry
7 Climate Change, Development, Poverty, and Economics
Sam Fankhauser and Nicholas Stern
The past three decades have seen an unprecedented increase in world liv-
ing standards and a fall in poverty across many fundamental dimensions.
Increased confidence in what was possible together with greater acceptance
of moral responsibilities led to the adoption of the Millennium Develop-
ment Goals at the turn of the century. They provided a real basis for inter-
national cooperation and development. In the Sustainable Development
Goals (SDGs), agreed on in September 2015, there is now a common plat-
form for the next phase of the fight against poverty.
The SDGs make it clear that environmental protection will be a key fea-
ture of this next phase, since it is increasingly intertwined with poverty
reduction. Thirteen of the seventeen SDGs are directly concerned with the
natural environment, climate, or sustainability. Environment, climate, and
sustainability were not prominent in the Millennium Development Goals.
With hindsight, we can now see that this omission was a mistake.
A key factor in all this is climate change. Climate change is not the only
environmental problem we face, nor is it the only threat to global pros-
perity. But climate change is unique in its magnitude and the vast risks it
multiplier for other urgent concerns, such as
poses. It is a potent threat-
habitat loss, disease, and global security (IPCC 2014). And it puts at risk
We thank Gael Girard, Mike Toman, Bob Ward, and the participants of the World
Bank conference on The State of Economics, the State of the World (Washington,
DC, June 2016) for their thoughtful comments. Patrick Curran and Isabella Neuweg
have provided outstanding research support. We also acknowledge financial support
from the Grantham Foundation for the Protection of the Environment and the UK
Economic and Social Research Council (ESRC), through its support of the Centre for
Climate Change Economics and Policy (CCCEP).
296 Sam Fankhauser and Nicholas Stern
the development achievements of the past decades (Hallegatte et al. 2016).
If unchecked, climate change could fundamentally redraw the map of the
planet, and where and how humans and other species can live.
Climate change is also unique in the scale of the response that is needed.
Reducing climate risks requires cooperation from all countries, developed
and developing, to reorient their economic systems away from fossil fuels
use practices. This reorientation is urgent. Our activities
and harmful land-
in the next two decades will determine whether our successes in develop-
ment will be sustained or advanced, or whether they will be undermined or
reversed in a hostile environment.
The nature of the climate problem has implications for economic analy-
sis. Economics has much to offer, and indeed continues to provide impor-
tant insights, but there has been a dangerous tendency to force climate
change into narrow conventional ways of thinking. This must change. We
need to construct theories and models that reflect the structure and scale of
the problem and the contexts in which it occurs.
Climate change also has implications for development policy. In the
negotiated at the end of 2015—
Paris Agreement— there is now an interna-
tional platform through which global climate action can be advanced and
coordinated. The Paris Agreement has been ratified by 185 countries (as
of April 2019). It sets out a process through which the rise in global mean
industrial
temperatures may be curtailed to “well below” 2° C above pre-
levels and perhaps as low as 1.5° C. In 2018 the Intergovernmental Panel
on Climate Change advised that 1.5° C would have substantial benefits for
people and the natural environment, compared with 2° C (IPCC 2018).
Meeting the Paris objectives requires sustained action over many decades.
It also requires the reorientation of investment. At least US$100 trillion will
be invested over the next two or three decades in buildings and urban infra-
structure, roads, railways, ports, and new energy systems. It is imperative
that these investment decisions are taken with climate change in mind.
If they are, there will be substantial benefits for development and pov-
living spaces where we can move, breathe, and be produc-
erty reduction—
tive and better protection for fragile ecosystems, as well as the fundamental
reduction of the risks of climate change.
Putting the SDGs and Paris together, the agreements of 2015 have given
us, for the first time, a global agenda for sustainable development apply-
ing to all countries. This chapter sets out the implications of this agenda,
Climate Change, Development, Poverty, and Economics 297
and climate change in particular, for development economics and develop-
ment policy. It emphasizes the nature of the required changes and their
implications. We start with an examination of what economics has had to
say about the link between economic prosperity and the environment. We
then explain why climate change is a different kind of problem, and why it
requires a new approach to both analysis and policy. The final two sections
explore how this new approach might look.
Prosperity and the Environment
Environmental concerns entered development policy relatively late. The
World Bank created the Office of the Environmental Advisor in 1970, but
in the early years, this was very much an advisory function. Over time, the
role evolved and the environment grew in importance, culminating in the
creation of the Environmentally Sustainable Development vice presidency
in 1993.1 In parallel, environmental economics began to emerge as a new
field of academic study (Pearce 2002).
Understanding the interactions between economic growth and environ-
mental protection is crucial to development in all countries, but especially
in poor ones. Careful environmental management is a critical ingredient
of any viable path to poverty reduction. Bad environmental management
results in environmental degradation, poor public health, and lost eco-
nomic output. Poor people are the primary victims of these trends, though
we should recognize that poverty also contributes to them (Pearce and
Warford 1993).
Environment and Growth
Knowledge about the link between economic development and the envi-
ronment of course goes back much further than the 1970s. The econom-
ics pioneers of the eighteenth and ninetenth centuries were well aware of
environmental resources as an essential source of wealth, and indeed as a
potential constraint to economic growth. For David Ricardo, differences in
Malthus,
land quality were the main source of rent for landowners. Thomas
more pessimistically, predicted widespread poverty as a consequence of
1. See https://archivesholdings.worldbank.org/.
298 Sam Fankhauser and Nicholas Stern
population growth and decreasing returns to agriculture. Montesquieu spec-
ulated at length about the influence of the climate on society and the “tem-
per of the mind” (Montesquieu [1748, Book XIV] 2011), but the link to
economic performance was cursory. The early economists were more inter-
ested in resource endowments than climate factors.
Unlike Montesquieu’s theories on climate, Malthus’s concern about nat-
ural resource constraints has remained a constant feature of the growth
debate. In the 1860s, William Stanley Jevons worried about the future of
industrial England when its coal reserves would run out. In the 1970s, the
Club of Rome made headlines with The Limits to Growth (Meadows et al.
1972). Inspired by Kenneth Boulding’s (1966) notion of “spaceship Earth,”
the interdisciplinary field of ecological economics has continued to probe
the natural boundaries that the laws of science impose on economic pro-
cesses (e.g., Rockström et al. 2009).
So far, Malthus and the resource pessimists have generally appeared to be
wrong. Human ingenuity has mostly managed to outpace natural resource
constraints. This does not mean that environmental resources are not over-
exploited. They are, including not least in developing countries. However,
in most cases this overexploitation appears, in large measure, to be the
result of policy mismanagement and market failure rather than resource
scarcity per se.
The Management of Natural Resources
From the outset, economists have devoted considerable attention to the
effective management of natural resources. In the nineteenth century, Knut
Wicksell and Martin Faustmann were among the first to study the optimal
harvesting cycle for slow- Nyström
maturing resources like forests (Hedlund-
et al. 2006). However, it was Harold Hotelling (1931) who produced the
defining treatise on natural resource management. According to his Hotel-
ling rule, the value of natural resources, if optimally used, must rise at the
rate of interest. This insight has formed the basis of natural resource eco-
nomics to this day. It also informs the analysis of stock pollution problems
like climate change.
The Hotelling rule was revisited in the 1970s, when it became appar-
ent that it may not be consistent with an emerging development concept,
that of sustainable development. The notion of sustainable development
was popularized by the Brundtland Commission on Environment and
Climate Change, Development, Poverty, and Economics 299
Development, which defined it as “development which meets the needs
of current generations without compromising the ability of future genera-
tions to meet their own needs” (World Commission on Environment and
Development 1987).
For economists, this meant consumption (or utility) could not be allowed
to decrease over time. Robert Solow and John Hartwick worked out what
nondecreasing utility meant for resource depletion. The rents from natural
resource extraction had to be reinvested in other forms of capital, so that
the total stock of environmental, physical, and human capital remained
constant (Solow 1974; Hartwick 1977). The World Bank has been at the
Solow rule into practical policy advice
forefront of translating the Hartwick-
(World Bank 2011).
Environmental Management and Public Policy
If Harold Hotelling is the forefather of natural resource economics, Arthur
Cecil Pigou deserves the credit for incorporating environmental concerns
into welfare economics. Drawing on his teacher Alfred Marshall, Pigou sys-
tematically introduced into economics the notion of externalities, that is,
costs or benefits that are not captured in the market price of goods. Later
such as open access problems, com-
writers added nuance and extensions—
that refine our understanding
mon property resources, and public goods—
related market failures, but the core concept of externali-
of environment-
ties remains central to modern environmental economics.
Pigou’s observations on the environment were prescient. He discussed at
length the negative effects of pollution, which “inflicts a heavy uncharged
loss on the community” (Pigou (1920), as cited in Sandmo (2015, 53)). The
concern remains valid to this day. Urban air pollution, linked to particulate
matter and other pollutants, remains a major issue in most countries (New
Climate Economy 2014). In another perceptive comment, Pigou praised the
external value of forests, whose “beneficial effect on climate often extends
beyond the borders of the estates owned by the person responsible for the
forest,” though he probably had the local climate in mind (cited in Sandmo
(2015, 55)).
Pigou also identified the requisite remedy to address these market fail-
ures: a corrective tax levied in proportion to the externality. This was later
complemented by the work of Ronald Coase, who showed that problems
of externalities could also be managed via clearer (and perhaps tradable)
300 Sam Fankhauser and Nicholas Stern
property rights (Coase 1960). Both writers were drawing on John Stuart
Mill, who already in 1848 had called for government intervention to ensure
the “common enjoyment” of the world’s natural riches (Sandmo 2015).
Today, variants of Pigouvian taxes and Coasean trading schemes are in use
throughout the world (for an overview, see Sterner (2003); Freeman and
Kolstad (2007)).
Following in Pigou’s footsteps, John Hicks and Nicholas Kaldor devel-
oped the theory for a systematic comparison of the costs and benefits of
policy intervention. James Meade (1955) provided the defining general
equilibrium approach and analysis in his seminal book Trade and Welfare
(see also Drèze and Stern (1987, 1990)). Cost-
benefit analysis soon became
the standard tool for project appraisal, including in development organiza-
tions like the World Bank (e.g., Little and Mirrlees 1974).
In environmental economics, the extensive body of work on welfare
the use of
economics gave rise to the field of environmental valuation—
techniques that monetize the external value of the environment, so it can
benefit analysis (for an overview, see
be appropriately reflected in cost-
Hanley and Barbier (2009)).
It soon became clear that nature’s contribution to human welfare goes
well beyond the provision of food and materials, which had exercised
Malthus and the Club of Rome. The modern theory of ecosystem services
(e.g., TEEB 2010) distinguishes between provisioning services (food, water,
materials), cultural services (spiritual value, recreation, mental and physical
health), regulating services (air quality, water treatment, carbon sequestra-
tion) and support services (genetic diversity, habitats). The full extent of
or indeed, always
this rich range of services is not yet fully understood—
by policy makers. It remains an active and important area of
appreciated—
interdisciplinary research.
A central test for any economic prescription on environmental man-
agement is the health of the natural environment. Against this yardstick,
the economics of Hotelling, Pigou, Meade, and their successors has serious
limitations. There have been notable successes, but on the whole, environ-
mental protection in practice has been much harder than the solutions
embodied in simple theory. The political economy of poverty and the
environment is particularly complex and has to include factors like power,
exclusion, land rights, market access, and gender relations.
Climate Change, Development, Poverty, and Economics 301
development nexus has become more
Unfortunately, the environment–
first century
complex still. The environmental problems of the twenty-
could be of a different order of magnitude and generality than those of the
past, and none more so than climate change.
Why Climate Change Is Different
Climate change is different from past environmental problems in terms
of its scale, the magnitude of risks, and the urgency of action. We are all
involved both in the generation of the problems and in our vulnerability to
its impacts. Climate change is also different in terms of its complexity and
the difficulty of identifying a “solution.” To appreciate the nature and scale
of the challenge, it is necessary to set out some basic science about climate
change.
Science
The science of climate change is based on almost two centuries of theory
that there are
and evidence. The basic physics of the greenhouse effect—
trapping gases in the atmosphere, which leads to the earth retaining
heat-
were established by Jean-
heat— Baptiste Fourier and John Tyndall in the
second half of the nineteenth century. Studying the earth’s heat balance,
the former showed that something was preventing the escape of energy,
and the latter identified the key gases at work. At the start of the twenti-
based emissions
eth century, Svante Arrhenius made the link to fossil fuel-
by showing that they intensified the magnitude of the natural greenhouse
effect. In the first half of the twentieth century, with the rise of quantum
theory, it was established that the mechanism at work was the frequency
of oscillation of greenhouse gas molecules, which interfered with that of
infrared energy. The systematic monitoring of atmospheric CO2 concentra-
tions began in 1958.
This part of the physics and chemistry of the atmosphere is basic and
clear. Important uncertainties remain, but we increasingly understand the
main driving forces in the inherently complex and chaotic system that is
the earth’s climate. From this evidence, which continues to be gathered,
published, and presented, we understand that the current, unprecedented
climate change starts and ends with people.
302 Sam Fankhauser and Nicholas Stern
Human activity, through the extraction and combustion of fossil fuels,
removal of forests, or agricultural activities contributes to the emission (or
“flow”) of greenhouse gases. The increased flows lead to increased quanti-
ties (or “stocks”) of greenhouse gases in the atmosphere, and with them, an
increase in the amount of heat energy trapped by the atmosphere. As the
heat energy increases, so too do the average global land and sea tempera-
tures. With higher temperatures and more energy, there is increased inten-
sity and variability in the global climate system, leading to fluctuations or
changes in local and regional weather patterns.
Risks
The implications of this complex causal chain are difficult to comprehend
in their entirety, and the specifics cannot be predicted with certainty. How-
ever, it is clear that the effects in terms of human lives and livelihoods are
potentially severe.
1800s,
Since the beginning of the Industrial Revolution in the mid-
global mean surface temperatures have risen by about 0.9°C (IPCC 2018).
The atmospheric concentration of the main greenhouse gases has increased
from about 285 parts per million (ppm) of carbon dioxide equivalent (CO2e)
to more than 450 ppm of CO2e today, of which over 400 ppm is CO2. About
70 years ago, we were adding approximately 0.5 ppm of CO2e per year, and
now we are adding about 2.5 ppm of CO2e per year. If this trend continues,
the median temperature increase over the next one or two centuries would
be in the region of 4° C, with a substantial probability of well over 4° C
(IPCC 2013).
To put these numbers into context, our civilization has developed dur-
ing the climatically benign Holocene period, following the last ice age,
which came to an end about 9,000 or 10,000 years ago. The Holocene has
had relatively stable temperatures that fluctuated in a range of ±1−1.5° C
relative to the late nineteenth century benchmark. We are now near the
edge of that range. If the temperature increase reaches 3 or 4° C, we would
be outside the range of experience of our species, Homo sapiens, which is
about 250,000 years old. The planet has not seen a 3° C increase in tem-
perature for about 3 million years (when the sea level was about 20 meters
higher than it is today; IPCC (2013)), and 4° C for tens of millions of years.
Along with the physical science, the natural and social sciences are rap-
idly developing models to investigate the risks of rising temperatures for
Climate Change, Development, Poverty, and Economics 303
economies, ecosystems, cultures, and social structures. The specifics cannot
be known with certainty, but risks to people and the environment will rise
rapidly above 1.5°C of warming (IPCC 2018). There is an increased risk of
tipping points (Drijfhout et al. 2015) and of exacerbating and compound-
ing other threats, like habitat loss, political instability, and disease (IPCC
2014).
Poor countries and poor people would be hit particularly hard. They rely
sensitive economic activities like agriculture and
more heavily on climate-
have reduced capacity to adapt effectively. Poor people are also more likely to
live in hazard zones, such as floodplains, and their assets are more likely
to be damaged in extreme weather events. They are also more susceptible
to the pests and diseases that follow heat waves, floods, and drought (Hal-
legatte et al. 2016).
The Urgency
Limiting temperature rises to any specific level requires the restriction of
lived greenhouse gases in the atmosphere. The
the accumulation of long-
concentration of greenhouse gases in the atmosphere cannot exceed a cer-
tain threshold and must stabilize at a lower level. The lower the tempera-
ture target is, the lower the threshold and stabilization level will be and the
sooner emissions will have to peak.
zero,” that
Eventually, global annual emissions will have to reach “net-
is, a balance must be established between the release of greenhouse gases
into the atmosphere from human activities and their removal (for example,
through reforestation).
The 2° C upper temperature bound in the rise in global mean surface
temperature is associated with a remaining “budget” for carbon dioxide,
1100 gigatons of CO2
the most important greenhouse gas, of maybe 600–
over the period to 2100, depending on the probability we seek of keeping
to the 2º C target; the higher the probability the lower the budget. A 1.5º C
target would involve lower budgets in the order of 400–750 gigatons of Co2
and require reaching net zero by around 2050 (IPCC 2018).
1100 gigatons CO2, global
To remain within an emissions budget of 600–
emissions would have to peak before 2020 and decline rapidly from then
on. Negative emissions technology (not just expanded forest cover but also,
e.g., bioenergy combined with carbon capture and storage) will likely be
required later in the century to avoid warming of more than 2º C.
304 Sam Fankhauser and Nicholas Stern
sum game. The higher one
The global emissions budget creates a zero-
country’s emissions are, the lower those of other countries will have to
be. It is here that disagreements occur. Developed countries are responsible
for the majority of historical greenhouse gas emissions. But the balance of
annual emissions has shifted in recent years. Developing countries (led by
China) now account for about 60 percent of total annual emissions and will
be responsible for most future emissions growth (New Climate Economy
2014). Six of the top 10 emitters are developing countries (World Resources
Insititute 2014).
Cooperation
Tackling climate change thus requires efforts from all countries and strong
international cooperation. Experience tells us that such cooperation can be
hard to secure. International cooperation on climate change has histori-
cally been difficult,.
The benefits that accrue from reduced climate risks are a global public
good. Countries cannot be excluded from profiting and have incentives to
free ride if they perceive reducing emissions to be costly to themselves and
disregard the benefits to others. Moreover, the group that would benefit
is large and diverse, and the impacts of accelerated climate change affect
countries unevenly. These are strong reasons for why reaching an agree-
ment is difficult, but they are also the reasons that international coopera-
tion is needed (Barrett 2003).
Against this backdrop, the Paris Agreement is a remarkable breakthrough
in international climate cooperation. To illustrate this, compare Paris to
another agreement that seemed almost impossible at the time. The Bretton
Woods Agreement brought together 44 countries in an attempt to rebuild
the international economic and financial system after World War II in a
more cooperative form.
In 1944, Keynes (cited in Braithwaite and Drahos (2001, 98)) described
four nations
it as “forty- … actually able to work together at a constructive
task in amity and unbroken concord. Few believed it possible. If we can
continue in a larger task as we have begun in this limited task, there is hope
for the world.”
Although the Bretton Woods agreement should be regarded as a crucial
achievement, it is important to recognize that the urge for collaboration
World War II era and the call for international coordination
in the post–
Climate Change, Development, Poverty, and Economics 305
were almost omnipresent. The grave experience of two world wars and a
great depression in 30 years taught some clear and strong lessons. The con-
sequences of the failure to work together were demonstrated to be cata-
strophic; the evidence was hard and real. Furthermore, the United States
was in a dominant position. In contrast, the Paris Agreement brings together
more than 180 countries in anticipation of future harm, which makes it all
the more remarkable. And no one country was dominant.
That an agreement was formed lies not only in the increased understand-
ing of the gravity of the risks but also, and crucially, in an understanding of
the attractiveness of alternative pathways to sustainable development. This
interested action. But the agreement also
has changed the calculus of self-
includes features that enhance the willingness to cooperate by increasing
the benefits of cooperation and realizing them more quickly, such as inter-
carbon research and development (Keohane
national collaboration on low-
and Victor 2016). Moreover, transfers between country coalitions (in the
form of funds, commitments, etc.) helped make the agreement more profit-
able to participants. However, we should also not underestimate a shared
sense of responsibility. Much of the motivation appeared to be beyond nar-
interest and was about responsibility to future generations.
row self-
Yet, however remarkable, the deal struck in Paris must be seen as only
the beginning of a long process of international cooperation. The effective-
ness of the agreement is yet to be tested. The building blocks that have
led to the agreement will need to be expanded and deepened. The pledges
submitted ahead of Paris, if fully implemented, still put the world on an
emissions path that is closer to 3° C warming than the Paris objective of
“well below” 2° C, let alone 1.5° C (Rogelj et al. 2016). Without even closer
cooperation by and action from all countries over the next 10−15 years, the
chance of remaining well below 2° C is slim.
The Analytical Challenge: Beyond the Marginalist Approach
Economists were slow to recognize the enormity of climate change and
its relevance to economic development. Climate change has yet to reach
the mainstream in many economics departments. Yet a small number of
pioneers have engaged with the topic from an early stage (Nordhaus 1982,
1991a, 1991b; Edmonds and Reilly 1983; Cline 1992; Manne and Richels
1992; Schelling 1992).
306 Sam Fankhauser and Nicholas Stern
The authors of those early works applied the tools of their trade. The
groundbreaking work of William Nordhaus was inspired by the growth
theory of Ramsey and Solow.2 The accumulation of greenhouse gases in
the atmosphere was understood as an exhaustible resource problem in the
spirit of Hotelling. The likely impacts of climate change were enumerated,
monetized, and aggregated in the tradition of Pigou and Meade. To correct
the externality, economists advocated Pigouvian carbon taxes or Coasean
emissions trading schemes (see Fankhauser 1995 for an overview of early
climate economics).
Their contributions were essential to building the argument for action.
However, by placing a strong focus on the marginalist tools of welfare
economics, economists have tended to underestimate both the poten-
tial impacts of climate change and the wider benefits of a transition to
carbon growth, to the point where their models were increasingly at
low-
odds with the science. They have focused on fairly marginal perturbations
term growth when the question at hand is the management of
to long-
immense risk and the longer term. Growth itself could be severely disrupted
not simply perturbed on the margin.
and reversed—
The Precautionary Economics of Climate Change Risks
Initial estimates of the economic costs of climate change began to emerge in
the 1990s. They were both derived from and provided input into integrated
assessment models. These models attempt to combine the key elements of
biophysical and economic systems and represent the full cycle from socio-
economic activity to emissions, temperature change, and impacts that then
feed back into the socioeconomics. It was a valiant endeavor, but the early
models suffered from a poor evidence base. Many important impacts either
had to be omitted or were extrapolated from single data points (Tol and
Fankhauser 1998). This had the effect of marginalizing or ignoring some of
the most worrying risks identified by scientists.
Today, our evidence base is much better (IPCC 2014). More solid empiri-
cal evidence is beginning to emerge on the impacts of moderate climate
change, for example, in regard to agricultural impacts (e.g., Schlenker, Hane-
mann, and Fisher 2005; Schlenker and Lobell 2010) and labor productivity
2. Nordhaus’s work on climate change economics was recognized with the 2018
Nobel Prize in economics.
Climate Change, Development, Poverty, and Economics 307
(e.g., Heal and Park 2013; Burke, Hsiang, and Miguel 2015). Case study
evidence also links climate and conflict (Hsiang and Burke 2014; Kelley
et al. 2015).
However, there are inherent limits to the empirical investigation of
severe climate impacts on people. The nature of the problem is precisely
that it will take us outside the range of the empirically observed in the his-
tory of Homo sapiens (see above). To understand the consequences of the
large temperature changes, we might have to go back further in time and
study the evidence from paleoclimatology, for example, on sea levels.
The Intergovernmental Panel on Climate Change therefore concluded
that the results of integrated assessment models depend on a number of
“disputable” assumptions (IPCC 2014). This is hard to disagree with, when,
in one common specification, a temperature increase of 5°C is associated
10% of GDP. Temperatures at that level
with damages equivalent to just 5–
have not been seen for tens of millions of years. The transformation would
likely be traumatic.
Integrated assessment models still have a role to play. However, their-
value does not lie in producing specific estimates of economic damage,
which can be profoundly misleading. Instead it lies in documenting the
high levels of risk we face. Multiple model runs and some understand-
ing of the omitted impacts show that the balance of uncertainty is heav-
ily tilted toward the downside. Negative surprises relative to the effects
that are incorporated are much more likely than positive ones. Economic
tools can be used to translate these uncertainties into prescriptions for risk
management.
An important strand of research, pioneered by Martin Weitzman, is
demonstrating the importance of looking not just at the most likely out-
comes but also at the tail of the distribution (Weitzman 2012). However,
although the focus on the tails is welcome, the central estimates of poten-
beyond past human experience—
tial change over the long term— are
themselves deeply worrying and offer sufficient grounds for strong action
(Stern 2016).
Carbon Transition
The Dynamic Economics of a Low-
carbon development paths
The economic models available to study low-
often, in structure and approach, predate the debate on climate change and
have their origin in energy sector planning. At the core of many models
308 Sam Fankhauser and Nicholas Stern
are estimates of marginal abatement costs, that is, the incremental costs
of reducing emissions by an additional ton. Models based on marginal
carbon strategies
abatement costs have been useful in informing the low-
of many countries. However, by focusing on emission reduction efforts at
the margin, they often ignore the inherently systemic nature and dynamic
force of transformative change.
Some systemwide effects will make carbon abatement more expensive
than would be the case in their absence. We should not underestimate the
difficulty of deep structural change. One key concern is rigidities in the
labor market, both in terms of labor mobility and wages (Bowen and Kural-
intensive
bayeva 2015). There are also rigidities in the capital stock. Carbon-
capital is often long lived, and assets might get stranded unless investment
decisions are sufficiently forward looking (Pfeiffer et al. 2016). And finally,
inertia is associated with innovation, which appears to be heavily path
dependent (Aghion et al. 2016). Few of these effects are properly modeled
carbon capital and
as yet, but they point to the dangers of locking in high-
infrastructure.
However, there are potentially very large gains from future innovations
on cheaper and sustainable paths. We have the potential to harness the
carbon innovation—
large dynamic benefits of low- unlocking the process
of “creative destruction,” which Joseph Schumpeter described back in the
1940s. This includes not just technological innovation but also changes
in business practices and social behavior (Stern 2016). As engineers learn
how to install, connect, and repair technology cheaply, unit costs fall faster
for many new technologies than for existing ones. Also influential will be
vehicle
the emergence of new networks, such as the integration of electric-
energy storage into smart grids. Dechezleprêtre, Martin, and Mohnen (2014)
find that clean technology innovation creates much higher spillovers than
conventional innovation does, on a par with those in transformative sec-
tors like information technology and nanotechnology. New technologies
plus wise management and investment can both produce very large gains
in energy efficiency. Indeed, nearly half of the required action on climate
change could come from energy efficiency.
carbon transition also has other environmental benefits, from
The low-
fuel pollution (air and water) to the preservation of the
reduced fossil-
world’s forests. In China and India, probably close to 2 million people die
each year as a result of poor air quality (New Climate Economy 2014). These
Climate Change, Development, Poverty, and Economics 309
are environmental priorities of immense significance that could and should
carbon transition offers opportu-
be pursued in their own right, but the low-
nities for synergies and coordination.
The Ethics of Intervention
The magnitude of climate risks and the lasting impact of policy choices on
lives and livelihoods, both today and in the future, raise issues of equity and
justice that are more consequential and difficult than we usually encounter
in policy analysis.
Different ethical approaches guide the actions of individuals and com-
munities, but they all provide consistent normative support for strong
action (Stern 2007, 2015). Moral guidance is also offered in the teachings
of major religions. Concern about future generations, deep respect for the
environment, and the duties of the current generation as stewards of the
earth are consistent themes.3
The ethics discourse in economics has, for the most part, made little
accommodation or room for these wider philosophical, ethical, and reli-
gious perspectives. It has focused heavily on technical issues, unusually
narrowly defined, in particular on the intergenerational question of dis-
counting and the intragenerational issue of burden sharing or dividing up
the remaining carbon space.
Discounting is of course a central issue and requires rigorous, analyti-
cal scrutiny from economic, philosophical, and political perspectives. It is
discussed in great detail elsewhere, and readers are referred to Stern (2007,
discounting, because it
2015). Those works argue strongly against pure time-
is essentially “discrimination by date of birth” that would be unacceptable,
for example, in criminal courts, voting procedures, and human rights. If it
were to be introduced as an ethical criteria, it would require direct and con-
vincing argument: Such argument is usually conspicuous by its absence.
These writings also point out that speaking of “the discount rate” as if
it were something introduced entirely from outside the debate is a serious
conceptual mistake. The discount factor is a relative price between goods
3. This can be seen from the Papal encyclical Laudato Si: On Care for Our Common
Home, the Islamic Declaration on Global Climate Change, the Bhumi Devi Ki Jai!
(A Hindu Declaration on Climate Change), and the Buddhist Climate Change State-
ment to World Leaders.
310 Sam Fankhauser and Nicholas Stern
now and in the future. It depends on which goods and which dates. It is
a relative price logically prior to the concept of the discount rate, which
is the rate of fall of the discount factor. Discount factors, and thus, dis-
count rates, like other prices and values, depend on where we turn out
to be, and that depends on our decisions. They are endogenous to our
decision-making.
The ethics of “burden sharing” are also often misconstrued. There is a
powerful argument that developed countries have a moral obligation, from
their history, their wealth, and their technology, to take a strong lead in
cutting emissions. However, the current arguments tend to see rights and
allocations only in terms of a single dimension: greenhouse gas emissions.
The focus on this one dimension ignores a multitude of other relevant
benefits of the alternative
influencing factors and the dynamics and co-
low-carbon transition.
There is no evidence that greenhouse gas emissions are needed for
development. Although energy is an essential requirement for develop-
ment (Fankhauser and Jotzo 2017), it does not necessarily, at least in a tech-
nical sense, have to be associated with greenhouse gas emissions, because it
is possible to source energy with low or zero emissions. It can be argued that
each country or individual has a right to development, a right to energy,
and a right to basic human needs, but these rights neither separately nor
together imply a right to emit or degrade the environment.
The Policy Challenge: Beyond Incremental Action
The development community is increasingly aware of the risks of climate
change (e.g., World Bank 2010, 2012; Hallegatte et al. 2016). However, it
has yet to respond to the threat with sufficient purpose and scale. Climate
policy is not about incremental initiatives that can be attached to exist-
ing development plans. It requires deep structural and systemic change,
implemented over many decades, both to reduce emissions and to adapt to
remaining climate risks.
Climate-Resilient Development
It is well recognized that even a moderate degree of climate change can pose
risks to development. What is less appreciated is the extent to which the
for
rapid development that many developing countries are undergoing—
Climate Change, Development, Poverty, and Economics 311
is shaping their
example, along urban coastlines (Hanson et al. 2011)—
future vulnerability to climate change.
The pace of development means that the greatest opportunities for
akers
achieving climate resilience lie in influencing these trends. Policy m
term development, infrastruc-
should incorporate climate risks into long-
level approach is an
ture, and spatial planning decisions. This macro-
important departure from traditional analysis, which has tended to treat
specific
adaptation to climate change as a set of independent, threat-
responses, such as coastal protection schemes.
resilient development differ from conventional devel-
How does climate-
opment? Thomas Schelling, one of the first economists to engage with cli-
mate change, famously claimed that economic development was the best
resilient devel-
form of adaptation, implying that conventional and climate-
opment are one and the same (Schelling 1992, 1997).
Climate resilience and economic progress are indeed heavily intertwined.
However, not all forms of development have the same effect on climate
resilience. As countries develop, the structure of their economy evolves,
typically away from agriculture. Sectors become more productive, and the
location of economic activity may shift to urban centers. Income per capita
rises, and with higher incomes the demand for climate protection goes up.
Of these changes, only the increased demand for adaptation unequiv-
ocally reduces climate change risks. The net effect of the other trends is
unclear. Although agriculture is highly sensitive to climate change, a struc-
tural shift into industry and urban living improves resilience only if those
sectors and locations encounter fewer climate risks than agriculture, which
they may not do (Fankhauser and McDermott 2014, 2016). For example,
much urban development has involved building on flood plains.
resilient development at the macro scale has institu-
Pursuing climate-
tional consequences. The responsibility for adaptation shifts from envi-
meteorological offices to planning and
ronment departments and hydro-
economic ministries. These tend to be more powerful and better able to
instigate the necessary reforms. This shift is an important and sometimes
level adaptation to climate-
overlooked side effect of moving from project-
resilient development.
When integrating development and climate action, we should recognize
that development (conventionally understood), mitigation, and adaptation
till agriculture and approaches
are closely intertwined. For example, low-
312 Sam Fankhauser and Nicholas Stern
like Sustainable Rice Intensification save energy and water, reduce emis-
sions and are more resilient. There are many further examples in energy,
urban planning, and building design.
Carbon Transition
The Low-
based energy has been such a powerful force of growth and
Fossil fuel-
poverty reduction that it seems reasonable to ask, in the words of Dercon
(2012), whether “green growth is good for the poor.” It is a longstanding
concern. The original text of the UN Framework Convention on Climate
Change deals extensively with the question of who bears the incremental
costs, implying that there is a “horse race” between growth and environ-
mental responsibility.
We now know that the notion of a “horse race” represents a false dichot-
omy. We have highlighted above the dynamic benefits of an innovation-
driven growth model, where learning processes and economies of scale
create investment and employment opportunities. We have also outlined
the environmental benefits of such a course of action, for example, in terms
of air quality, and the great scope for improving resource efficiency. We have
emphasized the intertwining of development, mitigation, and adaptation.
The challenge for development policy is to guide economic decisions in
this new direction. Even if it is beneficial, structural transformation is never
easy. Policy makers will have to tackle fundamental market failures not just
in relation to greenhouse gases, but also in networks, capital markets, clean
innovation, and the provision of information, and with respect to the local,
regional, and global environment. There are harmful policy distortions, not
least the subsidization of fossil fuels and the underpricing of energy, which
amount to hundreds of billions of dollars each year (Coady et al. 2015;
OECD 2015). The vested interests can be very powerful. Political skills and
systems will be tested severely.
The choice of policies is important. Carbon pricing has proven to be an
effective tool to incentivize emission reductions with very limited effects, so
far, on competitiveness (Dechezleprêtre and Sato 2014). The breakthrough
carbon technology requires additional support for clean research
of low-
and early deployment (Dechezleprêtre, Martin, and Bassi 2016). Thought-
ful regulation (and its enforcement) also has a role to play, for example,
in the form of efficiency standards, planning rules, and building codes.
Another essential part of the policy mix is strategies to reduce structural
Climate Change, Development, Poverty, and Economics 313
adjustment costs by supporting labor mobility, providing social safety nets,
income households.
and protecting low-
carbon growth requires the redirection of financial flows and
Spurring low-
investment. Private investors will only do this if the balance of risks and
returns is attractive, and the direction of travel is clear. The consistency,
clarity, and credibility of climate policies therefore matter hugely. This is
not something current political processes always deliver. Government-
induced policy risk is an immense disincentive around the world. However,
it is possible to reduce policy uncertainty, for example, through statutory
carbon targets enshrined in legislation and monitored by an independent
nonpolitical body (Fankhauser 2013).
A key concern is infrastructure. Over the next 20 years, the required
investment in infrastructure will be in the region of US$100 trillion or more
(Bhattacharya, Oppenheim, and Stern 2015). This new capital will be long
lasting, and the choices made now will have enduring consequences for
growth, development, and the climate. Currently about 60 percent of global
annual greenhouse gas emissions can be attributed to the investment in
and use of infrastructure. Very rapid urbanization (likely to rise from about
century) demon-
3.5 billion people now to about 6.5 billion people by mid-
in of wasteful and polluting structures.
strates the immense dangers of lock-
These numbers show that investment over the next 20 years will shape the
future profoundly: It will determine whether we have cities where we can
move and breathe, and whether we can hold the global temperature rise to
well below 2° C.
Conclusions
Human ingenuity has succeeded in overcoming natural resource constraints
that were once thought binding. That extraordinary progress has not been
sufficient to eradicate global poverty, and the natural environment has suf-
fered, but human welfare has improved markedly. However, the environ-
first century are likely to
ment and development challenges of the twenty-
be more difficult than those of the past.
Nowhere is this more evident than for climate change. Climate change
is a threat of a completely different magnitude and character from those of
the past. To continue our progress in the face of climate risks, we need both
strong policy action and a radical deepening of economic analysis. We need
314 Sam Fankhauser and Nicholas Stern
to construct theories and models that reflect the unique challenges we now
face and the contexts in which they arise.
The response to the threat is not the cessation of economic growth (Jack-
son 2011; Klein 2015). It is possible to advance economic prosperity and
combat climate change at the same time. We argue that an approach to
growth driven by clean innovation and investment can create new growth
and employment opportunities. The economic, structural, and technologi-
cal challenges of sustainable growth are massive, but the opportunities are
real and very attractive.
However, time is short. Over the next two decades, the emerging mar-
kets of Asia, Africa, and Latin America will build their cities, infrastructure,
and energy systems. Developed nations will need a major renewal of theirs.
The way we make decisions on these issues will determine whether we have
a chance of keeping climate change well below 2° C.
There is some reason for optimism. In the Paris Agreement (December
2015) and the Sustainable Development Goals (September 2015), the inter-
national community now has a platform through which climate change,
environment, and development can be integrated into planning, financ-
ing, and investment decisions. We have a global agenda for the first time in
which virtually all countries are involved.
To guide these decisions, we call for a radical deepening of economic analy-
sis. Climate change is the biggest and most important example of systemic
global risk, but it is not the only one, and we, in economics, have to learn
to think about and investigate these issues much more carefully. Standard
growth theory, general equilibrium, and marginal methods will, as ever,
have much to contribute. But they will not be sufficient. We should seek
a dynamic economics where we tackle directly issues involving pace and
scale of change in the context of major and systemic risks.
We also call for a departure from development business as usual. Poor coun-
up demand for modern forms of energy, transport,
tries have a large pent-
carbon
and essential consumption goods that must now be met in a low-
way. They will suffer most from the adverse effects of climate change and
need a form of economic development that manages their climate expo-
sure and increases their capacity to adapt. A key focus must be investment
in sustainable infrastructure. The world needs strong and clear policies to
foster those investments and a major expansion in finance to undertake
them. With their range of instruments, the confidence inspired by their
Climate Change, Development, Poverty, and Economics 315
term view, the development banks
presence, and the ability to take a long-
have a vital role to play.
Managing climate change and reducing poverty are the defining chal-
first century. Both can be tackled, and the alternative
lenges of the twenty-
paths to sustainable growth are very attractive. We know what needs to be
done, we know how to begin, and we will learn along the way.
References
Aghion, Philippe, Antoine Dechezleprêtre, David Hémous, Ralf Martin, and John
Van Reenen. 2016. “Carbon Taxes, Path Dependency, and Directed Technical
51.
Change: Evidence from the Auto Industry.” Journal of Political Economy 124 (1): 1–
Barrett, Scott. 2003. Environment and Statecraft: The Strategy of Environmental Treaty-
Making. Oxford: Oxford University Press.
Bhattacharya, Amar, Jeremy Oppenheim, and Nicholas H. Stern. 2015. “Driving
Sustainable Development through Better Infrastructure: Key Elements of a Trans-
formation Program.” Global Working Paper 91, Brookings Institution, Washing-
ton, DC.
Boulding, Kenneth E. 1966. “The Economics of the Coming Spaceship Earth.” In
Environmental Quality in a Growing Economy: Essays from the Sixth RFF Forum, edited
14. Baltimore: Resources for the Future, Johns Hopkins Univer-
by Henry Jarrett, 3–
sity Press.
Bowen, Alex, and Karlygash Kuralbayeva. 2015. “Looking for Green Jobs: The Impact
of Green Growth on Employment.” Policy brief, Grantham Research Institute on
Climate Change and the Environment, London School of Economics and Political
Science, London.
Braithwaite, John, and Peter Drahos. 2001. Global Business Regulation. Cambridge:
Cambridge University Press.
Linear
Burke, Marshall, Solomon M. Hsiang, and Edward Miguel. 2015. “Global Non-
Effect of Temperature on Economic Production.” Nature 527 (7577): 235.
Cline, William. 1992. The Economics of Global Warming. Washington, DC: Peterson
Institute for International Economics.
Coady, David, Ian Parry, Louis Sears, and Baoping Shang. 2015. “How Large Are
Global Energy Subsidies?” IMF Working Paper 15/105, International Monetary
Fund, Washington, DC.
Coase, Ronald H. 1960. “The Problem of Social Cost.” Journal of Law & Economics
3 (October): 1–44.
316 Sam Fankhauser and Nicholas Stern
Dechezleprêtre, Antoine, and Misato Sato. 2014. “The Impacts of Environmental
Regulations on Competitiveness.” Policy brief, Grantham Research Institute on
Climate Change and the Environment, London School of Economics and Political
Science, London.
Dechezleprêtre, Antoine, Ralf Martin, and Samuela Bassi. 2016. “Climate Change
Policy, Innovation and Growth.” Policy brief, Grantham Research Institute on
Climate Change and the Environment, London School of Economics and Political
Science, London.
Dechezleprêtre, Antoine, Ralf Martin, and Myra Mohnen. 2014. “Knowledge Spill-
overs from Clean and Dirty Technologies: A Patent Citation Analysis.” Working
Paper 135, Grantham Research Institute on Climate Change and the Environment,
London School of Economics and Political Science, London.
Dercon, Stefan. 2012. “Is Green Growth Good for the Poor?” Policy Research Work-
ing Paper 6231, World Bank, Washington, DC.
Drèze, Jean, and Nicholas H. Stern. 1987. “The Theory of Cost- Benefit Analysis.”
In Handbook of Public Economics, volume 2, edited by Alan J. Auerbach and Martin
989. Amsterdam: North-
Feldstein, 909– Holland.
Drèze, Jean, and Nicholas H. Stern. 1990. “Policy Reform, Shadow Prices, and Market
45.
Prices.” Journal of Public Economics 42 (1): 1–
Drijfhout, Sybren, Sebastian Bathiany, Claudie Beaulieu, Victor Brovkin, Martin
Claussen, Chris Huntingford, Marten Scheffer, Giovanni Sgubin, and Didier Swinge-
douw. 2015. “Catalogue of Abrupt Shifts in Intergovernmental Panel on Climate
Change Climate Models.” Proceedings of the National Academy of Sciences 112 (43):
E5777–E5786.
Edmonds, Jae, and John Reilly. 1983. “A Long-Term Global Energy- Economic Model
88.
of Carbon Dioxide Release from Fossil Fuel Use.” Energy Economics 5 (2): 74–
Fankhauser, Sam. 1995. Valuing Climate Change: The Economics of the Greenhouse.
London: Earthscan.
Carbon Economy: Lessons
Fankhauser, Sam. 2013. “A Practitioner’s Guide to a Low-
362.
from the UK.” Climate Policy 13 (3): 345–
Fankhauser, Sam, and Frank Jotzo. 2017. “Economic Growth and Development with
Low-Carbon Energy.” Wiley Interdisciplinary Review Climate Change: e495.
Fankhauser, Sam, and Thomas K. J. McDermott. 2014. “Understanding the Adapta-
tion Deficit: Why Are Poor Countries More Vulnerable to Climate Events Than Rich
18.
Countries?” Global Environmental Change 27 (Supplement C): 9–
Fankhauser, Sam, and Thomas K. J. McDermott, eds. 2016. The Economics of Climate-
Resilient Development. Cheltenham, UK: Edward Elgar.
Climate Change, Development, Poverty, and Economics 317
Freeman, Jody, and Charles D. Kolstad. 2007. Moving to Markets in Environmental
Regulation: Lessons from Twenty Years of Experience. Oxford: Oxford University Press.
Hallegatte, Stephane, Mook Bangalore, Laura Bonzanigo, Marianne Fay, Tamaro
Schilb. 2016.
Kane, Ulf Narloch, Julie Rozenberg, David Treguer, and Adrien Vogt-
Shock Waves: Managing the Impacts of Climate Change on Poverty. Washington, DC:
World Bank.
Hanley, Nick, and Edward B. Barbier. 2009. Pricing Nature: Cost-
Benefit Analysis and
Environmental Policy. Cheltenham, UK: Edward Elgar.
Hanson, Susan, Robert Nicholls, N. Ranger, S. Hallegatte, J. Corfee-Morlot, C. Her-
weijer, and J. Chateau. 2011. “A Global Ranking of Port Cities with High Exposure to
Climate Extremes.” Climatic Change 104 (1): 89–111.
Hartwick, John M. 1977. “Intergenerational Equity and the Investing of Rents from
974.
Exhaustible Resources.” American Economic Review 67 (5): 972–
Heal, Geoffrey, and Jisung Park. 2013. “Feeling the Heat: Temperature, Physiology &
the Wealth of Nations.” NBER Working Paper 19725, National Bureau of Economic
Research, Cambridge, MA.
Gustaf Löfgren, and Bo Sandelin. 2006.
Hedlund-Nyström, Torun, Lars Jonung, Karl-
“Knut Wicksell on Forestry: A Note.” In Swedish Economic Thought: Explorations and
65. London and New York: Routledge.
Advances, edited by Lars Jonung, 46–
Hotelling, Harold. 1931. “The Economics of Exhaustible Resources.” Journal of Politi-
175.
cal Economy 39 (2): 137–
Hsiang, Solomon M., and Marshall Burke. 2014. “Climate, Conflict, and Social Sta-
55.
bility: What Does the Evidence Say?” Climatic Change 123 (1): 39–
IPCC (Intergovernmental Panel on Climate Change). 2007. Climate Change 2007:
Synthesis Report. Contribution of Working Groups I, II and III to the Fourth Assessment
Report of the Intergovernmental Panel on Climate Change. Geneva: Intergovernmental
Panel on Climate Change.
IPCC (Intergovernmental Panel on Climate Change). 2013. Climate Change 2013:
The Physical Science Basis: Working Group I Contribution to the Fifth Assessment Report
of the Intergovernmental Panel on Climate Change. Cambridge: Cambridge University
Press.
IPCC (Intergovernmental Panel on Climate Change). 2014. Climate Change 2014:
Impacts, Adaptation and Vulnerability. Part A: Global and Sectoral Aspects. Contribution
of Working Group II to the Fifth Assessment Report of the Intergovernmental Panel on
Climate Change. Cambridge: Cambridge University Press.
IPCC (Intergovernmental Panel on Climate Change). 2018. Global Warming of 1.5°.
An IPCC Special Report, Intergovernmental Panel on Climate Change, Geneva.
318 Sam Fankhauser and Nicholas Stern
Jackson, Tim. 2011. Prosperity without Growth: Economics for a Finite Planet. London:
Routledge.
Kelley, Colin P., Shahrzad Mohtadi, Mark A. Cane, Richard Seager, and Yochanan
Kushnir. 2015. “Climate Change in the Fertile Crescent and Implications of the Recent
3246.
Syrian Drought.” Proceedings of the National Academy of Sciences 112 (11): 3241–
Keohane, Robert O., and David G. Victor. 2016. “Cooperation and Discord in Global
575.
Climate Policy.” Nature Climate Change 6 (6): 570–
Klein, Naomi. 2015. This Changes Everything: Capitalism vs. The Climate. New York:
Simon & Schuster.
Little, Ian, Malcolm David, and James A. Mirrlees. 1974. Project Appraisal and Plan-
ning for Developing Countries. New York: Basic Books.
Manne, Alan Sussmann, and Richard G. Richels. 1992. Buying Greenhouse Insurance:
The Economic Costs of Carbon Dioxide Emission Limits. Cambridge, MA: MIT Press.
Meade, James Edward. 1955. Trade and Welfare. London, New York: Oxford Univer-
sity Press.
Meadows, Donella H., Dennis L. Meadows, Jørgen Randers, and William W. Behrens.
1972. The Limits to Growth. New York: Universe Books.
Montesquieu, Charles Baron de. [1748] 2011. The Spirit of Laws. Translated by
Thomas Nugent. New York: Cosimo Classics.
New Climate Economy. 2014. Better Growth, Better Climate: The New Climate Economy
Report. London: Global Commission on the Economy and Climate.
Nordhaus, William D. 1982. “How Fast Should We Graze the Global Commons?”
246.
American Economic Review 72 (2): 242–
Nordhaus, William D. 1991a. “A Sketch of the Economics of the Greenhouse Effect.”
150.
American Economic Review 81 (2): 146–
Nordhaus, William D. 1991b. “To Slow or Not to Slow: The Economics of the Green-
937.
house Effect.” Economic Journal 101 (407): 920–
OECD (Organisation for Economic Co- operation and Development). 2015. OECD Com-
panion to the Inventory of Support Measures for Fossil Fuels 2015. Paris: OECD Publishing.
Pearce, David W. 2002. “An Intellectual History of Environmental Economics.”
81.
Annual Review of Energy and the Environment 27: 57–
Pearce, David W., and Jeremy J. Warford. 1993. World without End: Economics, Envi-
ronment, and Sustainable Development. Oxford: Oxford University Press.
Pfeiffer, Alexander, Richard Millar, Cameron Hepburn, and Eric Beinhocker. 2016.
“The ‘2°C Capital Stock’ for Electricity Generation: Committed Cumulative Carbon
Climate Change, Development, Poverty, and Economics 319
Emissions from the Electricity Generation Sector and the Transition to a Green
1408.
Economy.” Applied Energy 179 (Supplement C): 1395–
Pigou, Arthur C. 1920. The Economics of Welfare. London: Macmillan.
Rhode, Robert A., and Richard A. Muller. 2015. “Air Pollution in China: Mapping of
Concentrations and Sources.” PLoS ONE 10 (8): e0135749.
Rockström, Johan, Will Steffen, Kevin Noone, Åsa Persson, F. Stuart Chapin III, Eric
F. Lambin, Timothy M. Lenton, et al. 2009. “A Safe Operating Space for Humanity.”
475.
Nature 461 (7263): 472–
Rogelj, J., M. Den Elzen, N. Höhne, T. Fransen, H. Fekete, H. Winkler, R. Schaeffer, F.
Sha, K. Riahi and M. Meinshausen. 2016. “Paris Agreement Climate Proposals Need
a Boost to Keep Warming Well Below 2 C.” Nature 534 (7609): 631–639.
Sandmo, Agnar. 2015. “The Early History of Environmental Economics.” Review of
63.
Environmental Economics and Policy 9 (1): 43–
Schelling, Thomas C. 1992. “Some Economics of Global Warming.” American Eco-
14.
nomic Review 82 (1): 1–
Schelling, Thomas C. 1997. “The Cost of Combating Global Warming.” Foreign
14.
Affairs 76 (6): 8–
Schlenker, Wolfram, and David B. Lobell. 2010. “Robust Negative Impacts of
Climate Change on African Agriculture.” Environmental Research Letters 5 (1):
014010.
Schlenker, Wolfram, W. Michael Hanemann, and Anthony C. Fisher. 2005. “Will
U.S. Agriculture Really Benefit from Global Warming? Accounting for Irrigation in
406.
the Hedonic Approach.” American Economic Review 95 (1): 395–
Solow, Robert M. 1974. “Intergenerational Equity and Exhaustible Resources.”
45.
Review of Economic Studies 41 (5): 29–
Stern, Nicholas H. 2007. The Economics of Climate Change: The Stern Review. Cam-
bridge: Cambridge University Press
Stern, Nicholas H. 2015. Why Are We Waiting?: The Logic, Urgency, and Promise of
Tackling Climate Change. Cambridge, MA: MIT Press.
Stern, Nicholas H. 2016. “Economics: Current Climate Models Are Grossly Mislead-
409.
ing.” Nature News 530 (7591): 407–
Sterner, Thomas. 2003. Policy Instruments for Environmental and Natural Resource Man-
agement. Washington, DC: Resources for the Future.
TEEB (The Economics of Ecosystems and Biodiversity). 2010. The Economics of Eco-
systems and Biodiversity. London: Earthscan.
320 Sam Fankhauser and Nicholas Stern
Tol, Richard S. J., and Sam Fankhauser. 1998. “On the Representation of Impact in
Integrated Assessment Models of Climate Change.” Environmental Modeling & Assess-
2): 63–
ment 3 (1– 74.
Weitzman, Martin L. 2012. “GHG Targets as Insurance Against Catastrophic Climate
244.
Damages.” Journal of Public Economic Theory 14 (2): 221–
World Bank. 2010. World Development Report 2010: Development and Climate Change.
Washington, DC: World Bank.
World Bank. 2011. The Changing Wealth of Nations: Measuring Sustainable Develop-
ment in the New Millennium. Washington, DC: World Bank.
World Bank. 2012. Turn Down the Heat: Why a 4°C Warmer World Must Be Avoided.
Washington, DC: World Bank.
World Commission on Environment and Development (Brundtland Commission).
1987. Our Common Future. Oxford: Oxford University Press.
World Resources Institute. 2014. “CAIT Climate Database.” http://cait.wri.org/.
Comment: Michael Toman
In their paper, Fankhauser and Stern (hereafter FS) do a fine job of dem-
onstrating the urgent need to address the threat of global climate change,
a view that I very much share. Climate change will be a particular threat
for World Bank client countries with greater vulnerabilities due to their
lying coastal areas); the prevalence of at-
location (e.g., in low- risk sectors
productivity subsistence agriculture);
in their economic activity (e.g., low-
a lower level of access to more resilient technologies; and less developed
institutional capacities for adapting to climate change (e.g., in delivering
public health programs). Similar arguments have also been made in a flag-
ship report by the Bank on climate change and poverty risks (Hallegatte et
al. 2016). Although many of the major risks will materialize in the future,
inertia in the earth’s climate system and in the adjustment of capital stocks
in the economy mean that actions need to start in earnest now to stem the
risks, even though their magnitudes are uncertain.
FS also argue that now is the time for a major push to stem climate change
risks through deep and rapid cuts in global greenhouse gas emissions. For
though
reasons described below, I am less sanguine about this possibility—
I would be glad to be wrong. FS base their conclusion on several premises:
1. The ethical argument for the responsibility of this generation to protect
future generations from the serious adverse effects of climate change is
unambiguous.
2. The political aspects of obtaining international agreement on concerted
action to mitigate greenhouse gas emissions have become more favor-
able (New Climate Economy 2015), particularly in light of the Paris
Agreement established at the United Nations conference on climate
change in late 2015 (UNFCCC 2015, Addendum).
322 Comments by Michael Toman and Gaël Giraud
3. Rapid decarbonization can be undertaken in ways that actually create
economic opportunity over the medium and longer terms, for develop-
ing and developed countries alike, through new opportunities for tech-
nical advance and creation of markets for new goods and services (New
Climate Economy 2014).
term benefits of mitigating greenhouse gas emissions as
4. There are near-
most notably, “co-
well— benefits” obtained when switching to renewable
energy and improving energy efficiency reduce local pollutants from fos-
sil fuel burning that damage human health and the environment.
With respect to the first point, I do not think there is yet a widely shared
view of what it means in practice to assume an intergenerational responsi-
bility. Is the obligation of the current generation to do as much as possible
to mitigate cumulative emissions in an attempt to forestall catastrophic
impacts of climate change, an option discussed in (Barrett 2013)? What is
the responsibility to reduce noncatastrophic risks as well? Are there differ-
ent ethical obligations between mitigating greenhouse gas emissions and
strengthening resilience to climate change?
Greater complexity comes in addressing unavoidable questions about
term costs of emissions mitigation and improved resilience to
how nearer-
climate change are to be shared among members of the current generation.
Almost 25 years of analytical work and policy wrangling have not led to
sharing issue, other than the general rec-
practical resolution of the burden-
off countries should carry more of the burden. Funding
ognition that better-
sharing remains inadequate, and there continues to be advocacy
for cost-
carbon energy projects in low-
for expensive low- income developing coun-
tries whose contributions to global emissions are minimal.
With respect to the second point, the degree of engagement among
developing and developed countries in the 2015 Paris Agreement is indeed
a significant achievement. Going forward, it remains to be seen how well
countries do in implementing their “Nationally Determined Contribu-
tions” (NDCs) to reducing global greenhouse gas emissions. Moreover,
mitigating the serious risks from climate change will require substantially
deeper cuts than will follow even under full implementation of NDCs. The
basic paradox of international agreements holds: Finding ways to agree on
and deliver significant mitigation commitments across many countries is
quite difficult.
Climate Change, Development, Poverty, and Economics 323
With respect to the third point, analysis reported in the most recent IPCC
assessment indicates relatively modest cumulative effects on consumption
over time from greenhouse gas mitigation, if everything goes right (IPCC
2014, table SPM2). That means the ready availability and public accep-
effective decarbonization technologies that remain controver-
tance of cost-
sial (notably geological carbon sequestration, as well as greatly expanded
effective coordinated imple-
nuclear power). It also means extremely cost-
mentation of national policies to curb greenhouse gas emissions. Costs are
considerably higher if these strong assumptions do not hold.
Beyond these challenges, it is important to be circumspect about the
economics of rapidly and massively scaling up decarbonization. A great
deal can be accomplished with improvements in energy efficiency. On the
other hand, although solar power in particular seems to be increasingly
through
inexpensive these days, the cost of overcoming intermittency—
up fossil fuel generation, smart grids (which help
combinations of back-
costly storage—
only for uncorrelated intermittency), and evolving but still-
also must be taken into account.
The 2014 New Climate Economy report makes much of the broader pos-
sibilities for “creative destruction” from more stringent limits on green-
house gas emissions, leading to economic gains from increased innovation
and new markets. I think the breadth of applicability of this argument
needs further validation. Although retiring a significant amount of fossil-
based power generation capacity would lead to expanded markets for
fuel-
replacement technologies and competitive gains for some suppliers, such
term win across the board. How much
a policy is not likely to be a near-
innovation would take place also depends critically on the extent to which
greenhouse gases are appropriately priced, and what complementary poli-
cies for supporting basic and applied R&D are deployed.
The fourth point is a popular argument in climate policy debates, but I
think we need to consider more carefully the economic and ethical aspects
benefits as an argument for greenhouse gas
of counting environmental co-
mitigation. Developing countries currently face numerous environmen-
tal challenges, including major public health threats from air pollution.
effectively with established
However, air pollution can be reduced cost-
technologies, without the delay or uncertain cost associated with scaling-
up low-
polluting renewable alternatives to fossil energy. Why not make
the strong economic case for cutting these emissions anyway, regardless of
324 Comments by Michael Toman and Gaël Giraud
carbon energy? From an ethical perspec-
what is done with respect to low-
tive, there are intense debates about who has a greater responsibility to pay
for steps to cut current greenhouse gases in order to protect the welfare of
future generations. What can we say about the morality of not pushing
saving pollution control
for readily available and relatively affordable life-
measures today?
FS make the valid and important point that macro and micro scales of
analysis need to be better integrated for assessing greenhouse gas mitiga-
tion possibilities and for enhancing resilience to climate change. What is
needed is more of an “environmental macroeconomics” than is currently
within the scope of environmental and natural resource economics. They
also argue that at this juncture, it is important to “get the big decisions
like how to implement carbon pricing and increase assistance for
right”—
adaptation measures.
To have a realistic chance to make the deep cuts in future global green-
house gas emissions that FS rightly advocate, the development of a favor-
able technological environment is crucial. There is a vital need especially
competitive low-
to provide more cost- carbon energy technology options.
carbon energy sources—
Low- renewables, nuclear, and fossil energy use
must increase from less than 20 per-
with carbon capture and storage—
cent of total energy use to more than 70 percent or even 90 percent by
2100, depending on the stringency of the limit on temperature increase
sought (IPCC 2014, figure 7.16). Such a transformation will not be possible
carbon
without fundamental changes in the cost and performance of low-
energy technologies.
The call by some prominent observers (including Stern) for a “Global
Apollo Programme to Tackle Climate Change” (King et al. 2015) draws
welcome attention to the need for greatly expanding international R&D
for greenhouse gas mitigation. The proposal is to do this through volun-
Carbon Technology Innovation Club.”
tary participation in a kind of “Low-
Keohane and Victor (2016) describe in more detail such an approach for
international cooperation to develop technologies needed for deep cuts in
greenhouse gas emissions, as part of a larger framework for different types
of climate change policy coordination. However, the initial target proposed
by King et al. (2015) of $15 billion per year, or about 0.02 percent of global
GDP, is roughly an order of magnitude smaller than the required invest-
ment levels per year that the International Energy Agency has calculated to
Climate Change, Development, Poverty, and Economics 325
carbon transition (IEA 2014). How to mobilize such
be necessary for a low-
large sums of money in order to make rapid and deep cuts in global green-
unanswered question.
house gas emissions is an urgent but still-
References
Barrett, Scott. 2013. “Climate Treaties and Approaching Catastrophes.” Journal of
250.
Environmental Economics and Management 66 (2): 235–
Hallegatte, Stephane, Mook Bangalore, Laura Bonzanigo, Marianne Fay, Tamaro
Schilb. 2016.
Kane, Ulf Narloch, Julie Rozenberg, David Treguer, and Adrien Vogt-
Shock Waves: Managing the Impacts of Climate Change on Poverty. Washington, DC:
World Bank.
IEA (International Energy Agency). 2014. World Energy Investment Outlook. Paris:
International Energy Agency. https://www.iea.org/publications/freepublications/public
ation/WEIO2014.pdf.
IPCC (Intergovernmental Panel on Climate Change). 2014. Climate Change 2014:
Mitigation of Climate Change. Contribution of Working Group III to the Fifth Assessment
Report of the Intergovernmental Panel on Climate Change. Cambridge: Cambridge Uni-
versity Press.
Keohane, Robert, and David Victor. 2016. “Cooperation and Discord in Global Cli-
575.
mate Policy.” Nature Climate Change 6 (6): 570–
King, David, John Browne, Richard Layard, Gus O’Donnell, Martin Rees, Nicholas
Stern, and Adair Turner. 2015. “A Global Apollo Programme to Combat Climate
Change.” London: London School of Economics, Centre for Economic Performance.
http://cep.lse.ac.uk/pubs/download/special/Global_Apollo_Programme_Report.pdf).
New Climate Economy. 2014. Better Growth, Better Climate: Charting a New Path for
Carbon Growth and a Safer Climate. London: Global Commission on the Econ-
Low-
omy and Climate. http://newclimateeconomy.report/2014/.
New Climate Economy. 2015. Seizing the Global Opportunity: Partnerships for Better
Growth and a Better Climate. London: Global Commission on the Economy and Cli-
mate. http://newclimateeconomy.report/2015/.
UNFCCC (United Nations Framework Convention on Climate Change). 2015.
“Report of the Conference of the Parties on Its Twenty- First Session,” Paris, Novem-
ber 30– December 15, Addendum. Document FCCC/CP/2015/10/Add.1. http://
unfccc.int/resource/docs/2015/cop21/eng/10a01.pdf.
Comment: Gaël Giraud
The Trouble with Climate Economics
Here I briefly comment on the main points raised by the nice and thought-
provoking paper by Sam Fankhauser and Nicholas Stern (FS hereafter). The
next section deals with the ongoing debate on the seriousness of economic
damages induced by climate change. I argue in section 2 that the gravity of
the physical risk creates a funding problem that can hardly be expected to
be solved solely by conventional means, such as national budgets. Section 3
provides some brief thoughts on the ethical questions raised by FS. Section
4 echoes the strong call made by the authors for a “radical deepening” of
integrated economic models aimed at assessing the impact of global warm-
ing (and how we can avoid its disastrous effects). For that purpose, the last
section offers a tentative suggestion of a dynamic model that could be used
or an alternative—
as a complement— to more conventional ones.
Climate: It’s Serious!
The first and main lesson to be taken away from the FS paper is pretty
clear: Economic damages caused by global warming are probably going to
be considerably greater than our current economic models predict. This
makes it more important than ever to take urgent and drastic action to
curb temperature change by reducing carbon emissions. What is more,
the authors emphasize the “double inequity” that plagues the challenge
This work benefited from the support of the Energy and Prosperity Chair, under
the aegis of the Fondation du Risque (Institut Louis Bachelier, 28 place de la Bourse,
75002 Paris, France). All errors are, of course, mine.
Climate Change, Development, Poverty, and Economics 327
of coping with climate change: Rich countries are responsible for most of
the current stock of greenhouse gases in the atmosphere, but poor people
in southern countries (and to a lesser extent, in northern ones) will be hit
earliest and hardest. On this issue, the index for physical vulnerability to
climate change provides an interesting, albeit perfectible, tool for measur-
ing the exposure of poor countries to the consequences of global warming
(Guillaumont 2013). Figure 7.1 illustrates the geographical distribution of
physical climate risk, as estimated according to this index.
Even a country like France is acutely concerned, through its overseas
geographies (Goujon, Hoarau, and Rivière 2015) of course, but also with
respect to its metropolitan territory (Le Treut 2013). Hallegatte et al. (2016)
estimate that about 100 million people in the world may be relegated to
below the poverty line by 2030 because of climate change. Obviously, as
stressed by FS, “mitigation, adaptation, and development are intertwined,”
race” between climate policy and development repre-
such that the “horse-
sents a “false dichotomy.” Some concrete experiences confirm that devel-
and actually ought to—
opment and climate policy can— be achieved at
the same time. Many of the projects in which Agence Française de Dével-
oppement (AFD) is involved reflect this conjugacy, from urban planning
(in Porto Novo, Benin, or the Philippines) and addressing rising sea levels,
to building the solar power plant near Ouarzazate (Morocco). Additional
projects in Zimbabwe or sanitation
examples include agroecological micro-
programs in the slums of Santo Domingo’s Barquita district, aimed at chil-
dren suffering from leptospirosis, a disease spread by alternating periods of
drought and devastating typhoons. As a consequence, adaptation to global
warming and resilience are of utmost importance for southern countries,
whereas mitigation should be a priority for emerging and advanced econo-
mies. Unfortunately, this does not mean that developing countries could be
exempted from any efforts regarding mitigation. Greenhouse gas emissions
Saharan Africa today represent less than 3.4 percent
stemming from Sub-
of the world’s emissions. But Liousse et al. (2014) suggest that by 2030,
this continent’s contribution could account for up to 20 percent of global
at least in a business-
emissions, or even more— usual scenario. Thus,
as-
even for some countries that have not yet emerged, a path toward emer-
gence that would simply mimic Western “dirty” production modes and life
style should not be considered a valid option. This is particularly true in
fired power plants—
Asia, where the already planned coal- if they do indeed
Afghanistan
HaiƟ
The Gambia
Guinea
Sierra Somalia
Leone
Nigeria Rwanda
Chad Burundi
Legend
Extreme risk
High risk
Medium risk
Low risk
No Data
Figure 7.1
Physical vulnerability to climate change
Source: Guillaumont (2013).
Climate Change, Development, Poverty, and Economics 329
would absorb the entire carbon budget
start operating in the near future—
left available at world level, if we want the average planetary temperature
increase to have reasonable chances of remaining below 2° C.
On this count, my feeling is that we urgently need more data on the
regional and local impacts of climate change: Global integrated assessment
models, however powerful they might be, will remain of middling help for
the political agenda as long as we are not able to increase the granularity of
our understanding of the consequences of global warming. Climatologists
are devoting valuable efforts to this central issue: Vautard et al. (2014) and
Le Treut (2013), among many others, show that, at least for some territories,
it is possible to get a relatively clear picture of the consequences of climate
change in the foreseeable future, provided a truly interdisciplinary meth-
odology is adopted.
Reducing greenhouse gases is far from easy, but efficient adaptation
is actually an even more challenging task, because resilience to climate
change means shaping infrastructure and institutions so that they evolve
according to a phenomenon that is itself dynamic and highly nonlinear. A
single example can illustrate this point: The coast of Danang and Hoïan,
in Vietnam, is heavily eroded by the rise of the sea level. One immedi-
inspired, say, by the secular experience
ate answer that comes to mind—
would involve building dikes so as to protect the coast.
of Dutch polders—
sighted and even counterproduc-
This, however, might prove to be a short-
tive answer. Indeed, as the sea level rises, the direction of flows and waves
might change in the coming decades. Being the result of complex turbu-
Stokes partial differential
lence phenomena related to the nonlinear Navier-
equation, these changes are hard to predict. Dikes that would be efficient
term might promote a disaster in the medium run. A smart
in the short-
answer therefore calls for some kind of adaptive process. It seems to me
that we are just beginning to realize how demanding this challenge is.
Let me close this section on the physical risks arising from the coming
related events by
increased frequency and severity of climate-and weather-
stressing one particularly important point that might well be overlooked
in a hasty reading of the FS paper. Mentioning the celebrated debate about
Malthusian pessimism, the authors rightly argue: “So far, Malthus and the
resource pessimists have generally appeared to be wrong. Human ingenuity
has mostly managed to outpace natural resource constraints.”
330 Comments by Michael Toman and Gaël Giraud
That the carefulness of this statement is not a mere rhetorical precaution
third report to the Club of
is confirmed by the conclusions of the thirty-
Rome (Bardi 2014): Today, the world’s mining industry is already starting
to show worrying signs of difficulty. The mineral resources that are the least
expensive to extract and process have mostly been exploited and depleted.1
Whilst there are plenty of minerals left to extract, they will come at higher
financial and energy cost and be increasingly difficult to extract. Thus, the
depletion of minerals (in the economic rather than geological sense, mean-
ing the unsustainable cost of today’s plundering of the planet) has to be
weighed up when planning the path towards societies based on renewable
energies (Vidal, Goffé, and Arndt 2013; Giraud 2014).2
Mobilizing Climate Finance
Insurers are on the frontline of physical risks. This engagement is illustrated
a partnership formed in 2015
by the Insurance Development Forum—
between the UN Development Programme, the World Bank, and the insur-
ance sector with the intention of using the industry’s expertise to insure
people in developing countries who are unprotected but vulnerable to cli-
mate change risk. According to Bank of England’s Governor Mark Carney,
“this protection gap currently represents 90 percent of the economic costs
of natural disasters that are uninsured.”3
But beyond the physical risk, and because of its very gravity, the finan-
cial stake should not be neglected. As argued by Carney, too rapid a move-
carbon economy could materially damage financial
ment toward a low-
assessment of prospects, as climate-
stability: “A wholesale re- related risks
evaluated, could destabilise markets, spark a pro-
are re- cyclical crystallisa-
tion of losses and lead to a persistent tightening of financial conditions: a
climate Minsky moment” (Carney 2016). Conversely, insufficient adoption
1. To take the example of copper (a widely used mineral still difficult to substitute
in many industrial applications), the density of copper resources exploited so far had
been greater than 5 percent on average. That of today’s remaining resources is at
most 1 percent (Vidal, Goffé, and Arndt 2013).
2. Depletion is not the only problem: pollution induced by mining takes many forms
and produces many consequences, including the aggravation of climate change.
3. Carney (2016).
Climate Change, Development, Poverty, and Economics 331
of adequate financial tools may prevent the world economy from investing
at the required scale.
The strong warnings expressed by FS are in line with those of the Bank
of England’s governor, as well as with the message put forward by the New
Climate Economy (2014) report. According to the latter, US$90 trillion are
needed at the world level over the next 15 years to fund clean infrastruc-
income countries, and between US$3
tures; US$2 trillion per year in high-
income countries. These numbers prompt
and 4 trillion in low-and middle-
a daunting question: How will the world economy finance such monetary
flows? The first difficulty lies probably in the huge Knightian uncertainty
benefit analysis of the opportunity to devote costly
that plagues any cost-
efforts today to addressing climate change challenges.
FS rightly claim that the international community needs now to “get the
big decisions right.” One could object, however, that given the pervasive
deep uncertainty we are facing, big decisions might also lead to big mis-
takes. At the analytical level at least, this issue has been successfully tackled
in the field of financial measures of risk. Value at risk, as is well known,
provides a poor measure of the tail of a risk distribution. However, Artzner
et al. (1999) laid the axiomatic foundation of a family of alternative coher-
ent risk measures, whose essence is the following. In a situation where we
do not even know with sufficient accuracy the probability distribution of
risk, a rational approach consists of envisaging the worst distribution of risk
and optimizing our expected outcome according to it. Thus it would not be
fair, I believe, to claim that deep uncertainty prevents us from taking action
along the lines advocated by FS.
That being said, the question as to how the international community is
going to fund the required financial efforts remains open. The Green Cli-
mate Fund established at the Conference of the Parties (known as COP 16)
in Cancun in 2011 is quite a promising tool but, in its current design, its
size may not suffice to reach an adequate order of magnitude, even when
due account is taken of the leverage effect of additional private capital mar-
kets. Thus, complementary solutions are called for. Two reports published
before and after the Paris agreement (Canfin et al. 2015; Canfin, Grandjean,
and Mestrallet 2016) consider some alternative proposals. Let me just men-
tion two of them.
Canfin, Grandjean, and Mestralle (2016) make a strong case in favor
of orienting international negotiations toward a corridor of carbon prices.
332 Comments by Michael Toman and Gaël Giraud
Indeed, the quest for a unique, universally relevant price is probably a dead
end: Why should the (real) marginal costs of producing 1 ton of carbon be
sectional differences between
equal across countries? Beyond obvious cross-
national industry and agricultural sectors, the lack of methodological
robustness surrounding the purchasing power parity calculus and the long-
standing noncoincidence of these rates with market exchange rates are well
known. There is probably very little hope of ever being able to identify “the”
market carbon price that would provide the right incentives for efficient
decarbonization in Maputo, Buenos Aires, or Osaka, for example. Moreover,
the financial transfers from northern to southern countries that would be
required to compensate for the losses incurred by the latter seem to exceed
the limits of any politically reasonable transaction. In contrast, the corridor
approach requires the international community to agree on three variables:
a cap, a floor, and the slope of the tubular neighborhood (i.e., the speed at
and-
which the median price would increase, keeping the cap- floor diam-
50 interval,
eter constant). At the time these lines were written, a US$20–
together with a 5 percent yearly growth rate seem to be reasonable figures
on which an international consensus would not be out of reach.
Next, Canfin et al. (2015) suggest setting up a financing tool that uses
the ability of the International Monetary Fund to create new international
reserve money in the form of Special Drawing Rights (SDRs). In contrast to
some proposals dealing with SDRs (e.g., Bredenkamp and Pattillo 2010),
the plan of Canfin et al. (2015) is not to create new and additional SDRs
but rather to use already existing ones. In fact, in 2009, the International
Monetary Fund “printed” about US$300 billion to sustain countries shack-
2009 global financial crisis. A large
led by the financial turmoil of the 2008–
fraction of this “money” is stored today as currency reserves and could
blown money provided the countries that received this
be turned into full-
manna in 2009 would agree to convert it and thus pay the (low) interest due
option is exercised.4
to the International Monetary Fund as soon as the SDR-
This is admittedly quite an unconventional proposal, and more analysis is
needed to understand its macroeconomic implications.5 It should never-
4. An SDR can indeed be viewed as a call on one of the four currencies into which
SDRs are convertible— with
the US dollar, the euro, the pound sterling, and the yen—
an unspecified maturity.
5. See, however, the section on “The Trouble with Macroeconomics” in these
comments.
Climate Change, Development, Poverty, and Economics 333
theless be clear from FS’s paper that overcoming the climate challenge will
not be cheap. As most countries currently confronted with huge public def-
term climate-
icits are reluctant to spend money on medium- related issues,
a genuinely effective climate policy to reduce global warming as much as
still possible probably has to rely on unconventional tools.
Can Ethical Traditions Cooperate?
As pointed out by FS, when assessing financial risks associated with the tran-
carbon economy, ethical issues inevitably come to the fore.
sition to a low-
Indeed, due to the intergenerational gap between polluters and victims,6
standard incentives (e.g., carbon taxes) are key tools, as ever, but are prob-
ably insufficient to provide the right impetus. Some spiritual or moral
at the cost, however, of having to face today’s prolif-
resources are needed—
eration of spiritual experimentations in our globalized postsecular societies
(Giraud 2015). Could the rich diversity of ethical traditions prevent these
efforts from unifying on the front of the climate change “tragedy” (Carney
2016), and therefore from providing a clear call to action?
On this aspect of the climate change problem, social choice theory
can be helpful. In fact, at least to a first analytical approximation, mod-
ern consequentialist theories of distributive justice can be encapsulated in
two extremal points. On the one hand is the utilitarian viewpoint, which
claims that justice consists of maximizing the average welfare of people’s
normalized utility functions (see, e.g., Dhillon and Mertens 1999);7 on the
other is the Rawlsian (maximin) approach, which asserts that fairness is
best captured by optimizing the fate of the less advantaged citizens (Fleur-
baey and Maniquet 2008). A continuum of intermediate theories of justice
can be conceived, lying somewhere between these two extreme standpoints
6. One could also add the geographic gap that prevailed until recently between
polluters (mostly in the north) and their contemporaneous victims (mostly in the
south). But the magnitude of this second gap is currently shrinking, as emerging
economies are now contributing more to greenhouse gas emissions than countries
from the Old World, as FS remind us.
7. Citizens’ utility functions need to be normalized in some way or other, because
otherwise, the arbitrariness of the cardinal representation of ordinal preferences
potentially leads to distortions in the respective weight of each individual utility. In
a broad sense, Dhillon and Mertens (1999) essentially offer a quite general axiomatic
that leads to a unique, well-defined normalization procedure.
334 Comments by Michael Toman and Gaël Giraud
(Giraud and Gupta 2016). With each of these theories, a specific social wel-
fare function can be associated, whose optimization (under standard con-
straints) potentially leads to diverging guidelines for action.
For the sake of concreteness, let us examine this point with respect to
the specific (but decisive) issue of choosing the “right” discount rate with
which future expected profits and losses can be valued. As argued by Sterner
and Persson (2008), there actually is no reason to assume a priori that the
discount rate must be constant across time. Let us nevertheless assume that
it is, for the sake of simplicity (and because this is still the current practice
in the financial industry today). Then, if one is utilitarian (in the sense
of Jean-François Mertens’s relative utilitarianism), the discount rate, r, that
should be adopted ought to be equal to the real growth rate, g, of the econo-
my.8 In the context of our current debates, this choice means that the dis-
cussion about the “correct” discount measure boils down to the plausibility
of secular stagnation. If there are good reasons to believe that g will remain
low (and even close to zero) in the future, then there are equally good
at least in a utilitarian Weltanschauung—
reasons— to choose a low (or even
zero) discount rate. For those who, on the contrary, adhere to the Rawlsian
perspective, things might seem to be completely different. But in fact they
are not. Indeed, Roemer (2011) has shown that the “correct” discount rate
that should be deduced from a normative maximin approach is zero. As a
result, the practical difference between two apparently antagonistic ethical
postures, such as utilitarianism versus the Rawlsian viewpoint, might not
be as large as initially suspected.
The Trouble with Macroeconomics
Beyond warning that emissions are presumably going to be very high and,
on top of that, that the economic damage from temperature change will
presumably be much worse than most of the literature would so far admit,
Fankhauser and Stern (2016, 23) argue that the economic models that have
8. In other words, the normalization of citizens’ utility functions boils down to the
unitary normalization of the risk aversion premium (or, geometrically, the curvature of
utility functions), γ, in the “golden rule” formula, r = θ + γg, with θ being the normative
exchange rate between the welfare of today’s generations and that of future generations
(or, equivalently, the psychological rate of time preference). I assume here that θ = 0.
Climate Change, Development, Poverty, and Economics 335
been used to calculate the fiscal fallout from climate change are woefully
inadequate and severely underestimate the scale of the threat: “This is why
we call for a radical deepening of economic analysis, including a develop-
ment economics that begins to understand and incorporate climate change.
Standard growth theory, general equilibrium and marginal methods will, as
ever, have much to contribute but they will be nowhere near sufficient. This
is about immense risks and radical change where time is of the essence. We
should seek a dynamic economics where we tackle directly issues involving
pace and scale of change in the context of major and systemic risks.”
Indeed, several of the standard economic models used so far to assess the
impact of global warming rest on assumptions that simply do not reflect
current knowledge about climate change. The difficulty encountered today
by the community of physicists in their dialog with the scientific tribe of
economists (e.g., in the UN Intergovernmental Panel on Climate Change
circles) is not new, however. It was already acknowledged by Wassily
Leontief in the early 1980s “How long will researchers working in ajoining
…
fields abstain from expressing serious concern about the splendid isola-
tion in which academic economics now finds itself?” (Leontief 1982, 104).
FS’s call for a “radical deepening” is also in line with the even harsher
considerations recently expressed by Narayana Kocherlakota (2016) on
macroeconomics as such:
The premise of “serious” modelling is that macroeconomic research can and
should be grounded in an established body of theory. My own view is that, after
the highly surprising nature of the data flow over the past ten years, this basic
premise of “serious” modelling is wrong: we simply do not have a settled suc-
cessful theory of the macroeconomy. The choices made 25– 40 years ago— made
then for a number of excellent reasons—should not be treated as written in stone
or even in pen. By doing so, we are choking off paths for understanding the
macroeconomy.
The former president of the Federal Reserve of Minneapolis concludes
that we should prefer toy models to “serious modeling.” The difference
between the two lies in their relationships to data and their normative
usage: “Users of toy models can often gauge the magnitude of key forces
using simple calculations. (Mehra and Prescott 1985 is a nice example
of what I have in mind.) But toy models are not designed to allow users
to reach definitive quantitative answers to policy questions of interest”
(Kocherlakota 2016).
336 Comments by Michael Toman and Gaël Giraud
real
The criticism expressed by Romer (2016) about what he calls “post-
macroeconomics” rather nicely complements Kocherlakota’s viewpoint. At
the core of Romer’s critique lies the idea that “macroeconomists got com-
fortable with the idea that fluctuations in macroeconomic aggregates are
caused by imaginary shocks, instead of actions that people take, after Kyd-
land and Prescott (1982) launched the real business cycle (RBC) model.”
Regarding dynamic stochastic general equilibrium (DSGE) models, the
harsh judgment recently formulated by Blanchard (2016) suggests that,
despite being widely used in advising policy makers, this specific class of
depth questioning of contem-
quantitative tools is not immune to the in-
poraneous macroeconomics raised by the past decade of evidence. Even
though, to the best of my knowledge, DSGE models are rarely used for
assessing the economic impact of global warming, some of the critiques
that Blanchard (2016) levels at them also hold for alternative (computable)
in particular, the difficulty of providing a convincing
equilibrium models—
story for price inertia, the lack of robustness of certain Bayesian estima-
tions, and the relative neglect of issues related to the distribution of wealth.
These critiques suggest that FS’s call for a “radical deepening” is actually
part of a larger revision of current macroeconomics. In this context, how-
ever, it raises specific challenges linked to climate and development eco-
nomics. Which features should realistic macro models share if they are to
related assessments?
be used for climate-
First, they probably ought to be based on some nonlinear dynamics.9
Why dynamics? Because, as underlined by FS, the timing of mitigation is
key: We need to find the correct speed at which our economies must transit
carbon institutions. This issue can hardly be dealt with in a
toward low-
static framework. One might add a second reason: because economic resil-
ience requires an adaptive process, as I suggested above. And a third reason:
because fluctuation of most macroeconomic variables is a trivial matter of
9. By this, I mean an out- equilibrium dynamics in the sense given to this word in
of-
the mathematics of dynamical systems after Poincaré, or in recent developments of
thermodynamics. Indeed, although the Boltzman- Gibbs law of classical thermody-
namics is an equilibrium theory, out- equilibrium thermodynamical systems had
of-
only been understood, until recently, in the vicinity of an equilibrium, thanks to
Onsager’s linear formalism. To the best of my knowledge, the first consistent theory
from-
of far- equilibrium (and therefore nonlinear) thermodynamics goes back to
any-
Mallick (2009) (see also the references therein).
Climate Change, Development, Poverty, and Economics 337
fact and, as advocated by Romer (2016), should not be explained by imagi-
which are assumed to temporarily perturb some otherwise
nary shocks—
but rather by the interplay of endogenous forces.
stable fixed point—
Why nonlinear? Because, as also stressed by FS, we unfortunately need
much more than marginal adjustments to address climate issues. The size of
the shift required from our economies is potentially large. Although linear-
ity is often a good proxy for small changes, we need to take due account of
the full nonlinearity of the phenomena at stake when studying the possibil-
ity of large disruptions.
Second, we certainly need these models to make explicit the dynamics of
be it public or private. As already stated, the cost of the energy transi-
debt—
carbon economy might reach US$90 trillion. Undoubt-
tion toward a post-
edly, this immense amount of wealth will require more debt in significant
segments of the world economy. The potentially depressing consequences
of this additional leverage need to be addressed if we want to have a realis-
tic narrative of the energy shift. Moreover, given the nontrivial role played
by money and debt, our models should be able to capture Fisherian debt-
deflation (see Eggertsson and Krugman 2010; Giraud and Pottier 2016) and
the Minskian instability hypothesis (Minsky 1992). This is important for at
least two reasons. In the first place, because Japan, southern Europe, and
possibly a larger number of advanced economies are stuck in a liquidity trap
(mostly resulting from the financial crisis) or are on the verge of becoming
so. This specific situation might impede the funding of the needed green
investments alluded to in the third section above. Any analysis of the way
in which the world economy might address the climate issue but which
neglects the essence of today’s “new normal” (negative interest rates, saving
glut, etc.) would indeed be of little help.
Third, despite its enormous influence on the literature over about four
decades, we may have to give up the mathematical elegance of the ratio-
nal expectations hypothesis. Why? Because of the huge (Knightian) uncer-
tainty surrounding climate change issues. I have already touched on this
topic in section 3 above, but because relaxing rational expectations is so
known) example. As recalled
controversial, let me illustrate it with a (well-
by FS, there is still no consensus in the scientific community regarding the
climate sensitivity that links the increase in CO2 concentration in the atmo-
sphere and the change in average temperature at the surface of the planet.
The parameter capturing this sensitivity (economists would speak in terms
338 Comments by Michael Toman and Gaël Giraud
of elasticity) varies between 1 and 6, depending on the climate model we
are referring to.10
cut indication as to which value is the most
Today, there is no clear-
model that would provide the prob-
probable one. Nor do we have a meta-
ability distribution telling us how likely it is that this parameter takes any
given value. We just do not know.11 So how can prices publicly convey
information that is held by nobody? As public transmission of privately
held information is what rational expectations are all about (Dubey, Geana-
koplos, Shubik 1987), this suggests that rational expectations cannot be the
relevant concept for analyzing climate change issues.
Fourth, markets should not be assumed prima facie to clear automati-
cally. As Joseph Stiglitz made evident in chapter 3 of this book, asymmet-
ric information, hence price stickiness, may prevent markets from clearing
instantaneously, the labor market in particular. Again, a simple example
might help explain why this is crucial for the global warming issue. Some
emerging countries ran large computable models to assess their inten-
tional Nationally Determined Contribution (NDC) for the Paris summit
by December 2015. By now, most of these contributions are no longer just
intentional, but have become genuine NDCs. Almost all of the macro mod-
els that have been used for this exercise fail to specify private debts (often
simply because they rely on the “representative consumer” assumption,
despite ubiquitous emergence phenomena in economics; more on this
below) and, moreover, assume full employment throughout. Now, what
will happen if the path that one of these countries wants to follow to keep
its promises requires its private debt to skyrocket up to, say, 400 percent of
its GDP, together with a 70 percent rate of unemployment (which is hard
to believe will be entirely voluntary)? This country will simply never put its
NDC into practice, because the path that would lead to its fulfilment is sim-
ply politically infeasible. Thus, it is of utmost importance to check whether
carbon economies is compatible with
our narratives of the transition to low-
actual political feasibility. This might require abandoning the elegance of
10. Snyder (2016) even recently argued that climate sensitivity could reach the cata-
strophic value of 9.
11. This contrasts even with quantum mechanics, where Heisenberg’s uncertainty
principle goes hand in hand with a probabilistic theory of where and how fast par-
ticles move.
Climate Change, Development, Poverty, and Economics 339
point theory (e.g., Giraud 2001), but it might be the price
topological fixed-
to pay for making economic science relevant to today’s climate challenges.
That said, we should certainly not throw the baby of general equilib-
rium theory out with the bathwater of unsatisfactory macroeconomics.
and this is my fifth point—
Indeed— we should probably not forget the wis-
fashioned Arrow-
dom of the old- Debreu theory, namely, that economics
exactly in the same way as statistical
does admit emergence phenomena—
physics does. “Emergence” should be understood here more or less as a syn-
onym for complexity, that is, in the following, rather weak, sense: aggregate
behavior may lead to macro-
micro- behavior that cannot be reduced a priori
to that of any “representative” creature. This was precisely the content of
the celebrated results of Sonnenschein, Mantel, and Debreu, published in
pointing continuous vec-
the 1970s (e.g., Sonnenschein 1972): Any inward-
tor field on the positive part of the unit sphere (of normalized prices) can be
chosen economy. My
viewed as the aggregate excess demand of some well-
viewpoint is that there are at least two escape routes from this quandary:
based models (see, e.g., Axelrod 1997)
the numerical simulations of agent-
or a more phenomenological standpoint based on the empirical estimation
of aggregate behavioral functions. I shall end these comments by briefly
introducing this second perspective.
The Nonlinear Dynamics of Debt with Global Warming Economics
Giraud et al. (2016) introduce a toy model (in the sense of Kocherlakota;
flow consistent, nonlinear dynamics. Its
see above) based on some stock-
run Phillips curve relating the
basic building blocks are provided by a short-
growth rate of nominal wages to underemployment (Mankiw 2001, 2014)
and an aggregate investment function. The mere reduction of the aggregate
investment function to a finite sum of individual outputs induced by some
maximizing program would be problematic, because
intertemporal profit-
Colell (1989) that the analog of a Sonnenschein-
we know from Mas-
Debreu theorem holds on the production side as well. Thus, one
Mantel-
lets the data speak, and aggregate investment is empirically estimated. Of
course, investment may happen to exceed current profits, and we know
that this will presumably be the rule in the coming years for the required
green investments. Private debt therefore finances investment in excess
of profits. In the monetary sphere, sticky prices in the sense of Guillermo
340 Comments by Michael Toman and Gaël Giraud
Calvo (see chapter 4 in this book) dynamically relax along the (endoge-
nously determined) unitary production cost augmented by some markup,
which reflects the imperfect competitiveness of the commodity market.
Finally, the model is completed by adopting the UN median scenario for
world population growth.
dimensional nonlinear dynamics of the
The model boils down to a three-
Kolmogorov type, where the wage share and underemployment rate play
beyond the mere evolution of GDP—
a key role. Thus, welfare issues— lie at
the heart of the dynamics, as recommended by Blanchard (2016). Some-
what more precisely, the dynamical system can be paraphrased by the three
following and hardly disputable statements:
1. Employment will rise (resp. decline) if output growth exceeds (resp.
remains lower than) the sum of population plus labor productivity
growth.
2. Wage share of output will rise (resp. decline) if wage rise exceeds (resp.
remains lower than) growth in labor productivity.
3. Private debt ratio will rise (resp. decline) if the rate of growth of debt
exceeds (resp. remains lower than) that of GDP.
The simplicity of this presentation of the core dynamics differs sharply
from that of DSGE models, for example, which, in the words of Blanchard
run
(2016), “are bad communication devices.” More importantly, its long-
analysis shows that, in general circumstances, it admits several locally sta-
ble equilibria whose basin of attraction can be geometrically described.
Depending on the initial conditions, and absent any exogenous shocks,
the state of the economy will be trapped in one of these basins and ulti-
mately converge toward its associated attractor (Grasselli and Costa Lima
2012; Bastidas, Fabre, and Mclsaac 2016). This methodological simplicity
stands in sharp contrast to the equilibrium literature of monetary econo-
mies, for which, as Guillermo Calvo reminds us in chapter 4 in this vol-
ume, multiple equilibria are also the rule, but where one is often at pains
to explain how a static economy can switch from one equilibrium to the
other. Next, the interaction between the monetary and the real spheres of
the economy in Giraud et al. (2016) leads to endogenous monetary business
cycles without relying on exogenous shocks. Furthermore, the good piece
of news provided by the empirical estimation of the model at the world
scale is that, absent climate change, the world economy would presumably
Climate Change, Development, Poverty, and Economics 341
run equilibrium. Simulations suggest,
converge to some relatively safe long-
loop induced by global warming could drive
however, that the climate back-
the world economy out of the basin of attraction of this safe steady state,
which is a scenario with disastrous consequences.
To grasp the circumstances under which this might happen, let us first
assume that labor productivity grows exponentially at a rate of 1.5 percent
per year, the climate damage function is quadratic, and climate sensitivity
is 2.9 (its average estimation according to IPCC), as in Nordhaus and Sztorc
(2013). We then get a reassuring view on the future of the planet as shown
in figure 7.2: World real GDP grows exponentially and reaches 4.62 times
its 2010 level by the end of this century. Inflation stabilizes at about 2 per-
75 percent (close
cent, the employment rate oscillates in the vicinity of 70–
to-
to its current value), and the private debt- GDP ratio converges slowly
toward a stationary level slightly below 200 percent. By 2050, the average
yearly CO2e emission per capita is 5.6 tons. The temperature change in
+
2100 is 4.95 °C, and the CO2 concentration is 732.8 ppm. Despite these
last frightening numbers, the world economy seems to be doing rather well:
Damages induced by global warming are reducing the final world real GDP
Figure 7.2
Scenario 1: exponential growth
Source: Giraud et al. (2016).
342 Comments by Michael Toman and Gaël Giraud
a fraction higher than the 5 percent losses first envis-
by only one quarter—
aged by Stern (2007), but a much smaller relative loss than the one experi-
enced, say, by Russia in the 1990s. As a consequence of this hardly credible
scenario of exponential growth, CO2e emissions peak only in about the
second century, and the zero-
middle of the twenty- emission level reached
one century later!
The picture changes dramatically as soon as growth is made endog-
enous. Suppose, indeed, that the growth rate of labor productivity is
affected by the rise in temperature, as empirically estimated by Burke et
al. (2015): The hotter the planet becomes, the slower average productiv-
ity growth will be. Keeping all other parameters of the model unchanged,
this endogenization of technological progress suffices to provoke a forced
growth (figure 7.3): Around 2100, world real GDP peaks at 225 per-
de-
cent of its 2010 value and then inexorably declines. By the end of the
second century, it becomes even lower than its 2010 value. As a
twenty-
to-
counterpart, debt- GDP ratio explodes: It is already greater than 300 per-
cent by 2100 and grows exponentially after that. Due to a lower pace of
growth, the temperature increase in 2100 is lower than in the exponential
growth scenario (+4.92°C). De-
growth, however, has no disruptive effect
on the labor market, because the employment rate only decreases slightly
second century. As for inflation,
below 70 percent at the end of the twenty-
it remains wisely close to 2 percent.12 If such a scenario is considered a
plausible outcome, it logically implies that, above a certain maturity, the
term discount rate should be negative (cf. the discussion in section 4
long-
above). Do the negative rates exhibited by financial markets today reflect
the fact that investors are correctly forecasting the potentially disastrous
as-
consequences of the business- usual path most of the world economy is
still following?
12. Of course, de-growth is an implausible scenario given the astonishingly inno-
vative character of advanced economies and especially the ICT (Information and
Communications Technology) revolution of the past two decades or so. The ongoing
debate on secular stagnation initiated, among others, by Robert Gordon and Larry
Summers does not, however, take climate change into account. That the coupling of
a lack of substantial technological innovation in the coming decades and damages
provoked by climate change might lead to de- growth (by disaster, not by design)
should, at the least, sound like a warning.
Climate Change, Development, Poverty, and Economics 343
Figure 7.3
Scenario 2: Forced de- growth
Source: Giraud et al. (2016).
So what happens if one takes due account of the probable strong con-
vexity of the damage function, as advocated by Dietz and Stern (2015),
together with a climate sensitivity equal to 6? This time, numerical simu-
deflationary collapse of the world economy starting
lations lead to a debt-
not later than in the 2050s (figure 7.4). As for the employment rate, this
fluctuates around 70 percent up to the middle of this century, and then
plunges below 50 percent around 2100. Twenty years earlier, the world
has entered a strongly deflationary phase, as the inflation rate stabilizes
around −5 percent at the turn of the century. At this time, the debt-
to-
output ratio is above 800 percent. This disaster, however, is not even
good news for the climate, as the peak of emissions around 2045 does not
prevent the temperature from rising up to +4.62°C in 2100—
essentially
because of the strong inertia of the response of the world’s ecosystem to
carbon emissions.
Again, such a breakdown might seem inconceivable, given the current
prosperity of so many people, both in advanced and emerging economies.
344 Comments by Michael Toman and Gaël Giraud
Figure 7.4
Scenario 3: Debt-deflationary collapse
Source: Giraud et al. (2016).
And it is not the intention of Giraud et al. (2016) to claim that such a
simulated scenario is even probable. But it could be used as a tool to better
understand how the world economy is going to avoid such a collapse. In
particular, the public sphere is absent from the model envisaged in Giraud
et al. (2016). At the very least, this quite pessimistic perspective means that
the funding of the US$90 trillion investment identified in New Climate
Economy (2014) can presumably not rely solely on the private sector. The
public sphere will have to be involved at some stage. Numerical simulations
in Giraud et al. (2016) also suggest that a strongly increasing carbon price
at least within the
would be sufficient to allow an escape from a collapse—
clearly narrow limits of this model. Converted into 2005 US dollars, a value
of $74 per ton of CO2e in 2015 and $306 in 2055 would suffice to drive the
world economy onto a safe trajectory in the third scenario sketched above.
Note that this implies a price of about $900 for a ton of carbon before the
middle of this century.
Of course, Giraud et al. (2016) is definitely a toy model: It aims to gauge
“the magnitude of key forces using simple calculations” and “is not designed
to allow users to reach definitive quantitative answers to policy questions of
Climate Change, Development, Poverty, and Economics 345
interest” (Kocherlakota 2016). It should not be perceived as a tool to fore-
first century. Not only
cast the path of the world economy in the twenty-
because of its evident modeling limitations, but also because institutional
changes, technological shocks, and political complications will most prob-
ably play a major role in the future, just as they have always done in the
past. In this modest perspective, however, Giraud et al. (2016) undoubtedly
confirm some of the points forcefully made by FS:
as-
The business- usual scenario might look uglier than many of us believe.
A “radical deepening” of macroeconomics may shed light on issues that, so
far, have remained largely ignored by standard approaches, such as the
role of private debt along the path toward resilient economies.
or for that matter, the correct barycenter
The “correct price of carbon”—
is probably much higher
of the corridor of prices (see section 3 above)—
than more standard simulations would suggest.
References
Marc Eber, and David Heath. 1999. “Coher-
Artzner, Philippe, Freddy Delbaen, Jean-
228.
ent Measures of Risk.” Mathematical Finance 9 (3): 203–
Axelrod, Robert. 1997. The Complexity of Cooperation: Agent-Based Models of Competi-
tion and Collaboration. Princeton, NJ: Princeton University Press.
Bardi, Ugo. 2014. Extracted: How the Quest for Mineral Wealth Is Plundering the Planet.
White River Junction, VT: Chelsea Green Publishing.
Bastidas, Daniel, Adrien Fabre, and Florent McIsaac. 2016. “Minskyan Classical
Flow Consistent Macrodynamic Model.”
Growth Cycles: Stability Analysis of a Stock-
Mathematics and Financial Economics website, https://doi.org/10.1007/s11579-018
-0231-6
11,
Blanchard, Olivier. 2016. “Do DSGE Models Have a Future?” Policy Brief 16-
Peterson Institute for International Economics, Washington, DC.
Bredenkamp, Hugh, and Catherine A. Pattillo. 2010. “Financing the Response to
Climate Change.” IMF Staff Position Note 10/06, International Monetary Fund,
Washington, DC.
Linear
Burke, Marshall, Solomon M. Hsiang, and Edward Miguel. 2015. “Global Non-
Effect of Temperature on Economic Production.” Nature 527 (7577): 235.
Canfin, Pascal, Alain Grandjean, and Gérard Mestrallet. 2016. “Proposals for Align-
ing Carbon Prices with the Paris Agreement.” Ministère de l’environnement, de
346 Comments by Michael Toman and Gaël Giraud
l’énergie et de la mer, July. https://www.ladocumentationfrancaise.fr/rapports-publics
/164000418/index.shtml.
Canfin, Pascal, Alain Grandjean, Ian Cochran, and Mireille Martini. 2015. “Mobiliz-
ing Climate Finance— A Road Map to Finance a Low- Carbon Economy.” Report of
the Canfin-Grandjean Commission, June (INIS-FR--15-0680), France.
Carney, Mark. 2016. “Resolving the Climate Paradox.” Arthur Burns Memorial
Lecture, September 22, Berlin. http://www.bankofengland.co.uk/publications/Docu
ments/speeches/2016/speech923.pdf.
François Mertens. 1999. “Relative Utilitarianism.” Econo-
Dhillon, Amrita, and Jean-
498.
metrica 67 (3): 471–
Dietz, Simon, and Nicholas H. Stern. 2015. “Endogenous Growth, Convexity
of Damage and Climate Risk: How Nordhaus’ Framework Supports Deep Cuts in
620.
Carbon Emissions.” Economic Journal 125 (583): 574–
Dubey, Pradeep, John Geanakoplos, and Martin Shubik. 1987. “The Revelation of
Information in Strategic Market Games: A Critique of Rational Expectations Equilib-
137.
rium.” Journal of Mathematical Economics 16 (2): 105–
Eggertsson, Gauti B., and Paul Krugman. 2010. “Debt, Deleveraging, and the Liquidity
Minsky-
Trap: A Fisher- Koo Approach.” Mimeo, Princeton University, Princeton, NJ.
Fankhauser, Sam, and Nicholas Stern. 2016. “Climate Change, Development,
Poverty and Economics.” Grantham Research Institute on Climate Change and
Environment.
Fleurbaey, Marc, and Francois Maniquet. 2008. “Fair Social Orderings.” Economic
45.
Theory 34 (1): 25–
Giraud, Gaël. 2001. “An Algebraic Index Theorem for Non-
smooth Economies.”
269.
Journal of Mathematical Economics 36 (4): 255–
Giraud, Gaël. 2014. “Le Découplage Énergie- PIB, Ou Le Rôle (Sous-Estimé) de
l’Energie dans la Croissance du PIB.” In Produire Plus, Polluer Moins: L’Impossible
Découplage? by Thierry Caminel, Philippe Frémeaux, Gaël Giraud, Aurore Lalucq,
and Philippe Roman, 27– 43. Paris: Les Petits Matins.
Giraud, Gaël. 2015. “Préface.” In L’Économie En Question. Regards et Apports des
Luc Castel and Vincent Pilley, 1–
Spiritualités et des Religions, edited by Jan- 35. Paris:
L’Harmattan.
Giraud, Gaël, and Rakesh Gupta. 2016. “Coherent Multidimensional Poverty.”
Chair Energy and Prosperity Working Papers. http://www.chair-energy-prosperity
.org/publications/coherent-multidimensional-poverty-measurement/.
Giraud, Gaël, and Antonin Pottier. 2016. “Debt-
Deflation versus the Liquidity Trap:
The Dilemma of Nonconventional Monetary Policy.” Economic Theory 62 (1): 383– 408.
Climate Change, Development, Poverty, and Economics 347
Giraud, Gaël, Florent McIsaac, Emmanuel Bovari, and Ekaterina Zatsepina. 2016.
Flow Consistent Monetary Macrodynamics of
“Coping with the Collapse: A Stock-
Global Warming.” AFD Research Paper 2016/29, Agence Française de Développe-
ment, Paris.
Goujon, Michaël, Jean-François Hoarau, and Françoise Rivière. 2015. “Vulnérabilités
Comparées Des Économies Ultramarines.” AFD Working Paper 45, Agence Française
de Développement, Paris.
Grasselli, Matheus R., and B. Costa Lima. 2012. “An Analysis of the Keen Model
for Credit Expansion, Asset Price Bubbles and Financial Fragility.” Mathematics and
210.
Financial Economics 6 (3): 191–
Guillaumont, Patrick. 2013. “Measuring Structural Vulnerability to Allocate Devel-
opment Assistance and Adaptation Resources.” FERDI Working Paper 68, Fondation
por les Etudes et Recherches sur le Développement International, Paris.
Hallegatte, Stephane, Mook Bangalore, Laura Bonzanigo, Marianne Fay, Tamaro
Schilb. 2016.
Kane, Ulf Narloch, Julie Rozenberg, David Treguer, and Adrien Vogt-
“Shock Waves: Managing the Impacts of Climate Change on Poverty.” Washington,
DC: World Bank.
t
Kocherlakota, Narayana. 2016. “Toy Models.” July 17. https:// co
. / .
8dS85Nlpg9
Kydland, Finn E., and Edward C. Prescott. 1982. “Time to Build and Aggregate Fluc-
1370.
tuations.” Econometrica 50 (6): 1345–
107.
Leontief, Wassily. 1982. “Academic Economics.” Science 217 (4555): 104–
Le Treut, Hervé. 2013. Les Impacts du Changement Climatique en Aquitaine: Un Etat des
Lieux Scientifique. Pessac, France: Presses Universitaires de Bordeaux.
Liousse, Catherine, Eric Assamoi, Patrick Criqui, Claire Granier, and Robert Rosset.
2014. “Explosive Growth in African Combustion Emissions from 2005 to 2030.”
9326/9/3/035003.
Environmental Research Letters 9 (3): 035003, doi:10.1088/1748–
Mallick, K. Pramana. 2009. J Phys 73: 417. https://doi.org/10.1007/s12043-009-0098
-4 ----- source: https://link.springer.com/article/10.1007/s12043-009-0098-4#citeas.
Mankiw, N. Gregory. 2001. “The Inexorable and Mysterious Tradeoff between Infla-
61.
tion and Unemployment.” Economic Journal 111 (471): 45–
Mankiw, N. Gregory. 2014. Principles of Macroeconomics. Boston: Cengage Learning.
Mas-Colell, Andreu. 1989. “Capital Theory Paradoxes: Anything Goes.” In Joan Rob-
inson and Modern Economic Theory, edited by George R. Feiwel, 505– 520. London:
Palgrave Macmillan.
Mehra, Rajnish, and Edward C. Prescott. 1985. “The Equity Premium: A Puzzle.”
161.
Journal of Monetary Economics 15 (2): 145–
348 Comments by Michael Toman and Gaël Giraud
Minsky, Hyman P. 1992. “The Financial Instability Hypothesis.” Levy Economics
on-
Institute Working Paper 74, Levy Economics Institute, Annandale- Hudson, NY.
New Climate Economy. 2014. “Better Growth, Better Climate: The New Climate
Economy Report.” London: Global Commission on the Economy and Climate.
Nordhaus, William D., and Paul Sztorc. 2013. DICE 2013R: Introduction and User’s
Manual. https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved
=2ahUKEwjkzZ2poPLgAhXn2eAKHX58CVkQFjAAegQICRAC&u rl =http%3A%
2F%2Fwww.econ.yale.edu%2F~nordhaus%2Fhomepage%2Fhomepage%2Fdocument
s%2FDICE_Manual_100413r1.pdf&usg=AOvVaw2D1La96WGlSgoIm2Ka-Q1a.
Roemer, John E. 2011. “The Ethics of Intertemporal Distribution in a Warming
390.
Planet.” Environmental and Resource Economics 48 (3): 363–
Romer, Paul M. 2016. “The Trouble with Macroeconomics.” American Economist 20:
1–20.
Snyder, Carolyn W. 2016. “Evolution of Global Temperature over the Past Two Mil-
228.
lion Years.” Nature 538 (7624): 226–
Sonnenschein, Hugo. 1972. “Market Excess Demand Functions.” Econometrica 40 (3):
549–563.
Stern, N. H. 2007. The Economics of Climate Change: The Stern Review. Cambridge:
Cambridge University Press.
Sterner, Thomas, and Martin U. Persson. 2008. “An Even Sterner Review: Introduc-
ing Relative Prices into the Discounting Debate.” Review of Environmental Economics
76.
and Policy 2 (1): 61–
Vautard, Robert, Françoise Thais, Isabelle Tobin, François-Marie Bréon, Jean-Guy
Devezeaux de Lavergne, Augustin Colette, Pascal Yiou, and Paolo Michele Ruti.
2014. “Regional Climate Model Simulations Indicate Limited Climatic Impacts
by Operational and Planned European Wind Farms.” Nature Communications 5,
doi:10.1038/ncomms4196.
Vidal, Olivier, Bruno Goffé, and Nicholas Arndt. 2013. “Metals for a Low-
Carbon
896.
Society.” Nature Geoscience 6 (10): 894–
8 Behaviorally Informed
Cass R. Sunstein
In recent decades, behavioral economists have been incorporating empiri-
cal findings about human behavior into economic models (Kahneman
2011; Thaler 2015). Those findings have transformed our understandings of
economic theory. They have also greatly affected our understandings of the
role of economic incentives (Chetty et al. 2012) and the content of policy
instruments. At the same time, they are providing instructive lessons about
low-
the appropriate design of “nudges”— preserving, behav-
cost, choice-
iorally informed approaches to regulatory problems, including disclosure
requirements, default rules, and simplification (Thaler and Sunstein 2008;
Halpern 2015).
Economists have long emphasized the importance of incentives. Behav-
ioral economists do not disagree that incentives matter, but they empha-
size the need to see that choice architecture, understood as the background
against which decisions are made, can have major consequences for both
decisions and outcomes (Thaler 2015). Small, inexpensive policy initia-
tives, making modest design changes, can have large and highly beneficial
effects in areas that include health, energy, the environment, savings, and
much more. My main purposes here are to explore relevant evidence, to
explore its implications for standard economic theory, to catalog behavior-
ally informed practices and reforms, and to discuss some lessons for policy.
In the United States, numerous policies have been directly informed by
behavioral findings, and behavioral economics has played an unmistakable
role in countless domains (Sunstein 2013).
The relevant initiatives enlist such tools as disclosure, warnings, norms,
and default rules, and they can be found in multiple areas, including fuel
economy, energy efficiency, consumer protection, financial regulation,
environmental protection, health care, and obesity prevention (Sunstein
350 Cass R. Sunstein
2013). As a result, behavioral findings have become an important reference
point for regulatory and other policy making in the United States (Sunstein
2016).
Prime Minister Cameron created a Behav-
In the United Kingdom, then–
ioural Insights Team with the specific goal of incorporating an understand-
ing of human behavior into policy initiatives (Halpern 2015). The team
has used these insights to promote initiatives in numerous areas, including
smoking cessation, energy efficiency, organ donation, consumer protec-
tion, and compliance strategies in general (Halpern 2015). A great deal of
money is being saved. Other nations have expressed keen interest in the
work of the team, and its operations are expanding (Halpern 2015).
Behavioral economics has drawn attention in Europe more broadly. The
operation has published
Organisation for Economic Development and Co-
a consumer policy toolkit that recommends initiatives rooted in behavioral
General for
findings (OECD 2010). In the European Union, the Directorate-
Health and Consumers has also shown the influence of behavioral econom-
ics (DG SANCO 2010). A report from the European Commission, called
“Green Behavior,” enlists behavioral economics to outline policy initiatives
.c
to protect the environment (European Commission 2012; iNudgeYou om
n.d.). Private organizations are making creative use of behavioral insights
related, and other goals (see
to promote a variety of environmental, health-
iNudgeYou.com n.d.).
It is clear that behavioral findings have greatly affected economic theory
(Thaler 2015) and are having a large impact on regulation, law, and public
policy all over the world (Sunstein 2016). With increasing global interest
cost tools, that impact will inevitably grow over the next decades.
in low-
In these circumstances, it is particularly important to have a sense of what
we know, what we do not know, and how emerging understandings can
inform sensible policies and reforms.
What We Know
Findings
Consider a simple view: Human beings try to maximize utility. To under-
stand their behavior, two questions are important. (1) What do they care
about? (2) What incentives do they face? On one view, if you can answer
those questions, that is all ye need to know on earth (more or less).
Behaviorally Informed 351
Behavioral economics has cast serious doubt on that view. Even if ana-
lysts have full information about (1) and (2), they may have little or no
idea about what people will choose. At a minimum, there are two more
questions. (3) How do people deviate from full rationality? (4) What is the
relevant choice architecture? Without answers to (3) and (4), we might be
at sea, or make predictions that go badly wrong.
For purposes of policy, the central findings of behavioral economics
fall into four categories. What follows is not meant to be a comprehensive
account; the focus is on those findings that have particular importance to
what governments do.
Inertia and procrastination
a) Default rules often have a large effect on social outcomes. Both private and
rules that determine the
public institutions often establish “default rules”—
result if people make no affirmative choice at all (Sunstein 2015). Accord-
known view in economics and the economic analysis of law,
ing to a well-
default rules have no effect, at least when transactions costs are zero: People
will bargain their way to the efficient result, and that result will be the
same, whatever the content of the default.
That view is not correct. In part because of the power of inertia, default
rules can be extremely important, because they tend to stick. If the goal is
to affect behavior, the right advice is often simple: Create a default rule that
puts people in the situation that you favor. Where they start will often be
where they end up.
In the domain of retirement savings, for example, the default rule has
significant consequences. When people are asked whether they want to opt
in to a retirement plan, the level of participation is far lower than if they
are asked whether they want to opt out. Automatic enrollment significantly
increases participation (Thaler 2015). Something similar is true in the envi-
ronmental context. If people are automatically enrolled in green energy,
there can be major effects on pollution levels (Sunstein 2016).
More generally, people may decline to change from the status quo even
if the costs of change are low (or essentially zero) and the benefits sub-
stantial. In the context of energy and the environment, for example, we
efficient alterna-
might predict that people might neglect to switch to fuel-
tives even when it is in their interest to do so (Sunstein 2015). It follows
that complexity can have serious adverse effects by increasing the power of
inertia, and that ease and simplification (including reduction of paperwork
352 Cass R. Sunstein
burdens) can produce significant benefits. These benefits include increased
compliance with law and greater participation in public programs. Often
people do not act in advisable ways, not because they do not want to do
so, but because the best path is obscure or difficult to navigate. Behavioral
economists suggest that people will often use a GPS device, even when
rational people might be expected not to need one.
b) Procrastination can have significant adverse effects, even when it is in peo-
ple’s interest not to procrastinate. According to standard economic theory,
people will consider both the short term and the long term. They will take
account of relevant uncertainties; the future may be unpredictable, and sig-
nificant changes may occur over time. They will appropriately discount the
future; it may be better to have money, or a good event, a week from now
than a decade from now. In practice, however, some people procrastinate
term costs but that would
or neglect to take steps that impose small short-
term gains (Thaler 2015). They may, for example, delay
produce large long-
enrolling in a retirement plan, starting to exercise, ceasing to smoke, or
saving technology.
using some valuable, cost-
When procrastination is creating significant problems, automatic enroll-
ment in relevant programs might be helpful. Moreover, complex require-
ments, inconvenience, and lengthy forms are likely to make the situation
worse and perhaps unexpectedly so.
c) When people are informed of the benefits or risks of engaging in certain
actions, they are far more likely to act in accordance with that information if
they are simultaneously provided with clear, explicit information about how to
do so (Leventhal, Singer, and Jones 1965; Nickerson and Rogers 2010). On
one view, such information should not matter, at least if it is easy to find.
People will consider the costs of search, of course, but if those costs are low
and the potential benefits are high, they will search.
But not always. For example, those who are informed of the benefits of
a vaccine are more likely to become vaccinated if they are also given spe-
cific plans and maps describing where to go (Leventhal, Singer, and Jones
1965). Similarly, behavior has been shown to be significantly affected if
people are informed, not abstractly of the value of “healthy eating,” but
specifically of the advantages of buying 1 percent milk as opposed to whole
milk (Heath and Heath 2010). In many domains, the identification of a
specific, clear, unambiguous path or plan has an important effect on social
Behaviorally Informed 353
outcomes; complexity or vagueness can ensure inaction, even when people
are informed about risks and potential improvements. What appears to be
skepticism or recalcitrance may actually be a product of ambiguity.
Framing and presentation
a) People are influenced by how information is presented or “framed” (Levin,
Schneider, and Gaeth 1998). According to standard theory, “frames” should
not matter. What matters is expected value. But psychologists and behav-
ioral economists have found otherwise (Kahneman 2011).
If, for example, people are informed that they will gain a certain amount
of money by using energy efficient products, they may be less likely to
change their behavior than if they are told that they will lose the same
amount of money by not using such products. When patients are told that
90 percent of those who have a certain operation are alive after 5 years, they
are more likely to elect to have the operation than when they are told that
after 5 years, 10 percent of patients are dead (Redelmeier, Rozin, and Kahn-
free”
eman 1993). It follows that a product that is labeled “90 percent fat-
may well be more appealing than one that is labeled “10 percent fat.” It also
follows that choices are often not made based solely on their consequences;
assessments may be affected by the relevant frame.
b) Information that is vivid and salient usually has a larger impact on behav-
ior than information that is statistical and abstract. With respect to public
health, vivid displays can be more effective than abstract presentations of
statistical risks. This point bears on the design of effective warnings. Atten-
tion is a scarce resource, and vivid, salient, and novel presentations may
trigger attention in ways that abstract or familiar ones cannot.
far more so than standard eco-
In particular, salience greatly matters—
nomic theory has predicted. Why, for example, do people pay bank over-
draft fees? One of the many possible answers is that such fees are not
sufficiently salient to people, and the fees are incurred as a result of inat-
tention or inadvertent mistakes. One study suggests that limited attention
is indeed a source of the problem, and that once overdraft fees become
salient, they are significantly reduced (Stango and Zinman 2011). When
people take surveys about such fees, they are less likely to incur a fee in the
following month, and when they take multiple surveys, the issue becomes
sufficiently salient that overdraft fees are reduced for as much as 2 years.
In many areas, the mere act of being surveyed can affect behavior by, for
354 Cass R. Sunstein
example, increasing the use of water treatment products (thus promoting
health) and the take up of health insurance; one reason is that being sur-
veyed increases the salience of the action in question (Zwane et al. 2011).
A more general point is that many costs (or benefits) are less salient than
purchase prices; they are “shrouded attributes,” to which some consumers
on costs may matter a great deal but
do not pay much attention. Such add-
receive little consideration, because they are not salient.
c) People display loss aversion; they may well dislike losses more than they like
corresponding gains (Thaler, Kahneman, and Knetsch 1991; McGraw et al.
2010; Card and Dahl 2011). Standard economic theory emphasizes the
importance of expected value. A 90 percent chance of gaining $500 is not
any more good than a 90 percent chance of losing $500 is bad. But human
beings turn out to be loss averse; they much dislike losses, and they will do
a great deal to avoid them (Kahneman 2011).
Whether a change counts as a loss or a gain depends on the reference
point, which can be affected by mere description or by policy decisions,
for example, on grocery
and which is often the status quo. A small tax—
can have a large effect on behavior, even if a promised bonus has no
bags—
effect at all; one reason is loss aversion. It follows that very small charges or
fees can be a surprisingly effective policy tool. Partly as a result of loss aver-
sion, the initial allocation of a legal entitlement can affect people’s valua-
tions. Those who have the initial allocation may value a good more than
they would if the allocation were originally elsewhere, thus showing an
endowment effect (Thaler 2015).
Social influences
a) In multiple domains, individual behavior is greatly influenced by the perceived
behavior of other people (Hirshleifer 1995). With respect to obesity, proper
exercise, alcohol consumption, smoking, becoming vaccinated, and much
more, the perceived decisions of others have a significant influence on indi-
vidual behavior and choice. The behavior of peers has been found to have
a significant effect on risky behavior among adolescents, including tobacco
smoking, marijuana use, and truancy (Bisin, Moro, and Topa 2011; Card
and Giuliano 2011).
In particular, food consumption is greatly affected by the food consump-
tion of others, and indeed, the body type of others in the relevant group
can affect people’s responses to their food choices, with a greater effect from
those who are thin than from those who are heavy (McFerran et al. 2011).
Behaviorally Informed 355
Perception of the norm in the pertinent community can affect risk taking,
safety, and health (Sunstein 2015; Thaler 2015). The norm conveys signifi-
cant information about what ought to be done; for that reason, those who
lack private information may follow the apparent beliefs and behavior of
relevant others, sometimes creating informational cascades.
In addition, people care about their reputations. Thus they may be
influenced by others so as not to incur their disapproval. In some con-
texts, social norms can help create a phenomenon of compliance without
as, for example, when people comply with laws forbidding
enforcement—
indoor smoking or requiring buckling of seat belts, in part because of social
norms or the expressive function of those laws. These points bear on the
public partnerships.
value and importance, in many domains, of private–
b) In part because of social influences, people are more likely to cooperate with
one another, and to contribute to the solution of collective action problems, than
standard economic theory predicts (Camerer 2003). People’s willingness to
cooperate is partly a product of an independent commitment to fairness,
but it is partly a product of a belief that others will see and punish a failure
to cooperate or to act fairly. Norms of reciprocity can be exceedingly impor-
tant. In many contexts, the result is a situation in which people cooperate
and might punish
on the assumption that others are cooperating as well—
those who fail to do so.
Difficulties in assessing probability
a) In many domains, people show unrealistic optimism (Jolls 1998; Sharot
2011). Standard economic theory does not see human beings as having
systematically skewed probability judgments. But there is a systematic ten-
dency toward optimism (Sharot 2011). The “above average” effect is com-
mon (Weinstein 1987); many people believe that they are less likely than
others to suffer from various misfortunes, including automobile accidents
and adverse health outcomes. One study found that although smokers do
not underestimate statistical risks faced by the population of smokers, they
nonetheless believe that their personal risk is less than that of the average
smoker (Slovic 1998). Unrealistic optimism has neurological foundations,
with people incorporating good news far more readily than bad news (see
Sharot (2011) for an overview). A predictable result of unrealistic optimism
is a failure to take appropriate precautions.
b) People often use heuristics, or mental shortcuts, when assessing risks (Kahne-
man and Frederick 2002; Kahneman 2011). For example, judgments about
356 Cass R. Sunstein
probability are often affected by whether a recent event comes readily to
mind (Tversky and Kahneman 1973). If an event is cognitively “available,”
people may well overestimate the risk. If an event is not cognitively avail-
able, people might underestimate the risk. In short, “availability bias” can
lead to inaccurate judgments about the probability of undesirable outcomes.
c) People sometimes do not make judgments on the basis of expected value, and
they may neglect or disregard the issue of probability, especially when strong emo-
tions are triggered (Loewenstein et al. 2001). When emotions are strongly
felt, people may focus on the outcome and not on the probability that it
will occur (Loewenstein et al. 2001). (This point obviously bears on reac-
tions to extreme events of various sorts.) Prospect theory, which does not
depend on emotions at all, suggests that for low and moderate changes,
people may be risk averse with respect to gains but risk seeking with respect
to losses; for very large changes, people may be risk seeking with respect to
gains but risk averse for losses (Kahneman and Tversky 1979; Kahneman
2011).
Incentives and Choice Architecture
These various findings are hardly inconsistent with the conventional eco-
nomic emphasis on the importance of material incentives; actual and per-
ceived costs and benefits certainly matter. When the price of a product
rises, or when it becomes clear that use of a product imposes serious health
risks, the demand for the product is likely to fall (at least, and this is a
significant qualification, if these effects are salient). But apart from strictly
material incentives of this kind, evidence suggests the independent impor-
tance of (1) the social environment and (2) prevailing social norms. If, for
example, healthy foods are prominent and easily accessible, people are
more likely to choose them; one study finds an 8 to 16 percent decrease in
intake simply by making food more difficult to reach (as, for example, by
varying its proximity by 10 inches or altering the serving utensil; Rozin et al.
2011). The problem of childhood obesity is, at least in part, a result of the
easy availability of unhealthy foods. The same point bears on smoking and
alcohol abuse.
In fact, small nudges can have surprisingly large effects (Halpern 2015;
Thaler 2015). For example, automatic enrollment in savings programs can
a clear tes-
have far larger effects than significant economic incentives do—
timonial to the potential power of choice architecture and its occasionally
Behaviorally Informed 357
larger effect than standard economic tools (Chetty et al. 2012). Some
evidence suggests that if people are asked to sign forms first rather than
an especially minor change—
last— the incidence of honesty increases sig-
nificantly (Shu et al. 2012).
Markets, Government, and the Vexing Problem of Paternalism
It is natural to wonder whether an understanding of the findings outlined
above justify paternalism or operate as a defense of more regulation (Conly
2013). With respect to paternalism in particular, it is true that some of the
relevant findings supplement the standard accounts of market failures, sug-
gesting that in some settings, markets may fail, in the sense that they may
not promote social welfare even in the presence of perfect competition and
full information. We are now in a position to identify a series of behavioral
market failures, and these do appear to justify regulatory controls (Sun-
stein 2016). Responses to behavioral market failures might be counted as
paternalistic.
term costs and neglect long-
If, for example, people focus on short- term
benefits, it is possible that disclosure policies that specifically emphasize the
long term, or even regulatory requirements (involving, for example, energy
efficiency), may be justified. It is also possible to identify “internalities”—
control and errors in judgment that produce within-
problems of self-
person harms, as, for example, when smoking behavior leads to serious
term considerations over the longer
risks because of the victory of short-
view. These too count as behavioral market failures, and responses may be
paternalistic in character.
Richard Thaler and I have argued in defense of “libertarian paternal-
ism” (Thaler and Sunstein (2008); see also Sunstein (2013)), understood
as approaches that preserve freedom of choice while also steering people
in directions that will make their lives go better (by their own lights). And
it would be possible to think that at least some behavioral market failures
justify more coercive forms of paternalism.
It should not be necessary to emphasize that public officials are subject
to error as well. Indeed, errors may result from one or more of the find-
ings traced above; officials are human and capable of error, too. Behavioral
public choice explores this problem. The dynamics of the political process
may or may not lead in the right direction. It would be absurd to say that
behaviorally informed regulation is more aggressive than regulation that is
358 Cass R. Sunstein
not so informed, or that an understanding of recent empirical findings calls
for more regulation rather than less. The argument is instead that such an
understanding can help inform the design of regulatory programs.
Behaviorally Informed Disclosure
Actually Informing Choice
Examples Many statutory programs recognize that information disclo-
sure can be a useful regulatory tool, replacing or complementing other
approaches. Recent initiatives have drawn directly from behavioral econom-
ics, emphasizing the importance of plain language, clarity, and simplicity.
a) Credit cards. The Credit Card Accountability, Responsibility, and Disclo-
sure Act of 2009 (Credit CARD Act 2009) is designed in large part to ensure
that credit card users are adequately informed. Among other things, the Act
prohibits an increase in annual percentage rates without 45 days’ notice,
prohibits the retroactive application of rate increases to existing balances,
and also requires clear notice of the consumer’s right to cancel the credit
card when the annual percentage rate is raised.
The Act also requires several electronic disclosures of credit card agree-
ments. Specifically, it requires that (1) “each creditor shall establish and
maintain an Internet site on which the creditor shall post the written agree-
ment between the creditor and the consumer for each credit card account
end consumer credit plan”; (2) “each creditor shall provide
under an open-
to the Board, in electronic format, the consumer credit card agreements
that it publishes on its Internet site”; and (3) the “Board shall establish and
maintain on its publicly available Internet site a central repository of the
consumer credit card agreements received from creditors pursuant to this
subsection, and such agreements shall be easily accessible and retrievable
by the public” (Credit CARD Act 2009). The overall effect of the CARD Act
has been extremely impressive, with more than $20 billion in annual sav-
ings for consumers (Agarwal et al. 2013).
b) Nutrition. In the domain of nutrition, various disclosure requirements
are in place. To take just one example, a final rule has been issued by the US
Department of Agriculture (USDA), requiring provision of nutritional infor-
mation to consumers with respect to meat and poultry products. Nutrition
facts panels must be provided on the labels of such products. Under the
Behaviorally Informed 359
rule, the panels must contain information with respect to calories and both
total and saturated fats (9 CFR § 317.309).
The rule clearly recognizes the potential importance of framing. If a
product lists a percentage statement such as “80% lean,” it must also list
its fat percentage. This requirement should avoid the confusion that can
result from selective framing; a statement that a product is 80 percent lean,
standing by itself, makes leanness salient, and may therefore be misleading.
c) Health care. The Patient Protection and Affordable Care Act of 2010
(Affordable Care Act) contains many disclosure requirements designed to
promote accountability and informed choice with respect to health care.
Indeed, the Affordable Care Act is, in significant part, a series of disclosure
requirements, many of which are meant to inform consumers and to do
so in a way that is alert to behavioral findings. Under the Act, a restau-
rant that is part of a chain with twenty or more locations doing business
under the same name is required to disclose calories on the menu board.
Such restaurants are also required to provide in a written form (available
to customers on request) additional nutrition information pertaining to
total calories and calories from fat, as well as amounts of fat, saturated fat,
cholesterol, sodium, total carbohydrates, complex carbohydrates, sugars,
dietary fiber, and protein (Affordable Care Act 2010). Early results suggest
significant effects from calorie labels, concentrated among people who are
overweight (Deb and Vargas 2016).
How, not only whether As social scientists have emphasized, disclosure as
such may not be enough; regulators should devote care and attention to
how, not only whether, disclosure occurs. Clarity and simplicity are often
critical. In some cases, accurate disclosure of information may be ineffective
if the information is too abstract, vague, detailed, complex, poorly framed, or
overwhelming to be useful. If disclosure requirements are to be helpful, they
must be designed to be sensitive to how people actually process information.
A good rule of thumb is that disclosure should be concrete, straight-
forward, simple, meaningful, timely, and salient. If the goal is to inform
people about how to avoid risks or to obtain benefits, disclosure should
avoid abstract statements (such as, about “healthy eating” or “good diet”)
and instead clearly identify the steps that might be taken to obtain the rel-
evant goal (by specifying, for example, what specific actions parents might
take to reduce the risk of childhood obesity).
360 Cass R. Sunstein
In 2010, the Department of Health and Human Services emphasized the
importance of clarity and salience in connection with its interim final rule
titled “Health Care Reform Insurance Web Portal Requirements,” which
“adopts the categories of information that will be collected and displayed
as Web portal content, and the data we will require from issuers and request
from States, associations, and high risk pools in order to create this con-
tent.” (Department of Health and Human Services 2010). That web portal
can be found at http://www.healthcare.gov/.
Behavioral economics, cognitive illusions, and avoiding confusion
If not carefully designed, disclosure requirements can produce ineffective,
confusing, and potentially misleading messages. Behaviorally informed
approaches are alert to this risk and suggest possible improvements. For
instance, automobile manufacturers are currently required to disclose the
fuel economy of new vehicles as measured by miles per gallon (MPG). This
disclosure is useful for consumers and helps promote informed choice. As
the Environmental Protection Agency (EPA) has emphasized, however,
MPG is a nonlinear measure of fuel consumption (Environmental Protec-
tion Agency 2009). For a fixed travel distance, a change from 20 to 25
MPG produces a larger reduction in fuel costs than does a change from
30 to 35 MPG, or even from 30 to 38 MPG. To see the point more dra-
matically, consider the fact that an increase from 10 to 20 MPG produces
more savings than an increase from 20 to 40 MPG, and an increase from
10 to 11 MPG produces savings almost as high as an increase from 34 to
50 MPG.
Evidence suggests that many consumers do not understand this point
and tend to interpret MPG as linear with fuel costs. When it occurs, this
error is likely to produce inadequately informed purchasing decisions when
people are making comparative judgments about fuel costs. For example,
MPG car for
people may well underestimate the benefits of trading a low-
one that is even slightly more fuel efficient. By contrast, an alternative fuel
economy metric, such as gallons per mile, could be far less confusing. Such
a measure is linear with fuel costs and hence suggests a possible way to help
consumers make better choices.
Recognizing the imperfections and potentially misleading nature of
the MPG measure, the Department of Transportation and EPA proposed
in 2010 two alternative labels that are meant to provide consumers with
Behaviorally Informed 361
clearer and more accurate information about the effects of fuel economy
on fuel expenses and on the environment (Environmental Protection
Agency 2009). After a period of public comment, the Department of
Transportation and EPA ultimately chose a label that borrows from both
proposals (Environmental Protection Agency 2009). This approach calls
for disclosure of the factual material included in the first option but adds
year
a clear statement about anticipated fuel savings (or costs) over a 5-
period.
In a related vein, the USDA has abandoned the “Food Pyramid,” used
for decades as the central icon to promote healthy eating. The Pyramid has
long been criticized as insufficiently informative; it does not offer people
any kind of clear “path” with respect to healthy diet. According to one criti-
cal account (Heath and Heath 2010, 61),
its meaning is almost completely opaque. … To learn what the Food Pyramid has
to say about food, you must be willing to decipher the Pyramid’s markings. … The
language and concepts here are so hopelessly abstracted from people’s actual
experience with food … that the message confuses and demoralizes.
In response to these objections, and after an extended period of delibera-
tion, the USDA replaced the Pyramid with a new, simpler icon, consisting
of a plate with clear markings for fruit, vegetable, grains, and protein (Sun-
stein 2013).
The plate is accompanied by straightforward guidance, including “make
half your plate fruits and vegetables,” “drink water instead of sugary
free or low-
drinks,” and “switch to fat- fat (1%) milk.” This approach has
the key advantage of informing people what to do, if they seek to have a
healthier diet.
In some circumstances, the tendency toward unrealistic optimism may
lead some consumers to downplay or neglect information about statistical
risks associated with a product or an activity. Possible examples include
smoking and distracted driving. In such circumstances, disclosure might
be designed to make the risks associated with the product less abstract,
more vivid, and salient. For example, the Family Smoking Prevention and
Tobacco Control Act of 2009 requires graphic warnings with respect to the
risks of smoking tobacco, and the Food and Drug Administration has final-
ized such warnings for public comment, with vivid and even disturbing
pictures of some of the adverse outcomes associated with smoking.
362 Cass R. Sunstein
Behaviorally Informed Tools: Summary Disclosure and Full Disclosure
Disclosure requirements of this kind are designed to inform consumers
at the point of purchase, often with brief summaries of relevant infor-
mation. Such summary disclosures are often complemented with more
robust information, typically found on public or private websites. For
example, the EPA offers a great deal of material on fuel economy online,
going well beyond the information that is available on stickers, and the
nutrition facts label is supplemented by a great deal of nutritional infor-
mation on government websites. Approaches of this kind provide infor-
mation that private individuals and institutions can adapt; reassemble;
and present in new, helpful, imaginative, and often unanticipated ways.
Some of the most valuable and creative uses of full disclosure are made by
the private sector.
Other disclosure requirements are not specifically directed at consumers
or end users at all. They promote public understanding of existing problems
and help produce possible solutions by informing people about current
practices. One example is the Emergency Planning and Community Right-
Know Act (1986). At first, this law seemed to be largely a bookkeeping
to-
measure, requiring a “Toxic Release Inventory,” in which firms reported
what pollutants they were using. But available evidence indicates that it has
had beneficial effects, helping spur reductions in toxic releases throughout
the United States (Hamilton 2005). One reason involves public account-
ability: Public attention can help promote behavior that fits with statutory
purposes.
To be sure, mandatory disclosure can impose costs and burdens on both
private and public institutions, and to the extent permitted by law, those
costs and burdens should be considered when deciding whether and how
to proceed. Empirical evidence on the actual effects of disclosure policies
is indispensable (Greenstone 2009; Sunstein 2010; Schwartz et al. 2011).
Default Rules and Simplification
Social science research provides strong evidence that starting points, or
“default rules,” greatly affect social outcomes. Default rules are one way of
easing people’s choices, and they are used in countless domains by both
public and private institutions.
Behaviorally Informed 363
Automatic Enrollment and Default Rules: Examples
Savings In the United States, employers have long asked workers
whether they want to enroll in 401(k) plans; under a common approach,
the default rule is nonenrollment. Even when enrollment is easy, the
number of employees who enroll, or opt in, has sometimes been rela-
tively low (Madrian and Shea 2001; Gale, Iwry, and Walters 2009). In
the United States, some employers have responded by changing the
default to automatic enrollment, by which employees are enrolled unless
they opt out. The results are clear: Significantly more employees end up
out design than with opt-
enrolled with an opt- in (Gale, Iwry, and Walters
2009). This is so even when opting out is easy. Importantly, automatic
enrollment has significant benefits for all groups, with increased antici-
pated savings for Hispanics, African Americans, and women in particular
(Chiteji and Walker 2009; Orszag and Rodriguez 2009; Papke, Walker,
and Dworsky 2009).
The Pension Protection Act of 2006 (Pension Protection Act 2006) draws
directly on these findings by encouraging employers to adopt automatic
enrollment plans. The Pension Protection Act does this by providing non-
discrimination safe harbors for elective deferrals and for matching contri-
butions under plans that include an automatic enrollment feature, as well
withholding laws to allow
as by providing protections from state payroll-
President Obama
for automatic enrollment. Building on these efforts, then-
asked the Internal Revenue Service and the Treasury Department to under-
take initiatives to make it easier for employers to adopt such plans (Internal
Revenue Service 2009; Obama 2009).
School meals The National School Lunch Act (Healthy, Hunger-
Free
Kids Act 2012) takes steps to allow “direct certification” of eligibility, thus
reducing complexity and introducing what is a form of automatic enroll-
ment. Under the program, children who are eligible for benefits under
certain programs will be “directly eligible” for free lunches and free break-
fasts and hence will not have to fill out additional applications (Healthy,
Free Kids Act 2012). To promote direct certification, the USDA
Hunger-
has issued an interim final rule that is expected to provide up to 270,000
children with school meals (Department of Agriculture 2011). In total,
the program is enrolling more than 12 million children in the relevant
program.
364 Cass R. Sunstein
Payroll statements The Department of Homeland Security has changed
the default setting for payroll statements to electronic from paper, thus
reducing costs (Orszag 2010). In general, changes of this kind may save
significant sums of money for both the private and public sectors.
Automatic Enrollment and Default Rules: Mechanisms and Complexities
A great deal of research has attempted to explore exactly why default rules
have such a large effect on outcomes (Carroll et al. 2009; Dinner et al. 2009;
Gale, Iwry, and Walters 2009). There appear to be three contributing fac-
tors. The first involves inertia and procrastination. To alter the effect of the
default rule, people must make an active choice to reject the default. In
view of the power of inertia and the tendency to procrastinate, people may
simply continue with the status quo.
The second factor involves what might be taken to be an implicit
endorsement of the default rule. Many people appear to conclude that the
default was chosen for a reason; they believe that they should not depart
from it unless they have particular information to justify a change.
Third, the default rule might establish the reference point for people’s deci-
sions; the established reference point has significant effects, because people
dislike losses from that reference point. If, for example, the default rule favors
efficient light bulbs, then the loss (in terms of reduced efficiency) may
energy-
efficient light
loom large, and the tendency will be to continue with energy-
bulbs. But if the default rule favors less efficient (and initially less expensive)
light bulbs, then the loss in terms of upfront costs may loom large, and the
tendency will be to favor less efficient light bulbs. In a significant number of
domains, it might be possible to achieve regulatory goals, and to do so while
maintaining freedom of choice and at low cost, by selecting good default
rules and avoiding harmful ones (Sunstein 2015).
Some default rules apply to all of the relevant population, subject to
the ability to opt out. Other default rules are personalized, in the sense
that they draw on available information about which approach best suits
individuals in the relevant population. A personalized default might be
based on geographical or demographic variables; for example, income and
age might be used in determining appropriate default rules for retirement
plans. Alternatively, a personalized default might be based on people’s own
past choices to the extent that they are available.
Behaviorally Informed 365
An advantage of personalized default rules is that they may well
be more accurate than “mass” default rules. As technology evolves, it
should be increasingly possible to produce personalized defaults, based
on people’s own choices and situations; such rules are likely to be far
more accurate than more general ones. There will be excellent opportu-
nities to use default rules to promote people’s welfare (Sunstein 2016).
To be sure, any such rules must respect the applicable laws, policies, and
regulations involving personal privacy and should avoid unduly crude
proxies.
Simplification
Where it is not possible or best to change the default, a similar effect might
be obtained merely by simplifying and facilitating people’s choices. Com-
plexity can have serious unintended effects (including indifference, delay,
and confusion), potentially undermining regulatory goals by reducing
compliance or by decreasing the likelihood that people will benefit from
various policies and programs (Sunstein 2013).
For example, a series of steps have been taken recently toward simpli-
fying the Free Application for Federal Student Aid (FAFSA), reducing the
number of questions through skip logic (a survey method that uses previ-
ous responses to determine subsequent questions) and allowing electronic
retrieval of information (Office of Management and Budget 2010). Use of
a simpler and shorter form is accompanied by a pilot initiative to permit
online users to transfer data previously supplied electronically in their tax
forms directly into their FAFSA applications.
These steps are intended to simplify the application process for financial
aid and thus to increase access to college; there is good reason to believe
that such steps will enable many students to receive aid for attending col-
lege when they previously could not do so. Similar steps might be taken in
many other domains. And indeed, there is reason to believe that imperfect
up of existing benefit programs, including those that provide income
take-
support, is partly a product of behavioral factors, such as procrastination
and inertia. It follows that efforts to increase simplicity, including auto-
matic enrollment, may have substantial benefits.
366 Cass R. Sunstein
Well Beyond Incentives
My goals here have been to outline some of the key findings in behav-
ioral economics, to show how they depart from standard economic the-
ory, and to sketch some lessons for policy. A general conclusion is that
although material incentives (including price and anticipated health
effects) greatly matter, outcomes are independently influenced by choice
architecture, including (1) the social environment and (2) prevailing
social norms.
Because complexity can often have undesirable or unintended side
including high costs, noncompliance with law, and reduced partic-
effects—
simplification helps promote regulatory goals.
ipation in useful programs—
Indeed, simplification can often have surprisingly large effects.
filling burdens (as, for example, through
Reduced paperwork and form-
fewer questions, use of skip patterns, electronic filing, and prepopulation)
can produce significant benefits, not merely by reducing burdens but also
by making programs more readily available. It is thus desirable to take
steps to ease participation in such programs by increasing convenience
and by giving people clearer signals about what, exactly, they are required
to do.
References
Affordable Care Act. 2010. The Patient Protection and Affordable Care Act of 2010.
Pub L No 111–148, 124 Stat 119, codified in various sections of Title 42.
Agarwal, Sumit, Souphala Chomsisenghphet, Neale Mahoney, and Johannes Stroe-
bel. 2013. “Regulating Consumer Financial Products: Evidence from Credit Cards.”
NBER Working Paper 19484, National Bureau of Economic Research, Cambridge,
MA.
Bisin, Alberto, Andrea Moro, and Giorgio Topa. 2011. “The Empirical Content of
Models with Multiple Equilibria in Economies with Social Interactions.” NBER
Working Paper 17196, National Bureau of Economic Research, Cambridge, MA.
Camerer, Colin F. 2003. Behavioral Game Theory: Experiments in Strategic Interaction.
Princeton, NJ: Princeton University Press.
Card, David, and Gordon B. Dahl. 2011. “Family Violence and Football: The Effect
of Unexpected Emotional Cues on Violent Behavior.” Quarterly Journal of Economics
1907.
126 (4): 1879–
Behaviorally Informed 367
Card, David, and Laura Giuliano. 2011. “Peer Effects and Multiple Equilibria in the
Risky Behavior of Friends.” NBER Working Paper 17088, National Bureau of Eco-
nomic Research, Cambridge, MA.
Carroll, Gabriel D., James J. Choi, David Laibson, Brigitte C. Madrian, and Andrew
Metrick. 2009. “Optimal Defaults and Active Decisions.” Quarterly Journal of Econom-
1674.
ics 124 (4): 1639–
Chetty, Raj, John N. Friedman, Soren Leth- Petersen, Torben Heien Nielsen, and
Tore Olsen. 2012. “Active vs. Passive Decisions and Crowdout in Retirement Savings
Accounts: Evidence from Denmark.” NBER Working Paper 18565, National Bureau
of Economic Research, Cambridge, MA.
Chiteji, Ngina, and Lina Walker. 2009. “Strategies to Increase the Retirement Savings
of African American Households.” In Automatic: Changing the Way America Saves,
edited by William G. Gale, J. Mark Iwry, David C. John, and Lina Walker, 231– 260.
Washington, DC: Brookings Institution Press.
Conly, S. 2013. Against Autonomy. Cambridge: Cambridge University Press.
24, 123 Stat 1734, codified in various sections
Credit CARD Act. 2009. Pub L No 111–
of Titles 15 and 16.
Deb, Partha, and Carmen Vargas. 2016. “Who Benefits from Calorie Labeling? An
Analysis of its Effects on Body Mass.” NBER Working Paper 21992, National Bureau
of Economic Research, Cambridge, MA.
Department of Agriculture. 2011. “Direct Certification and Certification of Home-
less, Migrant and Runaway Children for Free School Meals.” 76 Federal Register
22785–02, 22793.
Department of Health and Human Services. 2010. Centers for Disease Control and
Prevention, Community Health Status Indicators (CHSI) to Combat Obesity, Heart
Disease and Cancer, May 1. Retrieved from http://www.data.gov/raw/2159.
DG SANCO (European Commission’s Directorate General for Health and Consum-
ers). 2010. Consumer Behaviour: The Road to Effective Policy- Making. Retrieved from
http://e c.e uropa.e u/c onsumers/d ocs/1 dg-s anco-b rochure-c onsumer-b ehaviour
-final.pdf.
Dinner, Isaac, Daniel G. Goldstein, Eric J. Johnson, and Kaiya Liu. 2009. “Partition-
ing Default Effects: Why People Choose Not to Choose.” Unpublished manuscript.
499,
Emergency Planning and Community Right to Know Act. 1986. Pub L No 99–
100 Stat 1728, codified at 42 USC § 11001 et seq.
Environmental Protection Agency. 2009. “Fuel Economy Labeling of Motor Vehicles:
Revisions to Improve Calculation of Fuel Economy Estimates.” 74 Federal Register
61537– 53 (amending 40 CFR Parts 86, 600).
01, 61542, 61550–
368 Cass R. Sunstein
European Commission. 2012. “Green Behavior.” Future Brief 4, Science for Environ-
ment Policy, October. http://ec.europa.eu/environment/integration/research/news
alert/pdf/FB4.pdf.
Gale, William G., J. Mark Iwry, and Spencer Walters. 2009. “Retirement Savings for
Middle-and Lower- Income Households: The Pension Protection Act of 2006 and
the Unfinished Agenda.” In Automatic: Changing the Way America Saves, edited by
27. Washington,
William G. Gale, J. Mark Iwry, David C. John, and Lina Walker, 11–
DC: Brookings Institution Press.
Greenstone, Michael. 2009. “Toward a Culture of Persistent Regulatory Experimen-
tation and Evaluation.” In New Perspectives on Regulation, edited by David Moss and
125. Cambridge: The Tobin Project.
John Cisternino, 111–
Halpern, David. 2015. Inside the Nudge Unit. London: Ebury.
Hamilton, James T. 2005. Regulation through Revelation: The Origin, Politics, and Impacts
of the Toxics Release Inventory Program. Cambridge: Cambridge University Press.
Free Kids Act. 2012. Pub L No 111–
Healthy, Hunger- 296, 124 Stat 3183.
Heath, Chip, and Dan Heath. 2010. Switch: How to Change Things When Change Is
Hard. New York: Broadway.
Hirshleifer, David. 1995. “The Blind Leading the Blind: Social Influence, Fads, and
Informational Cascades.” In The New Economics of Human Behavior, edited by Mariano
Tommasi and Kathryn Ierulli, 188– 215. Cambridge: Cambridge University Press.
Internal Revenue Service. 2009. “Retirement and Savings Initiatives: Helping Ameri-
cans Save for the Future.” http://www.irs.gov/pub/irs-tege/rne_se0909.pdf.
iNudgeYou.com. n.d. “Resources.” http://www.inudgeyou.com/resources.
Jolls, Christine. 1998. “Behavioral Economics Analysis of Redistributive Legal Rules.”
Vanderbilt Law Review 51 (6): 1653– 1677.
Kahneman, Daniel. 2011. Thinking, Fast and Slow. New York: Farrar, Straus, and Giroux.
Kahneman, Daniel, and Shane Frederick. 2002. “Representativeness Revisited: Attri-
bute Substitution in Intuitive Judgment.” In Heuristics and Biases, edited by Tom
Gilovich, Dale Griffin, and Daniel Kahneman, 49–81. Cambridge: Cambridge Uni-
versity Press.
Kahneman, Daniel, and Amos Tversky. 1979. “Prospect Theory: An Analysis of Deci-
292.
sion under Risk.” Econometrica 47 (2): 263–
Leventhal, Howard, Robert Singer, and Susan Jones. 1965. “Effects of Fear and Speci-
ficity of Recommendation upon Attitudes and Behavior.” Journal of Personality and
29.
Social Psychology 2 (1): 20–
Behaviorally Informed 369
Levin, Irwin P., Sandra L. Schneider, and Gary J. Gaeth. 1998. “All Frames Are Not
Created Equal: A Typology and Critical Analysis of Framing Effects.” Organizational
188.
Behavior and Human Decision Processes 76 (2): 149–
Loewenstein, George F., Elke U. Weber, Christopher K. Hsee, and Ned Welch. 2001.
286.
“Risk As Feelings.” Psychological Bulletin 127 (2): 267–
Madrian, Brigitte C., and Dennis F. Shea. 2001. “The Power of Suggestion: Inertia in
401(k) Participation and Savings Behavior.” Quarterly Journal of Economics 116 (4):
1149–1187.
McFerran, Brent, Darren W. Dahl, Gavan J. Fitzsimons, and Andrea C. Morales.
2011. “How the Body Type of Others Impacts Our Food Consumption.” In Lever-
aging Consumer Psychology for Effective Health Communications, edited by Rajeev
170. Armonk, NY: M. E.
Batra, Punam Anand Keller, and Victor J. Strecher, 151–
Sharpe.
McGraw, A. Peter, Jeff T. Larsen, Daniel Kahneman, and David Schkade. 2010.
1445.
“Comparing Gains and Losses.” Psychological Science 21 (10): 1438–
Nickerson, David W., and Todd Rogers. 2010. “Do You Have a Voting Plan? Imple-
mentation Intentions, Voter Turnout, and Organic Plan Making.” Psychological Sci-
199.
ence 21 (2): 194–
Obama, Barack H. 2009. Weekly address, September 5.
Office of Management and Budget, Office of Information and Regulatory Affairs.
2010. Information collection budget of the United States government. http:// www
.whitehouse.gov/sites/default/files/omb/inforeg/icb/icb_2010.pdf.
OECD (Organisation for Economic Co- operation and Development). 2010. Con-
sumer Policy Toolkit. Paris: OECD Publishing.
www
Orszag, Peter. 2010. OMB, Director, SAVEings, March 29. http:// whitehouse
.
.gov/omb/blog/10/03/29/SAVEings/.
Orszag, Peter, and Eric Rodriguez. 2009. “Retirement Security for Latinos: Bolstering
Coverage, Savings, and Adequacy.” In Automatic: Changing the Way America Saves,
edited by William G. Gale, Mark J. Iwry, David C. John, and Lina Walker, 173– 198.
Washington, DC: Brookings Institution Press.
Papke, Leslie E., Lina Walker, and Michael Dworsky. 2009. “Retirement Security for
Women: Progress to Date and Policies for Tomorrow.” In Automatic: Changing the
Way America Saves, edited by William G. Gale, J. Mark Iwry, David C. John, and Lina
Walker, 199–230. Washington, DC: Brookings Institution Press.
280, 120 Stat 780, codified in various
Pension Protection Act. 2006. Pub L No 109–
sections of Titles 26 and 29.
370 Cass R. Sunstein
Redelmeier, Donald A., Paul Rozin, and Daniel Kahneman. 1993. “Understanding
Patients’ Decisions: Cognitive and Emotional Perspectives.” Journal of the American
76.
Medical Association 270 (1): 72–
Rozin, Paul, Sydney Scott, Megan Dingley, Joanna K. Urbanek, Hong Jiang, and Mark
Kaltenbach. 2011. “Nudge to Nobesity I: Minor Changes in Accessibility Decrease
332.
Food Intake.” Judgment and Decision Making 6 (4): 323–
Schwartz, Janet, Jason Riis, Brian Elbel, and Dan Ariely. 2011. “Would You Like to
Downsize That Meal? Activating Self- Control Is More Effective Than Calorie Labeling
in Reducing Calorie Consumption in Fast Food Meals.” Unpublished manuscript.
Sharot, Tali. 2011. The Optimism Bias: A Tour of the Irrationally Positive Brain. New
York: Knopf.
Shu, Lisa L., Nina Mazar, Francesca Gino, Dan Ariely, and Max H. Bazerman. 2012.
“Signing at the Beginning Makes Ethics Salient and Decreases Dishonest Self-Reports
in Comparison to Signing at the End.” Proceedings of the National Academy of Sciences
109 (38): 15197–15200.
Slovic, Paul. 1998. “Do Adolescent Smokers Know the Risks?” Duke Law Journal
47 (6): 1133–1141.
Stango, Victor, and Jonathan Zinman. 2011. “Limited and Varying Consumer Atten-
tion: Evidence from Shocks to the Salience of Bank Overdraft Fees.” Working Paper
11–17, Federal Reserve Bank of Philadelphia.
Sunstein, Cass R. 2010. “Administrator, OIRA, Memorandum for the Heads of Execu-
tive Departments and Agencies, Disclosure and Simplification as Regulatory Tools.”
http://www.whitehouse.gov/sites/default/files/omb/assets/inforeg/disclosure_prin
ciples.pdf.
Sunstein, Cass R. 2013. Simpler. New York: Simon and Schuster.
Sunstein, Cass R. 2015. Choosing Not to Choose. Oxford: Oxford University Press.
Sunstein, Cass R. 2016. The Ethics of Influence. New York: Cambridge University Press.
Thaler, Richard H. 2015. Misbehaving. New York: Norton.
Thaler, Richard H., and Cass R. Sunstein. 2008. Nudge. New Haven, CT: Yale Univer-
sity Press.
Thaler, Richard H., Daniel Kahneman, and Jack L. Knetsch. 1991. “Experimental
Tests of the Endowment Effect and the Coase Theorem.” In Quasi Rational Econom-
188. New York: Russell Sage.
ics, edited by Richard H. Thaler, 167–
Tversky, Amos, and Daniel Kahneman. 1973. “Availability: A Heuristic for Judging
232.
Frequency and Probability.” Cognitive Psychology 5 (2): 207–
Behaviorally Informed 371
Weinstein, Neil D. 1987. “Unrealistic Optimism about Susceptibility to Health Prob-
lems: Conclusions from a Community- Wide Sample.” Journal of Behavioral Medicine
500.
10 (5): 481–
Zwane, Alix Peterson, Jonathan Zinman, Eric Van Dusen, William Pariente, Clair
Null, Edward Miguel, Michael Kremer, et al. 2011. “Being Surveyed Can Change
Later Behavior and Related Parameter Estimates.” Proceedings of the National Academy
1826.
of Sciences 108 (5): 1821–
9 CFR § 317.309.
Comment: Robert Hockett
Choice Architectures: An Appreciation and a Provisional Suggestion
I.
I have long been intrigued, and occasionally maddened, by certain idiom-
atic crazes or fads that seem constantly to break out and spread through
American society. Advertisers now hawk “solutions,” for example, rather
than goods and services. And of course, it has been decades by this point
that we have been bringing past conversations up into the present by say-
…”
ing “I’m like …,” not “I said.
A recent development along these lines that I find especially amusing is
heard expression, “a thing.” Each of “Benghazi” and “the 47
the now oft-
percent,” for example, for a time was said to have become “a thing.” Like-
wise Hillary Clinton’s emails and Donald Trump’s “Tweets.” Pretty much
every new entrant to the Grand Guignol theater of public consciousness
and conversation these days is a “thing” in the requisite sense. By this cri-
perhaps a sort of
terion, I suppose that “a thing” is itself now a thing—
referential thing.
recursive, reflexive, or self-
II.
In the academy, behavioralism seems to have become “a thing” by the late
1970s or early 1980s at latest, notwithstanding the fact that discoveries such
as the Allais and Ellsberg “paradoxes,” then Herbert Simonian “bounded
Broad thanks to participants at the “State of the Economy, State of the World” con-
ference held at the World Bank in Washington, DC, in June 2016. Special thanks to
Kaushik Basu and Cass Sunstein.
Behaviorally Informed 373
rationality,”1 evidenced certain systematic departures from orthodox mod-
els of choice behavior much earlier.
In my own case, I think behavioralism became something of “a thing”
with the reading of two authors during the late 1990s: first, a man who later
became one of my dissertation advisors, Bob Shiller at Yale; and second,
the man on whose vast and still growing body of work I am to comment
the phenomenal Cass Sunstein.
today—
Bob first got me to thinking about the work of Dick Thaler in particular—
especially what I call “endowment psychology” (not to mention Cornell
coffee mugs),2 which I thought a helpful way of explaining my own long-
standing intuition that what is now coming to be called “predistribution”
might prove more politically stable than redistribution as a means of
redressing distributive injustice.3 This in turn harmonized well with what
had drawn me to Bob as a mentor in the first place, for my aim was to
improving predistributive
develop means of financially engineering justice-
schemes, the ultimate upshot of which is a book now forthcoming from
Yale University Press.4
Cass came into the picture for me with what I suppose was then
hundredth book—
merely his eight- Free Markets and Social Justice, pub-
lished in 1998. I virtually devoured this rich, rich collection of previ-
ously published essays and articles, and learned much from it. But what
I think stuck with me most was Cass’s emphasis on the endogeneity of
preferences, as well as his patient tracing of normatively interesting con-
sequences therefrom.
endogeneity as an objection
Now of course, I’d been aware of preference-
to certain attempts at theorizing justice, thanks to Amartya’s celebrated
“tame housewife” objection, and Jerry Cohen’s cognate “Tiny Tim” objec-
tion, to certain assumptions that figured centrally in liberal accounts of jus-
tice.5 (John Roemer and Jerry were, like Bob Shiller, very patient mentors.)
And I’d been aware of Gary Becker’s work on tastes in micro theory. But it
1. See Allais (1953), Ellsberg (1961), and Simon (1991). It should be noted that Ells-
berg’s paradox effectively appears earlier in Keynes (1921, 75–76, n. 2).
2. See, for example, Hockett (2005, 2006, 2007, 2008a).
3. See sources cited in Hockett (2005, 2006, 2007, 2008a).
4. See Hockett (2017, forthcoming).
5. See Cohen (1989) and Sen (1995).
374 Comments by Robert Hockett and Varun Gauri
was Cass and his reflections that most aided me in thinking comprehen-
sively, in both a broadly transdisciplinary and a more systematically pro-
endogeneity and its implications. So
grammatic manner, about preference-
my remarks here will be one part encomium, one part elaboration, and one
perhaps in the direc-
part halfway provocative suggestion for further work—
tion of what I’ll call a sort of “behavioral macro” or “liberal collectivism.”6
III.
Let me begin, then, by noting a certain family resemblance between classi-
cal liberalism in political theory and the classical choice model in welfare-
economic theory. If we take Rawlsian justice theory as emblematic of
liberalism in the modern era, then in liberalism we find a political ideal
that is essentially indifferent to the origins or nature of preferences and is
concerned instead with what Rawlsians call “the basic structure” in which
satisfactions or “lifeplans” are pursued or executed.7 This con-
preference-
though, as I shall claim, misleading—
cern finds partial— programmatic
expression in the Rawlsian doctrine’s commitment to what Rawls called
“the priority of the right over the good.”8
Analogously, in classical welfare theory, we find preferences to be like-
placed outside of—
wise bracketed— the field of disciplinary inquiry. They
no more subject to rational cri-
are, that is to say, treated as exogenous—
tique than Rawlsian life plans are subject to normative political critique.
Discussion and disputation accordingly center on the formal properties of
the social welfare function or functional that aggregates preferences. The
social welfare function aggregation rule, pursuant to the dominant research
program, accordingly plays a role here analogous to “basic structure” in
normative liberal political theory à la Rawls.9
Now, as is well known, Rawlsian liberalism came under sustained scru-
tiny and critique during the 1970s and after. One grounds for criticism was
the account’s implausibly denuded conception of the choosing liberal self
behind the Rawlsian (or should we say Harsanyian)10 veil of ignorance. The
6. See, for example, Hockett (2013a).
7. See generally Rawls (1971).
8. Rawls (1971).
9. For more on this link, see Hockett (2008b, 2009).
10. See Harsanyi (1953, 1955).
Behaviorally Informed 375
“unsituated self,” as Michael Sandel later canonically dubbed it,11 became
something of an albatross for liberal justice theory, both for reasons of
normative attractiveness (cf. Cohen 1897; Sen 1995) and for reasons of
theoretic intelligibility (cf. Sandel (1982) and others).12 So-
called commu-
nitarians and, more broadly, communicative action theorists, actuated by
critiques of this general form, in consequence steadily wrought a manner
a revolution whose best-
of “contextualizing” revolution in justice theory—
known exponents at present are probably Jürgen Habermas, Axel Honneth,
and Rainer Forst.13
Against this backdrop, I think, one helpful way of viewing the behav-
ioralist revolution in normative economics and economic analysis of law—
is
particularly as systematized, interpreted, and further developed by Cass—
minded choice-
as a thoroughly and programmatically- theoretic analog to
the “communitarian” revolt against liberal justice theory. Situating the
Rawlsian unsituated self is, perhaps, best and most thoroughly done by first
theoretic chooser.
comprehensively endogenizing the classical choice-
This is, in part, precisely what Cass’s thoroughly cataloging, system-
atizing, and further advancing of behavioralist learning does. For what
inertia, framing, salience-
are careful attention to choice- attending, loss-
aversion, social influences, heuristics, implicit probability assumptions,
and so forth if not ways of thoroughly endogenizing preferences and,
therefore, more fully situating actual choosing selves? And if, with Cass
authors, we can do this both comprehensively and with an
and his co-
eye to normative significance, then we stand to develop both better posi-
tive and better normative microeconomic, welfare economic, and justice
theory. Pretty exciting stuff!
IV.
But now here is what I think might be most exhilarating of all in Cass’s
recent work: His achievements, although they began as theoretic advances,
have rapidly opened the door to more practical, “applied” advances as well.
11. See Sandel (1982).
12. My colleague Steve Shiffrin often says that “children are the Achilles Heal of
liberalism.” This seems to me nicely to capture both preference-endogeneity and
intelligibility objections in a single slogan.
13. See, for example, Habermas (1996), Forst (2002), and Honneth (2014).
376 Comments by Robert Hockett and Varun Gauri
By attending to the whole of the “choice architecture,” as Cass dubs it,
shaping he studies jointly constitute,
which the many forms of preference-
we soon spot a novel way to skirt a particularly vexed clash of values in
modern Western and, especially, US intellectual and political history.
I allude to the clash between what Rawls would call “liberalism and perfec-
tionism,” and what Cass and Thaler call “libertarianism and paternalism.”14
In effect, Cass and Thaler note, we can, by carefully studying and incre-
mentally improving choice architecture, both improve aggregate welfare—
something like what Rawls would call “the good”15—
and avoid any serious,
what Rawls would call
non de minimus affront to individual freedom—
respect for “the priority of liberty.”16
We can, in other words, act on a sort of commonsense, nonperfectionist
and nondogmatic view of the collective good while still allowing for indi-
outs by those who, upon consideration, still prefer to choose as
vidual opt-
they would have done under an earlier architecture. In this way, we get to
have a bit of our cake while eating it, too, sidestepping irresoluble conflicts
over totalizing visions of “the Good,” rather as Cass recommended long
ago, in a different context, under the rubric of what he called “incompletely
theorized agreements.”
We encourage or facilitate the making of choices that most would think
wise, in other words, without outright coercing them. This is an achieve-
ment on par, in my view, with Lock’s classic work on toleration and Mill’s
on liberty many decades ago. And it is apt to be rather more effective, in
my humble opinion, than Rawls’s late 1990s offering of a “political, not
metaphysical” account of liberal justice.17
All right, so there’s the encomium. Now for a brief closing suggestion
but I think only a little—
that might be a little— provocative. I want to sug-
gest that we might also encourage some socially beneficial choices with-
out outright coercing them through means additional to Cass’s style of
14. See Thaler and Sunstein (2008).
15. Though Rawls himself of course tends not to aggregate, since he brackets “the
good.” (A possible exception comes in the form of “the good of the worst- off,” whose
off” embraces a class
lot Rawls’s “difference principle” aims to optimize. If the “worst-
rather than a person— Rawls doesn’t tell us which— then of course there is aggrega-
tion at least with respect to the good of this class.)
16. Rawls (1971).
17. See Rawls (1996).
Behaviorally Informed 377
architecture reconstruction. Here I allude to work I’ve been doing
choice-
in recent years, some with my colleague Saule Omarova, on what I call
“private means to public ends.” In particular, I have in mind making more
acting roles that government
thoughtful, deliberate use of certain market-
instrumentalities often play in our macroeconomy.
Here’s what I mean. I’ve worked on and off at the Federal Reserve Bank
of New York (or “New York Fed”) in the past, and I am struck by how few
people seem to know anything about what is, by any measure, the most
critical function discharged by this remarkable institution each day. I mean
by-
the actual implementation of monetary policy, on a literal day- day
basis, by the New York Fed trading desk in lower Manhattan. By transacting
in massive quantities of (mainly) US Treasury securities with private dealer
banks each morning, this desk injects money into, or retracts money out of,
our banking and broader financial markets each day, thereby determining
borrowing costs and, we hope, the pace of activity throughout the broader
economy.18
Now, one way to conceive and then generalize from this literally quo-
governmental activity is to think of it as something that I call
tidian quasi-
moving.” A particularly important variable—
“market- what in other work
is deliberately
I call a “systemically important price or index,” or “SIPI”—
“moved” by a government instrumentality that acts pursuant to the same
modalities as do other, nongovernmental actors in the very same markets.
All that differs is the object of the activities in question.
Once we recognize that prevailing interest rates are but one of many
publicly cognizable SIPIs out there in our markets, it is easy to imagine
why and how we might wish to generalize from the New York Fed’s open
market operations to something that I call “open market operation plus” in
connection with other SIPIs.19 We might wish to move particularly impor-
tant commodity prices (e.g., foodstuffs or fuel) during a period of danger-
ous volatility,20 for example, or prevailing wage rates during a deflationary
slump.21 Or we might have acted to put downward pressure on secondary
backed
credit or mortgage markets during the junk bond and mortgage-
18. See, for example, Hockett and Omarova (2014).
19. See Hockett and Omarova (2015).
20. See Hockett (2011).
21. Hockett and Omarova (2014, 2015).
378 Comments by Robert Hockett and Varun Gauri
security (MBS) hyperinflations of the late 1980s and early 2000s, respec-
tively, or on health insurance prices right now through a “public option”
on to “Obamacare.”22
add-
enhancing market-
Once you start thinking about it, broadly welfare-
moving strategies of this kind come quite rapidly to mind. But my taxon-
moving.”
omy includes other modalities additional to what I call “market-
making,” in the sense meant by financial mar-
One such I call “market-
ket participants. This is partly what Fannie Mae was established to do in
to make a secondary market in mortgage loans so as to lower credit
1938—
era real
costs in the primary markets and thereby stabilize Depression-
estate markets and the home construction industry while raising home
ownership rates.23 That was a system that worked wonderfully for nearly
60 years until underregulated private investment banks got into the act
and blew everything up.24 The New York Fed’s Maiden Lane funds, spe-
cially created for the purpose, acted similarly in connection with MBSs to
stem an individually rational but collectively irrational run on MBSs from
preserving” role that was effec-
2008 into 2012, in what I call a “market-
tively taken over by the Fed Board itself via the third round of quantitive
easing in October 2012.
V.
These are but a few of the many examples that I elaborate elsewhere. I won’t
bore you with more of them here; those who are interested can take a look
at the works I cite in the footnotes. My object for present purposes is simply
to suggest that in some cases, there might be other avenues, additional to
Cass’s style of choice architecture, through which to influence preferences
in what nearly all would agree to be socially desirable ways, without out-
right coercing them.
It is true that my “big market actor” strategy might, if used for some
conceivable purposes, edge closer to coercion than do Cass’s strategies,
inasmuch as it imposes higher costs on contrarians than do Cass’s default-
ins to opt-
switches from opt- outs. But these seem to me differences of
22. Hockett (2010) and Hockett and Omarova (2014, 2015).
23. See Hockett (2006).
24. Hockett (2006) and also Hockett (2013b).
Behaviorally Informed 379
degree rather than of kind. And because most (if not all) entries on my
proposed menu of market actor roles aim to solve what I call “recursive col-
lective action problems” that everyone can plausibly be presumed to wish
to solve, rather than systematically to coerce choice,25 it might even be the
case that my proposals “impose” no more on individual choosers than do
Cass’s.
We have barely begun to explore these proposals’ potentials. I suspect
now that once we do, we shall see quickly that they can both complement
and supplement the impressive array of entries on Cass’s proposed menu.
VI.
And with that I shall close. To the vanishingly few of you here who might not
astonishing, proceeding as it does from
be familiar with Cass’s vast oeuvre—
I’ll say no more at present than please take a look! And
one still so young—
to Cass himself, I say one more time: Thank you, and please keep it coming!
References
Allais, Maurice. 1953. “Le Comportement de L’homme Rationnel Devant le Risque:
Critique des Postulats et Axiomes de L’Ecole Américaine.” Econometrica 21 (4):
503–546.
944.
Cohen, G. A. 1989. “On the Currency of Egalitarian Justice.” Ethics 99 (4): 906–
Ellsberg, Daniel. 1961. “Risk, Ambiguity, and the Savage Axioms.” Quarterly Journal
669.
of Economics 75 (4): 643–
Forst, Rainer. 2002. Contexts of Justice. Berkeley, Los Angeles: University of California
Press.
Habermas, Jürgen. 1996. Between Facts and Norms. Cambridge, MA: MIT Press.
Harsanyi, John. 1953. “Cardinal Utility in Welfare Economics and in the Theory of
435.
Risk-Taking.” Journal of Political Economy 61 (5): 434–
Harsanyi, John. 1955. “Cardinal Welfare, Individualistic Ethics, and Interpersonal
321.
Comparisons of Utility.” Journal of Political Economy 63 (4): 309–
Hockett, Robert C. 2005. “Whose Ownership? Which Society?” Cardozo Law Review
103.
27 (1): 1–
25. See Hockett (2015).
380 Comments by Robert Hockett and Varun Gauri
Hockett, Robert C. 2006. “A Jeffersonian Republic by Hamiltonian Means.” Southern
164.
California Law Review 79 (1): 45–
Hockett, Robert C. 2007. “What Kinds of Stock Ownership Plans Should There Be?”
952.
Cornell Law Review 92 (5): 865–
Hockett, Robert C. 2008a. “Insource the Shareholding of Outsourced Employees: A
426.
Global Stock Ownership Plan.” Virginia Law & Business Review 3 (2): 357–
Hockett, Robert C. 2008b. “Pareto Versus Welfare.” Cornell Legal Studies Research
031, Cornell University, Ithaca, NY.
Paper 08–
Hockett, Robert C. 2009. “Why Paretians Can’t Prescribe: Preferences, Principles
and Imperatives in Law and Policy.” Cornell Journal of Law and Public Policy 18 (2):
391–476.
Hockett, Robert C. 2011. “How to Make QE More Helpful: By Fed Shorting of Com-
modities.” Benzinga, October 11. https://www.benzinga.com/news/11/10/1988109
/how-to-make-qe-more-helpful-by-fed-shorting-of-commodities.
114.
Hockett, Robert C. 2013a. “The Libertarian Welfare State.” Challenge 56 (2): 100–
Hockett, Robert C. 2013b. “Paying Paul and Robbing No One: An Eminent Domain
Solution for Underwater Mortgage Debt.” Current Issues in Economics and Finance
19 (5): 1–12.
Hockett, Robert C. 2015. “Recursive Collective Action Problems.” Journal of Financial
128.
Perspectives 3 (2): 113–
Hockett, Robert C. 2017. A Republic of Producers. Forthcoming.
Hockett, Robert C., and Saule Omarova. 2014. “‘Private’ Means to ‘Public’ Ends.”
576.
Theoretical Inquiries in Law 15 (1): 530–
Hockett, Robert C., and Saule Omarova. 2015. “‘Public’ Actors in ‘Private’ Markets.”
176.
Washington University Law Review 93 (1): 103–
Honneth, Axel. 2014. Freedom’s Right. New York: Columbia University Press.
Keynes, John Maynard. 1921. A Treatise on Probability. London: Macmillan & Co.
Rawls, John. 1971. A Theory of Justice. Cambridge, MA: Belknap Press of Harvard
University Press.
Rawls, John, 1996. Political Liberalism. New York: Columbia University Press.
Sandel, Michael J. 1982. Liberalism and the Limits of Justice. Cambridge: Cambridge
University Press.
Sen, Amartya K. 1995. “Equality of What?” In Equal Freedom: Selected Tanner Lectures on
Human Values, edited by Stephen Darwall. Ann Arbor: University of Michigan Press.
Behaviorally Informed 381
Simon, Herbert. 1991. “Bounded Rationality and Organizational Learning.” Organi-
134.
zation Science 2(1): 125–
Thaler, Richard H., and Cass Sunstein. 2008. Nudge. New Haven, CT: Yale University
Press.
Comment: Varun Gauri
Nudging Goes Global
All over the world, policy making is being nudged. A partial list of govern-
ments that have begun, systematically, to use behavioral economics in their
policies and programs comprises the United Kingdom, the United States,
Chicago, New York, Washington, DC, Rio de Janeiro, New South Wales, New
Zealand, the Western Cape, Guatemala, the Netherlands, France, Peru,
Canada, Denmark, Indonesia, Lebanon, the UAE, Poland, Latvia, Moldova,
Japan, Germany, Singapore, and India. World Bank teams, including the
Mind, Behavior, and Development Unit (eMBeD), are involved in dozens of
ongoing projects that incorporate social and behavioral insights. Cass Sun-
stein’s work, crystallized in his book Nudge with Richard Thaler, has been
seminal; it has genuinely changed policy making the world over.
As the use of behavioral economics has moved from the periphery to the
mainstream, it is worth reflecting on some of the outstanding questions
and criticisms that confront the practice. Sunstein’s essay in this volume is,
like his work more broadly, not only thorough (in the sense that it success-
fully organizes a wide range of theory and evidence), but also thoughtful
(in the sense that it rewards close reading). In what follows, I use excerpts
from Sunstein’s essay as a point of departure to raise, in a preliminary way,
four issues related to the behavioral economics and policy making agenda.
It is also the case, as I will make clear, that Sunstein’s own work has antici-
pated the pathways through which one can make advances on some of
these questions.
Behaviorally Informed 383
For Which People Are Nudges Liberty Preserving?
Suppose, for example, that a particular default rule would place a strong majority
of the relevant population in the situation that they would favor if they made
an informed choice. If so, there is a legitimate decision reason to adopt that
default rule (with the understanding that for those who differ from the majority,
it remains possible to opt out).
Cass Sunstein, (forthcoming)
—
Because most people are myopic and/or otherwise inattentive, because
they view default savings plans as authorized or as important reference
points, automatic enrollment in a retirement savings plan increases mean
retirement savings. Subsequently allowing people to opt out preserves their
liberty to make significant choices regarding their own lives. Because there
either individuals are not enrolled
must be a default rule of some sort—
why not choose
and can opt in, or they are enrolled and can opt out—
the default rule that increases savings? This is the logic of libertarian
paternalism.
Notice, however, that the formulation trades on two different under-
standings of liberty: positive and negative (Berlin 1969). Automatic enroll-
ment appeals to positive liberty: Myopia and inattention are external
sources of “control or interference,” to use Berlin’s language, that affect
what people do. Automatic enrollment helps them achieve their true objec-
tives. But the power to opt out, once one is automatically enrolled, is a
negative liberty: Factors external to the will, such as myopia or inattention,
still limit the capacity of an automatically enrolled saver to opt out. These
enrolled savers are free in the negative sense that they can choose to disen-
roll without any obstruction by other persons.
Although space is insufficient to spell out the argument in detail, it
seems to be the case, then, that automatic retirement savings is not “liberty
preserving” in a simple way. Elsewhere, Sunstein (2012) comes to a similar
conclusion by referencing a continuum between soft and hard paternal-
ism, which is scaled by the sum of material and psychic costs imposed. He
preserving)
describes most “nudges” as a kind of soft (if not entirely liberty-
paternalism and argues that most people in fact opt out of defaults that are
decreasing (Beshears et al. 2010).
welfare-
term “nudge” agenda, particu-
But the paternalism challenge to the long-
larly in developing countries, will require further elaboration on the part
384 Comments by Robert Hockett and Varun Gauri
of those of us engaged in it. To take up just two points. First, it will not be
enough to say that most people opt out of bad defaults. We also need to
know who opts out, and much more about how the capacity to identify
improving choices and take advantage of information disclosure is
welfare-
related to poverty (Mani et al. 2013), as well as to gender and other norma-
tively important social categories.
Second, as nudging goes global and begins to work in cultural environ-
ments very different from the United States and the United Kingdom, where
it began, it may be that in many contexts, what is ethically salient is not the
extent to which a behavioral intervention constrains liberty, understood as
the sum of material and psychic costs imposed by a policy, but the intrinsic
ethical value of the program itself. Indeed, informal conversations suggest
that policy makers in many countries are not particularly troubled by the
paternalism question, because liberalism is not the assumed background of
such as “development” or “harmony” or
ethical evaluation. Other goals—
are often more prominent. Indeed, it might be helpful if
“social justice”—
policy makers were more troubled than they are by nudging policies. Those
working in the field might make a contribution to democratic policy mak-
ing around the world by insisting that nudgers disclose and debate their
nudge policies.
How Social Norms Change
Consider as well the problem of distracted driving. On October 1, 2009, the presi-
dent issued an executive order that bans federal employees from texting while
driving. Such steps can help promote a social norm against texting while driving,
thus reducing risks.
Cass Sunstein, (forthcoming)
—
We know that social norms are crucial drivers of behavior, but how can policy
makers shift them? One approach has been to activate existing social norms,
particularly empirical knowledge or expectations regarding modal behavior
in a group. Interventions in that vein have reduced road accidents (Habyari-
mana and Jack 2011), increased tax compliance (Hallsworth et al. 2014), and
successfully promoted energy conservation (Allcott and Rogers 2014).
Sunstein’s interpretation of the White House order on texting while
driving, however, is more ambitious. It is about creating a new social norm,
Behaviorally Informed 385
not merely activating an existing one. One analogy in developing countries
is a law requiring candidates for village council elections in Haryana, India,
to have a functioning toilet in their homes.1 The idea, as in the texting
law, is that public officials can serve as role models or otherwise inspire a
shift in the behavior in the general population. But social norms operate
in reference groups, and if public officials are not in the reference group of
the target population, their behavior might not motivate people to behave
differently, or may even even backfire. For instance, villagers in Haryana
might come to think that toilets are just for government workers and other
important people, not for ordinary folk.
Another analogy is early legislation in American states that made it
illegal for anyone who had engaged in a duel from holding public office
(Lessig 1995). Those laws, though not successfully enforced, were intended
to allow a gentleman to decline a challenge to a duel by appealing to, rather
he could say that because honor
than shrinking from, the honor code—
required him to serve the public, and dueling would make public service
impossible, he had no choice but to decline.
The target of the antidueling rules was elite behavior, but duels were
highly visible events, so it was possible that their disappearance would pro-
mote democratic sensibilities and the ethos of nonviolence more generally.
In contrast, texting is not easily observed; even if public officials stop tex-
ting, the general public may not realize it.
The general point is that scholars have taught us some things about
social norms (Sunstein 1996), and policy makers are coming to recog-
nize the value of activating them. But we know much more about the
comparative statics of social norms, and about norm unraveling through
bandwagon effects and pluralistic ignorance, than about norm emergence
and creation.
1. The law also requires minimum educational qualifications, not having defaulted
in cooperative loans or having outstanding dues on rural domestic electricity con-
nections, and not having been charged by a court for a grave criminal offense. The
Supreme Court of India upheld the law in December 2015. See http:// www livemint
.
.com/Politics/KTRLWs6xYd6OlfSKC3SRHL/Supreme-Court-upholds-Haryana-law
-on-Panchayat-polls.html.
386 Comments by Robert Hockett and Varun Gauri
Outcomes
If people learn that they are using more energy than similarly situated others,
their energy use may decline— saving money while also reducing pollution.
Cass Sunstein, (forthcoming)
—
A full assessment of the effects of home energy reports includes, in addition
to lower pollution and savings, the expenditures associated with efficiency-
improving capital investments (as when a homeowner purchases new
appliances or windows) and, to the extent it can be accurately measured,
the hedonic cost of tolerating a hotter or colder living environment in the
home (Allcott and Kessler 2015). Although everyone might agree that in
theory, those factors should also be included when evaluating the overall
effects of a behavioral intervention, they are not usually included in prac-
tice. Too often, evaluations of behavioral policies focus almost exclusively
on the intended behavioral change. When possible, assessments of behav-
being, and not
ioral policies should focus on the effects on overall well-
just on behavior itself. Similarly, although there are good reasons to think
term impact (Madrian
that some behavioral interventions can have long-
and Shea 2001; Yeager and Walton 2011), practitioners would like to know
more about the kinds of interventions and circumstances under which
term as opposed to ephemeral effects are achieved.
long-
Nudging the Nudgers
It should not be necessary to emphasize that public officials are subject to error
as well.
Cass Sunstein, this volume
—
Although the potential value of behavioral insights in developing countries
is substantial (World Bank 2014), one concern is that the successful formu-
lation and implementation of all policies, behaviorally informed or not,
requires the capacity to recruit, motivate, and supervise an effective bureau-
out retirement savings plans, for example, are built on finan-
cracy. Opt-
cial, regulatory, and informational infrastructure that cannot be taken for
granted in many countries. More generally, few now question the negative
and not just market failure—
impact of government failure— on economic
Behaviorally Informed 387
development (Bardhan 2015). Bureaucrats are subject to many of the cog-
nitive biases that everyone else is, including sunk cost bias, cultural cogni-
tion, and inaccurate assessment of risks (Banuri, Dercon, and Gauri 2016).
Can social and behavioral insights improve governance? Some preliminary
evidence suggests that they can. For instance, unexpected payments can
motivate workers, even if the money is not tied to performance (Hossain
and List 2012); peer effects seem to improve productivity (Mas and Moretti,
2009); and social recognition can improve performance (Ashraf, Bandiera,
and Lee 2014).
There remains to be developed an extremely interesting and potentially
very useful agenda related to the use of social and behavioral insights to
promote professional norms, bureaucratic identities, impartial and sound
decision making, and productivity in the public sector. As elsewhere, Sun-
stein’s writings have anticipated this line of research (Sunstein and Hastie
2015). With luck, this commentary will nudge him to expand it.
References
Allcott, Hunt, and Judd B. Kessler. 2015. “The Welfare Effects of Nudges: A Case
Study of Energy Use Social Comparisons.” NBER Working Paper 21671, National
Bureau of Economic Research, Cambridge, MA.
Run and Long-
Allcott, Hunt, and Todd Rogers. 2014. “The Short- Run Effects of
Behavioral Interventions: Experimental Evidence from Energy Conservation.”
3037.
American Economic Review 104 (10): 3003–
Ashraf, Nava, Oriana Bandiera, and Scott S. Lee. 2014. “Awards Unbundled: Evidence
from a Natural Field Experiment.” Journal of Economic Behavior and Organization 100
(April): 44–63.
Banuri, Sheheryar, Stefan Dercon, and Varun Gauri. 2017. “Biased Policy Profession-
als.” World Bank Policy Research Working Paper WPS 8113.
Bardhan, Pranab. 2015. “State and Development: The Need for a Reappraisal of the
Current Literature.” Journal of Economic Literature 54 (3): 862–892.
Berlin, Isaiah. 1969. “Two Concepts of Liberty.” In Four Essays on Liberty, edited by
Isaiah Berlin, 118–172. Clarendon Press.
Beshears, John, James Choi, David Laibson, and Brigitte Madrian. 2010. “The Limi-
tations of Defaults.” NBER Retirement Research Center Paper NB 10- 02, National
Bureau of Economic Research, Cambridge, MA.
388 Comments by Robert Hockett and Varun Gauri
Habyarimana, James, and William Jack. 2011. “Heckle and Chide: Results of a Ran-
domized Road Safety Intervention in Kenya.” Journal of Public Economics 95 (11):
1438–1446.
Hallsworth, Michael, John A. List, Robert D. Metcalfe, and Ivo Vlaev. 2014. “The
Behavioralist as Tax Collector: Using Natural Field Experiments to Enhance Tax
Compliance.” NBER Working Paper 20007, National Bureau of Economic Research,
Cambridge, MA.
Hossain, Tanjin, and John A. List. 2012. “The Behavioralist Visits the Factory:
Increasing Productivity Using Simple Framing Manipulations.” Management Science
2167.
58 (12): 2151–
Lessig, Lawrence. 1995. “The Regulation of Social Meaning.” The University of Chi-
cago Law Review 62: 943.
Madrian, Brigitte C., and Dennis F. Shea. 2001. “The Power of Suggestion: Inertia in
401(k) Participation and Savings Behavior.” Quarterly Journal of Economics 116 (4):
1149–1187.
Mani, Anandi, Sendhil Mullainathan, Eldar Shafir, and Jiaying Zhao. 2013. “Poverty
980.
Impedes Cognitive Function.” Science 341 (6149): 976–
Mas, Alexandre, and Enrico Moretti. 2009. “Peers at Work.” American Economic
Review 99 (1): 112–145.
Sunstein, Cass R. 1996. “Social Norms and Social Roles.” Columbia Law Review 96:
903.
Sunstein, Cass R. 2012. “Storrs Lectures: Behavioral Economics and Paternalism.”
1899.
Yale Law Journal 122 (7): 1826–
Sunstein, Cass R. Forthcoming. Nudges.gov: Behavioral Economics and Regula-
tion (February 16, 2013). In Oxford Handbook of Behavioral Economics and the Law,
edited by Eyal Zamir and Doron Teichman. Available at SSRN: https://ssrn.com/
abstract=2220022 or http://dx.doi.org/10.2139/ssrn.2220022.
Sunstein, Cass R., and Reid Hastie. 2015. Wiser: Getting beyond Groupthink to Make
Groups Smarter. Boston: Harvard Business Review Press.
World Bank. 2014. World Development Report 2015: Mind, Society, and Behavior. Wash-
ington, DC: World Bank.
Yeager, David S., and Gregory M. Walton. 2011. “Social-Psychological Interventions
301.
in Education: They’re Not Magic.” Review of Educational Research 81 (2): 267–
9 Morality: Evolutionary Foundations
and Policy Implications
Ingela Alger and Jörgen W. Weibull
Act only according to that maxim whereby you can, at the same time, will that it
should become a universal law.
—Immanuel Kant, Groundwork of the Metaphysics of Morals, 1785
One general law, leading to the advancement of all organic beings, namely, mul-
tiply, vary, let the strongest live and the weakest die.
—Charles Darwin, On the Origin of Species, 1859
The academic discipline of economics has over many years provided policy
makers all over the world with a powerful toolbox. Conceptual, philosophi-
cal, and methodological disagreements are relatively rare, and the disci-
pline is not torn by fights between disparate schools of thought. Whether
this monolithic character of the field is a sign of strength or weakness is not
easy to say, but this methodological unity and power has, arguably, given
the discipline great influence on policy. The strong methodological core of
in the 1950s−1960s epitomized by general equilibrium theory
economics—
has enabled positive and normative
and later incorporating game theory—
analysis of a wide range of economic and social issues.
So what, more exactly, does this core consist of? In a nutshell, it has
who
two main components. The first is that it views economic agents—
as goal-
may be individuals, households, firms, or organizations— oriented,
This manuscript was prepared for the conference “The State of Economics, The State
of the World,” held at the World Bank in Washington, DC, on June 8– 9, 2016. The
authors thank Daniel Chen, Jean- François Laslier, Assar Lindbeck, Erik Mohlin, Paul
Seabright, Jean Tirole, Nicolas Treich, Yu Wen, and Peter Wikman for valuable com-
ments and suggestions.
390 Ingela Alger and Jörgen W. Weibull
as if they each had some goal function that they strive to maximize under
the constraints they face, given the information they have, and given their
beliefs about relevant aspects of the world they live in. The second compo-
nent is that interactions between these economic agents are taken to meet
certain consistency requirements, formalized as equilibria, that is, collec-
tions of action plans, one for each agent, such that no agent can unilaterally
improve the expected value of her goal function (usually profit or utility).
Both components can and have been contested. Individuals may not be
so systematic and consistent, and interactions may be chaotic and volatile.
founded and empirically accurate understand-
Having a theoretically well-
ing of human motivation is, arguably, in any case of utmost relevance for
analysis and policy recommendations.
Among the more noticeable new methodological developments in eco-
nomics is the emergence of behavioral and experimental economics, where
the first strand endows economic agents with richer motivations than in
regarding
traditional economics, usually in the form of prosocial or other-
preferences. The second strand tests such models, old and new, in con-
trolled laboratory experiments and in randomized field experiments. The
external validity of laboratory experiments can be questioned, and field
experiments may depend on local and historical factors with little general-
ity, but this development of the discipline of economics toward becoming
an empirically founded science appears to be essentially very healthy. It was
not long ago that economics was thought of as similar to meteorology and
astronomy: All it could do was to observe what is happening, without the
possibility of experimenting. Moving away from mere observation of data
that happen to come about to carefully designed controlled experiments
is reminiscent of how Galileo Galilei once lead the way from Aristotelean
scholastic discourse to modern science.
Behavioral and experimental economics no doubt will improve the pre-
dictive power and the usefulness of economics, but further improvements
could certainly be made if the underlying factors that shape human moti-
vation were better understood. The literature on the evolutionary foun-
dations of human motivation aims at providing such understanding by
asking: What preferences should humans be expected to have if these are
transmitted in society from generation to generation? If certain prosocial or
antisocial preferences, or moral values, give their carriers on average better
material outcomes than other preferences or values (all else being equal),
Morality: Foundations and Implications 391
then one would expect the former to spread in the population (be it by
biological or cultural mechanisms). Our aim in this chapter is to discuss a
recent theoretical result concerning such evolutionary preference selection
and to examine its implications for a range of social and economic issues.
Milton Friedman (1953, 22) claimed that “unless the behavior of busi-
nessmen in some way or other approximated behavior consistent with the
maximization of returns, it seems unlikely that they would remain in busi-
ness for long.” In a similar vein, one may claim that unless the behavior of
an individual is consistent with the maximization of own material payoffs,
other, materially more successful behaviors will take over in the interacting
population. Economists have shown that this claim is theoretically valid
when (1) the population at hand is very large, (2) interacting individuals
do not know each other’s goal functions, and (3) interactions are perfectly
random in the sense that each encounter is just as likely (Ok and Vega-
Redondo 2001; Dekel, Ely, and Yilankaya 2007).
In reality, however, populations are not always large, and interact-
ing individuals sometimes know or learn about each other’s preferences
(for instance, think of the great number of interactions that take place in
families or small communities). It has been shown that in such settings,
preferences or goal functions can usually serve as effective commitment
devices, and evolution will almost always favor goal functions that differ
from own material payoffs.1 Furthermore—
and this is what we will focus
encounters are only rarely perfectly random; geographic location,
on here—
language, culture, and religion often have an impact on the likelihood of
specific encounters. For example, business partners may know each other
from college, and neighbors may have chosen to live in the same place
because they share socioeconomic or cultural backgrounds or have simi-
lar location preferences and so forth. In such structured populations, some
encounters are more likely than others, even if the overall population is
large. In two recent theoretical studies (Alger and Weibull 2013, 2016), we
show that such assortative matching makes evolution favor individuals who
interested but who attach some value to “doing the right
are not purely self-
thing,” even though the population is large, and interacting individuals
1. Seminal articles on preference evolution, or indirect evolution, are Frank (1987)
and Güth and Yaari (1992). See also Banerjee and Weibull (1995); Heifetz, Shannon,
and Spiegel (2007), and Alger and Weibull (2010).
392 Ingela Alger and Jörgen W. Weibull
do not know one another’s preferences. This (for us, initially surprising)
finding suggests an evolutionary foundation for a psychologically plausible
form of morality, in line with Immanuel Kant’s categorical imperative.
In the next section, we describe this novel class of preferences and their
evolutionary foundations. In the second section, we discuss the impli-
studied social and economic
cations of such preferences for some much-
behavior and policy issues, including public goods provision and behaviors
that affect the environment. The third section discusses other social prefer-
ences and contrasts morality with altruism. The final section concludes.
Evolution and Kantian Morality
Imagine a population that has evolved for many generations in a stationary
environment and that in each generation, individuals engage in some social
subsistence
or economic interaction. For instance, in a population of self-
farmers, the interaction could be teamwork in the fields, the extraction of
resources from a commonly owned lake or piece of land, lending activities,
or the maintenance of institutions. In Alger and Weibull (2013, 2016), we
propose a theoretical model of precisely such populations. We formalize
the interaction by assuming that individuals are now and then randomly
matched into groups of arbitrary (but fixed and given) size n to interact
with each other in the group. (There are no interactions between groups
and hence no group selection takes place.) The interaction may involve
elements of cooperation and/or conflict, asymmetric information, repeti-
tion or interaction of arbitrary duration, possibility of helping, rewarding
and/or punishing others, and so forth. Essentially only two restrictions are
imposed on the interaction. First, the material payoff consequences for
a participant depend only on the participant’s own actions and on some
aggregate of other group members’ actions (not on who of them does what).
In game theory, such interactions are called aggregative games. Examples are
market competition where only competitors’ aggregate output or lowest price
matters, contributions to public goods where only the sum of others’ contri-
butions matters, some environmental externalities, and the like. Second, the
material payoff function is the same for all individuals.
We follow standard economic theory by assuming that each individual
acts so as to maximize some goal function. Different goal functions may
be present in the population where each goal function represents some
Morality: Foundations and Implications 393
preference. Depending on the preference distribution and the process by
which interaction groups are formed, individuals may end up in more or
less homogeneous groups. For a given material interaction, a given prefer-
ence distribution, and a given group formation process, the average mate-
rial payoff consequences for individuals with a particular goal function are
well determined in each equilibrium. In our evolutionary stability analysis,
we ask: What kind of goal function, if any, would be favored by natural
selection? Specifically, we determine which such functions are evolution-
arily stable in the sense that, if almost all individuals in the population
have such preferences, these individuals would materially outperform indi-
viduals with other preferences. Thus, the material payoffs are taken to be
the drivers of evolution.
This approach is a generalization of the work of Maynard Smith and
Price (1973), from the notion of an evolutionarily stable strategy (ESS), to
that of an evolutionarily stable goal function.2 A major challenge arises with
the preference distribution in
this generalization. In any population state—
there may be multiple equilibrium behaviors, and hence
the population—
several possible material payoff allocations. We define a goal function to be
evolutionarily stable against another goal function if in every population state
where the latter goal function is rare, individuals equipped with the former
goal function outperform those with the latter in terms of the resulting
material payoffs in all equilibria.3 Conversely, a goal function is evolution-
arily unstable if there exists another goal function such that, no matter how
small its population share, there is some equilibrium in which the latter
goal function materially outperforms the former. In both definitions, the
test scenario is to let in a small population share of “mutants,” who may
be migrants or carriers of spontaneously and randomly arising alternative
goal functions, into the population of incumbents or residents. We impose
minimal constraints on the nature of potential goal functions. They are
not required to take any particular parametric form or even to depend on
the material payoffs. Hence, individuals may be selfish, altruistic, spiteful,
2. In our approach, it is thus as if “mother nature” delegates to individuals to choose
their actions, and instead equips them with goal functions that will guide their
choice of action.
3. By “equilibrium” we mean Bayesian Nash equilibrium under incomplete
information.
394 Ingela Alger and Jörgen W. Weibull
minded, inequity averse, environmentalists, moralists, and the
fairness-
like. Our only assumption is that each individual’s goal function is continu-
ous in all group members’ courses of action.
A second key feature of our approach is that it allows the random match-
ing to be assortative. Geographic, cultural, linguistic, and socioeconomic
distances impose (literal or metaphoric) transportation costs, which imply
that (1) individuals tend to interact more with individuals in their (geo-
graphic, cultural, linguistic, or socioeconomic) vicinity,4 and (2) cultural or
genetic transmission of types (say, behavior patterns, preferences, or moral
values) from one generation to the next also has a tendency to take place in
the vicinity of where the type originated.5 Taken together, these two ten-
dencies imply that individuals who interact with each other are likely to be
of the same type. We formalize such potential assortativity in the random
matching process in terms of a vector we call the assortativity profile. This
vector consists of probabilities for the events that none, some, or all indi-
viduals in a vanishingly rare mutant’s group also are mutants.6
Our analysis delivers two main results. First, although we impose vir-
tually no restrictions on permissible utility functions, evolution favors a
particular class of utility functions, which we call Homo moralis. Individu-
als with preferences in this class attach some weight to their own material
payoff but also to what can be interpreted as a probabilistically general-
ized version of Kantian morality. In his Grundlegung zur Metaphysik der
Sitten, Immanuel Kant ([1785] 2002, 37) wrote: “Act only according to that
4. Homophily has been documented by sociologists (e.g., McPherson, Smith- Lovin,
and Cook 2001; Ruef, Aldrich, and, Carter 2003) and economists (e.g., Currarini,
Jackson, and Pin 2009, 2010).
5. In biology, the concept of assortativity is known as relatedness, and the propensity
to interact with individuals locally is nicely captured in the infinite island model,
originally due to Wright (1931). Hamilton (1964) provided a first formalization of
what is now known as Hamilton’s rule: Evolution will select for behaviors whereby
the external effects on others are internalized at a rate provided by the relatedness
(see also Dawkins (1976), for a popular account of this idea, as well as Rousset (2004),
for a comprehensive treatment). In an article on the evolution of behaviors in inter-
actions among siblings, Bergstrom (1995) was probably the first to bring Hamilton’s
rule into the economics literature.
6. This concept generalizes Bergstrom’s (2003) definition of the index of assortativity
for pairwise encounters. See also Bergstrom (2012) and Alger and Weibull (2013) for
further discussions of assortativity under pairwise matchings.
Morality: Foundations and Implications 395
maxim whereby you can, at the same time, will that it should become a
universal law.” Similarly, paraphrasing Kant, Homo moralis attaches some
weight to the goal of acting according to that maxim whereby you can, at
the same time, will that it should become a universal law, even if followed
only probabilistically by others. More precisely, a Homo moralis individual
in a group of arbitrary size n maximizes a weighted average of equally many
terms, indexed j = 0, … , n −
1, where each term is the material payoff that
she would obtain if, hypothetically, she could replace the strategies of j
other individuals in the group by her strategy. We call the vector of these
probability weights the individual’s morality profile.
The class of Homo moralis preferences has two extremes: Homo oeco-
nomicus, who considers only her own material payoff,7 and Homo kantiensis,
who considers only the material payoff that she would obtain if all others
were to act like she does. In between these two extremes is a whole range
of Homo moralis preferences with different morality profiles, whereby an
individual examines what would happen if some but not all the others were
to act like him-or herself. Homo moralis partly evaluates her own actions in
this probabilistic Kantian sense. In other words, she is to some extent con-
cerned with the morality of her own acting, irrespective of what others do.
She asks herself, before taking her action, what action she would prefer if,
hypothetically, others would also probabilistically choose the same action
in her situation.
Our first main result is that Homo moralis with a morality profile identi-
cal to the assortativity profile is evolutionarily stable. The intuition behind
this result is not based on group selection, an old argument (appearing
already in Charles Darwin’s writings; see also Alexander (1987)) that essen-
tially says that evolution will lead to behaviors that enhance the survival of
the group. Quite on the contrary; the intuition is that natural selection will
lead to utility functions that preempt entry into the population in the sense
that the best a potential rare mutant can do, if striving for material payoff,
is to mimic the residents.
7. Note that we define Homo oeconomicus as individuals who always seek to maxi-
mize their own material payoff. Some writers define Homo oeconomicus, or “economic
man,” more generally as an individual who always acts in accordance with some goal
interest or not. All agents in the present study are
function, whether this be pure self-
varieties of Homo oeconomicus in this broad sense.
396 Ingela Alger and Jörgen W. Weibull
Our second main result is that any preferences that are behaviorally
distinct from those of Homo moralis with the stable morality profile are
evolutionarily unstable. Hence, although we made no parametric or struc-
tural assumption about utility functions, it appears that natural selec-
as represented by evolutionary stability in our abstract and simplified
tion—
favors the utility function of Homo moralis. In particular, our
framework—
results imply that Homo oeconomicus—pure material self-interest—is evolu-
tionarily unstable under any random matching process with positive assor-
tativity. Rare mutants may indeed garner a higher material payoff than
Homo oeconomicus, on average, by behaving somewhat prosocially, because
when there is positive assortativity, the benefits of this prosocial behavior
are sometimes bestowed on other mutants, whereas the residents almost
never benefit from it.
Homo moralis is easily defined for pairwise interactions, n = 2. Let π(x, y)
denote the material payoff to an individual who plays strategy x when the
opponent plays strategy y. Then the utility function of Homo moralis is
Uκ(x, y) = (1 − κ) · π (x, y) + κ · π (x, x), (1)
where 0 ≤ κ ≤ 1 is the individual’s degree of morality. The two extreme degrees
of morality represent Homo oeconomicus (κ = 0) and Homo kantiensis (κ = 1),
respectively, and intermediate degrees of morality correspond to individu-
als who attach some weight to their own material payoff, π(x, y), and some
weight to “the right thing to do if everyone were to choose the same behav-
ior,” π(x, x).
For n > 2, the precise definition of Homo moralis is fairly involved,8 but it
is analytically straightforward in the special case where the random match-
ing is such that the types of any other two group members are statisti-
cally independent, given the member’s own type. The morality profile is
then a binomial distribution, and the utility function of a Homo moralis
individual i is the expected value of i’s material payoff if, hypothetically,
other members of the group would randomly and statistically indepen-
dently switch to use i’s strategy with probability κ, which is then i’s degree
of morality. At one end of the interval of such Homo moralis, κ = 0, we
find Homo oeconomicus; at the other end, κ =
1, we find Homo kantiensis.
8. The general definition of Homo moralis is given in Alger and Weibull (2016).
Morality: Foundations and Implications 397
Moreover, in large groups, the share of mutants in a mutant’s group is,
Laplace theorem, approximately normally distributed
by the de Moivre–
with mean value κ and variance κ(1 − κ)/(n − 1). Hence, the share of other
mutants is then almost deterministic and is equal to κ. A Homo moralis
with degree of morality κ then acts (approximately) as if she hypotheti-
cally assumed that her behavior were to become, if not a “universal law,”,
then a “random law” applying to a randomly sampled share of size κ out
of her group’s other members.9
It is worth noting that the utility function of Homo moralis differs
sharply from any utility function that only depends on the payoffs to all
participants, such as altruism, inequity aversion, or a concern for social
efficiency. We illustrate this by way of a simple example at the end of the
third section.
Morality and ethics in connection with economics have been dis-
cussed at great length by many economists and philosophers, including
Smith ([1759] 1976), Edgeworth (1881), Rawls (1971), Arrow (1973), Sen
(1977), and Harsanyi (1980), to mention a few. But to the best of our
knowledge, Homo moralis preferences have not been previously studied,
or even known, with one exception. Bergstrom (1995) shows that evo-
lutionary stability of strategies in interactions between siblings induces
Kantian,” which corresponds to κ = 1/2 in
behavior that he calls “semi-
our equation (1).10
Kantian Morality and Economics
Economists’ policy advice traditionally relies on models in which individu-
als have Homo oeconomicus preferences. What if economists’ models instead
were populated by the more general Homo moralis? In this chapter, we will
merely scratch the surface by studying only a few examples.
9. This claim is not fully general and deserves further analysis, because even small
perturbations of continuous (utility) functions may lead to “jumps” in behavior.
10. Bergstrom thus differs from us in studying stability of strategies rather than of
utility functions. However, in Alger and Weibull (2013, corollary 5), we establish a
link between these approaches by showing that Homo moralis equilibrium strategies
are stable under strategy evolution. For a discussion of several ethical principles in
relation to strategy evolution, see Bergstrom (2009).
398 Ingela Alger and Jörgen W. Weibull
Trust
There is variation across countries in the extent to which people are trust-
ing, and trust is correlated with economic growth (Algan and Cahuc
2010).11 In economics, the so-
called trust game has been used extensively
in controlled laboratory experiments as a way to measure trust and trust-
worthiness in different countries and cultures. This literature was pioneered
by Berg, Dickhaut, and McCabe (1995) and has received a lot of attention
among behavioral economists and experimentalists. The trust game is suc-
cinctly described by Cesarini et al. (2008, 3721):
Many mutually beneficial transactions involve an element of interpersonal trust
and may fail to materialize in the absence of an expectation that trust will be
reciprocated. The prevalence of trust in a society has therefore been assigned pri-
macy in a number of domains, for instance empirical and theoretical studies of
economic growth. In recent years, the trust game has emerged as a favorite instru-
ment to elicit an individual’s interpersonal trust and willingness to reciprocate
trust. More generally, the game has been widely used to study cooperative behav-
ior. In a trust game, an individual (the investor) decides how much money out of
an initial endowment to send to another subject (the trustee). The sent amount is
then multiplied by some factor, usually three, and the trustee decides how much
of the money received to send back to the investor. The standard game- theoretic
prediction for a single anonymous interaction between two purely self- interested
individuals is for the investor to send nothing, rationally anticipating that the
trustee will not reciprocate. Yet, experiments consistently show that cooperation
flourishes in the trust game; the average investor sends a significant share of her
endowment, and most trustees reciprocate.
What will Homo moralis do in such an interaction? Consider a situation
in which two ex ante identical individuals are randomly paired. With equal
chance, one of them is offered an endowment and an investment opportu-
nity as described above. The other individual then has to act in the role of
11. A situation where trust is key is that of informal personal lending. In many
developing countries, large fractions of the populations are still shut out from formal
credit markets; see, for example, Kendall, Mylenko, and Ponce (2010). Then informal
lending, in the form of not legally binding loans between individuals, can sometimes
be enforced by the threat of future nonrenewal of lending (Ghosh and Ray 2016),
social disapproval, or both. Evidence from laboratory experiments suggests that such
informal lending may in fact even take place in one- shot interactions (Charness and
Dufwenberg 2006). The trust game we analyze here can be interpreted as informal
lending.
Morality: Foundations and Implications 399
the trustee. A strategy for an individual in such a symmetric interaction then
has two components. First, if given the endowment, what share s ∈ [0, 1]
of it is to be invested? Second, if not given the investment opportunity,
what “payback rule” p ∈ [0, 1] is to be used? Here, such a payback rule pre-
scribes for any invested share t ∈ [0, 1] chosen by the other party what share
p of the gross return to pay back. Let u(c) be an individual’s hedonic utility
from own consumption c, and take this to represent the material payoff in
our evolutionary framework. In the standard version of the trust game, the
material payoff from using a strategy x = (s, p) when the other individual
uses strategy y = (t, q) is then
1 1
π ( x, y ) = u (1 − s + 3sq ) + u (3t − 3tp ). (2)
2 2
In an interaction between two Homo oeconomicus, no party is trustworthy;
they will choose p = q = 0 for all s, t > 0. Thus, if each party knows the other’s
type, no investment is made in equilibrium (t = s = 0). The resulting expected
material payoff to each party is u(1)/2, the probability of being given the
initial endowment times the utility from keeping it. If instead both parties
were Homo kantiensis, then they would each invest all the money if given
the opportunity (t = s = 1) and return half of the gross return (i.e., use pay-
back rules p and q such that p = q = 0.5). The resulting expected material pay-
off to each party is then u(1.5), much higher than what Homo oeconomicus
obtains.
Full morality is not necessary to induce full investment, however. For
a pair of equally moral Homo moralis, full investment (t = s =
1) obtains in
equilibrium for any sufficiently high degree of morality, although as soon
as morality is less than full (κ < 1), the trustee pays back less than half the
gross returns from investment, in which case the trustee ends up being
better off than the investor. As the degree of morality κ falls, the amount
paid back decreases, and it eventually falls short of the amount originally
invested, in which case the investor makes a material loss; nonetheless,
morality makes the investor accept this loss and invest anyway, up to some
point.12 Indeed, for sufficiently low degrees of morality, the investor invests
12. To see this, note that the derivative of Uκ (x, y) with respect to s, where x = (s, p)
and y = (t, q), and evalutated when t = s = 1, is positive even for p < 1/3 for κ < 1 large
enough.
400 Ingela Alger and Jörgen W. Weibull
less than his full endowment, and eventually, when morality drops below a
certain level, he invests nothing.
Public Goods
Many situations that are important for economic growth may be represented
as situations in which people can make voluntary contributions to a public
good, including the generation and dissemination of knowledge, and institu-
tion building. We examine the behavior of individuals in a community of n
members, each of whom is in a position to make a voluntary contribution to
a public good (the contribution may be monetary or in kind). A standard con-
cern in economics is that free riding is enhanced as groups become larger, so
our aim here is to analyze how group size affects the behavior of Homo moralis.
Suppose, then, that i obtains material payoff
( )
π ( xi , y ) = B xi + ∑ j ≠i y j − C( xi ) (3)
if she makes the contribution xi and the sum of the contributions from the
other community members is Σyj. Here B is a production function for the
public good; and C a cost function for a contributing individual, represent-
ing forgone private consumption, income, or leisure. We take the marginal
cost of making a contribution to be increasing, and the marginal benefit of
the aggregate contribution to be decreasing.
Consider first the socially optimal individual contribution, x*. With
a conventional production function of the power form B(X) = Xa, where
0 < a < order condition for the sum of all members’
1, the necessary first-
material payoffs to be maximized,
nB′(nx*) = C′(x*), (5)
implies that the socially optimal individual contribution x* is increasing in
order condi-
n. By contrast, in a community of Homo oeconomicus, the first-
ˆ
tion for the unique Nash equilibrium contribution, x0 , is
ˆ0 ) = C ′( x
B ′(nx ˆ0 ), (5)
which implies that in communities with more members, each individual
the tendency for people to
contributes less. As a consequence, free riding—
provide public goods—
under- is exacerbated when group size increases. The
intuition is that if all contributions were to remain unchanged, then the
marginal benefit from each contribution would fall. Thus, each individual
will have a weaker incentive to contribute.
Morality: Foundations and Implications 401
x
0.7
κ=1
0.6
κ=0.75
0.5 κ=0.5
κ=0.25
0.4
κ=0
0.3
n
2 3 4 5 6 7
Figure 9.1
goods game for different
The unique Nash equilibrium contribution in the public-
degrees of morality
Suppose now instead that everyone in the community is a Homo moralis
with the same degree of morality κ ∈ [0, 1]. Then their unique individual
ˆκ , can be shown to satisfy
equilibrium contribution, x
[1 + (n − 1)κ ] ⋅ B ′(nx
ˆκ ) = C ′( x
ˆκ ), (6)
For any positive degree of morality, group size has two counteracting
effects on the individual contribution. The negative effect is, as before,
due to the decreasing marginal productivity. The positive effect is that
in larger groups, each individual’s contribution benefits a larger num-
ber of individuals. The “right thing to do,” as the group increases, is
thus to increase one’s contribution. The positive effect may outweigh
the negative.
To see this, consider again the conventional production function used
above, and note that for purely Kantian individuals (κ =
1), the individ-
ual contribution always increases with n. For intermediate values of κ, the
individual contribution decreases with n when small but increases with n
when large. See figure 9.1, which shows the equilibrium contribution of
Homo moralis with degree of morality κ as a function of community size
n, with higher curves for higher degrees of morality (when B( X ) = X and
C(x) = x2).
402 Ingela Alger and Jörgen W. Weibull
These predictions may potentially help explain observations made in
laboratory experiments, in which group size sometimes has a positive effect
and sometimes a negative effect on individual contributions (see Nosenzo,
Quercia, and Sefton (2015) for a review).
Does the extent of free riding increase or decrease as group size increases?
In the parametric specification used in figure 9.1, the individual contribu-
best contribution is
tion relative to the first-
23
ˆκ ⎛
x 1−κ ⎞
= ⎜κ + ⎟ , (7)
x* ⎝ n ⎠
a ratio that decreases as group size n increases (for any given degree of moral-
ity κ < 1).13 A smaller ratio indicates more free riding, so this equation shows
that as morality κ increases, the effect of group size n on the extent of free
riding declines.14 Moreover, the extent of free riding is bounded from below;
ˆκ x* exceeds κ2/3 for all group sizes n. Hence, com-
as seen in (7), the ratio x
pared to the outcome under Homo oeconomicus, an important policy implica-
tion is that, when κ is positive, the contributions from Homo moralis decline
less with group size and remain positive even in infinitely large groups.
Environmental Economics
According to World Bank president Jim Yong Kim, “If we don’t confront cli-
mate change, we won’t end poverty.”15 Some instruments have been pro-
posed to help mitigate climate change, such as a carbon tax, regulation of
production technologies, subsidies to public transportation, and support for
R&D concerning environmentally friendly technologies for different forms
of green energy. Determining the “right” carbon tax requires knowing how
it will affect behavior and welfare. Here we briefly analyze the behavior of
Homo oeconomicus and more generally, Homo moralis, in a standard model
of consumption that has an external effect on the environment (Musgrave
1959, Arrow 1970). In this model, the group is taken to be so large that each
individual’s impact on the group’s environment is negligible.
More specifically, there is a continuum of consumers, indexed i ∈
I = [0, 1], and there are two consumption goods, goods 1 and 2, where good 1
13. Formally, d ( x ˆκ x * ) dn < 0 when 0 < κ ≤ 1.
14. Formally, d 2 ( x ˆκ x * ) ( dndk ) > 0 .
15. See http://www.worldbank.org/en/news/feature/2014/03/03/climate-change
-affects-poorest-developing-countries.
Morality: Foundations and Implications 403
is environmentally neutral (that is, its consumption has no effect on the envi-
ronment) and good 2 is environmentally harmful. Aggregate consumption of
these goods are
X1 =
∫ x (i) d µ
I
1 and X2 =
∫ x (i) d µ
I
2
where x(i) = (x1(i), x2(i)) is the consumption bundle of individual i, and μ is
a density on I. Because all consumers are infinitesimally small, aggregate
consumption is unaffected by any individual’s personal consumption.
We take the material payoff to each individual i to be that individu-
al’s hedonic utility from own consumption, x(i), and from the quality of
the environment, which in turn depends on aggregate consumption, X2,
of the environmentally harmful good. We write u(x1(i), x2(i), X2) for this
hedonic utility and assume that it is increasing in consumption of each
good and decreasing in aggregate consumption of the environmentally
harmful good. Using good 1 as the numeraire, writing p for the price of
good 2, and assuming that all individuals have the same income, a socially
efficient consumption bundle, x*, the same for all individuals i, satisfies
* , X2
* , x2
u2 ( x1 *) *)
u ( x * , x * , X2
= p− 3 1 2 ,
* , X2
* , x2
u1 ( x1 *) * , X2
* , x2
u1 ( x1 * ) (8)
where subscripts on the personal utility function denote partial derivatives.
The marginal rate of substitution between the environmentally harmful
and environmentally neutral goods should thus equal the relative price of
the harmful good net of the marginal rate of substitution between the util-
ity from the quality of the environment and the neutral good. In other
words, social efficiency requires that, at given prices, consumers consume
less of a good the more harmful it is to the environment.
By contrast, in a population consisting entirely of Homo oeconomicus, an
(interior) equilibrium allocation in which everybody consumes the same
bundle x0 necessarily satisfies the first-
order condition
u2 ( x1
0
, x2
0
, X2
0
)
= p. (9)
u1 ( x1 , x2 , X2 )
0 0 0
Under decreasing marginal utility of consumption, this means that Homo
oeconomicus, not surprisingly, consumes more of the environmentally
harmful good than required by socially efficiency.
As observed above, for interactions in infinitely large groups, the util-
ity function of an individual Homo moralis with degree of morality κ ∈ [0,
404 Ingela Alger and Jörgen W. Weibull
1] is the material payoff that would obtain if a share κ of the group would
behave in the same way as the individual himself. In the present context,
if an individual consumes the bundle x = (x1,x2) and all the others consume
some bundle y = (y1,y2), then the utility to a Homo moralis with degree of
morality κ would be
Uκ(x, y) = u (x1, x2, (1 − κ) y2 + κx2), (10)
where, in this expression, we have normalized the total mass of individu-
als in the group (which could be a village, region, country, continent, or
the whole world) to unity. In a group consisting entirely of Homo moralis
with the same degree of morality κ, an (interior) equilibrium allocation,
everybody consumes the same bundle xκ , and this satisfies the first-
order
condition
κ κ κ
u2 ( x1 , x2 , x2 ) u ( xκ , x2κ κ
, x2 )
κ κ κ
= p−k⋅ 3 1 κ κ κ
. (11)
u1 ( x1 , x2 , x2 ) u1 ( x1 , x2 , x2 )
Compared to Homo oeconomicus, for any positive degree of morality κ, each
individual refrains somewhat from consuming the environmentally harm-
knowing that she is negligible—
ful good, although each individual— is fully
aware that her own consumption has no effect on the overall quality of
the environment! Hence, if people are in fact somewhat moral, then policy
advice based on models inhabited by Homo oeconomicus may exaggerate the
need for pecuniary incentives, such as carbon taxes. If people are more like
Homo moralis with some positive degree of morality, then, in addition to
some carbon taxes, it may be effective to provide individuals with informa-
tion about how aggregate consumption (and production) creates carbon
dioxide and what we know about how this affects the climate.16 By contrast,
such information in this stylized example would have no effect at all on the
behavior of Homo oeconomicus.17
16. Note that equations (8) and (9) are the special cases of (10) when κ = 0 (Homo
oeconomicus) and κ = 1 (Homo kantiensis). Laffont (1975) considers these two extreme
interested individuals (our Homo oeconomicus) and “Kantian individuals”
cases of self-
(our Homo kantiensis).
17. Note further that if good 2 does not cause any externality (u3 = 0), then Homo
moralis would behave precisely as the classical Homo oeconomicus; equation (10)
would boil down to equation (9). For such goods, there is no “right thing to do,” and
hence, morality has no bite.
Morality: Foundations and Implications 405
Voting
Another class of situations in which Homo moralis may make a difference
making by voting. By and large, countries with more
is collective decision-
developed economies tend to have more democratic political systems (see,
e.g., Persson and Tabellini (2006) and Acemoglu et al. (2014)). For democ-
racy to work, it is important that citizens participate in elections, committee
work, and related activities; it is still much debated in economics and politi-
cal science why and how people vote. As has been pointed out by econo-
mists, high participation rates in large elections appear incompatible with
rational Homo oeconomicus behavior. The reason being that the act of voting
usually has some personal cost, say, lost income or leisure, and this cost eas-
ily outweighs the expected benefit to the individual of participating in the
election, because the probability of being pivotal is virtually nil. This is the
well-known voters’ paradox. Despite this, the turnout in general and local
elections in many countries is often impressive. So what then motivates peo-
ple to participate in elections? Can Homo moralis provide an explanation?
A closely related and arguably equally important issue is participation
and voting in committees, such as parliamentary bodies, company boards,
Smith and Banks
court juries, and central bank boards. As shown by Austen-
(1996), when committee members have private information and are Homo
oeconomicus, then voting may fail to aggregate information efficiently, even
when the members have the same preferences. This observation challenges
called Condorcet jury theorem (Condorcet 1785), which states that
the so-
democracy in the form of majority rule in such situations is a great institu-
tion, because it implies that the right decision is almost always taken if the
electorate is large enough. How would Homo moralis vote in such committees?
Other Social Preferences
Theoretical work on the evolutionary foundations of human motivation
the
provides insights about potential ultimate causes of human behavior—
forces in the environment that have shaped our preferences, not only for
the foods that contain the nutrition that we need to survive but also for
behaviors in social interactions. This line of research is complementary
to behavioral economics, the branch of economics that investigates the
interest. In the
explanatory power of richer motivations than mere self-
language of evolutionary biology, the focus in behavioral economics is
406 Ingela Alger and Jörgen W. Weibull
the neurological,
on the proximate causes of observed human behaviors—
hormonal, and psychological mechanisms and triggers that induce us to
behave in certain ways. Here we briefly discuss how Homo moralis prefer-
ences compare with those considered in this literature, which is inspired by
research in psychology and sociology.
In the 1970s and 1980s, altruistic preferences were proposed to explain
intra-family transfers, transfers to the poor, and contributions to public goods
(Becker 1974, 1976; Andreoni 1988; Lindbeck and Weibull 1988). However,
altruism turned out to be insufficient to explain the data, and “warm glow”
was then proposed to enhance the understanding of voluntary contribu-
tions to public goods (Andreoni 1990). In the 1990s, inequity aversion, or
a preference for fairness, was introduced by Fehr and Schmidt (1999) as an
explanation for why people have a tendency to turn down low offers in the
ultimatum bargaining game (Güth, Schmittberger, and Schwarze 1982). Still
other forms of human motivation that have been proposed, and sometimes
tested, include conformity (Bernheim 1994), conditional altruism (Levine
1998), identity (Akerlof and Kranton 2000), and honesty and truth telling
(Alger and Ma 2003; Alger and Renault 2006; Demichelis and Weibull 2008).
Although conceptually very different from Homo moralis, these prefer-
ences would be compatible with evolutionary stability if they gave rise to
the same equilibrium behaviors as those of Homo moralis.18 For what class
of material payoff functions such behavioral equivalence obtains remains
to be analyzed. Here we limit ourselves to pointing out that Homo moralis
preferences sometimes give rise to radically different behaviors compared to
preferences that may appear to be similar. For example, consider altruistic
preferences. An altruistic individual’s preferences are usually represented as
aterial
a utility function that attaches unit weight to the individual’s own m
payoff and a positive weight, less than 1, to other individuals’ material
payoffs. An altruist hence internalizes some of the external effects of her
behavior on others. Let the latter weight be denoted α, the individual’s
degree of altruism toward the other party.19 For some material payoff func-
18. However, the preferences of Homo moralis are the only ones that are evolutionarily
stable in the whole class of interactions analyzed in Alger and Weibull (2013, 2016).
19. For n = 2, an altruist’s utility is uα(x, y) = π(x, y) + απ(x, y). Note that this function
can also be interpreted as the individual having a concern for efficiency, because it is
α
a monotone transformation of vα ( x , y ) = π ( x , y ) + [π ( x , y ) + π ( y , x )].
1−α
Morality: Foundations and Implications 407
tions, an altruist with degree of altruism α behaves exactly like Homo moralis
with a degree of morality κ = α (see Alger and Weibull 2013). Hence, in some
interactions, one cannot discriminate between moralism and altruism as
explanations for observed behavior. However, the two classes of preferences
are conceptually quite distinct and induce radically different behaviors in
some interactions. This difference is particularly striking in interactions
with many participants and in coordination problems among few or many
participants.
To illustrate the first case, consider again the environmental economics
and the public goods examples. In the environmental example, morality
induced consumers to reduce their consumption of the harmful good, even
though the effect of each individual’s consumption was negligible. In the
public goods example, as the number of participants tends to infinity, the
individual contribution to the public good tends to a positive amount for
any positive degree of morality. By contrast, Andreoni (1988) has shown
that in a population of altruists, the proportion of individuals who make
positive donations shrinks to zero as the number of individuals grows infi-
nitely large, because each individual donation then has a negligible effect
on the total value of the public good. There is thus a sharp distinction
between morality and altruism when groups are large. Even if an individual
is highly altruistic and cares about the consequences of her behavior for
others, she will behave very much like Homo oeconomicus if her impact is
marginal. By contrast, Homo moralis cares directly about his own behavior,
beyond the effects that this behavior has on his own material payoff, and
this consideration for “the right thing to do” makes him behave differently
from both selfish and altruistic individuals in these situations.
This observation may have important implications for other policy
issues as well, such as tax compliance. It has been noted by some econo-
mists (see Sandmo 2005), that less tax evasion appears to occur in certain
countries than would be compatible with Homo oeconomicus’s behavior.
The risk of being caught is often small and the penalties mild, so maxi-
mization of expected personal utility would suggest rampant tax evasion.
So why do people in those countries, and perhaps many in other coun-
tries, not evade taxes more? Because the marginal effect of any change
in an individual’s tax payment is, with few exceptions, negligible, proso-
cial preferences such as altruism or inequity aversion may fail to explain
why individuals evade taxes. However, as suggested by the analysis above,
408 Ingela Alger and Jörgen W. Weibull
Homo moralis may supply an explanation, because a Homo moralis may, to
a certain extent, prefer to pay taxes, since she cares about the moral qual-
ity of her actions.
Let us now turn to the second situation in which Homo moralis prefer-
ences give rise to radically different behaviors compared to altruism, namely,
coordination problems. Consider an example from Alger and Weibull
(2013), a simple 2 × 2 coordination game in terms of material payoffs:
A B
A 2,2 0,0 . (12)
B 0,0 1,1
When individuals pair up to play this game, two alternative potential
societal “conventions” are available: Either both parties take action A, or
both parties take action B. Clearly, the first convention is Pareto superior
to the second. However, under each convention, Homo oeconomicus has no
incentive to unilaterally deviate. Granted that a sufficiently large popula-
tion share act according to the going convention, an individual deviator
would lose material payoff and, in addition, inflict a payoff loss on the
unfortunate opponent.20 Therefore, an altruist would also stick to the going
convention, even if this happened to be the socially inferior convention to
always take action B. But not so a Homo moralis of high enough degree of
morality. For suppose a Homo kantiensis were to visit a country where (by
and large) every citizen takes action B in every encounter, and suppose that
the visitor is indistinguishable from a citizen. Then Homo kantiensis would
take action A in each encounter, because this would be “the right thing to
do” if upheld as a universal law of conduct.21 This moralistic visitor will
earn material payoff zero in each encounter, and so will the unfortunate
citizens who meet him. The citizens would very much wish that the visitor
instead had been a Homo oeconomicus or an altruist.
20. These are strict Nash equilibria in terms of material payoffs. The game also has a
mixed equilibrium, in which each individual plays A with probability 1/3. However,
this equilibrium is unstable in all plausible population dynamics. See Young (1993)
and Myerson and Weibull (2015) for formal models of stable conventions in large
populations.
21. Indeed, to take action A is optimal for all Homo moralis individuals with degree
of morality κ ≥ 1/3.
Morality: Foundations and Implications 409
odels
A final point before concluding. Some researchers have developed m
in which individuals care about norms, have a concern for their image (in
the eyes of others and perhaps also in their own eyes), or a desire to avoid
social stigma (Lindbeck, Nyberg, and Weibull 1999; Brekke, Kverndokk, and
Nyborg 2003; Bénabou and Tirole 2006; Ellingsen and Johannesson 2008;
Huck, Kübler, and Weibull 2012). In these models, individuals are assumed
to have a baseline intrinsic wish to “behave well” and in addition a wish
to be viewed favorably by others, image concerns that may strengthen the
wish to behave well (Bénabou, Falk, and Tirole 2018). Evidently, we humans
are very complex creatures, and our behavior is most likely driven by many
motives (what biologists would call “proximate causes” for our actions).
Biologists distinguish such proximate causes from ultimate causes, by which
is meant the reasons we exist in the evolutionary race. Our derivation of
Homo moralis was based entirely on such ultimate causes. A closer examina-
tion of relations between proximate and ultimate causes in human motiva-
tion is an avenue for future research. Eventually, evolutionary theory may
endedness of behavioral economics by providing test-
help close the open-
able predictions regarding which preferences are more likely to be sustained.
Conclusion
In this chapter, we have discussed the evolutionary foundations for human
motivation, how evolution favors the class of Homo moralis preferences,
and the implications for economics and policy of such preferences com-
pared to other preferences. We have presented the following main points:
Economics possesses powerful analytical tools that enable positive and
1.
normative analyses of a wide range of social and economic phenomena.
These tools should not be abandoned but instead brought to more gen-
eral use.
The conventional assumption among economists, since the days of
2.
Adam Smith’s ([1776] 1976) Wealth of Nations, is that economic agents
interested and focused on their own consumption. Yet
are purely self-
behavioral and experimental economics, insights from the other social
and behavioral sciences, everyday observation, and introspection sug-
gest that human motivation is much more complex, sometimes system-
interest.
atically deviating from narrow self-
410 Ingela Alger and Jörgen W. Weibull
First principles in evolutionary biology, formalized in terms of evolution-
3.
ary stability along the lines of Maynard Smith and Price (1973) suggest
that, in our simple model framework, evolution favors human motivation
in the form of Homo moralis, a generalization of Homo oeconomicus that
interest.
allows for varying degrees of morality along with self-
Applying the powerful analytical tools of economics to Homo moralis
4.
results in new predictions and policy recommendations. In particular,
because Homo moralis is not only motivated by her material gains and
losses, policy based on Homo oeconomicus may lead to exaggerated use
of pecuniary incentives, such as distortionary taxes. If people do have a
natural inclination for moral concerns, it may be more effective to pro-
vide the public with information about the consequences of our actions,
for ourselves and others.
Our results being purely theoretical, empirical and experimental work will
be necessary to determine the empirical validity of Homo moralis. To this
end, further theoretical analysis is also needed: Even though we have here
examined the behavior of Homo moralis in some common situations, we
have only scratched the surface. Moreover, many fundamental questions
have not been addressed at all. In particular, one fundamental issue that
we have not (yet) addressed is welfare. For economic and social policy, this
is a most important and philosophically nontrivial issue, especially when
individuals have social preferences. If individuals have Homo moralis pref-
erences, perhaps idiosyncratic degrees of morality, should welfare then be
defined in terms of the material payoffs or in terms of individuals’ utility
functions?
This philosophically and methodologically difficult issue may be related
to that addressed by John Harsanyi in two wonderful essays that deal with
game theory, utilitarianism and ethics (Harsanyi (1980, 1992). In these
essays, he advocates what he calls “rule utilitarianism,” an approach we
find also appealing for Homo moralis. Harsanyi distinguishes between an
individual’s “personal preferences” and his or her “moral preferences.” He
advocates that, when defining welfare in a society, one should only consider
personal preferences. When individuals’ preferences can be represented by
an additive utility function, where one term can be taken to represent “per-
sonal utility,” Harsanyi argues that welfare should be defined as the sum of
all individuals’ expected personal utilities, behind the veil of ignorance as to
Morality: Foundations and Implications 411
what societal position each individual will end up in. This appears to be in
line with Homo moralis. If we take the material payoff function to represent
personal utility, then welfare in a society consisting of Homo moralis indi-
viduals (each with his or her degree of morality) should be defined simply
as the sum of their expected material payoffs, just as in ordinary utilitarian
welfare theory.
To wit, suppose a parent has one selfish and one altruistic child, and has
a cake to divide between them. Suppose also that both children have the
same hedonic utility from consumption, and that this is increasing in the
amount consumed, with decreasing marginal utility.22 Should the parent
give a bigger slice to the selfish child, thus maximizing the sum of their
altruistic and selfish utilities, or should the parent give them equally large
slices, thus maximizing the sum of only their hedonic utilities? The sec-
ond alternative undoubtedly seems more appealing. The same could be said
with one selfish and one spiteful child; taking into account both children’s
total utility, a bigger slice should be given to the spiteful child, but equal
division is, arguably, more reasonable. By contrast, if one child is selfish,
and the other instead is inequity averse or a Homo moralis (with any degree
of morality), it makes no difference if the parent considers the children’s
total or hedonic (personal) utilities; in every case, their joint welfare is max-
imized by equal division. Further study of the welfare economics of Homo
moralis and other social preferences is a topic for future research.
A final point we make concerns the status of economics as a discipline,
in the eyes of the general public and among the other behavioral and social
sciences. Conventional economics textbooks may give the false impression
that selfishness is part of economic rationality (see the discussion in Rubin-
stein (2006) and the references therein). This misreading of conventional
economics probably hurts the reputation of economists. If economists
would instead use partly morally motivated agents, such as Homo moralis,
then such misunderstandings could be avoided, and the critique would fall
flat to the ground. The economist’s analysis would then not be prejudiced
in favor of either selfishness or morality. Instead it would allow for the
whole spectrum of intermediate degrees of morality, spanning from pure
interest to pure Kantian morality.
self-
22. This example is due to Peter Diamond, discussed in a conversation many years
ago with one of the authors.
412 Ingela Alger and Jörgen W. Weibull
References
Acemoglu, Daron, Suresh Naidu, Pascual Restrepo, and James A. Robinson. 2014.
“Democracy Does Cause Growth.” NBER Working Paper 20004, National Bureau of
Economic Research, Cambridge, MA.
Akerlof, George, and Rachel E. Kranton. 2000. “Economics and Identity.” Quarterly
753.
Journal of Economics 115 (3): 715–
Alexander, Richard D. 1987. The Biology of Moral Systems. New York: Aldine De Gruyter.
Algan, Yann, and Pierre Cahuc. 2010. “Inherited Trust and Growth.” American Eco-
2092.
nomic Review 100 (5): 2060–
Alger, Ingela, and Albert Ma. 2003. “Moral Hazard, Insurance, and Some Collusion.”
247.
Journal of Economic Behavior and Organization 50 (2): 225–
Alger, Ingela, and Régis Renault. 2006. “Screening Ethics When Honest Agents Care
85.
about Fairness.” International Economic Review 47 (1): 59–
Alger, Ingela, and Jörgen W. Weibull. 2010. “Kinship, Incentives, and Evolution.”
1758.
American Economic Review 100 (4): 1725–
Alger, Ingela, and Jörgen W. Weibull. 2013. “Homo Moralis—Preference Evolution
2302.
under Incomplete Information and Assortativity.” Econometrica 81 (6): 2269–
Alger, Ingela, and Jörgen W. Weibull. 2016. “Evolution and Kantian Morality.”
67.
Games and Economic Behavior 98: 55–
Andreoni, James. 1988. “Privately Provided Public Goods in a Large Economy: The
73.
Limits of Altruism.” Journal of Public Economics 35 (1): 57–
Andreoni, James. 1990. “Impure Altruism and Donations to Public Goods: A Theory
Glow Giving.” Economic Journal 100 (401): 464–
of Warm- 477.
Arrow, Kenneth. 1970. “Political and Economic Evaluation of Social Effects and
23. New
Externalities.” In The Analysis of Public Output, edited by Julius Margolis, 1–
York: Columbia University Press.
Arrow, Kenneth. 1973. “Social Responsibility and Economic Efficiency.” Public Policy
21: 303–317.
Austen-Smith, David, and Jeffrey S. Banks. 1996. “Information Aggregation, Rational-
45.
ity, and the Condorcet Jury Theorem.” American Political Science Review 90 (1): 34–
Banerjee, Abhijit, and Jörgen W. Weibull. 1995. “Evolutionary Selection and Ratio-
nal Behavior.” In Learning and Rationality in Economics, edited by Alan Kirman and
Mark Salmon, 343– 363. Oxford: Basil Blackwell.
Becker, Gary S. 1974. “A Theory of Social Interaction.” Journal of Political Economy
82 (6): 1063–1093.
Morality: Foundations and Implications 413
Becker, Gary S. 1976. “Altruism, Egoism, and Genetic Fitness: Economics and Socio-
826.
biology.” Journal of Economic Literature 14 (3): 817–
Bénabou, Ronald, and Jean Tirole. 2006. “Incentives and Prosocial Behavior.” Ameri-
1678.
can Economic Review 96 (5): 1652–
Bénabou, Ronald, Armin Falk, and Jean Tirole. 2018. “Narratives, Imperatives and
Moral Reasoning.” Mimeo, Toulouse School of Economics, France.
Berg, Joyce, John Dickhaut, and Kevin McCabe. 1995. “Trust, Reciprocity, and Social
142.
History.” Games and Economic Behavior 10 (1): 122–
Bergstrom, Theodore C. 1995. “On the Evolution of Altruistic Ethical Rules for Sib-
81.
lings.” American Economic Review 85 (1): 58–
Bergstrom, Theodore C. 2003. “The Algebra of Assortative Encounters and the Evo-
228.
lution of Cooperation.” International Game Theory Review 5 (3): 211–
Bergstrom, Theodore C. 2009. “Ethics, Evolution, and Games among Neighbors.”
Working Paper, University of California, Santa Barbara.
Bergstrom, Theodore C. 2012. “Models of Assortative Matching.” Working Paper,
University of California, Santa Barbara.
Bernheim, B. Douglas. 1994. “A Theory of Conformity.” Journal of Political Economy
877.
102 (5): 841–
Brekke, Kjell A., Snorre Kverndokk, and Karine Nyborg. 2003. “An Economic Model
1983.
of Moral Motivation.” Journal of Public Economics 87 (9): 1967–
Cesarini, David, Christopher T. Dawes, James H. Fowler, Magnus Johannesson,
Paul Lichtenstein, and Björn Wallace. 2008. “Heritability of Cooperative Behav-
ior in the Trust Game.” Proceedings of the National Academy of Sciences 105 (10):
3721–3726.
Charness, Gary, and Martin Dufwenberg. 2006. “Promises and Partnership.” Econo-
1601.
metrica 74 (6): 1579–
Condorcet, Nicolas de. 1785. Essai sur l’Application de l’Analyse à la Probabilité des
Décisions Rendues à la Pluralité des Voix. Paris: L’Imprimerie Royale.
Currarini, Sergio, Matthew O. Jackson, and Paolo Pin. 2009. “An Economic Model
of Friendship: Homophily, Minorities and Segregation.” Econometrica 77 (4):
1003–1045.
Currarini, Sergio, Matthew O. Jackson, and Paolo Pin. 2010. “Identifying the Roles
Based Choice and Chance in High School Friendship Network Formation.”
of Race-
4861.
Proceedings of the National Academy of Sciences 107 (11): 4857–
Darwin, Charles. 1859. On the Origin of Species by Means of Natural Selection. London:
John Murray.
414 Ingela Alger and Jörgen W. Weibull
Dawkins, Richard. 1976. The Selfish Gene. Oxford: Oxford University Press.
Dekel, Eddie, Jeffrey C. Ely, and Okan Yilankaya. 2007. “Evolution of Preferences.”
704.
Review of Economic Studies 74 (3): 685–
Demichelis, Stefano, and Jörgen W. Weibull. 2008. “Language, Meaning, and
Games: A Model of Communication, Coordination, and Equilibrium.” American
1311.
Economic Review 98 (4): 1292–
Edgeworth, Francis Y. 1881. Mathematical Psychics: An Essay on the Application of
Mathematics to the Moral Sciences. London: Kegan Paul.
Ellingsen, Tore, and Magnus Johannesson. 2008. “Pride and Prejudice: The Human
1008.
Side of Incentive Theory.” American Economic Review 98 (3): 990–
Fehr, Ernst, and Klaus Schmidt. 1999. “A Theory of Fairness, Competition, and
868.
Cooperation.” Quarterly Journal of Economics 114 (3): 817–
Frank, Robert H. 1987. “If Homo Economicus Could Choose His Own Utility Func-
tion, Would He Want One with a Conscience?” American Economic Review 77 (4):
593–604.
Friedman, Milton. 1953. Essays in Positive Economics. Chicago: University of Chicago
Press.
Ghosh, Parikshit, and Debraj Ray. 2016. “Information and Enforcement in Informal
90.
Credit Markets.” Economica 83 (329): 59–
Güth, Werner, Rolf Schmittberger, and Bernd Schwarze. 1982. “An Experimental
Analysis of Ultimatum Bargaining.” Journal of Economic Behavior and Organization 3
(4): 367–388.
Güth, Werner, and Menahem E. Yaari. 1992. “Explaining Reciprocal Behavior
in Simple Strategic Games: An Evolutionary Approach.” In Explaining Process and
24. Ann
Change: Approaches to Evolutionary Economics, edited by Ulrich Witt, 23–
Arbor: University of Michigan Press.
Hamilton, William D. 1964. “The Genetical Evolution of Social Behaviour.” Journal
16.
of Theoretical Biology 7 (1): 1–
Harsanyi, John C. 1980. “Rule Utilitarianism, Rights, Obligations and the Theory of
133.
Rational Behavior.” Theory and Decision 12 (2): 115–
Harsanyi, John C. 1992. “Game and Decision Theoretic Models in Economics.” In
Handbook of Game Theory with Economic Application, volume 1, edited by Robert J.
707. Amsterdam: North-
Aumann and Sergiu Hart, 669– Holland.
Heifetz, Aviad, Chris Shannon, and Yossi Spiegel. 2007. “What to Maximize If You
57.
Must.” Journal of Economic Theory 133 (1): 31–
Morality: Foundations and Implications 415
Huck, Steffen, Dorothea Kübler, and Jörgen W. Weibull. 2012. “Social Norms and
Economic Incentives in Firms.” Journal of Economic Behavior and Organization 83 (2):
173–185.
Kant, Immanuel. [1785] 2002. Groundwork for the Metaphysics of Morals. Edited and
translated by Allen W. Wood. New Haven, CT: Yale University Press.
Kendall, Jake, Nataliya Mylenko, and Alejandro Ponce. 2010. “Measuring Financial
Access around the World.” World Bank Policy Research Working Paper 5253, World
Bank, Washington, DC.
Jacques. 1975. “Macroeconomic Constraints, Economic Efficiency and
Laffont, Jean-
437.
Ethics: An Introduction to Kantian Economics.” Economica 42 (168): 430–
Levine, David K. 1998. “Modelling Altruism and Spitefulness in Experiments.”
622.
Review of Economic Dynamics 1 (3): 593–
Lindbeck, Assar, and Jörgen W. Weibull. 1988. “Altruism and Time Consistency: The
1182.
Economics of Fait Accompli.” Journal of Political Economy 96 (6): 1165–
Lindbeck, Assar, Sten Nyberg, and Jörgen W. Weibull. 1999. “Social Norms and Eco-
35.
nomic Incentives in the Welfare State.” Quarterly Journal of Economics 114 (1): 1–
Maynard Smith, John, and George R. Price. 1973. “The Logic of Animal Conflict.”
18.
Nature 246 (November): 15–
Lovin, and James M. Cook. 2001. “Birds of a Feather:
McPherson, Miller, Lynn Smith-
444.
Homophily in Social Networks.” Annual Review of Sociology 27: 415–
Musgrave, Richard A. 1959. The Theory of Public Finance: A Study in Public Economy.
New York: McGraw-Hill.
Myerson, Roger, and Jörgen W. Weibull. 2015. “Tenable Strategy Blocks and Settled
976.
Equilibria.” Econometrica 83 (3): 943–
Nosenzo, Daniele, Simone Quercia, and Martin Sefton. 2015. “Cooperation in Small
14.
Groups: The Effect of Group Size.” Experimental Economics 18 (1): 4–
Redondo. 2001. “On the Evolution of Individualistic
Ok, Efe A., and Fernando Vega-
Preferences: An Incomplete Information Scenario.” Journal of Economic Theory 97 (2):
231–254.
Persson, Torsten, and Guido Tabellini. 2006. “Democracy and Development: The
324.
Devil in the Details.” American Economic Review 96 (2): 319–
Rawls, John. 1971. A Theory of Justice. Cambridge, MA: Harvard University Press.
Rousset, François. 2004. Genetic Structure and Selection in Subdivided Populations.
Princeton, NJ: Princeton University Press.
416 Ingela Alger and Jörgen W. Weibull
Rubinstein, Ariel. 2006. “A Sceptic’s Comment on the Study of Economics.” Eco-
C9.
nomic Journal 116: C1–
Ruef, Martin, Howard E. Aldrich, and Nancy M. Carter. 2003. “The Structure of
Founding Teams: Homophily, Strong Ties, and Isolation among U.S. Entrepreneurs.”
222.
American Sociological Review 68 (2): 195–
Sandmo, Agnar. 2005. “The Theory of Tax Evasion: A Retrospective View.” National
663.
Tax Journal 58 (4): 643–
Sen, Amartya K. 1977. “Rational Fools: A Critique of the Behavioral Foundations of
344.
Economic Theory.” Philosophy & Public Affairs 6 (4): 317–
Smith, Adam. [1759] 1976. The Theory of Moral Sentiments. Edited by David D.
Raphael and Alec L. Macfie. Oxford: Oxford University Press.
Smith, Adam. [1776] 1976. An Inquiry into the Nature and Causes of the Wealth of
Nations. Edited by William B. Todd. Oxford: Oxford University Press.
159.
Wright, Sewall. 1931. “Evolution in Mendelian Populations.” Genetics 16 (2): 97–
84.
Young, H. Peyton. 1993. “The Evolution of Conventions.” Econometrica 61 (1): 57–
Comment: Lawrence E. Blume
The topic of this chapter has been central to the research agendas of Ingela
Alger and Jörgen Weibull. Both separately (Lindbeck, Nyberg, and Weibull
1999; Alger 2010) and together (Alger and Weibull 2012), they have made
important contributions to the study of prosocial preferences and behavior.
Weibull research program on the social construction of other-
The Alger-
regarding preferences (Alger and Weibull 2013, 2016, this chapter) is excit-
ing both for its formal development of the foundations of evolutionary
game theory and for its findings concerning a cultural evolution model
regarding preferences. It is conventional to
of the development of other-
assume that individuals’ preferences over social states are concerned only
with their own material outcomes. This assumption makes possible the
powerful duality between social optimality and market outcomes expressed
in the “welfare theorems” and is the baseline environment in which social
policy is examined. It is nonetheless naïve in its assumption that prefer-
ences are primitives, exogenous in a model of social behavior. The recogni-
tion that preferences are to some degree socially constructed challenges
many fundamental findings of economic theory; in particular, anything
having to do with welfare conclusions. Alger and Weibull have significantly
enriched the literature on the social construction of preferences by examin-
ing the evolutionary foundations of preference relations.
Evolutionary Game Theory
The Alger and Weibull research program develops new evolutionary game
theory tools to say something about the kinds of preferences that would
persist in a social system. In the conventional noncooperative theory of
418 Comments by Lawrence E. Blume and Xavier Giné
person symmetric games, a symmetric Nash equilibrium is a strategy
N-
that is a best response for any one individual if it is being used by all the
other participants. The fundamental equilibrium concept of evolutionary
game theory is that of an evolutionarily stable strategy (ESS).1 Intuitively,
an ESS is a strategy that cannot be “invaded” by another strategy. What
does “invasion” mean? Suppose that a large population of individuals are
matched at random to participate in the game. An ESS has the property that
if a sufficiently large fraction of the population uses it while the remain-
der of the population uses any other strategy, the expected payoffs to the
ESS players are greater than those of the residual population. Evolutionary
game theory arose first in biology, and so the strategy alternative to the ESS
is said to be invading, and the motivation for the concept is that payoffs
measure fitness. Higher payoffs mean higher fitness, and the part of the
population using the ESS will have higher average fitness and therefore
will outreproduce the group using the invading strategy. To see how the
two concepts of Nash equilibrium and ESS fit together, one can check that
in any finite and symmetric game, every ESS is a Nash equilibrium of the
game. The converse, however, is false.
Alger and Weibull’s program is in the tradition of Güth and Yaari’s
(1992) indirect evolutionary approach.2 Whereas in traditional evolution-
ary game theory, preferences are fixed and the evolution of the distribu-
tion of actions is governed by the distribution of utility payoffs, in indirect
evolutionary models, selection pressure on actions causes the distribution
of utilities to evolve. This is analogous to biological models in which selec-
tion on phenotypes regulates the distribution of genotypes. In the Güth
and Yaari program, behaviors correspond to phenotypes and preferences to
genotypes. Payoffs in the game correspond to reproductive fitness.
Weibull program, a strategic interaction is described by a
In the Alger-
π
material payoff function that assigns to each strategy profile a material
payoff, (e.g., profit in a model of firm competition).3 Players’ choices are
1. Maynard Smith and Price (1973).
2. See also Güth and Kliemt (1998).
3. Material payoffs, like von Neumann- Morgenstern payoff functions, are linear
in the distribution of pure strategy profiles. The present chapter mostly discusses
symmetric two- person interactions, but Alger and Weibull (2016) considers multi-
player interactions under an aggregative assumption that in the material payoff of
an agent’s choices, the choices of others are exchangeable.
Morality: Foundations and Implications 419
governed not by material payoff π, however, but by a payoff function u that
represents subjective expected utility preferences over outcomes.
Alger and Weibull repurpose the ESS solution concept from evolution-
ary game theory as an equilibrium concept for the distribution of von
Morgenstern payoff functions in the population rather than for
Neumann–
Weibull research program, a pay-
the distribution of strategies. In the Alger-
off function u is an ESS if when a sufficiently large fraction of the popula-
tion uses it while the remainder of the population uses any other payoff
function, the expected material payoffs to those with the u payoff function
are greater than those of the residual population. A second feature of the
and this is key for their results—
Alger and Weibull program— is that match-
ing is not random but assortative: Like tends to match with like.4
The indirect evolution of preferences with assortative matching pro-
duces novel results. The authors label a payoff function a Homo moralis
payoff function if u(x, y) is an average of π(x, y), the material benefit of play-
ing x when others play y, and π(x, x), the material benefit when everyone
plays according to x. One end of this class is Homo oeconomicus, where the
averaging weights puts all weight on π(x, y), and the other end they label
Homo kantiensis, where all weight is put on the material benefits assuming
everyone plays the same.
Other work on the evolution of preferences is close in spirit to the
Alger and Weibull program, but different assumptions lead to different
outcomes. For instance, Ely and Yilankaya (2001) consider the evolution
of preferences in a population using a static stability concept motivated
much as is ESS. Because they consider only random matching, they find
that outcomes are stable if and only if they are equilibria of the game
described by material payoffs; that is, the stable preferences are those of
Homo oeconomicus. The evolution of social behavior, as opposed to other-
regarding preferences, is by now an old topic in evolutionary biology.
Hamilton (1964) sees inclusive fitness as an explanation for prosocial
behavior, and Grafen (1979) attempts to provide formal support for this
idea by considering ESS with nonrandom matching. Bergstrom (1995)
considers nonrandom matching for the evolution of altruistic play in a
4. This is not simple to describe in depth, so following the Alger and Weibull essay
in this chapter, I shall not attempt to describe it. It is clearly defined in Alger and
Weibull (2016, 61).
420 Comments by Lawrence E. Blume and Xavier Giné
explicitly biological context and derives Homo moralis preferences with
κ = Kantian.” There is also some
1/2. He called these preferences “semi-
support favoring antisocial preferences. Koçkesen, Ok, and Sethi (2000)
introduce a class of payoff functions that depend increasingly on material
returns and on relative material returns. Thus if everyone else’s material
returns decline while mine do not, then my utility increases. They find
that in every equilibrium in a class of games much like those considered
by Alger and Weibull but with complete rather than incomplete informa-
tion, those players with antisocial preferences do materially better than do
players who maximize material returns. This is not an evolutionary analy-
sis, but it suggests one. Finally, the Alger and Weibull results work because
those with the “right” payoff function receive more material benefits than
do others, and sometimes the “right” payoff function is not that of Homo
oeconomicus. Bester and Güth (1998) and Eshel, Samuelson, and Shaked
regarding preferences do materially
(1998) develop models where other-
worse than does Homo oeconomicus, and yet they survive because of group
selection effects. The conclusion to draw from this is that details matter for
the results of evolutionary models, and we are far from having a complete
understanding of how different configurations of environmental charac-
teristics collectively determine evolutionary outcomes. Thus conclusion
3 in the final section of the Alger and Weibull essay in this chapter5 is an
overstatement. Natural selection does not “favor human motivation in the
form of Homo oeconomicus.” Different models of natural selection favor dif-
ferent preference relations. Homo moralis and Homo oeconomicus are two.
Nonetheless, Alger and Weibull are to be commended for filling in a new
and important part of this landscape.
The promise of the indirect evolutionary approach goes far beyond
selection over payoff functions. In evolutionary game theory as received
from the biologists, selection forces act on payoffs, and the distribution
of strategies evolve. A second level of selection is the indirect evolution-
ary paradigm. In this case, preferences (which is to say the game itself)
evolve in some fashion. Mechanisms for preference evolution include such
phenomena as social learning, imitation and other adaptive processes, and
5. p. 410.
Morality: Foundations and Implications 421
the sorting of individuals across roles. These processes operate on a sys-
tem level rather than at the level of the individual. For instance, Blume
and Easley (1992, 2006) show how the redistribution of wealth through
repeated trading can drive some kinds of traders from the market. Thus
although there are nearly as many behavioral models of choice as there are
behavioral economists, only some of them can pass the market survival
test. Yet a third level has both strategy choices and the strategic environ-
ment coevolve through time. For instance, some papers look at evolution
where the community structure, represented by a social network, coevolves
with strategic choice and not payoff functions (e.g., Ely 2002; Goyal and
Redondo 2005; Staudigl and Weidenholzer 2014). A novel paper by
Vega-
Sandholm (2002) applies the idea of coevolution of strategies and the game
to mechanism design. Moving beyond the Alger and Weibull program, the
coevolution of preferences and game forms could contribute much to cen-
tral questions in political economy in particular and, more generally, the
analysis of institutions.
Symmetry
One limitation of their current essay and indeed, the research program, is
that Alger and Weibull have so far studied only symmetric environments;
player games, those in which the roles of player 1 and player
that is, for two-
2 are identical. This limitation is disappointing, because ESS can certainly
be generalized to asymmetric games.6 Knowing that Homo moralis arises in
a “one-
symmetric models— population” model, one wonders what would
population model.
emerge from a multiple-
Alger and Weibull consider an asymmetric problem in the third sec-
tion of their essay in this chapter. I am dissatisfied with their treatment
for reasons that foreshadow issues I raise in the sections below having to
do with the distinction between positive and normative claims. The stra-
tegic situation of their third section imagines a borrower and a lender; the
lender has to decide whether to make the loan, and the borrower has to
decide whether to pay it back. This is a great example (despite my qualms),
6. See, for instance, Fishman (2008) and citations therein.
422 Comments by Lawrence E. Blume and Xavier Giné
because one can see the surprising power of the ESS in preferences. A more
conventional analysis would consider repeated interactions between bor-
rowers and lenders. Loans would be made and paid back, because in ongo-
ing relations, reciprocity has value. The borrower understands that if he
pays back today, he may be able to get a loan tomorrow. In equilibrium,
the lender understands that the borrower understands this, and so she is
willing to make the loan. Furthermore, it is her willingness to make future
loans that validates the borrower’s belief. Alger and Weibull consider only
shot interactions—
one- dependent behav-
there is no possibility of history-
ior. Nonetheless, lending and borrowing can be sustained.
So what is wrong with this? To apply their tools, Alger and Weibull must
symmetrize the situation. They state that a canonical way to do this is to
initially cover the interaction under a “veil of ignorance” as to who will be
in what role. They assume that these roles are contingent. At any moment,
a given individual from a single population can either be a borrower or a
lender; essentially determined by the flip of a coin. The justification for this
move is hinted at by the phrase “veil of ignorance.” They call on the usual
Harsanyi (1953), Rawls (1958), and Vickrey (1945)—
suspects— who intro-
duced this move in the analysis of social systems. However, the suspects
introduced the veil of ignorance, the original position, ex ante randomness,
for purposes of normative analyses. The original position, behind the veil
of ignorance, is a counterfactual hypothetical that provides a frame outside
the social system for evaluating the moral consequences of its outcomes.
We do not pretend that individuals are actually randomized in such a way.
The evolutionary model, however, is concerned with real environments
rather than counterfactuals. Of course, there could be situations where roles
really are random; a given individual could play one role today and another
role tomorrow. But I do not believe that this is a useful way to think about
the evolution of preferences where each individual’s role is known and cer-
tain, set in stone. The use of normative analyses to justify positive claims
is one example of the conflation of positive and normative that, I believe,
obscure the significance of Alger and Weibull’s findings.7
7. Their Kantian claims would be much more compelling if a given individual consid-
ered the situation of the other party even though she will never ever be in that role.
This seems to be required by several of Kant’s expressions of his fundamental law.
Morality: Foundations and Implications 423
Welfare Economics
Alger and Weibull have uncovered some powerful results in the positive
theory of socially constructed preferences. Their treatment of normative
questions, however, and the distinctly normative cast of their entire essay,
raises some issues. For instance, how should we view Homo moralis prefer-
ences from the consequentialist perspective that is traditional in econom-
ics? The examples of their third section suggest that a Homo moralis world
may be materially better than an Homo oeconomicus world. To see that this
is not the case, consider a variant of the public goods game they discuss. In
this variant, N individuals can give, an outcome that is either 0 or 1. The
material benefit of the public good is ϱ to each person, the material cost of
< c <
giving is 0 1, and the public good will be provided if and only if the
sum of the contributions is at least 1.
Thus, letting y − i = ∑ j ≠i x j,
Suppose that Nϱ > c, so that the aggregate material benefit exceeds the cost
of provision. It is socially optimal for one individual to provide the good, and
the net benefit will be Nϱ − c.
The analysis breaks down into three cases (ignoring boundaries). If ϱ > c,
one person on his or her own should be willing to give. At one extreme,
Homo kantiensis chooses to maximize π(x, x). The optimum is, x = 1, every-
one gives, and the public good will be massively oversubscribed. If utilities
are interpersonally comparable, the optimum achievable welfare is Nϱ − c,
and Homo kantiensis society achieves N(ϱ − c), for a material payoff loss of
(N − 1)c. At the other extreme, Homo oeconomicus can achieve efficiency in
N distinct asymmetric Nash equilibria. In each equilibrium, one and only
one individual gives.
If ϱ < c, then Homo kantiensis gives zero. The asymmetric pure Nash equi-
libria of Homo oeconomicus also disappear, and Homo oeconomicus also gives
zero. Both, then, are inefficient.
When ϱ > c, there is also a symmetric mixed Nash equilibrium in which
the probability of choosing zero is c 1/( n − 1). In this case, the expected value
of the equilibrium to an individual Homo oeconomicus is ρ − c − ( ρ − 1)c n /( n −1) .
Comparing this payoff to that of Homo kantiensis, we see that it is materially
worse when ϱ > 1 but materially better when c/N < ϱ < 1.
In summary, for ϱ <
c, both preference types achieve the efficient out-
come. For c < ϱ < 1, a Homo kantiensis society does materially worse than
424 Comments by Lawrence E. Blume and Xavier Giné
every Nash equilibrium outcome of a Homo oeconomicus society. And for
ϱ > 1, some H. oeconomicus equilibria are efficient, with higher material pay-
off than that of the H. kantiensis society, but the symmetric mixed oeco-
nomicus equilibrium is worse.8
The general point is that there are problems that, despite being posed
symmetrically, have optimal solutions that are asymmetric. Minority games
and the related El Farol game provide further examples. This example serves
as a caveat to conclusion 4 of the Alger and Weibull essay in this chapter
that designing policies for Homo oeconomicus when individuals are in fact
Homo moralis may overincentivize them. Yes, it can, but it may not.
Alger and Weibull’s examples in their third section raise the interest-
ing question of how welfare economics should be conducted when prefer-
regarding. They follow Harsanyi (1980, 1992) and argue
ences are other-
that welfare should be measured as the sum of individual material utili-
ties. I followed them in my preceding public good example for purposes of
comparison, but this is controversial. To see why, ask: Why exactly is one’s
desire for a drink of water for herself more necessary to the social welfare
calculation then her desire to offer her companion a drink? I can think of
two arguments in favor of this claim: one, that water is a necessity for life,
and if anything is fundamental, survival needs should be; the other, that to
count the companion’s welfare in her utility is to double count it. The first
argument is nothing more than a statement about marginal rates of substi-
tution at the boundary of the consumption set. At the survival boundary,
water for one’s self is critical. The second argument says that the utility a
decision maker gets from a drink is different from the utility she gets from
giving someone else utility. If you take a drink of water, you get some util-
ity. If I offer you that drink, the utility that I get does not count in the social
calculation. But if I expend my own resources to do it, the opportunity cost
8. One can derive similar results for the middle- ground cases. The treatment of mix-
ing with Homo moralis preferences is unusual, except in the extreme oeconomicus case.
In Alger and Weibull (2016), we are told that the set X on which π(x, x) is defined
is the set of mixed strategies in the material game. I understand this to mean that
if I were, say, kantiensis (just for clarity), and if I chose 1 with probability p, then I
assume everyone else is too, and when I consider what happens if I were to choose
1 with probability p' instead, I assume everyone else chooses p' too. This leads to a
symmetric randomized equilibrium with an expected social net material benefit that
converges upward to n(ρ − c) as n increases.
Morality: Foundations and Implications 425
of providing the gift does again count. Apparently, only certain actions
are allowed to generate utility for welfare purposes. In my view, neither of
these arguments holds water.
Alger and Weibull adopt Harsanyi’s distinction between personal and
social preferences, and they note that one might understand Homo mora-
lis as an individual whose personal preferences are the material prefer-
ences π and whose social preferences are given by the Homo moralis utility
function with its degree κ of morality. Ken Arrow famously wrote,9 “I am
fashioned enough to retain David Hume’s view that one can never
old-
derive ‘ought’ propositions from ‘is’ propositions.” The findings of evolu-
tionary game theory are “is” propositions. Alger and Weibull are eager to
derive from them “oughts.” The conflation of “is” and “ought” perhaps
undercuts the “is” exercise of their research program.
Alger and Weibull write:
If we take the material payoff function to represent personal utility, then welfare
in a society consisting of Homo moralis individuals (each with his or her degree of
morality) should be defined simply as the sum of their expected material payoffs,
just as in ordinary utilitarian welfare theory.
Harsanyi takes personal preferences to be those preferences that guide
individuals’ choices, their “everyday behavior.”10 If this is what Alger and
Weibull mean by personal preferences, then the moralis payoff function
should represent personal preferences and not material payoffs, and Alger
and Weibull’s and my welfare calculations are incorrectly done. Harsanyi’s
description of personal preferences can certainly allow for externalities. If
Alger and Weibull believe, following Harsanyi’s paradigm, that moralis pref-
erences represent what he calls “moral preferences,” then I do not under-
stand why they would appear in an evolutionary analysis; decisions are not
made based on moral preferences, and so they cannot be selected on.11
If Homo moralis preferences are the right preferences to undertake calcu-
lations with, then one cannot make welfare comparisons across populations
with different degrees κ of morality. By analogy, we might consider two dif-
ferent production economies that differ only in consumers’ preferences.
9. In his Ely Lecture, Arrow (1994, 1).
10. Harsanyi (1992, 675).
11. Harsanyi (1992, 671) says that “rational behavior is not a descriptive concept but
rather is a normative concept.” So he is an odd partner for evolutionary game theory.
426 Comments by Lawrence E. Blume and Xavier Giné
We might observe that one economy has a higher GDP than the other, but
this gives no guide for comparing the welfare of the two economies, even if
utility is interpersonally comparable.
Homo moralis as a Moral Theory
Alger and Weibull write that their work “suggests an evolutionary founda-
tion for a psychologically plausible form of morality, in line with Immanuel
Kant’s categorical imperative” (page 392, this chapter). Strictly speaking, they
provide “an evolutionary foundation for” preferences that describe behavior
consistent with “a psychologically plausible form of morality.” What kind of
moral theory? They suggest it is “in line with Immanuel Kant’s categorical
Kantian” to describe
imperative.” Bergstrom (1995) uses the phrase “semi-
Homo moralis preferences with κ = 1/2. I believe this Kantian affiliation comes
from a misreading of Kant. The idea of Kantian preferences exists outside
evolutionary game theory. Roemer (2010) calls a strategy profile in a certain
class of games “Kantian” if it is immune to simultaneous proportional devia-
tions from all the players.
Broadly speaking, moral theories fall into one of three classes: conse-
quentialist theories, deontological theories, and virtue theories. Conse-
quentialism emphasizes the consequences of actions. Welfare economics is
consequentialist. Deontological theories emphasize duties, rules, and obli-
gations. Most philosophers, including Kant, consider(ed) Kantian theories
to be deontological.12 Virtue ethics emphasizes virtues or moral character.
To illustrate, suppose someone’s life is in danger and can be saved by my
telling a lie. A consequentialist would lie, because he believes that saving
a life is a good outcome. A deontologist would lie if he believed that sav-
ing a life when one can without doing injury to others is a universal law.
However, if he believed “never lie” is a universal maxim, then he would
not lie even to save a life. A virtue ethicist would lie because saving a life is
benevolent; a virtue. I claim that Homo moralis has much more to do with
virtue ethics than with any deontological moral theory.
12. Kagan (2002, 112).
Morality: Foundations and Implications 427
The fundamental moral principle, according to Kant, is a categorical
imperative: imperative because it is a command, and categorical because
it is required of us unconditionally. That moral principle is, “act only in
accordance with that maxim through which you can at the same time will
that it become a universal law,” or, in another formulation by Kant, to “act
as if the maxim of your action were to become through your will a universal
law of nature.” Where does this come from? Kant wrote:13
Everyone must admit that a law, if it is to be valid morally, i.e., as the ground of
an obligation, has to carry absolute necessity with it; that the command “You
ought not to lie” is valid not merely for human beings, as though other rational
beings did not have to heed it; and likewise all the other genuinely moral laws;
hence that the ground of obligation here is to be sought not in the nature of the
human being or the circumstances of the world in which he is placed, but a priori
solely in concepts of pure reason, and that every other precept grounded on prin-
ciples of mere experience, and even a precept that is universal in a certain aspect,
insofar as it is supported in the smallest part on empirical grounds, perhaps only
as to its motive, can be called a practical rule, but never a moral law.
In other words, it is to be rationally derivable, assuming that every human
were to heed it. The law is based entirely on reason and is not a conse-
quence of any facts on the ground. In particular, moral propositions are to
be independent of whom they are applied to; their preferences make no
difference. These propositions are independent of our desires and uncou-
pled from the consequences that ensue. Clearly, however, the rules that one
would derive from Homo moralis preferences depend on what the material
payoffs are: Consequences matter. To put this somewhat differently, Kant’s
theoretic nature: An assumption about
categorical imperative has a game-
the behavior of others enters into your calculation about how you should
behave. But Homo kantiensis is not Kantian, because his evaluation of the
act is independent of his preferences. If a given maxim survives the cat-
egorical imperative test, one is obliged to act according to it, even if it is
preference minimal. Thus the moral theory for which Alger and Weibull
“provide an evolutionary foundation” is not Kantian. Quite the opposite.
Harsanyi (1980) calls individuals who maximize a class of utility functions
13. Kant ([1785] 2002, 5).
428 Comments by Lawrence E. Blume and Xavier Giné
containing Homo moralis payoff functions rule “utilitarians.” It appears to
be consequentialist.14
To the extent that we use the language of choice theory to talk about
moral choices of individuals, any such theory will appear to be consequen-
tialist. One can read virtue ethics this way. Our preferences are shaped by
our character. Thus in some situations, preferences of individuals who
have internalized particular virtues will look different than those of indi-
viduals who have not. And so the choices of those of virtuous character—
will reflect these virtues. These are moral
sympathetic, charitable, etc.—
choices. One school of modern virtue ethics, so- based ethics,15
called agent-
“understands rightness in terms of good motivations and wrongness in
terms of the having of bad (or insufficiently good) motives.” Alger and
Weibull’s evolutionary account of preference evolution supports this view.
They tell us that, as a consequence of the social condition, as a result of
social interaction, preferences must in the long run take on a certain form,
regarding.
and that form is other-
Adam Smith ([1759] 2004, 1) begins The Theory of Moral Sentiments by
claiming the universality of certain virtues:
How selfish soever man may be supposed, there are evidently some principles in
his nature, which interest him in the fortune of others, and render their happi-
ness necessary to him, though he derives nothing from it except the pleasure of
seeing it. Of this kind is pity or compassion, the emotion which we feel for the
misery of others, when we either see it, or are made to conceive it in a very lively
manner. That we often derive sorrow from the sorrow of others, is a matter of fact
too obvious to require any instances to prove it; for this sentiment, like all the
other original passions of human nature, is by no means confined to the virtuous
and humane, though they perhaps may feel it with the most exquisite sensibil-
ity. The greatest ruffian, the most hardened violator of the laws of society, is not
altogether without it.
He goes on to argue16
feeling for the misery of others, that it is by
that this is the source of our fellow-
changing places in fancy with the sufferer, that we come either to conceive or to
14. In fairness, I should say that the contrast between deontology and consequen-
tialism is not as sharp as it is often made out to be and is somewhat contested. See
Kagan (2002) and Cummiskey (1990).
15. Slote (2001, 14).
16. Smith ([1759] 2004, 4).
Morality: Foundations and Implications 429
be affected by what he feels, may be demonstrated by many obvious observations,
if it should not be thought sufficiently evident of itself.
This expression of sympathy is, for Smith, the source of our moral decision-
making. In a passage that is reminiscent of Homo moralis, he states:17
The principle by which we naturally either approve or disapprove of our own
conduct, seems to be altogether the same with that by which we exercise the like
judgments concerning the conduct of other people. We either approve or disap-
prove of the conduct of another man according as we feel that, when we bring
his case home to ourselves, we either can or cannot entirely sympathize with the
sentiments and motives which directed it.
Assuming others behave as x, how do we feel about x?
Finally, it is interesting to note that perhaps Smith in the Theory of Moral
Sentiments would be sympathetic to the Alger and Weibull program. He
writes: 18
It is thus that the general rules of morality are formed. They are ultimately founded
upon experience of what, in particular instances, our moral faculties, our natural
sense of merit and propriety, approve, or disapprove of. We do not originally
approve or condemn particular actions; because, upon examination, they appear
to be agreeable or inconsistent with a certain general rule. The general rule, on the
contrary, is formed, by finding from experience, that all actions of a certain kind,
or circumstanced in a certain manner, are approved or disapproved of.
Our moral views emerge from experience, a social process. It would be ask-
seventeenth-
ing too much of the mid- century Smith to distinguish between
social learning and social evolution, and even today, it is not clear that, as
classes, these are observationally distinct. But Alger and Weibull need not
commit to a mechanism for their ESS analysis beyond the fact that it is
monotone in payoffs, and so they are not inconsistent with Smith.
Conclusion
Although I have reservations about Alger and Weibull’s (and many other
economists) assertions about moral theory, the Alger and Weibull research
program is among the most ambitious and promising to date on the explo-
regarding preferences. The results are
ration of the evolution of other-
17. Smith ([1759] 2004, 151–152).
18. Smith ([1759] 2004, 206).
430 Comments by Lawrence E. Blume and Xavier Giné
exciting both for what they find and for the extent of the environments
in which they hold.19 Received game and market theory is of the take-
all-
comer’s variety; equilibrium exists no matter what preferences agents hold.
But if preferences are socially constructed, the forces described by Alger
and Weibull should limit the kinds of preferences that are prevalent. Game
and market theory should take advantage of this fact to make sharper pre-
dictions about the behavior of social systems. Finally, Jörgen Weibull has
contributed significantly to the literature on evolutionary dynamics, and
so I look forward to seeing this program progress from the static analysis of
ESS to the much harder (but potentially richer) dynamic analyses that have
emerged in recent years.
References
Alger, Ingela. 2010. “Public Goods Games, Altruism, and Evolution.” Journal of Public
Economic Theory 12 (4): 789–813.
Alger, Ingela, and Jörgen W. Weibull. 2012. “A Generalization of Hamilton’s Rule—
Love Others How Much?” Journal of Theoretical Biology, Evolution of Cooperation 299:
42–54.
Alger, Ingela, and Jörgen W. Weibull. 2013. “Homo moralis—Preference Evolution
under Incomplete Information and Assortative Matching.” Econometrica 81 (6):
2269–2302.
Alger, Ingela, and Jörgen W. Weibull. 2016. “Evolution and Kantian Morality.”
67.
Games and Economic Behavior 98: 56–
Arrow, Kenneth. 1994. “Methodological Individualism and Social Knowledge.”
9.
American Economic Review 84 (2): 1–
Bergstrom, Theodore C. 1995. “On the Evolution of Altruistic Ethical Rules for Sib-
81.
lings.” American Economic Review 85 (1): 58–
Bester, Helmut, and Werner Güth. 1998. “Is Altruism Evolutionarily Stable?” Journal
209.
of Economic Behavior and Organization 34 (2): 193–
Blume, Lawrence, and David Easley. 1992. “Evolution and Market Behavior.” Journal
40.
of Economic Theory 58 (1): 9–
19. This is not apparent in the present chapter, but can be seen in Alger and Weibull
(2016).
Morality: Foundations and Implications 431
Blume, Lawrence, and David Easley. 2006. “If You’re So Smart, Why Aren’t You Rich?
966.
Belief Selection in Complete and Incomplete Markets.” Econometrica 74 (4): 929–
615.
Cummiskey, David. 1990. “Kantian Consequentialism.” Ethics 100 (3): 586–
32.
Ely, Jeffrey C. 2002. “Local Conventions.” Advances in Theoretical Economics 2 (1): 1–
Ely, Jeffrey C., and Okan Yilankaya. 2001. “Nash Equilibrium and the Evolution of
272.
Preferences.” Journal of Economic Theory 97 (2): 255–
Eshel, Ilan, Larry Samuelson, and Avner Shaked. 1998. “Altruists, Egoists, and Hooli-
gans in a Local Interaction Model.” American Economic Review 88 (1): 157– 179.
linear Pure
Fishman, Michael A. 2008. “Asymmetric Evolutionary Games with Non-
90.
Strategy Payoffs.” Games and Economic Behavior 63 (1): 77–
Redondo. 2005. “Network Formation and Social
Goyal, Sanjeev, and Fernando Vega-
207.
Coordination.” Games and Economic Behavior 50 (2): 178–
Dove Game Played between Relatives.” Animal
Grafen, Alan. 1979. “The Hawk-
907.
Behaviour 27 (3): 905–
Güth, Werner, and Hartmut Kliemt. 1998. “The Indirect Evolutionary Approach:
Bridging the Gap between Rationality and Adaptation.” Rationality and Society 10
(3): 377–399.
Güth, Werner, and Menahem E. Yaari. 1992. “Explaining Reciprocal Behavior
in Simple Strategic Games: An Evolutionary Approach.” In Explaining Process and
34. Ann
Change: Approaches to Evolutionary Economics, edited by Ulrich Witt, 23–
Arbor: University of Michigan Press.
Hamilton, William D. 1964. “The Genetical Evolution of Social Behavior, I and II.”
52.
Journal of Theoretical Biology 7 (1): 1–
Harsanyi, John C. 1953. “Cardinal Utility in Welfare Economics and in the Theory
435.
of Risk-Taking.” Journal of Political Economy 61 (5): 434–
Harsanyi, John C. 1980. “Rule Utilitarianism, Rights, Obligations and the Theory of
133.
Rational Behavior.” Theory and Decision 12 (2): 115–
Harsanyi, John C. 1992. “Game and Decision Theoretic Models in Ethics.” In
Handbook of Game Theory with Economic Applications, volume 1, edited by Robert J.
707. Amsterdam: North-
Aumann and Sergiu Hart, 669– Holland.
Kagan, Shelly. 2002. “Kantianism for Consequentialists.” In Groundwork for the Meta-
156. New Haven, CT: Yale Univer-
physics of Morals, edited by Allen W. Wood, 111–
sity Press.
Kant, Immanuel. [1785] 2002. Groundwork for the Metaphysics of Morals. Edited and
translated by Allen W. Wood. New Haven, CT: Yale University Press.
432 Comments by Lawrence E. Blume and Xavier Giné
Koçkesen, Levent, Efe A. Ok, and Rajiv Sethi. 2000. “Evolution of Interdepen-
dent Preferences in Aggregative Games.” Games and Economic Behavior 31 (2):
303–310.
Lindbeck, Assar, Sten Nyberg, and Jörgen W. Weibull. 1999. “Social Norms and
Economic Incentives in the Welfare State.” Quarterly Journal of Economics 114 (1):
1–35.
Maynard Smith, John, and George R. Price. 1973. “The Logic of Animal Conflict.”
18.
Nature 246 (November): 15–
194.
Rawls, John. 1958. “Justice as Fairness.” Philosophical Review 67 (2): 164–
Roemer, John E. 2010. “Kantian Equilibrium.” Scandinavian Journal of Economics 112
(1): 1–24.
Sandholm, William H. 2002. “Evolutionary Implementation and Congestion Pric-
689.
ing.” Review of Economic Studies 69 (3): 667–
Slote, Michael. 2001. Morals from Motives. Oxford: Oxford University Press.
Smith, Adam. [1759] 2004. The Theory of Moral Sentiments. New York: Barnes and
Noble.
Staudigl, Mathias, and Simon Weidenholzer. 2014. “Constrained Interactions and
63.
Social Coordination.” Journal of Economic Theory 152: 41–
Vickrey, William. 1945. “Measuring Marginal Utility by Reactions to Risk.” Econo-
333.
metrica 13 (4): 319–
Comment: Xavier Giné
Morality: Evolutionary Foundations and Policy Implications
Positive economic theories typically model agents interacting in markets
interested individuals. This chapter summarizes the work of the
as self-
authors in Alger and Weibull (2013, 2016), which questions this assump-
tion by investigating the preferences that humans would exhibit if these
preferences were transmitted from generation to generation.
In the case of interactions between two agents, the utility function of
humans would be a linear combination of the material payoff of a Homo
interested individual) and that of a Homo kantiensis (an
oeconomicus (or self-
individual who “does the right thing” by assuming the other agent will
behave as he or she does). Facing a set of choices similar to that of the other
agent, in practice, Homo kantiensis does not solve the Nash equilibrium
but rather chooses the payoff with highest value from the diagonal of the
matrix of the game. Remarkably, these preferences called Homo moralis pref-
erences are the only ones that are evolutionarily stable when the matching
protocol among agents is exogenous.
But where does the weight that defines the linear combination between
the two payoff functions come from? Alger and Weibull (2013) argue that
the weight, or the “degree of morality,” is related to the probability that
individuals are matched with others of same type relative to the probabil-
ity that they are matched with others of a different type. The model thus
interested agents as long as
predicts that individuals will not behave as self-
this probability is positive and that everyone in a society will share same
preferences (with the same degree of morality).
434 Comments by Lawrence E. Blume and Xavier Giné
Taking Predictions to Data
The first prediction is consistent with experimental evidence showing
robust deviations in behavior from the assumption of Homo oeconomicus
agents. It is unclear, however, whether these deviations reflect universal
social preferences or whether instead social preferences are shaped by the
economic, social, and cultural environment. Henrich et al. (2004) set out
to distinguish between these two hypotheses by conducting a large cross-
cultural study of behavior using several standard experimental games in
scale societies, ranging from foraging to sedentary agricultural
fifteen small-
societies.
The results confirm that there are violations of Homo oeconomicus, as
individuals seem to care about fairness and reciprocity. In addition, there is
dispersion across and within societies (of roughly equal magnitude) in the
degree to which the assumption of Homo oeconomicus is violated. The dis-
persion across societies can be explained by the Homo moralis preferences if
we assume that different societies exhibit different degrees of morality. But
the dispersion within societies cannot be explained, because all individuals
of a society share the same preferences.
Henrich et al. (2004) also suggest that prosocial behavior is correlated
with market integration. Homo moralis preferences, however, are correlated
with the degree of morality. It is unclear whether market integration is posi-
tively or negatively correlated with the degree of morality. One argument
suggests they are negatively correlated: market integration may increase the
probability of matching with individuals of another type, thus decreasing
the degree of morality.
The findings from Henrich et al. (2004) should perhaps be taken with
caution, as the relationship between market integration and values in the
section may suffer from endogeneity, because institutions and val-
cross-
Schündeln (2007) com-
ues may coevolve. For example, Alesina and Fuchs-
pare the attitudes toward redistribution of East and West Germans after the
reunification. They find that communism instilled in people the view that
being. This suggests that institutions
the state was essential for their well-
and political regimes can shape preferences, and therefore the degree of
morality may change even if the matching protocol did not.
Falk and Szech (2013) run an experiment in which individuals choose
between keeping money or saving a mouse. Decisions are made individually
Morality: Foundations and Implications 435
(involving the simple choice of getting money or saving the mouse) or
through a market mechanism involving many buyers and sellers. Sellers
are endowed with the mice and buyers with money. The mouse was killed
if a trade occurred, the seller kept the sale price, and the buyer the endow-
ment minus the sale price. If no trade occurred, the mouse survived, and
earnings for both players were zero. The authors find that the willingness
to keep the money (and thus to kill the mouse) is higher when decisions are
made through a market with many buyers and sellers compared to when
the market only has one buyer and seller. Put differently, Falk and Szech
(2013) suggest that market interaction may be negatively correlated with
the degree of morality.
Role of Institutions and Incentives
Contracts, subsidies, taxes, and other public policy issues are designed to
interested individuals to act in the common interest. David
induce self-
Hume, the Scottish philosopher and economist (and friend of Adam
Smith), said it best when arguing that public policy should be designed for
“knaves” motivated by the private interest.1 As Bowles (2008) puts it, the
invisible hand needs a helping hand.
In this chapter, Homo moralis individuals are not knaves. But could insti-
tutions designed for knaves end up turning individuals into knaves? In other
words, when individuals are not knaves, can incentives backfire? There is
certainly a literature suggesting that this is the case. One example is the
known study of six day-
well- care centers in Haifa by Gneezy and Rustichini
care centers decided to impose a fine on parents who were
(2000). The day-
late picking up their kids at the end of the day. Parents reacted to the fine by
doubling the fraction of time they arrived late. More importantly, once the
fine was removed, parents continued to be late when picking up the kids.
In another example, Giné, Mansuri, and Sreshtra (2018) study the impacts
of a monetary incentive given to the staff of a microfinance institution if
1. The quote is “in contriving any system of government […] every man ought to be
supposed a knave, and to have no other end, in all his actions, than private interest.
By this interest we must govern him, and, by means of it, make him, notwithstanding
his insatiable avarice and ambition, co- operate to public good.” (David Hume 1777),
http://oll.libertyfund.org/titles/hume-essays-moral-political-literary-lf-ed.
436 Comments by Lawrence E. Blume and Xavier Giné
they achieved certain “social” goals related to the empowerment and well-
being of their clients. For staff who worked in teams, such incentives led to
a worsening of social outcomes.
The critical assumption when designing incentive schemes is that
regarding motives may be present, they are not affected
although other-
by the schemes individuals face. This “separability” assumption fails in
both examples above, as they underscore the fact that monetary incentives
may diminish the intrinsic motivation of individuals to comply with social
norms (Bowles 2008; Bowles and Hwang 2008).
The preferences of Homo moralis individuals discussed in this chapter
maintain the separability assumption and therefore predict that policies
interested individuals will not backfire when applied to
designed for self-
Homo moralis. But one cannot help but wonder whether the degree of regu-
lation should be different across societies with different degrees of morality.
Indeed, although regulation may be essential in a society of Homo oeco-
nomicus, it may not be needed in a society of Homo kantiensis. This observa-
tion points to another hypothesis that could be tested in future empirical
research.
References
Schündeln. 2007. “Good-
Alesina, Alberto, and Nicola Fuchs- Bye Lenin (or Not?):
The Effect of Communism on People’s Preferences.” American Economic Review
97 (4): 1507–1528.
Alger, Ingela, and Jörgen W. Weibull. 2013. “Homo moralis—Preference Evolution
Under Incomplete Information and Assortative Matching.” Econometrica 81 (6):
2269–2302.
Alger, Ingela, and Jörgen W. Weibull. 2016. “Evolution and Kantian Morality.”
67.
Games and Economic Behavior 98: 56–
Interested Citizens May Under-
Bowles, Samuel. 2008. “Policies Designed for Self-
mine ‘the Moral Sentiments’: Evidence from Economic Experiments.” Science 320
(5883): 1605–1609.
Bowles, Samuel, and Sung- Ha Hwang. 2008. “Social Preferences and Public Econom-
ics: Mechanism Design When Social Preferences Depend on Incentives.” Journal of
1820.
Public Economics 92 (8): 1811–
Falk, Armin, and Nora Szech. 2013. “Morals and Markets.” Science 340 (6133):
707–711.
Morality: Foundations and Implications 437
Giné, Xavier, Ghazala Mansuri, and Slesh Sreshtra. 2018. “Mission and the Bottom
goal Development Organization.” Working
Line: Performance Incentives in a Multi-
Paper, World Bank, Washington, DC.
Gneezy, Uri, and Aldo Rustichini. 2000. “A Fine Is a Price.” Journal of Legal Studies
29 (1): 1–17.
Henrich, Joseph, Robert Boyd, Samuel Bowles, Colin F. Camerer, Ernst Fehr, Herbert
Gintis, and Richard McElreath. 2004. “Overview and Synthesis.” In Foundations of
Human Sociality: Economic Experiments and Ethnographic Evidence from Fifteen Small-
Scale Societies, edited by Joseph Henrich, Robert Boyd, Samuel Bowles, Colin F. Cam-
erer, Ernst Fehr, and Herbert Gintis. New York: Oxford University Press.
Hume, David. 1777. “Essay VI: Of the Independency of Parliament.” In Essays and
Treatises on Several Subjects. http://oll.libertyfund.org/titles/hume-essays-moral-political
-literary-lf-ed.
10 The Influence of Randomized Controlled Trials on
Development Economics Research and on
Development Policy
Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
Many (though by no means all) of the questions that development econo-
mists and policy makers ask themselves are causal in nature: What would
be the impact of adding computers in classrooms? What is the price elas-
ticity of demand for preventive health products? Would increasing inter-
est rates lead to an increase in default rates? Decades ago, the statistician
Fisher proposed a method to answer such causal questions: randomized
controlled trials (RCTs; Fisher 1925). In an RCT, the assignment of different
units to different treatment groups is chosen randomly. This ensures that
no unobservable characteristic of the units is reflected in the assignment,
and hence that any difference between treatment and control units reflects
the impact of the treatment. Although the idea is simple, the implementa-
tion in the field can be more involved, and it took some time before ran-
domization was considered to be a practical tool for answering questions
in social science research in general and in development economics more
specifically.
About 20 years ago, the idea of randomized controlled trials was just
starting to make its way into development economics. Starting in 1994,
started the use of randomized
Glewwe, Kremer, and Moulin (2009) kick-
evaluations among development economists and practitioners (Kremer
The views expressed in this document express the personal opinions of the author
and are entirely the authors’ own. They do not necessarily reflect the opinions of the
U.S. Agency for International Development (USAID) or the United States Govern-
ment. USAID is not responsible for the accuracy of any information supplied herein.
We thank Alison Fahey, Noor Iqbal, Sasha Gallant, Joaquin Carbonell, Adam Trow-
bridge, and Anne Healy for their support. We thank Rachel Glennerster for useful
comments, and Francine Loza and Laura Stilwell for excellent research assistance.
440 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
2003). In 1997, the PROGRESA randomized controlled trial began, mark-
scale policy effort in a developing
ing the first evaluation of a large-
country. With the launch of these randomized evaluations, we, perhaps
naively, expressed the hope that RCTs would revolutionize social policy
first century, much as they had revolutionized medicine in
in the twenty-
the twentieth century (Duflo and Kremer 2005; Duflo 2004; Banerjee et al.
2007). With the century less than 20 years old, it seems a little premature
to evaluate this claim. Randomized evaluations clearly take a larger place in
the policy conversation now than they did at the turn of the century, and
they receive substantially more funding from donor organizations and local
governments. Policy innovations that have been tested with RCTs have
reached millions of people. However, the amount of money involved is
still small. Development policy, moreover, is known for its twists and turns;
many have predicted that RCTs are just the current fad and, soon enough,
will have their comeuppance.
Something that we did not anticipate, however, has undoubtedly hap-
pened: Randomized controlled trials have, if not revolutionized, at least
profoundly altered, the practice of development economics as an academic
discipline. Some scholars applaud this change (we are obviously in that
camp), while others rue it (Deaton 2010; Ravallion 2012), but the fact is not
really in dispute. In this essay, we start by quantitatively documenting this
remarkable evolution. Here we discuss the ways in which the field has been
affected by the practice of RCTs and what we see as their main contribu-
tions to the practice of development economics.
The popularity of RCTs as a research tool has sometimes been seen as con-
flicting with their potential (or ambition) for changing the world. The view is
that the “academic” desire to come up with the cleverest research design may
not line up with the practitioners need to identify scalable innovations (the
next cell phone), or change “systems” (health care) or reform institutions
(democracy). Using the USAID Development Innovation Ventures (DIV)
portfolio as a case study, we identify the policy innovations tested with DIV
scale reach (more than 100,000
funding that have eventually led to large-
people). The analysis suggests that the proposed opposition between inter-
esting and important is not particularly pertinent. In practice, many of the
interventions supported by DIV that have reached this scale started as small
research projects driven by academics. These projects also had the greatest
The Influence of Randomized Controlled Trials 441
300
250
200
150
100
50
0
1975 1980 1985 1990 1995 2000 2005 2010 2015
PublicaƟon Year
Figure 10.1
Number of published RCTs
Source: Cameron, Drew B., Anjini Mishra, and Annette N. Brown. 2016. “The Growth
of Impact Evaluation for International Development: How Much Have We Learned?”
21.
Journal of Development Effectiveness 8 (1): 1–
“bang for the buck” evaluated in terms of lives eventually reached per USAID
initial funding dollars.1 We conclude this essay by discussing what this tells
us about the policy process and the role RCTs can have in it.
Rapid Growth
Over the past 15 years, the use of experiments has expanded in academia
and in international organizations: The DIME group at the World Bank lists
more than 200 studies, nearly all of them randomized, and Arianna Lego-
vini, the head of DIME, estimates that if we take the World Bank as a whole,
there are at least 475 RCTs going on (Legovini, personal communication).
Tables 10.1 and 10.2 and the figures in the chapter summarize some trends
in the use of experiments over time.
We start with a review of impact evaluations conducted by Cameron,
Mishra, and Brown (2016; figures 10.1 and 10.2). They compiled a reposi-
tory of 2,259 impact evaluation studies in development economics that
were published between 1981 and 2012 by searching all major academic
databases in health, economics, public policy, and the social sciences. They
1. This does not necessarily imply they have the highest social return.
442 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
Differences in Differences
2.40% 16.10% 16.70%
Instrumental Variables
8.30%
RCT
Regression DisconƟnuity
66.40%
Propensity Score Matching, or
Other Matching Method
Figure 10.2
Evaluations by type
Source: Cameron, Drew B., Anjini Mishra, and Annette N. Brown. 2016. “The Growth
of Impact Evaluation for International Development: How Much Have We Learned?”
21.
Journal of Development Effectiveness 8 (1): 1–
supplemented this with an online crowdsourcing effort, which offered a $10
gift certificate per qualifying paper that was not already in the database. They
then classified the papers by sector and by type. Overall, 66 percent (1,491)
of those evaluations are RCTs. Figure 10.1 shows that the number of RCTs
has grown rapidly over time.
Next, we look at the data compiled by Aidgrade (Vivalt 2015). Aidgrade
compiles the results of impact evaluations of development interventions.
According to Vivalt:
The evaluations included in the AidGrade database were carefully selected from a
number of different databases and online sources, the detailed process for which
is outlined in Vivalt (2015). AidGrade org employees first chose 30 topics they
.
felt were important development issues. Those lists were combined and made
into one large list of topics. The list was then narrowed down based on whether
or not there were likely to be enough evaluations for a meta- analysis. The search
universe includes search aggregators, such as Google Scholar and EBSCO, but also
PAL, IPA, CEGA, and 3ie online databases.
includes the J-
Figure 10.3a shows the number of evaluations per year, and figure 10.3b
shows how the evaluations are distributed over time among RCTs in eco-
RCTs. Both fig-
nomics, RCTs in other fields (e.g., medical trials), and non-
ures show a clear trend in both the number and the fraction of RCTs among
the impact evaluations that are surveyed.
The Influence of Randomized Controlled Trials 443
A
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
1982
1989
1990
1991
1992
1993
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Percent RCTs in Econ Percent RCTS in Other Fields Non-RCTs
B
25
20
15
10
5
0
1980 1985 1990 1995 2000 2005 2010 2015 2020
Total Number of EvaluaƟons Total # of Econ EvaluaƟons
Total # of Other EvaluaƟons
Figure 10.3
Aidgrade.org evaluations
Source: Aidgrade.org.
RCTs are particularly popular among younger researchers. Figures 10.4
and 10.5 show the number and the fraction of researchers who carry out
RCTs among the fellows and associates of the Bureau for Research and Eco-
nomic Analysis of Development (BREAD), the association of development
economists, by the year in which they obtained their PhDs. The number
clearly increases among the recent PhDs, and although this is in part driven
444 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
1980 or earlier 1981-1990 1991-2000 2001-2005 2006-today
PhD Year
Figure 10.4
Fraction of BREAD affiliates and fellows with one or more RCTs
Source: Aidgrade.org.
70%
62%
60%
50%
40%
29%
30%
20%
10%
0%
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
Year
Figure 10.5
Percentage of BREAD conference papers using an RCT
Source: Aidgrade.org.
by a larger number of recent fellows and associates, the fraction of them
who conduct RCTs increases as well.
The number of RCTs presented at development economics conferences
grew rapidly until 2010 and then stabilized (or decreased) after that. At the
annual conference of BREAD (the flagship conference in development eco-
nomics), the fraction of papers featuring RCTs increased from 8 percent in
The Influence of Randomized Controlled Trials 445
Table 10.1
North East Universities Development Consortium conference papers
Year Total number of RCTs Share of RCTs (percent)
2015 40 18.20
2014 36 17.90
2013 49 24.30
2012 27 16.00
Source: Data from neudc.org.
50 percent after that
2005 to 63 percent in 2010, and hovered around 40–
(except for the last conference, at Georgetown, where it was 28 percent). At
the North East Universities Development Consortium Conference, a larger
conference attended by many junior researchers, the fraction of RCTs has
been fairly stable, ranging between 16 and 24 percent for the years 2012 to
2015 (the years for which we could get the papers) and showing no particu-
lar trend (table 10.1).
RCTs have made a clear entry in top academic journals. Looking at the
American Economic Review (AER), the Quarterly Journal of Economics (QJE),
Econometrica, Review of Economic Studies, and the Journal of Political Economy
(JPE), the number of RCT studies was 0 in 1990, 0 in 2000, and 10 in 2015
(table 10.2). At the same time, the number of development papers pub-
lished in these journals almost doubled (from 17 in 1990 to 32 in 2015).
Table 10.2 also provides the details by journal. This is not driven by any
particular journal (except that Econometrica does not seem to contribute
much). Note that this does not mean that RCT studies have supplanted
other types of work: Nearly all published work on development is still non-
ranked journals), and even in top journals, the
RCT (if we look at lower-
experiments have been in addition to the (limited number of) papers that
were published on development.
Beyond the growth in the number of experiments and in the number of
researchers who carry them out, what also stands out is the range and the
ambition of the projects that are attempted: Few topics seem off limits, and
scale does not seem to be a barrier.
Researchers work directly with governments to randomize aspects of
their work. Finan, Olken, and Pande (2015) describe several of these ambi-
tious experiments. For example, Dal Bó, Finan, and Rossi (2013) randomize
the wages at which new government employees are hired; Khan, Khwaja,
and Olken (2016) randomize incentives for tax collectors in Pakistan; and
446 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
Table 10.2
Papers in top journals
Number of Number of
Total number development which are
Journal Year of papers papers RCTs
American Economic 2015 101 15 4
Review 2000 48 6 0
1990 57 2 0
Quarterly Journal of 2015 40 1 1
Economics 2000 43 5 0
1990 52 3 0
Journal of Political 2015 36 4 3
Economy 2000 51 7 0
1990 65 9 0
Restud 2015 48 7 2
2000 36 3 0
1990 40 1 0
Econometrica 2015 46 5 0
2000 37 0 0
1990 64 2 0
Total 2015 271 32 10
2000 215 21 0
1990 278 17 0
Source: Data from neudc.org.
Ashraf, Bandiera, and Lee (2015) work on how government health workers
are recruited for their jobs. In experiments covering several districts and
millions of workers, Muralidharan, Niehaus, and Sukhtankar (2016) and
Banerjee et al. (2016) evaluate two separate process changes in the pay-
ment of wages of India’s major workfare program the Mahatma Gandhi
National Rural Employment Guarantee Act (MGNREGS), while Banerjee et
al. (2014) randomize reforms in the police department in India, and Duflo
et al. (2013a, 2013b) randomize the enforcement of pollution regulation on
industrial firms in India.
Researchers work at a scale that is sufficient to capture market equilib-
rium effects: Muralidharan and Sundararaman (2015) randomize a private
school voucher at the school market level, while Muralidharan, Niehaus,
and Sukhtankar (2016), in their aforementioned experiment, are able to
look at the impact of MGNREGS on wages and productivity.
The Influence of Randomized Controlled Trials 447
The range of topics keeps expanding. Development economists study
alcohol addiction (Schilbach 2015), electoral fraud in Afghanistan (Callen
combatants (Blatt-
and Long 2015), Cognitive Behavioral Therapy for ex-
man, Jamison, and Sheridan 2015), and early childhood stimulation and
development (Attanasio et al. 2014).
In summary, randomized experiments have become not so much the
“gold standard” as just a standard tool in the toolbox. Running an experi-
ment is now sufficiently commonplace that by itself, it does not guarantee
that the paper will get into a top journal or even the BREAD conference.
However, researchers from all sorts of perspectives have come to consider
RCTs as a feasible option for answering the questions they are interested
in. This level of comfort is in part due to the growth of several entities that
help researchers with their fieldwork, including by codifying and standard-
izing experimental practices, and training enumerators. The leader for this
is Innovation for Poverty Action, with its vast network of country offices
PAL, CEGA, and the World Bank.
and experienced staff workers, but also J-
There is also more funding available, from USAID (DIV in particular), the
World Bank (SIEF and DIME), DFID, The Bill and Melinda Gates Founda-
tion, The William and Flora Hewlett Foundation, The International Ini-
tiative for Impact Evaluation, in particular and, more recently, the Global
Innovation Fund. But part of it also has to do with the appeal of the tech-
nique. In the next section, we reflect on the influence that RCTs have had
on development economics research and why.
The Influence of RCTs on Development Economics Research
The remarkable growth in the number of RCTs, and more generally in the
importance of empirical development economics as a field, are in themselves
dramatic changes. The type of development research that is carried out today
is significantly different from research conducted even 15 years ago. A reflec-
tion of this fact is that many researchers who were openly skeptical of RCTs, or
simply belonged to an entirely different tradition in development economics
(e.g., Daron Acemoglu, Derek Neal, Martin Ravallion, and Mark Rosenzweig)
have become involved in one or more RCTs in a developing country.
Early discussions of the merits (or lack thereof) of randomization put
a lot of emphasis on its role in the reliable identification of internally
valid causal effects and the external validity of such estimates. We, and
448 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
others, have had these discussions in other places (Heckman 1992; Banerjee
2008; Duflo, Glennester, and Kremer 2007; Banerjee and Duflo 2009; Dea-
ton 2010), and we will not reproduce them here. As we began to argue in
Banerjee and Duflo (2009), we actually think that these discussions some-
what miss the point about why RCTs are really valuable and why they have
become so popular with researchers.
A Greater Focus on Identification across the Board
The original motivation of randomized experiments, starting with Neyman
([1923] 1990; as a theoretical device) and Fisher (1925; who was the first to
propose physically randomizing units), was a focus on the credible iden-
tification of causal effects. As Athey and Imbens (2017, 78) write in their
chapter for The Handbook on Field Experiments:
There is a long tradition viewing randomized experiments as the most credible
of designs to obtain causal inferences. Freedman (2006) writes succinctly “experi-
ments offer more reliable evidence on causation than observational studies.” On
the other hand, some researchers continue to be skeptical about the relative mer-
its of randomized experiments. For example, Deaton (2010) argues that “evidence
from randomized controlled trials can have no special priority. … Randomized
controlled trials cannot automatically trump other evidence, they do not occupy
any special place in some hierarchy of evidence … ” Our views align with that of
Freedman and others, who view randomized experiments as playing a special role
in causal inference. Whenever possible, a randomized experiment is unique in
the control that the researcher has over the assignment mechanism, and by virtue
of this control, selection bias in comparisons between treated and control units
can be eliminated. That does not mean that randomized experiments can answer
all causal questions. There are a number of reasons randomized experiments may
not be suitable to answer particular questions.
For a long time, observational studies and randomized studies progressed
on largely parallel paths: In agricultural science and then biomedical stud-
ies, randomized experiments were quickly accepted, and a vocabulary
and statistical apparatus to think about them were developed. Despite the
adoption of randomized studies in other fields, in the social sciences, most
researchers continued to reason exclusively in terms of observational data.
The main approach was to estimate associations and then to try to assess
the extent to which these associations reflect causality (or to explicitly
give up on causality). Starting with Rubin’s (1974) fundamental contribu-
tion, researchers started to use the experimental analog to reason about
The Influence of Randomized Controlled Trials 449
observational data, which set the stage for thinking about how to analyze
observational data through the lens of the “ideal experiment.”
Through the 1980s and 1990s, motivated by this clear thinking about
causal effects, labor economics and public finance were transformed by
the introduction of new empirical methods for estimating causal effects
in-
(matching, instrumental variables, difference- differences, and regres-
sion discontinuity designs). Development economics also embraced those
methods starting in the 1990s, but unlike in labor economics and public
finance, some researchers also decided that it may be possible to go directly
to the “ideal” experiment or to go back and forth between experimental
and nonexperimental studies. As a result, the two literatures developed in
fertilizing each other.
close relationship, constantly cross-
The nonexperimental literature was completely transformed by the exis-
tence of this large RCT movement. When the gold standard is not just a
twinkle in someone’s eyes but the clear alternative to a particular empirical
strategy and a benchmark for it, researchers feel compelled to think harder
about identification strategies, and to be more inventive and rigorous about
them. As a result, researchers have become increasingly clever at identifying
and using natural experiments, and at the same time, much more cautious
in interpreting the results from them. Not surprisingly, the standards of
the nonexperimental literature have improved tremendously over the past
few decades without necessarily sacrificing their ability to ask broad and
important questions. For example, Alesina, Giuliano, and Nunn (2013) use
run determinants of the social atti-
suitability to the plow to study the long-
tudes toward the role of women; Padró i Miquel, Qian, and Yao (2014) use
in-
a difference- difference strategy to study village democracy; and Banerjee
and Iyer (2005) and Dell (2010) use a spatial discontinuity to look at the
run impact of extractive institutions. In each of these cases, the ques-
long-
tions are approached with the same eye for careful identification as other
more standard program evaluation questions.
Meanwhile, the RCT literature was also influenced by work done in the
nonexperimental literature. The understanding of the power (and limits)
of instrumental variables allowed researchers to move away from the basic
experimental paradigm of the completely randomized experiment with
up and use more complicated strategies, including encour-
perfect follow-
agement designs. Techniques developed in the nonexperimental literature
offered ways to handle situations in the field that are removed from the
450 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
ideal setting of experiments (e.g., imperfect randomization, noncompli-
ance, attrition, spillovers, and contamination). Structural methods were
combined with experiments to estimate counterfactual policies (Todd and
Wolpin 2006; Attanasio, Meghir and Santiago 2012).
More recently, machine learning techniques have also been combined
with experiments to model treatment effect heterogeneity (see Athey and
Imbens 2017 for a recent review of the econometrics of experiments).
Of course, the broadening offered by these new techniques comes at the
cost of making additional assumptions on top of the original experimental
assignment, and those assumptions may or may not be valid. Thus the
identified,
difference in the quality of identification between a very well-
nonexperimental study and a randomized evaluation that ends up facing
lots of constraints in the field or tries to estimate parameters beyond pure
treatment effects is a matter of degree. In this sense, there has been a con-
vergence across the empirical spectrum in terms of the quality of identifica-
tion, mostly because experiments have pulled the remaining study designs
up with them.
Assessing External Validity
In the words of Athey and Imbens (2017, 79): “External validity is con-
cerned with generalizing causal inferences, drawn for a particular popula-
tion and setting, to others, where these alternative settings could involve
different populations, different outcomes, or different contexts.”
The question of the external validity of RCTs is even more hotly debated
than that of their internal validity. This is perhaps because, unlike internal
validity, there is no clear endpoint to the debate: Heterogeneity in treat-
ment effects across different types of individuals could always occur, or
so-
heterogeneity in the effect may result from ever- slightly different treat-
ments. As Banerjee, Chassang, and Snowberg (2016, 25) acknowledge:
“External policy advice is unavoidably subjective. This does not mean that
it needs to be uninformed by experimental evidence, rather, judgment will
unavoidably color it.”
It is worth noting that very little here is specific to RCTs (Banerjee 2008).
The same problem afflicts all empirical analysis with the one exception
of what Heckman (1992) calls the “randomization bias.” “Randomization
bias” refers to the fact that experiments require the consent of both the
subjects and the organization that is carrying out the program, and these
The Influence of Randomized Controlled Trials 451
people may be quite different. Glennerster (2017), in her chapter in the
Handbook of Field Experiments, provides the list of the characteristics of the
ideal partner, and they are clearly not representative of the typical nongov-
ernmental organization (NGO). But it is worth pointing out that any natu-
rally occurring policy that gets evaluated (i.e., not an RCT) is also selected:
The evaluation requires that the policy did take place, and that was presum-
ably because someone thought it was a good idea to try it out.
In general, any study takes place at a particular time and place, and that
affects results. This does not imply that subjective recommendations by
experts, based both on their priors and the results of their experiments,
should not be of some use to policy makers. Most policy makers know how
to combine the data that is presented to them with their own prior knowl-
edge of their settings. From our experience, we have often observed that
when presented with evidence from an RCT on a program of interest, the
immediate reaction of a policy maker is to ask whether an RCT could be
done in their own context.
There is one clear advantage that RCTs do offer for external valid-
ity, although it is not often discussed and has not been systematically
exploited as yet. To assess any external validity issues, it is helpful to have
identified causal studies in multiple settings. These settings should vary
well-
and possibly in
in terms of the distribution of characteristics of the units—
in
terms of the specific nature of the treatments or the treatment rate—
order to assess the credibility of generalizing to other settings. With RCTs,
because we can, in principle, control where and over what sample experi-
ments take place (and not just how to allocate the treatment in a sample),
we can get a handle on how treatment effects might vary by context. By
itself, this is not sufficient to say anything much, if we account for the
infinite unstructured variation in the world. But there are several ways to
make progress.
A first approach is to combine existing evaluations and make assump-
tions about the possible distribution of treatment effects. Rubin (1981) pro-
poses modeling treatment effect heterogeneity as stemming from a normal
specific
distribution: At each site, the causal effect of the treatment is a site-
effect drawn from a normal distribution. The goal is to estimate the mean
and variance of the treatment effect, and the implied specific site effect,
taking into account the fact that we have other effects, too. An interesting
case study is the effect of microfinance programs. Meager (2016) analyzes
452 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
data from seven randomized experiments, including six published in a spe-
cial issue of the American Economic Journal: Applied Economics in 2015. She
finds remarkable consistency in the mean effects across these studies, but
much more heterogeneity in their variance. Of course, to carry out this
exercise properly, we need access to an unselected sample of studies, and
because there is publication bias in economics, the sample of published
studies may not be representative of all studies that exist. This is where
another advantage of RCT kicks in: Because they have a defined beginning
and end, they can in principle be registered. To this end, the American
Economic Association recently created a registry of randomized trials (www
socialscienceregistry
. org), which, as of June 1, listed 699 studies. The hope
.
is that all projects will be registered, preferably before they are launched,
and that results will be clearly linked to the study, so that in the future,
analysts can work from the full universe of studies.
meta-
A second approach is to conceive projects as multisite projects from
the start. One recent example of such an enterprise is the “graduation”
an integrated, multifaceted program with livelihood promotion
approach—
at its core that aims to “graduate” individuals out of extreme poverty and
term, sustainable higher consumption path. BRAC, the world’s
onto a long-
largest nongovernmental organization, has scaled up this program in Ban-
gladesh (Bandiera et al. 2013), and NGOs around the world have engaged
based efforts. Six randomized trials were undertaken
in similar livelihood-
over the same period around the world (in Ethiopia, Ghana, Honduras,
India, Pakistan, and Peru). The teams regularly communicated with one
another and with BRAC to ensure that their local adaptations remained
true to the original program. The results suggest that the integrated mul-
term income, where
tifaceted program was “sufficient” to increase long-
term” is defined as 3 years after the productive asset transfer (Baner-
“long-
jee et al. 2015a). Using an index approach to account for multiple hypoth-
eses testing, positive impacts were found for consumption, income and
revenue, asset wealth, food security, financial inclusion, physical health,
mental health, labor supply, political involvement, and women’s decision-
making after 2 years. After a third year, the results remained the same in
by-
eight of ten outcome categories. There is country- country variation (e.g.,
the program was ineffective in Honduras), and the team is currently work-
analysis to quantify the level of heterogeneity.
ing on a meta-
The Influence of Randomized Controlled Trials 453
One issue is that there is little that the researcher can do ex post to
reliably identify the source of differences in findings across countries. A
third possible approach would be to take guidance from the first few sites
to make a prediction on what the next sites would find. To discipline this
process, researchers would be encouraged to use the results from existing
trials to make some explicit predictions about what they expect to observe
in other samples (or with slightly different treatments). These can serve as
a guide for subsequent trials. This idea is discussed in Banerjee, Chassang,
and Snowberg (2016), who call it “structured speculation.” They propose
the following broad guidelines for structured speculation:
Experimenters should systematically speculate about the external valid-
1.
ity of their findings.
Such speculation should be clearly and cleanly separated from the rest of
2.
the paper, maybe in a section called “speculation.”
3. Speculation should be precise and falsifiable.
Structured speculation has three advantages, according to Banerjee, Chas-
sang, and Snowberg (2016, 27). First, it ensures that the researcher’s specific
knowledge is captured. Second, it creates a clear sense of where else experi-
ments should be run. Third, it creates incentives to design research that has
greater externality. They write:
To address scalability, experimenters may structure local pilot studies for easy
comparison with their main experiments. To identify the right sub- populations
for generalizing to other environments, experimenters can identify ahead of time
the characteristics of groups that can be generalized, and stratify on those. To
extend the results to populations with a different distribution of unobserved char-
acteristics, experimenters may elicit the former using the selective trial techniques
discussed in Chassang, Padró- Miquel, and Snowberg (2012), and run the experi-
i-
ments separately for each of the groups so identified.
As this approach was just proposed recently, there are few examples as
yet. A notable example is Dupas (2014). Dupas (2014) studies the effect
term subsidies on long-
of short- run adoption of new health products and
term subsidies had a significant impact on the adoption
reports that short-
of a more effective and comfortable class of bed nets. The paper then pro-
vides a clear discussion of external validity. It first spells out a simple and
run subsidies
transparent argument relating the effectiveness of short-
to: (1) the speed at which various forms of uncertainty are resolved and
454 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
(2) the timing of user’s costs and benefits. If the uncertainty over benefits is
run subsidies can have a long-
resolved quickly, short- term effect. If uncer-
tainty over benefits is resolved slowly, and adoption costs are incurred early
run subsidies are unlikely to have a long-
on, short- term effect.
Dupas (2014) then answers the question: For what types of health prod-
ucts and contexts would we expect the same results? The paper does so by
classifying potential technologies into three categories based on how short-
time) subsidies would change adoption patterns. Clearly, there
run (or one-
could be such discussions at the ends of all papers, not just ones featuring
RCTs. But because RCTs can be purposefully designed and placed, there is a
up in this case.
higher chance of follow-
Observing the Unobservable
If the main benefit of randomization is not the identification of causal effect,
what is it? And what explains its remarkable success among researchers?
We agree with Athey and Imbens (2017, 78) that “a randomized experi-
ment is unique in the control that the researcher has over the assignment
mechanism,” and we would take the argument one step further: Random-
ization is also unique in the control that the researcher (often) has over
the treatment itself. In observational studies, however beautifully designed,
the researcher is limited to evaluating what has been implemented in the
world. In a randomized experiment, she can manipulate the treatment in
ways that we do not observe in reality. This has a number of advantages.
First, she can innovate (i.e., design new policies or interventions that she
thinks will be effective based on prior knowledge or theory) and test these
innovations, even if no policy maker is thinking about putting them in
practice yet. Development economists have many ideas, often inspired by
what they have read or researched, and many of the randomized experi-
ment projects come out of those ideas: They test in the field an interven-
tion that simply did not exist before (for example, a kilogram of lentil for
parents who vaccinate their kids; stickers to encourage riders to speak up
against a bad driver; free chlorine dispensers).
Second, the researcher can introduce variations that will help her estab-
known Negative
lish facts that could not otherwise be established. The well-
Income Tax (NIT) experiment was designed with precisely that idea in mind:
In general, a raise in wages creates both income and substitution effects that
cannot easily be separated (Heckman 1992), but randomized manipulation
The Influence of Randomized Controlled Trials 455
of the slope and the intercept of a wage schedule makes it possible to estimate
both together. Interestingly, after the initial NIT and the Rand Health Insur-
ance experiment, the tradition of social experiments in the United States, as
Judy Gueron (2017) describes in her chapter in the Handbook of Field Experi-
ments, has mainly been to obtain causal effects of social policies that were
often fairly comprehensive packages. In contrast, development economists
have worked both on evaluations of real policies (e.g., the PROGRESA evalu-
ation, or, more recently, the evaluation of the graduation program) but also
on what Congdon et al. (2017, 394) describe as “mechanism experiments”:
Broadly, a mechanism experiment is an experiment that tests a mechanism— that
is, it tests not the effects of variation in policy parameters themselves, directly,
but the effects of variation in an intermediate link in the causal chain that con-
nects (or is hypothesized to connect) a policy to an outcome. That is, where there
is a specified policy that has candidate mechanisms that affect an outcome of pol-
icy concern, the mechanism experiment tests one or more of those mechanisms.
There can be one or more mechanisms that link the policy to the outcome, which
could operate in parallel (for example when there are multiple potential mediat-
ing channels through which a policy could change outcomes) or sequentially (if
for example some mechanisms affect take- up or implementation fidelity). The
central idea is that the mechanism experiment is intended to be informative
about some policy but does not involve a test of that policy directly.
In other words, mechanism experiments do not confine themselves to test-
ing feasible (or desirable) policies. For example, cars with broken windows
could be put in the street to test the broken window theory. Once we realize
that we are not limited to a set of realistic policy options (though we are
constrained by what is ethically acceptable), this opens up a wide range of
possibilities.
Banerjee and Duflo (2009) discuss some examples of mechanism experi-
ments. One prominent example in development is a project conducted by
Karlan and Zinman (2008) in collaboration with a South African lender that
risk borrowers at high interest rates. The experi-
makes small loans to high-
ment was designed to test the relative weights of ex post repayment bur-
den (including moral hazard) and ex ante adverse selection in loan default.
Potential borrowers with the same observable risk are randomly offered
a high or a low interest rate in an initial letter. Individuals then decide
whether to borrow at the solicitation’s offer rate. Of those who apply at the
higher rate, half are randomly offered a new, lower contract interest rate
when they are actually given the loan, whereas the remaining half continue
456 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
at the offer rate. Individuals did not know ex ante that the contract rate
could differ from the offer rate. The researchers then compared repayment
performance of the loans in all three groups. The comparison of those who
offer interest rate with those who responded to the
responded to the high-
offer interest rate in the population that received the same low contract
low-
rate allows the identification of the adverse selection effect; comparing
those who faced the same offer rate but differing contract rates identifies
the repayment burden effect. The basic idea of varying prices ex post and
ex ante to identify different parameters has since been replicated in several
different studies (e.g., Ashraf, Berry, and Shapiro 2010; Cohen and Dupas
2010). The experimental variation was key here, and not only to avoid bias:
In the world, we are unlikely to observe a large number of people who face
different offer prices but receive the same actual price.
Experiments can also be devised to understand how institutions func-
tion. An example is Bertrand et al. (2007), who set up an experiment to
understand the structure of corruption in the process of obtaining a driving
license in Delhi. They recruited people who were aiming to get a driving
license and set up three groups, one that receives a bonus for obtaining a
driving license quickly, one that gets free driving lessons, and a control
group. They found that those in the “bonus” group got their licenses faster,
but those who received the free driving lessons did not. They also found
that those in the bonus group were more likely to pay an agent to get the
license (who, they conjecture, bribed someone). They also found that the
applicants who hired an agent were less likely to have taken a driving test
before getting a license. Although they did not appear to find that those in
the bonus group who get licenses are systematically less likely to know how
to drive than those in the control group (which would be the litmus test
that corruption does result in an inefficient allocation of driving licenses),
this experiment provides suggestive evidence that corruption in this case
does more than “grease the wheels” of the system.
Such designs do not always directly lead to actionable policy, but they
have allowed us to describe or understand how the world works. For exam-
ple, in the seminal Bertrand and Mullainathan (2004) study, researchers
sent resumes to prospective employers. The resumes are paired, such that
there are identical resumes, except for the name of the job applicants, who
can either be white sounding or African American sounding. They find that
“applicants” with black sounding names are half as likely to be called back
The Influence of Randomized Controlled Trials 457
as those with white sounding names. Furthermore, being highly educated
does not help, which suggests that something other than statistical dis-
crimination is at play. This design has been replicated hundreds of times in
different settings, providing extensive evidence of discrimination against
different people and in different markets. This large body of evidence does
not necessarily point to a specific solution to this problem, or even help
determine the root of this behavior, but, unlike the previous literature, it
provides clear evidence that the phenomenon exists.
Data Collection
Experiments have also spurred creativity in measurement. In principle,
there is no automatic link between careful and innovative collection of
microeconomic data and the experimental method. And, indeed, it is a
long tradition in development economics to collect data that is specifically
designed to test theories: Both the breadth and the quantity of microeco-
nomic data collected in development economics has exploded in recent
decades, and not only in the context of experiments (see Udry 1995 for a
prominent early example).
However, one specific feature of experiments that serves to encourage
up rates and a
the development of new measurement methods is high take-
specific measurement problem. In many experimental studies, a large frac-
tion of those who are intended to be affected by the program are actually
affected. Thus, the number of units on which data needs to be collected to
assess the impact of the program does not have to be very large, and the
data are typically collected especially for the purpose of the experiment.
Elaborate and expensive measurement of outcomes is therefore easier to
obtain than in the context of a large multipurpose household or firm sur-
vey. By contrast, observational studies must often rely on variation for iden-
induced variation, natural variation,
tification (e.g., policy changes, market-
and supply shocks) that cover large populations, requiring the use of a
large data set often not collected for a specific purpose. This makes it more
difficult to fine tune the measurement to the specific question at hand.
Moreover, even if it is possible ex post to do a sophisticated data collection
exercise specifically targeted to the question, it is generally impossible to
do it for the preprogram situation. This precludes the use of a difference-
differences strategy for these types of outcomes, which again limits the
in-
incentives to collect them ex post.
458 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
Some of the most exciting recent developments in empirical develop-
ment economics have to do with measurement. Researchers have turned to
other subfields of economics, as well as entirely different fields, to borrow
tools for measuring outcomes. Examples include soil testing and remote
sensing in agriculture (see de Janvry, Sadoulet, and Suri 2017 for a review of
to-
agriculture); techniques developed by social psychologists for difficult-
measure outcomes, such as audit and correspondence studies, implicit asso-
ciation tests, Goldberg experiments, and List experiments (see Bertrand and
Duflo 2016 for a review of their use to measure discrimination); tools devel-
oped by cognitive psychologists for child development (Attanasio et al.
DeGroot-
2014); tools inspired by economic theory, such as Becker- Marshak
games to infer willingness to pay (see a discussion in Dupas and Miguel
(2017)); biomarkers in health, beyond the traditional height, weight, and
hemoglobin (cortisol to measure stress, for example); and wearable devices
to measure mobility or effort (Kreindler 2018; Rao, Schilbach, and Scho-
field n.d.).
Specific methods and devices that exactly suit the purpose at hand have
also been developed for experiments. Olken (2007) is one example of the
kind of data that can be collected in an experimental setting. The objective
was to determine whether audits or community monitoring were effective
ways to curb corruption in decentralized construction projects. Getting a
reliable measure of actual levels of corruption was thus necessary. Olken
focused on roads and had engineers dig holes in the road to measure the
material used. He then compared that with the level of material reported
to be used. The difference is a measure of how much of the material was
stolen or never purchased but invoiced, and thus is an objective measure of
corruption. Olken then demonstrated that this measure of “missing inputs”
is affected by the threat of audits, but not, except in some circumstances,
by encouraging greater attendance at community meetings. Rigol, Hus-
sam, and Regianni (n.d.) provide another example of clever data collection
methods. For their experiment, they designed soap dispensers that could
track when the pump was being pushed in order to accurately measure
whether and when people wash their hands and hired a Chinese company
to manufacture the dispensers. Similar “audit” methodologies are used to
measure the impact of interventions in health, such as patients posing with
specific diseases to measure the impact of training (Banerjee et al. 2016) or
The Influence of Randomized Controlled Trials 459
Ross et al. 2017).
ineligible people attempting to buy free bed nets (Dizon-
Even a partial list of such examples would be very long.
In parallel, greater use is being made of administrative data, which are
scale experiments. For example, Banerjee et al.
often combined with large-
(2016) make use of both publicly available administrative data on a workfare
program in India and restricted expenditure data made available to them as
part of the experiment; Khan, Khwaja, and Olken (2016) use administrative
tax data; and Attanasio et al. (2017) use unemployment insurance data to
term effect of job training in Colombia.
measure the long-
The bottom line is that great progress has been made in our understand-
ing of how to creatively and accurately collect or use existing data that go
beyond the traditional survey, and these insights have led both to better
projects and to innovations in data collection that have been adopted in
nonrandomized work as well.
Iterate and Build on Previous Research in the Same Settings
The next methodological advantage of RCTs also relates to the control
that researchers have over the assignment and, often enough, over the
identified policy evaluations often leave us
treatments themselves. Well-
with many questions about why things turned out the way they did. For
example, some papers using regression discontinuity designs find that the
impact of “elite” schools on the marginal child who is admitted tends to
be very low. These results seem to hold both in rich and in poor coun-
tries (Clark 2009; Abdulkadiroglu, Angrist, and Pathak 2014; Dobbie and
Fryer 2014; Lucas and Mbiti 2014; Dustan, de Janvry, and Sadoulet 2015).
But these results leave some questions pending: Does this mean that the
impact is zero for all students or just the marginal student? Is it because
peers don’t matter and curriculum doesn’t matter, or because they both
matter but cancel out?
Although some progress can be made (e.g., Abdulkadiroglu, Angrist, and
Pathak (2014) exploit the fact that students take two different tests to get
a handle on the impact of the program for different types of students),
one is necessarily limited by the type of policy variation that is actually
available. The result from a single RCT often likewise raises more questions
than it can actually answer. For example, when Duflo, Kremer, and Robin-
son (2008) found that the return to fertilizer appears to be very large, even
460 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
when used by the farmers themselves on their own fields (and not just on
experimental plots), one possible policy response might have been to fol-
low Jeff Sachs’s idea of distributing fertilizer for free. But this was not their
next step. Instead, they started wondering why farmers are not using more
fertilizer. This set them down a path that led them to set up experiments
in the same setting: Some focused on learning and social networks, and
some on the difficulty to save even over short periods of time. This latter
inquiry led them down the path of designing and implementing a specific
product, for which the household was offered the option of buying fertil-
izer in advance (Duflo, Kremer, and Robinson 2008). The social network
interventions found surprisingly little diffusion of agricultural innovation
to immediate friends, and this observation set the experimenters down
another path: How could it be the case, given all we know about how much
people talk about agriculture? To unpack this further, they introduced a
simple device designed to address a problem that they noticed in their first
set of experiments: Households tend to overuse fertilizer (conditional on
maximizing application
using it), relative to what appears to be the profit-
rate. They then set up experiments to study in what conditions this device
does spread, and what this tells us about how farmers decide whether to
talk to and trust one another (Duflo et al. 2017).
Analyzing these results will no doubt spur new questions and experi-
ments. All empirical science is of course iterative, with studies building on
each other. But the ability to work in the same setting, with the same out-
come and measurement, is extremely precious and is not available outside
a controlled setting.
Unpacking the Interventions
Finally, RCTs, allow the possibility to “unpack” a program to its constitu-
ent elements. Here again, the work may be iterative. For example, all the
initial evaluations of the BRAC ultra poor program were done using their
“full package,” as were a large number of evaluations of the Mexican con-
ditional cash transfer (CCT) program PROGRESA. But both for research and
for policy, once we know that the full program works, it is clearly of interest
to know why it works. In recent years, some papers have looked “inside”
CCT, relaxing the conditionality, for example. Some work has been con-
ducted on the role and the type of conditionality (see Baird, McIntosh, and
The Influence of Randomized Controlled Trials 461
Özler 2011; Bursztyn and Coffman 2012; and Benhassine et al. 2015 for
examples), followed by many papers experimentally varying other features
(we return to the impact of this work below).
Similarly, the early results of the evaluation of the ultra poor program
have set the stage both for a more theoretically grounded understanding
of exactly which market failures led to a poverty trap, as well as for a more
practically grounded understanding of whether all the interventions were
truly necessary or if certain components could be removed. In the event
that some components are unnecessary, costs could be lowered consider-
ably, allowing the program to reach more people using the same budget.
540) discuss how one could go from the ini-
Hanna and Karlan (2017, 539–
tial “full package” evaluation to this greater understanding:
The ideal method, if unconstrained by budget and organizational constraints, is a
complex experimental design that randomizes all permutations of each component.
The productive asset transfer, if the only issue were a credit market failure,
may have been sufficient to generate these results, and if no other component
enabled an individual to accumulate sufficient capital to acquire the asset, the
transfer alone may have been a necessary component. The savings component on
the other hand may have been a substitute for the productive asset transfer, by
lowering transaction costs to save and serving as a behavioral intervention which
facilitated staying on task to accumulate savings. Clearly it is not realistic in one
setting to test the necessity or sufficiency of each component, and interaction
across components: Even if treated simplistically with each component either
present or not, this would imply 2x2x2x2 = 16 experimental groups.
Several studies have tackled pieces of the puzzle, and more are underway
(see the review in Hanna and Karlan 2017). The way forward is clearly
going to be the development of a mosaic, rather than any one defini-
tive study that both tests each component and also includes sufficient
contextual and market variations that it can help set policy for myriad
countries and populations. More work is needed to tease apart the differ-
ent components: asset transfer (addresses capital market failures), savings
account (lowers savings transaction fee), information (addresses informa-
coaching (addresses behavioral constraints, and perhaps
tion failures), life-
changes expectations and beliefs about possible return on investment),
health services and information (addresses health market failures), con-
based poverty traps), among other
sumption support (addresses nutrition-
possibilities. Furthermore, for several of these questions, there are key,
462 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
coaching can
open issues about how to address them; for example, life-
take on an infinite number of manifestations. Some organizations conduct
coaching through religion, others through interactive problem solv-
life-
ing, and others through psychotherapy approaches (Bolton et al. 2003,
2007; Patel et al. 2010). Much remains to be learned not just about the
coaching components but also about how to make
promise of such life-
them work (if they work at all).
In some settings, particularly when working on a large scale with a gov-
ernment, it is actually possible to experiment from the beginning with vari-
ous versions of a program. This serves two purposes: It gives us a handle
on the theory behind the program; and it has operational value for the
effective combination. Banerjee
government, which can pick the most cost-
et al. (2015b) is an example of this approach. The government of Indone-
sia was interested in reducing corruption in their rice distribution program
(Raskin), which is infamous for reaching few of its intended beneficiaries
and for not always being sold at the right price. They thought that deliv-
ering a card to the beneficiaries with the eligibility information might
ameliorate this problem and lead to greater benefits. Working with the
Government of Indonesia, the authors designed a set of field experiments
to provide information directly to eligible households. In 378 villages (ran-
domly selected from among 572 villages spread over three provinces), the
central government mailed “Raskin identification cards” to eligible house-
holds to inform them of their eligibility and the quantity of rice that they
were entitled to. To unbundle the mechanisms through which different
forms of information may affect program outcomes, the government also
experimentally varied how the card program was run along three key
whether an additional rule (the copay price) was also listed
dimensions—
on the card, whether information about the beneficiaries was also made
very public, and whether cards were sent to all eligible households or only
to a subset of them. The researchers then collected data on eligible and
ineligible rice purchases and prices paid for all villages. On net, they found
that the card did lead to large increases in the amount of subsidies received
by the households. Further, they found that the information on the card
mattered: the price paid was lower when the price was indicated on the
card. They also found that the card was more effective when the informa-
tion was made public. Finally, public information was not sufficient on its
own: The physical card also mattered.
The Influence of Randomized Controlled Trials 463
Knowing all of this is important for understanding the mechanisms at
play. It was also immediately actionable for the government, which pro-
ceeded to scale up the program and to provide cards with price information
to all eligible households accompanied by posters. Cards were distributed to
more than 65 million individuals. This is one occasion where the research-
ers’ and the government’s interests were exactly aligned. Is it more gener-
ally true?
Have RCTs Become Too Academic to Lead to Any Real World Changes?
RCTs have changed development economics, but have they also had sig-
nificant influence in the world? If RCTs are pushing forward the frontiers of
academic research by seeking to understand mechanisms and testing ideas
generated by academics themselves, does this make them too academic and
less useful for policy?
In this section, we argue that RCTs can contribute to policy not only
by providing evidence on specific programs that can be scaled but also by
changing the general climate of thinking about an issue. We then exam-
ine a case study of a funder, Development Innovations Ventures at USAID.
Some of the innovations that it has funded were driven by social entrepre-
neurs without researcher involvement and some were tested using RCTs or
had close involvement with development economics researchers. A review
of this portfolio suggests that several programs involving development eco-
world influence.
nomics researchers and RCTs had substantial real-
Are RCTs That Are More “Academic” Less Useful for Policy?
Many studies seek not just to test a particular program but also to contrib-
ute to a body of literature that seeks to test different theories of human
behavior. If citizens vote for candidates based on their ethnicity or caste,
is that because of very strong preferences, clientelistic networks, or a com-
binations of weak preferences and no alternative information on candi-
date quality? Do people only value what they pay for? How important are
liquidity constraints, as opposed to lack of information or low human capi-
tal, in explaining poor child health and low business profitability in low-
income families?
The studies that seek to answer these questions do not always test stan-
dard development programs, although some may become development
464 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
ideas. De Mel, McKenzie, and Woodruff (2012) gave cash to businesses in Sri
Lanka without conditions, repayment requirements, or mentoring, some-
thing unheard of in finance programs at the time (of course, eventually, the
idea of unconditional cash transfers caught on as a realistic policy option,
as indicated by the success of GiveDirectly). As we have discussed above, a
series of studies that focused on pricing of health goods first asked house-
holds whether they were willing to purchase a good at one price and then
gave them the good at a lower price or for free, not something a regular
program would do. Researchers pushed to test unconditional cash transfers
(Baird, McIntosh, and Özler 2011; Haushofer and Shapiro 2013; Benhassine
et al. 2015; Blattman, Fiala, and Martinez 2014), even though at the time,
the political consensus favored conditional transfers.
The reason this is potentially important for policy, and not just for aca-
demic curiosity, is that even where certain program specifics do not gener-
alize, underlying patterns in human behavior may. The finding that small
incentives are effective in encouraging people to take actions that have
run costs but long-
short- run benefits is more likely to generalize than the
finding that lentils are a successful incentive for vaccination in Rajasthan
(Banerjee et al. 2010). Kremer and Glennerster (2011) review more than
seventy health economics RCTs and find strong similarities in consumer
behavior across countries and products, including sharp reductions in
up of nonacute care health products with small increases in price, big
take-
up of nonacute products with small incentives (negative
increases in take-
prices), and no evidence that paying for something makes people more
likely to use it (Kremer and Miguel 2007; Ashraf, Berry, and Shapiro 2010;
Cohen and Dupas 2010; Dupas 2014a).
This body of work on prices was taken up by advocates of free distri-
bution of insecticide treated bednets (ITNs). For many years, there had
been a fierce debate on the merits of free distribution, with free distribu-
tion advocates arguing that even small prices deter the poor, while others
argued that small copayments were important to ensure ITNs were utilized.
Armed with the evidence from RCTs, advocates of mass free distribution
have successfully pushed this approach, resulting in a dramatic rise in ITN
coverage across Africa from roughly 2009 to 2015. The World Health Orga-
three of forty-
nization reports that forty- Saharan
seven countries in sub-
Africa with ITN distribution programs provide them for free (World Malaria
Report, World Health Organization 2015). A recent article in Nature (Bhatt
The Influence of Randomized Controlled Trials 465
Saharan
et al. 2015) examines the sharp decline in malaria infections in sub-
Africa and estimates that between 2000 and 2015, malaria interventions
prevented 663 million malaria cases, most of which is attributable to the
sharp rise in ITN coverage: 450 million cases of malaria and roughly 4 mil-
lion deaths were prevented by ITNs from 2000 to 2015.
Beyond the specific example of malaria, the policy community is com-
ing to a more general realization that higher prices for preventive health
up and that price elasticity of demand
products can sharply decrease take-
can be very high (Kremer and Holla 2009; Kremer and Glennerster 2011;
Dupas 2014b). These results are changing the entire approach to pricing of
these products.
Another area where a body of evidence from RCTs has produced both
specific policy changes and given rise to more general lessons that have
profoundly changed the policy debate is on attitudes toward cash trans-
fer programs. Arguably the biggest innovation in antipoverty and social
protection policies in developing countries over the past 20 years is the
growth of conditional cash transfer programs (CCTs). Beginning in Mex-
ico, these programs have now spread to more than thirty countries, and
they have arguably played an important role in the decline in poverty in
Osorio et al. 2011; Alzúa, Cru-
Latin America (Attanasio et al. 2005; Barrera-
ces, and Ripani 2013; Galiani and McEwan 2013). Although many factors
were at play in the spread of CCTs, we and many others think that the
PROGRESA experiment (Gertler 2004; Schultz 2004) and the many subse-
quent experiments in other contexts2 played a significant role. These pro-
grams influenced Mexico’s decision to continue and expand CCTs after the
inauguration of a new administration, the active promotion of CCTs by the
American Development Bank and the World Bank, and the adoption
Inter-
of CCTs by many countries.
More recently, additional examination of how CCTs work is further
changing the policy debate. CCTs have been shown by RCTs to not
only increase the behavior on which the cash is conditional but to also
improve outcomes, such as height, weight, and cognitive development
(Barham, Macours, and Maluccio 2013) and reduce HIV infection (Baird,
2. See Glewwe and Olinto (2004), Maluccio and Flores (2005), Galiani and McEwan
(2013), World Bank (2013), Benhassine et al. (2015), among others, as well as the
review in Fiszbein and Schady (2009).
466 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
McIntosh, and Özler 2011). No evidence indicates that poor households
spend increased cash on alcohol or other temptation goods (Haushofer
and Shapiro 2013; Masterson and Lehmann 2014; Evans and Popova
2014). Indeed, the evidence suggests that the income elasticity of demand
for food out of cash transfers is surprisingly high (see a review in Banerjee
2016), and food transfers do not improve nutrition more than cash trans-
fers (Cunha 2014).
This evidence is causing a movement from a situation in which policy
makers would almost never consider cash transfers to one in which cash
transfers, conditional or not, are becoming an accepted tool in develop-
ment policy. For example, as the world struggles to cope with refugees from
war, groups such as the International Rescue Committee have drawn on
RCTs of cash distributions in stable environments and with refugees (Mas-
kind
terson and Lehmann 2014) to strongly push for cash rather than in-
support for refugees. In an IRC press release, David Miliband, IRC president
and CEO, said:
The spate of man- made and natural disasters enveloping innocent civilians
raises profound questions not just for international politics, but for NGOs and
the humanitarian sector, as well. If we keep doing “business as usual,” the gap
between need and provision will continue to grow. Cash distribution— alongside
clear humanitarian “floor” targets in the revised Millennium Development Goals,
more sustainable local partnerships and better use of evidence overall— could be
part of a vital renewal of the humanitarian sector.
Early in the introduction of RCTs, Lant Pritchett (2002) argued that RCTs
would never become particularly popular with policy makers, because they
have reason to prefer ignorance over rigorous knowledge to continue favor-
ing their preferred program: “It pays to be ignorant.” Although in some
cases policy makers may have incentives to preserve ignorance, in others
they are aware of the holes in their knowledge and would like to learn more.
They may have a strong attachment to a favorite program, either due to
inertia or a political imperative. But the experience of running the program
often persuades them that they could do it better, and they are surprisingly
open to ideas about how to improve their programs. The Raskin and MGN-
REGS programs mentioned above, where several teams of researchers have
worked with the government, are good examples: although it was clear that
the programs would continue, finding ways to make them work better was
of interest.
The Influence of Randomized Controlled Trials 467
How to Assess the Policy Success (or Not) of the RCT Agenda
It is somewhat difficult to assess the causal effect of RCTs on policy adop-
tion. Interventions subject to RCTs are not themselves randomized, and
many factors influence whether and when a particular intervention is
adopted. When a program is taken up after an RCT showed it has worked,
it is not always because of the RCT, and it is never just because of the RCT.
Nevertheless, some have argued that the influence of RCTs on policy is
actually quite low, compared to the volume of RCTs. For example, Shah et
al. (2015) point out that despite the 489 completed evaluations by J-PAL
up or policy influence sto-
affiliated researchers, there were only nine scale-
PAL’s website at the time. But this number per se is not particularly
ries on J-
informative: for example, it is not a census of the studies that have some
impact. Not all RCTs conducted by J-PAL affiliated researchers are system-
atically followed up. These stories are chosen precisely because of the size
of their impact and because they can be documented clearly. The absolute
the J-
number of lives reached by them is quite significant— PAL website
tells us that more than 400 million people were reached by these programs.
But the main concerns with any statistic like this are conceptual:
The J-
1. PAL website does not carry statistics on studies conducted by
researchers outside the J-PAL network for the very good reason that,
PAL, it
based on our experience collecting information from DIV and J-
is far from straightforward to collect information on the extent to which
RCTs have influenced policy. For example, the number does not include
the hundreds of millions of people who have been reached by CCTs.
Many RCTs are fairly recent. Taking these to the policy level requires a
2.
lot of care, especially given the external validity issues. (Would it work
in government? Would it work in a different place?) The process is there-
fore often slow, again for good reasons. Therefore, we should not expect
a lot of these to be scaled as yet.
Many of the most valuable RCTs are those that test popular and highly
3.
scale and show
touted policies that already exist in the world on a large-
that they are in fact much less effective than previously claimed or
stoves are two obvious exam-
believed. Microfinance and improved cook-
ples. In such cases, success would be to slow down the spread of such
policies. In such cases, one would not expect something to appear on the
PAL scale-
J- up page, but these are two cases where the work has probably
been quite influential.
468 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
In some cases, the primary purpose of an RCT is not to directly affect
4.
policy, but instead to investigate an underlying theoretical mechanism,
which may, in turn, indirectly influence policy. However, such cases
ups, even though the knowledge
would not appear on a list of scale-
they have provided has impacted, albeit indirectly, a large number of
people. For example, the orthodoxy in development economics had
long been that the poor are “poor but rational.” The accumulating evi-
dence from RCTs has undoubtedly hastened the diffusion of the idea
into development economics and development policy that poor people
are not always rational. This idea is reflected for example, in both the
content and the number of RCTs in the World Development Report 2016
(World Bank 2016) on psychology and poverty. In turn, publications like
this and the associated discussions influence the design of policies.
It is not clear what the right benchmark for success should be. We sus-
5.
pect that if one looked at other areas of economics, one would find that
research projects influenced policy at a much lower rate than RCTs have
in development policy in recent years. Moreover, one would not want
to say that rapid policy influence is the sole or even the major metric
think of
by which the worth of economic research should be assessed—
the idea of congestion pricing for road use (Vickrey 1969), which is only
beginning to find real world applications.
Perhaps most importantly, it is worth realizing that the payoff to RCTs is
6.
likely to be the average of a highly skewed distribution. Looking at the
fraction of RCTs that scale, rather than the average payoff, is therefore as
misleading as looking at the fraction of any research and development
effort that succeeds in terms of, say, generating a successful marketed
product, because the payoff to research and development in general is
typically very highly skewed. As is well known, citations across scien-
tific disciplines appear to follow a power law distribution, with a small
fraction of papers accounting for the majority of citations. This peak is
followed by a steep decay, as a large portion of research papers are never
cited (Radicchi, Fortunato, and Castellano 2008).3 As we mentioned, the
3. For instance, in the social sciences in general, papers receive on average 0.5 cita-
citations (Klamer and Dalen
tions in the first 2 years after publication, including self-
2002), whereas in mathematics, medicine, and education, the number is estimated to
be less than 1 (Mansilla et al. 2007). The skewed distribution implies that the median
The Influence of Randomized Controlled Trials 469
nine policy innovations that were listed on the J-PAL website in 2015
reached more than 200 million people, and this did not include the more
than 100 million people who have been reached through India’s most
recent round of deworming, the millions of people who have received
PAL lists it as policy influence but does not provide
free bed nets (since J-
a count), and the 60 million people whose water and air is less polluted
because of the statewide adoption of better regulation of industrial pol-
lution in Gujarat (again, not counted).
For this reason, pointing out that many R&D efforts yield low payoffs
7.
does not suggest that these are bad investments ex ante. The correct
analytical question to ask is whether the expected average or marginal
payoff to R&D effort in RCTs is positive or greater than that in other
areas of research if one takes overall research budgets as fixed. Of course,
measuring the payoff to research is inherently a difficult exercise for all
sorts of conceptual reasons. There is also the added statistical difficulty
that a large amount of data is needed to accurately measure the mean of
tailed distribution.
a fat-
What Have We Learned from the DIV Experience?
Keeping all of this in mind, we now turn to one particular example, the
experience of the investments made by USAID’s DIV between 2010 and
2012.
round grant competition for innovative solutions to
DIV holds a year-
a range of development challenges, pilots and tests them using analyti-
cal methods, and scales solutions that demonstrate widespread impact and
effectiveness. DIV supports novel business or organizational models;
cost-
operational, behavioral or production processes; and products or services
funding model
that can help address development challenges. DIV’s tiered-
sized
provides small grants to pilot innovations in development; medium-
effectiveness (often using RCTs)
grants to rigorously test for impact and cost-
scale grants to help transition
or ability to pass a market test; and larger-
innovations to scale that have passed a market test or that have rigorous
evidence of impact and cost effectiveness.
paper is never cited. Similarly, most new patents have extremely low value with a
small fraction of patents accounting for much of the overall value of patents.
470 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
When DIV was established, two targets were set for the program: (1) a
15 percent social rate of return on investment, and (2) a reach of at least
75 million people worldwide, through direct investment and through
broader influence on the rest of USAID. Preliminary work by DIV staff sug-
2012 portfolio easily met the first goal, even under the
gests that the 2010–
conservative assumptions that all innovations supported by DIV yielded
no further benefits, and even looking at only a subset of innovations that
yielded financial benefits or health benefits that could be valued in terms
of DALYs. Although social return is a more conceptually comprehensive
measure for evaluating DIV, it is difficult to measure. By considering social
returns we do not seek to evaluate DIV, but rather to look at the narrower
question of whether RCTs can have real world influence. We therefore focus
on examining the number of people reached by innovations supported by
DIV (as well as by later adapted versions of these innovations). (Note that
substantial reach is a necessary but not sufficient condition for high social
return because the total social benefit of an innovation equals the net ben-
efit per person reached times the number of people reached.) This exercise
is inherently limited, so readers will have to make their own judgements
about the likely impact per person reached, the likely future reach of these
innovations (sustainability), and the extent to which DIV funding played
an important role in the reach achieved by innovations in the DIV port-
folio. What we are doing here is rather the descriptive exercise of system-
2012
atically tracking a portfolio. Nevertheless, following the entire 2010–
DIV portfolio is interesting for a paper that explores the influence of RCTs,
because the premise of DIV is specifically to fund innovations in develop-
effectively reach a large number of
ment that have the potential to cost-
people through either the public or the private sector.
down approach
In particular, whereas many other programs have a top-
in which program staff identify problems in advance, choose sectors
up
on which to focus, or set strategy in sectors, DIV follows a bottom-
approach that is deliberately open across sectors: supporting innovations
that will scale commercially, innovations designed to scale through the
public sector, and startups and organizations proposing to change behav-
ior within existing large organizations. Although the bulk of DIV’s out-
reach effort has been oriented toward traditional social entrepreneurs,
DIV has also made an effort to be open to proposals from development
The Influence of Randomized Controlled Trials 471
economics researchers. To balance this openness, DIV employs a staged
scale support
finance approach in which innovations only receive larger-
scale support
after they have passed rigorous tests. DIV provides large-
(stage 3) only for innovations that have rigorous evidence of impact and
cost effectiveness or have demonstrated market viability. At the piloting
(stage 1) and testing stages (stage 2), however, DIV has historically been
open to proposals that have the potential to scale based on their cost-
effectiveness, for example, even if they do not necessarily already have a
management team in place capable of scaling internally or written com-
mitments from scaling partners.4
This combination of approaches thus helps us ask whether the engage-
ment with the development economics research community, and the will-
stage investments even without a fully proven
ingness to consider early-
capacity to scale, came at the cost of scaling success. We can shed light on
these questions by comparing the scaling record across types of projects,
stages of funding, and of course by looking at the scaling record of DIV.
In the online Appendices, we provide a list of all the DIV awards from
this period and a description of the innovations that have, subsequent to
DIV’s funding, reached more than 100,000 people. Table 10. 3 shows the
results of this exercise.
Here are some key insights:
DIV has been relatively successful in supporting innovations that scale.
1.
A relatively high fraction of DIV awards, and an even higher fraction of
DIV total investment, has gone to projects that have already reached
more than 100,000 people (and a smaller but still high fraction of the
awards went to projects that reached more than a million people). Thirty
percent of DIV awards (13/43) have so far reached more than 100,000
5 years.5 These awards account for 57 percent of the
people within 3–
total value of DIV awards in this period, or $10.98 million in total fund-
ing. Fourteen percent of DIV awards (5/43) have so far reached more
4. Although DIV does not require a proven pathway to scale at stages 1 or 2, a promis-
ing pathway to scale through the public or private sector (or a hybrid of the two) and
strong potential demand is one of its main selection criteria, particularly at stage 2.
5. Two innovations (that reached over 100,000 people) received both a stage 1 and a
stage 2 award. Thus, these twelve awards support ten separate innovations.
472 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
Table 10.3
Future reach of DIV projects, by award type
Fraction Fraction
Reaching Reaching DIV
Number Total more than more than Expenditure
of Awarded 100,000 1,000,000 People per Person
Award Stage Awards Value people people Reached* Reached
Stage 1 23 $2,353,136 17% (4/24) 8% (2/24) 6,723,733 $0.35
(<$100,000)
Stage 2 19 $9,557,926 44% (8/18) 11% (2/18) 16,931,044 $0.56
(<$100,000,000)
Stage 3 1 $5,516,606 100% (1/1) 100% (1/1) 1,750,000 $3.15
(<$15 million)
*Two innovations (Voter Information Report Cards and CommCare) that reached more than
100,000 people received both stage 1 and stage 2 awards. In both cases, people reached by
those innovations are counted as people reached by stage 2 awards.
than 1 million people. These awards account for 33 percent of the total
value of DIV awards in this period, or $6.38 million in total funding.
Why do we say that 30 percent is “relatively successful”? A rule of
thumb in the venture capital world is that 10 percent of investments
yield modest success, and 1 percent yield large successes. Although we
have not yet identified other funders that publish data that would allow
for computation of comparable statistics, our reading of the literature
and our examination of websites of some other organizations suggests
that these rates compare well with those achieved over a much longer
investing organizations. These results are all
time frame by other impact-
the more striking because, although some organizations provide funding
only after a certain level of scale is reached (e.g., Acumen, Skoll Founda-
tion), DIV often supported innovations at an early stage (as well as tests
to know whether they were worth scaling up), rather than waiting until
innovations had already reached a certain scale and had attracted earlier
support before investing.
Stage 1 and stage 2 awards have a particularly low DIV expenditure per
2.
person reached and account for more than 90 percent of people reached
by innovations supported by DIV during this period.
One of these early stage innovations (Consumer Action and Matatu
Safety) recently received a stage 3 DIV award, but in general, stage 1 and
The Influence of Randomized Controlled Trials 473
stage 2 innovations attained high levels of reach because other funders/
entities provided support based in part on the information generated
funded project.
from the DIV-
Although the estimated DIV expenditure per person is lower for earlier
3.
stage grants, it is fairly low across the board. This is because most of the
supported innovations was attained without the applicants
reach of DIV-
returning to DIV for additional financial support.
Though many past awardees apply for additional funding, only 7 per-
2012 portfolio of grantees received follow-
cent of DIV’s 2010– on fund-
ing after the initial period of performance. More than 40 percent of DIV’s
2012 grantees received follow-
2010– on funding from either the public
or private sector after DIV’s investment. DIV’s capacity to be catalytic of
course partly derives from the rich funding ecosystem in which it oper-
ates, where other entities (governments, NGO, private sector firms) can
adopt innovations.
Cost was a key determinant of which innovations scaled. The largest
4.
scale was achieved by innovations with very low costs per person.
In some cases, the innovations involved the provision of informa-
tion by media or phone (including voter report cards, election mon-
itoring), or provided behavioral “nudges” in large, existing systems
(e.g., Zambian community health workers). Of course, it’s important to
recognize that total impact depends on the benefit per person reached
times the number of people reached, and some innovations with mod-
erate cost per person (e.g., Vision Spring) and moderate reach may
generate high total social benefit because the benefit per person is
very high.
Although some innovations reached more than 100,000, or in one case,
5.
more than 1,000,000 people through the creation and growth of a new
organization designed to scale the innovation, the vast majority of reach
was delivered through adoption by existing large organizations, includ-
ing large firms, NGOs, and governments.
supported innovations that reached 100,000 or more
Four of the DIV-
consumers involved the creation of new organizations that scaled from
scratch. Seven involved adoption of the innovation by existing entities
that already had high levels of reach.
Of the six innovations that reached more than one million people,
one was scaled by an NGO that constructed and built operations around
474 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
the innovation (Evidence Action in the case of chlorine dispensers), and
four did so by adoption by existing organizations (an insurance company
and the Kenyan National Transport and Safety Authority in the case of
stickers in matatus, the Government of India in the case of biometric
time efforts to send
monitoring, political campaigns in the case of real-
polling station outcomes to central locations by mobile phones, and news-
papers in the case of voter report cards). Existing organizations with large
supported innovations or modified versions of
reach that adopted DIV-
these innovations included private sector firms, NGOs, and governments.
Innovations tested with RCTs scale not only through adoption by
6.
governments, but also through adoption by private sector firms and
NGOs.
Of the ten DIV awards for innovations with RCTs that have reached
more than 100,000 people, there were two clear cases in which develop-
ing country governments played the lead role (scaling of an improved
approach to community health worker recruitment by the government
of Zambia and biometric monitoring in India). The Kenyan government
seems likely to play an important role alongside the insurance industry
in scaling the Kenyan matatu safety program. Donors played a key role
in provision of Potential Energy’s improved cookstoves in Darfur. NGO
partners played a role in a number of projects. A major lesson of this
analysis is that large private firms played a major role as well (e.g., an
insurance company played a key role in the matatu stickers project and
newspapers published the free content when an NGO provided them
with voter report cards).
Innovations involving RCTs or developed in part by researchers (often
7.
working in close conjunction with implementers), reach 100,000 or
1,000,000 users at a particularly high rate.
three percent (10/23)6 of awards for which an RCT was used
Forty-
for evaluation or development economics researchers were involved in
6. Projects were coded as having development economics researchers involved if the
initial proposal that was funded by DIV explicitly included the efforts of researchers.
Although d light’s initial proposal included an RCT on the impacts of their products,
.
this RCT did not take place and funding strictly supported the development of a
new solar home system as well as an ex post impact evaluation of these systems. Due
to these circumstances, we have not included d light in our calculation of projects
.
The Influence of Randomized Controlled Trials 475
design of the innovation reached more than 100,000 people.7 Twenty-six
percent (6/23) of these awards supported innovations that had reached
more than one million people in the original or adapted form (including
voter report cards, election monitoring, stickers in matatus, chlorine dis-
pensers, and biometric attendance verification). In contrast, among the
innovations not including an RCT component or a strong role for devel-
opment economics researchers, only 16 percent (3/19) reached 100,000
.
people (Vision Spring, Mera Gao, dlight), and none reached more than
one million people.8
One could imagine multiple hypotheses for this difference in the
rates of success. First, it might be easier to reach many people by per-
suading large organizations and governments to adopt the innovation,
and in this process, the evidence from the RCTs might have played an
important role. By contrast, those innovations that did not come from
the academic RCT side tried to scale by directly implementing or selling
their product, which may be harder, as these innovations do not have
large preexisting policies, programs, or institutions as initial partners.
Second, it is often argued that academic researchers mainly want to pub-
lish, and this conflicts with their incentives to get involved in projects
that are socially useful but not as creative (e.g., replication, tinkering
with design). But it is also argued that journals have a strong publica-
tion bias, and it is easier to publish ideas that have worked. Ergo, devel-
opment economists should have strong incentives to develop and test
innovations that have a reasonable chance of success.
developed in part by researchers in this point. If we were to include d light, this fig-
.
ure would be 11/24, or 46 percent.
7. Voter information report cards (two awards), election monitoring technology,
digital attendance and medical information systems in primary health care centers,
mobile tools for community health care workers (two awards), consumer action on
Matatu safety, bringing safe water to scale, improved cookstoves, and recruiting com-
munity health workers.
8. Twenty- four awards incorporated an RCT component or were based on an RCT.
This excludes two cases in which the initial proposal included an RCT but the ulti-
mate actual project funded by DIV did not include an RCT: Psychometric Analysis
for Entrepreneurs (AID- F-
OAA- 13–00028) and Affordable Access to Energy for All:
Innovative Financing for Solar Systems (AID- OAA- 13–
F- 00007). Note that because
there is a lot of overlap between researcher-led projects and projects with an RCT, we
cannot easily separate their impact.
476 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
Moreover, perhaps economics actually gives them some useful
insights into the design of projects. Third, it may also be that the recent
focus on information and behavioral economics makes them particu-
larly interested in innovations with a low cost per user (“nudges”),
which seems to be a strong predictor of success. Fourth, when research-
ers were involved, they were typically not just evaluators: They were
fully involved in the development of the innovation (e.g., voter report
cards, chlorine dispensers, a monitoring project in Afghanistan), worked
closely with implementing organizations, and remained closely involved
in the details of the implementation. They were in fact “researcher-
entrepreneurs.” Many of the ideas developed by researchers drew on the
latest ideas in the field, and the data suggest that the researchers who
developed these ideas were then relatively successful in working with
others to scale these innovations.
Innovations that had already been tested through RCTs and found to
8.
have impact and potential for cost effectiveness prior to applying for
DIV support accounted for three of the five innovations that reached
more than one million people.
Three of the five innovations that reached more than one million
people (voter report cards, Consumer Action and Matatu Safety, and
Chlorine Dispensers for Safe Water) had already been subject to RCTs
before applications were submitted to DIV. Although we have not yet
coded the data, we believe that there were very few applications in this
category, so the rate at which proposals in this category reached more
than one million people was very high (possibly 100 percent).
Although some DIV-
9. supported innovations have been applied in mul-
tiple countries, most have not.
supported innovations have typically not been applied
So far, DIV-
much beyond the country where they have been tested. This may be an
area where future work is needed.
Conclusion
The previous discussion on the role that RCTs play in policy suggests that
RCTs have influenced policy both by providing evidence on individual proj-
ects and programs and by changing thinking in development more broadly.
The Influence of Randomized Controlled Trials 477
The biotech and information technology industries routinely build on
innovations developed by researchers using frontier techniques in those
fields. The evidence from DIV awards is consistent with the idea that a simi-
lar approach may be effective in development, with innovations developed
in part by researchers or involving RCTs reaching 100,000 or 1,000,000
users at a particularly high rate. This is absolutely not to say that work is
not needed to fine tune interventions for different contexts, or that it is not
world programs that have not yet been evaluated
important to evaluate real-
using an RCT. But the development of new ideas that are grounded in basic
life change.
science actually can lead to real-
One striking lesson of this analysis is that the projects that are scaled up
cost, well-
tend to be low- defined, and simple. Other examples, not in this
list, also fit this bill (e.g., deworming, the Raskin card). There are notable
counterexamples of programs that are neither particularly cheap nor simple
and have scaled up: Conditional Cash Transfers and the BRAC ultra poor
programs are two examples. Furthermore, those two programs were not
only scaled up where they had been tested but were also implemented in
many other countries as well. Interestingly, they were initially replicated
as RCTs.
defined interventions are also the ones that are more likely to lead
Well-
to successful research projects because they can more easily pin down a
specific mechanism and be construed as a test for a theory. So the reasons
RCTs have been so successful as a research tool may also be what makes
world changes.
them successful at leading to real-
Looking forward, we don’t know what the most important pathways
of influence for RCTs might turn out to be. One route is that simple, clear
cost interventions, or low-
insights, low- cost modification to promising
existing programs get adopted, as the DIV case study suggests. That these
cost of course does not mean that they have low
innovations are low-
identified development research
impact. One lesson from decades of well-
is that details are incredibly important, and that the distinction between
“big” and “small” questions can be very misleading (see Banerjee and Duflo
2011, chapter 10) for a more detailed discussion).
An alternative pathway is one in which more complex interventions are
replicated in many contexts and then widely adopted, following the PRO-
GRESA or the BRAC model. The third pathway is that rather than focusing
478 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
only on the results, policy makers and other actors adopt the experimental
attitude by allowing for innovations and learning perhaps inside a special-
department fund
ized unit (like the White House “nudge” unit) or a cross-
(like the Tamil Nadu innovation fund).
But to really get the full benefits of the RCT revolution, it is not enough
to do more RCTs and get some of them scaled up. A range of complemen-
tary institutions are also necessary to more effectively translate research
into policy. For example, we need better systems for the production of
meta-analyses and review articles and for the creation of expert panels to
review the evidence. Medicine has a quite involved system for this, but
even setting aside the question of how well that system works in medicine
(Sim et al. 2001; Kawamoto et al. 2005), the institutions that are appropri-
ate for medicine are not necessarily appropriate for social science and devel-
opment economics in particular. These institutions are just starting to be
built: The American Economic Association registry of RCTs is an example
of a successful effort to build a registration platform. Its popularity suggests
that the development community is receptive to these efforts.
In addition to the purely scientific infrastructure for learning, the pro-
cess of going from an idea to a program at scale requires appropriate insti-
tutional support. Funders are needed to finance iterative piloting before an
RCT to work out the implementation details.9 Once an RCT has been con-
ducted, institutional support is also needed for iterating on the interven-
tion to prepare it for transition to scale. This includes testing ways to bring
unit costs down (because the first RCT often evaluates a small pilot with
high unit costs); collaborate with potential implementing partners; and
mitigate potential cost increases or reduced benefits that may result from
up
institutional and personnel differences between the pilot and scaled-
versions of an innovation (due to, for example, government procurement
systems with higher transaction costs or limited government capacity to
up ver-
implement the intervention effectively). To get to the right scaled-
sion therefore involves trying them out at scale and measuring the impact
at scale. Indeed, multiple iterations may be needed until something that
9. Development Innovation Ventures and the Global Innovation Fund— a private
fund modeled after DIV and to which DIV and other bilateral donors and impact
explicitly encompass such a piloting phase.
investors contribute—
The Influence of Randomized Controlled Trials 479
is appropriate for policy can work. Figuring out how best to do the scaling
in each case or how to do so in additional countries takes time, specialized
human capital, and additional funding.
References
Abdulkadiroglu, Atila, Joshua Angrist, and Parag Pathak. 2014. “The Elite Illusion:
Achievement Effects at Boston and New York Exam Schools.” Econometrica 82 (1):
137–196.
Alesina, Alberto, Paola Giuliano, and Nathan Nunn. 2013. “On the Origins of Gender
530.
Roles: Women and the Plough.” Quarterly Journal of Economics 128 (2): 469–
Alzúa, María Laura, Guillermo Cruces, and Laura Ripani. 2013. “Welfare Programs
and Labor Supply in Developing Countries: Experimental Evidence from Latin
1284.
America.” Journal of Population Economics 26 (4): 1255–
Gooders and Go-
Ashraf, Nava, Oriana Bandiera, and Scott S. Lee. 2015. “Do- Getters:
Career Incentives, Selection, and Performance in Public Service Delivery.” Working
Paper, Harvard Business School, Cambridge, MA.
Ashraf, Nava, James Berry, and Jesse M. Shapiro. 2010. “Can Higher Prices Stimu-
late Product Use? Evidence from a Field Experiment in Zambia.” American Economic
2413.
Review 100 (5): 2383–
Athey, Susan, and Guido W. Imbens. 2017. “The Econometrics of Randomized
Experiments.” In Handbook of Field Experiments, volume 1, edited by Abhijit V.
140. Amsterdam: North-
Banerjee and Esther Duflo, 73– Holland.
Attanasio, Orazio P., Costas Meghir, and Ana Santiago. 2012. “Education Choices in
Mexico: Using a Structural Model and a Randomized Experiment to Evaluate PRO-
66.
GRESA.” Review of Economic Studies 79 (1): 37–
Hernandez.
Attanasio, Orazio P., Erich Battistin, Emla Fitzsimons, and Marcos Vera-
2005. “How Effective Are Conditional Cash Transfers? Evidence from Colombia.”
IFS Briefing Note BN54, Institute for Fiscal Studies, London.
Attanasio, Orazio P., Camila Fernández, Emla O. A. Fitzsimons, Sally M. Grantham-
McGregor, Costas Meghir, and Marta Rubio- Codina. 2014. “Using the Infrastructure
of a Conditional Cash Transfer Program to Deliver a Scalable Integrated Early Child
Development Program in Colombia: Cluster Randomized Controlled Trial.” BMJ 349
(September): g5785.
Attanasio, Orazio P., Arlen Guarín, Carlos Medina, and Costas Meghir. 2017. “Voca-
tional Training for Disadvantaged Youth in Colombia: A Long- Term Follow- Up.”
143.
American Economic Journal: Applied Economics 9 (2): 131–
480 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
Baird, Sarah, Craig McIntosh, and Berk Özler. 2011. “Cash or Condition? Evidence
1753.
from a Cash Transfer Experiment.” Quarterly Journal of Economics 126 (4): 1709–
Bandiera, Oriana, Robin Burgess, Narayan Das, Selim Gulesci, Imran Rasul, and
Munshi Sulaiman. 2013. “Can Basic Entrepreneurship Transform the Economic Lives
of the Poor?” IZA Discussion Paper 7386, Institute for the Study of Labor, Bonn.
Banerjee, Abhijit V. 2008. “Big Answers for Big Questions: The Presumption of Mac-
roeconomics.” Paper presented at Brookings Global Economy and Development
Conference: What Works in Development? Thinking Big and Thinking Small, Wash-
ington, DC, May.
Fed World.” Review of World Econom-
Banerjee, Abhijit V. 2016. “Policies for a Better-
17.
ics 152 (1): 3–
Banerjee, Abhijit V., and Esther Duflo. 2009. “The Experimental Approach to Devel-
opment Economics.” Annual Review of Economics 1: 151– 178.
Banerjee, Abhijit V., and Esther Duflo. 2011. “Policies, Politics.” In Poor Economics: A
Radical Rethinking of the Way to Fight Global Poverty, edited by Abhijit V. Banerjee and
Esther Duflo, 235–265. New York: Public Affairs.
Banerjee, Abhijit V., Esther Duflo, Rachel Glennerster, and Dhruva Kothari. 2010.
“Improving Immunization Coverage in Rural India: A Clustered Randomized Con-
trolled Evaluation of Immunization Campaigns with and without Incentives.” Brit-
ish Medical Journal 340: c2220.
Banerjee, Abhijit V., and Lakshmi Iyer. 2005. “History, Institutions, and Economic
Performance: The Legacy of Colonial Land Tenure Systems in India.” American Eco-
1213.
nomic Review 95 (4): 1190–
Banerjee, Abhijit V., Sylvain Chassang, and Erik Snowberg. 2016. “Decision Theo-
retic Approaches to Experiment Design and External Validity.” NBER Working Paper
22167, National Bureau of Economic Research, Cambridge, MA.
Banerjee, Abhijit V., Alice H. Amsden, Robert H. Bates, Jagdish N. Bhagwati, Angus
Deaton, and Nicholas Stern. 2007. Making Aid Work. Cambridge, MA: MIT Press.
Banerjee, Abhijit V., Raghabendra Chattopadhyay, Esther Duflo, Daniel Keniston,
and Nina Singh. 2014. “Improving Police Performance in Rajasthan, India: Experi-
mental Evidence on Incentives, Managerial Autonomy and Training.” NBER Work-
ing Paper 17912, National Bureau of Economic Research, Cambridge, MA.
Banerjee, Abhijit V., Esther Duflo, Nathanael Goldberg, Dean Karlan, Robert Osei,
William Parienté, Jeremy Shapiro, Bram Thuysbaert, and Christopher Udry. 2015a.
“A Multifaceted Program Causes Lasting Progress for the Very Poor: Evidence from
1260799.
Six Countries.” Science 348 (6236): 1260799–
Banerjee, Abhijit V., Rema Hanna, Jordan C. Kyle, Benjamin A. Olken, and Sudarno
Sumarto. 2015b. “The Power of Transparency: Information, Identification Cards and
The Influence of Randomized Controlled Trials 481
Food Subsidy Programs in Indonesia.” NBER Working Paper 20923, National Bureau
of Economic Research, Cambridge, MA.
Banerjee, Abhijit, Esther Duflo, Clément Imbert, Santhosh Mathew, and Rohini
Pande. 2016. “Can e-Governance Reduce Capture of Public Programs? Experimental
Evidence from India’s Employment Guarantee.” Mimeo, MIT, Cambridge, MA.
Barham, Tania, Karen Macours, and John A. Maluccio. 2013. “More Schooling and
Year Conditional Cash Transfer Program in Nica-
More Learning? Effects of a Three-
WP-
ragua after 10 Years.” IDB- American Development Bank, Washing-
432, Inter-
ton, DC.
Osorio, Felipe, Marianne Bertrand, Leigh L. Linden, and Francisco Perez-
Barrera-
Calle. 2011. “Improving the Design of Conditional Transfer Programs: Evidence
from a Randomized Education Experiment in Colombia.” American Economic Journal:
195.
Applied Economics 3 (2): 167–
Benhassine, Najy, Florencia Devoto, Esther Duflo, Pascaline Dupas, and Victor Pou-
liquen. 2015. “Turning a Shove into a Nudge? A ‘Labeled Cash Transfer’ for Educa-
125.
tion.” American Economic Journal: Economic Policy 7 (3): 86–
Bertrand, Marianne, and Esther Duflo. 2016. “Field Experiments on Discrimina-
tion.” NBER Working Paper 22014, National Bureau of Economic Research, Cam-
bridge, MA.
Bertrand, Marianne, and Sendhil Mullainathan. 2004. “Are Emily and Greg More
Employable Than Lakisha and Jamal? A Field Experiment on Labor Market Discrimi-
1013.
nation.” American Economic Review 94 (4): 991–
Bertrand, Marianne, Simeon Djankov, Rema Hanna, and Sendhil Mullainathan.
2007. “Obtaining a Driver’s License in India: An Experimental Approach to Studying
1676.
Corruption.” Quarterly Journal of Economics 122 (4): 1639–
Bhatt, S., D. J. Weiss, E. Cameron, D. Bisanzio, B. Mappin, U. Dalrymple, K. E.
Battle, et al. 2015. “The Effect of Malaria Control on Plasmodium falciparum in Africa
between 2000 and 2015.” Nature 526 (7572): 207– 211.
Blattman, Christopher, Nathan Fiala, and Sebastian Martinez. 2014. “Generating
Employment in Developing Countries: Experimental Evidence from
Skilled Self-
752.
Uganda.” Quarterly Journal of Economics 129 (2): 697–
Blattman, Christopher, Julian C. Jamison, and Margaret Sheridan. 2015. “Reducing
Crime and Violence: Experimental Evidence from Cognitive Behavioral Therapy
in Liberia.” NBER Working Paper 21204, National Bureau of Economic Research,
Cambridge, MA.
Bolton, Paul, Judith Bass, Theresa Betancourt, Liesbeth Speelman, Grace Onyango,
Kathleen F. Clougherty, Richard Neugebauer, Laura Murray, and Helen Verdeli.
2007. “Interventions for Depression Symptoms among Adolescent Survivors of War
482 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
and Displacement in Northern Uganda: A Randomized Controlled Trial.” JAMA 298
(5): 519–527.
Bolton, Paul, Judith Bass, Richard Neugebauer, Helen Verdeli, Kathleen F. Clough-
erty, Priya Wickramaratne, Liesbeth Speelman, Lincoln Ndogoni, and Myrna Weiss-
man. 2003. “Group Interpersonal Psychotherapy for Depression in Rural Uganda: A
Randomized Controlled Trial.” JAMA 289 (23): 3117–3124.
Bursztyn, Leonardo, and Lucas C. Coffman. 2012. “The Schooling Decision: Family
Preferences, Intergenerational Conflict, and Moral Hazard in the Brazilian Favelas.”
397.
Journal of Political Economy 120 (3): 359–
Callen, Michael, and James D. Long. 2015. “Institutional Corruption and Election
Fraud: Evidence from a Field Experiment in Afghanistan.” American Economic Review
381.
105 (1): 354–
Cameron, Drew B., Anjini Mishra, and Annette N. Brown. 2016. “The Growth of
Impact Evaluation for International Development: How Much Have We Learned?”
21.
Journal of Development Effectiveness 8 (1): 1–
Chassang, Sylvain, Gerard Padró i Miquel, and Erik Snowberg. 2012. “Selective
Agent Approach to Randomized Controlled Experiments.” Ameri-
Trials: A Principal-
1309.
can Economic Review 102 (4): 1279–
Clark, Damon. 2009. “The Performance and Competitive Effects of School Auton-
783.
omy.” Journal of Political Economy 117 (4): 745–
Sharing?
Cohen, Jessica, and Pascaline Dupas. 2010. “Free Distribution or Cost-
Evidence from a Randomized Malaria Prevention Experiment.” Quarterly Journal of
45.
Economics 125 (1): 1–
Congdon, William J., Jeffrey R. Kling, Jens Ludwig, and Sendhil Mullainathan. 2017.
“Social Policy: Mechanism Experiments and Policy Evaluations.” In Handbook of
Field Experiments, volume 2, edited by Abhijit V. Banerjee and Esther Duflo, 389–
426. Amsterdam: North- Holland.
Kind Transfers.” Ameri-
Cunha, Jesse M. 2014. “Testing Paternalism: Cash versus In-
230.
can Economic Journal: Applied Economics 6 (2): 195–
Dal Bó, Ernesto, Frederico Finan, and Martín A. Rossi. 2013. “Strengthening State
Capabilities: The Role of Financial Incentives in the Call to Public Service.” Quarterly
Journal of Economics 128 (3): 1169–1218.
Deaton, Angus. 2010. “Instruments, Randomization, and Learning about Develop-
455.
ment.” Journal of Economic Literature 48 (2): 424–
de Janvry, Alain, Elisabeth Sadoulet, and Tavneet Suri. 2017. “Field Experiments in
Developing Country Agriculture.” In Handbook of Field Experiments, volume 2, edited
466. Amsterdam: North-
by Abhijit V. Banerjee and Esther Duflo, 427– Holland.
The Influence of Randomized Controlled Trials 483
Dell, Melissa. 2010. “The Persistent Effects of Peru’s Mining Mita.” Econometrica
78 (6): 1863–1903.
de Mel, Suresh, David McKenzie, and Christopher Woodruff. 2012. “One- Time
Lasting Effects on Microenterprises in Sri
Transfers of Cash or Capital Have Long-
966.
Lanka.” Science 335 (6071): 962–
Ross, Rebecca, Pascaline Dupas, and Jonathan Robinson. 2017. “Governance
Dizon-
and the Effectiveness of Public Health Subsidies: Evidence from Ghana, Kenya and
169.
Uganda.” Journal of Public Economics 156: 150–
Dobbie, Will, and Roland G. Fryer, Jr. 2014. “The Impact of Attending a School with
Achieving Peers: Evidence from New York City Exam Schools.” American Eco-
High-
75.
nomic Journal: Applied Economics 6 (3): 58–
Duflo, Esther. 2004. “Scaling up and Evaluation.” Paper presented at the Annual
Bank Conference on Development Economics (ABCDE), Bangalore, May 21– 22.
Duflo, Esther, and Michael Kremer. 2005. “Use of Randomization in the Evaluation
of Development Effectiveness.” In Evaluating Development Effectiveness, edited by
George Keith Pitman, Osvaldo N. Feinstein, and Gregory K. Ingram, 205– 230. New
Brunswick, NJ: Transaction.
Duflo, Esther, Rachel Glennerster, and Michael Kremer. 2007. “Using Randomiza-
tion in Development Economics Research: A Toolkit.” In Handbook of Development
3962.
Economics, volume 4, edited by T. Paul Schultz and John A. Strauss, 3895–
Amsterdam: Elsevier.
Duflo, Esther, Michael Kremer, and Jonathan Robinson. 2008. “How High Are Rates
of Return to Fertilizer? Evidence from Field Experiments in Kenya.” American Eco-
nomic Review 98 (2): 482–488.
Duflo, Esther, Michael Greenstone, Rohini Pande, and Nicholas Ryan. 2013a. “What
Party Auditors.” Ameri-
Does Reputation Buy? Differentiation in a Market for Third-
319.
can Economic Review 103 (3): 314–
Duflo, Esther, Michael Greenstone, Rohini Pande, and Nicholas Ryan. 2013b.
Telling by Third-
“Truth- Party Auditors and the Response of Polluting Firms: Experi-
1545.
mental Evidence from India.” Quarterly Journal of Economics 128 (4): 1499–
Duflo, Esther, Michael Kremer, Jonathan Robinson and Frank Schilbach. 2017.
“Technology Diffusion and Appropriate Use: Evidence from Western Kenya.” Work-
ing Paper, MIT, Cambridge, MA.
Run Subsidies and Long-
Dupas, Pascaline. 2014a. “Short- Run Adoption of New
228.
Health Products: Evidence From a Field Experiment.” Econometrica 82 (1): 197–
Dupas, Pascaline. 2014b. “Getting Essential Health Products to Their End Users: Sub-
sidize, But How Much?” Science 345 (6202): 1279– 1281.
484 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
Dupas, Pascaline, and Edward Miguel. 2017. “Impacts and Determinants of Health
Levels in Low-Income Countries.” In Handbook of Field Experiments, volume 2, edited
93. Amsterdam: North-
by Abhijit V. Banerjee and Esther Duflo, 3– Holland.
Dustan, Andrew, Alain de Janvry, and Elisabeth Sadoulet. 2015. “Flourish or Fail?
The Risky Reward of Elite High School Admission in Mexico City.” Department of
Economics Working Paper 15– 00002, Vanderbilt University, Nashville.
Evans, David K., and Anna Popova. 2014. “Cash Transfers and Temptation Goods:
A Review of Global Evidence.” Policy Research Working Paper 6886, World Bank,
Washington, DC.
Finan, Frederico, Benjamin A. Olken, and Rohini Pande. 2015. “The Personnel Eco-
nomics of the State.” NBER Working Paper 21825, National Bureau of Economic
Research, Cambridge, MA.
Fisher, Ronald Aylmer. 1925. Statistical Methods for Research Workers. Guildford, UK:
Genesis.
Fiszbein, Ariel, and Norbert Schady. 2009. Conditional Cash Transfers: Reducing Pres-
ent and Future Poverty. Washington, DC: World Bank.
Freedman, David A. 2006. “Statistical Models for Causation: What Inferential Lever-
713.
age Do They Provide?” Evaluation Review 30 (6): 691–
Galiani, Sebastian, and Patrick J. McEwan. 2013. “The Heterogeneous Impact of
96.
Conditional Cash Transfers.” Journal of Public Economics 103 (Supplement C): 85–
Gertler, Paul. 2004. “Do Conditional Cash Transfers Improve Child Health? Evi-
dence from PROGRESA’s Control Randomized Experiment.” American Economic
341.
Review 94 (2): 336–
Glennerster, Rachel. 2017. “The Practicalities of Running Randomized Evaluations:
Partnerships, Measurement, Ethics, and Transparency.” In Handbook of Field Experi-
243. Amster-
ments, volume 1, edited by Abhijit V. Banerjee and Esther Duflo, 175–
dam: North-Holland.
Glewwe, Paul, and Pedro Olinto. 2004. “Evaluating the Impact of Conditional Cash
Transfers on Schooling: An Experimental Analysis of Honduras PRAF Program.”
Manuscript, University of Minnesota, St. Paul.
Glewwe, Paul, Michael Kremer, and Sylvie Moulin. 2009. “Many Children Left
Behind? Textbooks and Test Scores in Kenya.” American Economic Journal: Applied
135.
Economics 1 (1): 112–
Gueron, Judy. M. 2017. “The Politics and Practice of Social Experiments: Seeds of a
Revolution.” In Handbook of Field Experiments, volume 1, edited by Abhijit V. Baner-
69. Amsterdam: North-
jee and Esther Duflo, 27– Holland.
Hanna, Rena, and Dean Karlan. 2017. “Designing Social Protection Programs: Using
Theory and Experimentation to Understand How to Help Combat Poverty.” In
The Influence of Randomized Controlled Trials 485
Handbook of Field Experiments, volume 2, edited by Abhijit V. Banerjee and Esther
553. Amsterdam: North-
Duflo, 515– Holland.
Haushofer, Johannes, and Jeremy Shapiro. 2013. “Household Response to Income
Changes: Evidence from an Unconditional Cash Transfer Program in Kenya.”
Mimeo, Massachusetts Institute of Technology, Cambridge, MA.
Heckman, James J. 1992. “Randomization and Social Policy Evaluation.” In Evaluat-
ing Welfare and Training Programs, edited by Charles Manski and Irwin Garfinkel,
230. Cambridge MA: Harvard University Press.
201–
International Rescue Committee. 2014. “IRC releases evaluation: Cash transfers
work for refugees in emergencies.” International Rescue Committee, New York.
Developed
Karlan, Dean S., and Jonathan Zinman. 2008. “Credit Elasticities in Less-
Economies: Implications for Microfinance.” American Economic Review 98 (3):
1040–1068.
Kawamoto, Kensaku, Caitlin A. Houlihan, E. Andrew Balas, and David F. Lobach.
2005. “Improving Clinical Practice Using Clinical Decision Support Systems: A Sys-
tematic Review of Trials to Identify Features Critical to Success.” BMJ 330 (7494): 765.
Khan, Adnan Q., Asim I. Khwaja, and Benjamin A. Olken. 2016. “Tax Farming
Redux: Experimental Evidence on Performance Pay for Tax Collectors.” Quarterly
271.
Journal of Economics 131 (1): 219–
Klamer, Arjo, and Hendrik P. van Dalen. 2002. “Attention and the Art of Scientific
315.
Publishing.” Journal of Economic Methodology 9 (3): 289–
Kreindler, Gabriel. 2018. “The Welfare Effect of Road Congestion Pricing: Experimen-
tal Evidence and Equilibrium Implications.” Job Market Paper, MIT, Cambridge, MA.
Kremer, Michael. 2003. “Randomized Evaluations of Educational Programs in Devel-
106.
oping Countries: Some Lessons.” American Economic Review 93 (2): 102–
Kremer, Michael, and Rachel Glennerster. 2011. “Improving Health in Developing
Countries: Evidence from Randomized Evaluations.” In Handbook of Health Econom-
ics, volume 2, edited by Mark V. Pauly, Thomas G. Mcguire, and Pedro P. Barros,
201– 315. Amsterdam: Elsevier.
Kremer, Michael, and Alaka Holla. 2009. “Pricing and Access: Lessons from Random-
ized Evaluations in Education and Health.” In What Works in Development? Thinking
129. Wash-
Big and Thinking Small, edited by William Easterly and Jessica Cohen, 91–
ington, DC: Brookings Institution Press.
Kremer, Michael, and Edward Miguel. 2007. “The Illusion of Sustainability.” Quar-
1065.
terly Journal of Economics 122 (3): 1007–
Lucas, Adrienne M., and Isaac M. Mbiti. 2014. “Effects of School Quality on Stu-
dent Achievement: Discontinuity Evidence from Kenya.” American Economic Journal:
263.
Applied Economics 6 (3): 234–
486 Abhijit Vinayak Banerjee, Esther Duflo, and Michael Kremer
Maluccio, John A., and Rafael Flores. 2005. “Impact Evaluation of a Conditional
Cash Transfer Program: The Nicaraguan Red de Protección Social.” IFPRI Research
Report 141, International Food Policy Research Institute, Washington, DC.
Mansilla, Ricardo, Elke Köppen, Germinal Cocho, and Pedro Miramontes. 2007.
Order Distribution.” Journal of
“On the Behavior of Journal Impact Factor Rank-
160.
Informetrics 1 (2): 155–
Masterson, Daniel, and Christian Lehmann. 2014. “Emergency Economies: The
Impact of Cash Assistance in Lebanon.” International Rescue Committee, New York.
Meager, Rachael. 2016. “Understanding the Impact of Microcredit Expansions: A
Bayesian Hierarchical Analysis of 7 Randomised Experiments.” Working Paper, MIT,
Cambridge, MA.
Muralidharan, Karthik, and Venkatesh Sundararaman. 2015. “The Aggregate Effect
Stage Experiment in India.” Quarterly Journal
of School Choice: Evidence from a Two-
1066.
of Economics 130 (3): 1011–
Muralidharan, Karthik, Paul Niehaus, and Sandip Sukhtankar. 2016. “Building State
Capacity: Evidence from Biometric Smartcards in India.” American Economic Review
2929.
106 (10): 2895–
Neyman, Jerzy. [1923] 1990. “On the Application of Probability Theory to Agricul-
tural Experiments. Essay on Principles. Section 9.” Translated and edited by Dorota
472.
M. Dabrowska and Terence P. Speed. Statistical Science 5 (4): 465–
Olken, Benjamin A. 2007. “Monitoring Corruption: Evidence from a Field Experi-
249.
ment in Indonesia.” Journal of Political Economy 115 (2): 200–
Padró i Miquel, Gerard, Nancy Qian, and Yang Yao. 2014. “Social Fragmentation,
Public Goods and Elections: Evidence from China.” NBER Working Paper 18633,
National Bureau of Economic Research, Cambridge, MA.
Patel, Vikram, Helen A. Weiss, Neerja Chowdhary, Smita Naik, Sulochana Pednekar,
Sudipto Chatterjee, Mary J. De Silva, et al. 2010. “Effectiveness of an Intervention Led
by Lay Health Counsellors for Depressive and Anxiety Disorders in Primary Care in Goa,
India (MANAS): A Cluster Randomised Controlled Trial.” Lancet 376 (9758): 2086– 2095.
Pritchett, Lant. 2002. “It Pays to Be Ignorant: A Simple Political Economy of Rigor-
469.
ous Program Evaluation.” Journal of Policy Reform 5 (4): 251–
Radicchi, Filippo, Santo Fortunato, and Claudio Castellano. 2008. “Universality of
Citation Distributions: Toward an Objective Measure of Scientific Impact.” PNAS
104 (45): 17268–17272.
Rao, Gautam, Frank Schilbach, and Heather Schofield. n.d. “Sleepless in Chennai:
The Economic Effects of Sleep Deprivation among the Poor.” Working Paper, Uni-
versity of Pennsylvania, Philadelphia.
The Influence of Randomized Controlled Trials 487
Ravallion, Martin. 2012. “Fighting Poverty One Experiment at a Time: A Review of
Abhijit Banerjee and Esther Duflo’s ‘Poor Economics: A Radical Rethinking of the
114.
Way to Fight Global Poverty.’” Journal of Economic Literature 50 (1): 103–
Rigol, Natalia, Reshmaan Hussam, and Giovanni Regianni. 2017. “Habit Formation and
030, Cambridge, MA.
Rational Addiction.” Harvard Business School Working Paper 18-
Rubin, Donald B. 1974. “Estimating Causal Effects of Treatments in Experimental
670.
and Observational Studies.” Journal of Educational Psychology 66 (5): 668–
Rubin, Donald B. 1981. “Estimation in Parallel Randomized Experiments.” Journal of
401.
Educational Statistics 6 (4): 377–
Control: A Field Experiment in India.”
Schilbach, Frank. 2015. “Alcohol and Self-
Working Paper, MIT, Cambridge, MA.
Schultz, T. Paul. 2004. “School Subsidies for the Poor: Evaluating the Mexican
250.
PROGRESA Poverty Program.” Journal of Development Economics 74 (1): 199–
Shah, Neil Buddy, Paul Wang, Andrew Fraker, and Daniel Gastfriend. 2015. “Evalu-
Focused Impact Evaluation as a Practical Policymaking
ations with Impact: Decision-
Tool.” 3ie Working Paper 25, International Initiative for Impact Evaluation, New
Delhi.
Sim, Ida, Paul Gorman, Robert A. Greenes, R. Brian Haynes, Bonnie Kaplan, Harold
Lehmann, and Paul C. Tang. 2001. “Clinical Decision Support Systems for the Prac-
tice of Evidence-Based Medicine.” Journal of the American Medical Informatics Associa-
534.
tion 8 (6): 527–
Todd, Petra E., and Kenneth I. Wolpin. 2006. “Assessing the Impact of a School Sub-
sidy Program in Mexico: Using a Social Experiment to Validate a Dynamic Behavioral
1417.
Model of Child Schooling and Fertility.” American Economic Review 96 (5): 1384–
Udry, Christopher. 1995. “Risk and Saving in Northern Nigeria.” American Economic
1300.
Review 85 (5): 1287–
Vickrey, William S. 1969. “Congestion Theory and Transport Investment.” American
260.
Economic Review 59 (2): 251–
Vivalt, Eva. 2015. “How Much Can We Generalize from Impact Evaluations? Are
They Worthwhile?” Mimeo, Stanford University, Palo Alto, CA.
World Bank. 2013. Philippines Conditional Cash Transfer Program: Impact Evaluation
2012. Washington, DC: World Bank.
World Bank. 2016. World Development Report 2016: Digital Dividends. Washington,
DC: World Bank.
World Health Organization. 2015. World Malaria Report 2015. Geneva: World Health
Organization.
Comment: David McKenzie
The rise and normalization of randomized controlled trials (RCTs) as an
important part of the toolkit of development economists has been rapid,
with much debate as to whether this is a cause for celebration or concern.
Abhijit, Esther, and Michael have been the early pioneers and proponents
of the use of RCTs in development economics, and their paper represents
an important stocktaking exercise, documenting this rise and attempting
to draw out some of the consequences of this process for both research
and policy. I group my comments around three themes: putting the rise
of RCTs in perspective, considering how they have affected the practice
of research, and attempting to understand how they have and have not
influenced policy.
The Rise of RCTs in Perspective
Their paper documents the rapid growth in the number of RCTs published
in top journals, from 0 papers in 2000 to 32 papers in 2015. In table 10.4,
I extend this analysis by also considering development economics papers
published in three leading general interest economics journals considered
five journals (American Economic Journal:
to be in the next tier below the top-
Applied Economics, Review of Economics and Statistics, and Economic Journal),
papers published in three leading development economics journals (Journal
of Development Economics, Economic Development and Cultural Change, and
World Bank Economic Review), and in World Development, the leading mul-
tidisciplinary journal of development. I consider papers published in 2015
and define development economics papers in the general interest journals
as those with an “O” (development economics) Journal of Economic Litera-
ture classification code.
The Influence of Randomized Controlled Trials 489
Table 10.4
RCTs as a share of development papers published in 2015, by journal type
Number of
Development Number that
Papers are RCTs Percent RCT
Top five journals 32 10 31.3
Good general interest 32 14 43.8
American Economic Journal: 16 10 62.5
Applied Economics
Economic Journal 8 1 12.5
Review of Economics and Statistics 8 3 37.5
Leading development journals 115 15 13.0
Journal of Development Economics 70 9 12.9
Economic Development and 24 5 20.8
Cultural Change
World Bank Economic Review 21 1 4.8
World Development 275 5 1.8
All development papers 454 44 9.7
Source: Top five journals data are from Banerjee, Duflo and Kremer (2016). Data for
other journals collected by author. Papers at good general interest journals classified
as development if they have an “O” code in the Journal of Economic Literature (JEL)
classification system. Counts exclude editorials, comments, rejoinders, corrigendum,
the papers and proceedings issue of the World Bank Economic Review (WBER), and the
125th anniversary issue of the Economic Journal (EJ).
Several points emerge from this table that I believe are important for put-
ting the rise of RCTs in perspective and for considering their influence on
policy. First, despite the rapid growth, the majority of development econom-
five journals are not RCTs. Second, RCTs
ics papers published in even the top-
make up a much higher share of development papers in general interest jour-
nals than they do in development journals. Third, most published develop-
ment papers are not being published in the top journals but in field journals.
As a result, out of the 454 development papers published in these fourteen
journals in 2015, only 44 are RCTs (9.7 percent). The consequence is that
RCT studies are only a small share of all development research taking place. I
believe this is evidence against the (perhaps strawman) argument that RCTs
have crowded out other development research, and policy makers looking
for advice on questions RCTs can’t answer are missing out as a result.
490 Comments by David McKenzie and Martin Ravallion
Their paper also documents how RCTs have become more common
among younger researchers, showing that BREAD members who gradu-
ated more recently are more likely to have done RCTs than those who
graduated longer ago. This observation has led to a second caricature or
strawman argument: that the “best and brightest talent of a generation
of development economists have been devoted to producing rigorous
impact evaluations about topics that are easy to randomize (e.g., Pritchett
2014) and that they take a “randomize or bust” attitude, whereby they
turn down many interesting research questions if they can’t randomize
(e.g., Ravallion 2009).
five
To explore this, I examined the publication records of the sixty-
BREAD affiliates (this is the group of more junior members), restricting
three researchers who had graduated in 2011 or ear-
attention to the fifty-
lier (to give them time to have published). The median researcher had pub-
lished nine papers, and the median share of their papers which were RCTs
was 13 percent. Focusing on the subset of those who have published at
least one RCT, the mean (median) percentage of their published papers that
90 range is 11–
are RCTs is 35 percent (30 percent), and the 10– 60 percent.
So young researchers who publish RCTs also do write and publish papers
that are not RCTs. Indeed, this is also true of Abhijit, Esther, and Michael—
although known as the leaders of the “randomista” movement, the top-
cited papers of all three researchers are not RCTs.
The Influence of RCTs on Development Research
Abhijit, Esther, and Michael document several important ways that RCTs
have affected the way development economics research is done. I agree
with their claims that RCTs have raised the bar for nonexperimental
research in terms of thinking about credible identification, and that RCTs
have spurred creative new ways of measurement. I want to note two other
areas of influence.
The first, extremely positive, influence has been making it common-
place for researchers to actually talk to the people and firms they are study-
ing. This is a big change from the era when most development research
consisted of researchers downloading a dataset like the Penn World Tables
or Living Standards Measurement Surveys, attempting to estimate some
model or test some theory, and then writing the paper without ever talking
The Influence of Randomized Controlled Trials 491
to anyone in the country being studied. Indeed, this categorizes well my
dissertation research: I was interested in understanding why people in Tai-
wan continued to save so much when their incomes had been rising rapidly
for years. I carefully worked out new econometric theory and estimated
and tested models of several competing consumption theories, but never
asked directly any households in Taiwan “Why do you save so much?” I
likewise have been on several World Bank missions where projects were
being designed by talking to policy makers and perhaps a handpicked set
of existing beneficiaries, and the idea of just walking into an average neigh-
borhood and talking to some randomly selected small businesses was seen
as a surprising thing to do. RCTs make this more commonplace, and they
also make it much more likely that researchers actually talk to the imple-
menters of the programs they are trying to study.
However, I also think that RCTs do affect to some extent which ques-
tions researchers work on. As noted above, there are many researchers, and
most research done in development economics is still not done via RCTs.
I think it is fair to say that probably some questions have been answered
only because they could be answered cleanly by an experiment, and these
questions would not have otherwise had researchers working on them. As
I argue in the next section, it is unclear whether this is necessarily a bad
thing, as it has resulted in researchers getting much more involved in the
messy business of understanding how policies are implemented, which
otherwise had not received much research attention.
The Influence of RCTs on Development Policy
I think it is fair to say that RCTs have had much more influence on how
development economics policy is implemented rather than on what is
done. Many of the questions answered by RCTs fall into the category of
helping policy makers better target, or better implement, a policy they have
already decided to do. For example, should grants be given conditionally or
unconditionally? How can government workers be incentivized to provide
the services they are meant to provide? Should mosquito nets be given
out for free or offered at a price? Will people use savings products more if
offered commitments or reminders? This use of RCTs is very similar to the
B testing is used to fine
main use of RCTs in a lot of businesses, where A-
tune products and decide how to best target customers.
492 Comments by David McKenzie and Martin Ravallion
When it comes to what is done, I make the distinction between efforts to
try to make marginal improvements in the lives of people and firms, given
the economic structure they operate in, and attempts to spur the types of
changes from a stagnant, largely rural, agrarian economy to a vibrant, inno-
based economy that we
vative, largely urban manufacturing and services-
associate with the process of development. Much of the early RCT research
up cases profiled by
was focused on the former, and many of the DIV scale-
how can we make traffic a little less
Banerjee et al. also fall into this case—
risky, water a bit cleaner, poor households get a little more electricity, and
so forth. Success here is largely in terms of making poor people a little bit
less poor, or making life a little easier for them. This is an important class of
policies, and one where RCTs have had some policy influence.
In contrast, until recently there have been far fewer RCTs that help
policy makers attempt to test policies associated with a more structural
how do we get more firms innovating and growing?
transformation—
How do we get people to move out of poor places with few job prospects
to places with better prospects? However, this is an area where RCTs are
rapidly expanding, with examples like Bryan, Chowdhury, and Mobarak
(2014), Atkin, Khandewal, and Osman (2017), McKenzie (2015), Beam,
McKenzie, and Yang (2016), and Cusolito, Dautovic, and McKenzie (2018)
showing that RCTs can also provide useful policy advice for these ques-
tions as well.
A final point I want to make is to argue against the idea that policy
by-
makers can easily substitute for RCTs by rapid, iterative learning- doing
processes. Such an approach may be possible in some environments, but it
is very difficult to learn by doing in some situations. One reason for this is
that people often find it hard to generate accurate counterfactuals for them-
selves, even when they have gone through a program, so McKenzie (2018)
finds that both treatment and control groups overestimate the effect that
winning a business plan competition would have, even after the fact. Sec-
ond, so many factors influence outcomes that RCTs often need hundreds
or thousands of observations to detect an effect, and it is impossible for
individuals to extract signal from noise to determine whether their actions
are working. As an extreme example, Lewis and Rao (2015) show that firms
often cannot know whether their marketing campaigns are working, even
when testing on millions of customers.
The Influence of Randomized Controlled Trials 493
References
Atkin, David, Amit Khandewal, and Adam Osman. 2017. “Exporting and Firm Per-
formance: Evidence from a Randomized Experiment.” Quarterly Journal of Economics
132(2): 551–615.
Banerjee, Abhijit V., Esther Duflo, and Michael Kremer. 2016. “The Influence of
Randomized Controlled Trials on Development Economics Research and on Devel-
opment Policy.” Paper prepared for the World Bank’s “The State of Economics, The
State of the World” Conference, September 2016.
Beam, Emily A., David McKenzie, and Dean Yang. 2016. “Unilateral Facilitation
Does Not Raise International Labor Migration from the Philippines.” Economic Devel-
368.
opment and Cultural Change 64 (2): 323–
Bryan, Gharad, Shyamal Chowdhury, and Ahmed Mushfiq Mobarak. 2014. “Under-
investment in a Profitable Technology: The Case of Seasonal Migration in Bangla-
1748.
desh.” Econometrica 82 (5): 1671–
Cusolito, Ana Paula, Ernest Dautovic, and David McKenzie. 2018. “Can Government
Intervention Make Firms More Investment- Ready? A Randomized Experiment in the
Western Balkans.” Policy Research Working Paper 8541, Impact Evaluation Series,
World Bank, Washington, DC.
Lewis, Randall A., and Justin M. Rao. 2015. “The Unfavorable Economics of Measur-
1973.
ing the Returns to Advertising.” Quarterly Journal of Economics 130 (4): 1941–
McKenzie, David. 2017. “Identifying and Spurring High-Growth Entrepreneurship:
Experimental Evidence from a Business Plan Competition.” American Economic Review
107(8): 2278–2307.
McKenzie, David. 2018. “Can Business Owners Form Accurate Counterfactuals?
Eliciting Treatment and Control Beliefs about Their Outcomes in the Alternative
Treatment Status.” Journal of Business and Economic Statistics 36 (4): 714–722.
Pritchett, Lant. 2014. “Is Your Impact Evaluation Asking Questions That Matter? A
Four Part Smell Test.” Views from the Center (blog), November 6. http:// www .cgdev
.org/blog/your-impact-evaluation-asking-questions-matter-four-part-smell-test.
5.
Ravallion, Martin. 2009. “Should the Randomistas Rule?” Economists’ Voice 6 (2): 1–
Comment: Martin Ravallion
Randomized Trials and Development Policy
Measure what is important, don’t make important what you can measure.
Robert McNamara, president of the World Bank, 1968–
— 1981
Randomized controlled trials (RCTs) are on the menu of options for devel-
opment impact evaluation. That is not news, for it has been true for at least
40 years.1 What has changed over the past 10–
15 years is the academic
popularity of RCTs. The chapter by Banerjee, Duflo, and Kremer (BDK)
describes and reflects on the expanding use of RCTs in development eco-
nomics. The authors have been at the forefront of this change.
In theory, the idea of an RCT is simple enough. Access to the program is
randomly assigned to some units, with others set aside as a control group.
The impact is then estimated by the difference in the sample mean out-
comes between treated and control groups. This converges toward the true
mean impact in the population as the sample sizes increase.
In practice, RCTs are rarely perfect, their internal validity is rarely assured,
and their external validity is often questionable. As argued by Deaton and
Cartwright (2018), these limitations do not appear to be well understood
among practitioners. It does not help that prominent advocates often make
unguarded claims that exaggerate the virtues of RCTs. For example, it is
clearly not true that “any difference between treatment and control units
reflects the impact of the treatment,” as BDK say, because there is always
some experimental error (including, of course, sampling error).
1. The earliest development RCT that I know of was done in 1978 by the World Bank
and was published in 1981, namely, Jamison et al. (1981).
The Influence of Randomized Controlled Trials 495
The concerns go deeper. Not even the theoretical rationale for random-
ization is as clear as advocates claim. Indeed, quite generally, there exists a
deterministic (nonrandom) assignment of treatment status (based on con-
tinuous covariates) that minimizes the expected error variance, as shown
by Kasy (2016). This holds for a given sample size. Comparing methods,
it makes more sense to fix the budget for the evaluation than to fix the
sample size. RCTs can be costly. With a given budget, RCTs will often have
lower sample sizes than are possible with observational studies (OSs). An OS
can then turn out to be closer to the truth in practice, even if it comes with
a bias (Ravallion 2018).
Has the new popularity of RCTs in development research helped inform
development policy making? That is not the only reason we might do
to identify
RCTs; another is to better understand how an economy works—
key structural parameters. However, policy making is an important reason.
BDK clearly agree. Indeed, that is explicitly the goal of the premier institu-
tion for promoting RCTs in development, namely, the Abdul Latif Jameel
PAL), founded by two of the authors. On the bio page
Poverty Action Lab (J-
PAL’s mission is to reduce
of Banerjee and Duflo (2011), it is said that “J-
poverty by ensuring that policy is based on scientific evidence.” (“Scientific
PAL and other advocates of
evidence” can be taken as code for RCTs.) J-
RCTs have framed their task as that of figuring out what works and what
does not, to scale up the former and scale down the latter. Is that what is
happening now?
To inform antipoverty policy making, researchers ideally should be fill-
ing the gaps between what we know about the effectiveness of policies and
what policy makers need to know. As economists, we should first ask our-
selves: Why do such gaps exist? Imperfect information plays a role. Here
the problem is that development practitioners cannot easily assess the
quality and expected benefits of an evaluation, to weigh against the costs.
Compared to the complex econometric methods used in some OSs, the
simplicity of an RCT helps practitioners understand what is being done.
However (as already noted), that understanding is not always as deep as
it needs to be for practitioners to properly assess the lessons from an RCT,
including its limits.
There are also important externalities. The benefits of an evaluation are
rarely confined to that specific project but instead spill over to future proj-
ects. These external benefits are probably greater for OSs than for RCTs,
496 Comments by David McKenzie and Martin Ravallion
for which external validity has been a recurrent concern (see, for example,
Pritchett and Sandefur 2015). In addition, current project managers cannot
be expected to take proper account of the external benefits to other proj-
ects when deciding how much to spend on their own project’s evaluation.
Thus there may well be an underinvestment in OSs, which generate more
externalities, relative to RCTs.
Knowledge gaps also stem from misalignments of evaluative effort. One
aspect is that development evaluators often ignore the scope for fungibil-
ity. Recipients (governmental or not) can reallocate their own efforts in
response to new funding, such as development aid. As a consequence of
such fungibility, donors are often implicitly supporting something else and
evaluating the wrong program from the point of view of assessing their
impact. Then evaluative efforts are not aligned well with development
efforts. This applies as much to RCTs as to OSs.
Methodological preferences on the part of evaluators can reinforce such
misalignments, and here the emphasis on RCTs may well be hurting our
progress in addressing important knowledge gaps. There are both output
and substitution effects of the RCT boom. The output effect is obvious, as
documented by BDK. The substitution effect relates to the methods used.
There has been a marked increase in the share of journal articles on devel-
opment economics that use RCTs. But that is not where a methodological
substitution is worrying; instead, it is in policy evaluation. We have seen
a marked switch in favor of RCTs in institutions such as the World Bank.
The Bank’s own Independent Evaluation Group reports that more than 80
2010 used randomiza-
percent of the impact evaluations starting in 2007–
2006 and only 19 percent in prior
tion, compared with 57 percent in 2005–
years (World Bank 2012).
A problem in overall policy evaluation stems from the fact that random-
ization is clearly only feasible for a nonrandom subset of policies and set-
tings, so we lose our ability to comprehensively address our knowledge gaps
(Ravallion 2009, 2018). For example, it is rarely feasible to randomize the
scale infrastructure projects and sectoral and
location of medium-to large-
wide reforms, which are core activities in almost any poor coun-
economy-
try’s development strategy. Indeed, the very idea of randomized assignment
is antithetical to the goals of many development programs, which typi-
cally aim to reach certain types of people or places. Governments will
(hopefully) be able to do better in reaching poor people than a random
The Influence of Randomized Controlled Trials 497
assignment would. Randomization also tends to be better suited to relatively
simple programs, with clearly identified participants and nonparticipants,
rather short time horizons, and little scope for the costs or benefits to spill
over to nonparticipants.
There are both supply and demand sides to this misalignment. On the
supply side, the reality today is that graduate students and their teachers
are wandering around looking for something they can randomly assign. If
randomization is not feasible for the question being posed, then research-
ers are often drawn to ask other questions. Governments in the developing
world are having a harder time finding someone to help evaluate those
public programs for which randomization is not a feasible option.
The potential biases go further. On the demand side, governments (and
development agencies) are largely free to choose what gets evaluated. Even
when they agree to an RCT, they can choose those programs for which they
do not care what the verdict will be. Other programs will not get evaluated
in equilibrium. (And, as noted, they may include what was really being
funded by aid.) The risks are plain.
If we are really concerned about obtaining reliable estimates of the
impact of the portfolio of development policies, we should choose a rep-
resentative sample from that portfolio and then find the best method for
each of the selected programs/policies, with randomization as only one of
a number of options. That is not what is happening now.
References
Banerjee, Abhijit V., and Esther Duflo. 2011. Poor Economics: A Radical Rethinking of
the Way to Fight Global Poverty. New York: Public Affairs.
Deaton, Angus, and Nancy Cartwright. 2018. “Understanding and Misunderstand-
21.
ing Randomized Controlled Trials.” Social Science and Medicine 210: 2–
Jamison, Dean, Barbara Searle, Klaus Galda, and Stephen P. Heyneman. 1981.
“Improving Elementary Mathematics Education in Nicaragua: An Experimental
Study of the Impact of Textbooks and Radio on Achievement.” Journal of Educational
567.
Psychology 73 (4): 556–
Kasy, Maximilian. 2016. “Why Experimenters Might Not Always Want to Random-
338.
ize, and What They Should Do Instead.” Political Analysis 24: 324–
Pritchett, Lant, and Justin Sandefur. 2015. “Learning from Experiments When Con-
475.
text Matters.” American Economic Review: Papers and Proceedings 105 (5): 471–
498 Comments by David McKenzie and Martin Ravallion
5.
Ravallion, Martin. 2009. “Should the Randomistas Rule?” Economists’ Voice 6 (2): 1–
Ravallion, Martin. 2018. “Should the Randomistas (Continue to) Rule?” Working
Paper 492, Center for Global Development, Washington, DC.
World Bank. 2012. World Bank Group Impact Evaluations: Relevance and Effectiveness.
Independent Evaluation Group, World Bank, Washington, DC.
Contributors
Philippe Aghion is a professor at the College de France and at the London School of
Economics.
Ingela Alger is research faculty at the Toulouse School of Economics, research direc-
tor at Center National de la Recherche Scientifique (CNRS), and director of the biol-
ogy program at the Institute for Advanced Study (IAST) in Toulouse.
Kenneth Arrow was Joan Kenney Professor of Economics and professor of operations
research, emeritus at Stanford University. He was the joint winner of the 1972 Nobel
Memorial Prize in Economic Sciences.
Abhijit Vinayak Banerjee is the Ford Foundation International Professor of Econom-
ics at MIT and co-director of J-PAL.
Kaushik Basu is Carl Marks Professor of International Studies at Cornell University
and former senior vice president and chief economist of the World Bank.
Lawrence E. Blume is Visiting Research Professor at the Institute for Advanced Stud-
ies, Vienna, and member of the Santa Fe Institute’s external faculty.
Guillermo Calvo is professor of economics, international, and public affairs at
Columbia University.
Francesco Caselli is Norman Sosnow Chair in Economics at the London School of
Economics.
Aslı Demirgüç-Kunt is director of research at the World Bank.
Shantayanan Devarajan is senior director for development economics at the World
Bank.
Esther Duflo is the Abdul Latif Jameel Professor of Poverty Alleviation and Develop-
ment Economics at MIT and co-director of J-PAL.
Sam Fankhauser is at Grantham Research Institute on Climate and the Environment
and Centre of Climate Change Economics and Policy (CCCEP) at the London School
of Economics and Political Science.
500 Contributors
James E. Foster is Oliver T. Carr Professor of International Affairs and professor of
economics at the George Washington University.
Varun Gauri is senior economist in the Development Economics Vice Presidency of
the World Bank. He co- leads the Mind, Behavior, and Development Unit (eMBeD),
and was co-director of the World Development Report 2015: Mind, Society, and
Behavior.
Xavier Giné is a lead economist on the Finance and Private Sector Development
Team of the Development Research Group at the World Bank.
Gaël Giraud is chief economist and executive director of the Innovation, Research
and Knowledge Directorate at the Agence Française de Développement (AFD).
Gita Gopinath is the economic counsellor and director of research at the Interna-
tional Monetary Fund (IMF) and John Zwaanstra Professor of International Studies
and of Economics at Harvard University.
Robert Hockett is Edward Cornell Professor of Law and professor of public policy at
Cornell University.
Karla Hoff is lead economist in the Development Research Group at the World Bank
and co-director of the World Development Report 2015: Mind, Society, and Behavior.
Ravi Kanbur is T. H. Lee Professor of World Affairs, international professor of applied
economics and management, and professor of economics at Cornell University.
Aart Kraay is senior advisor in the Macroeconomics and Growth Group of the Devel-
opment Research Group at the World Bank.
Michael Kremer is Gates Professor of Developing Societies in the department of eco-
time scientific director of develop-
nomics at Harvard University and serves as part-
ment Innovation Ventures at USAID.
David McKenzie is a lead economist on the Finance and Private Sector Development
Team of the Development Research Group at the World Bank.
Célestin Monga is the chief economist and vice president, Economic Governance
and Knowledge Management, African Development Bank.
Maurice Obstfeld is professor of economics at University of California, Berkeley, and
a former economic counsellor and director of research at the International Monetary
Fund.
Hamid Rashid is chief, Development Research, Economic Analysis and Policy Divi-
sion, the United Nations Department of Economic and Social Affairs in New York.
Martin Ravallion holds the inaugural Edmond D. Villani Chair of Economics at
Georgetown University. Prior to joining Georgetown he was the director of the
World Bank’s research department.
Contributors 501
David Rosenblatt is manager of strategy and operations in the Development Unit at
the World Bank.
Amartya Sen is Thomas W. Lamont University Professor, and professor of econom-
ics and philosophy at Harvard University. He is also a senior fellow at the Harvard
Society of Fellows. He is the winner of the 1998 Nobel Prize in Economic Sciences.
Claudia Sepúlveda is a lead economist in the Development Economics Vice Presi-
dency at the World Bank.
Luis Servén is senior advisor in the Macroeconomics and Growth Group of the
Development Research Group at the World Bank.
Hyun Song Shin is economic adviser and head of research at the Bank of Interna-
tional Settlements.
Nicholas Stern is the IG Patel Professor of Economics and Government, London
School of Economics and Political Science, and former president of the British
Academy.
Joseph Stiglitz is university professor at Columbia University. He is the joint winner
of the 2001 Nobel Memorial Prize in Economic Sciences.
Cass R. Sunstein is the Robert Walmsley University Professor at Harvard University.
Michael Toman is research manager of the Sustainable Development Team in the
Development Research Group at the World Bank.
Jörgen W. Weibull is A. O. Wallenberg Professor of Economics at Stockholm School
of Economics and visiting research fellow at Institute for Advanced Study (IAST) in
Toulouse.
Index
Figures and tables are denoted by “f” and “t” following page numbers.
Abdul Latif Jameel Poverty Action Lab Alger, Ingela, 389, 392, 397n10, 408,
(J-PAL), 442, 447, 467, 469, 495 417–430, 433
“Above average” effect, 355 Alp, Harun, 271–272
ABS. See Asset-backed securities Altruism, 406–408, 419–420
Acemoglu, Daron, 266, 268, 269n13, American Economic Association, 259,
273–274 452, 478
Adaptive expectations, 160, 161, 161n3 Andreoni, James, 407
AD securities, 105, 106 Angrist, Joshua, 273–274
Adverse selection, 112, 115–117, 199, Antisocial preferences, 390, 420
455, 456 Anti-trust policies, 105–106, 126
Africa. See also specific regions and Appiah, Kwame Anthony, 91
countries Arbitrariness in social choice, 48
bed net distribution programs in, Argentina
464–465 bank lending in, 160n1
emergence of reason in, 90 productivity growth in, 267
overvaluation of CFA franc in, 34 Arnott, Richard, 108n8, 116
African Americans, 59, 363 Arrhenius, Svante, 301
Agent-based ethics, 428 Arrow, Kenneth J.
Aggregative games, 392 on aggregating preferences, 125
Aghion, Philippe, 253, 258–260, on asymmetric information, 44
268–269, 273, 275, 284 on behavioral economics, 37–38,
Agriculture Department, US, 358–359, 40, 45
361, 363 collaboration with Debreu, 31
Aguiar, Mark, 194 on conditions for market efficiency,
Aidgrade database, 442, 443f 105, 120
Air pollution, 299, 323 convexity in proofs of, 114
Akcigit, Ufuk, 262, 271–272 death of, 23
Akerlof, George A., 5, 117, 124 on general equilibrium theory, 2, 5,
Alesina, Alberto, 269, 434 34, 36
504 Index
Arrow (cont.) Austen-Smith, David, 405
impossibility theorem of, 6, 48–51, 58, Authoritarianism, 48
62, 65–66, 82 Automatic enrollment
on interpersonal comparison of utility, default rules and, 363–365
61–62 for electronic payroll statements, 364
on propositions, 425 in retirement plans, 351, 363, 383
on risk markets, 105, 106 in savings programs, 356–357, 363
Social Choice and Individual Values, 6 in school lunch programs, 363
on social choice theory, 52, 58, as solution to procrastination, 352
60–63, 82 Automatic thinking, 38
on welfare economics, 35, 49 Availability bias, 356
Artzner, Philippe, 331 Avdjiev, Stefan, 221, 223
Asia. See also specific regions and countries
coal-fired power plants in, 327, 329 Banerjee, Abhijit Vinayak
economic growth in, 273, 286 data collection by, 459
financial crisis in (1997–1998), 154, 177 on external policy advice, 450
Asset-backed securities (ABS), 168, 169, Indonesian rice distribution study by,
176, 186 462–463
Assets as J-PAL founder, 495
distribution of, 118 on mechanism experiments, 455
financial, 160, 168 poverty research by, 266
global, 214 on RCTs, 448
liquid (see Liquid assets) on structured speculation, 453
price of, 201 USAID DIV case study by, 492
private-label, 199–200, 203 Bankruptcy law, 114, 126
safe, 183, 194–196, 199–200, 203, Banks, Jeffrey S., 405
204, 234 Banks and banking sector. See also
toxic, 176, 196 Federal Reserve
underpriced, 106 competition in, 240
Assortative matching, 391–392, 394, 419 conflicts of interest in, 145–148
Assortativity profiles, 394, 395 cross-border lending to emerging
Asymmetric information markets, 247, 247f
bargaining with, 116 European Central Bank, 167, 173, 176,
endogeneity of, 109–111, 118–119 181–182, 194
in financial sector, 148, 150, 239 fiscal authority lending by, 160, 160n1
incentives for exploitation of, 107, global, 212–219, 215–218f, 220f,
107n5, 119, 145 239–242
in insurance sector, 44, 117, 118 incentives for complexity, 124
market efficiency and, 106–107, 118 information used by, 118
market failures caused by, 145 lending in relation to GDP growth,
technology and, 127 240, 241f
uncertainty and, 32 market share of, 148, 149f
Athey, Susan, 448, 450, 454 new product development by, 10
Index 505
regulation of, 110, 146–148, 241 on simplification, 351–352, 365, 366
risk-taking capacity of, 211n3 on social influences, 40, 354–355
shadow, 178, 229 on subjective information
stock-based compensation by, 147–149, processing, 44
149t Bentham, Jeremy, 56, 81
Bargaining Bergeaud, Antonin, 262
with asymmetric information, 116 Bergson, Abram, 6, 58
collective, 126 Bergstrom, Theodore C., 397, 397n10,
default rules and, 351 419–420, 426
ultimatum bargaining game, 406 Bernoulli, Daniel, 2
Barter equilibria, 169, 198 Bernoulli, Nicolaus, 2
Basu, Kaushik, 1 Bertrand, Marianne, 456–457
Baumol, William, 58 Bertrand competition, 26, 115
Bed nets, 453, 459, 464–465 Bester, Helmut, 420
Behavioral development economics, Bias
42, 44 availability, 356
Behavioral economics, 349–391. See also cognitive, 387
Nudges confirmatory, 127
actor types in, 38, 44f, 45 in decision-making, 127
choice architecture in, 349, 356–357, in industrial policy, 266
366, 376–378 publication, 452, 475
on default rules, 349, 351, 362–365, randomization, 450–451, 495, 497
383 selection, 448
defined, 37 sunk cost, 387
development policy and, 10–11, Black, Duncan, 52
349–350, 382 Blacks, 59, 363
equilibrium in, 11 Blanchard, Olivier, 245, 336, 340
on framing and presentation, 45, Blanche, Francis, 92
353–354, 359 Bloom, Nick, 257, 271, 288
history and development of, 372– Blume, Lawrence E., 417, 421
373, 375 Blundell, Richard, 256, 257, 262
on incentives, 356–357, 366 Boldrin, Michele, 259
on inertia, 351–352, 364 Bolton, Patrick, 273
influences on development of, Borda, Jean-Charles de, 47, 84
126–127, 130 Bourguignon, François, 11n11
on information disclosure, 358–362 Bowles, Samuel, 435
on motivation, 390, 409 BRAC, 452, 460, 477
on probability judgments, 355–356 Brazil
on procrastination, 352, 364 fertility rates in, 39–40
on proximate causes of behavior, 406 growth diagnostics approach as
radical uncertainty and, 141 applied to, 273
on regulation, 357–358 inflation in, 163, 165
research strands in, 38–39, 350–358 social assistance in, 12n13
506 Index
Bretton Woods Agreement, 304–305 Cartwright, Nancy, 494
Brown, Annette N., 441–442 Caselli, Francesco, 280
Bruno, Valentina, 229, 238, 239 Cash-in-advance model, 170, 172, 181,
Buchanan, James, 51 182, 186
Burden sharing, 309, 310, 322 Cash transfer programs, 460–462, 4 70–
Bureau for Research and Economic 472, 477
Analysis of Development (BREAD), Categorical imperative, 89, 392,
443–445, 444f, 490 426–427
Burke, Edmund, 142, 143 Causality, 7, 234, 448
Business cycles, 28, 30, 240, 336, 340 CCT (conditional cash transfer)
programs, 460–462, 465–466, 477
Caballero, Ricardo, 195 CDS (credit default swap), 225–226, 226f
Cagan, Phillip, 160 Cesarini, David, 398
Calvo, Guillermo Cette, Gilbert, 261
on bank lending, 160n1 Chassang, Sylvain, 450, 453
on devaluation expectations, 166 Chile, inflation in, 165
on expectations, 162, 193, 194 China
on inflation indexation, 165 air quality-related deaths in, 308
on liquidity deflation, 179 banking expansion in, 241–242
on price stickiness, 339–340 debt concerns for, 196
on price theory of money, 196 entry into world economy, 250
on public debt configurations, 160, 163 Choice architecture, 349, 356–357, 366,
on regulatory requirements, 200 376–378
on reserve currencies, 172 Churchill, Winston, 85
on stabilization programs, 161 Cigarettes. See Smoking
on staggered prices, 167 Cioran, Emil, 78, 79, 86–87
on supply-side liquidity trap, 183, 202 CIP. See Covered interest parity
Cameron, Drew B., 441–442 Clark, John Bates, 29
Canfin, Pascal, 331–332 Classical liberalism, 374–375
Capability approach, 56, 68, 69, 97 Classical welfare theory, 374
Capital flows Climate change, 301–348. See also
contraction to emerging markets, 178 Greenhouse gas emissions
from low- to high-return countries, 240 assessment models for, 306–307, 329,
net and gross, 227, 235–237, 236f, 245 334–338
sudden stops in, 154, 169n12, 228 economic development and, 12–13,
Capitalism, 103, 111, 143, 168 314
Carbon pricing, 312, 324, 331–332, ethics of intervention, 309–310, 321,
344, 345 333–334
Carbon taxes, 306, 333, 402, 404 finances for mitigation of, 313,
Cardinal utility, 60, 65–66 330–333, 337
Carney, Mark, 330 geographic vulnerability to, 327, 328f
Carroll, Lewis, 48 Intergovernmental Panel on, 296, 307,
Carry trade, 221, 229, 238–239 323, 335
Index 507
international cooperation on, 304–305, regulation of, 268
314, 321, 322 Schumpeterian, 120, 256–259, 274
low-carbon transition and, 307–309, technology and, 127
312–313, 324–325, 336–338 Competitive equilibrium, 35
marginalist approach to, 306 Complementary slackness, 29–30
nonlinear dynamics of debt and, Complements, 29
339–345, 341f, 343–344f Composition effect, 258
Paris Agreement on, 13, 296, 304–305, Conditional cash transfer (CCT)
314, 321, 322 programs, 460–462, 465–466, 477
policy challenges related to, 310–313 Condorcet, Nicolas de, 47–48, 62, 84, 405
poverty and, 327, 402 Confidentiality. See Privacy
risks posed by, 295–296, 302–303, Confirmatory bias, 127
306–307 Conflicts of interest, 58, 110, 145–148
science of, 301–302 Congdon, William J., 455
uncertainty regarding, 307, 331, Consequentialism, 426–428, 427n14
337–338 Conservatism, 142–143
UN Framework Convention on, 312 Consumption booms, 158, 161–163,
urgency of, 303–304, 314, 326 163n6, 194
Climate-resilient development, 310–312 Contracts
Coase, Ronald, 116, 299–300 competition and, 115
Cognitive bias, 387 financial, 165–167
Cognitive tools, 38, 39 information economics on, 117–118
Collective action problems, 355, 379 insurance, 140
Collective bargaining, 126 optimal design of, 121
Collective Choice and Social Welfare (Sen), pooling, 140
63, 95 purpose of, 435
Collective decision-making, 84 regulation of, 110, 123
Collective rationality, 49–51 as response to market failures, 116
Communitarians, 375 social, 12n13
Comparative advantage, 26, 181–182, Coordination game, 408, 408nn20–21
248 Corrective taxation, 105, 109, 116, 299
Competition Corruption, 12, 12n13, 14, 456, 458, 462
in aggregative games, 392 Cost-benefit analysis, 77–79, 300, 331
Bertrand, 26, 115 Cournot, Antoine Augustin, 2, 5, 26
decrease in, 103 Covered interest parity (CIP), 209–212,
in domestic banking industry, 240 209n1, 221, 235, 244–248
imperfect, 101, 102, 105, 114–115, 125 Creative destruction
innovation-led growth and, 256–259, business-stealing effect of, 256
288 defined, 255, 288
for liquidity services, 202 firm dynamics and, 265, 272
perfect, 102, 105, 107n5, 125 innovation from, 260, 262, 263, 274,
price and, 25, 114–115 284, 323
productivity growth and, 256, 257 in low-carbon transition, 308
508 Index
Credit Card Accountability, foreign-currency-denominated, 177
Responsibility, and Disclosure Act of nominal, 160, 194
2009, 358 of nonbanks in emerging markets,
Credit default swap (CDS), 225–226, 226f 223, 225, 229
Cronyism, 34–35 nonlinear dynamics of, 339–345, 341f,
Cross-currency basis, 210–211, 212–213f, 343–344f
219, 223, 246, 246f public, 160, 163, 170, 199–200,
Currency. See also Money; specific 203–204
currencies sovereign, 167, 203
appreciation (see Currency Debt-deflation theory, 112n15, 165,
appreciation) 175, 177, 337
depreciation, 208, 221, 234 Debt securities, 217f, 229
devaluation of, 166, 177 Decarbonization efforts, 322, 323, 332
in foreign exchange swaps, 209–210 Dechezleprêtre, Antoine, 308
international funding, 223 Decision-making. See also Social choice
invoicing, 172, 214, 244 theory
reserves, 172–173, 177–180, 185, 332 behavioral economics and, 37
safe haven, 212, 248 bias in, 127
Currency appreciation collective, 84
contractionary nature of, 207 default rules and, 364, 383
in emerging markets, 208, 225, 234 discount rate in, 310
financial conditions impacted by, 227 distributional issues and, 96
real, 158, 161 ethics in, 79, 86
Swiss francs, 173 interactive, 48
US dollars, 172, 208, 221, 228, 234, 238 majority decisions, 52–53
Currency substitution, 178–179 moral, 428
rationality in, 82–83n1
Darwin, Charles, 389 social influences on, 354
Data collection methods, 457–459 switcher’s curse in, 142–143
Davidson, Donald, 63, 85 Default rules, 349, 351, 362–365, 383
Deaton, Angus, 68, 494 Deflation
Debreu, Gerard consequences of, 155
on aggregate demand, 339 debt-deflation theory, 112n15, 165,
collaboration with Arrow, 31 175, 177, 337
on conditions for market efficiency, in developed markets, 155
105, 120 in emerging markets, 154
convexity in proofs of, 114 explanations for, 179
on general equilibrium theory, 2, 34 generation of, 170, 182, 184, 186, 187
on risk markets, 105, 106 of liquidity, 179–184
The Theory of Value, 5 monetary policy and, 202
Debt as persistent threat, 183
configuration of public debt, 160, 163 spillovers from, 185
default on, 166–167 Delegation, 119, 123, 271, 272
Index 509
Demand governance challenges in, 290
aggregate, 111, 180–184, 202, 259, 339 greenhouse gas emissions in, 304, 327
changes in structure of, 102–103, 127 maturity of bond issues in, 237–238,
for climate protection, 311 238f
in equilibrium, 30 natural resource exploitation in, 298
in information economics, 104 policy evaluation in, 440
for liquid assets, 204 research agenda for, 240–242
for money, 159, 163, 181–183, 186 Development economics
price in relation to, 29, 356 as academic discipline, 440
for safe assets, 195–196, 203 behavioral, 42, 44
Demand curves, 26 causal questions in, 439
Demand functions, 25, 26, 197 climate change in, 335
Demand-side liquidity traps, 202 experimental random methods of
de Mel, Suresh, 464 analysis in, 266
Demirgüç-Kunt, Asli, 233 history and development of, 3–4, 4n3
Democracy, 85, 269, 269n13, 384, impact evaluation studies in, 441–442,
385, 411 443f, 490, 494, 496
Deontology, 426, 427n14 information revolution and, 139
Department of ___. See specific name of RCTs in (see Randomized controlled
department trials)
Dercon, Stefan, 312 research strands in, 447
Descriptive domain of inquiry, 6–7 Schumpeterian paradigm as bridge
Desiderata, 98, 99 between growth and, 267
Devarajan, Shantayanan, 12n13, 34 Development policy
Developed markets barriers to, 9–10
deflation in, 185 behavioral economics and, 10–11,
financial systems in, 175, 193 349–350, 382
inflation in, 153, 155 causality in, 7
interest rates in, 185 climate-resilient, 310–312
quantitative easing in, 176 data and theory in, 4–7
stability of, 154 environmental concerns in, 297
Developing countries. See also Emerging growth diagnostics approach to,
markets 273–275
antipoverty and social protection imperfect information in, 495
policies in, 465 information revolution and, 139
behavioral economics in, 386 innovations in, 440, 469–476
climate change vulnerability in, institutions and governance in, 11–12,
310–311 11n11, 12n12
economic and political interface in, 15 intuition in, 7–8, 15
environmental challenges in, 323 RCTs in, 463–469, 476–477, 491–492
favorable terms of trade shock in, 35 Schumpeterian paradigm in
firm dynamics in, 269–272, 270f formulation of, 272–276, 284,
formal credit markets in, 398n11 289–290
510 Index
Dewey, John, 88–89n6 firm dynamics and, 265–266, 269–272,
Dhillon, Amrita, 333n7 270f
Diamond, Peter A., 120 industrialization and, 261, 281–282
Dietz, Simon, 343 innovation and, 255–259, 263,
Difference principle, 65, 65n4, 68, 274–275
376n15 local lending opportunities during, 240
Disclosure of information, 103, 358–362. long-term, 142, 254, 255
See also Privacy methodological core of, 389–390
Discount rate, 78, 92, 309–310, 334, neoclassical model of, 253
342 political economy of, 255
Distortionary taxation, 410 post-global financial crisis, 201
DIV. See USAID Development resource allocation and, 240
Innovation Ventures Solow model of, 253–255
Dodd-Frank Wall Street Reform and taxonomy of growth experiences,
Consumer Protection Act of 2010, 280–283
147, 148, 149 technology and, 254, 282
Dominance axioms, 98 trust as factor in, 398
Draghi, Mario, 167, 194 Economic inequality, 129
Dual economy model, 3–4, 6 Economic rationality, 411
465–
Duflo, Esther, 266, 439, 448, 455, Economics
466, 495 behavioral (see Behavioral economics)
Dupas, Pascaline, 453–454 development (see Development
Dupuit, Jules, 27 economics)
Dutch disease, 34 environmental, 297, 299, 300, 324,
Dynamic stochastic general equilibrium 402–404
(DSGE) models, 336, 340 experimental, 390, 409
Dysfunctional institutions, 42, 44, 116 history and development of, 1–3, 25
of information (see Information
Easley, David, 421 economics)
East Asia institutional, 121, 130
financial crisis in, 237 Kantian morality and, 397–405
industrialization in, 281 of knowledge, 119–120
productivity growth in, 286 labor, 449
ECB. See European Central Bank macro (see Macroeconomics)
Econometrics, 12n13, 28, 32, 261, microeconomics, 5, 9–10, 112, 155
491, 501 natural resource, 298, 299, 324
Economic growth and development. neoclassical, 3, 44–45, 44f, 253, 285
See also Development policy; normative, 6, 81, 375
Schumpeterian growth paradigm politics and, 15–16
climate change and, 12–13, 314 positive, 81, 433
constraints on, 273–274, 297 publication bias in, 452, 475
environment and, 297–298 welfare (see Welfare economics)
finance in, 233 The Economics of Welfare (Pigou), 27, 56
Index 511
Eden, Maya, 205 Endogenous liquidity, 178–179
Edgeworth, Francis T., 56 Enlightenment, 47, 50
Edlin, Aaron, 142–143 Environmental economics, 297, 299,
Education 300, 324, 402–404
FAFSA procedures, 365 Environmental issues. See also Climate
growth-enhancing, 269, 273–275 change; Natural resources
investment in, 287 economic growth and, 297–298
lunch program enrollment, 363 externalities and, 105, 107n7, 392
productivity waves and, 261 pollution, 107n7, 299, 308, 323,
rate of return on, 273, 273 330n2, 351
for women, 40–42, 43f prosperity and, 297
El Salvador, growth diagnostics public policy and, 105, 299–301
approach as applied to, 273 regulatory, 105, 312
Ely, Jeffrey C., 419 Environmental Protection Agency
Emergency Planning and Community (EPA), 360–362
Right-to-Know Act of 1986, 362 Environmental valuation, 300
Emerging markets. See also Developing Equilibrium. See also General
countries equilibrium theory
appreciation of currency in, 208, 225, allocation, 403, 404
234 barter, 169, 198
CDS spreads for, 225–226, 226f in behavioral economics, 11
corporate borrowing by, 219, 229–230, competitive, 35
237–239 in evolutionary game theory, 418
cross-border lending to, 247, 247f goal functions in, 393
currency substitution in, 178 Homo moralis strategies, 397n10
deflation in, 154, 185 market, 37, 108n9, 446
fear of floating in, 172, 173, 177 mixed, 408n20
financial crises in, 154, 193 Nash, 400, 401f, 408n20, 418, 423, 433
greenhouse gas emissions in, 333n6 pooling, 108n9, 117, 140
inflation in, 153–155, 158 price, 117, 273
liquidity crunch in, 177 quantity, 117
maturity of bond issues in, 237–238, rational expectations, 164–166, 181
238f steady-state, 265, 341
risk-taking channel for, 225, 226f stock and flow, 249–250
safe assets in, 204 supply and demand in, 30
stabilization programs in, 161, 194 Escape competition effect, 257–259, 274
staggered prices in, 167 Eshel, Ilan, 420
strains in, 211 ESS. See Evolutionarily stable strategy
sudden stops in, 177 Ethics. See also Morality
US monetary policy effects on, 251 agent-based, 428
Empirical knowledge, 384 changes in standards of, 91
Employment. See Labor of climate change intervention,
Enculturated actors, 38, 44f, 45 309–310, 321, 333–334
512 Index
Ethics (cont.) motivation and, 390, 405–406,
in decision-making, 79, 86 409–410
situational, 88–89, 88–89n6 natural selection and, 8, 90, 393,
virtue ethics, 426, 428 395–396, 420
Europe. See also Eurozone; specific of preferences, 418–422, 428, 429
countries reason and, 90
behavioral economics in, 350 Evolutionarily stable goal functions, 393
cross-border lending to emerging Evolutionarily stable strategy (ESS), 393,
markets, 247, 247f 418–422, 429, 430
growth diagnostics approach as Evolutionary game theory, 417–421, 425
applied to, 273 Exchange rates
liquidity trap in, 337 bilateral, 172, 211, 225–226
on monopolies, 128 current account deficits and, 245
recovery from Great Recession in, 185 devaluation and, 166
European Central Bank (ECB), 167, 173, for euros, 211, 213f
176, 181–182, 194 financial channel of, 225
Euros flexibility of, 201
cross-currency basis and exchange rate in foreign exchange swaps, 209–210
for, 211, 213f real, 161
exchange rate for, 211, 213f risk-taking channel and, 221–223,
Federal Reserve policy effects on, 248 222f, 224f
in foreign exchange swaps, 209, 210f stabilization programs based on, 158,
as international funding currency, 223 161, 163, 194
as reserve currency, 182 for US dollars, 211–212, 212–213f
risk-taking channel for, 223, 224f Expectations. See also Rational
share of cross-border lending to expectations (RE)
emerging markets in, 247, 247f adaptive, 160, 161, 161n3
Eurozone of expectations, 164, 201
bankruptcy prevention in, 173 forward-looking, 160
debt crisis in, 194–195 for inflation, 160, 164–166, 171,
deflationary forces in, 157 194, 203
Great Recession in, 175 management of, 162, 163
interest rates in, 167 of price, 32, 199
productivity waves in, 261 for women, 40
risk premiums in, 181, 182 Expectations dominance, 164–166,
sovereign 10-year yields in, 194, 195f 193, 196
Evolution Expected utility, 82–83n1, 140–141
assortative matching and, 391–392 Expected value, 78, 353–354, 356, 390,
behavioral selection in, 394n5, 395 396, 423
drivers of, 393 Experimental economics, 390, 409
indirect approach to, 418–421 Exploitation
intuition and, 8 of asymmetric information, 107,
Kantian morality and, 392–397 107n5, 119, 145
Index 513
of conflicts of interest, 147 Financial crises. See also Global financial
of knowledge, 7 crisis
of markets and market power, 124, amplification mechanisms in,
128 200–201
Marx on, 103–104 Asian (1997–1998), 154, 177
of natural resources, 298, 330n1 in emerging markets, 154
technology and, 127 indicators of, 208
Externalities Lehman (2008), 168, 172, 173, 177, 184
absence in market efficiency, 105 resilience of money during, 169
competitive equilibrium and, 35 self-fulfilling, 193, 194, 198
defined, 299 sluggish recovery from, 170, 185–187,
environmental, 105, 107n7, 392 201–202
financial intermediation and, 205 Financial sector. See also Banks and
informational, 111–112, 115, 116 banking sector
intertemporal, 273 architecture of, 113–114
pecuniary (see Pecuniary externalities) asymmetric information in, 148,
technological, 27–28 150, 241
External validity of RCTs, 450–454, corporate governance in, 119
494, 502 globalization in, 239, 242
market failures in, 110
FAFSA (Free Application for Federal market value of equity in, 149–150,
Student Aid), 365
150f
Falk, Armin, 434–435 regulation of, 102, 110–111, 124,
Family Smoking Prevention and 150, 202
Tobacco Control Act of 2009, 361 transparency in, 103
Fankhauser, Sam, 295, 321–322, underpriced assets in, 106
334–335 Firms
Farhi, Emmanuel, 195 complexity of, 109–110, 124
Faustmann, Martin, 298 disclosure requirements for, 103
Fear of floating, 172, 173, 177 dynamics of, 265–266, 269–272, 270f
Federal Reserve (US) in emerging markets, 214
bank regulation by, 148 frontier, 257, 258, 268–269
Great Depression and, 155, 175 insurance market and, 106
in Lehman crisis, 173 laggard, 257, 258
as lender of last resort, 181–182 liquidity of, 202
monetary policy of, 246, 248, 249 marketing campaigns for, 492
quantitative easing by, 176, 378 market power of, 102, 114, 119
Fehr, Ernst, 406 pricing practices of, 114–115
Females. See Women profit for, 25, 184
Ferreira, Francisco H. G., 12n13 property rights of, 108
Fertility rates, 39–40 turnover in, 255
Fiat monies, 167, 168, 179, 186, Fiscal dominance, 163, 164
198, 207 Fisher, Irving, 112n15, 165, 175, 177
514 Index
Fisher, Ronald Aylmer, 439, 448 Gauri, Varun, 382
Flight-to-money phenomenon, 169 Gender inequality, 70
Food and Drug Administration, US, 361 General equilibrium theory
Foreign exchange markets as basis for models, 33
anomalies in, 209–212, 235, 245 for climate change analysis, 339
interest rates in, 209–211 criticisms of, 36
over-the-counter derivatives in, 219, econometrics and, 28
220f history and development of, 2, 5
Foreign exchange swaps, 209–210, inconsistency of equations in, 30
210f, 237 limitations of, 32
Forward rate of exchange, 209 money in, 13–14
Fossil fuels, 296, 301–302, 308, 312, profits and, 26–27
322, 323 proof of existence of, 2, 29, 30,
Foster, James E., 95 34, 35
Fourier, Jean-Baptiste, 301 returns to scale in, 26
Framed utility, 41, 43f on system effects, 29
Framing, 45, 353–354, 359 under uncertainty, 36
Fraud, 123, 123n32, 124, 447 General Theory (GT)
Free Application for Federal Student Aid conceptual foundations of, 141
(FAFSA), 365 on demand for money, 181
Free riding, 304, 400, 402 expectations of expectations in, 164
Freund, Caroline, 34, 290 flight-to-money phenomenon in, 169
Friedman, Milton, 155, 158–160, history and development of, 155
175, 397 on value of money, 171
Frontier firms, 257, 258, 268–269 Germany
Fuchs-Schündeln, Nicola, 434 behavioral economics in, 382
Fuel economy disclosures, 360–362 inflation in, 13
as lender, 167, 181
Gaertner, Wulf, 55 outward portfolio flows of insurance
Game forms, 55, 421 companies from, 217f, 219
Game theory Ghana, economic and political interface
aggregative games, 392 in, 15–16
coordination game, 408, 408nn20–21 Giné, Xavier, 433, 435–436
evolutionary, 417–421, 425 Gini coefficient, 264, 264f
history and development of, 2 Giraud, Gaël, 326, 339, 340, 344–345
measurement and aggregation Glass-Steagall Act of 1933, 148
techniques in, 85 Glennerster, Rachel, 451, 464
public goods game, 423–424 Glewwe, Paul, 439–440
social system predictions from, 430 Global assets, 214
symmetry in, 399, 418, 421–422 Global Financial Development Report
trust game, 398–400 2015/16, 237
ultimatum bargaining game, 406 Global Financial Development Report
Garcia-Macia, Daniel, 288 2017/18, 239
Index 515
Global financial crisis (2008–2009) of information, 103
CIP deviations during, 209 institutional frameworks for, 125
collapse in safe assets following, 195 negative impact of failures in, 386–387
discipline of economics impacted by, Grafen, Alan, 419
1, 193, 197 Grandjean, Alain, 331–332
explanations for, 200, 202 Great Depression, 111, 155, 157, 165, 175
lessons learned from, 145–148 Great Inflation, 153, 155
post-crisis economy, 201–204 Great Recession. See also Global
securitization market and, 123, financial crisis
123n32, 145–147 deflation and, 154
sluggish recovery from, 201–202 General Theory and, 155
US dollar appreciation in, 228, 234 macroeconomics impacted by, 197
Globalization monetary policy prior to, 168, 173
banking and, 212–219, 215–218f, persistence of, 175
220f, 239–242 securitization market prior to, 123,
in financial sector, 239, 242 123n32
income distribution and, 103 sluggish recovery from, 185
technology and, 260, 262 sudden stops during, 169n12
Global liquidity, 237, 240, 244, 246 Greece, debt crisis in, 194–195
Global warming. See Climate change Green Climate Fund, 331
Gneezy, Uri, 435 Greenhouse effect, 301
Goal functions, 390–394, 393n2, 395n7 Greenhouse gas emissions
The Good, the Bad, and the Ugly (film), 88 accumulation of, 303, 306
Goods and services. See also Public goods atmospheric concentration of, 302, 303
availability of, 24–25 in burden sharing arguments, 310
as complements and substitutes, 29 in developed vs. developing countries,
consumption booms, 158, 161–163, 304, 327
163n6, 194 in emerging markets, 333n6
demand curves for, 26 human activity and, 302
free, 29–30 infrastructure and, 313
markets for, 31–32 market failures related to, 312
nontradable, 165 mitigation of, 321–325, 329
primary, 68 Greenspan, Alan, 249
Gopinath, Gita, 193 Greenwald, Bruce, 107, 107–108n8,
Gordon, Robert, 259 112, 118, 121
Gorton, Gary, 178 Griffith, Rachel, 256, 257
Gossen, Hermann Heinrich, 2 Grossman, Sanford J., 108
Governance Growth diagnostics approach, 273–275
behavioral economics for, 387 GT. See General Theory
corporate, 119, 125, 126 Gueron, Judy, 455
in developing countries, 290 Guidotti, Pablo, 160
in development policy, 11–12, Güth, Werner, 418, 420
11–12nn11–12 Guzman, Martin, 113
516 Index
Hahn, Frank, 169, 170, 174 in evolutionary game theory, 418, 419
Hallegatte, Stephane, 327 material payoffs and, 394–396
Hamilton, William D., 394n5, 419 as moral theory, 426–429
Hamilton’s rule, 394n5 preferences of, 394–397, 406–408,
Hammond, Peter J., 81 420, 425–427, 433–436
Handbook of Field Experiments in symmetric models, 421
(Banerjee & Duflo), 448, 451, 455 tax compliance and, 408
Hanna, Rena, 461 in trust game, 399
Hansen, Alvin, 259 utility functions of, 394, 396, 397,
Hare, Richard M., 83 403–404
Harsanyi, John, 63, 82–83, 410, 422, voluntary contribution to public
424–427, 425n11 good, 400–402
Hartwick, John, 299 voting behavior of, 405
Hausmann, Ricardo, 273–275 welfare economics of, 410–411
Hayek, Friedrich, 31 Homo oeconomicus
Head-count measure of poverty, 68 consumption of harmful goods,
Health and Human Services 403, 410
Department, US, 360 in coordination game, 408
Heckman, James J., 450–451 defined, 395n7
Hedonic utility, 399, 403, 411 in evolutionary game theory, 418, 419
Hemous, David, 262 material payoffs and, 395, 396, 433
Henrich, Joseph, 434 policy based on, 410
Hicks, John, 2, 31, 32, 300 preferences of, 395–397
High-frequency trading, 122, 122n29 in public goods game, 423–424
Hirshleifer, Jack, 108n9 regulation for, 436
Hispanics, 363 tax compliance and, 407
Hockett, Robert, 372 in trust game, 399
Hoff, Karla, 37, 41 violation of assumptions for, 434
Homeland Security Department, US, 364 voluntary contribution to public
Homo kantiensis good, 400
in coordination game, 408 voting behavior of, 405
in evolutionary game theory, 419 Hotelling, Harold, 298, 299
material payoffs and, 395, 396, 433 Howitt, Peter, 253, 255, 259, 284
preferences of, 395, 396, 427 Hsieh, Chang-Tai, 269, 271, 272, 288
in public goods game, 423–424 Human capital, 240, 285–287, 299,
regulation for, 436 463, 485
in trust game, 399 Hume, David, 425, 435, 435n1
Homo moralis Hungary, inflation in, 13
consumption of harmful goods, Hussam, Reshmaan, 458
403–404 Hyperinflation, 13, 161, 162n5, 194, 255
in coordination game, 408, 408n21
empirical validity of, 410 Ilzetzki, Ethan, 184
equilibrium behaviors, 397n10, 406 Imbens, Guido W., 448, 450, 454
Index 517
IMF. See International Monetary Fund private, 113–114, 205
Impact evaluation studies, 441–442, for public good contributions, 401
443f, 490, 494, 496 for quasi-money creation, 173, 205
Imperfect competition, 101, 102, 105, rational expectations and, 159
114–115, 125 separability assumption and, 436
Imperfect information. See also in sharecropping, 121
Asymmetric information social, 113–114, 205, 435–436
competition and, 102, 107n5, for time inconsistency, 156
114–115 Income
consequences of, 121, 125 distribution of, 98, 103, 254, 285, 333
in development policy, 495 generation of utility from, 56
importance of, 104 inequality in, 262–264, 280
incentives for creation of, 109 innovation and, 262, 263f
market efficiency and, 107, middle-income trap, 274, 281–282n1,
107–108n8 281–283
radical uncertainty as, 141–143 in nonlinear dynamic model, 340
second welfare theorem and, 102n2 in poverty measurement, 68–69
Imperfect policy credibility, 158–161 India
Impossibility theorem air quality-related deaths in, 308
as foundation for research agenda, 6 behavioral economics in, 382
in social choice theory, 48–51, 58, 62, corruption in, 456
65–66, 82, 86 economic growth in, 267, 275
social welfare and, 49–50, 58, 65–66 education for women in, 41–42, 43f
Incentives firm dynamics in, 269–272, 270f, 288
choice architecture and, 356–357, 366 laws for village council candidates in,
for decarbonization, 332 385, 385n1
for exploitation of asymmetric poverty rates in, 267
information, 107, 107n5, 119, 145 RCTs in, 446
for firm complexity, 109–110, 124 social impact of hiring female villagers
for free riding, 304 in, 40, 41t
for growth, 272 Indirect evolutionary approach,
for human capital investment, 287 418–421
for imperfect information creation, Individual preferences
109 materiality of, 425
importance of, 349 measurement of, 79, 80
for inflation, 159, 160, 194 orderings and, 49, 50, 83
for information collection and relationship to social preferences,
processing, 126 51, 58
for innovation, 129, 256, 284, 288 social welfare and, 50, 60, 410
in insurance sector, 108 value restriction and, 52
for liquid asset creation, 178 Indonesia, rice distribution program in,
material, 356, 366 462–463
pecuniary, 404, 410 Industrialization, 261, 281–283, 302
518 Index
Inequality Information. See also Knowledge
under capitalism, 103 asymmetric (see Asymmetric
economic, 129 information)
gender, 70 barriers to dissemination of, 108–109
general equilibrium theory and, 35 broadening bases of, 96–97
growth of, 103, 126 cognitive tools for processing, 38, 39
income, 262–264, 280 disclosure of, 103, 358–362 (see also
innovation and, 262–265, 264f Privacy)
measurement of, 70 distributive effects of, 108, 108n9
political, 129 economics of (see Information
of utility, 62 economics)
Inequity aversion, 406, 407 framing and presentation of, 45,
Inertia 353–354, 359
default rules and, 351, 364 governance of, 103
in earth’s climate system, 321 imperfect (see Imperfect information)
of ecosystem response to carbon objective processing of, 44
emissions, 343 perfect, 104, 105, 110, 115
innovation and, 308 production of, 104, 111
intellectual, 168 as public good, 104–105, 104–105n3,
in price, 336 108, 109
Inflation reliability of, 201
in developed markets, 153, 155 in social choice theory, 57, 60–65,
in emerging markets, 153–155, 158 70–71
expectations for, 160, 164–166, 171, subjective processing of, 44
194, 201 in utilitarian welfare economics, 56
fiscal dominance and, 163 Informational externalities, 111–112,
hyperinflation, 13, 161, 162n5, 115, 116
194, 257 Information economics, 101–131
imperfect policy credibility and, asymmetric information and,
158–161 109–111
incentives for, 159, 160, 194 on contracts, 117–118
indexation of, 165, 165n8, 166 delegation in, 119
in Keynesian context, 155–156 financial architecture and, 113–114
monetary policy and, 202 history and development of, 101,
in nonlinear dynamic model, 104–106
341–343 on imperfect competition, 114–115,
runaway, 13, 178 125
spikes in, 159 institutions and, 101, 121, 130
stabilization programs, 161–167, macroeconomic contributions of,
162n5, 163n6, 194 111–112, 112n15
theory and practice, 158 market efficiency and, 101–103,
uncertainty of, 165 105–109, 120–121
during World War II, 155 on market failures, 115–116, 118–119
Index 519
policy implications of, 122–124, in development policy, 11–12, 11n11,
130–131, 150 12n12
production in, 111 disclosure requirements for, 362
robustness of, 117–118 dysfunctional, 42, 44, 116
technology and, 127–129 in emerging markets, 154, 163
theoretical impacts of, 126–127, 130 of growth, 255, 267–268, 289
theory of second best and, 112–113 information economics and, 101,
Information revolution, 104–106, 126, 121, 130
139–140, 143, 150 low-carbon, 336
Innovation in mechanism experiments, 456
in antipoverty policy, 465 preferences shaped by, 434
barriers to, 109, 255 RCT support from, 478–479
competition and, 256–259, 288 rise of, 115
from creative destruction, 260, 262, social, 42, 44, 116
263, 274, 284, 323 transparency of, 99
in data collection, 457–459 Insurance Development Forum, 330
democracy and, 269, 269n13 Insurance sector
in development policy, 440, 469–476 asymmetric information in, 44,
economic growth and, 255–259, 263, 117, 118
274–275 contracts in, 140
financial, 10 in emerging markets, 203, 204
firm dynamics and, 265, 266 liquidity crunch and, 175
incentives for, 129, 256, 284, 288 market failures in, 115, 116, 140
income and, 262, 263f outward portfolio flows, 217–218f,
inequality and, 262–265, 264f 219
in information economy, 127 Pareto efficiency in, 44, 106
in low-carbon transition, 308 risk markets and, 106, 108
patent protection for, 258–259 Intellectual property rights, 103, 105n3,
productivity growth from, 260, 268, 108–109, 120
274, 288 Inter-American Development Bank,
profitability from, 255–258, 274 465
property rights and, 255, 289 Interest rates
in RCTs, 454 of central banks, 183, 184
social mobility from, 263, 264f, 265 covered interest parity and, 209–212,
spillovers from, 120, 256, 308 209n1, 221
taxation and, 265 in developed markets, 185
theory of, 119 in foreign exchange markets,
Institutional economics, 121, 130 209–211
Institutions global, 249, 250f
analysis of, 421 on government bonds, 165
climate resilient, 309 in growth diagnostics approach, 273
default rules of, 351, 362 mechanism experiments and,
455–456
520 Index
Interest rates (cont.) IRC (International Rescue Committee),
as monetary policy instruments, 466
158, 172 Ivashina, Victoria, 247, 248
negative, 14, 259
nominal, 162, 163n6, 195 Japan
normalization of, 208 competition in, 268
peso problem and, 166 deflationary forces in, 157
in post-crisis economy, 202 liquidity of firms in, 202
real, 165, 176, 195, 249 liquidity trap in, 337
for reserve currencies, 178–179 monetary easing in, 223
strategies for lowering, 167 outward portfolio flows of insurance
of US dollars in foreign exchange companies from, 217f, 219
swaps, 209–210, 210f productivity waves in, 261
Intergovernmental Panel on Climate total-factor productivity growth in,
Change (IPCC), 296, 307, 323, 335 261, 262
Internal validity of RCTs, 450, 494 Jefferson, Thomas, 91
International Monetary Fund (IMF), Jensen, Robert, 40
160n2, 204, 247, 332 Jevons, William Stanley, 2, 29, 80–81,
International Rescue Committee 298
(IRC), 472 Jobs. See Labor
Interpersonal comparisons Jones, Benjamin F., 286
analytical framework for, 63–64, 83, J-PAL. See Abdul Latif Jameel Poverty
89, 91 Action Lab
extent of, 64–65 Justice, theories of, 57, 68, 333–334,
impartiality in, 82–83 374–376
limitations of, 85, 86
of mental states, 63, 87–88, 92 Kahneman, Daniel, 38
multidimensional, 70 Kaldor, Nicholas, 300
opposition to, 81 Kanbur, Ravi, 139
self-reported measures in, 86–87 Kant, Immanuel, 389, 394–395, 426–427
in social choice theory, 62–67, 70, Karlan, Dean, 455, 461
81–83 Kasy, Maximilian, 495
of utility, 56, 57, 60–62, 66, 81–82 Keohane, Robert, 324
in welfare economics, 56, 57, 60–66, Keynes, John Maynard. See also General
70, 85 Theory
of well-being, 59, 64, 66, 79, 80, 86 on Bretton Woods Agreement, 304
Interventionism, 142, 143 on current interest parity, 246
Intuition, 7–8, 11, 15, 38 on flight-to-money phenomenon, 169
Invariance axioms, 98 interventionism of, 142, 143
Invariance conditions, 64 macroeconomic analysis and, 6, 111
Invoicing currencies, 172, 214, 244 on power of ideas, 16
IPCC. See Intergovernmental Panel on on skepticism, 8
Climate Change Kim, Jim Yong, 402
Index 521
King, David, 324–325 Lancaster, Kelvin, 112
Klenow, Peter, 269, 271, 272, 288 Latin America. See also specific countries
Klette, Tor Jakob, 265, 271 financial crisis responses in, 177, 178
Knowledge. See also Information growth diagnostics approach as
acquisition of, 7–8 applied to, 273–274
dissemination of, 400 poverty reduction in, 465
economics of, 119–120 Latinos/Latinas, 363
empirical, 384 Lecat, Remy, 261
exploitation of, 7 Legovini, Arianna, 441
gaps in, 7, 12, 496 Lehman crisis (2008), 168, 172, 173,
intertemporal, 256, 273 177, 184
production of, 109, 120 Leite, Phillipe G., 12n13
as public good, 105, 120 Lenders of last resort, 167, 174, 177,
restrictions on use of, 109, 120 181–182, 194, 199
scientific, 8, 81 Leontief, Wassily, 335
Koch, Catherine, 221, 223 Le Treut, Hervé, 329
Kocherlakota, Narayana, 335 Levine, David K., 259
Koçkesen, Levent, 420 Lewis, Arthur, 3–4, 6, 15–16
Kortum, Samuel, 265, 271 Lewis, Randall A., 492
Kosenko, Andrew, 117 Lexicographic form, 65n4
Kraay, Aart, 284 Liberalism, 374–376
Kremer, Michael, 266, 439–440, 459– Liberal paradox, 54
460, 464 Libertarian paternalism, 357, 376,
Krugman, Paul, 5 383–384
Liberty, 53–55
Labor Lieberman, Joseph, 84
excess supply of, 5–6 Life-coaching, 461, 462
financial crises and, 111–112, 202 Lindahl, Eric, 31
fixed quantity of, 3 Liousse, Catherine, 327
inflation and, 156 Lipsey, Richard G., 112
liquidity in restoration of, 184 Liquid assets
in low-carbon transition, 308, 313 defined, 154
in nonlinear dynamic model, demand for, 204
340–343 efficient supply of, 194
opportunities for, 312 fragility of, 179
Pigou effect and, 180 incentives for creation of, 178
as primary factor of production, 27 in operation of modern economics, 197
rates of return on, 273 output backstop of, 185
reallocation of, 274 price flexibility of, 174
Labor economics, 449 reliability of, 175, 180
Labor productivity growth, 261, 306, supply of, 204
340–342 as transactions facilitators, 169
Laggard firms, 257, 258 valuation of, 198
522 Index
Liquidity informational externalities in, 111–112
of assets (see Liquid assets) information economics and impact
crunch (see Liquidity crunch) on, 111–112, 112n15
definitions of, 154–155 interest rates and, 249
deflation of, 179–184 Keynesian, 6, 111
endogenous, 178–179 liquidity in, 157, 168, 197
fragility of, 185, 195, 197, 199–200 poverty and, 267
global, 237, 240, 244, 246 rational expectations in, 153
high-frequency trading justified by, reassessment of, 197
122n29 relationship with microeconomics,
in macroeconomics, 157, 168, 197 9–10, 155
of money, 169, 179–181 Majority decisions, 52–53
of reserve currencies, 172, 180 Malthus, Thomas, 12, 297–298, 300, 329
shortages of, 170, 184, 185, 193, 202 Manchester School (Lewis), 4
sudden stops and, 155, 169, 175 Mansuri, Ghazala, 435–436
traps (see Liquidity traps) Mantel, Rolf, 339
Liquidity crunch Manuelli, Rodolfo, 286, 287
credit flow interruption through, Marginal utility, 2, 403, 411
175, 178 Market efficiency
defined, 155, 186 asymmetric information and, 106–107,
in emerging markets, 177 118
in global financial crisis, 202 conditions for, 105, 105n4, 120–121
in Lehman crisis, 168 in decentralized price systems, 104
policy responses to, 176 information economics and, 101–103,
sluggish recovery due to, 187 105–109, 120–121
Liquidity traps Pareto efficiency, 37, 105, 107
advanced economies in, 337 price and, 104, 106–108, 108n8
arguments against relevance of, 180 in provision of public goods, 105
demand-side, 202 Market equilibrium, 37, 108n9, 446
flight-to-money phenomenon and, 169 Market failures. See also Global financial
shocks and, 155 crisis
supply-side, 181–185, 202–203 absence of risk markets and, 106
use of term, 157 adverse selection and moral hazard
Logical positivism, 57, 63, 81, 85 in, 112
Lopez, Jimmy, 261 asymmetric information as cause of,
Low-carbon transition, 307–309, 145
312–313, 324–325, 336–338 behavioral, 357
Lucas Paradox, 255 environment-related, 298, 299, 312
explanations of, 115–116, 121n27
Machina, Mark J., 140 in financial sector, 110
Macroeconomics information, 107, 108, 118–119
criticisms of, 197, 283, 334–339 poverty and, 461
in environmental economics, 324 responses to, 116
Index 523
Market-moving strategies, 377–378 Measurement
Market power, 102, 109, 114, 119, of cross-country and over-time
124–130, 124n34 differences, 285–287
Markets desiderata and, 98, 99
anomalies in, 209–212, 235, 245 of inequality, 70
developed (see Developed markets) normative, 68–70
efficiency of (see Market efficiency) of poverty, 68–70, 97
emerging (see Emerging markets) in RCTs, 457–459
exploitation of, 124 social choice theory and, 68–70, 95–99
failures in (see Market failures) of variable properties, 97
foreign exchange, 209–212 Mechanism experiments, 455–456
for goods and services, 31–32 Meghir, Costas, 269
imperfections in, 102 Mendoza, Enrique G., 184
integration of, 434 Menger, Carl, 29–30
perfect markets paradigm, 101 Mental states, 63, 67, 87–88, 92
risk, 105, 106 Mercier, Hugo, 90
Marshall, Alfred, 29, 56, 299 Mertens, Jean-François, 333n7
Martin, Ralf, 308 Mestrallet, Gérard, 331–332
Marx, Karl, 27, 103–104 Metrick, Andrew, 178
Material incentives, 356, 366 Mexico
Material payoffs CCT programs in, 460–462, 465
altruism and, 406–407 currency devaluation in, 166
consequences for, 392, 393 firm dynamics in, 269, 270f
in consumption models, 403, 404 “Tequila” crisis (1994/1995), 154
in coordination game, 408 Mian, Atif, 202
in evolutionary game theory, 418–420 Microeconomics, 5, 9–10, 112, 155
Homo kantiensis and, 395, 396, 433 Middle East, breakdown of social
Homo moralis and, 394–396 contract in, 12n13
Homo oeconomicus and, 395, 396, 433 Middle-income trap, 274, 281–283,
linear, 418n3 281–282n1
maximization of, 391, 395n7, 400 Miliband, David, 466
Nash equilibria in terms of, 408n20, Mill, John Stuart, 26, 53–54, 300
433 Millennium Development Goals, 295,
as personal utility, 411, 425 466
in public goods game, 423 Mincerian wages, 273
in trust game, 399 Minsky moment, 200, 330
Maynard Smith, John, 8, 393, 410 Mishra, Anjini, 441–442
MBS. See Mortgage-backed securities Mixed equilibrium, 408n20
McKenzie, David, 464, 488, 492 Mohnen, Myra, 308
McKenzie, Lionel, 2 Monetary policy. See also Money
McNamara, Robert, 494 deficiencies in understanding of, 13–14
Meade, James E., 112, 300 expansionary, 202–203
Meager, Rachael, 451–452 General Theory on, 155
524 Index
Monetary policy (cont.) reflective, 89n6
governance of, 125 trust and, 398–400
instruments of, 158, 170 voluntary contributions to public
prior to Great Recession, 168, 173 goods and, 400–402, 401f, 406, 407
staggered prices in, 167 voting behavior and, 405
trade-offs in, 239 Morality profiles, 395, 396
unconventional, 170, 184, 196 Moral preferences, 410, 425
of US Federal Reserve, 246, 248, 249 Morgenstern, Oskar, 30
Money. See also Currency; Monetary Mortgage-backed securities (MBS), 146,
policy 147, 174, 378
backing of, 169, 179, 180, 198–199 Motivation, evolutionary foundations
demand for, 159, 163, 181–183, 186 of, 390, 405–406, 409–410
fiat, 167, 168, 179, 186, 198, 205 Mottaghi, Lili, 12n13
flight-to-money phenomenon, 169 Moulin, Sylvie, 439–440
in general equilibrium theory, 13–14 Mullainathan, Sendhil, 37, 456–457
liquidity of, 169, 179–181 Mundell-Fleming model, 225, 228
output value of, 170–171, 173
price theory of, 169, 171–172, 172n17, Nash equilibrium, 400, 401f, 408n20,
179, 196, 198–199 418, 423, 433
quasi, 173–174, 180, 205 Nationally Determined Contributions
resilience of, 169, 198 (NDCs), 322, 338
supply of, 161, 163, 176, 179–183, National School Lunch Act of 2012,
186–187, 202 363
Monga, Célestin, 77 Natural monopolies, 106, 128
Monopolies Natural resource economics, 298, 299,
distortions in, 115, 128 324
grabbing of rents by, 125 Natural resources
natural, 106, 128 constraints on, 298, 313, 329
power of, 34–35, 128–129 depletion of, 299, 330
pricing in, 114 exploitation of, 298, 330n1
regulation of, 106, 128 management of, 298–299
Montesquieu, Charles Baron de, 298 Natural selection, 8, 90, 393, 395–396,
Moral hazard, 112, 114–117, 455 420
Morality. See also Ethics NDCs (Nationally Determined
altruism vs, 407–408 Contributions), 322, 338
consumption of harmful goods and, Necker, Jacques, 48
402–404, 407 Negative emissions technology, 303
in decision-making, 428 Negative Income Tax (NIT) experiment,
degree of, 396–404, 433–436 454–455
economics and, 397–405 Neoclassical economics, 3, 44–45, 44f,
ethical standard changes and, 91 253, 285
evolution and, 392–397 Neutrality, 61, 62
general rules of, 429 Newbery, David M. G., 112
Index 525
New Climate Economy report (2014), Pareto principle, 50, 54, 65n4
323, 331, 344 Paris Agreement (2015), 13, 296,
New welfare economics, 57–58, 96 304–305, 314, 321, 322
Neyman, Jerzy, 448 Partial orderings, 98–99
Nkrumah, Kwame, 15, 16 Patent protection, 258–259
Nominal interest rates, 162, 163n6, 195 Paternalism, 357, 376, 383–384
Nordhaus, William, 306, 306n2 Patient Protection and Affordable Care
Normative economics, 6, 81, 375 Act of 2010, 359
Normative measurement of well-being, Patriarchal social order, 41, 42, 43f
68–70 Pattanaik, Prasanta K., 55
Norms. See Social norms Payroll statements, paper to electronic
North East Universities Development transition for, 364
Consortium Conference, 445, 445t Pecuniary externalities
Nozick, Robert, 54, 55 asymmetric information and, 107
Nucifora, Antonio, 34, 290 for atomistic agents, 179
Nudges internalization of, 182
design of, 349 manifestation of, 112
in development policy, 473 monetary policy and, 202
effectiveness of, 38, 356 in proof of market efficiency,
framing and, 45 107–108n8
global, 384 technological externalities vs., 27
as paternalism, 383 Pecuniary incentives, 404, 410
Nutritional disclosures, 358–359, 361, 364 Pension Protection Act of 2006, 363
Perfect competition, 102, 105, 107n5,
Obesity, 354, 356, 359 125
Obstfeld, Maurice, 244 Perfect information, 104, 105, 110, 115
Ok, Efe A., 420 Perfect markets paradigm, 101, 113
Olken, Benjamin A., 458 Personal preferences. See Individual
Opportunity aspect of liberty, 53–55 preferences
Orderings, 49, 50, 98–99 Persons of influence, 14
Ordinal utility, 60 Persson, Martin U., 334
Organisation for Economic Peso problem, 166
Development and Co-operation Peters, Michael, 271–272
(OECD), 350 Phillips curve, 155–156, 339
Phishing for Phools (Akerlof & Shiller), 124
Pareto comparison, 57–58 Pigou, Arthur Cecil, 27, 31, 56, 299
Pareto efficiency Pigou effect, 180, 181
institutional interventions and, 116 Policy credibility, 156, 158–163
in insurance sector, 44, 106 Politics
liberty and, 54 African American influence on, 59
of markets, 37, 105, 107 economics and, 15–16
in new welfare economics, 58 inequality in, 129
welfare theorems and, 102n2, 105, 118 market power and, 129
526 Index
Pollution, 107n7, 299, 308, 323, 330n2, peer influences on, 37
351 in population state, 393
Pooling equilibrium, 108n9, 117, 140 rationality of, 82
Population state, 393 single-peaked, 52
Positive economics, 81, 433 social construction of, 417, 423, 429
Positivism, logical, 57, 63, 81, 85 of voters, 59
Poverty Price. See also Deflation; Inflation
antipoverty policy, 465, 495 of assets, 201
behavior of those in, 266 competition and, 25, 114–115
climate change and, 327, 402 demand in relation to, 29, 356
consequences of, 90 discovery of, 122
as deprivation of capabilities, 69 Dutch disease and, 34
graduation approach to, 452 equilibrium and, 117, 273
macro and systemic factors in, 267 expectations of, 32, 199
market failures and, 461 flexibility in (see Price flexibility)
measurement of, 68–70, 97 in Great Depression, 165, 175
micro interventions and macro effects impersonal nature of, 37
on, 9 indexation of, 160
multidimensional approach to, 99 market efficiency and, 104, 106–108,
orderings, 98, 99 108n8
political economy of, 300 as reflection of economic state, 274
reduction of, 295–297, 312, 315, 465 setting in advance, 171, 171n16
in Sustainable Development Goals, 295 staggered, 167–169, 171, 196, 198
Prantl, Susanne, 259 sticky (see Price stickiness)
Preferences. See also Social preferences; Price, George R., 8, 393, 410
Individual preferences Price flexibility
aggregating, 125 inflation control and, 167
altruistic, 406–407 of liquid assets, 174
antisocial, 390, 420 monetary fragility resulting from, 198
changes in, 40 nominal, 169, 182
in classical welfare theory, 374 perfect, 172, 186
endogeneity of, 373–374 Pigou effect and, 180
evolution of, 418–422, 428, 429 in price theory of money, 172
expected utility representation of, 140 Price stickiness
Homo kantiensis and, 395, 396, 427 backing of money and, 179, 180, 199
Homo moralis and, 394–397, 406–408, consumption boom and, 163n6
420, 425–427, 433–436 explanations for, 196
Homo oeconomicus and, 395–397 nominal, 198
institutional shaping of, 434 relaxation of, 339–340
Kantian, 426 Price theory of money (PTM), 169,
methodological, 496 171–172, 172n17, 179, 196, 198–199
methods for influencing, 378 Primary goods, 68
moral, 410, 425 Pritchett, Lant, 466
Index 527
Privacy PROGRESA experiment, 440, 460–462,
debates regarding, 103 465, 477
default rules and, 365 Property rights
rights to, 53–54, 129 innovation and, 255, 289
Private incentives, 113–114, 205 intellectual, 103, 105n3, 108–109, 120
Private-label assets, 199–200, 203 patent protection, 258–259
Probability judgments, 355–356 Prospect theory, 356
Process aspect of liberty, 53–55 Pseudo-wealth, 113
Procrastination, 352, 364 PTM. See Price theory of money
Procyclicality, 219, 240 Publication bias, 452, 475
Production Public debt, 160, 163, 170, 199–200,
fixed coefficients in, 29 203–204
of information, 104, 111 Public goods
of knowledge, 109, 120 in aggregative games, 392
management of, 271 defined, 104n3
nonconvexity in, 114 information and knowledge as,
primary factors of, 26 104–105, 104–105n3, 108, 109, 120
of quasi-monies, 205 Kantian morality and, 400–402
time considerations in, 30–31 management of, 48
Productivity growth reduced climate risks as, 304
competition and, 256, 257 voluntary contributions to, 400–402,
constraints on, 287–288 401f, 406, 407
correlation with firm/job turnover, 255 Public goods game, 423–424
cross-country and over-time Public policy
differences in, 287 analysis of, 273
education and, 269 anti-trust, 105–106
financial system structures and, 275 behavioral economics and, 350, 382
industrial policy impact on, 266 cost-benefit analysis of, 77, 300
from innovation, 260, 268, 274, 288 environmental, 105, 299–301
labor, 261, 306, 340–342 necessity of, 102
peer effects and, 387 purpose of, 435, 435n1
technology and, 261, 286 Purchasing power, 26, 29, 158, 169,
total-factor, 260–262 179, 332
waves of, 261
Profits Quantitative easing (QE), 176, 183,
of banks on risky bets, 147, 149 203, 383
excessive, 25 Quantity equilibrium, 117
front running and, 122 Quasi-monies, 173–174, 180, 205
general equilibrium theory and, 26–27 Quasi-rational actors, 38, 44f, 45
information asymmetries and, 107
from innovation, 255–258, 274 Radical uncertainty, 141–143
from market exploitation, 124 Rand Health Insurance experiment,
maximization of, 36, 110, 176 455
528 Index
Randomization bias, 450–451, 495, 497 misconceptions regarding, 161
Randomized controlled trials (RCTs), policy credibility and, 156
439–498 usefulness of, 153
academic nature of, 463 Rationality
assessment of policy success of, assumptions of, 89–91
467–469 collective, 49–51
data collection in, 457–459 in decision-making, 82–83n1
development policy influenced by, economic, 411
463–469, 476–477, 491–492 individual, 180
development research influenced by, of preferences, 82
447–463, 490–491 tyrannies of, 92
growth of, 441–445, 441–446f, 488– Ravallion, Martin, 12n13, 494
490, 489t Rawls, John
history of, 439–440, 494n1 difference principle of, 65, 65n4, 68,
identification of causal effect in, 376n15
448–450 theory of justice and, 57, 68, 374, 376
innovation in, 454 on veil of ignorance, 83, 374, 422
institutional support for, 478–479 RCTs. See Randomized controlled trials
limitations of, 495–497 RE. See Rational expectations
mechanism experiments, 455–456 Real interest rates, 165, 176, 195, 249
nonexperimental influences on, Reasoned intuition, 7, 8
449–450 Recession of 2007–2008. See Great
observing the unobservable in, Recession
454–457 Recursive collective action problems,
output and substitution effects of, 379
496 Reflections on the Revolution in France
pathways of influence for, 477–478 (Burke), 142
range of projects using, 445–447 Refugees, cash transfer programs
registry of, 452, 478 for, 472
in same settings, 459–460 Regianni, Giovanni, 458
structured speculation in, 453 Regulation
unpacking interventions with, anti-trust policies, 105–106, 126
460–463 bankruptcy law, 114, 126
USAID DIV case study, 440–441, 469– of banks, 110, 146–148, 241
476, 472t, 492 in behavioral economics, 357–358
validity of, 450–454, 494, 496 of competition, 268
Rao, Justin M., 492 of contracts, 110, 123
Rashid, Hamid, 145 degree of, 436
Rational actors, 44f, 45 environmental, 105, 312
Rational expectations (RE) of financial sector, 102, 110–111, 124,
for climate change analysis, 337–338 150, 200
equilibrium, 164–166, 181 of monopolies, 106, 128
incentives and, 159 productivity waves and, 261
Index 529
structural, 111 Sarbanes-Oxley Act of 2002, 146–147
Reinhart, Carmen M., 172 Sargent, Thomas J., 162n5
Relatedness, 394n5 Savings programs, automatic enrollment
Religion, 54, 309, 462 in, 356–357, 363
Reserve currencies, 172–173, 177–180, Scharfstein, David S., 247, 248
185, 332 Schelling, Thomas, 311
Retirement plans, automatic enrollment Schlesinger, Karl, 30
in, 351, 363, 383 Schmidt, Klaus, 406
Returns to scale, 26 School. See Education
Reverse Yankee borrowing, 223 School lunch programs, automatic
Ricardo, David, 12, 25, 26, 297 enrollment in, 363
Rigol, Natalia, 458 Schumpeter, Joseph, 129, 255, 284, 308
Rijkers, Bob, 34, 290 Schumpeterian effect, 258
Risk markets, 105–107, 107–108n8, Schumpeterian growth paradigm,
112–113, 121, 125, 130 253–277
Risk-sharing, 101, 121, 240 as bridge between growth and
Risk-taking channel, 221–227, 222f, development economics, 267
224f, 226f on competition, 120, 256–259, 274
Robbins, Lionel, 57, 67 firm dynamics in, 265–266, 269–272,
Robinson, James, 269n13 270f
Robinson, Jonathan, 459–460 implications for policy development,
Rodrik, Dani, 273–275 272–276, 284, 289–290
Roemer, John E., 334, 426 on inequality and social mobility,
Rogoff, Kenneth, 204 262–265, 263–264f
Role model effects, 39, 43f on innovation-led growth, 255–259,
Romer, Paul M., 259, 336, 337 263, 274–275
Rothschild, Michael, 117, 140 motivations for development of,
Rubin, Donald B., 448–449, 451 253–255
Rule utilitarianism, 410, 427 productivity growth in, 255–257,
Russia, 1995–1996 financial crisis in, 267–269, 288
154, 175, 177 on secular stagnation, 259–262
Rustichini, Aldo, 435 Schwartz, Anna J., 155, 175
Scientific knowledge, 8, 81
Sachs, Jeff, 460 SDGs (Sustainable Development Goals),
Safe assets, 183, 194–196, 199–200, 203, 295, 296, 314
204, 234 SDRs (Special Drawing Rights), 332,
Safe haven currencies, 212, 248 332n4
Safety traps, 203–204 Second best theory, 112–113
Salary. See Income Secular stagnation, 170, 201, 259–262,
Samuelson, Larry, 420 334, 342n12
Samuelson, Paul, 2, 6, 58, 104n3 Securities and Exchange Commission
Sandel, Michael, 375 (SEC), 147
Sandholm, William H., 421 Securities and securitization market
530 Index
AD, 105, 106 The Theory of Moral Sentiments,
asset-backed, 168, 169, 176, 186 428–429
collapse of, 124 on time for production, 30
conflicts of interest in, 145–148 on universality of certain virtues, 428
debt, 217f, 229 The Wealth of Nations, 1, 409
fraud in, 123, 123n32 Smoking, 55, 108, 352, 354–357, 361
mortgage-backed, 146, 147, 174, 378 Snelling, A. W., 15
Selection bias, 448 Snowberg, Erik, 450, 453
Self-fulfilling financial crises, 193, Social Choice and Individual Values
194, 200 (Arrow), 6
Sen, Amartya, 6, 47, 79–80, 83–92, Social choice theory, 47–71
95–98 aggregative assessment in, 47
Senior, Nassau, 27 arbitrariness in, 48
Servén, Luis, 197 authoritarianism and, 48
Services. See Goods and services challenges related to, 47, 49, 78–79, 91
Seshadri, Ananth, 286, 287 climate change and, 333–334
Sethi, Rajiv, 420 history and development of, 47–49, 84
Shadow banks, 178, 229 impossibility theorem in, 48–51, 58,
Shah, Neil Buddy, 467 62, 65–66, 82, 86
Shaked, Avner, 420 information in, 57, 60–65, 70–71
Sharecropping, 121, 139 intellectual framework for, 80
Shared prosperity, 13, 98 internal consistency of, 51
Shiller, Robert J., 124, 373 interpersonal comparisons in, 62–67,
Shin, Hyun Song, 207, 221, 223, 229, 70, 81–83
233–239, 245–251 liberty and rights in, 53–55
Simplification, 351–352, 365, 366 majority decisions in, 52–53
Singapore mental states in, 63, 67
behavioral economics in, 382 metrics and measurements in, 68–70,
productivity growth in, 286 95–99
Single-peaked preferences, 52 rationality assumptions in, 89–91
SIPI (systemically important price or voting in, 52–53, 56, 59, 62, 84–85
index), 377 Social contracts, 12n13
Situational ethics, 88–89, 88–89n6 Social incentives, 113–114, 205,
Skepticism, 8, 89n6, 91, 353, 447 435–436
Skidelsky, Robert, 142 Social institutions, 42, 44, 116
Sluggish recovery, 170, 185–187, Social mobility, 263, 264f, 265
201–202 Social norms, 11, 355, 356, 366, 384–
Smith, Adam 385, 436
economic theory development by, 25 Social preferences
invisible hand concept of, 103, 105 altruism and, 406–408
on market equilibrium, 37 as-if, 51
on poverty, 69 assessment of, 78
informational penury and, 60
Index 531
orderings and, 49, 83, 89 Stern, Nicholas H., 12, 295, 321–322,
relationship to individual preferences, 334–335, 342, 343
51, 58 Sterner, Thomas, 334
universal, 434 Sticky prices. See Price stickiness
utility function and, 425 Stiglitz, Joseph E.
Social welfare. See also Social welfare on asymmetric information, 338
functions; Welfare economics general equilibrium under uncertainty,
aggregative assessment of, 59 36
functionals and, 64 on institutional interventions, 116
impossibility theorem and, 49–50, 58, on insurance sector, 140
65–66 on market equilibrium, 108n9
individual preferences and, 50, 60, on market failures, 112
410 on market inefficiency, 107, 107–
maximization of, 35–36, 156 108n8, 108, 118, 121
Social welfare functions microeconomic contributions of, 5, 6
aggregation rule for, 374 Pareto-ranked equilibria, 41
axioms of, 50, 97–98 on pseudo-wealth, 113
defined, 49 on quantity equilibrium, 117
development of, 58 on sharecropping, 121
formal structures of, 66 on theory of second best, 112
restricted domain of, 52 Stock-based compensation, 147–149,
theories of justice and, 334 149t
Solow, Robert, 299, 306 Strategic voting, 59
Solow model of economic growth, Structural regulations, 111
253–255 Structured speculation, 453
Sonnenschein, Hugo, 339 Subgroup axioms, 98
Sovereign debt, 167, 203 Sub-Saharan Africa
Special Drawing Rights (SDRs), 332, bed net distribution programs in,
332n4 464–465
Sperber, Dan, 90 greenhouse gas emissions in, 327
Spillbacks, 223 malaria interventions in, 465
Spillovers Substitutes, 29
from deflation, 185 Sudden stops
from dollar borrowing, 229 in capital flows, 154, 169n12, 228
from innovation, 120, 256, 308 defined, 169n12
monetary, 223 in emerging markets, 177
Sreshtra, Slesh, 435–436 liquidity and, 155, 169, 175
Sri Lanka, unconditional cash transfers systemic, 154, 169
in, 464 Sufi, Amir, 202
Staggered prices, 167–169, 171, 196, Summers, Larry, 259
198 Sunk cost bias, 387
Steady-state equilibrium, 265, 341 Sunstein, Cass R., 349, 373–376, 378–
Stein, Jeremy C., 247, 248 379, 382–387
532 Index
Suppes, Patrick, 63 economic growth and, 273
Supply innovation and, 265
aggregate, 111 purpose of, 435
in equilibrium, 30 redistributive, 118
in information economics, 104 reform efforts, 262
of liquid assets, 204 revenue from, 164
of loanable funds, 273 Technology
of money, 161, 163, 176, 179–183, asymmetric information and, 127
186–187, 202 changes in, 102–103, 124
secular stagnation as problem of, 259 competition and, 127
Supply-side liquidity traps, 181–185, cross-country and over-time
202–203 differences in, 285–287
Sustainable development decarbonization, 323
alternative pathways to, 305, 315 economic growth and, 254, 282
benefits of, 308 externalities and, 27–28
defined, 299 globalization and, 260, 262
emergence of concept, 298 information economics and, 127–129
global agenda for, 295, 296, 314 in low-carbon transition, 308, 312, 324
in infrastructure, 314 negative emissions, 303
Sustainable Development Goals (SDGs), productivity growth and, 261, 286
295, 296, 314 theoretical basis for growth of, 284
Suzumura, Kotaro, 55 TFP (total-factor productivity) growth,
Sweden 260–262
outward portfolio flows of insurance Thaler, Richard, 37, 357, 373, 376, 382
companies from, 218f, 219 Theory of Capital (Hayek), 31
total-factor productivity growth in, The Theory of Moral Sentiments (Smith),
261–262 428–429
Swiss francs The Theory of Value (Debreu), 5
appreciation of, 173 Time inconsistency, 156, 159, 194
deviations from CIP for, 210 Time in production, 30–31
in foreign exchange swaps, 209, 210f Tinbergen, Jan, 28
as safe haven currency, 212, 248 Tobacco use. See Smoking
Switcher’s curse, 142–143 Tobin, James, 249, 250
Systemically important price or index Toman, Michael, 321
(SIPI), 377 Total-factor productivity (TFP) growth,
Szech, Nora, 434–435 260–262
Toxic assets, 176, 196
Taiwan, savings behavior in, 491 Tract on Monetary Reform (Keynes), 142
Taxation Trade. See also Foreign exchange
of carbon, 306, 333, 402, 404 markets
compliance with, 407–408 carry, 221, 229, 238–239
corrective, 105, 109, 116, 299 comparative advantage in, 26
distortionary, 410 intertemporal, 163, 167, 174
Index 533
invoicing currencies for, 172, 214 United Nations Development
liberalization of, 162, 267, 289 Programme, 330
nondiscrimination provisions in, 112 United Nations Framework Convention
Trade and Welfare (Meade), 300 on Climate Change, 312
Transition traps, 267–269 United States. See also US dollars
Transitivity, 49 account deficit of, 208, 228
Transparency, 99, 103, 109, 128, 129 African American political influence
Transportation Department, US, 360–361 in, 59
Trebbi, Francesco, 269 bank regulation in, 148
Triple coincidence reasoning, 228 behavioral economics in, 382
Trust cross-border lending to emerging
economic growth and, 398 markets, 247, 247f
firm growth and, 271, 272 economic growth in, 273
inflation stabilization programs employment collapse in, 202
and, 165 firm dynamics in, 269–271, 270f
in informal personal lending, 398n11 income inequality in, 263
Kantian morality and, 398–400 innovation in, 288–289
reason and, 90 liquidity of firms in, 202
in reserve-pooling arrangements, 204 political influence of monopolies in,
Trust game, 398–400 128
Tunisia productivity waves in, 261
barriers to entry in, 290 recovery from Great Recession in, 185
cronyism in, 34–35 securitization market collapse in, 124
Turgot, Anne Robert Jacques, 47–48 social experiments in, 455
Tversky, Amos, 38 social mobility in, 263
Tyndall, John, 301 Universalizability principle, 83
USAID Development Innovation
Ultimatum bargaining game, 406 Ventures (DIV), 440–441, 469–476,
Uncertainty 472t, 492
asymmetric information and, 32 US dollars
in climate change, 307, 331, 337–338 appreciation of, 172, 208, 221, 228,
complexity and, 110 234, 238
general equilibrium theory under, 36 CDS spreads against, 225–226, 226f
of inflation, 165 cross-border bank claims in, 212–213,
intrinsic, 113 215–216f, 218–219, 227
narcissism based on, 87 cross-currency basis and exchange
radical, 141–143 rate for, 211, 212–213f, 223, 246,
United Kingdom 246f
behavioral economics in, 350, 382 in debt of nonbanks in emerging
productivity waves in, 261 markets, 229
Swiss and euro area bank activity depreciation of, 221
in, 220 exchange rate for, 211–212, 212–213f
vote to exit European Union, 8 flight to, 168–169
534 Index
US dollars (cont.) preemption of entry into population
in foreign exchange swaps, 209–210, by, 395
210f social preferences and, 425
in global banking system, 212–216, von Neumann–Morgenstern, 83
215–216f, 218–219, 234 zero-output value of money and, 171
indexation of, 165
as international funding currency, Validity of RCTs, 450–454, 494, 496
223 Value and Capital (Hicks), 31
as reserve currency, 172 Value restriction, 52
resilience of, 167 Vandenbussche, Jérôme, 269
risk-taking channel for, 221, 222f Van Reenen, John, 256
share of cross-border lending to Vautard, Robert, 329
emerging markets in, 247, 247f Végh, Carlos A., 167, 183
Utilitarianism Veil of ignorance, 83, 374, 410–411, 422
defined, 82 Velasco, Andrés, 273–275
discount rate and, 334 Vickrey, William, 422
on interpersonal comparisons, 66, 81 Victor, David, 324
reason in, 90 Vietnam, coastal erosion in, 329
rule, 410, 427 Viner, Jacob, 27
theory of justice in, 333 Virtue ethics, 426, 428
in welfare economics, 56–57, 60, 411, Vivalt, Eva, 442
424–425 von Neumann–Morgenstern payoff
Utility. See also Utility functions functions, 418n3, 419
cardinal, 60, 65–66 von Neumann–Morgenstern utility
expected, 82–83n1, 140–141 function, 82–83n1, 83
framed, 41, 43f Voting
of general equilibrium theory, 34 Kantian morality and, 405
hedonic, 399, 403, 411 in social choice theory, 52–53, 56, 59,
inequality of, 62 62, 84–85
interpersonal comparison of, 56, 57, strategic, 59
60–62, 66, 81–82 welfare economics and, 53, 59–60
intertemporal, 253 Voting paradox, 48, 62, 405
marginal, 2, 403, 411
maximization of, 36, 56, 81, 350, 407 Wages. See Income
nondecreasing, 299 Wald, Abraham, 30
ordinal, 60 Walras, Léon, 2, 5, 26, 27, 29
Utility functions The Wealth of Nations (Smith), 1, 409
additive, 410 Weibull, Jörgen W., 389, 392, 397n10,
altruism and, 406 408, 417–430, 433
Homo moralis and, 394, 396, 397, Weitzman, Martin, 307
403–404 Welfare economics. See also Social welfare
normalization of, 333, 333n7, 334n8 aggregative assessment in, 47, 59
personal, 403 classical choice model in, 374
Index 535
consequentialism in, 426 shared prosperity goal of, 98
crisis in, 55–58 in translation of Hartwick-Solow rule
defined, 55–56 into policy advice, 299
development of, 56 World Development Report on Governance
distributional issues in, 61, 85 and the Law, 11–12
environmental concerns in, 299 World Development Report on Mind,
of Homo moralis, 410–411 Society, and Behavior, 10–11, 127
interpersonal comparisons in, 56, 57, World Health Organization, 464
60–66, 70, 85
marginalist tools of, 306 Yaari, Menahem E., 418
new, 57–58, 96 Yen
normative measurement in, 68–70 deviations from CIP for, 209–210
scope of, 81 in foreign exchange swaps, 209, 210f
theorems of, 35, 102n2, 105, 118, 417 as international funding currency,
utilitarian, 56–57, 60, 411, 424–425 223
voting information and, 53, 59–60 as safe haven currency, 212
Wicksell, Knut, 12, 298 Yilankaya, Okan, 419
Women Young, Allyn, 27
automatic enrollment benefits for, 363 Young, Alwyn, 286
education for, 40–42, 43f Yun, Jungyoll, 117
expectations for, 40
fertility rates for, 39–40 Zambia, favorable terms of trade shock
gender inequality and, 70 in, 35
social impact of hiring, 40, 41t Zilibotti, Fabrizio, 268
Woodruff, Christopher, 464 Zinman, Jonathan, 455
Work. See Labor
World Bank
behavioral economics used by, 382
CCT program promotion by, 465
Commission on Global Poverty, 99
cost-benefit analysis used by, 300
general equilibrium theory used by, 34
Global Financial Development Report
2015/16, 237
Global Financial Development Report
2017/18, 239
Insurance Development Forum
and, 332
measurement methods used by, 99
objectives of, 3, 6, 130–131
Office of the Environmental Advisor,
297
RCTs and, 441, 447, 494n1, 496