The Global Macro Economy and Finance (PDFDrive)

The Global Macro Economy and Finance
This is IEA conference volume no. 150-III

The Global Macro Economy
and Finance
Edited by
Franklin Allen
University of Pennsylvania, USA
Masahiko Aoki
Stanford University, USA
Jean-Paul Fitoussi
Sciences Po, France
Nobuhiro Kiyotaki
Princeton University, USA
Roger Gordon
University of California, San Diego, USA
Joseph E. Stiglitz
Columbia University, USA
© International Economics Association 2012
Softcover reprint of the hardcover 1st edition 2012 978-1-137-03423-6
All rights reserved. No reproduction, copy or transmission of this
publication may be made without written permission.
No portion of this publication may be reproduced, copied or transmitted
save with written permission or in accordance with the provisions of the
Copyright, Designs and Patents Act 1988, or under the terms of any licence
permitting limited copying issued by the Copyright Licensing Agency,
Saffron House, 6–10 Kirby Street, London EC1N 8TS.
Any person who does any unauthorized act in relation to this publication
may be liable to criminal prosecution and civil claims for damages.
The authors have asserted their rights to be identified as the authors of this work
in accordance with the Copyright, Designs and Patents Act 1988.
First published 2012 by
PALGRAVE MACMILLAN
Palgrave Macmillan in the UK is an imprint of Macmillan Publishers Limited,
registered in England, company number 785998, of Houndmills, Basingstoke,
Hampshire RG21 6XS.
Palgrave Macmillan in the US is a division of St Martin’s Press LLC,
175 Fifth Avenue, New York, NY 10010.
Palgrave Macmillan is the global academic imprint of the above companies
and has companies and representatives throughout the world.
Palgrave® and Macmillan® are registered trademarks in the United States,
the United Kingdom, Europe and other countries
ISBN 978-1-137-03424-3 ISBN 978-1-137-03425-0 (eBook)
DOI 10.1057/9781137034250
This book is printed on paper suitable for recycling and made from fully
managed and sustained forest sources. Logging, pulping and manufacturing
processes are expected to conform to the environmental regulations of the
country of origin.
A catalogue record for this book is available from the British Library.
A catalog record for this book is available from the Library of Congress.
10 9 8 7 6 5 4 3 2 1
21 20 19 18 17 16 15 14 13 12
Contents
List of Tables vii
List of Figures viii
Notes on Contributors x
Foreword xiii
Introduction 1
Franklin Allen, Masahiko Aoki, Jean-Paul Fitoussi, Roger Gordon, Nobuhiro
Kiyotaki and Joseph E. Stiglitz
Part I Beyond GDP

1 On the Measurement of Social Progress and Wellbeing: Some
Further Thoughts 13
Jean-Paul Fitoussi and Joseph E. Stiglitz
2 The Role of Statistics in the United States’ Economic Future 25

J. Steven Landefeld and Shaunda Villones
3 Measuring Equitable and Sustainable Wellbeing in Italy 38

Enrico Giovannini and Tommaso Rondinella
Part II Structural Imbalances, Financial

Frictions and Externalities
4 Sectoral Imbalances and Long-run Crises 61
Domenico Delli Gatti, Mauro Gallegatti, Bruce C. Greenwald, Alberto
Russo and Joseph E. Stiglitz
5 Capital Flows, Crises and Externalities 98

Anton Korinek
6 Liquidity Shocks and Asset Prices in the Business Cycle 118

Shouyong Shi
7 Whither Capitalism? Financial Externalities and Crisis 131

Marcus Miller and Lei Zhang
v
vi Contents
8 Bank Lending and Credit Supply Shocks 154

Simon Gilchrist and Egon Zakrajšek
9 A Mechanism Design Approach to Financial Frictions 177

Nobuhiro Kiyotaki
Part III Behavior of Financial

Institutions and Prudential Regulations
10 Systemic Risk and Macroprudential Regulation 191
Franklin Allen and Elena Carletti
11 Filling the Gaps – the Vienna Initiative and the Role of the
International Financial Institutions in Crisis Management
and Resolution 211
Erik Berglof
12 Some Recent Progresses on Financial Structure and

Development 224
Justin Yifu Lin and Lixin Colin Xu
13 The Race to Zero 245

Andrew G. Haldane
14 A Model of Private Equity Fund Compensation 271

Wonho Wilson Choi, Andrew Metrick and Ayako Yasuda
Part IV Taxation in a Globalized Economy

15 How Should Income from Multinationals Be Taxed? 289
Roger Gordon
16 Taxing Multinationals in a World with International

Mergers and Acquisitions: Should the Home Country
Exempt Foreign Income? 313
John Douglas Wilson
Index 329
List of Tables
3.1 Goals of progress 44

3.2 What Italians think is important for their wellbeing 55
5.1 Externalities imposed by different financial instruments in
Indonesia 1997–98 113
7.1 Calibration of the capital framework: capital requirements and
buffers 144
7.2 Regulatory measures for capital, liquidity, and separation 146
13.1 Round-trip latencies for selected exchanges, April 2010 252
14.1 Parameter values for the simulation model 279
14.2 Baseline model results 280
14.3 The effect of carry timing rules on PV of carry 281
14.4 The effect of inflated (reported) values of un-exited
investments on the PVs of carry 282
14.5 Sensitivity analysis 284
vii
List of Figures
2.1 Reconciling measures of income 27

2.2 Healthcare price indexes 30
2.3 Housing and personal income 32
2.4 Household asset values and savings 32
2.5 Growth in equity prices relative to GDP and NIPA profits 33
2.6 Financial business sector leverage (total financial assets/total
liabilities) 34
2.7 Sustainable investment 35
3.1 Adjusted disposable income and GDP, Italy 1995–2009 47
4.1 Equilibrium combination of urban employment (wages) and
rural prices 76
4.2 The effects of an increase in agricultural productivity 77
4.3 Impact of Keynesian stimulus 79
4.4 The effects of downward wage adjustments 80
4.5 The effects of migration subsidy 81
5.1 Conditional probability of crisis after capital flow bonanza 99
5.2 Balance sheet crises and financial amplification in emerging
economies 100
5.3 Private and social valuation of liquidity 109
5.4 Constrained planner’s second-best intervention 110
6.1 Steady-state effect of reducing φ ∗ : partial equilibrium 126
7.1 The financial accelerator as pecuniary externality 134
7.2 Private and social valuation of liquidity 135
7.3 Risk concentration, excessive leverage, and global imbalances 141
7.4 Timeline for the implementation of Basel III 145
7.A1 A partially connected network 148
7.A2 Optimal network 150
7.A3 Expected utility of a node with circuit-breaker 150
8.1 Cyclical dynamics of private nonfinancial debt and its main
components 158
8.2 The relative importance bank lending to households and
businesses 159
8.3 Cyclical dynamics of household and business loans at
commercial banks 160
8.4 The excess bond premium 162
8.5 Credit supply shocks, economic activity, and bank lending 164
8.6 Core loans and unused commitments at commercial banks 166
8.7 Growth in business lending capacity at commercial banks 167
8.8 The excess bond premium and banks’ willingness to lend 168
viii
List of Figures ix
8.9 The excess bond premium and business lending, 2005–2010 169
8.10 Credit supply shocks and business lending capacity 170
9.1 The Green-Oh model 181
9.2 Hidden storage and limited commitment 185
10.1 Housing prices in Ireland, Spain, and the US 192
10.2 A comparison of foreign exchange reserves in different regions 193
11.1 Share of foreign banks owned by Eurozone-based groups 212
11.2 Capital flows 2008–2010 from major world regions 217
13.1 Market capitalization by region, relative to GDP 247
13.2 Equity market turnover by region 247
13.3 NYSE-listed trading volume shares, January 2005 248
13.4 NYSE-listed trading volume shares, February 2011 249
13.5 FTSE 100 trading volume shares, June 2008 249
13.6 FTSE 100 trading volume shares, June 2011 250
13.7 Access to multiple trading venues by European HFTs 251
13.8 Median end-of-day bid-ask spread for largest 20 stocks in FTSE
100 as a proportion of realized volatility 254
13.9 Volatility and correlation of S& P 500 255
13.10 Excess volatility 256
13.11 Simulated price series under different Hurst coefficients 257
13.12 Distribution of simulated returns at different time horizons
with H = 0.9 261
13.13 Difference in maximum and minimum price of GE shares
across different exchanges on 6 May 2010 262
Notes on Contributors
Franklin Allen is Nippon Life Professor of Finance, Professor of Economics and

Co-director of the Financial Institutions Center at the Wharton School of the
University of Pennsylvania. He is former President of the American Finance
Association and a Fellow of the Econometric Society.
Masahiko Aoki is Henri and Tomoye Takahashi Professor Emeritus in the
Department of Economics and a senior fellow of Stanford Institute of Economic
Policy Research and Freeman Spolgi Institute of International Studies at Stan-
ford University. He is a Fellow of the Econometric Society and former President
of the International Economic Association (2008–11).
Erik Berglof is Chief Economist and Special Adviser to the President at the
EBRD. He was formerly Professor and Director of SITE (SSE) and Founder and
President of the Centre for Economic and Financial Research (Moscow).
Elena Carletti is Professor of Economics at the European University Institute,
Research Fellow at CEPR, Extramural fellow at TILEC, and Fellow at CESifo, the
Center for Financial Studies and the Wharton Financial Institutions Center.
Domenico Delli Gatti is Professor of Economics at Catholic University in Milan
(Italy). He is a member of CESifo, associate editor of the Journal of Economic
Behaviour and Organization (1998–2010) and of the Journal of Economic Interaction
and Coordinaton.
Jean-Paul Fitoussi is Professor Emeritus at Sciences-po, Paris, Professor of Eco-
nomics at Luiss University, Rome, Director of Research at OFCE, Paris, and
former Secretary General of the International Economic Association.
Mauro Gallegati is Professor of Economics at the Polytechnic University of
Marche, Italy.
Simon Gilchrist is Professor of Economics at Boston University, and a Research
Associate of the National Bureau of Economic Research.
Enrico Giovannini is Full Professor of Economic Statistics at the University
of Rome ‘Tor Vergata’. He has been President of the Italian Statistical Institute
(Istat) since August 2009. From 2001 to July 2009 he was Chief Statistician of the
OECD where he launched the Global Project on the ‘Measurement of Progress in
Societies’.
Roger Gordon is Professor of Economics at University of California San Diego,
Adjunct Professor at CKGSB, China, and a Research Fellow at NBER. He is past
editor of the Journal of Economic Literature, the Journal of Public Economics, and
x
Notes on Contributors xi
the American Economic Review. He is a Fellow of the Econometric Society and

AAAS.
Bruce Greenwald is Heilbrunn Professor of Finance and Economics at Columbia
Business School, New York, USA.
Andrew Haldane is Executive Director, Financial Stability and a member of the
Financial Policy Committee, Bank of England.
Nobuhiro Kiyotaki is Professor of Economics at Princeton University. He is an
Academic Consultant of the Federal Reserve of New York and a Fellow of the
British Academy.
Anton Korinek is Assistant Professor at the University of Maryland and Visiting
Scholar at the International Monetary Fund.
J. Steven Landefeld is Director, Bureau of Economic Analysis at the US Depart-
ment of Commerce.
Justin Yifu Lin is Professor of Economics at Peking University, China Cen-
ter for Economic Research (CCER) where he was previously Founding Director
and professor (1994–2008). Previously, he was Senior Vice President and Chief
Economist of the World Bank (2008–2012).
Andrew Metrick is the Michael H. Jordan Professor of Finance and Manage-
ment at the Yale School of Management.
Marcus Miller is Professor at Department of Economics and Research Associate
of the Centre for Competitive Advantage in the Global Economy at the Univer-
sity of Warwick, UK. He is a former adviser to the Treasury Committee of the
UK House of Commons.
Tommaso Rondinella is a Researcher at Istat, Italy.
Alberto Russo is Assistant Professor of Economics at the Department of Eco-
nomics and Social Sciences, Università Politecnica delle Marche, Ancona, Italy.
Shouyong Shi is Professor of Economics at the University of Toronto and the
Canada Research Chair Research Fellow of the Bank of Canada.
Joseph E. Stiglitz is University Professor at Columbia University. In 2001, he
was awarded the Nobel Prize in economics for his analyses of markets with asym-
metric information. He is currently the President of the International Economic
Association (2011–14).
Shaunda M. Villones is Senior Economist and Assistant to the Director at the
Bureau of Economic Analysis, US Department of Commerce.
Wonho Wilson Choi is an Assistant Professor at KAIST College of Business,
South Korea.
John D. Wilson is Professor of Economics at Michigan State University. He was
previously Editor-in-Chief of International Tax and Public Economics, and served
xii Notes on Contributors
as a member of the Board of Editors of the American Economic Review, and as

Co-editor of the Journal of International Economics.
Lixin Xu is an economist at the World Bank.
Ayako Yasuda is Associate Professor of Management (Finance Area) at the
Graduate School of Management, University of California at Davis.
Egon Zakrajšek is Deputy Associate Director, Division of Monetary Affairs,
Board of Governors of the Federal Reserve System.
Lei Zhang is Associate Professor in the Department of Economics and
Research Associate of the Centre for Competitive Advantage in the Global
Economy at the University of Warwick, UK.
Foreword
The International Economic Association (IEA), the association of national

economic associations/societies, held its 16th Triennial World Congress in
Beijing, China, from July 4–8, 2011. This four-volume IEA publication, IEA
Conference Volume No. 150–I~IV, comprises 55 papers selected from those pre-
sented at the Congress and, in many cases, revised afterwards incorporating
discussions that took place at the event.
The Congress was organized in the aftermath of the financial crisis, the pro-
longed economic consequences of which continued to cast shadows over the
global economic landscape. At the same time significant changes in economic
structure, mechanisms and institutions were taking place globally – albeit in
uncertain directions. It was thus a particularly appropriate time for economists
from around the world to gather together in order to present their ongoing
analyses about the evolving economic situations and to offer their prescriptions
for the future orientation of economic policies. Further, the shared sense of
uncertainty about the transformation of the world economy seems to impel
economists to reflect on and debate the fundamental premises and focus of
their discipline, and to seek advancements in methodology, conceptual terms
and so on so that economics can once again become relevant, insightful and
constructive.
Dynamic China provided a felicitous meeting place for economists from across
the world to engage in this academic enterprise, and more than 1,000 papers
were submitted in response to the Call for Papers by the Program Committee.
From these submissions the academic program committee selected some 400
papers for presentation. In addition, the Committee organized five Lectures,
four Plenary Panels and 22 invited sessions on various subject matters, inviting
about 90 leading scholars to participate in them. The contributions to this IEA
conference series are comprised mostly of invited papers, selected and organized
into four volumes according to the following themes.
Volume I, Institutions and Comparative Economic Development (IEA Conference
Volume No. 150–I), edited by Masahiko Aoki, Timur Kuran and Gérard Roland,
contains 16 chapters. In the process of the rapid globalization of economies
over the course of the past few decades, the landscape of the world economy
as divided into developed, developing and underdeveloped terrains is being
reshaped. Some of the economies that had stagnated for a number of years are
now emerging as fast-growing economies; by contrast, developed economies
are now facing unprecedented challenges to sustain growth under the rapidly
xiii
xiv Foreword
evolving environmental, demographic and global market conditions. Will these

changes lead to a new convergence of economic structure and performance
across economies? Or will the differences in structure and institutional frame-
work remain in spite of global market integration, entailing differentiated
developmental performances? A plausible prediction about these issues would
require, among other things, a deep understanding of the nature of the long-
term developmental processes of economies from both a comparative and a
historical perspective. Accordingly, the chapters in this volume consider themes
such as: Why did divergences in development performance occur in the past
across different regions: the West, Asia, the Islamic world, sub-Saharan Africa
and so on? What role did institutional, political and cultural factors play in
these divergences? Will differences in these factors persist, or will they become
malleable in response to emergent global conditions? Whichever is the case,
what will be the consequences for development? Why and how is the issue of
food security still a matter of concern for some parts of the world economy?
This volume presents a host of original contributions to these and related top-
ics of development economics with historical, comparative and institutional
underpinnings.
Volume II, Complexity and Institutions: Markets, Norms and Corporations (IEA
Conference Volume No. 150–II), edited by Masahiko Aoki, Ken Binmore, Simon
Deakin and Herbert Gintis contains 11 chapters. The totality of economic
exchanges among agents in the society is complex, leading economists to grap-
ple with questions such as: How do the agents in society deal with the complex
reality in order to pursue their own materialistic and other possible preferences?
How could economists construct a good theoretical explanation of complex real-
ity and predict its future performance? According to one view, institutions may
be identified with salient features of stable states of play of the societal game that
are publicly represented in such forms as norms, laws, markets, organizations,
political and corporate governance structures and so on. As such, institutions
can provide an order to economic and other types of social exchange, helping
agents to cope with the complexity of the real world. However, how can such
an order, or an equilibrium state of play, be chosen from among the many possi-
ble, and in turn recognized and observed by agents? What are the relationships
between such social orders and individual rationality? How do institutional
forms, as mentioned above, mediate between the two and how do they evolve?
Does the complexity of real life refute the traditional economists’ premise of
rationality? The contributions in this volume adopt a variety of perspectives to
consider these and other related questions.
Volume III, The Global Macro Economy and Finance (IEA Conference Volume
No. 150–III), edited by Franklin Allen, Masahiko Aoki, Jean-Paul Fitoussi,
Nobuhiro Kiyotaki, Roger Gordon and Joseph E. Stiglitz, comprises 16 chapters.
The financial crisis and the resulting persistent shortfall between the level
Foreword xv
of production in the USA, Europe and Japan and potential output indicates
that macroeconomic mechanisms, theory and policy are not well tuned to,
or in step with, rapidly changing global economic environments. Important
advances in microeconomics over the past three decades have shown that when-
ever information is imperfect and asymmetric and markets are incomplete – that
is, always – markets are not even constrained Pareto-efficient. These advances
have created a presumption that markets are inefficient, but, unfortunately,
these insights have not been built into most of the standard macroeconomic
models. This may help explain their poor performance, in terms of predicting
major downturns, and in providing coherent interpretations of the downturns
and their persistence. Similarly, they have offered little of practical use in terms
of providing advice about how to respond to crises such as the current global
one, how to foster a robust recovery, and how to prevent a recurrence of such
events. The chapters in this volume confront issues such as: Can the tradi-
tional measurement of GDP be a good measure for gauging and promoting social
progress and global welfare? What was wrong with the financial mechanism that
had been thought of as promoting economic development prior to the crisis?
Is there a way to reconnect the financial and real sectors in a more stable and
sustainable way? Alternatively, is it right to presume that the current economic
crisis is essentially a financial crisis? Is there not a more fundamental structural
problem in the real sector that caused the economic crisis on a global scale? If
so, what is it? What kinds of prudent financial regulations, financial institu-
tion reforms, macroeconomic performance measures, and taxation on global
activities are desirable for ever-interconnected but nevertheless diverse national
economies to be stabilized and develop together?
Volume IV, The Chinese Economy: A New Transition (IEA Conference Volume
No. 150–IV), edited by Masahiko Aoki and Jinglian Wu, contains 12 chapters.
The rise of industrialized China and her resurgence as an economic power-
house is a transformative event in the history of the world economy. However,
there now appears to be an emergent consensus that the Chinese economy
is facing a turning point, that is to say, another transition after the thirty
years of successful transition from the command economy to the market econ-
omy. The nature of this new transition may be understood to some extent
within a recent conceptual and analytical framework that unifies development
economics and demography in a long-term perspective. After a rather long tran-
sition out of the Malthusian state (1911–the late 1940s) and then the phase of
government-mediated initial industrialization (the early 1950s–the late 1970s),
the era of high growth ensued, driven by favorable demographic factors such
as the demographic dividend (an increase in the ratio of working-age popu-
lation in the total population) and the massive domestic migration of labor
from the rural agricultural sector to the industrial sector. However, this phase
xvi Foreword
of development is now about to end. Simply put, improvements in the qual-

ity of human resources rather than a mere expansion of their quantity, as
well as an increase in consumption relative to investment in physical capital,
needs to become a major driving force of further development. This transition
to the phase of human capital-based, market-oriented development is the key
for China to sustain per capita income growth, albeit at a slower rate than in
the past three decades, and to avoid a stall referred to as the ‘middle-income
trap’ in the policy arena of development economics. As argued by authors of
almost all the chapters included in this volume, this transition would call for
a further deepening of the market mechanism driven by private incentives,
governed by the rule of law and complemented by sustainable public policy.
More specifically, the chapters ask and attempt to answer questions such as:
How has modern economic science helped guide the Reform and Open Pol-
icy of China in the past and what suggestions can it offer for China’s future
direction? How does the Chinese (state) market economy operate in the areas
of labor markets, social welfare systems, banking and corporate governance,
foreign exchange and capital account controls, environmental control, and
also in the area of relations between central and local governments? What
kinds of future policy agenda and institutional reforms may be needed in these
areas to make the transition to the new phase of development smooth and
effective?
As noted, the papers compiled in this IEA World Conference series are
selected not only for their academic merit but also on the basis of the selected
themes. It is therefore regrettable that many excellent papers presented and dis-
cussed at the Congress have not been included in this series, especially those
in the fields of experimental economics (both field and laboratory), interna-
tional trade and foreign exchange, income and wealth inequality, environment
and energy, econometric methods (especially those dealing with unobserved
heterogeneity), country-specific empirical studies and so on. The whole pro-
gram of the Congress may be accessed by visiting the Congress website:
www.iea-congress-2011.org/Homepage.html.
The Program Committee responsible for the design and implementation of
the program was comprised of leading economists from all over the world
and from broad fields of economics. Their wisdom, academic expertise and
leadership, organizing competence and generous sharing of time made the
World Congress an enormously successful academic event. As the Chairman
of the Program Committee ex officio, I would like to express sincere gratitude
to them by listing their names below in alphabetical order with their insti-
tutional affiliations at the time of the Congress (those with an asterisk have
also made additional contributions to the compilation of these IEA conference
Volumes).
Foreword xvii
* Agarwal, Bina, India (Delhi University)

* Allen, Franklin, USA (University of Pennsylvania)
* Aoki, Masahiko, Japan (Stanford University)
Araujo, Aloisio, Brazil (Instituto Nacional de Matematica)
Ayogu, Melvin, Nigeria (Standard Bank Group, Johannesburg)
* Bai, Chong-En, China (Tsinghua University)
Bardhan, Pranab, India (University of California, Berkeley)
Basu, Kaushik, India (Cornell University/Indian Government)
Berglof, Erik, Sweden (European Bank for Reconstruction and
Development)
Binmore, Ken, UK (University College London)
* Cai, Fang, China (Chinese Academy of Social Sciences)
Calvo, Guillermo, Argentina (Columbia University)
de Cecco, Marcello, Italy (Scuola Normale Superiore – Pisa)
* Deakin, Simon, UK (University of Cambridge)
Englund, Peter, Sweden (Stockholm School of Economics)
Esteban, Joan, Spain (Universitat Autonoma Barcelona)
* Fang, Hanming, China (University of Pennsylvania)
* Fitoussi, Jean-Paul, France (Sciences Po – Paris)
Freeman, Richard, USA (Harvard University)
* Gintis, Herbert, USA (Santa Fe Institute)
Goulder, Lawrence, USA (Stanford University)
* Gordon, Roger, USA (University of California at San Diego)
Greif, Avner, Israel (Stanford University)
Hong, Han, China (Stanford University)
Ito, Takatoshi, Japan (University of Tokyo)
* Jorgenson, Dale, USA (Harvard University)
Kandori, Michihiro, Japan (University of Tokyo)
* Kiyotaki, Nobuhiro, Japan (Princeton University)
Kornai, Janos, Hungary (Harvard University/Hungary Academy)
* Kuran, Timur, Turkey (Duke University)
Lau, Lawrence, China (Chinese University – Hong Kong)
Lee, Joung-wha, Korea (Asian Development Bank)
Li, Yang, China (Chinese Academy of Social Sciences)
* Lin, Justin Yifu, China (World Bank)
Ndikumana, Leonce, USA (African Development Bank)
Paganetto, Luigi, Italy (Università Roma ‘Tor Vergata’)
Papademos, Lucas, Greece (European Central Bank)
Piketty, Thomas, France (Ecole d’Economie de Paris)
Polterovich, Victor, Russia (Central Economics and Mathematics
Institute (CEMI), Russian Academy of Sciences)
* Qian, Ying-yi, China (Tsinghua University)
xviii Foreword
* Roland, Gérard, Belgium (University of California, Berkeley)

Sheshinski, Eytan, Israel (Hebrew University)
* Stiglitz, Joseph, USA (Columbia University)
* Svejnar, Jan, Czech Rep. (University of Michigan)
Tommasi, Mariano, Argentina (Universidad de San Andrés)
Uygur, Ercan, Turkey (Ankara University)
Vilamill, Winfred, Philippines (De La Salle University-Manila)
Wei, Shang-Jin, China (Columbia University)
* Wu, Jinglian, China (Development Research Center)
Xu, Chenggang, China (University of Hong Kong)
Zhou, Lin, China (Shanghai Jiaotong University)
The meeting of the Congress was held in the excellent academic and histor-
ical atmosphere of Qinghua University which was simultaneously celebrating
its centenary. The collegial academic discussions that took place in this ambi-
ence certainly marked one of highlights of the 50-year history of the IEA. The
Association would like to express sincere gratitude to the university, headed by
then-President Gu Binglin, for providing an amiable environment, excellent
facilities, efficient administrative help and warm hospitality. The IEA also owes
a great debt of gratitude to those who organized the Congress on site: Executive
Vice President Xie Weihe of Tsinghua University, Chairman of the Local Orga-
nizing Committee, Professor Bai Chong-En of Tsinghua University, its Secretary
General, and all the other members of the Local Organizing Committee:
Cai, Hongbin (Peking University)

Che, Jiahua (Chinese University of Hong Kong)
Ju, Jiandong (Tsinghua/University of Oklahoma)
Li, Hongbin (Tsinghua University)
Li, David (Tsinghua University)
Qian, Ying-yi (Tsinghua University)
Wen, Yi (Tsinghua University/Federal Reserve Bank at St Louis)
Xiao, Meng (Bijiao)
Xue, Lan (Tsinghua University)
Zhao, Yaohui (Peking University)
Zhou, Lian (Peking University)
In addition, the hard work of the administrative staff and student assistants
at Tsinghua University coordinated by Mr Yu Jiang, ensured that the logistic
operations of the Congress ran in an impeccably smooth manner, something
for which the executive committee of the IEA would like to express great thanks.
The Congress was financially supported by CITIC Group, the China Invest-
ment Corporation, the China Construction Bank and the China International
Foreword xix
Capital Corporation. The IEA would like to express deep gratitude to these
donors for their generous support.
And finally, we are deeply indebted to Nick Brock and Rick Bouwman for
their careful editing of the entire manuscripts, and to all the staff at Palgrave
Macmillan for their great help in shepherding the volumes from the contracting
through the production process.
July 27, 2012
Masahiko Aoki
General Editor, IEA Conference Volume No. 150: I–IV
President (2008–11), the International Economic Association
Introduction
Franklin Allen
Masahiko Aoki
Stanford University, USA
Jean-Paul Fitoussi
Sciences-po, Paris
Roger Gordon
University of California, San Diego, USA
Nobuhiro Kiyotaki
Joseph E. Stiglitz
In the event of 2007 financial crisis and its prolonged aftermath, it has become
increasingly evident that macroeconomic mechanisms and financial institu-
tions, as well as macroeconomic theory and policy relevant to them, are not
well tuned to, or in step with, rapidly changing global economic environ-
ments. National economies, developed, emerging and developing alike, are
ever more closely interconnected through market and financial integration,
fast information transmission and human resource mobility, restructuring of
the global division of labor, and so on. Macroeconomics, as well as financial
economics, appear urged to re-examine, and possibly re-orient, its accepted
premises, research foci, analytical method, and even underlying philosophy, in
order to understand the nature of problems that the global economy is facing,
improve on economic policy and prudent regulations, and reform global and
national economic institutions. Understandably, meeting this challenge would
be hardly easy because of the complexity of the issues involved.
Chapters in this volume ambitiously attempt to face this challenge in one
way or other, by focusing on research questions such as: can the traditional
measurement of GDP be a good measure for gauging social progress and global
welfare? What was wrong with the financial mechanism that had been thought
of as promoting economic development prior to the crisis and how should it be
reformed? Is there a way to reconnect the financial and real sectors in a more
stable and sustainable way? Alternatively, is it right to presume that the current
economic crisis is essentially a financial crisis? Is there not a more fundamental
1
2 F. Allen, M. Aoki, J.-P. Fitoussi, R. Gordon, N. Kiyotaki and J. E. Stiglitz
structural problem in the real sector that caused the economic crisis on global
scale? If so, what is it? What kinds of prudent financial regulations, financial
institution reforms, and taxation of international economic activities are desir-
able for ever-interconnected but nevertheless diverse national economies to be
stabilized and develop together? The chapters in this volume are organized into
five Parts according to these issues. In the paragraphs below, we summarize each
of the chapters.
Beyond GDP
The financial crisis revealed that we (especially in the West) were not doing as
well as we had thought we had by looking at the available metrics. The output
measures had been exaggerated by bubble prices in real estate and by virtual
profits in the financial sector, while these states had been generated by the pre-
vailing pursuit of maximizing benefits measured by market prices. It is true that
what we do depends on what we measure, but we may not be really happy with
the consequences of actions based on the adopted measure. Therefore what
we measure should be conditioned by what we are really aiming at. But what
should our aims be? How do we agree on them? To explore these issues, the
Commission on Measurement of Economic Performance and Social Progress
was set up by the French government in January 2008. The Commission pro-
duced the so-called Stliglitz-Sen-Fitoussi Report (2009/2011), which called for a
‘shift [of] emphasis from measuring economic production to measuring people’s
well-being’. This was a timely restatement of an agenda put forward, inter alia,
four decades ago by William Nordhaus and James Tobin (1972), and a number
of international and national initiatives have been undertaken in an attempt
to follow the recommendations of the Report since then (most recently by the
OCED). Three chapters included in the first Part of this volume, taken together,
cover the most important dimensions of the agenda: the setting of an analyt-
ical framework, the selection of relevant instruments, and their application to
specific case studies.
The first chapter, ‘On the Measurement of Social Progress and Wellbeing’,
by two of the authors of the Report, Jean-Paul Fitoussi and Joseph Stiglitz,
offers some further reflections on the subject after the delivery of the Report,
by confronting it with some important geopolitical developments that have
taken place since its release. The authors take a fresh look at some important
global events, such as the Arab revolutions, the disaster of Fukushima and the
aftermath of the financial crisis, and bring new thoughts to old issues such as
mass unemployment, the measurement of economic products, wellbeing and
sustainability, and European economic policy-making. The chapter illustrates
how economics as a science of measurement can impact policy when a simple
truth is acknowledged: measuring is, in some way, governing.
Introduction 3
The second chapter, ‘The Role of Statistics in the United States’ Economic
Future’, by J. Steven Landefeld and Shaudra Villones of the US Bureau of Eco-
nomic Analysis, applies the Stiglitz-Sen-Fitoussi approach to the US, where social
phenomena of great importance often go unnoticed because of the insufficiency
of the current economic and social measurement system. They make a power-
ful case for better measures of income distribution, an improved accounting
framework able to take stock of the immaterial economy, natural resources, the
healthcare sector and public services. Of particular importance and relevance
is their call for new measures able to better identify and manage unsustainable
trends in financial and housing markets.
The third chapter, ‘Measuring Equitable and Sustainable Wellbeing in Italy’
by Enrico Giovannini and Tommaso Rondinella, sheds light on a much inter-
esting case, that of Italy. Italy, like France and a number of other countries, has
started to implement the Stiglitz-Sen-Fitoussi Report’s recommendations. As the
authors point out, it is crucial that this process takes place in a democratic set-
ting, where citizens are invited to deliberate on new indicators that are aimed at
better reflecting the realities of their lives. Without such a social appropriation,
the redefinition of economic and social measurement will lack legitimacy. And
without democratic legitimacy, this effort will be without raison d’être. One of
the triggers of this new area of research has been indeed the disparities between
what government agencies say and what individual citizens feel or know about
their own status.
Structural imbalances, financial frictions and externalities
This Part comprises six chapters dealing with structural imbalances, financial
frictions, and externalities in the context of the (global) macroeconomy.
There has been a widespread presumption that the current economic crisis is
a financial crisis. The most important policy implication of this view could be to
suggest that if the financial system were repaired, the economy would return to
health. However, the depth and duration of the current downturn, in contrast
to other postwar recessions, may call for an alternative explanation. Perhaps the
dominant strand in modern macroeconomics has focused on models in which
markets could be efficient. But important advances in microeconomics over
the past three decades have shown that whenever information is imperfect and
asymmetric and markets are incomplete – that is, always – markets are not even
constrained Pareto-efficient. These advances have created a presumption that
markets are inefficient, but unfortunately, these insights have not been built
into most of the standard macroeconomic models, which helps explains their
poor performance, in prediction of major downturns, in providing coherent
interpretations of the downturns and their persistence, and in providing advice
about how to respond to the crisis, how to foster robust recovery, and how to
prevent a recurrence.
The challenge facing modern macroeconomics is thus to identify the key mar-
ket imperfections that provide insight into the economy’s aggregate behavior.
The chapters in this section are important contributions to a growing literature
that attempts to do this. Each focuses on a different market failure. Chapter 4,
‘Sectoral Imbalances and Long-run Crises’, by Domenico Delli Gatti, Mauro
Gallegatti, Bruce C. Greenwald, Albert Russo, and Joseph E. Stiglitz, examines
the current financial crisis, breaks with more common analyses that attribute
the crisis mostly to overleveraging, poor regulation, and the subprime spark. The
chapter proposes an alternative interpretation of the current global financial cri-
sis that emphasizes sectoral dislocation following localized technical change in
the presence of barriers to labor mobility. Delli Gatti et al. thus suggest that
underlying the Great Recession are structural problems. The tale echoes that of
the Great Depression: in the 1930s, technical change was localized in agricul-
ture; wages fell dramatically, but because of the costs of moving out of the rural
sector, farm workers were ‘trapped’. Shrinking income in agriculture reverber-
ated in the other sectors, causing a large depression. Now, it is manufacturing
that plays the role of the epicenter of technical change. Increasing productivity
in that sector in excess of the increase in demand implies shrinking employ-
ment, but mobility constraints and costs make it difficult for workers to shift
out of that sector to the sectors that should be growing, especially the service sec-
tor. The authors argue that this may be the underlying cause of the long-lasting
slump, helping to explain the painfully slow recovery – and why unemployment
has remained so high even after the financial sector has largely been repaired. It
suggests that weaknesses will remain even after households finish deleveraging
and banks and firms have rebuilt their balance sheets. And the chapter suggests
clear policy prescriptions: while the crisis was caused in a sense by a structural
problem (amplified by financial sector excesses), Keynesian aggregate demand
policies can stimulate the economy, and even more so if they are designed to
help address the underlying structural problems. By the same token, policies
aimed at decreasing wages can aggravate the downturn.
In Chapter 5, ‘Capital Flows, Crises, and Externalities’, by Anton Korinek, we
see yet another side of crisis and contagion that affects even economies where
risky financial practices are not the norm. There is no aberrant behavior. Korinek
shows that unencumbered capital flows to emerging market economies – some-
times celebrated as engines of development – create externalities that make the
affected economies more vulnerable to financial fragility and crises. The value
of an emergent economy’s collateral and the health of private sector balance
sheets depend on exchange rates and asset prices. They deteriorate in bad times
when exchange rates depreciate but when access to finance is most needed. By
analyzing a model to illustrate the point, he shows that under the condition
Introduction 5
of tight financial constraint, a constrained social planner takes on less debt

than decentralized agents. This may suggest that policy-makers can make their
economies better off by imposing capital controls as a form of Pigouvian taxa-
tion. For example, regulating and discouraging the use of risky forms of external
finance such as short-term dollar-denominated debts could reduce macroeco-
nomic volatility and lead to a global Pareto-improvement. Local currency debts,
portfolio investments, and non-financial direct investment would also mitigate
externalities.
Chapter 6, ‘Liquidity Shocks and Asset Prices in the Business Cycle’, by
Shouyong Shi, theoretically re-examines the plausibility of a popular hypothesis
that liquidity shocks in the equity market are important causes of the business
cycle because they affect firms’ ability to finance investment. To do so, Shi con-
structs a tractable macro model to incorporate the frictions that restrict firms’
ability to issue new equity and sell existing equity. He formulates the recursive
equilibrium and analyzes how the equilibrium responds to liquidity shocks in
the equity market. A main result is that a negative liquidity shock alone gen-
erates an equity price boom. Because this response is opposite to the observed
behavior of equity price in the business cycle, it indicates that shocks to equity
market liquidity are not the primary driving force of the business cycle. For
equity prices to fall after a negative liquidity shock, as is typically seen in reces-
sions, the equity liquidity constraint must become less tight, which is possible
only if there are other concurrent shocks that sufficiently reduce the need for
investment in the economy. Thus, if a shortfall in liquidity is driven by purely
non-fundamental events, then supplying liquidity may be a good policy. But if
a shortfall in liquidity is generated by a deterioration in the quality of invest-
ment, supplying liquidity to the market does not seem a good policy, because it
acts as a subsidy to low-quality investment.
Chapter 7, ‘Whither Capitalism? Financial Externalities and Crisis’, by Marcus
Miller and Lei Zhang, attacks the issue of the ongoing crisis by looking at it from
the perspective of failures in the financial markets. They find that externalities
are essential to understanding what went wrong – and what seems to be con-
tinuing to go wrong in Western countries that have hardly rebounded, if at
all, from the crisis. They focus on three main sources of externalities in the
financial sector. One is ‘fire sales’ of collateral assets, which they argue has led
to the undervaluing of liquidity. The second is the use of limited liability that
allows highly leveraged financial institutions to take on the benefits of risk-
taking while passing the losses and hazards on to others. Thirdly, they examine
the way in which financial networks actually amplify shocks – despite their sup-
posed design to do the contrary – if there is no ring-fence or circuit-breaker. To
show the degree to which these externalities matter, Miller and Zhang compare
the aftermath of the crisis in hard-hit Western countries, where the financial
capitals are, to the briskly and so far sustainably growing Asian economies.
The last two chapters in this Part deal with interactions of financial factors and
aggregate economic activity, with focus on roles of financial frictions in business
fluctuations. Chapter 8, ‘Bank Lending and Credit Supply Shocks’, by Simon
Gilchrist and Egon Zakrajšek, empirically analyzes the linkage between credit
supply and macroeconomic activity via bank lending. Building on their earlier
work (Gilchrist and Zakrajšek 2011), they use the excess bond premium – a com-
ponent of corporate credit spreads designed to measure shifts in the risk attitudes
of financial intermediaries – to empirically identify credit supply shocks. Their
results show that shocks thus measured lead to a pronounced and protracted
contraction in economic activity, a decline in nominal interest rates, a sharp
fall in equity valuations, and an eventual decline in outstanding business loans.
To shed light on the lagged response of banks’ lending to credit supply shocks,
they further examine the joint response of outstanding loans and unused com-
mitments. They found that in the initial phase of the downturn, the capacity of
businesses to borrow from the banking sector shrinks primarily through reduc-
tions in unused commitments and only eventually through a reduction in loans
outstanding. They argue that these differential dynamics between on- and off-
balance sheet credit exposure suggest that macroprudent policy should take into
account banks’ off-balance sheet exposures when assessing the overall risk of the
financial sector.
Chapter 9, ‘A Mechanism Design Approach to Financial Frictions’, by
Nobuhiro Kiyotaki, studies how financial frictions become endogenous out-
comes of underlying environments such as private information, limited com-
mitment on the side of agents, and limited contract enforcement on the side of
the intermediary. By extending and modifying the mechanism design approach
of Green and Oh (1991) to financial intermediation, he characterizes different
contractual forms as an equilibrium outcome of the underlying environments.
Specifically, he shows that limitations on commitment and private information
of individual income and trades lead to the financial structure under which
the present value of individual consumption is equal to individual income –
there is no insurance as in the case of the Arrow-Debreu Economy. Moreover,
the low-income agents face a binding borrowing constraint – arguably the most
common contract used in practice. Given that financial intermediaries expe-
rienced significant financing constraints during the recent financial crisis, he
proposes that the explicit account of incentive constraints of the intermediary
should be on the imminent research agenda of the mechanism design approach
to financial frictions.
Behavior of financial institutions and prudential regulations
Chapters in this Part focus on the behavior of financial institutions and regu-
lations on them. The financial services industry is the most regulated industry
Introduction 7
in most countries. Much of this regulation was designed to prevent financial

crises. Unfortunately, it failed to do this. The basic idea behind much of it was
microprudential in that, if individual banks and institutions were prevented
from taking risks, then the financial system itself would not be risky. The prob-
lem is that this ignores systemic risk that can come about in a number of ways.
Chapter 10, ‘Systemic Risk and Macroprudential Regulation’, by Franklin Allen
and Elena Carletti, identifies six different types of systemic risk and discusses
the potential for controlling them. The first is exposure to asset price bubbles,
particularly real estate bubbles. This was the systemic risk that set off the crisis in
the US in 2007. The second systemic risk considerably exacerbated the problems
caused by falling real estate prices. This was the problem of liquidity provision
and mispricing of assets in financial markets due to limits to arbitrage. The third
type of systemic risk is the traditional view of panics based on the notion of
multiple equilibria. If everybody thinks the financial system is sound, it will be.
However, if everybody thinks there is a crisis they will all attempt to withdraw
and the crisis will be self-fulfilling. The fourth kind of systemic risk is contagion.
If one institution fails then others connected to it may also fail, and so on, so
that the whole financial system melts down. The fifth is sovereign default and
the sixth is currency mismatches in the banking system. This chapter considers
in detail macroprudential regulations aimed at counteracting them.
Chapter 11, ‘Filling the Gaps: The Vienna Initiative and the Role of the Inter-
national Financial Institutions in Crisis Management and Resolution’, by Erik
Berglof, considers a successful intervention by international financial institu-
tions in central and eastern Europe to prevent the financial crisis escalating. This
was known as the Vienna Initiative. During the 1990s foreign banks expanded
through subsidiaries in central and eastern Europe. In many countries this pro-
cess culminated with foreign subsidiaries having more than 80 per cent of total
banking assets. However, there was effectively no cross-border regulation by
the European Union or other international bodies and when the crisis struck
there was a significant chance of a disaster if foreign banks withdrew funds and
took them to their parent countries. By facilitating coordination between home
and host countries, international financial institutions such as the International
Monetary Fund and the European Bank for Reconstruction and Development
helped to introduce the Vienna Initiative. Under this, large multinational banks
agreed to limit their withdrawals from the region. The general view is that it
was very successful. The feared large-scale uncoordinated withdrawals did not
occur and no subsidiaries failed. Berglof argues that this intervention provides
an example of effective intervention by international financial institutions in
future crises.
One of the important issues that has been debated for many years in the lit-
erature is whether there is an optimal financial structure for development. The
traditional view has been that financial depth rather than financial structure
determines how rapidly an economy develops. Chapter 12, ‘Some Recent Pro-
gresses in Financial Structure and Development’, by Justin Yifu Lin and Lixin
Colin Xu, reconsiders anew the question based on a number of recent papers that
suggest there is in fact an optimal capital structure that depends on a country’s
stage of development. There is evidence that financial structure is endogenously
determined by the demand for financial services that in turn is determined by
industrial structure. Thus optimal financial structure is specific to a country’s
stage of development. As countries become richer, their economic development
becomes more sensitive to stock market development and less sensitive to bank
development. Evidence based on firm-level data shows that bank development
has particularly strong effects in relatively poor countries. Banks are relatively
effective in reducing poverty, particularly in countries with weak institutions.
One important policy implication of this research is that financial structure
becomes an independent financial policy consideration that should be matched
to a country’s level of development.
Chapter 13, ‘The Race to Zero’, by Andrew Haldane, sharply focuses on secu-
rities markets and outlines how shifts in the structure and speed of trading have
increased abnormalities in securities pricing and potentially increased systemic
risk. In particular, the change in speed has been driven by the dominance of
high-frequency trading (HFT) in a number of markets. There is evidence that
these developments have led to more frequent periods of higher volatility and
correlation across markets since 2005. The chapter outlines a framework for
understanding these effects, drawing on Mandelbrot’s fractal geometry tech-
niques. It notes that the cause of price dislocation may be the disappearance of
liquidity in stressed situations. The advent of HFT may have made this more
likely, with HFT firms more inclined to withdraw liquidity and longer-term
investors either unable or unwilling to fill the gap. This leads to a potential
double liquidity void and a greater dislocation of prices in times of stress. In
addition, the changing structure of trading means that high-frequency, local-
ized price disturbances could be magnified across time, markets, and assets. The
chapter goes on to discuss several policy options for mitigating the impact of
these developments on market dynamics. These include circuit-breakers to halt
trading and establish a level informational playing field; and a speed limit on
trades through minimum resting periods to make bid-ask spreads less variable,
especially in situations of stress.
Chapter 14, ‘A Model of Private Equity Fund Compensation’, by Wonho
Wilson Choi, Andrew Metrick, and Ayako Yasuda, considers a narrower but
important issue, namely, how compensation in private equity investments
should be structured. Typically private equity funds are set up as limited part-
nerships that last for ten years. The general partners provide the investment
expertise while the limited partners provide the financing. The basic principle in
the structuring of funds’ compensation is that the limited partners must recover
Introduction 9
the money they put up, known as their ‘basis’, before the general partners receive
compensation. One of the most commonly used forms of contract is the fair-
value test carried interest (FVT) scheme. This allows early payments to general
partners before the limited partners have received their basis. However, there are
limits that depend on the value of the remaining investments in the fund and
there is almost always a ‘clawback’ provision that allows funds to be reclaimed
from general partners if necessary at the end of the fund’s life. The chapter uses
simulation techniques to compare the FVT scheme with other commonly used
arrangements. The authors find that the FVT scheme is favorable for general
partners but does not provide significant incentives for them to overstate the
values of un-exited investments in the fund.
Corporate taxation in a globalized economy
With globalization, there is an increasing amount of cross-border activity by

multinationals. How should this activity be taxed in the home country? Answers
have been in flux, both in practice and in the academic literature. In practice,
tax structures are limited by OECD conventions, which allow countries either
to exempt from domestic corporate taxes the foreign-source income earned by
multinationals or alternatively to tax it at repatriation but with a credit for any
taxes paid abroad on this income. While the US follows the second conven-
tion, the UK and Japan recently shifted to the exemption method and there are
proposals in the US for it to make the same change.
The traditional recommendation in the public finance literature, dating back
to Richman (1964), is to tax all net income of multinationals at the same rate,
regardless of source, a concept known as ‘capital-export neutrality’. More recent
papers, such as Desai and Hines (2004), note that there need not be any capital
exports when a foreign subsidiary is acquired, for example, when needed funds
were borrowed. Any tax on the resulting productivity gains, arising from use of
the multinational’s intellectual property, distorts the acquisition choice, leading
to a recommendation of tax exemption in order to achieve ‘capital ownership
neutrality’. The two chapters in this Part each examine different aspects of this
debate.
The first, ‘How Should Income from Multinationals be Taxed?’, by Roger
Gordon, focuses on the interactions between the corporate taxes on a foreign
subsidiary and the domestic personal income taxes in the home country. The
income of the subsidiary can represent not only a return to capital invested
in the subsidiary but also a return to the intellectual property developed by the
multinational. A small country would have no incentive to tax the return to cap-
ital financed from abroad, and there are many papers arguing against taxes on
the return to the savings undertaken by domestic residents, whether invested
at home or abroad. However, domestic-source labor income remains taxable.
Unless the return to the creative efforts of domestic entrepreneurs accruing

abroad is taxed at the same rate as other forms of labor income, the tax sys-
tem introduces added efficiency costs from distorting the form and location of
labor effort. Strikingly, neutral taxation then leads to the second convention
under OECD rules: full taxation at repatriation with a credit for any taxes paid
abroad.
The final chapter, ‘Taxing Multinationals in a World with International Merg-
ers and Acquisitions: Should the Home Country Exempt Foreign Income?’, by
John Wilson, focuses on the appropriate tax treatment of income accruing to a
foreign subsidiary, given that domestic profits are taxable. If opportunities are
available to multinationals for deducting the costs of their foreign investments,
the tax system effectively lowers these costs and thereby reduces or eliminates
the disincentive effects of taxes on investing abroad. Activity by the foreign sub-
sidiary interacts with the domestic operations of the multinational, interactions
that will be taken into account by the firm. A higher tax rate on foreign-source
income (net of investment costs) leads to more investment abroad if this invest-
ment raises the firm’s domestic productivity, and vice versa. The tax increases
domestic productivity and the domestic tax base, generating an efficiency gain
and leading to a finding that the foreign profits should be taxed by the home
country when such spillovers exist.
References
Desai, M. and J. R. Hines, Jr (2004) ‘Old Rules and New Realities: Corporate Tax Policy in
a Global Setting’, National Tax Journal, vol. 57, no. 4, pp. 937–960.
Gilchrist, S. and E. Zakrajšek (2011) ‘Credit Spreads and Business Cycle Fluctuations,
forthcoming in American Economic Review.
Green, E., and S.-N. Oh (1991) ‘Can a “Credit Crunch” Be Efficient?’, Federal Reserve Bank
of Minneapolis Quarterly Review, vol. 15, no. 4, pp. 3–17.
Nordhaus, W. D. and J. Tobin (1972) ‘Is Growth Obsolete?’, in Fiftieth Anniversary
Colloquium, Vol 5: Economic Growth (Washington, DC: NBER), pp 1–80.
Richman, P. (1963) Taxation of Foreign Investment Income: An Economic Analysis (Baltimore:
Johns Hopkins University Press).
Stiglitz, J. E., A. Sen and J.-P. Fitoussi (2009) Report by the Commission on the Measurement of
Economic Performance and Social Progress, at: www.stiglitz-sen-fitoussi.fr/en/index.htm.
Published as J.-P. Fitoussi, A. Sen and J. E. Stiglitz (2009) Mismeasuring Our Lives: Why
GDP Doesn’t Add Up (New York: The New Press).
Part I
Beyond GDP
1
On the Measurement of Social
Progress and Wellbeing: Some
Further Thoughts
Jean-Paul Fitoussi
Sciences-po, Paris
Joseph E. Stiglitz
There is not a single year where our measurement systems are not called into
question, and as a consequence it will take more time than we would like to
understand what is going on in the world economy.
The ‘financial’ crisis revealed that we (and especially the United States) were
not doing as well as we thought we were when looking at the available metrics.
That is, we realized that economic growth was not sustainable, and the output
measures had been exaggerated by bubble prices in real estate and by fictional
profits in the financial sector. The fact that in some countries (such as the United
States) GDP has returned to pre-crisis level does not capture, in any way, the
diminution in the sense of wellbeing. With almost one out of six Americans who
would like a full-time job unable to get one – and others facing high anxiety at
the risks they face of the loss of a home or a job – and cutbacks threatened in the
basic public expenditures programs, the loss in wellbeing is enormous. The situ-
ation in Spain is even worse, with an unemployment rate higher than 24 per cent
on average and more than one out of two young Spanish unemployed.
The events in Japan can be seen as a metaphor of our measurement prob-
lems. Some suggest that while in the short run GDP may decrease, in the long
run it will rise as a result of the reconstruction efforts. The nuclear disaster has
increased anxieties – and may well have significant health effects on large num-
bers of the population. Again, the expenditures required to respond may raise
GDP, perhaps enough to get Japan out of its long-standing economic malaise.
But no one would claim that Japan is better off as a result. It would require a
huge increase in GDP to compensate for the destruction of capital, of all kinds
of assets, that the event has caused, and to offset the increased anxieties that
so many in the country face today. And we are not good – our metrics are not
13
14 J.-P. Fitoussi and J. E. Stiglitz
adapted – at measuring the value of the lost assets. Even if it were, the arithmetic
of compensation will not tell us much about the way the wellbeing of Japanese
people has evolved. The mechanical nature of our economic models will tell
us nothing about the immaterial consequences of the irreversible losses of the
people. In the aftermath of the crisis, we now realized that our measurements
before the crisis were also not accurate. GDP may have been higher because
of the greater efficiency (cost savings) as a result of the reliance on nuclear (as
opposed to, say, renewable) energy. The placement of the spent nuclear material
in a way that exposed the entire country to risks that are now so evident may
also have contributed to a seemingly higher GDP then. But just as accounting
frameworks before the financial crisis mispriced risk, so they did in Japan. The
Japanese case is thus a metaphor because it underlines the three shortcomings
of our metrics: the measurement of the ‘economic product’, the measurement
of wellbeing, and the measurement of sustainability.
Another universal fact well documented is the intra-country increase in
inequality, which has characterized at least the past quarter of century. Look-
ing at the growth of GDP or at that of net income would tell us nothing about
this fact, and would certainly give us a wrong impression about the evolution
of societal wellbeing. A striking fact is that in OECD countries the increase of
income of 80 per cent of the population has been lower than the rate of growth
of the overall economy (which is, obviously, an average), and the more so the
lower is the decile considered. If we seek numbers that assess the impact of eco-
nomic growth on society as a whole, surely we want to know what is happening
to most citizens. GDP tells us nothing about that.
Another example is provided by the revolution in the Arab world – especially
in Tunisia – which opened the process. Some economists1 think that political
freedom is a luxury good that leads to a lower rate of growth, because of the quest
for redistribution to which it leads. Setting aside whether such claims rest either
on sound theoretical or empirical foundations (at least in the case of Tunisia, the
lack of democracy contributed to corruption, which had an enervating effect on
growth), here too the concept used is misleading. GDP is not a measure of well-
being. Even if it could be shown in regressing growth of GDP on some indices
of political freedom that limiting political freedom leads to increased GDP –
and quite apart from the fragility of such empirical exercises – the conclusion
that countries would be well-advised to postpone democratization until they
can afford this luxury makes no sense. It may well be that wellbeing increases
more from an increase in political freedom than from an increase in GDP, espe-
cially given the way GDP is measured. In debating about the effect of political
freedom on the evolution of GDP, we are missing an essential point: the risk
taken by the people to fight for freedom is a testimony that it is a fundamental
component of wellbeing.
The Measurement of Social Progress and Wellbeing 15
These are just some examples of how our present statistical system, with both
flaws in the available metrics and the absence of alternatives, may implicitly
lead to erroneous policy conclusions.
All of this is important because what we measure affects what we do. Reducing
wellbeing to increase whatever imperfect measure of material wealth gives rise
to totally flawed policies.
1 Metrics and policies
For economists, these concerns are especially important, because we often rely
on statistical (econometric) analyses to make inferences about what good poli-
cies are. Those inferences are only as reliable as the data that they are based on.
Some studies suggested that financial market or capital market liberalization
contributed to higher economic growth. It is now clear that such studies’ con-
clusions were flawed because: i) GDP numbers in the growth spurts were exagger-
ated by the bubbles that are often associated with such liberalizations; ii) unless
an adequate time horizon is taken, the losses that follow the crashes will not be
taken into account – and these losses may more than offset the short-term gains
arising from the bubbles to which liberalization often gives rise; iii) the distribu-
tive consequences of those policies are not taken into account – so that even if
GDP goes up, it may be the case that most citizens are worse off; and iv) the costs
to wellbeing – from, for example, the insecurity that follows the volatility that
typically accompanies such liberalization measures – are not taken into account.
More generally, the empirical studies conducted to demonstrate the beneficial
effect of financial market liberalization on growth and employment are vul-
nerable to the same kind of limitations. There is thus a hiatus between some
of the usual policy recommendations and the weaknesses of the evidence to
support them.
To take another example, there has been a wealth of econometric studies
aimed at showing how certain labor market institutions and the adverse effects
they have on the flexibility of the labor market affect unemployment and
growth. These studies of the impacts of the particular institutions under study
on unemployment are, at best, able to explain effects that are of second order of
importance. Two studies based on a sample of 19 OECD countries independently
conducted on the subject2 reached the same conclusion. Capitalism is evidently
sufficiently robust to accommodate rather different institutional settings.3
Putting aside for the moment an evaluation of the contention that more flex-
ible labor markets do increase GDP (reducing the cumulative disparity between
actual and potential GDP), of what moment is this observation if GDP is not
the right measure of societal wellbeing?4
Many of the advocates of unfettered markets – who see any interven-

tion, whether through, say, public policy or private institutions, as welfare-
decreasing – ignore the extensive and well-documented ‘market failures’, which
are especially widespread in labor and capital markets. Some institutions have
been created to make up for failures in insurance markets and training. If our
metrics do not capture the benefits of the greater security that unemployment
insurance provides, the reform that is needed is not the abolition of unemploy-
ment insurance, but of the flawed metrics. Some of these institutions can be
thought of as reflecting a social contract that arises from a democratic pro-
cess. There are winners and losers to any structural reform, so such reform
is unlikely to lead to a Pareto-improving outcome, or even one supported
by a majority of the electorate. But, and even more importantly, the move
towards a greater flexibility of the labor market could negatively affect at least
two of the main objective determinants of wellbeing: the quality of jobs (the
quest for a decent job) and economic security.5 In short, out of very weak
evidence drawn from ill-measured phenomena, we could draw policy recom-
mendations the implementation of which may reduce the wellbeing of the
people.
Of special concern are the econometric inferences about good policies we
draw from cross-country regressions. Whether we like it or not, international
comparisons of levels and more importantly of rates of growth play a very impor-
tant role in the design of policy. To some, cross-country regression provides us
greater confidence in making these comparisons, by isolating the effects of cer-
tain factors that are thought to help explain differences in performance across
countries.
Many of the critiques of this methodology are well known. For instance, usu-
ally this type of exercise proceeds by constraining coefficients to take the same
value across countries as if it were a single model (both economic and social)
that is able to explain economic and welfare outcomes all over the world quite
independently of the specific choices and institutional arrangements made by
different countries. If the equation fits for one group of countries but not for
another the panel estimation may yield significant results due only to the inclu-
sion of the first group. The implications are obvious: it would be wrong to extend
the inferences to a country that belongs to the second group.
Our concern here with cross-country regressions is about another caveat.
Comparisons are only meaningful if the procedures and definitions used to
compute the accounts are comparable and if there are not inbuilt biases in
the construction of the data series themselves. Yet there are still ‘large differ-
ences in the ways national accounts calculations are carried out even among
European countries, let alone between Europe and the US’ (Hartwig 2005).
This may have far-reaching consequences. It makes no sense, for instance,
to structural reforms intended to import the ‘best practice’ of the country
performing the best in terms of growth rate, if the growth rates of the two
countries differ mainly because of differences in the ways national accounts are
computed.
Another example, reflective of a failing in GDP measurement that has long
been recognized, arises in analyses of the effect of the size of government on
growth. Because output in the public sector is typically measured by its inputs,
there is an implicit assumption of no productivity growth, when in fact in
some cases (where detailed studies have been conducted, or on the basis of
casual empiricism) we know there is rapid productivity growth. Inevitably, such
assumptions bias cross-country regressions, to suggest that a larger public sector
is associated with smaller rates of productivity growth. The result is not a deep
empirical insight; it is simply a statistical artefact of measurement. Consider, for
instance, what might happen if one were to privatize America’s social security
(the public old age pension system). We know that transactions costs for that
system are an order of magnitude lower than for private annuity programs. It is
extraordinarily efficient, and surveys have shown that it is also very ‘customer
responsive’. Privatization would result in higher profits for America’s financial
services industry and lower benefits for America’s retirees. The higher profits
would likely be reflected in an increase in GDP. But the wellbeing of Ameri-
cans would be decreased, and the gains of the financial industry would be at
the expense of the average retirees. Wellbeing, appropriately defined, would go
down. But it is easy to see how in the mindless cross-country regressions that
have become the fashion, one might conclude that such a privatization would
be good for ‘growth’. And more importantly, even if it were – and it is not –
should we conclude that democracies all over the world should choose a small
state (and become impotent)?
2 Use and misuse of the concept of sustainability
We can do better. The Commission on the Measurement of Economic Perfor-

mance and Social Progress identified a number of reforms (Fitoussi et al. 2009).
Some would lead to a better GDP metric – so that even if GDP is not a measure
of wellbeing, it is a better measure of whatever it is that it is attempting to mea-
sure. In effect, the purposes of our statistical systems are multiple, and a metric
that is adapted to one purpose may be ill-suited to another. Sometimes confu-
sion is engendered when a measure adapted to one purpose is used to highlight
another. For example, GDP is neither a measure of income nor a measure of
wellbeing. What we want to measure is the key question. We may want to mea-
sure, for instance, the levels of market activity – one of the original objectives
of national income measurement.
But, increasingly, there is a demand to go beyond measures of market activity
to measures of wellbeing. And even before the crisis, there were worries about
sustainability – and that our metrics did not tell us anything about whether
what we were doing was sustainable.
That is why some would encourage more focus on other metrics. Looking
at the (real) income of the median individual would give us a better picture
of what is happening to the typical individual in society than GDP per capita
does. Before the crisis, many thought that the US had been performing well.
But if they had looked at median income, they would have seen that incomes
were stagnating or declining – and such measures did not even account for the
greater insecurity resulting from reduced health insurance coverage or weaker
retirement protection, which in turn is a result of the shift from defined benefit
programs to defined contribution programs. Should we pursue growth at the
expense of the sustainability of one’s life?
We care about the future and that the living standards that we enjoy today
should be enjoyed by future generations. Our statistical systems should tell us
whether or not what we are doing is sustainable – economically, environmen-
tally, politically, or socially. There is reason to believe that, at least in certain
dimensions, what we are doing is not sustainable, but current statistics do not
reflect this – just as they gave little indication of the unsustainability of the US
economic growth in the years preceding the crisis.
It is important for any society to form an assessment, no matter how imper-
fect, about whether its current consumption or wellbeing is sustainable, and
whether this is coming at the expense of future generations. We can ascertain
whether a society’s wealth is increasing or decreasing (per capita). If (appropri-
ately measured) it is increasing, then presumably society can do in the future
whatever it did today, that is, it can sustain its per capita income. But we need a
comprehensive measure of wealth, and we need to be sure that the valuations are
correct. A comprehensive measure obviously includes measures of physical cap-
ital, human capital, and natural capital (including the environment). Changes
in capital include those arising from investment in plant and equipment, edu-
cation, the depletion of natural resources, depreciation of physical capital, and
environmental degradation.
We know that prices do not adequately reflect the true social costs of carbon
emissions and the risks that a major change in carbon prices would impose on
all asset prices. Thus, we felt reluctant to use, or at least rely on, market prices
to assess environmental sustainability, suggesting instead the concurrent usage
of physical metrics.
One of the problems encountered in the aftermath of the financial crisis is the
misuse of the concept of sustainability. The lack of an indicator of sustainability
may lead us to an unsustainable path, but a partial measure may lead us to wrong
policies which would eventually jeopardize the sustainability of an economy. A
case in point is Europe. Sharing a common currency in a global crisis, Eurozone
countries are currently looking for sustainability indicators in order to assess

the financial sustainability – by which they essentially mean the sustainability
of the public debt – of each member country: they are trying to define sustain-
ability objectives, implement economic policies that are ‘sustainability friendly’
and spread information to financial markets in order to reduce pressure on pub-
lic and private sector borrowing. The problem is that European countries are
focussing on a very partial view of sustainability, namely the sustainability of
public debt, which led them to impose on peripheral countries austerity pro-
grammes, that is, procyclical policies, which would likely result in a much lower
rate of growth and may eventually lead to financial unsustainability both in the
public and the private sector. A comprehensive measure should at least take
into account the destruction of human capital to which these policies lead in
increasing both the rate and the duration of unemployment. Whatever mea-
sure we devise, we have to recognize that it will be grounded on our present
imperfect knowledge of the future.
Inevitably, metrics are partially grounded in models: many of the metrics
upon which we focus are not ends in themselves, but are viewed as intermediate
variables of interest because they provide insight into things we really care about.
But the relationship between these intermediate variables and the things we
really care about is often uncertain, and depends on the model. However, there
is a great deal of uncertainty about the right model. For instance, prior to the
crisis, many believed that all a country needed to do to maintain high and stable
growth was to maintain low and stable inflation. In the aftermath of the crisis,
there is broad consensus that low and stable inflation was certainly not sufficient
for economic stability. Those policy conclusions were themselves predicated on
wrong models. And those wrong models encouraged economists to focus on a
single variable, inflation, as an indicator of the country’s future prospects. We
now know that there should have been more focus on indicators of financial
stability.
Similarly, those who argue a second metric of focus should be debt/GDP, and
that debt/GDP ratios above a given level (say 80 per cent) are not sustainable are
basing this conclusion on models. Given the present level of wealth (both pub-
lic and private), even higher levels of debt might be sustainable, if there is rapid
enough technological progress. Greece is not bankrupt, but threatened to be so.
Financial markets’ judgment is a (possibly) educated guess about what will be the
future, but can’t be more than that, given their fantasy in the valuation of asset
prices. Keeping that in mind, it may not be a terribly good idea to force the gov-
ernment of Greece to privatize public assets to reduce its public debt. A fire sale
would actually worsen the government’s balance sheet. It will also lead to a dis-
mantling of the public sector, which will limit the scope for future intervention,
and through this limitation may impede policies geared toward sustainability.
3 Assessing wellbeing
The members of our Commission felt strongly that GDP did not provide a good
measure of wellbeing, even contemporaneously – setting aside the question of
whether current standards of living were sustainable. We urged the construction
of broader measures of wellbeing that would take some account of some of the
most important factors that affect wellbeing that were not yet included in GDP
metrics, like connectedness.
But in our report, and in the discussion surrounding its presentation, we
also highlighted another one of the debates over measurement: While a focus
on false measures might distort policy, a dialogue around what we, as a society,
care about, and whether these concerns are adequately reflected in our statistics,
could contribute not only to an enhanced understanding of the limitations of
these standard statistical measures, but also to the formulation of better policies,
more reflective of the concerns and values of citizens.
We believe that this has, in fact, been the case. In many countries – notably
in France, Germany, Italy, and the UK – steps have been taken to implement
some of the recommendations of our report. But the most comprehensive exer-
cise undertaken has been the OECD study released in May 2011, the OECD
Better Life Initiative. It shows the will of the Organization to dialogue with the
civil society by allowing each citizen to build his or her own aggregate index
of the quality of life (see http://oecdbetterlifeindex.org). Eleven indicators have
been selected by the OECD for its 34 member countries and some emerging
countries, according to the domains identified in our report. People are asked
to compute their own index by selecting the weight of each determinant of
wellbeing through an interactive Internet tool called ‘Your Better Life Index’.
All but one of the determinants considered are objective (Health, Employment,
Education, Housing Conditions and others); but one pertains to the subjective
category ‘life satisfaction’. It is obtained through surveys. The subjective deter-
minants of wellbeing are obviously important. In effect, a long philosophical
tradition views individuals as the best judges of their own conditions. But they
are subject to a kind of ‘time inconsistency’ problem, as they may evaluate their
circumstances (or even a particular event) in a different way at different peri-
ods of time. While some people might describe parenthood as painful at the
moment they are raising young children, 20 years later they may remember
this period as the most satisfactory of their life.
How to interpret and use these different results in developing wellbeing met-
rics is a subject on which there is ongoing research.6 We are hopeful that not
only will this research lead to better metrics, but active engagement with civil
society will result in policies that are directed at the improvement of societal
wellbeing – reflected not in some flawed measure of GDP, but of the newly
constructed measures.
Surely in the present circumstances, most of our countries need more growth;
But growth of what? The usual answer is the growth of GDP but a better answer
should be the growth of wellbeing, that is, of what really matters for citizens. The
shift from the former to the latter objective probably requires an enrichment of
the instruments of public policy and a much more selective approach to policies
aimed at increasing GDP: it will serve no purpose if this aim is achieved at the
expense of wellbeing.
4 Wellbeing and the business cycle
This broader and redefined perspective on measurement is relevant not only

for assessing the long-term progress of society, but also for understanding cycli-
cal fluctuations, such as the one that much of the world is now experiencing.
(Ironically, it was precisely to understand cyclical fluctuations that the narrower
GDP measure was originally developed.) Earlier we noted that before the crisis,
GDP was exaggerated. But one can argue that in the crisis, the loss in well-
being may also be underestimated. The Commission’s initial report emphasized
the importance of employment itself. One point where various subjective mea-
sures of people’s wellbeing agree is that unemployment has a very adverse effect
on people’s quality of life. People who become unemployed report lower life-
evaluations, even after controlling for their lower income. The adverse effects
persist over time. The unemployed also report higher prevalence of various neg-
ative feelings (sadness, stress, and pain) and lower levels of positive ones (joy).
One may also suspect that the adverse effects of unemployment are felt even
by those who are not themselves unemployed, especially in societies where
there is high unemployment. These subjective measures suggest that the costs
of unemployment exceed the income loss suffered by those who lose their jobs,
reflecting the existence of non-pecuniary effects among the unemployed, and
of fears and anxieties generated by unemployment in the rest of society.
In the US, some seven million families have already lost their home. Both this
and unemployment are contributing to increased levels of anxiety, even among
those still with homes and jobs. There are identifiable effects of home ownership
on individuals’ sense of and participation in community and investments in
local schools – and possibly therefore in the future wellbeing of their children.
There is evidence too of adverse health effects (see Currie and Tekin 2011).
All of this suggests that economic fluctuations may have strong asymmetric
effects on wellbeing – something we should already intuitively have known.
Moreover, some of the consequences (such as on health and education) may
be irreversible. However, the prevalent use of GDP as the intermediate indicator
results in our not taking, explicitly, account of these adverse effects, not only
on the current level of wellbeing, but also on the ‘stock’ of human capital. The
economics of the business cycles should be rethought in light of the probable
discrepancies between the fluctuation of output and that of wellbeing. It may

well be that it would be better for governments to devise policies aimed more
at minimizing the rate of unemployment and its variation over the business
cycle rather than policies aimed at maximizing output growth (as measured by
GDP). Some of the instruments for implementing these two strategies may be
the same – employment concerns are central to overall macroeconomic strat-
egy – but surely the first strategy needs specific, supplementary instruments
to smooth the evolution of unemployment. Risks, too, may be very asymmet-
ric: an extended period of high unemployment may have far higher long-term
consequences than those that might arise from a slightly overheated econ-
omy. Moreover, in light of the preceding arguments, such a strategy will surely
enhance wellbeing even if there were some adverse effects on growth, as mea-
sured by GDP.7 The design of good policies can’t be grounded on the artificial
separation between social policies and macroeconomic ones: if the wellbeing
of the people is the ultimate end, employment, labor market analysis, and
income distribution must be central components of the macroeconomic analysis
supporting stabilization policies.
5 Beyond GDP: the experience of Bhutan
We should note one further example – Bhutan – whose quest for better metrics
began long before the work of our Commission. Some 40 years ago, then-King
Jigme Singye Wangchuck announced that the country’s goal was not to maxi-
mize GDP but to increase GNH, gross national happiness. Rather than turning to
an Economic Development and Planning Agency for formulating development
strategies, the country established its Gross National Happiness Commission.
It was more than a matter of words. Questions were raised that typically do
not get raised in a single-minded focus to increase GDP: a) What is the impact
on the environment? This is not typically priced correctly within GDP. Greater
forest cover might increase GNH, even if cutting down forests might create, in
the short run, increased GDP; b) What is the impact on ‘social capital’ (social
cohesion)? This is something that is virtually never priced into GDP. Trust in
government can allow better compliance with environmental regulations (with-
out which restrictions on cutting down forests would be very hard to enforce)
or more responsiveness to government efforts to improve the education and
health of children – actions that almost surely will improve GDP in the future,
but with benefits that do not show up in today’s GDP.
Out of this has grown a more holistic approach to development, which sees it
as a transformation of society, receiving the benefits of modernization (such as
greater literacy, more political participation, and better health) while retaining
traditional values and a national sense of identity. Development is seen as more
than just the accumulation of more factors of production or an increase in static
efficiency. New questions are asked, and out of these new approaches are taken:
what is the impact on entrepreneurial and societal learning? Opening up con-
struction bids for a new school to all contractors, foreign or domestic, might lead
to lower short-run costs – a seemingly better economic performance today – but
encouraging local builders using local materials and techniques, and designs
that conform to local preferences and which might have relevance for other
construction activities might have much greater long-term growth benefits.
Bhutan is consciously involved in a process of societal transformation, and
so, for that country, it was imperative that they think deeply about the ways in
which this was happening. But while other countries may not be consciously
planning for societal change, it is nevertheless occurring everywhere, even if
slowly and in a far more evolutionary way. If well designed, our metrics can
give us indicators of where we are, and over time can provide a picture of where
we are going. They can give us information that can facilitate assessments of
whether we are achieving our objectives – and, even if we are succeeding in
the metrics that we set as our objectives, whether there are other less imminent
consequences that we need to address.
Our assessment of current metrics has left us convinced that, too often, they
have led countries to set off in the wrong direction, or at least to adopt policies
of ambiguous benefit. And our quest for better metrics has convinced us that
there are today available metrics that could provide better guidance. And our
research has shown that there is considerable scope for improvement in these
metrics and the development of new metrics that will correspond more closely
to societal objectives.
Notes
1. See, for example, Barro (1996).
2. See Fitoussi and Passet (2000) and Freeman (2000).
3. That conclusion is at odds with the common wisdom according to which the diversity
of institutional structures plays a determinant role in explaining both unemployment
and growth. Institutions do matter – the Scandinavian experience shows that active
labor market policies and the corresponding institutions may enable labor markets to
function better, at least in periods in which there is not a large deficiency in aggregate
demand. Our discussion here focuses on those institutions that allegedly lead to less
flexible labor markets.
4. In particular, research growing out of the Fisher-Greenwald-Stiglitz debt-deflation liter-
ature shows that with imperfectly indexed contracts, greater wage and price flexibility
may be associated with deeper downturns and slower recoveries. Indeed, in a cross
section study of volatility, wage and price rigidities were far less important than finan-
cial market factors. See Easterly et al. (2001a, 2001b, 2003). Weaker job security will
reduce workers’ willingness to invest in firm-specific capital, and thus can undermine
growth and productive efficiency.
5. Security is, of course, an important aspect of what is viewed as a good job.
6. For two earlier surveys, see Sunstein et al. (2002) and Kahneman and Krueger (2006).
7. We should emphasize, however, that there are some reasons to believe that a greater
focus on employment security might also enhance not just current wellbeing, but even
growth, for example, by facilitating greater investments in human capital and a greater
willingness to undertake risk.
References
Barro, Robert (1995) ‘Determinants of Economic Growth: A Cross-country Empirical

Study’, NBER Working Paper no. 5698, August (Cambridge, MA: NBER).
Currie, J. and E. Tekin (2011) ‘Health Consequences of the Foreclosure Crisis’, working
paper, George Mason University, Fairfax, VA, May.
Easterly, W., R. Islam and J. E. Stiglitz (2001a) ‘Shaken and Stirred: Explaining Growth
Volatility’, in Annual Bank Conference on Development Economics 2000 (Washington, DC:
World Bank), pp. 191–212.
Easterly, W., R. Islam and J. E. Stiglitz (2001b) ‘Shaken and Stirred: Volatility and
Macroeconomic Paradigms for Rich and Poor Countries’, in J. Drèze (ed.), Advances in
Macroeconomic Theory, IEA Conference Volume 133 (Basingstoke: Palgrave Macmillan),
pp. 353–372.
Easterly, W., R. Islam, and J. E. Stiglitz (2003) ‘Volatility and Macroeconomic Paradigms
for Rich and Poor Countries’, in J. H. Drèze (ed.), Advances in Macroeconomic Theory:
Volume 1 (Basingstoke: Palgrave Macmillan), pp. 352–372.
Freeman, R. (2000) ‘Single Peaked vs Diversified Capitalism: The Relation Between Eco-
nomic Institutions and Outcomes’, NBER Working Paper no. 7556 (Cambridge, MA:
NBER)
Fitoussi J.-P. and O. Passet (2000) Reformes structurelles et politiques. . .‘modèles’ de pays,
en Reduction du chômage: les réussites en Europe, Rapport du Conseil d’Analyse
Economique, n. 23 (Paris: La documentation Française).
Fitoussi, J.-P., A. Sen and J. E. Stiglitz (2009), Mismeasuring Our Lives: Why GDP Doesn’t Add
Up (New York: The New Press) (Original report, 2009; New Press edition, 2010) Also
available at: www.stiglitz-sen-fitoussi.fr/en/index.htm (accessed 16 May 2011).
Hartwig, Joachen (2005) ‘On Misusing National Account Data for Governance Purposes’,
Working Paper 05-101, KOF Swiss Economic Institute (Zurich: ETH).
Kahneman, D., A. B. Krueger, D. Schkade, N. Schwarz and A. Stone (2002) ‘Toward
National Wellbeing Accounts’, American Economic Review, vol. 94, no. 2, pp. 429–434.
Kahneman, D. and A. B. Krueger (2006) ‘Developments in the Measurement of Subjective
Wellbeing’, Journal of Economic Perspectives, vol. 20, no. 1, pp. 3–24.
Sunstein, C. R., D. Kahneman, D. Schkade and I. Ritov (2002) ‘Predictably Incoherent
Judgments’, Stanford Law Review, 54 , pp. 1153–1215.
2
The Role of Statistics in the
United States’ Economic Future
J. Steven Landefeld
Bureau of Economic Analysis, USA
Shaunda Villones
Bureau of Economic Analysis, USA
The need for better statistics
The national accounts were developed to address the paucity of comprehensive

and consistent data confronting decision makers during the Great Depression.
Over the intervening years, the accounts have served macroeconomic policy
makers well, contributing to the unprecedented period of post-World War II
economic growth and prosperity. Despite this success, there have been contin-
uing calls – including those by the founder of US accounts, Simon Kuznets –
for an expansion of the accounts to cover household production, environmen-
tal externalities, and other near-market and non-market activities that affect
households’ wellbeing. One of the most eloquent critiques of the focus on GDP
as a measure of society’s progress was made by Robert F. Kennedy:
Too much and too long, we seem to have surrendered community excel-
lence and community values in the mere accumulation of material things.
Our gross national product, if we should judge America by that, counts air
pollution and cigarette advertising, and ambulances to clear our highways of
carnage. It counts special locks for our doors . . . Yet the gross national product
does not allow for the health of our children, the quality of their education,
or the joy of their play . . . it measures everything, in short, except that which
makes life worthwhile. And it tells us everything about America except why
we are proud that we are Americans.1
Robert F. Kennedy, Address, University of Kansas,
Lawrence, Kansas, 18 March 1968
Over time there have also been calls for the inclusion of estimates of human
capital, natural resources, and R&D and other intangibles to provide a bet-
ter understanding of the sources of growth in GDP and productivity. One of
the most comprehensive set of recommendations in this area was a 2005 US
25
26 J. S. Landefeld and S. Villones
National Academy of Sciences report, Beyond the Market (Abraham and Mackie
2005). More recently the Stiglitz-Sen-Fitoussi Commission Report (2009) and
the global financial crisis have pointed to the need to expand the accounts to
better guide economic policy and business and household decisions.
The SSF Report noted that while many of the concerns about GDP as a mea-
sure of standards of living has long been recognized, changes in the economy –
including trends in the distribution of income, concerns about global warming,
and questions about the sustainability of growth – have heightened the mea-
surement problems. At the same time the financial crisis has highlighted the
need to better measure the sustainability of trends in saving, investment, asset
prices, and other data that are key to understanding business cycles, and the
sources of economic growth.
Many of these concerns can be addressed within the framework of the existing
national accounts, and those are described below. Another approach that has
been suggested is the measurement of ‘gross national happiness’ or more broadly
the development of subjective wellbeing accounts. While productive work has
been done in this area, particularly the work by Kreuger and his colleagues
(2009), such accounts may be best developed by academic and other researchers.
The degree of subjectivity and uncertainty in such measures present difficul-
ties for official statistical agencies. The concern is that the inclusion of such
measures – which are often very large in magnitude, would reduce the useful-
ness and credibility of the existing national accounts. Further, existing measures
of wellbeing are rather flat over time, fluctuating in a rather narrow range around
some ‘natural’ level of happiness.
For example, while the Gallup self-reported Well-Being Index (www.well-
beingindex.com) declined during the ‘Great Recession’ in the United States,
it rebounded quicker than most indicators of household economic wellbeing
and by the end of 2010 had returned to nearly its prerecession peak. Yet, the
US unemployment rate is still nearly twice its prerecession rate, and stock and
housing prices are 15 per cent below their prerecession levels. This example from
the recent recession, and the limited responsiveness in the post-World War II
era of subjective wellbeing indexes to major events, suggest that such indexes
may be of limited value in guiding economic or social policy, and legislation.2
The Bureau of Economic Analysis (BEA), however, has and continues to work
on household production accounts based on time use studies that could be
usefully linked to subjective wellbeing accounts, such as those developed by
Krueger et al. (2009).3
Distribution of growth and spending power
There is growing concern, heightened by the financial crisis, over the discon-
nect between the picture of the economy captured in the national accounts and
The Role of Statistics in the US Economic Future 27
Figure 2.1 Reconciling measures of income
Source: BEA, IRS Statistics of Income.
GDP and the individual experiences of households. The national accounts are
rich with data that can be used to construct measures that would highlight other
subcomponents of GDP that would provide better indicators of the household’s
personal experience. Figure 2.1 illustrates the performance of alternative mea-
sures of income over the period of the last US economic expansion, 2000–2007.
Within BEA’s existing accounts, an adjustment to personal income to exclude
taxes paid by households provides a measure of disposable personal income that
is the after tax income households have for consuming or saving. Then taking
that measure and putting it on a per capita basis begins reflects something closer
to the average household’s experience.
However, even per capita measures and other average measures may not
accurately reflect the economic position of most households. The disposable
income measures in Figure 2.1 reflect the growth in average income and
include the gains by those in higher income brackets. Adding information
on median income and income growth by deciles could provide a signifi-
cant amount of information on how the gains from economic growth are
distributed. The ability to decompose growth in the context of the national
accounts would be especially useful because the accounts provide a more com-
prehensive measure of income than captured by other household-survey-based
measures of income. Today it is not only important to understand how income
from current production is shared between capital and labor but also how
that income and associated purchasing power is shared among households
and why.
Expanded measures of the sources of growth
Although researchers ranging from Dennison and Madison to Jorgenson and

Kendrick have made significant progress in developing expanded measures of
human capital and other near market contributions to economic growth, little
progress has been made within official statistical agencies. Partially as a result,
a significant share of economic and productivity growth cannot be accounted
for by measured inputs, with nearly one-third of economic growth measured as
the residual ‘multi-factor productivity’.
Consistent with the guidance of several US National Academy of Sciences
reports on expanded national accounts, BEA’s work is focused on near-market
activities that can be priced using market prices or proxies thereof in a framework
consistent with the existing national accounts. This work includes measures of
R&D and other intangible assets, human capital, and natural resources.
The increasing role of innovation and intangibles in

measuring economic growth
One area in which BEA has made significant progress in developing new statistics
is in the area of measuring innovation and the impact of research and develop-
ment on the economy. Intangible investments in such activities as innovative
product design, brand promotion and cutting-edge drug therapies are critical
to competing in a global economy and have long been considered engines of
economic growth.
Since 2006, BEA, along with the National Science Foundation, has produced
a R&D satellite account to quantify the effects of treating R&D as an invest-
ment in GDP rather than as an expense. Results from the 2010 satellite account
suggest that R&D has accounted for a substantial share of the resurgence in US
growth in recent years. Between 1959 and 2007, R&D investment accounted
for 5.2 per cent of growth in real GDP. Between 1998 and 2007 its contribu-
tion to real growth rose to 6.8 per cent; in comparison, business investment in
commercial and other types of structures accounted for just over 1.3 per cent
of average annual growth in real GDP during the same time period (Lee and
Schmidt 2009). These results underscore the fact that much of the growth that
the US economy has experienced in the last decade is not being adequately cap-
tured by traditional economic measures. Understanding the role of intangible
investments on the economy will continue to be critical to accurately measuring
and encouraging a strong economy in the future.
Human capital
The very large magnitude of measures of human capital, dwarfing measures of
tangible assets, has made many reluctant to include such values in national
accounts. As pointed out by Abraham (2010) and Christian (2010), adjustment
of Jorgenson-Fraumeni estimates of human capital (based on the discounted

value of the returns to education) to exclude the returns to non-market pro-
duction, substitute a more conservative estimate of the returns to education,
and use of discount rate closer to market rates produces a still large but sig-
nificantly smaller estimate. This estimate is also more consistent with the
concepts and measures existing national accounts and the National Academies
recommendations.
In the past, BEA has also done work on prototype natural resource and
environmental satellite accounts.4 Unfortunately, the resource requirement in
producing such accounts is quite large and requires a significant degree of col-
laboration and data collection across disciplines ranging from economics and
epidemiology to geology and environmental sciences. Currently BEA has col-
laborative work underway in developing statistical infrastructure that needed to
improve the existing accounts as well as provide the building blocks for possible
natural resource and environmental accounts in the future.
These infrastructure projects include work with the US Energy Information
Administration on an expanded energy sector in the US input-output accounts;
work with US academics and epidemiologists on medical care spending and
health accounts; and preliminary work with the US National Oceanic and Atmo-
spheric Administration on extension of BEA’s regional impact models to better
estimate the effect of different policies on ocean economies.
More accurate measures of key services sectors for output, prices,

and productivity, especially for the healthcare sector
The Stiglitz-Sen-Fitoussi Commission Report (2009) and the National Academies

Beyond the Market report (Abraham and Mackie 2005) both call for improved
measures of government and private services and their contribution to economic
production and welfare. BEA is working on a number of projects to better mea-
sure services, especially for financial and healthcare services. The financial crisis
and mounting medical care costs underline the importance of countries better
measuring economic activity in these sectors to better inform policy. BEA’s work
on healthcare is highlighted below.
Rising healthcare costs

One of the most important challenges facing the United States and other coun-
tries is managing the fiscal implications of rising healthcare costs associated
with the aging baby boom generation. Unfortunately, we do not currently have
the statistical tools to accurately and consistently measure the drivers of med-
ical care spending and to assess the efficacy of different types of medical care
interventions. Developing such tools will require the collaborative efforts of
statisticians, economists, epidemiologists, and physicians.
Comparison of Price Indexes for Medical Care, 2001–2003

(compound annual growth rates)
5 5
4 4
3 3
2 2
1 1
0 0
Prescription Office Hospital Hospital Average provider-based Disease-based
drugs visits outpatient inpatient
Figure 2.2 Healthcare price indexes
One of most basic needs is accurate data on medical care inflation. Existing
price data measure costs by type of service and fail to adequately capture changes
in medical treatment that move from higher-cost to lower-cost treatments. Stud-
ies such as those by Cutler et al. (2001) and Berndt et al. (2001) have found that
conventional price statistics overstate medical care inflation by failing to fully
capture the switch from high-cost open heart surgery to drug therapy, or the
movement from expensive talk therapy to a combination of talk therapy and
drugs in psychotherapy.
To address this issue, BEA and BLS are working together to develop new price
indexes that capture such changes in medical care treatments. These require
data on the costs of treating diseases rather than the costs of specific medical
services. BEA’s work relies on public and private medical care insurance micro
data. As can be seen from Figure 2.2, such estimates produce a significantly
lower rate of medical care inflation.
BEA is also working with David Cutler and a team of other economists, physi-
cians, and epidemiologists to hopefully produce quality-adjusted price indexes
that take into account the efficacy of treatment.
In producing such disease-based statistics, BEA will be able to provide data on
costs of disease, cross-classified by disease and type of service, as well as by geog-
raphy. Such data will play a critical role in identifying the drivers of healthcare
costs, as well as sources of reductions in the costs of medical care. Such data
should play a key role in better projecting and managing health care costs.
Measures of financial and economic sustainability
The recent financial crisis has highlighted the need for up-to-date and trans-
parent information by type of instrument, currency, creditors, and debtors.
Emerging macro-prudential supervisors will need such information to identify

emerging risks. There may also be a role for collaborative work by statistical
agencies and regulators in developing alternative estimates of security values
through indirect methods.
The business and popular press seem to have decided that no one – other than
a few prescient souls, such as Noriel Rubini – anticipated the collapse in US hous-
ing and equity prices. What is probably more accurate is that few anticipated
the suddenness or depth of the drop in housing and stock markets. And few
anticipated the spread of the problems in US asset markets around the world.
There seemed to be an overall consensus at the time that US housing and stock
prices were ‘too high’ to be sustained and that consumers spent ‘too much’ and
saved ‘too little’. Since the mid-1990s, consumers depended on the appreciation
of their homes and portfolios to do their saving for them. However, most experts
seem to have thought that the correction would occur smoothly over time, with
a slowing in the rate of increase in house and equity prices below overall growth
and inflation, or through a normal (mild) cyclical correction.
Improved economic statistics could have helped in getting policy-makers,
analysts, and investors to focus on how out-of-line housing and equity prices
were, and how big an adjustment was required. Unfortunately, while much of
the information was available, it was not presented in a fashion that attracted
attention or affected policy in the way GDP, inflation, or the unemployment
rate affect monetary policy. While many attribute the collapse to the failure of
monetary policy to confront asset inflation and of regulatory policy to confront
excessive risk-taking, good statistics play a key role in forming public policy
by publicly highlighting the magnitude of emerging problems and aid in the
building of public consensus about the need for action.
Figure 2.3 highlights the rise in the value of US housing stock in relation to
personal income and GDP. Between 2000 and 2007, the value of housing stock
rose from 1.4 times personal income to 1.8 times personal income. However,
during the same time period there was an equal rise in household liabilities as
a share of personal income indicating households’ willingness to increase their
debt service burden encouraged partially by their expected future capital gains
on their housing investment.
Figure 2.4 shows the share of the increase in household’s net worth (saving)
came from saving out of current income as compared to capital gains on their
home or their investments. Between 2000 and 2007, households saw their net
worth rise from $42.0 trillion to $62.6 trillion.5 In response, households saw lit-
tle need to save out of current income and the personal saving rate dropped from
2.3 per cent to 0.6 per cent. There seemed to be little need for households to be
concerned about the future, because ‘saving’ through appreciation in their port-
folio was more than offsetting the drop in their saving out of current income,
and the ratio of net worth to disposable income was actually increasing. These
2.20 4.00
2.00
3.50
1.80
1.60 3.00
1.40 2.50
1.20
Index
Ratio
2.00
1.00
0.80 1.50
0.60 1.00
0.40
0.50
0.20
0.00 0.00
19 7
19 8
19 9
19 0
19 1
19 2
19 3
19 4
19 5
96
19 7
19 8
20 9
20 0
20 1
20 2
03
20 4
20 5
20 6
20 7
20 8
20 9
10
8
8
8
9
9
9
9
9
9
9
9
9
0
0
0
0
0
0
0
0
0
19
19
20
Ratio of the value of household real estate assets to personal income
Ratio of household total liabilities to personal income
Case-Shiller housing price index
Figure 2.3 Housing and personal income
Source: BEA NIPA data.
70000 1.60
60000 1.40
1.20
50000
Billions of dollars
Index, 80Q1 = 1
1.00
40000
0.80
30000
0.60
20000
0.40
10000 0.20
0 0.00
1980Q1
1981Q1
1982Q1
1983Q1
1984Q1
1985Q1
1986Q1
1987Q1
1988Q1
1989Q1
1990Q1
1991Q1
1992Q1
1993Q1
1994Q1
1995Q1
1996Q1
1997Q1
1998Q1
1999Q1
2000Q1
2001Q1
2002Q1
2003Q1
2004Q1
2005Q1
2006Q1
2007Q1
2008Q1
2009Q1
2010Q1
2011Q1
Net worth Net worth as % of DPI Personal savings rate
Figure 2.4 Household asset values and savings
Source: BEA NIPA data and Federal Reserve Board Flow of Funds data.
unsustainable trends – based as they were on the unsustainable rise in housing

and equity prices – not only had significant implications for the adequacy of
household retirement assets, but significant implications for the US and world
macro economy that would result from a rise in US saving out of current income
Index, 49=1
100
90
80
70
60
50
40
30
20
10
20
20
20
20
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
03
06
09
94
97
00
73
76
79
82
85
88
91
49
52
55
58
61
64
67
70
Nominal GDP S&P closing prices Domestic profits from current pdn
Figure 2.5 Growth in equity prices relative to GDP and NIPA profits
Source: BEA NIPA data, Standard and Poor’s data.
and the corresponding drop in the 70 per cent of US GDP accounted for by
consumer spending.
Figure 2.5 shows the rise in US equity prices relative to profits and GDP. For
most of the post-World War II era, the S&P price index rose at roughly the
same rate as GDP and corporate profits. This makes sense, because over time
growth in stock prices must come from growth in the economy, or a higher rate
of return to capital investments and growing share of GDP going to corporate
profits.6 However, after the mid-1990s, US stock prices – even after accounting
for the cyclical drop in profits in 2000 – soared relative to GDP and corporate
profits. Part of the rise was based on the perception that the United States had
entered a period of higher economic growth driven by technology. And as can
be seen from Figure 2.5, while there was a bump-up in economic growth, above
the slower growth experienced since the early 1970s, it was not sufficient to
explain the ‘irrational exuberance’ seen in financial market expectations, nor
was it particularly high in the context of long-term growth.
These figures – which are all based on available data – vividly illustrate how
far ‘out-of-line’ the prices were in housing and stock markets, and the extent to
which households’ saving rate out of current income was unsustainable. Unfor-
tunately, these charts, and associated ratios, were not produced, or highlighted,
by BEA – which produces the US GDP, personal income, and profits data –
and the Federal Reserve Board – which produces the US domestic financial and
household balance sheets – in the decentralized US statistical system. A paper by

Palumbo and Parker (2009), which examined how an integrated set of accounts
would have performed during the current crisis, found that the ‘facts’ related to
housing prices, household leveraging, and saving would have been highlighted
by such a set of accounts.
Where there was simply a gap was in macroeconomic data to warn of the
looming crisis in credit markets. The available data in Figure 2.6 shows data in
the financial sector only show a slightly higher average leverage ratio of 1.03
beginning in the late 1990s compared with an average ratio just over 1.0 over
the previous two decades, indicating the fact that the US data are too aggregate
to isolate the dramatic increase in leveraging that was taking place in mortgage
banks and other financial institutions, as well as in special purpose entities, that
were masked by decreasing leverage in other financial institutions.
What was also missed, because the data were too aggregate, were detailed data
on maturity to identify misalignment of assets and liabilities and detailed data
by type of instrument, such as how much of US international bond sales were of
collateralized sub-prime loans. Finally, there was a lack of data on the ownership
of these collateralized sub-prime loans. In the spring of 2008, as US financial
markets were experiencing significant difficulties, initially Europe viewed this
as a uniquely American problem but because of the interconnectedness of these
types of instruments it quickly grew into an international problem. Better data
[Ratio]
1.30
1.20
1.10
1.00
0.90
0.80
0.70
0.60
0.50
1970 1975 1980 1985 1990 1995 2000 2005 2010
Securities Brokers & Dealers Commercial Banks Finance Companies All financial business
Figure 2.6 Financial business sector leverage (total financial assets/total liabilities)
Source: BEA and Federal Reserve Board Flow of Funds data.

on the ownership of these assets might have contributed to earlier and better
coordinated international macroeconomic policy in the emerging international
recession.
Tangible investment
Since the national accounts were developed, there has been recognition of the
need to provide measures of sustainable growth and investment, specifically
net domestic product and net investment. Net domestic product (NDP) deducts
depreciation, or the amount required to replace the capital used up in produc-
tion, from GDP. NDP can also be thought of as the amount that can be consumed
without reducing the consumption of future generations. Similarly, net invest-
ment deducts depreciation from gross investment to measure the net addition to
the nation’s productive capital stock after deducting the amount of investment
necessary to replace the capital used up in production.
Such net production and investment measures are available in national
accounts, but more prominently featuring them among a suite of other sus-
tainable indicators would be useful to better understanding a guiding economic
growth. For example, during the last economic expansion real GDP grew at a 2.4
per cent annual rate, while net domestic product grew at a 2.2 per cent annual
rate, for a cumulative difference of $11.8 trillion over that period. As can be
seen from Figure 2.7, most gross investment has gone to replacing the capital
3000
2500
Billions, chained 2005 dollars
2000
1500
1000
500
0
2006 2007 2008 2009 2010
Real Gross Domestic Investment Real Net Domestic Investment
Figure 2.7 Sustainable investment
Source: BEA NIPA data.

used up in production with only an average of 28 per cent representing a net

addition to the US productive capital stock.
Notes
1. Gross National Product is the market value of goods and services produced by labor
and property supplied by US residents, regardless of where they are located. It was used
as the primary measure of US production prior to 1991, when it was replaced by gross
domestic product (GDP) which is the market value of goods and services produced
within the US.
2. For a discussion of these issues, see Deaton (2012).
3. For more on BEA’s work on household production see Landefeld and McCulla (2000)
and Landefeld et al. (2009).
4. For more information on BEA’s work on environmental satellite accounts see Landefeld
and Carson (1994) and Nordhaus (1999).
5. Federal Reserve Board, Flow of Funds data available at: www.federalreserve.gov
/econresdata/default.htm.
6. Since US rates of return to foreign investments are higher than foreign rates of return
to investment in the United States, a growing share of profits coming from overseas
investments can boost profits and stock prices, but this trend has not been significant
enough to explain the post-1995 run-up in stock prices.
References
Abraham, K. G. and C. Mackie (eds) (2005) Beyond the Market: Designing Nonmarket Accounts
for the United States (Washington, DC: National Academy Press).
Abraham, K. G. (2010) ‘Accounting for Investments in Formal Education’, unpublished
paper (College Park: University of Maryland).
Berndt, E. R., S. H. Busch and R. G. Frank (2001) ‘Treatment Price Indexes for Acute Phase
Major Depression’, in D. M. Cutler and E. R. Berndt (eds), Medical Care Output and
Productivity, Studies in Income and Wealth, vol. 62 (Chicago: University of Chicago
Press), pp. 463–505.
Christian, M. S. (2010) ‘Human Capital Accounting in the United States, 1994–2006’,
unpublished working paper (Madison: University of Wisconsin Center for Education
Research).
Cutler, D. M., M. McCullen, J. P. Newhouse and D. Remler (2001) ‘Pricing Heart Attack
Treatments’, in D. M. Cutler and E. R. Berndt (eds), Medical Care Output and Produc-
tivity, Studies in Income and Wealth, vol. 62 (Chicago: University of Chicago Press),
pp. 305–347.
Deaton, A. (2012) ‘The Financial Crisis and the Well-being of Americans’, Oxford Economic
Papers, vol. 64, no. 1, pp. 1–26.
Krueger, A. B., D. Kahneman, D. Schkade, N. Schwarz and A. A. Stone (2009) ‘National
Time Accounting: The Currency of Life’, in A. B. Krueger (ed.), Measuring the Subjective
Well-being of Nations: National Accounts of Time Use and Well-being (Chicago: NBER/
University of Chicago Press).
Lee, J. and A. G. Schmidt (2009) ‘Research and Development Satellite Account Update,
Estimates for 1959–2007’, Survey of Current Business, 90 (December), pp. 16–27.
Landefeld, J. S., B. M. Fraumeni and C. M. Vojtech (2009) ‘Accounting for Household

Production: A Prototype Satellite Account Using the American Time Use Survey’, Review
of Income and Wealth, vol. 52, no. 2, pp. 205–255.
Landefeld, J. S. and C. S. Carson (1994a) ‘Integrated Economic and Environmental Satellite
Accounts’, Survey of Current Business, 74 (April), pp. 33–49.
Landefeld, J. S. and C. S. Carson (1994b) ‘Accounting for Mineral Resources: Issues and
BEA’s Initial Estimates’, Survey of Current Business, 74 (April), pp. 50–72.
Landefeld, J. S. and S. H. McCulla (2000) ‘Accounting for Nonmarket Household Produc-
tion Within a National Accounts Framework’, Review of Income and Wealth, vol. 46,
no. 3, pp. 289–307.
Nordhaus, W. D. (1999) ‘The Future of Environmental and Augmented National Accounts,
An Overview’, Survey of Current Business, 79 (November), pp. 45–65.
Palumbo, M. G., and J. A. Parker (2009) ‘The Integrated Financial and Real System of
National Accounts for the United States: Does It Presage the Financial Crisis?’, American
Economic Review, vol. 99, no. 2, pp. 80–86.
Stiglitz, J. E., A. Sen and J.-P. Fitoussi (2009) Report by the Commission on the Measure-
ment of Economic Performance and Social Progress. available at the Commission on the
Measurement of Economic Performance and Social Progress website at: www.stiglitz-
sen-fitoussi.fr/en/index.htm.
3
Measuring Equitable and Sustainable
Wellbeing in Italy
Enrico Giovannini
The Italian National Statistical Office (Istat) and the University of Rome
‘Tor Vergata’, Italy
Tommaso Rondinella
The Italian National Statistical Office (Istat)
Introduction
Throughout history, various notions of wellbeing have been discussed depend-

ing on cultural influences and prevailing political regimes. In the 20th century,
wellbeing was often equated with economic welfare. After the Great Depres-
sion and World War II, national accounting, and in particular Gross Domestic
Product, came to be seen by many as the main way of measuring development.
Although several alternative measures of wellbeing and societal progress were
developed by researchers during the 1970s and the 1980s (for example, the ones
grouped under the so-called ‘social indicators movement’), it is only in the 1990s
that initiatives concerned with sustainable development and measuring human
development (such as the UNDP Human Development Index, and the UN’s Mil-
lennium Development Goals) have captured the attention of media and have
played a role in political debates. More recently, thanks to initiatives carried out
by (some) national and local political authorities, to the research on the mea-
surement of quality-of-life and happiness, and to initiatives undertaken by the
Organization for Economic Co-operation and Development (OECD) on mea-
suring and fostering societal progress, a new movement aiming at measuring
wellbeing is emerging.
In this context, measuring national wellbeing and societal progress in Italy
is one of the challenges that the Italian National Institute of Statistics (Istat)
is called to face. In recent years, Istat’s attention towards this issue has
taken the form of a number of activities aimed at strengthening the ability
of official statistics to measure specific dimensions of wellbeing. Such initia-
tives include objective and subjective measures of individual wellbeing, the
strengthening of environmental measures and accounts, and the adaptation
38
Measuring Equitable and Sustainable Wellbeing in Italy 39
of macroeconomic aggregates to provide distributional information and to

overcome GDP limitations in general.
During the past two years a further improvement towards the measurement of
wellbeing and societal progress has been made by publishing the annual report
‘Noi Italia’ which provides a brief presentation of the most relevant, according
to Istat, statistical information for describing the state of the country, and by
including in the Multipurpose Social Survey an internationally comparable sin-
gle question on overall life satisfaction as well as a question on how relevant
the different dimensions of wellbeing are for citizens. Finally, in 2011, Istat,
together with the National Council for Economics and Labor (CNEL), started
a national consultation to identify a shared set of indicators of the progress of
Italian society, therefore honoring the recommendations of the OECD and the
Stiglitz Commission (see Stiglitz et al. 2009).
This initiative needs to take into account at least two major challenges. First, a
sufficient and robust statistical production able to cover all relevant dimensions
of progress must be made available. Secondly, a democratic process is needed
to grant public legitimacy to the selected key indicators. Only if the latter con-
dition is satisfied, the set may (have the chance to) become a tool recognized
by different social components for assessing the state of the country and for
effectively supporting policy decisions towards progress.
This chapter initially describes the international context in which the Istat
initiative moves. Then, it presents the current statistical production by Istat
with respect to the measurement of wellbeing and societal progress assessing its
ability to answer to Stiglitz Commission recommendations. Finally, it illustrates
the Istat-CNEL initiative aiming at the identification of a set of key indicators
for measuring progress in Italy.
1 The context
Measuring the wellbeing of individuals and societies has been a concern of

statisticians for some time but over recent years the discussion on how to
measure progress is gathering momentum worldwide. It is being discussed by
policy-makers at all levels and has increasingly attracted media attention. For
many years, using GDP per capita as a proxy for the population’s wellbeing
made much sense. GDP per capita provides a good measure of a country’s
capacity to deal with the material needs of its residents. However, the relation-
ship between economic growth as measured by GDP and other dimensions of
societal progress is not straightforward. Academic researchers, civil society orga-
nizations, official statisticians, and international organizations have proposed
alternative measures of societal progress, which can be classified according to
various criteria.
40 E. Giovannini and T. Rondinella
Since 2001 the OECD has promoted several initiatives in an effort to raise
awareness about measuring and fostering the progress of societies, about the
need to develop new measures and to improve the use of the existing ones. A
consensus has not emerged yet on the best way to go, but the Istanbul Decla-
ration (OECD 2007) adopted in June 2007 by the European Commission, the
OECD, the Organization of the Islamic Conference, the United Nations, the
United Nation Development Programme, and the World Bank, highlighted an
international consensus on the need to ‘undertake the measurement of societal
progress in every country, going beyond conventional economic measures such
as GDP per capita’ and launched the Global Project on Measuring the Progress
of Societies as the worldwide reference point for those who wish to measure and
assess the progress of their societies.
The most influential work in this area has been the one by the Commis-
sion on the Measurement of Economic Performance and Social Progress (The
Stiglitz Commission; Stiglitz et al. 2009), set up by French President Nicolas
Sarkozy in January 2008. The Commission produced a final report in September
2009 calling for a ‘shift [of] emphasis from measuring economic production to
measuring people’s well-being’. The Commission’s aim has been to identify the
limits of GDP as an indicator of economic performance and societal progress; to
consider what additional information might be required for the production of
more relevant indicators of social progress; to assess the feasibility of alternative
measurement tools; and to discuss how to present the statistical information in
an appropriate way. The following recommendations arose from the report:
Recommendation 1: When evaluating material wellbeing, look at income and

consumption rather than production.
Recommendation 2: Emphasize the household perspective.
Recommendation 3: Consider income and consumption jointly with wealth.
Recommendation 4: Give more prominence to the distribution of income,
consumption and wealth.
Recommendation 5: Broaden income measures to non-market activities
Recommendation 6: Quality of life depends on people’s objective conditions
and capabilities. Steps should be taken to improve measures of people’s
health, education, personal activities, and environmental conditions. In par-
ticular, substantial effort should be devoted to developing and implementing
robust, reliable measures of social connections, political voice, and insecurity
that can be shown to predict life satisfaction.
Recommendation 7: Quality of life indicators in all the dimensions covered
should assess inequalities in a comprehensive way.
Recommendation 8: Surveys should be designed to assess the links between
various quality of life domains for each person, and this information should
be used when designing policies in various fields.
Recommendation 9: Statistical offices should provide the information needed

to aggregate across quality of life dimensions, allowing the construction of
different indexes.
Recommendation 10: Measures of both objective and subjective wellbeing
provide key information about people’s quality of life. Statistical offices
should incorporate questions to capture people’s life evaluations, hedonic
experiences and priorities in their own survey.
Recommendation 11: Sustainability assessment requires a well-identified dash-
board of indicators. The distinctive feature of the components of this
dashboard should be that they are interpretable as variations of some under-
lying ‘stocks’. A monetary index of sustainability has its place in such a
dashboard but, under the current state of the art, it should remain essentially
focused on economic aspects of sustainability.
Recommendation 12: The environmental aspects of sustainability deserve a
separate follow-up based on a well-chosen set of physical indicators. In partic-
ular there is a need for a clear indicator of our proximity to dangerous levels
of environmental damage (such as those associated with climate change or
the depletion of fishing stocks).
The issue was a theme for discussion even at the 2009 Pittsburgh Summit,
where the G20 Leaders asked for work on measurement methods that ‘better
take into account the social and environmental dimensions of economic devel-
opment’ as an inherent part of the implementation of the new Framework for
Strong, Sustainable and Balanced Growth (G20 2009). An important develop-
ment has also come with the European Commission communication GDP and
Beyond: Measuring Progress in a Changing World (European Commission 2009)
which fulfills the commitment made at the Beyond GDP conference, where was
clearly stated that ‘. . . It’s time to go beyond GDP’ (Barroso 2007). The Commu-
nication has moulded the ideas presented at the conference into a EU roadmap
for action committing itself to work in several areas to improve existing mea-
sures and to report on the implementation and outcomes of the listed actions
by 2012.
The key actions from the Commission Communication were:
• Complementing GDP with environmental and social indicators;

• Near real-time information for decision-making;
• More accurate reporting on distribution and inequalities;
• Developing a European Sustainable Development Scoreboard;
• Extending National Accounts to environmental and social issues.
It is against this background that within the European Statistical System

(ESS) the Sponsorship Group1 ‘Measuring Progress, well-being and sustainable
development’ has been established with the mandate of coordinating activities

on the issue and building on the recommendations from the abovementioned
Stiglitz Report and Commission Communication, taking also in consideration
the objectives of the European Commission Europe 2020 Strategy.2 The key
challenge within the ESS is to implement the recommendations arising from
these converging initiatives in order to deliver richer statistical information
and further enhance harmonization at the International level, in particular in
Europe.
Meanwhile, on 30 September 2010 the 96th conference of Directors General
of the National Statistical Institutes (DGINS) produced the ‘Sofia Memorandum’
recognizing the validity of the Stiglitz Commission recommendations, listing a
number of improvements that NSIs should adopt (such as to reconcile National
Accounts aggregates with household survey data, to give more attention to the
household perspective, to capture distributional aspects, to harmonize environ-
mental measures and improve timeliness of quality of life statistics) (DGINS
ESSC 2010).
The Human Development Report (HDR) and its Human Development Index
(HDI), from the United Nations Development Programme (UNDP) are an impor-
tant part of the measurement landscape; indeed the HDI is arguably the most
well known alternative to GDP (UNDP 2010). The index, which comprises mea-
sures of life expectancy, literacy and income, was devised explicitly as a rival to
GDP. The 2010 – 20th anniversary – report includes a chapter focussing on other
key dimensions of development which lie outside those within the HDI. There is
considerable correspondence between the UNDP’s and the Stiglitz-Sen-Fitoussi
and the OECD frameworks. The HDR, for instance, discusses the importance
of looking at inequality when assessing progress and also of both sustainability
and vulnerability. Empowerment – which is broadly defined as an increase in
people’s ability to bring about change – is also recognized as fundamental to
development and central to the capability approach (see below) which in turn
shaped the HDI.
Finally, in February 2010 the Franco-German Ministerial Council decided to
ask the French Conseil d’Analyse Économique (CAE) and the German Council
of Economic Experts (GCEE) to follow-up the Stiglitz Commission’s outcomes.
The CAE and GCEE have recently published a report on ‘Monitoring Economic
Performance, Quality of Life and Sustainability’ (CAE and GCEE 2011) which,
starting from the domains and indicators of the Stiglitz Commission, discusses
how the comprehensiveness and accuracy of an indicator set might be traded
off optimally with parsimony and cost to provide a reliable basis for regular,
timely, and digestible reporting on three key issues: economic performance,
quality of life, and sustainability. One of the conclusions of the Franco-German
report is that a single-indicator approach to measuring human progress is inher-
ently insufficient because the complexity of life and the demands on statistical
reporting are too diverse to allow a meaningful condensation of the current state
of affairs into a single comprehensive indicator. The report suggests that com-
prehensive statistical reporting requires a dashboard of indicators rich enough
to facilitate an informed discussion on the different aspects of human welfare,
without being overwhelmingly extensive.
2 A framework to measure equitable and sustainable wellbeing
The Stiglitz Commission’s recommendations for the measurement of progress

reduce the emphasis towards economic indicators in favor of a multi-
dimensional approach that considers social and environmental wellbeing as
being as important as economic wellbeing. Hall et al. (2009) have developed
a framework that aims at measuring societal progress defined as an increase in
‘equitable and sustainable wellbeing’. In particular, they propose to consider
final and intermediate goals, as well as the relationships between them and two
cross-cutting dimensions.
Italy is strongly committed to following the approach towards progress
defined at international level and a widespread consensus exists over the need
to widen the observation of key indicators for the evaluation of progress and cit-
izens’ wellbeing in our country. Yet, the definition of a shared set of key national
progress indicators needs to be broadly legitimated if it has to become a tool for
monitoring a shared vision of progress for the country.
A recent OECD working paper by Scrivens and Iasiello (2010) well identifies
the challenges that the definition of a set of societal progress indicators has
to face in order to be used and applied in decision-making processes. The first
challenge is in fact granting legitimacy to the tools, which means that ‘the
issues highlighted by the indicators are considered important’ and that selected
indicators ‘provide meaningful measures of those issues’. The general answer to
this is that ‘indicators must be developed with the participation of those who
will use – and learn from – them’.
The other two challenges identified by Scrivens and Iasiello lie in the choice
of the wider information system in which progress indicators are used (‘fit-
for-purpose information’) and in the ability to produce policy incentives. The
former implies the assessment of the opportunity to use either composite
indexes, a restricted set of headline indicators, or a broader and more compre-
hensive set, facing the trade-off between communication needs and avoiding
simplistic views of the issue. The latter is less straightforward since in order to
define a policy-relevant set of indicators it is not enough to ‘produce academ-
ically certified data and handling it to policy makers’ (Innes 1990: 8) and it is
politically very difficult to fix binding conditions for the fulfillment of objec-
tives implicitly or explicitly set by progress indicators. In order to guide policy
decisions, progress indicators need to be accompanied by sufficient incentives
Table 3.1 Goals of progress
Final goals Intermediate goals

Ecosystem condition: outcomes for Economy
the environment • national income
• land (geosphere) • national wealth
• fresh water, oceans, and seas
(hydrosphere)
• biodiversity (biosphere)
• air (atmosphere)
Human wellbeing: outcomes for people, Governance

individual aspects • human rights
• physical and mental health • civic and political engagement
• knowledge and understanding • security and violence
• work • institutional trust
• material wellbeing • access to services
• freedom and self-determination
Human wellbeing: outcomes for people, Culture

social aspects • cultural heritage
• social connections • arts and leisure
• social participation
• interpersonal trust
Links between the two sets of Goals Cross-cutting perspectives

Resource management, use, development • intra-generational aspects:
and protection equity/inequality
• resource extraction and • inter-generational aspects: sustain-
consumption ability/vulnerability/resilience
• pollution
• protection and conservation of
economic and environmental assets
Ecosystem services
• resources and processes provided
• impact of natural events
for policy-makers to enact change which can be fostered either by including

politicians in the selection process, identifying an independent watchdog or by
keeping public attention around the selected issues high. This is why Istat has
proposed to CNEL the launch of a joint initiative to measure societal progress
involving all components of society through a consultation process. Before
entering into such a presentation, it is useful to briefly describe the available offi-
cial statistics relative to the dimensions of wellbeing as selected by the Stiglitz
Commission Report and by the OECD. Therefore the next section, following
the current structure of the Report, presents the most relevant Istat statisti-
cal production concerning: a) economic performance; b) quality of life; and
c) environmental sustainability.
3 Istat current production
3.1 Measuring economic performance

GDP
Economic performance is classically represented by production growth. Istat
currently produces GDP estimates at both annual and quarterly frequency. How-
ever, this indicator can be expanded in order to include a number of neglected
components. Much has already been done in recent years within European offi-
cial statistics, namely the development of supply and use tables, the estimates of
the underground economy, the inclusion of all productive units’ activity within
national accounts (exhaustiveness), and the deflation of public services aggre-
gates embedding quality changes. With respect to the non-observed economy
(Calzaroni 2000), the pioneering work carried out by Istat has to be highlighted:
it has subsequently led to the definition of international standards as presented
in OECD guidelines (OECD 2002).
Another challenge is the estimation of public services’ values, which should
be based on actual production (output), rather than on the cost of production
(input). Hopefully it will be possible to overcome these limitations in a relatively
short time. Relying on time use survey data, a preliminary estimate (Baldassarini
and Romano 2006) of non-market activities of households has also been pro-
duced but, being still at an early stage, it cannot be included yet in the formation
of GDP.
According to the Stiglitz report, Net Domestic Product (NDP), that is, net
of depreciation, should be used rather than Gross Domestic Product (GDP). Of
course, Istat produces the net aggregate, even though the measurement of depre-
ciation presents some difficulties. For example, in recent years technological
innovation has grown quickly and that makes it harder to estimate the depreci-
ation rate of capital assets and depreciation of capital even without taking into
account environmental degradation.
One of the criticisms in considering GDP as the only indicator of welfare is

that the market prices used in its calculation do not reflect the use value of goods
and services, nor do they reflect the environmental damage that production and
consumption generate. Moreover, markets are not perfectly competitive and
consumers do not always possess the information essential for their choices. On
this front, the estimation of national accounts in real terms is guaranteed by the
treatment of quality change through the use of price deflators as well as a vari-
able weighting system for production and consumption price indexes which,
following international standards, take into account the changes in products’
quality.
Income
The Stiglitz-Sen-Fitoussi Commission suggests increasing attention in the obser-
vation of income whose key measure should be disposable income, that is, the
amount of current resources available to households for final use, consump-
tion and saving, less depreciation. Including all transactions that actually affect
consumption capacity of the beneficiaries means extending the concept of dis-
posable income. It should also include the use of goods and services freely
provided by the government and non-profit institutions, such as medical care,
hospital stays, housing allowance, crèches, and the like. This leads us to the def-
inition of the Adjusted Households’ Disposable Income aggregate that can bring a
sharper focus on the role of government in the process of income redistribution
and, more generally, the actual redistributive capacity of welfare systems. This is
an indicator neutral with respect to differences in coverage of social protection
systems across countries and to the public/private mix.
Istat already calculates this indicator, whose evolution (compared to GDP) is
shown in Figure 3.1. In addition, disposable income is sided by the Replacement
Rate for maintaining the standard of living guaranteed by the public pension
system, that is, the ratio between the first pension and the last pay from work.
Apart from being an indicator of extreme relevance to pension systems’ finan-
cial sustainability, this rate provides basic information on the adequacy of the
benefits it guarantees.
Consumption
The Istat Division for National Accounts produces quarterly and annually the
aggregate of Actual Household Final Consumption. It also includes, in the final con-
sumption of households, the expenses arising from private social institutions
and public social transfers in kind.
The extension of national accounts

National accounts are an essential tool for guiding decisions of economic
agents and for evaluating policy results. Their theoretical and conceptual
180
170
160
150
140
130
120
110
100
90
80
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
GDP Adjusted disposable income
Figure 3.1 Adjusted disposable income and GDP, Italy 1995–2009, 1995 = 100
Source: Istat.
framework was not specifically designed to analyze individual and societal

wellbeing, but to analyze the economic system of a country and some sheer
economic dimensions of wellbeing. In addition, the process of maintenance
of the system, which depends on international definition of the aggregates,
may lag behind the changes in economic and social reality, showing certain
stickiness in timely adaptation. Nevertheless, the Italian system of national
accounts has gradually opened to integrate information on the distribu-
tion and variability of stocks, such as wealth – in addition to the flows –
and on social and environmental phenomena – in addition to the tradi-
tional economic and financial dimensions. This is possible through satellite
accounts.
For some time the Istat Division for National Accounts has in fact worked
to develop new projects that meet the new perspectives of wellbeing analysis,
including: quarterly accounts by institutional sector; studies on globalization;
estimates for total and partial factor productivity; the first prototype estimates
of the stock of real assets for institutional sectors, in order to produce a complete
system of balance sheets; and national estimates of adjusted disposable income
and actual household consumption. In addition, an experimental work aimed
at building a satellite account for households includes data on unpaid work
and non-market activities as well as human and social capital. Within environ-
mental satellite accounts, pilot studies have been carried out on, among other
topics, physical and monetary environmental assets accounts, supply/use tables
in physical terms, and physical accounts for water.
3.2 Measuring quality of life

Over the past 20 years, social statistics in Italy have seen a continuous progress,
with the production of large amounts of data and indicators for the measure-
ment of quality of life. Istat has developed a strong set of information through
the development of the Multipurpose Survey (‘Indagine Multiscopo’) and other
household surveys, which today are widely used for designing and evaluating
social policies. Work, family life and relationships, the economic conditions of
families, leisure, political participation, social lifestyles, the relationship with
services, cultural enjoyment, security, and health conditions are investigated,
taking into account people’s actual behavior, as well as individual subjective
dimensions. Actually, the Multipurpose Survey is one of the most advanced
and ambitious social survey systems at international level: it is structured as an
annual survey named ‘Aspects of daily life’, integrated by five in-depth surveys
on family, health, leisure, security, and time-use carried out every five years and
a survey over travel and holidays carried out every three months.
The Stiglitz Report suggests considering measures that relate to subjective
experience in addition to objective measures of quality of life. The Multipurpose
Surveys integrate subjective measures in all different areas of investigation: from
health, family, work, leisure, and relationships with friends, to citizens’ security,
poverty, or economic conditions, just to name a few. In addition, in the latest
edition of the survey, Istat added an indicator on individuals’ perception of ‘life
as a whole’ on an internationally comparable scale from 0 to 10 and a question
about the degree of trust towards others.
Of course, the measurement of perceptions, opinions, and attitudes of people
do not replace events or behaviors measured in objective terms, but it manages
to capture information on issues and events of reality under investigation that
could not be otherwise obtained. The inclusion of subjective questions in official
surveys has not been a shift in focus from one level to another, but rather an
extension of the usual procedures of statistical collection and production of
social data, an enrichment of the spectrum of information that allows a better
reading of the phenomena, putting citizens at the center of official statistics.
In this way, Istat is able to produce most of the information needed for a
multidimensional measurement of wellbeing. Below we provide a description of
such a wealth of information according to the eight areas of wellbeing identified
by the Stiglitz Commission.
Material living standards

The measures of material wellbeing are fundamentally based on two surveys:
the EU-SILC survey (Eurostat 2010) and the national survey on household con-
sumption. The European survey gathers different measures of individual and
household income, as well as indicators of deprivation and social exclusion.
The survey includes a number of aspects of particular importance for a multidi-

mensional analysis of quality of life, such as participation in the labor market,
health, and education, characteristics of the dwelling and the area of human
habitation, expenditures on rent or mortgage, as well as the main economic
problems of families in a perspective of deprivation. In the Italian case, the size
of the sample is enlarged (Eurostat 2010) compared to the European standards,
so as to allow a regional disaggregation of data.
The national survey on household consumption covers the different models
of household spending and highlights the subjective assessment of families with
respect to changes in their purchasing behavior. The results of this survey are
used also for the calculation of relative and absolute poverty indicators. Thanks
to this survey, Italy is one of the few countries in the world that produces a
measure of absolute poverty. For both measures on income and consumption,
Istat produces median values, deciles and a Gini index, thus applying Stiglitz
Commission recommendations for enhanced attention towards distributional
issues.
Health
The Multipurpose Survey ‘Health conditions and use of health services’ investi-
gates aspects typically found in this kind of survey (acute and chronic diseases,
some types of disabilities, conditions of disability, use of drugs) along with
health-related quality-of-life indicators. These are tools used at the international
level that enable the identification of two synthetic indices of health status: the
Physical Component Summary measure (PCS) and the Mental Component Sum-
mary measure (MCS). Using the data from this five-year survey it is possible to
build one of the main indicators of quality of life – life expectancy free of dis-
ability – while an annual survey permits the estimate of life expectancy in good
health.
Education
A number of different sources provide a composite picture of human capital in
Italy. Apart from traditional statistics on formal education which are prepared
by the Ministry of Education, Istat annually tracks the number of early school
leavers and NEETs (Not in Employment, Education or Training) through the
Labor Force Survey, and provides information on school attendance, tuition
fees and English and computer science classes through the Multipurpose Survey.
Harmonized indicators on lifelong learning, one of the key issues of the Lisbon
strategy, will be produced by 2011 thanks to a European survey.
The skill levels of a population are an important piece of information for mea-
suring human capital; this is currently not included in Istat’s output, but should
soon be covered by the OECD ‘Programme for the International Assessment of
Adult Competencies’, which joins the already existing International Adult Lit-
eracy Survey (IALS) and the OECD survey PISA (Programme for International
Student Assessment) for 15-year-old students. Finally, the information frame-
work on human capital should be soon complemented by an education satellite
account on which Istat has already started to work.
Personal activities and work

The time use survey measures in detail the number of hours devoted to different
types of activities during the day. This is a key to addressing leisure, housework,
care activities, and work, allowing us to assess the weight they carry in citizens’
life. Following the Stiglitz Report, the survey may be strengthened by including
feelings experienced at certain times of the day. Using additional tools, an in-
depth analysis could be performed to take into account the quality of working
conditions. The Labor Force Survey measures non-standard employment, under-
employment, underutilization of human capital and the gender gap. Moreover,
the structural survey data on wages allows indicators to be developed on sectoral
and territorial pay differentials.
Labor indicators include also subjective measures such as a single question
on job satisfaction (which is also asked of housewives), while surveys on ‘crit-
ical aspects of women’s career paths’ and ‘family and social subjects’ collect
data on satisfaction in relation to several job characteristics (content, salary,
relationship with superiors, relationships with colleagues, and so on) and expec-
tations of individuals in relation to family and children, job opportunities,
career progression, stabilization, and change of employment.
Political voice and governance

A fundamental element to the exercise of the right to self-determination is the
presence of a flourishing and functioning democracy, universal suffrage, a free
press, a judicial system that guarantees the right to justice, and lively civil society
organizations. The measurement of these dimensions is particularly difficult
using the indicators traditionally produced by Istat. Today data are available on
the duration of civil and criminal trials, allowing the efficiency of the judicial
system to be evaluated, or on citizens’ fear of being victim of criminal acts or of
going out at night alone in the dark.
In terms of political participation since 1993 Istat has produced indicators of
invisible political participation (speaking and being informed about politics, by
frequency, channel information, reasons for not practicing) and visible political
participation (participation in demonstrations, rallies, or party meetings). The
availability of these indicators may allow the construction of composite indica-
tors of civic and political engagement. Nevertheless, it would be useful to have
indicators on press freedom similar to those published by non-governmental
agencies such as Freedomhouse.
Social connections and relationships

Interpersonal relationships influence quality of life in various ways. People with
stronger social relationships show a higher life satisfaction, better health, and
a higher probability of finding a job. Istat produces quite an important set of
information on this issue, including indicators about friendship, kinship, the
frequency with which people meet their relatives and friends, and informal sup-
port networks. Through the Multipurpose Survey it is possible to estimate the
fundamental characteristics of social networks of solidarity (care-givers and fam-
ily helpers), the type of help offered, any sharing of it with others, the number
of hours dedicated to it, all fundamental elements for helping to measure social
capital. This is also characterized by the relations one has with his/her neigh-
bors and by the perception that people have of whom they can really rely on
in case of need. Subjective indicators on the satisfaction of family and friends
relationships have been collected every year since 1993.
Insecurity
Indicators on safety and violence have been developed in recent years by Istat.
Next to security information from the objective point of view (crime suffered),
subjective opinions are sought in a specific survey that investigates such issues
as the sense of insecurity at home or out in the street at night, and the risk of
crime and social decay in the area where the family lives. Another important
survey from the standpoint of security and safety concerns women and has been
specifically conducted to detect physical and sexual violence, both inside and
outside the family. Economic security is then covered by objective and subjective
information related to living and working conditions.
3.3 Environmental statistics and accounts

From the standpoint of environmental statistics, Italy has seen a remarkable
growth in the availability of objective and subjective statistical information. The
information set on environmental issues has increased continuously. Indeed,
environmental statistics have been covering more and more issues, recently
thanks to the development of a survey on water and urban environment and
the implementation of agri-environmental indicators. Alongside pressure indi-
cators, which measure what is taken from the environment (withdrawal of
natural resources, energy consumption from different sources, occupation of
land) and released into it (emissions, pollution and waste production) and mea-
sures of impact, which describe the ultimate effects of environmental changes
and human activities, Istat produces response indicators that describe and quan-
tify the society’s efforts to solve environmental problems (protected areas, water
softening or green agricultural practices of firms). These performance indica-
tors are directly related to the implementation of policies or strategies that
have the reduction of impacts on nature as their ultimate goal. Moreover,
through surveys, Istat also covers the views that people have over environmental
conditions.
Istat has also been involved in the collection of indicators for the analysis of
sustainable development from the list of 140 indicators proposed by Eurostat.
In the past decade, a regular production on the side of environmental account-
ing has emerged in response to a request expressed at both national and
international level. Istat already includes in its production a number of envi-
ronmental accounts which will soon become compulsory at European level.
Time series for economy-wide material flows are published regularly covering
extraction of different materials and weights of imports and exports by product
along with Supply-Use and Input-Output tables and time series since 1980 for
the material flows indicators requested at international level: Domestic material
input (DMI); Domestic material consumption (DMC); Total material require-
ment (TMR); Total material consumption (TMC); and Physical Trade Balance
(PTB). National accounts also include aggregates of emissions associated with
production activities (NAMEA) available also at regional level for 10 pollutants.
In relation to the economic effort borne by the country to protect the natural
system, Istat records activities and financial transactions related to the envi-
ronment, such as costs for environmental protection, environmental taxes,
as well as the main economic aggregates for the field of eco-industries, for
which Istat is carrying out a reconnaissance of available sources. The concep-
tual framework and methodology of initiatives to protect the environment are
given by SERIEE (Système Européen de Rassemblement del’Information Economique
sur l’Environnement), which defines two separate satellite accounts: the satel-
lite account of the costs of ‘environmental protection’ (EPEA – Environmental
Protection Expenditure Account) and the satellite account of the costs of the
‘use and management of natural resources’ (RUMEA – Resource Use and Man-
agement Expenditure Account). These accounts can also be used to determine
unit costs to be applied for the quantification of interventions needed against
ecosystem degradation, an effort which began with the Italian contribution to
CICES, the Common International Classification of Ecosystem Services.
Further elements of assessment of natural resources are provided by the
accounts for monitoring fossil energy resources, in physical and monetary terms,
and the economic accounts for forestry (EAF). A feasibility study has been carried
out for the implementation of the European Framework for Integrated Environ-
mental and Economic Accounting for Forests (IEEAF). Istat is therefore aligned
with the most advanced standards in terms of environmental monitoring.
4 The launch of a national consultation
The wealth of data described so far will support an inter-institutional initiative

promoted by the National Council for Economics and Labor (CNEL) and Istat
for the identification of a set of indicators of societal progress and wellbeing

which was launched in December 2010. The initiative aims at carrying out a
process lasting 18 months which will involve all major representatives of Italian
civil society in the definition of the dimensions of progress and their related
indicators.
CNEL leadership of the process is a guarantee for its legitimacy: CNEL is in fact
a council established by the Italian Constitution and composed of representa-
tives of all major working categories, including representatives of entrepreneurs,
unions and of the third sector, for a total of more than 100 counselors repre-
senting different citizens’ groups. CNEL working groups and its assembly will
also serve as places for deliberation over controversial issues and trade-offs. The
initiative foresees the setting up of a Steering Committee jointly coordinated
by CNEL and Istat, including 20 participants from CNEL, non-governmental
organizations and public institutions, and the establishment of a scientific com-
mittee, hosted by Istat and composed of experts in the subject, as well as the
organization of a public consultation process.
The CNEL-ISTAT initiative adds Italy to the group of countries (France, Ger-
many, the United Kingdom, Canada, the United States, Australia, Ireland,
Luxembourg, Mexico, Switzerland, and the Netherlands) that decided to mea-
sure societal wellbeing by using a selected set of high-quality statistical indicators
identified through the joint efforts of union/management representatives and
civil society. As recommended by the OECD and by the Stiglitz Commission,
this approach will give the country a shared perspective on the evolution of
primary economic, social, and environmental dimensions. In particular, the
initiative’s objectives are to:
• develop a shared definition of progress in Italian society, by defining the most

relevant economic, social, and environmental dimensions;
• select a set of high-quality statistical indicators that are representative of the
different domains;
• communicate the results of this process, informing citizens of indicator val-
ues in the most thorough possible way. The set of indicators defined is in fact
intended for a broad public audience as well as for policy users.
The process will be divided into three phases fulfilling the objectives:
1. During the first phase the Steering Committee defined the dimensions to
be taken into account which were discussed by CNEL thematic working
groups and finally approved by CNEL assembly; the 12 selected domains
presented in November 2011 are: Environment, Health, Economic wellbe-
ing, Education and training, Work and life balance, Social relationships,
Security, Subjective wellbeing, Landscape and cultural heritage, Research and

innovation, Quality of services, Policy and institutions.
The Steering Committee is also in charge of defining the most appropriate
kind of information (fitness-for-purpose) and to evaluate the opportunity to
develop a tool explicitly designed for policy-making and in that case to define
the strategies for maximizing political commitment in following indications
emerging from progress measurement.
2. The second phase will be devoted to the analysis of the available indica-
tors which could be used to represent the various dimensions, stressing their
technical features and the opportunity of using them. It will be carried out
by the Scientific Commission composed of researchers and practitioners with
experience in the use and building of indicators. The number of indicators
in the final set should be small enough to facilitate their understanding by
non-experts. On this basis a first set of indicators will be identified in order to
be thoroughly discussed by CNEL workgroups and approved by the Assem-
bly. The Scientific Commission will treat separately the wellbeing indicators,
mainly focusing on outcomes, their distribution among social groups, and
the indicators of economic, social, and environmental sustainability.
3. The last phase will be dedicated to the drafting of a final report and the
development of various tools for dissemination/communication.
Apart from the inclusion of relevant stakeholders within the Steering Group
and into CNEL’s assembly, Italian citizens will be able to express their priori-
ties on the dimensions of wellbeing that are most relevant for individuals and
society through the dedicated website www.misuredelbenessere.it. The website
offers two major consultation tools: a short questionnaire and a blog. In the
questionnaire citizens are asked to give their opinion on the indicators to be
used for measuring wellbeing, on the 12 selected domains, and on the relation-
ships between then and the definition of public policies. This tool is similar to
the one proposed by ONS in the UK.
Through the blog a more in-depth discussion is possible, opening a national
debate among experts, practitioners, and anyone interested in the issue who can
in this way contribute to defining relevant dimensions to monitor progress and
wellbeing in Italy. During the spring of 2012, when a first set of indicators are to
be published, conferences in all Italian regions will be organized to debate the
measurement proposal, and auditions of relevant stakeholders will be organized
to finalize the tool.
A further consultation stream is represented by the inclusion of a specific
question in the Multipurpose Survey, ‘Aspects of daily life’, which is submit-
ted annually to 24,000 families (54,000 individuals). In the 2011 edition of the
Multipurpose Survey (carried out in spring 2011), Istat tried to assess the impor-
tance citizens attribute to different dimensions of wellbeing. Citizens have been
Table 3.2 What Italians think is important for their wellbeing
Aspects on which Italians will be asked to assess Average score Percentage

the importance for life and individual wellbeing of 10s
(0–10)
Be in good health 9.7 79.9

Guarantee the future of you children socially and 9.3 66.1
economically
Have decent work in which you are satisfied 9.2 59.5
Have an adequate income 9.1 56.0
Good relationships with friends and relatives 9.1 53.2
Feeling safe with respect to criminality 9.0 53.6
Be happy in love 9.0 56.3
Live in a society in which you can trust others 8.9 48.8
Good education 8.9 48.3
Present and future environmental conditions 8.9 48.8
Good governance 8.8 46.6
Good quality accessible services 8.7 43.9
Good quality and adequate time for leisure 8.5 37.4
Be able to influence local and national policies 7.8 30.6
Participation in community life through political and 7.1 18,7
associative structures
Source: Istat.
asked to assign a score from 0 to 10 according to the increasing ‘importance for

wellbeing and life’ people give to the aspects presented in Table 3.2. Such a tool
allows not only the adding of a robust assessment of people’s priorities to the
debate, but also to continuously monitor changes over time so to alert over the
need to modify the set of indicators. The high scores of all the selected dimen-
sions show clearly the multidimensionality of wellbeing. Almost no differences
emerge among age, gender, level of education and territory. Major variability
is shown by the percentage of people assigning a score of ten to the different
domains.
Politics will be formally excluded from the process. Nevertheless after each
phase of the initiative the Parliament will be informed about the major results
emerging from the consultation process.
As discussed in the second section, one of the critical aspects of this kind of
process is how to guarantee a solid legitimacy to the final output. In order to
produce a set of indicators which effectively represents people’s priorities for the
progress of the nation, such a set has to be widely discussed and recognized as
valid by most national actors: it is a one-shot opportunity for the creation of
a tool able to guide national policies. This will allow policy-makers and public
opinion to (hopefully) always refer to a shared national vision of progress, even
if it may change over time.
5 Conclusions
Italian official statistics appear fully capable of meeting the demand coming
from society and from international debate for the multidimensional measure-
ment of wellbeing and societal progress. Most of the issues raised by the Stiglitz
Report, in fact, are already satisfied by the information system developed by
Istat in terms of economic performance, objective and subjective quality of life,
and environmental measures. Italy is therefore ready to face the challenge of
building a shared measure of progress which may become a constant reference
for citizens, media and policy-makers.
To this end, Istat and CNEL launched a 18-month initiative for the measure-
ment of ‘equitable and sustainable wellbeing’. The process aims at producing a
set of indicators able to provide a shared vision of progress for Italy which will
be legitimated by a consultation of experts, relevant stakeholders and citizens
through dedicated meetings and workgroups, an online consultation and the
inclusion of a question in one of Istat’s major social surveys to identify people’s
priorities when dealing with individual and national wellbeing. The appoint-
ment is for the end of 2012, when the publication of the first joint Istat-CNEL
report is scheduled.
Notes
1. The Sponsorship Group was co-chaired by the Eurostat and FR-INSEE (National Statis-
tical Institute of France) Directors General, with the participation of 16 Member States
(Presidents/Directors General of NSIs: AT, BG, CH, DE, DK, ES, FR, IT, LU, NL, NO, PL,
SE, SI, SK, UK) as well as the OECD and UNECE.
2. The activities on the GDP and Beyond Communication and the Stiglitz Report in the
European Commission and in the European Statistical System (ESS) are also coordi-
nated by the Inter-departmental Co-ordination Group co-chaired by Eurostat and DG
Environment Directors General, with the participation of 11 Commission DGs and
three agencies.
References
Baldassarini, A. and M. C. Romano (2006) ‘Non-market Household Work in National

Accounts’, paper presented at the 18th Annual Meeting on Socio-Economics, Trier,
Germany, 30 June–2 July 2006.
Barroso, J. M. (2007) ‘José Manuel Barroso (President of the European Commission),
Beyond GDP International Initiative, 2007’, opening speech at the Beyond GDP
Conference, European Parliament, Brussels, 19–20 November, at: www.beyond-
gdp.eu/proceedings/bgdp_proceedings_intro_ses1.pdf
CAE and GCEE (2011) Monitoring Economic Performance, Quality of Life and Sustainability.
Joint Report as Requested by the Franco-German Ministerial Council, Conseil des ministres
franco-allemand, Fribourg-en-Brisgau, 10 December 2010.
Calzaroni, M. (2000) ‘The Exhaustiveness of Production Estimates: New Concepts and

Methodologies’, at: www.oecd.org/dataoecd/5/54/2464056.pdf
DGINS ESSC (2010) Sofia Memorandum. Measuring Progress, Well-being and Sustainable
Development, 96th DGINS Conference, Sofia, Bulgaria, 30 September, at: www.dgins-
sofia2010.eu
European Commission (2009) Communication from the Commission to the Council and
the European Parliament – GDP and Beyond: Measuring Progress in a Changing World
(COM/2009/0433 final) (Strasbourg: European Union)
Eurostat (2010) EU-SILC 2007 Comparative EU Final Quality Report Version 2 – June
2010, at: http://circa.europa.eu/Public/irc/dsis/eusilc/library?l=/quality_assessment/
comparative_quality_1/quality_2007_version/_EN_1.0_& a=d
G20 (2009) Leaders’ Statement: The Pittsburgh Summit, 24–25 September, at: www.pitts-
burghsummit.gov/mediacenter/129639.htm
Hall, J., E. Giovannini, A. Morrone and G. Ranuzzi (2009) ‘A Framework to Measure the
Progress of Societies’, Working Paper no. 34 (Paris: OECD Statistics Directorate).
Innes, J. (1990) Knowledge and Public Policy: the Search for Meaningful Indicators, 2nd edition
(New Brunswick and London: Transaction Publishers).
OECD (2002) Measuring the Non-observed Economy. A Handbook, at: www.oecd.org/dataoecd
/9/20/1963116.pdf
OECD (2007) Istanbul Declaration, Final declaration of the OECD 2nd World Forum
on ‘Measuring and Fostering the Progress of Societies’, Istanbul, 30 June, at:
www.oecd.org/dataoecd/14/46/38883774.pdf
Scrivens, K. and B. Iasiello (2010) ‘Indicators of “Societal Progress”: Lessons from Interna-
tional Experiences’, OECD Statistics Working Papers, 2010/4 (Paris: OECD Publishing),
online at: doi: 10.1787/5km4k7mq49jg-en
Stiglitz, J., A. Sen and J.-P. Fitoussi (2009) Report by the Commission on the Measurement of
Economic Performance and Social Progress, at: www.stiglitz-sen-fitoussi.fr
UNDP (2010) Human Development Report 2010, The Real Wealth of Nations: Pathways
to Human Development (New York: UNDP), at: http://hdr.undp.org/en/reports/global
/hdr2010/
Part II
Structural Imbalances, Financial
Frictions and Externalities
4
Sectoral Imbalances and
Long-run Crises∗
Domenico Delli Gatti
Università Cattolica, Italy
Mauro Gallegati
Università Politecnica delle Marche, Italy
Bruce C. Greenwald
Alberto Russo
Università Politecnica delle Marche, Italy
Joseph E. Stiglitz
There has been a widespread presumption that the current economic crisis is a
financial crisis, caused by the bursting of a credit bubble. Unjustified optimism
about asset prices and associated risks (primarily in housing but also in financial
industry equities and even in equities generally), accommodated by lax regula-
tion, careless private lending and loose monetary policies, led to unsustainable
levels of household and financial sector leverage. The inevitable collapse of the
underlying asset prices then caused widespread bankruptcies, foreclosures, and
impaired balance sheets among households, firms, and financial institutions.
Combined with consequent large increases in the incremental risks of lending
and investing, these balance sheet effects induced large declines in household
spending, firm output and investment, and bank lending.1
This perspective has contributed to an understanding of what should be done,
and the economy’s prospects. Consequently, US government efforts to revitalize
the economy focussed on pumping enormous sums into the banks. Now, more
than four years since the beginning of the recession, and more than three years
since the enactment of TARP, the economy is not back to health, and will likely
not return to full employment for years to come.
This chapter provides an explanation for the depth and duration of the cur-
rent downturn, in contrast to other postwar recessions.2 Our assessment of the
evidence suggests to us that the underlying cause of the current crisis lies in
61
62 D. Delli Gatti, M. Gallegati, B. C. Greenwald, A. Russo and J. E. Stiglitz
real economic imbalances, exacerbated by financial crises to which they gave

rise (or hidden, for a time, by bubbles which led to unsustainable levels of con-
sumption). We argue that the structural transformation of the global economy
from a manufacturing economy to a service sector economy necessitated, in
a sense, by the impressive growth in manufacturing productivity, is the real
underlying cause of current problems in the US and Europe. The consequent
adjustments have been exacerbated by globalization – the dramatic changes in
comparative advantages among countries of the world. But even had these not
occurred, the increase in productivity in manufacturing – at 6 per cent or more
per year, far outpacing growth in demand – would have meant a decline in
global manufacturing employment.3
While productivity gains are presumptively good for the economy, they are
not unambiguously so. There are losers and winners, and even if the winners
could compensate the losers they seldom do. But this chapter is not concerned
about the social consequences of these distributive impacts, but rather with their
macroeconomic consequences. Namely: In the absence of free migration, workers
can become trapped in the sector with rapidly increasing productivity. Especially if
the elasticity of demand for their output is limited, declining incomes in that sector
translate into declining demands for goods in other sectors, with ambiguous welfare
effects.4 Indeed, if there are efficiency wage effects in the other sector(s), productivity
shocks in one sector can give rise to unemployment in other sectors, and be welfare-
decreasing.
This chapter thus illuminates the potential role of structural factors in crises,
and highlights the interactions between cyclical factors and structural factors.
It distinguishes between ‘normal’ business cycles (typically generated by inven-
tory fluctuations or by central banks stepping too hard on the brakes in their
inflation-fighting zeal5 ) and extended downturns. In the former, informational
imperfections giving rise to financial constraints amplify economic shocks.6
The latter are attributed to the collapse of a major, often geographically iso-
lated, sector of the economy, though in these cases too financial constraints
amplify the consequences.7,8 The former are usually self-correcting and rel-
atively short-lived. The latter are long-lived and call for explicit government
intervention.
In the Great Depression, the collapsing sector was agriculture.9 Today, the sec-
tor in decline is manufacturing. Increase in manufacturing productivity at a rate
in excess of the rate of increase in demand leads naturally to steadily declining
manufacturing employment and income. Yet workers continue to be trapped
in the manufacturing economies both nationally, in countries like Japan, and
regionally, in US states like Michigan.
This chapter is organized into seven parts. The first describes in greater detail
the evidence that this crisis was not purely or fundamentally a financial one. The
second provides a motivation for the paper in terms of the Great Depression –
Sectoral Imbalances and Long-run Crises 63
arguing that just such a structural change, from agriculture to manufacturing,

was the real cause underlying the Great Depression. The third analyzes the
macroeconomics of a productivity increase in one sector in a simple model of
an economy with free migration among sectors. We structure the model in such
a way that not only is GDP increased, but everyone benefits. The fourth section
describes what happens when the conditions necessary to support migration no
longer hold and workers are trapped in the high-productivity but ‘dying’ sec-
tor – dying because it no longer generates jobs or because incomes are falling.
The fourth section shows that the problems are exacerbated if wages in the
urban (manufacturing) sector are rigid, for example because of efficiency wage
considerations. The sixth then analyzes three policy responses – fiscal policies,
wage policies, and migration policies. The seventh then expands the analysis
from a closed economy to an open economy. It shows that in this situation of
‘blocked’ migration, some individual nations/regions in a global economy can
(temporarily) improve their situation by, for instance, manipulating exchange
rates, with, however, adverse consequences for others. A brief conclusion
follows.
I An incomplete explanation for the crisis

and the Great Recession
Understanding the nature of the crisis – and why the economy has remained
weak – is essential not only for interpreting the events of the past few years, but
for ascertaining prospects and assessing policies going forward.
Initial assessments and policy interventions have focussed on the role of the
leveraged financial sector, and the subprime spark. Furthermore, recoveries from
financial crises, it is said, are slow, partially because bank and firm balance
sheets recover only slowly.10 Financial crises are typically associated with the
destruction of information, for example about who is creditworthy, as banks
are pushed into bankruptcy (Greenwald and Stiglitz 2003). In contrast to the
policies pushed by the IMF and the US Treasury in Indonesia and elsewhere
during the East Asia crisis, we congratulate ourselves in having preserved these
institutions, admittedly at some risk of moral hazard going forward. But there
is still the slow process of rebuilding bank balance sheets.
But, the failure of the strategies undertaken to end the crisis hints at the incom-
pleteness of the diagnosis that informed them. Since the crisis was deemed to
be a financial one, lawmakers and central bankers crafted policy on the assump-
tion that if the financial system were repaired, the economy would return to
health. This gave a sense of priorities to government. It provided justifica-
tion for the bank bailout and TARP; political leaders supporting this highly
unpopular bailout could feel virtuous because they put the wellbeing of the
economy over pursuing short-term political advantage. With a quick repair of
the financial system in the offing, only a short-term stimulus was required to
tide the economy over.
The weaknesses in the economy have, however, turned out to be more per-
sistent than this diagnosis would have suggested; the Fed has (as of early
2012) committed itself to leaving interest rates at near zero through the end
of 2014. Though there are still many concerns with the financial system (lack of
transparency, inadequate SME lending, weaknesses in many local and regional
banks), it is not apparent that finance is holding the economy back. (Of course,
in any crisis, the real and financial sectors are intertwined: any real crisis, lasting
long enough, will have consequences for the financial sector; and the subse-
quent weaknesses in the financial sector will have real consequences. That is
why ascertaining causality is always going to be difficult.11 )
If the financial sector were the cause of the economy’s current problems, it
should be reflected most strongly in investment. But business investment in
the United States, as a percentage of GDP, is not particularly low – certainly not
in a way that would be suggested if the availability of funds were the binding
constraint. Indeed, large businesses are reportedly awash with cash.12
Of course, investment in real estate is constrained – less than half the pre-crisis
level, but, with real estate prices down 30 to 40 per cent, that would presumably
be the case even with perfectly functioning financial markets. Indeed, the exces-
sive investment in real estate was really a symptom of a dysfunctional financial
market; one can hardly complain about a market that finally begins to show
some sense of ‘rationality’ after a prolonged period of excess.
There is another reason for suspecting that finance is not the major constraint
in the economy’s recovery – and therefore not the only explanation for its
weakness. If the financial sector were really broken, real lending rates would
presumably be very high. With inflation around 2 per cent, real T-bill rates are
now markedly negative, and even prime lending rates are very low (adjusted
for inflation, a little over 1 per cent).13 This is in marked contrast to the Great
Depression, in which prices were falling at 10 per cent a year, so real interest
rates were, in fact, very high. Indeed, the low (negative) real interest rates raise
questions about conventional monetary theory and policy, which focus on real
interest rates. Some economists have even suggested that the limitation of mon-
etary policy in restoring the economy is the ‘zero lower bound’, and some (such
as Krugman14 ) have made reference to a (Keynesian) ‘liquidity trap’. With real
interest rates already negative, it is hard to believe that high interest rates are
keeping the economy from recovering, and the data on investment cited earlier
is consistent with this perspective. It is hard to argue that with these low real
interest rates, finance is the critical constraint.15
Another aspect of the conventional wisdom is that if the economy is to
recover, households must deleverage. The fact that the process of deleveraging is
so slow – in the absence of some process of debt restructuring – adds pessimism

about the economy’s prospect. There is no doubt that the fact that nearly one
out of four Americans with a mortgage is underwater (an aggregate gap between
what is owed and the value of the underlying property estimated at some $700
billion16 ) causes anxiety and considerable misery among a substantial fraction
of American citizens, and one can argue that it was unconscionable – and politi-
cally unwise – to ignore this, especially as money was shoveled to the big banks.
Others hope that, somehow, even with the slow pace of deleveraging, the Amer-
ican consumer will return; they look carefully at the monthly sales data to see
some indication that that might be the case.
A closer look at the data, however, suggests that deleveraging, as desirable
as it might be from a welfare point of view, is not going to lead to signifi-
cant increases in consumption – and – certainly not be the basis of a strong
recovery.
Sustaining near-full employment in countries like the United States prior to
the crisis of 2008 seems to have depended on extraordinarily profligate con-
sumer behavior. Under ordinary circumstances, a near zero savings rate like that
of US households in the mid-2000s should have generated significant inflation-
ary pressure. But in spite of the absence of inflationary pressures, the low savings
rate was clearly unsustainable. High-income households, with roughly 40 per
cent of permanent income, typically save 15 per cent or more of their incomes.17
By themselves, they account for a 6 per cent savings rate out of total income:
15 per cent of 40 percent. An overall zero savings rate, therefore, required that
middle and lower-income households with 60 per cent of permanent income
dis-save at a rate of 6 per cent of total income per year. This, in turn, meant that
these households had to spend 110 percent of their incomes every year: –10 per
cent savings times 60 per cent for –6 percent of total income. The return to a
zero savings rate by middle and lower-income households in the wake of the
2008 financial crisis, which eliminated their continuing ability to borrow, led to
an increase in the overall savings rate to roughly 6 per cent.18 The consequent
decline in consumption demand appears to have been the proximate cause of
the recession. (Housing demand began to decline in early 2006.) Continued
prosperity pre-crisis seems to have depended on continued bubble-driven con-
sumption. If a return to ‘normal’ savings levels was inevitable, then the financial
crisis affected the timing rather than existence of a severe recession.19 By the
same token, ‘fixing’ the financial system, or even deleveraging, is not likely to
have a substantial, sustained effect on aggregate consumption, and therefore on
aggregate demand. The savings of the bottom 80 per cent are not likely again
to be negative, and those of the upper 20 per cent are not likely to fall much
below 15 per cent.20,21
Looking at this and other crises around the world throws further doubt on the
hypothesis that this is centrally a financial sector crisis. First, the severity of the
downturn has generally been unrelated across countries to the financial origins
of the crisis. The United States and the United Kingdom, both countries with
outsized financial sectors that failed spectacularly in the wake of widespread
financial misbehavior, suffered relatively less severe output declines than other
nations with sounder financial systems and no notable failures of financial insti-
tutions. Finland, Japan, Germany, Denmark, and Italy all suffered larger declines
in GDP than the United States and the United Kingdom. In other countries –
Spain, Ireland, Greece, and Portugal – financial difficulties and banking insol-
vencies appeared late in the crises following severe real economic contractions.
In these cases, financial crises appear to have been the consequence rather than
the cause of the recession, though weak financial systems are more likely to be
damaged by a ‘real’ economic downturn, and the consequent financial crisis
may serve to prolong the downturn.22
Moreover, for all the talk of a ‘great moderation’, the period since 1980
has been characterized by severe persistent crises outside the United States23
and, in many quarters, slow growth.24 Crises, in particular, have become
far more frequent and more severe. What is striking is that this was in a
period where economists claimed we knew more about economic manage-
ment, and more countries followed the precepts advocated by economists.
One explanation is that what was ‘learned’ was wrong, and the policy advice
was a move in the wrong direction. Another explanation (not necessarily
mutually exclusive) was that there were real changes which lead even well-
managed economies into crises, or at least increased the difficulties of economic
management.
While in some of the crises, bubble-like behavior played a relatively minor
role, Japan in the early 1990s did suffer from the collapse of a spectacular finan-
cial bubble and a badly impaired banking system. By 2000, however, these
problems were in the past, yet stagnant economic growth continued. The gen-
erally disappointing rate of recovery from the crisis in many countries besides
the US (with the important exception of the emerging markets) despite the
marked improvement in the financial sectors in these countries,25 suggests that
the Japanese experience may not be an isolated one.
The real changes in the economy that we believe are at the core of the problem
of economic adjustment are those caused by the enormous increase in pro-
ductivity in manufacturing. The issue has to be looked at, as we have noted,
from a global perspective. While the increase in manufacturing productivity in
excess of the increase in demand for manufactured goods will mean that global
manufacturing employment will decrease, there are, at the same time, shifting
comparative advantages.26 Countries that both have a large manufacturing sec-
tor and are losing their comparative advantage will face the largest structural
transformations – and thus may be the countries (ceteris paribus) most affected
by the crisis.
Not surprisingly, because the cause of this downturn is different from that of
other more recent recessions, it is plausible that the policy response might have
to be different. There should be structural policies to facilitate the movement
of labor that is ‘trapped’ in a dying sector, and that requires understanding the
economic forces that impede mobility. But even though structural policies are
part of the solution, traditional Keynesian policies play a role. The corrective
intervention that brought about the end of the Great Depression was World
War II – but not as it is generally interpreted. As we explain, the policies were
both Keynesian (a massive economic stimulus) and structural. Today, correcting
this situation will require a focused effort in managing the transition of workers
on a global basis out of manufacturing into services with an impact comparable
to that of World War II in moving workers off the farm. Our analysis, which
shows that well-designed Keynesian responses may be appropriate even when
there is a structural aspect to the underlying crisis, stands in marked contrast to
those who now claim that most of the remaining unemployment is structural –
there is a new ‘normal’ to which we must now accommodate ourselves – and
therefore policies designed to stimulate the economy may not only be useless,
they may be counterproductive.
II The Great Depression as a ‘model’ for the current downturn
The depth and duration of the present crisis is outside the normal range of post-
World War II experience.27 Not surprisingly, then, there is renewed interest in
the previous episode of a long downturn, the Great Depression (Temin 2010).28
Our thinking has also been greatly shaped by reflections on the Great Depres-
sion. Many attribute that economic downturn (like this one) to the financial
sector – a stock market bubble, supported by excessive margin, which, when
it broke, had large effects on balance sheets. Clearly, too, the banking failures
played an important role in the dynamics of the Great Depression. But was it not
possible that the stock market bubble itself was hiding underlying weaknesses
and more fundamental problems, just as the housing bubble did in the years
before the Great Recession? Indeed, the global banking crisis in 1931 appeared
relatively late in the global decline. A country like Canada, with no significant
banking issues, appears to have suffered as much in the Depression as countries
that experienced severe banking crises. The Depression, we believe, ultimately
arose from real factors rather than financial imbalances.29
In the case of the Great Depression, it is clear what the underlying real problem
was declining prices and incomes in the agricultural sector.30 In the United
States agricultural prices began to decline precipitously in August 1929, well
before the stock market crash in October of that year, and continued to fall
for years. It was to be another four years before banking failures reached their
zenith, with the enforced ‘bank holiday’ in 1933.31
It is easy to infer both the causes and consequences of the decline in agri-
cultural prices and incomes. Long-term increases in global farm productivity
coupled with increases in land under cultivation had, since the second half of
the 19th century, led to long-term increases in farm output above the rate of
increase of farm demand and, thus, secularly declining farm prices. Harvest and
demand fluctuations32 meant that this trend was far from uniform and there
were periods of high prices and farm prosperity. But by the late 1920s33 farm
prices were in steady decline, with consequent effects on farm income.
Gross farm income in current dollars fell from $17.9 billion in 1919 to $13.9
billion in 1929 to $6.4 billion in 1932 – a decline of more than 50 per cent in
three years – before recovering to $11.4 billion in 1937.34
Given the size of the agricultural sector (farm population represented 30 per
cent of the total in 1920) it is not surprising that a decline in that sector would
have macroeconomic consequences. These declines represent losses that are a
significant fraction of GDP. (The loss in gross farm income between 1929 and
1932 represented 13 per cent of 1932 GDP.)
One would have anticipated that declines of this magnitude would have led to
mass migration. And in the 1920s, it did, with farm population as a percentage
of the total falling from 29.9 to 24.8 per cent of the total from 1920 to 1929. But
remarkably, in the 1930s, migration was limited, with farm population falling,
proportionally, by just 1.4 percentage points in the 1930s (to 23.4 per cent by
1940.)35 The explanation is easy: the effective push and pull were far weaker
than the wage discrepancies would have suggested. Farmers had almost all their
capital invested in rural houses, farm equipment, land, local structures, and
related equipment. The sharp decline in the value of this capital coupled with
the simultaneous decline in farm income impaired the financial positions of
farmers and their local lending institutions. Thus, farmers could not afford to
migrate to the cities. Moreover, with high urban unemployment in the 1930s,
the migrant’s prospects were bleak.36 As a result of the inhibited migration, the
benefit that would have been enjoyed from higher farm productivity as a result
of reallocating labor was largely lost.
This loss of income to farmers itself led to weakening demand for urban goods,
leading to lower incomes and employment there, and this in turn led to declin-
ing demands for agricultural goods, in a downward vicious circle. In short, we
argue that the ‘shock’ to the economy which led to the low-level equilibrium
was a positive productivity shock in the agricultural sector – combined with
frictions that trapped workers in the rural sector.
There is one obvious objection to this analysis: The fall in agricultural prices
can be thought of as purely redistributive: farmers lose, those in the urban sector
gain.37 But there are several reasons that the net effect on aggregate demand in
the short run could be markedly negative. The resulting decline in rural demand
for industrial output would have outweighed any increase in urban demand as
long as the marginal propensity to increase consumption by urban households
was lower than the marginal propensity to reduce consumption by rural house-
holds. Several factors made such an outcome likely. First, budget-constrained
rural households would have been forced to reduce their consumption of indus-
trial goods sharply and immediately. Newly better-off industrial households
would have had the freedom to adjust more slowly to their higher real incomes.
Secondly, if they were uncertain about the permanence of the price changes
(that is, whether their permanent income had increased), they would not have
wanted to adjust their spending quickly. Thirdly, if the marginal propensity
to consume declines with income, the per dollar impact of declining incomes
among already relatively poor rural households would have been larger than the
impact of rising incomes for richer urban households.38 Fourthly, the failure of
rural financial institutions and the impairment of rural assets would have greatly
limited the ability of borrowing to offset the effects on demand of declining
rural incomes.39 The stimulating impact of lower lending constraints on largely
unconstrained urban households would have been far smaller. (Consumers’ abil-
ity to borrow for purposes of consumption smoothing was limited, far more so
than it was prior to this crisis.) Limitations in the ability of banks to lend to
farmers is the one aspect in which weaknesses in the financial sector contributed
to the underlying macroeconomic weaknesses – and to keeping workers trapped
in the dying sector; but for most farmers, it was probably not the operative con-
straint, for given the circumstances, even strong financial institutions would
have been reluctant to lend to farmers whose income was declining so rapidly.
In the formal model presented later, there is one further effect: the decline in
prices leads to a substitution away from manufactured goods. This strengthens
the adverse impacts on the urban sector. While the positive effect of the substi-
tution effect partially offsets the direct negative effect of the productivity shock
on rural incomes, so long as the system is stable it can only partially do so.
For all these reasons, the collapse of agriculture would have been expected to
lead to a parallel decline in overall urban industrial demand, and this is what
happened.40 At the very least, incomes in the industrial sector would have been
expected to decline markedly, as labor demand fell. If wages were flexible, there
would have been large redistributions from labor to capital in the urban sector,
with further adverse effects on aggregate demand. But if wages were at all rigid,
for instance, because of efficiency wage considerations, then unemployment
would increase markedly.
It is not surprising, given the magnitude of the negative shock from the
agriculture sector, that the limited increases in Federal spending during the
New Deal (partially offset by decreases in state and local spending) had limited
effects.41 Moreover, Federal spending was variable, declining markedly in 1937.
Our model also provides an interpretation of the recovery from the Great
Depression. The war provided the large Keynesian stimulus that the country
needed. But the war spending did far more than that. If the war merely rep-
resented a large Keynesian stimulus, then in the wake of the war, as Keynes
himself feared, the removal of that stimulus should have caused a return to
Depression-like conditions. In fact, the war appears to have been a uniquely, if
inadvertently, well-designed industrial policy.
Our earlier discussion identified several problems: in the wake of the collapse
of agricultural prices and incomes at the end of the 1920s, agricultural work-
ers no longer had sufficient income to finance a transition to manufacturing
employment. At the same time, the fall in agricultural incomes reduced demand
for manufactures so that manufacturing firms were also unable to finance this
transition, and the high urban unemployment rate made migration unattrac-
tive, even for those who might have had the resources to finance it. The war
solved these problems. It forcibly moved workers off the farms into both the
armed forces and war production plants. Together with the GI bill, the war pro-
vided the human and financial capital that made a transition from rural agrarian
employment to urban manufacturing employment possible. At the same time,
it created a transitional postwar demand for industrial workers through forced
wartime savings in the United States and the demands of reconstruction in
Europe and Japan.
Significantly, countries like Argentina that did not participate in the war
appear to have recovered from the Depression much more slowly. This is true
even though, because of flexible exchange rates, they may have weathered the
Depression better than the US. Without the war, the required restructuring
occurred only very slowly.
The model presented in the following sections tries to capture the spirit of
our analysis of the Great Depression. We begin with an analysis of what would
have happened as a result of agricultural productivity shock if there were perfect
labor mobility. We then extend the analysis to successfully more complicated
situations, where there is imperfect labor mobility and urban wage rigidities.
III The basic model, with free mobility
We consider a closed economy consisting of two sectors – agriculture and man-

ufacturing. Firms in each sector produce output using only labor as an input
with a constant-returns-to-scale technology. We denote variables pertaining to
the manufacturing sector with a superscript M and to the agricultural sector
with a superscript A.42 In manufacturing,
t = employment in period t,
M
mt = output per worker in period t,
wMt = manufacturing wage in period t,

ytM = manufacturing income in period t, = M M
t wt
We set the price of manufacturing output to unity as the numéraire good.

Thus, in competitive equilibrium
mt = w M
t (1)
so
ytM = M M
t mt (2)
In agriculture,
t = employment in period t,
A
at = output per worker in period t,
wAt = wage in period t, and
pt = price of agricultural output in period t.
For the agricultural sector, in equilibrium,
pt at = wA
t (3)
and
ytA = agricultural income in period t = A A A

t wt = pt t at . (4)
Under ‘normal’ conditions we will assume that agricultural wages are deter-
mined by a migration condition and manufacturing wages. Let
f · wM
t = the annualized value of the total cost of moving from agriculture to
manufacturing.
This represents the effect of transportation, job search, retraining, rehousing,

and acquisition of tools necessary for effective employment in manufacturing.
The assumption here, made largely for the sake of expositional simplicity, is that
these costs are proportional to manufacturing wages.43
The agricultural wage will then be determined, assuming that there is a
continuing migration from agriculture to manufacturing by the condition
that44
wA M
t = (1 − f )wt = (1 − f )mt (5)
and thus
pt = (1 − f )(mt /at ) (6)

Equations (1), (5) and (6) completely determine prices and wages in terms
of underlying productivities and the migration factor, f . It only remains to
determine the levels of labor employed in each sector.
The level of labor in agriculture is determined by the supply and demand for
agriculture output.45 Namely,
at A A AA
t = t d (pt , ytA ) + M
t d
AM
(pt, wM
t ) (7)
where d AA is the (per worker) demand function for agricultural output by agri-
cultural workers and d AM is the (per worker) demand function for agricultural
output by manufacturing workers,
ytA = (1 − f )mt A
t , (7a)
ytM = mt M
t , (7b)
and where, it will be recalled,
pt = (1 − f )(mt /at ). (7c)
In the ‘perfect markets’ equilibrium, with a fixed overall supply of labor,

employment levels adjust to ensure full employment. In this case
t = total labor supply in period t = A M

t + t (8)
For benchmark purposes we will assume that under ‘normal’ conditions full
employment characterizes the equilibrium with, in a more completely speci-
fied model, self-correcting fluctuations around this equilibrium in response to
random demand shocks (see Greenwald and Stiglitz 1993b).
Under these circumstances, what we are interested in is the impact of high
productivity growth in the agricultural sector on the overall economy. In order
to focus on this issue we will assume that manufacturing productivity, mt , and
the overall labor force, t , do not change. In the context of the assumption of no
manufacturing productivity growth, the natural way to think about changes in
agricultural productivity, at , is as changes in agricultural productivity growth rel-
ative to that in manufacturing since at appears in the equilibrium-determining
equations (7)–(7c) and (8) only as part of the ratio (mt /at ).
Substitution from (7a)–(7b) and (8) into equation (7) yields
at A A AA
t = t d [(1 − f )(mt /at ), (1 − f )mt A
t ]
+ M
t d
AM
[(1 − f )(mt /at ), mt (t − A
t )] (9)
This equation can be solved for A

as a function of at . The comparative statics
of this equation are captured by
dltA
lt
dat
= elasticity of agricultural employment with respect to
at agricultural productivity
= (εpA − 1)[1 − sA εyAA + (1 − sA )εyAM (A M
t /t ) − (d
AA
− d AM )]−1 (10)
where εpA is the overall price elasticity of the demand for agricultural products,46
εyAA is the income elasticity of agricultural demand for agricultural products by
those in the agricultural sector, εyAM is income elasticity of the manufacturing
demand for agricultural products, and
sA ≡ A
t d
AA
/(A
t d
AA
+ M
t d
AM
)
agricultural workers’ share in the demand for agricultural goods.

Since the demand for agricultural products should be inelastic (εpA less than
unity), the numerator in this expression will be negative. If demand for agri-
culture products grows less rapidly than income (εyAA < 1), and per capita food
consumption of agricultural workers is not too much greater than that of urban
workers (because urban workers’ incomes are typically much higher, the pre-
sumption is that d AA < d AM ) the denominator in the equation (10) will be
positive.
Thus, in the ‘normal’ case, higher agricultural productivity will shrink the size
of the agricultural labor force. However, overall output will grow rapidly for two
reasons. First, higher agricultural productivity by itself increases output. Indeed,
as long as agricultural output is not an inferior good, overall agricultural out-
put will grow despite the fact that agricultural employment is falling. Second,
because manufacturing wages are higher than agricultural wages, the shift of the
labor force to manufacturing increases overall output. In this virtuous cycle ris-
ing overall income should then lead to rising demand for manufacturing which
is sufficient to accommodate the workers moving into manufacturing.
Moreover, workers in both sectors are better off. Let V (wm , p) be the indirect
utility function of urban workers. While wm is unchanged, p (the agricultural
price) has fallen. Equation (5) ensures that those in the rural sector are also
better-off.
We can summarize the results of this section in
Theorem 1 In the normal case with free mobility a productivity improvement in

agriculture leads to lower rural prices and improves the wellbeing of all workers.47
However, constraints on mobility may dramatically alter this picture, and that
is what we examine next.
IV Mobility constraints
Workers moving from agriculture to manufacturing must usually be able to cover

the whole upfront cost F. Typically only a fraction of agricultural workers will
have the necessary savings.48 We will denote this fraction as γt in period t, where
γt will typically vary with fluctuations in agricultural prosperity. A sequence of
poor agricultural years will reduce the value of agricultural investments, espe-
cially local housing, and reduce the number of households able to finance a
move to manufacturing.
The amount of surplus labor in agriculture in any given year will depend
on the rate of productivity growth in agriculture. In the previous section we
showed that the higher productivity in agriculture, the lower the demand for
agricultural workers. This means that the greater the rate of productivity growth,
the higher the number of workers displaced. If the number of displaced workers
exceeds the number of agricultural workers able to finance the transition to
manufacturing, then the wage equalization conditions, equation (6), will no
longer apply. Formally, if at is the change in agricultural productivity between
t − 1 and t, then if
γ t A A
t < |dt /dat | · at (11)
the agriculture wage will be determined by market-clearing conditions in an iso-

lated agricultural labor market. The resulting equilibrium is self-reinforcing. The
limitation on migration increases the agricultural labor force and reduces agri-
cultural wages from what they would otherwise be. The reduction in agricultural
wages increases agricultural output and reduces agricultural prices and incomes
(under the likely circumstances derived below). The reduction in agricultural
incomes reduces the fraction of agricultural workers able to finance a transition
to manufacturing, which further increases the agricultural labor supply (from
what it would have been in the unconstrained migration equilibrium).49 In the
end migration may essentially evaporate and, in equilibrium, further increases
in agricultural productivity will lead to further immiseration of the now trapped
population of agricultural workers.
Lower agricultural incomes, in turn, undermine demand for manufactured
goods and may lead to overall economic stagnation. Thus, if productivity
growth in agriculture is high enough and/or agricultural workers’ ability to
finance migration is impaired enough (γt falls enough), a transition to a
long-lived inferior equilibrium may replace the virtuous cycle of the previous
section.
To model this equilibrium, we will no longer assume that agricultural wages
are determined by the migration condition of equation (6). Instead we will
assume that A t is fixed and wages in agriculture are set at a level necessary to
employ this labor force. We will continue to assume that there are constant-
returns-to-scale in production in agriculture and that labor is the only input.
In the ‘normal’ equilibrium, as surplus labor migrated from agriculture to
manufacturing, the rise in average wages generated sufficient income to absorb
the greater level of manufacturing output. In the no-mobility equilibrium, there
is no longer any need to absorb surplus agricultural labor into the manufacturing
sector. But the steady decline in agriculture incomes may, under circumstances
outlined below, actually reduce the overall demand for manufacturing in the
short run and, as the decline in agricultural incomes continues, in the longer run
as well. The low income in agriculture not only weakens demand in the urban
sector, but the weaker urban economy has repercussions back to the agricultural
sector.
In this section, we assume that wages in the urban sector adjust to main-
tain full employment. In the next, we assume that wage rigidities lead to
unemployment in the urban sector.
We generalize the model slightly to assume that different workers in the urban
sector have different reservation wages, so that while potential labor supply in
t , actual employment is E(wt ) ≤ t , and that output in the
the urban sector is M M M
urban sector is H(E). We focus on the situation where E = E(wM t ), that is, there
is full employment in the urban sector, in the sense that everyone who wants a
job at the going wage (wm− t ) can get one. Inverting, we obtain
−1
wm
t = E (Et ) ≡ ξ(Et )
Equilibrium agricultural prices are now determined by50
at A A AA
t = t d (pt , pt at ) + Et d AM (pt, ξ(Et )) (12a)
where A
t is now fixed by a no-migration constraint. Equilibrium requires
demand to equal supply:
H(Et ) = A
t d
AM
(pt , pt at ) + Et d MM (pt , ξ(Et )) + It (12b)
where d AM (d MM ) is demand from those in the agricultural (manufacturing) sec-

tor for manufactured goods, and I is the level of investment (assumed to be
industrial goods).
(12a)–(12b) can be solved simultaneously for {Et , pt } for any given value of at .
The curve AA gives the equilibrium combinations of urban employment (wages)
and rural prices which clear the agricultural market: for any given at , the higher
urban employment (wages), the higher rural prices.
The curve MM in Figure 4.1 gives the equilibrium combinations of urban
employment (wages) and rural prices which clear the urban (manufacturing)
market. Higher urban employment (wages) increases supply (assuming that all
workers who want to work are employed) and demand, but stability implies
A
p*
A
E
E*
Figure 4.1 Equilibrium combinations of urban employment (wages) and rural prices
that the effect on supply exceeds the effect on demand (otherwise, as labor is
hired to meet a shortfall of production and wages get bid up, demand would
increase more than supply, and the shortfall would increase).51 At the same time,
higher rural prices increase demand, both because farmers’ income is higher, and
because consumers substitute urban goods for rural goods. Hence, as p increases,
equilibrium urban employment and wages increase.
Since both curves are upward sloping, there can be more than one equilibrium
as illustrated in the figure – a high urban employment (wage)/higher rural price
equilibrium, and a low urban employment (wage)/low rural price equilibrium.
Clearly, from the perspective of rural workers the former is preferable to the
latter, and, under plausible conditions, so is it preferable for the urban workers.52
For the rest of this chapter, we focus on the case where there is a unique, stable
equilibrium. It can be shown that a ‘natural’ stability condition requires the AA
curve to be flatter than the MM curve. Thus we focus our analysis on the high
price and employment equilibrium in Figure 4.1, and how it shifts as agricultural
productivity increases.
The effect of rising agricultural productivity is to sharply decrease agricultural
incomes. The AA curve shifts down (at any level of urban employment and
income, the market-clearing level of agricultural prices is lower). At the same
time, the MM curve shifts to the right: at any given rural price, rural incomes are
higher, and so will the demand for urban goods. Under plausible conditions,
however, the new equilibrium entails both lower urban employment (wages)
and rural prices, and both urban and rural workers are worse-off. With restricted
mobility, the productivity improvement in the rural sector leads to universal
immiseration53 (See Figure 4.2).
M
M'
A
p* A'
p'
A
A'
M
M'
E
E' E*
Figure 4.2 The effects of an increase in agricultural productivity
We summarize the results of this section in
Theorem 2 Normally, with restricted mobility, an increase in agricultural produc-

tivity leads to both lower urban wages and lower rural prices, and both rural and urban
workers are worse-off.
V Efficiency wages and induced unemployment
In the previous section, we assumed only one market imperfection – there

was a constraint on labor mobility. Markets still cleared in both the urban
and rural sectors. If, however, wages are not fully flexible, for example,
because of efficiency wages, then there will be involuntary unemployment (as
opposed to the voluntary decision not to supply labor, discussed in the previous
section.)
For simplicity, consider first the case where wages are totally rigid,54 but there
is zero labor supply elasticity. We then replace equations (12) with equations (13)
at A A AA
t = t d (pt , ytA ) + Et d AM (pt , w∗ ) (13a)
∗
H (Et ) = A
t d AM
(pt , pt at ) + Et d MM
(pt , w ) + It (13b)
(13a) and (13b) can be solved simultaneously for {Et , pt }. The analysis parallels
that of the previous section, but now, the welfare costs of the productiv-
ity increase in agriculture are greater, because it gives rise to induced urban
unemployment.
It is possible to show (as before)
Theorem 3 ‘Normally’, with stability, equilibrium employment and prices will

decrease as agriculture productivity increases.
The equilibrium can be illustrated in a figure identical to Figure 4.2. Stability

again requires that the slope of the MM curve, giving the pair of {E, p} for which
the demand for manufactured goods equals the supply (eq. 13b is satisfied) be
greater than the slope of the AA curve (the set of {p, E} for which the agricultural
market clears). The footnote provides sufficient conditions for this condition to
be satisfied in terms of the underlying demand and supply functions.55
A special parameterization. The results can be seen especially easily in the
following special parameterization. Assume
Et mt = I + Ec M (w∗ − pt d AM ) + A A A
t c (yt − pt d
AA
)
where individuals first decide on how much food to eat (food is a necessity,
which is totally price inelastic) and then workers in the urban sector spend
a fraction of the residual, cM on urban goods, and similarly for agricultural
workers. Assume, further, that rural workers are limited in the amount that they
can spend on manufactured goods to what they receive from urban workers, and
that they do in fact spend that amount (a kind of balance-of-trade condition).56
Hence
Epd AM = A A
t c [pa − pd
AA
]
Substituting, we obtain
Et mt = I + Et c M (w∗t M − pt d AM ) + Epd AM
or
E = I/{mt − [c M (w∗t M − pt d AM ) + pd PM ]},
so
dE/dp = Ed AM (1 − c M )/{mt − c M (w∗ − pt d AM ) − pd AM }

= Ed AM (1 − c M )/{(mt − w∗ ) + (1 − c M )(w∗ − pt d AM )}
Assume, for instance, that those in the urban sector consume all of their ‘excess’
income, so c M = 1. Then,
dE/dp = 0.
The MM line is perfectly vertical at
E = I/(mt − w∗t m )
VI Policy
Government interventions can improve matters. In this section, we explore

three possible interventions: fiscal policy, wage policies, and policies to facilitate
migration.
1 Impact of Keynesian stimulus

Assume now that we introduce government spending, replacing equation (13b)
with
H(Et ) = A
t D
AM
(pt , pt at ) + Et DMM (pt , w∗ ) + It + G. (14)
It immediately follows that even though the origin of the crisis was structural, a
Keynesian (fiscal) stimulus (an increase in G) increases both employment and rural
prices. As G increases, MM shifts to the right, as in Figure 4.3.
Theorem 4 Under the stability condition,57 an increase in government expenditure

increases urban employment and raises agricultural prices and incomes.58
2 The effects of wage adjustments

We have assumed that urban wages are fixed, for example, as a result of efficiency
wages. Different efficiency wage theories provide different predictions for what
might happen to real wages denominated in manufactured goods when food
prices fall.59 Some versions (such as labor turnover theories) are consistent with
no changes in real wages, denominated in manufactured goods. Others, such
M
M'
A M''
p*
A'
p'
A
A'
M
M'
M''
E
E' E*
Figure 4.3 Impact of Keynesian stimulus: an increase of G shifts the MM curve from M
M to M M and increases both employment and rural prices
p
M'
M
A
p* A'
p'
A
A'
M'
M
E
E' E*
Figure 4.4 The effects of downward wage adjustments
as the Shapiro-Stiglitz incentives-based model, predict a decline in real wages,

denominated in manufactured goods.
By the same token, many economists have blamed unemployment on exces-
sively high wages. But in our model – as almost surely in the real world in a
severe economic downturn such as the Great Recession or Great Depression –
the effect on demand of falling wages worsens the downturn. A decrease in
urban real wages shifts down the AA curve (demand for food decreases with
the decrease in urban workers’ incomes), which lowers employment and prices
further. Such decreases did, of course, occur in the Great Depression. But now,
the second-round effect in the urban sector reinforces this effect. At any level
of employment, for demand for urban goods to equal supply the price of urban
goods has to fall, that is, the price of rural goods has to rise, implying that the
MM curve shifts up, so that in equilibrium urban employment is lowered even
more (and agricultural goods prices are further lowered).60
Theorem 5 Under the stability condition, a decrease in urban real product wages
increases urban unemployment and lowers agricultural prices and incomes.
3 Migration subsidies
Assume the government could facilitate migration, that is, reduce A t . What hap-
pens to the equilibrium? We focus our discussion here on the unemployment
equilibrium of section IV. From equation (13), it is clear that the AA curve shifts
upwards (at each E, the equilibrium agricultural price increases, since net sup-
ply – output minus consumption by agricultural workers – decreases). By itself,
p
M'
M
A'
p'
A
p*
A'
M'
M
E
E* E'
Figure 4.5 The effects of migration subsidy
this leads to more urban employment. The MM curve, on the other hand, shifts
to the left – at each p, demand for urban goods decreases, so employment falls.
The net effect would appear to be ambiguous, but under ‘normal’ conditions,
employment increases – given the price inelasticity, the decreased supply has a
larger impact on agricultural prices and income (and therefore urban employ-
ment) than the fact that, at any price, there are fewer rural workers demanding
urban goods.
VII Global interactions
In a closed economy agricultural demand is limited to local demand. In a global

economy individual local economies can sustain employment in their agri-
cultural sector by exporting into a global market. In particular by devaluing
its currency, any small country can increase output (although not the inter-
national price at which it sells that output). The advantages of doing so are
threefold. First, the local terms of trade between manufactures (urban goods)
and agricultural products (the rural sector) can be improved. To the extent
that there are manufactured goods locally produced and consumed (includ-
ing, and especially, non-traded manufactured goods), a devaluation of the local
currency which raises the local price of agricultural output will increase the
local currency value of rural income and have a positive effect on rural demand
for manufacturers. Second, if agricultural debts are denominated in local cur-
rency, the devaluation will reduce the level of these debts relative to agricultural
income and perhaps relax the agricultural mobility constraint. Third, the rise in
local currency agricultural incomes may by itself relax the mobility constraint,
specifically if migration costs are set in local currency terms.
These factors apply (mutatis mutandi) with special force today in the transition
from manufactures which are traded globally and are experiencing high cur-
rent rates of productivity growth to services which are overwhelmingly locally
produced and consumed.
The problem with attempting to capture a greater share of international
demand is that not all countries can succeed in doing so. Surpluses and deficits
across all countries must sum to zero. What matters is relative prices – exchange
rates – and a country’s trading partners can largely undo whatever a country
does to lower its exchange rate. It is only if some countries cannot devalue their
currencies either because they are reserve currencies against which all other cur-
rencies are measured (for example, the US) or because they are part of a common
currency area (for example, Greece, Spain, Portugal, Italy, and the other Euro
countries), that they can be taken advantage of in a global context.
In this connection, it is significant that relatively small countries like Australia
and Argentina, which devalued their currencies early in the Depression, were
able successfully to limit the short-run impact of increasing global agricultural
productivity. The adverse impact on reserve currency countries or those that
adhered strictly to the gold standard like the UK and the US was much more
severe.
In the present crisis reserve currency countries like the US with very limited
manufacturing employment (relative to total employment) have still suffered
from prolonged deflationary pressure. Other countries, such as Germany, Korea,
and China, which have large manufacturing sectors but have been able to limit
the appreciation of their currencies, have fared unexpectedly well.
VIII Conclusion
We argue here that the relatively rare long-lived deflationary episodes arise from
significant real structural economic dislocations. Data presented here show that
the Depression in the United States was related to the decline of agriculture
arising from the large increase in agricultural productivity. As farm income –
the income of almost 30 per cent of the population at that time – fell in the
1920s, workers left the sector. But then in current dollars it fell by more than half
from 1929 to 1932. Despite the relative decline in farm returns, the agricultural
population, which fell from 30 per cent to 24.8 per cent of the US population
over the course of the 1920s, actually rose slightly in 1930s (although it did fall as
a percentage of the total population). Workers were ‘trapped’ in the rural sector,
reinforcing the income declines, the effects of which were then felt strongly in
the urban sector. The limited increases in government spending could not offset
the impact of declining agricultural incomes. With trapped labor, productivity

growth, rather than raising standards of living for all, could lower welfare.
World War II and the policies that were pursued in its aftermath had a major
structural impact on the US economy. These policies succeeded in helping the
economy make the transformation that it was unable to do on its own. In the
1940s, with the impetus of war-driven dislocations, the agricultural population
fell sharply to 15.2 per cent, with the vast majority of the shift occurring between
1940 and 1945.61 Today the challenge is to shift workers out of manufacturing
into services. This will have to be done globally. But globalization has made
the global economy especially sensitive to shifting comparative advantages. In
the shift from agriculture to manufacturing, the US continued to be a strong
producer of agricultural goods, even as employment diminished, simply because
there was an abundance of one of the critical factors, productive land. That may
happen in some manufacturing sectors, where US production continues apace,
but with a fraction of the employment. But the dislocations may be larger, since
in some sectors, there may be little reason that production should continue at
all inside the US. Employment will then decrease both because of the increase in
productivity and because of the loss of comparative advantage. Some countries
may attempt to invest to restore their dynamic comparative advantage; but from
a global perspective, this cannot be a ‘solution’ to the problem of declining
global manufacturing employment. Indeed, to some extent, it will exacerbate
the problems, because the struggle to achieve dynamic comparative advantage
will normally be through further increases in productivity – meaning even less
manufacturing employment.
If global growth continues to remain as robust as it has been at its peak, at some
4 per cent,62 global growth in the demand for manufactured goods may fall only
slightly short of the amazing growth in manufacturing productivity (or may
even slightly exceed it). In that case, the fight will be over where the relatively
fixed set of manufacturing jobs are located. For the world as a whole, the share
of employment in manufacturing will almost surely decline significantly, and in
the longer run, almost surely, productivity growth will outpace demand growth.
Thus, in the long run, the countries that will face the most serious adjust-
ment problems are those that, in the short run, remain committed to strong
manufacturing sectors supported by exports and ultimately low exchange rates.
For most countries, continuing commitments to jobs in manufacturing (just as
many countries remained committed to agricultural employment throughout
the 1930s) in the face of adverse productivity and demand conditions is a recipe
for problems, if not now, then later.
Poorly managed transitions lead to financial sector problems – or crises –
which, in turn, impede the transition. The underlying weaknesses that our
analysis has uncovered can, temporarily, be covered up by a credit bubble, the
breaking of which can (as in the current crisis) make the transition all the more
difficult, compounding the challenges of transition with excess capacity in, say,
real estate.
As in the case of the transition out of agriculture, this transition will require
major, carefully conceived government intervention. Section V showed that
even though the underlying problem may be considered structural, austerity, of
the kind currently being contemplated in Europe and America, leads to higher
unemployment. Keynesian policies can help address the unemployment prob-
lem, even when the origins are linked to structural transformation. Well-targeted
industrial policies can have an even more positive effect. There is one more
aspect of the current transition which suggests that austerity is likely to have
an even more adverse effect than our analysis suggested. This is a transition
from manufacturing to services, and among the services most in demand, and
likely to provide employment for those transitioning out of manufacturing, are
health and education, sectors in which government finance has, for good rea-
sons, traditionally been important. Austerity will especially weaken demand in
these sectors, making the plight of those trapped in the declining sectors all the
worse.
This chapter has not only argued that underlying current economic difficul-
ties are a ‘real’ shock to the economy, but also explained why the single-minded
focus on the financial sector is misguided. The aftermath of a real estate bubble,
the overhang of debt and low real estate prices caused by financial sector mis-
management does make the transition all the more difficult. Moreover, failing
to fix the financial and real estate sectors almost surely will impede the transi-
tion. In the end, though, ‘fixing’ the financial sector does not itself ‘solve’ the
transition problem. That requires, first, recognizing that this is the underlying
problem facing the economy today, and second, designing a package of fiscal,
structural, and financial responses that will facilitate a more rapid transition.
Notes
* This is a revised version of a paper presented to the International Economic Associa-
tion meetings, Beijing, 8 July 2011. Research assistance from Laurence Wilse-Samson
and Eamon Kircher-Allen is gratefully acknowledged.
1. Asymmetric information concerns have ruled out many natural financial market
recapitalizations, like extensive new equity issues. Some recapitalization was provided
directly through government (TARP), but as we note later, most of the recapitaliza-
tion was through retained earnings. The underlying theory, with its implications
for banks, was set out several years before the crisis in Greenwald and Stiglitz
(1993b, 2003) based on their work in the 1980s and early 1990s (see Greenwald and
Stiglitz, 2003, for a complete list), resting on micro-foundations provided by, for
instance, Greenwald et al. (1984) and Majluf and Myers (1984). See also Bernanke
et al. (1999) on the working of the financial accelerator in an asymmetric infor-
mation framework. For surveys, refer to Stiglitz (1988, 2011), and Greenwald and
Stiglitz (1993a). Many models focusing on balance sheet effects of financial disrup-
tion look not just at the financial sector (see, for example, Greenwald and Stiglitz
1993b; Adrian and Shin 2008; and Shiller 2008), though disruptions in the finan-
cial sector have particularly large systemic effects, and especially after the repeal of
deposit rate restrictions, are particularly slow in reversing (Greenwald and Stiglitz
2003). Household and company balance sheets are restored only slowly over time
through accumulated savings and debt reductions associated with graduate declines
in real asset holdings by means of inventory liquidations and gross investments lev-
els below depreciation. The process has obvious adverse short-run macroeconomic
consequences.
2. There are other parts of the explanation that we cannot pursue here. For instance,
countries, such as China, that instituted strong Keynesian policies, counteracted the
decrease in global trade. Their policies were, at the same time, structural policies,
sensitive to the changed composition of demand.
3. Indeed, Greenwald and Kahn (2008) estimate that between 1980 and 1991, a period
of significant manufacturing job loss, 85 per cent of the decline in manufactur-
ing employment was due to productivity growth, and only 15 per cent was due to
increased imports. While more recently globalization has played a more important
role, still, over the longer period from 1991 to 2007, two-thirds of the decline in
manufacturing employment was due to productivity growth and only one-third to
imports – the growth in China notwithstanding. If manufacturing productivity grows
at 6 per cent, even if the global economy grows at the impressive rate that it has been
recently, say 4.5 per cent, and manufacturing demands grows roughly commensu-
rately, then global employment in manufacturing will decline. Globalization means
that there is a global fight over where remaining jobs will be located, that is, who has
to make the largest adjustments.
4. It is ambiguous because declining agricultural prices enhance urban worker welfare.
Real wages measured in manufactured goods may decline while real wages in agri-
cultural goods may increase. This in fact happened in the Great Depression. In that
case, urban real (consumption) wages, using the CPI, actually rose while manufactur-
ing real product wages fell. But the appropriate model for describing what happened
is one that incorporates unemployment, reflected in the efficiency wage model of
section IV (See Greenwald and Stiglitz 1988).
5. The 1991–1992 recession is often related to the banking crisis that preceded it (see
Stiglitz 2003; Greenwald and Stiglitz 2003), and the 2001 downturn is generally
related to the breaking of the tech bubble. But in both cases, as now, the bubbles
that preceded these crises can be related to underlying problems in the real sector.
6. See, for instance, Greenwald and Stiglitz (1993b, 2003), Bernanke et al. (1999),
Korinek (2011) or Stiglitz (2011). For surveys, see Stiglitz (1988) or Greenwald and
Stiglitz (1993b). As Greenwald and Stiglitz (1993b) point out, these financial con-
straints not only can explain amplification (why small shocks can have large effects),
but persistence, including why recoveries are so slow. These results stand in marked
contrast to Real Business Cycle Models (RBC) without financial constraints, where the
fluctuations simply reflect random real shocks to the economy. Indeed, in the absence
of financial constraints, there are a number of ‘buffers’, like inventories, the effect of
which is to dampen the impact of any real shock to the economy.
7. We should emphasize that there may be other real factors contributing to the insuffi-
ciency of aggregate demand. In particular, the period before this crisis, like the period
before the Great Depression, was marked by large increases in inequality (Atkin-
son et al. 2011; United Nations 2009; Rajan 2010). With the marginal propensity
to consume of high-income households being less than that of low-income house-

holds, such a redistribution would be expected to reduce aggregate demand. The
necessary adjustments of factor and goods prices to restore a full employment general
equilibrium may have been large, and evidently did not occur. Again, the financial
sector played a role in forestalling the impacts: lower-income individuals were able
to continue to consume as if their incomes were not stagnating by borrowing.
One other real factor may have played a major role in this crisis: high oil prices
redistributed income from the US and Europe to the oil-producing countries, whose
propensity to consume may be circumscribed by the recognition that given the high
volatility of oil prices and the risks of Dutch disease problems, oil producers should
set aside large fractions of their revenues.
Still another factor contributing to the global crisis was the large build-up of reserves
by emerging countries (motivated sometimes by a precautionary demand for savings,
sometimes by the pursuit of export-led development strategies in the presence of con-
straints on industrial policies). (See Greenwald and Stiglitz 2010a, 2010b.) Whatever
the motivation, the increase in reserves subtracted from global aggregate demand.
8. There is another major deficiency in RBC and related models. In those models, the
shock to the economy is exogenous and it is modeled by a simple stochastic process.
In reality, the most important disturbances are endogenous and episodic – whether
they are bubbles or the structural transformations upon which we focus here.
9. As well as, perhaps, marginal manufacturing using techniques pre-dating the rollout
of electrification and mass production.
10. See, for example, Reinhart and Rogoff (2009). The theory is set forth, for example, in
Greenwald and Stiglitz (1993b).
11. Financial sector problems arise both directly, as a result of a weak economy, and
indirectly, as a result of government responses to the fear of a weak economy. Govern-
ments may respond to what otherwise would be a weak economy by lowering interest
rates and weakening regulation, or undertaking other policies that help create future
financial crises. Arguably, this was the case in the US.
12. Private non-residential fixed investment as a percentage of GDP was around 10 per
cent in the second quarter of 2011, while the historical post-war average is 10.7 per
cent (though we note that GDP has fallen below trend). Equipment and software
investment by firms in real terms was about 8.2 per cent of GDP in early 2011 com-
pared to a high of 8.4 per cent in 2007 and 6.6 per cent at the peak of the crisis in
the fourth quarter of 2008. These investments, being difficult to collateralize, are the
first to suffer when bank lending is restricted. Their relatively high level suggests that
a shortage of bank lending has not had a significant dampening effect on business
investment. Business investment in structures has fallen sharply but this appears to
be due more to the overhang of empty buildings from the earlier boom than to lim-
itations on bank lending. The level of commercial and industrial lending for small
domestic banks rose in the second quarter of 2011 to second quarter 2007 levels,
after a prolonged period of being far lower. (Seasonally adjusted commercial and
industrial loans at all commercial banks were $1.293 trillion on 1 August 2011, and
$1.293 trillion on 1 July 2007, according to figures from the St Louis Fed. Available
at http://research.stlouisfed.org/fred2/categories/32389, accessed 24 January 2011).
The trend toward industrial firms holding more cash is not new. Bates et al. (2009)
document between 1980 and 2006 a doubling in the average cash-to-assets ratio for
US industrial firms, such that ‘at the end of the sample period, the average firm can
retire all debt obligations with its cash holdings’. They find (p. 2018) that the ‘main
reasons for the increase in the cash ratio are that inventories have fallen, cash flow
risk for firms has increased, capital expenditures have fallen, and R&D expenditures
have increased’ (where cash flow risk is measured as the standard deviation of industry
cash flow to assets). It may be that the post-crisis build-up in cash reflects increased
uncertainty and the consequences of the extreme credit conditions of 2008, which
many businesses fear may occur again.
13. Inflationary expectations, as reflected by TIPS, are also low. The average spread
between TIPS and 10-year treasuries (a good measure of expected inflation) was about
2 per cent from 2010 through the summer of 2011. The CPI increase between August
2010 and August 2011 (excluding food and energy) was also 2 per cent (although
the overall index including food and energy increased by 3.6 per cent). Data from
St Louis Fed, available at http://research.stlouisfed.org/fred2/series/CPILFESL and
http://research.stlouisfed.org/fred2/series/CPIAUCSL?cid=9.
14. Krugman (2009) and Eggertsson and Krugman (2010).
15. More accurately, it is hard to argue for this within the conventional models, in which
credit rationing does not exist. Stiglitz and Weiss (1981) explain why there may be
credit rationing, and Greenwald, Stiglitz and Weiss (1984) explain why the extent
of credit rationing may vary over the business cycle. But as we noted, the level of
investment in equipment and software and the magnitude of cash holdings by large
firms suggests that by mid-2011 finance was not the major constraint on recovery.
16. Moody’s estimates that some 14 million homeowners are in positions of negative
equity, ‘half by more than 30% …(and) the average underwater homeowner’s debt
exceeds market value by nearly $50,000’ (Zandi 2011: 2).
17. Dynan et al. (2004: 399–400) find savings rates varying from zero for the lowest
quintile of the income distribution to in excess of 25 percent for the top.
18. Personal savings rates were around 5 per cent in 2009, before rising towards 6 per cent
in 2010. At the end of 2011, rates dropped back down to 3.5 per cent, around what
they were in 2004.
19. The analysis of this paper does not deny the importance of the failings of the financial
sector in determining not only the timing of the crisis, but also the depth and duration
of the downturn. The legacy of excess investments in real estate and of excessive
indebtedness by households is playing a role, just as – as we argue below – the build-
up of ‘forced savings’ during World War II helped not just to prevent the US from
sliding back into recession or worse, but to propel the country into a new prosperity.
20. Deleveraging could have one important effect on aggregate demand: lower expendi-
tures servicing debt would leave more money to spend on real goods – illustrating
another way in which the excessive financialization of the economy may have con-
tributed to its weaknesses. But the data suggests that this effect is likely, at most, to be
small – perhaps because of the innovativeness of the financial sector in finding new
ways of extracting money from consumers, partly because some of the deleveraging
is taking the form of home foreclosures, forcing individuals into rental properties,
which over the longer run may actually reduce what can be spent on other goods
and services. Non-consumption household outlays, which include household inter-
est payments, fell from 3.94 per cent of total outlays at the peak of the borrowing
boom in 2007 to 3.45 per cent at the end of 2010. The resulting increase in funds
available for consumption was less than 0.5 per cent, and this includes the impact of
lower household interest rates as well as deleveraging.
21. Once the deleveraging process is completed, the rate of growth of consumption might
be restored to a more normal level. But full economic recovery, with a restoration of
full employment, would require still more rapid growth.
22. This data is only meant to be suggestive, because many factors contributed to the
depth of the downturn and the speed of recovery (and there are alternative measures of
the depth of the downturn – Germany had a larger downturn in output, but a smaller
downturn in employment). Some countries (such as China) took strong actions to
offset the downward pressures. Still, these experiences suggest that it is structural
factors (the composition of output and trade dependence), as much as weaknesses in
the financial sector, that determined the depth of the downturn. With the precipitous
fall in trade, especially in manufactured goods, countries that were more dependent
on exports of manufactured goods suffered more, ceteris paribus. To be sure, with
weak banking systems, precipitous declines in GDP can translate into financial sector
problems, making the challenge of recovery greater.
The evolution of the crisis has also thrown doubt on other shibboleths. A cen-
tral contention of some central bankers (and many strands of macroeconomics) has
been that it is wage rigidities which give rise to extended periods of unemployment.
Yet in this crisis the United States, supposedly the advanced industrial country with
the most flexible labor market, has been plagued with higher and more persistent
unemployment (especially relative to the drop in GDP) than, say, Germany. This
is consistent with both theoretical work (surveyed in Greenwald and Stiglitz 1993a)
that argues against the hypothesis that it is wage and price rigidities that are primarily
responsible for the magnitude of employment and output fluctuations (on the con-
trary, fluctuations may be greater with more flexible wages and prices) and with the
confirming empirical studies (Easterly et al. 2001a, 2001b).
23. Even the United States had one costly episode, the S & L crisis of the 1980s, and would
have had more had the government not engineered (through the IMF) bailouts, for
example as a result of the Latin American debt crisis.
24. Employing the definitions of Reinhart and Rogoff (2009), the proportion of countries
experiencing new external debt crises reached as high as 40 per cent in the mid-1980s,
and the proportion of countries experiencing banking crises reached 30 per cent in the
late 1990s. These were the highest since World War II and represented a precipitous
increase since the moderate period between 1945 and 1980 (Reinhart and Rogoff
2009: 74).
25. There have been extensive recapitalizations, both through the issue of new shares
and (sometimes forced) retention of high earnings (facilitated by the low interest
rates at which the banks can get access to funds). Still, critics argue that what has
been done is not enough, that banks continue with highly risky activities, that their
lack of transparency makes it difficult to judge the adequacy of their capital, and that,
as a result, weaknesses in the financial sector continue to plague the economy. The
lack of confidence in the financial sector is manifested by the high volatility of bank
share prices. Still, the most direct consequence of the weaknesses in the financial
sector should be on the level of investment, and, apart possibly for the availability
of finance to SMEs, this does not seem to be impaired by weaknesses in the financial
sector.
26. Far more important than relative resource endowments is knowledge, so that what
matters is dynamic comparative advantage, which is endogenous, and which can
change markedly over time (Greenwald and Stiglitz 2006, 2012).
27. As this chapter goes to press, it is far from clear that the crisis is over, despite fiscal
and monetary interventions that have also been without precedent in the postwar
era (and even the prewar period). Most projections suggest that it will be years before
unemployment returns to ‘normal’ levels.
28. There is also a resurgence of interest in the ways in which deep downturns differ from
ordinary downturns; see Stiglitz (2011). Interestingly, there is no consensus about
the causes of the Great Depression, including the relative role of monetary versus
real forces. See, for instance, Temin’s Lionel Robbins Lecture (1991). The explana-
tion we provide here focusses on the source of the underlying disturbance to the
economy, and one of the impediments to the economy’s adjustment to this distur-
bance. Greenwald and Stiglitz (1993b, 2003) discuss other factors that contribute to
the amplification and persistence of shocks, including financial market constraints
and imperfections. This analysis does not rule out that flawed monetary and reg-
ulatory policies and asset price bubbles might have contributed to the depth and
duration of the Depression. As we argue below, however, increasing wage flexibility
might have made matters worse (in contrast to much of standard New Keynesian
analysis where the focus of attention is on nominal wage [or price] rigidities.) The
analysis is also consistent with the hypothesis that asymmetries in adjustment speeds
across sectors played an important role in the evolution of the crisis. See Stiglitz
(1999).
29. Reinhart and Rogoff (2009) claim to have identified the unusually prolonged conse-
quences of business cycles associated with financial crises. But they make no serious
attempt to examine the original causes of these crises. To the extent that severe real
imbalances that take a long time to resolve ultimately lead to financial crises more
often than less severe ones, financial crises will be a symptom of severe real imbal-
ances. In this case, they have merely discovered that severe imbalances are more
prolonged than mild ones.
30. Contemporaneous work citing the importance of agriculture includes League of
Nations (1931) and Timoshenko (1933). In a public lecture given in October 1931,
Dennis Robertson ascribes the ‘primary cause’ (original emphasis), as the ‘glut’ of cap-
ital goods and, in particular, ‘(i) the rapid application of science to agriculture […]
leading […]to a decline in the total receipts even of [low-cost producers], (ii) the
decline of the rate of growth of the population […] (iii) the durable nature of some
new objects of consumption’ (Robertson 1956: 72).
31. Initial banking distress was particularly acute in rural areas. Chandler (1970: 62)
reports, ‘in the three years 1930–32, 5,096 banks failed in the United States, 3,448
of these […] were in places with populations below 2,500.’
32. Combined with supply effects associated with expectations of future prices, the
impact of financial constraints in limiting investment in agriculture, and speculative
hoarding.
33. Chandler (1970) cites the 23 million acres of farmland that became available from
the replacement of draft animals by automobiles, trucks and tractors. He also notes
as important (1970: 55), ‘Continued advances in technology [which] were a major
force tending to increase total farm output. These took many forms: improvements
in methods of farm management, better adaptation of crops to soils, development of
more efficient plants and animals, and so on.’
34. Net farm income after expenses fell from a peak of $9.6 billion in 1919 to $6.3 bil-
lion in 1929 to $1.9 billion in 1932. It recovered to $5.7 billion in 1937 and fell to
$4.2 billion in 1938 where it remained through 1940. Net farm income deducts farm
wages. Nominal GDP was: $84 billion in 1919, $103.1 billion in 1929, $58 billion
in 1932, and $84.7 billion in 1937; while at 1958 prices it was, 146.4, 203.6, 144.2
and 203.2, respectively. Source: United States Bureau of the Census, 1975, Historical
Statistics of the United States Colonial Times to 1970: 483–484.
There are, of course, two possible reasons for the dramatic decline in income –
a fall in prices or a fall in quantities. Our model is predicated on an increase in
productivity which would have generated a decline in prices even in the absence
of a recession/depression; but the recession/depression exacerbated the magnitude
of the decline, as the model in the following section illustrates, but given the low
income elasticity of food, the quantitative importance of this may be limited. Data
for internationally traded goods (cotton, corn, and wheat) show dramatic declines in
prices from 1929 to 1932. In some parts of the United States, these price declines were
reinforced by quantity declines as a result of environmental disaster (the dustbowl)
(Hornbeck 2011). However, in the aggregate, quantities actually increased. As detailed
by Chandler (1970: 58), ‘In contrast to behavior in most other industries, real output
in agriculture did not fall […] total farm output in 1931 and 1932 was slightly higher
than in 1929. The most important reasons […] were the recognition by each indi-
vidual farmer that he could not raise prices by reducing his output […].’ Chandler
concludes (1970: 59), ‘Thus, the entire decrease in the money incomes of farmers
resulted from declines in the prices of farm products […] by 1932, prices received by
farmers had fallen 56 percent below their levels in 1929, while prices paid by farm-
ers had declined only 32 percent.’ One might ask why, beside the fall in demand
and the price inelasticity of demand, there should have been a decline in prices of
this magnitude, given the limited rise in agricultural output. One explanation is that
prior to 1929, farmers had been hoarding, in the anticipation that prices would rise,
so that the flow of produce on the market was less than the output. Once storage
capacity constraints are reached, the flow of produce must equal that of production;
and if market participants anticipate that prices will not recover any time soon, or
can no longer finance large stocks in storage, de-hoarding will occur, so that the flow
of produce will exceed production.
35. In fact, between 1931 and 1934 there was net in-migration of around 700,000 (com-
pared to, for example, net outmigration of 6.4 million between 1942 and 1944). See
Carter et al. (2006).
36. The theory of migration from rural to urban sectors, taking account of the conse-
quences of urban unemployment, and rural credit constraints, is well developed in
the development literature. See, for instance, Harris and Todaro (1970) and Stiglitz
(1969, 1974).
37. In an open economy, there is a net loss (if the country is a food exporter, like Argentina
or the US) or a net gain (if the country is a net food importer).
38. Hansen (1941: 232–234) derives numbers from the 1939 Consumer Expenditures in the
United States indicating the propensity to consume for 1935–36. For income earners
earning less than $500, consumption as a per cent of income is found to be 149.4
per cent, for those earning in excess of $20,000, this falls to 49.3 per cent. Marriner
Eccles (1951: 76), appointed Chairman of the Federal Reserve by Roosevelt, framed
the problem as ‘by taking purchasing power out of the hands of mass consumers, the
savers denied to themselves the kind of effective demand for their products that would
justify reinvestment of their capital accumulations in new plants. In consequence, as
in a poker game where the chips are concentrated in fewer and fewer hands, the
other fellows could stay in the game only by borrowing. When their credit ran out,
the game stopped.’ Attempting to explain his high, back-of-the-envelope estimated
multipliers for the Depression period, Field (2011: 240) observes, ‘[the] Depression
reduced income, but it also reduced inequality, and this reduced saving both in the
aggregate and as a share of GDP. Gross saving as a share of GDP was 18.6 percent of
GDP in 1929, but fell to 5.6 percent in 1932. It recovered to 17.5 percent in 1937 and
had risen to 23.5 percent in 1941.’
39. Between 1930 and 1932, 68 per cent of bank failures were in rural areas – where
populations were below 2,500 people (Chandler 1970: 62). Friedman and Schwartz
describe the onset of the First Banking Crisis in October 1930 as beginning in the
agricultural sector: ‘A contagion of fear spread among depositors, starting from the
agricultural areas, which had experienced the heaviest impact of bank failures in
the twenties’ (Friedman and Schwartz 1963: 308). Madsen (2001: 328) makes similar
points across countries – ‘The declining real prices of agricultural products […] had
adverse effects on consumption and investment. First, the marginal propensity to
spend of those who lost income exceeded the marginal propensity to spend of those
who experienced income gains. Second […] declining real prices of farmland […]
increased the cost of borrowing for farmers, and thus adversely affected investment
and […] consumption. Third, the declining ability of farmers to honor their debt
obligation adversely affected the functioning of the banking sector […] For the United
States […] William Arthur Lewis argues that the declining agricultural prices, the fall
in real estate values, and the bankruptcy of farmers were the most important factors
behind the bank failures.’
40. For empirical data on the subject, see Bell (1940) and Swanson and Williamson (1972).
Note too that increased uncertainty of future income, as the crisis evolved, may have
reinforced these effects, as even urban workers who retained their jobs and benefitted
from lower agricultural prices faced a risk of a job loss, with poor prospects of reem-
ployment. The model below does not incorporate this effect, or one other, that may
be playing a role in the current crisis: the resulting weaknesses in the urban labor mar-
ket may lead to some lowering of real urban wages (even in the presence of efficiency
wage concerns), and the resulting intra-sectoral redistribution may have an adverse
effect on aggregate demand.
41. Romer (1992) finds almost no role for fiscal policy in the recovery from Great Depres-
sion between 1933 and 1942, ‘fundamentally due to the fact that the deviations of
fiscal policy from normal were not large during the 1930s’ (768). Cary Brown also
shows an increase in net taxes (taxes minus transfers) of $2.9 billion between 1936
and 1937. This appears to arise from a misreading of the difference between a ‘cash
budget’ and the administrative budget. Since there was no massive drop in transfers,
the change must have come from tax increases. But IRS collection data show no such
increase (also the data are for fiscal years ending June 30, not calendar years – Brown
is not clear about what period he is using).
IRS Receipts for the Relevant Fiscal Years (in $ billions)
Personal Income Social Security

Year Total Tax Corporate Tax Excise Tax Tax
1939 5.2 1.029 1.283 1.768 0.834

1938 5.7 1.286 1.482 1.731 0.742
1937 4.6 1.091 1.217 1.764 0.266
1936 3.5 .674 .848 1.547 0.048
The big rise in Social Security tax of about $0.5 billion occurs between FY1937 (end-
ing June 1937) and FY1938. Excise taxes (and customs receipts not included here)
are relatively flat. Corporate and personal taxes rise by about $0.8 billion but this
rise is between FY1936 and FY1937. Romer argues thus that the growth in real GNP
between 1933 and 1937 and 1938 and 1942 was primarily the result of an increase
in Aggregate Demand due to monetary expansion, but her analysis does not seem to
take into account the large increase in agricultural incomes in these periods as a result
of increased global agricultural prices (themselves, in part, a lagged supply response
to the very low prices in earlier years, lags of the kind that played such an important
role in the corn-hog cycle). Unemployment remained elevated at 10 per cent in 1941,
but down from a high of 25 per cent in 1933.
42. We ignore variations in the hours worked. The question of why downward adjust-
ments in aggregate hours worked takes the form of unemployment rather than just
hours worked per employee is one of the central questions of macroeconomics.
43. It would be easy to generalize these results, to make migration costs be a function of
urban and rural prices and wages. Training costs are naturally related to wM , rehous-
ing, to the cost of urban goods. One of the main costs is the opportunity cost of search
(as in the rural-urban migration models in the development literature.)
44. This assumes that there is full employment in the urban sector. Later, we will dis-
cuss conditions (like efficiency wages) which result in urban unemployment. Then,
workers contemplating migration from the rural to the urban sector have to take into
account the probability of finding a job (or the expectation of a period of unemploy-
ment.) The precise relationship depends on how vacancies are filled and the nature
of job search. See, for example, Stiglitz (1974).
If there is not ongoing migration because rural workers cannot obtain the capital
to finance migration, then we replace equation (5) by an inequality wA M
t < (1 − f )wt =
(1 − f )mt .
45. Equation (7) (and other demand equations used below) could easily be derived from
underlying utility maximization. But note that the hypothesis of rational expectations
is hardly plausible in the current context: structural transformations of the kind that
we are describing in this chapter occur very infrequently, sufficiently rarely that there
would be limited statistical bases for making inferences about the future evolution of
prices and employment; accordingly, different individuals are likely to have different
beliefs about the future. What matters for our analysis is not how those beliefs are
formed, but the resulting demand functions postulated in equations (7) and, e.g. (12)
below.
By the same token, our analysis does not explicitly incorporate the consequences
of dysfunctions in the financial sector. Presumably, these would be reflected in
lower investment, which under normal stability conditions, leads to a lower equi-
librium level of employment and lower prices. The demand curves are aggregate
demand curves, and to the extent that access to debt is constrained (or expected
to be constrained), current consumption, especially of urban goods, will be con-
strained. The demand for agricultural goods may also be constrained. The net effect is
again to lower further rural prices, exacerbating the adverse effects of the productivity
increases.
46. Total agricultural demand DA is given by A t d
AA ((1 − f )(m /a ), (1 − f )m A ) +
t t t t
M
t d
AM ((1 − f )(m /a ), m ( − A )). ε A is the partial elasticity of D with respect to
t t t t t p
p, keeping At fixed. There are both income and substitution effects; in particular, an
increase in price increases the income of agricultural workers, which, by itself, would
lead to increased demand for agricultural goods.
47. The last result is, however, not general, and depends strongly on our production
assumptions. More generally, an increase in the productivity of workers in agriculture
could have distributive consequences (between labor and other factors of production)
so that workers could be worse-off. We have formulated our model deliberately to
avoid these distributive issues.
48. For an early discussion of the role of financial constraints in determining migration
(and urban-rural equilibrium, in the context of a developing country), see Stiglitz
(1969). Note that in this model, the major effect of a disruption to the financial
system is that it would make the financing of moving more difficult, that is, a smaller
fraction of the population could obtain the funds required to move. In practice, few
individuals actually finance migration through loans.
Individuals differ, of course, not only in the access to funds, but in the returns to
migration. If all individuals of the same age cohort are identical, then it would be the
youngest people who would migrate first (in a world with perfect capital markets),
since they could amortize the fixed costs of moving over a longer period. In addition,
different individuals face different prospects of getting an urban job. The analogy
in terms of transferring from manufacturing (or construction) to a job in services is
the limited access to funds for human capital upgrading. Workers need to invest to
develop the new skills needed for the new job (as well as move to where job prospects
are better).
49. There is an alternative formulation that gives more ambiguous results, with the pos-
sibility of intermittent periods of migration. Assume that that there is a distribution
of costs of capital. Then migration occurs to the point where the annualized cost
of migration equals Wm − WA (assuming static expectations). If, for the individ-
ual with the lowest cost of capital, the cost of migration exceeds Wm − WA , then
there will be no migration. But a fall in the agricultural wage relative to the urban
wage might induce migration, even if worsening conditions in the agricultural sector
led to an increase in cost of capital even for the individual with the lowest cost of
capital. (In the efficiency wage version, to be discussed below, there is urban unem-
ployment; what matters for migration is expected lifetime income of an individual
who migrates to the city. That depends on how the urban labor market functions, for
example, if there is a daily labor market, so the expected wage is (1−U ) wm , or whether
there is a queue for jobs, with migrants coming at the end of the queue. See Stiglitz
(1974).
50. This equation might be interpreted as suggesting that those in the urban sector who
are not employed have zero demand for food. A better way of thinking of equation
(12a) is urban demand depends on urban income (Eξ(Et )) and relative prices, p : DAM =
DAM (Eξ(Et )), p), where DAM is total urban demand for agricultural goods. We have
simplified by assuming that DAM takes on the special form: Et d AM (pt, ξ(Et )). But the
analysis does not depend on this parameterization.
51. Technically, we assume that H (E) > Et dyMM ξ ‘(Et ) + d MM , or h > sMM (εdMMy μ + 1),
where h = d ln H/d ln E, the elasticity of output with respect to employment, sMM is the
share of total manufactured goods purchased by urban workers, εdMMy is the income
elasticity of demand for urban goods by urban workers, and μ ≡ d ln ξ(Et )/d ln E, the
percentage change in wages from a one per cent increase in employment along the
labor supply curve (the inverse of the labor supply elasticity).
52. From the indirect utility function, we require (1 + μ) > σ MA (d ln p/d ln E)AA where
σ MA is the average propensity of those in the urban sector to consume agricultural
goods (≡ pd MA /w) and (d ln p/d ln E)AA denotes the elasticity of the AA curve. If σ MA is
small enough, and the AA curve is flat enough (increases in urban employment have
a relatively small effect on the market-clearing price in the rural sector) it is clear that
urban workers are better-off at the high price equilibrium. The slope of the AA curve
can, in turn, be related to demand elasticities and labor supply elasticities.
53. The condition is simply that, at a fixed urban wage (employment level), the decrease
in price as at increases in the agricultural sector (that is, for the AA curve) is greater
than the decrease in the urban sector (that is, for the MM curve). For the former,
d ln p/dlna = −(1 − sA eyAA )/εpA and for the latter, d ln p/d ln a = −sM eyAM /ε1/p
M , where e Ai
y
is the income elasticity of sector i goods of agricultural workers, sM is the share of
urban goods consumed by rural workers, εpA is the (absolute value of) price elasticity
M is the (absolute value of) price elasticity of manufac-
of agricultural goods, and ε1/p
tured goods (noting that the relative price of urban goods is 1/p). Normally, we would
expect 1 − sA ≥ sM with strict inequality if there is savings; and low income elastic-
ities (especially for food) so that (1 − sA eyAA ) > sM eyAM , i.e., 1 > sA eyAA + sM eyAM and
εpA ε1/p
M – food consumption is also price inelastic. We will refer to the case where
(1 − sA eyAA )/εpA > sM eyAM /ε1/p

M and where the stability condition is satisfied as the
‘normal case’.
54. It is easy to generalize these results to the case where the efficiency wage is, itself, a
function of the employment (unemployment) level, as in the standard Shapiro-Stiglitz
no-shirking model. We then replace w∗ with a function, w∗ (E).
55. Define, as before, DA (p; E) ≡ A d AA (p, γ ) + Ed MA (p, w∗ ) as total demand in the agri-
cultural sector, and define DM (p, E) similarly. Then the slope (elasticity) of the AA
curve is given by – Ed MA /pDpA = (1 − sAA )/εpA where εpA is the price elasticity of total
demand in the agricultural sector (now, keeping E and A fixed) and sAA is the share
of consumption of food by those in the agricultural sector.
The elasticity of the MM curve is −(H E − Ed MM )/pdpM = h − sMM /εpM , where, it will
be recalled, sMM is the share of manufactured goods consumed by those in the urban
sector, and εpM is the (absolute value of the) elasticity of demand of manufactured
goods with respect to the agricultural price. When the price of agricultural goods
increases, there are two effects: a substitution effect away from food and towards
manufactured goods, and an increase in real incomes of those in the rural sector
and a decrease of those in the urban sector. In a representative agent model, the
redistribution effects cancel and there is only the substitution effect. If the elasticity
of substitution is low, εpM will be small. On the other hand, an increase in agricultural
prices results in a substitution effect against agriculture (again related to the elasticity
of substitution) combined with an income effect (farmers are better-off, urban workers
worse-off).
Hence we assume (h − sMM )/εpM > (1 − sAA )/εpA . If price elasticities (appropriately
defined) are approximately the same, then stability (AA being flatter than MM) simply
requires that (h − sMM ) > (1 − sAA ). This will be true if sAA > sMM and h is near unity,
that is, agriculture workers consume a larger fraction of their own goods than urban
workers’ share of consumption (production) of their own goods.
56. If it were not satisfied, it would mean that rural workers were getting increasingly
indebted to the urban sector – they would be buying more than they are selling – or
that they were saving (paying back prior debts). Given the constraints in financial
markets, the former does not seem plausible; and in the presence of large declines
in incomes, neither does the latter. This assumption allows us to greatly simplify the
analysis.
57. Defined earlier as the AA curve being flatter than the MM curve.
58. By hypothesis, we assume that Ricardian equivalence does not hold, so that the future
tax liabilities do not lead to an equal and offsetting reduction in consumption today.
This follows naturally from our assumptions of capital market imperfections, which
underlies the entire analysis.
59. For a discussion, see, for example, Sah and Stiglitz (1992).
60. Keynes anticipated this effect: ‘if labour were to respond to conditions of gradually
diminishing employment by offering its services at a gradually diminishing money-
wage, this would not, as a rule, have the effect of reducing real wages and might even
have the effect of increasing them, through its adverse influence on the volume of
output. The chief result of this policy would be to cause a great instability of prices,
so violent perhaps as to make business calculations futile in an economic society
functioning after the manner of that in which we live’ (Keynes 1936: Chapter 19). The
quotation makes clear that Keynes did not think that the ‘solution’ to unemployment
was to lower wages. Our model is consistent with this result, in that output falls, and
since agricultural prices fall, real wages in agricultural goods could actually rise, and
are likely to do so if the MM curve is very steep.
61. It continued to fall rapidly in the 1950s and 1960s.
62. Estimates of growth based on exchange rates are lower than those based on purchasing
power parity. The latter is probably more relevant for estimating the growth in the
demand for manufactures.
References
Adrian, T. and Shin, H. (2008) ‘Liquidity and Leverage’, Federal Reserve Bank of New York
Staff Report # 398.
Atkinson, A. B., T. Piketty and E. Saez (2011) ‘Top Incomes in the Long Run of History’,
Journal of Economic Literature, vol. 49, no. 1, pp. 3–71.
Bates, T. W., K. M. Kahle and R. M. Stulz (2009) ‘Why Do US Firms Hold So Much More
Cash than They Used To?’, Journal of Finance, vol. 64, no. 5, pp. 1985–2021.
Bell, S. (1940) Productivity, Wages and National Income (Washington, DC: The Brookings
Institution).
Bernanke, B., M. Gertler and S. Gilchrist (1999) ‘The Financial Accelerator in Quanti-
tative Business Cycle Framework’, in L. Taylor and M. Woodford (eds), Handbook of
Macroeconomics, Vol. 1C (Amsterdam: North-Holland).
Carter, S. B., S. S. Gartner, M. R. Haines, A. L. Olmstead, R. Sutch and G. Wright (eds) (2006)
Historical Statistics of the United States, Millennial Edition Online (Cambridge: Cambridge
University Press).
Chandler, L. V. (1970) America’s Greatest Depression (New York: Harper and Row).
Dynan, K. E., J. Skinner and S. P. Zeldes (2004) ‘Do the Rich Save More?’, Journal of Political
Economy, vol. 112, no. 2, pp. 397–444.
Easterly, W., R. Islam and J. E. Stiglitz (2001a) ‘Shaken and Stirred: Explaining Growth
Volatility’, Annual Bank Conference on Development Economics 2000 (Washington: World
Bank), pp. 191–212.
Easterly, W., R. Islam and J. E. Stiglitz (2001b) ‘Volatility and Macroeconomic Paradigms
for Rich and Poor Countries’, in J. H. Drèze (ed.), Advances in Macroeconomic Theory:
Volume 1, IEA Conference vol. 133 (Basingstoke: Palgrave Macmillan), pp. 353–372.
Eggertsson, G. B., and P. Krugman (2010) ‘Debt, Deleveraging, and the Liquidity Trap: A
Fisher-Minsky-Koo Approach’, mimeo, Princeton University.
Field, A. (2011) A Great Leap Forward: 1930s Depression and Economic Growth (New Haven:
Yale University Press).
Friedman, M. and A. J. Schwartz (1963) A Monetary History of the United States, 1867–1960
(Princeton: Princeton University Press).
Greenwald, B. and M. Kahn (2008) Globalization: The Irrational Fear that Someone in China
Will Take Your Job (New York: Wiley).
Greenwald, B. and J. E. Stiglitz (1988) ‘Examining Alternative Macroeconomic Theories’,
Brookings Papers on Economic Activity, vol. 1, pp. 207–270.
Greenwald, B. and J. E. Stiglitz (1993a) ‘New and Old Keynesians’, Journal of Economic
Perspectives, vol. 7, no. 1, , pp. 23–44.
Greenwald, B. and J. E. Stiglitz (1993b) ‘Financial Market Imperfections and Business
Cycles’, Quarterly Journal Of Economics, vol. 108, no. 1, pp. 77–114.
Greenwald, B. and J.E. Stiglitz (2003) Towards a New Paradigm in Monetary Economics
(Cambridge: Cambridge University Press).
Greenwald, B. and J. E. Stiglitz (2006) ‘Helping Infant Economies Grow: Foundations of
Trade Policies for Developing Countries’, American Economic Review: AEA Papers and
Proceedings, vol. 96, no. 2, pp. 141–146.
Greenwald, B. and J. E. Stiglitz (2010a) ‘A Modest Proposal for International Monetary
Reform’, in S. Griffith-Jones, J. A. Ocampo, and J. E. Stiglitz (eds), Time for a Visible
Hand: Lessons from the 2008 World Financial Crisis, Initiative for Policy Dialogue Series
(Oxford: Oxford University Press), pp. 314–344.
Greenwald, B. and J. E. Stiglitz (2010b) ‘Towards a New Global Reserves System’, Journal
of Globalization and Development, vol. 1, no. 2, Article 10.
Greenwald, B. and J. E. Stiglitz (2012) ‘Creating a Learning Society: A New Paradigm for
Development and Social Progress’, working paper, Columbia University.
Greenwald, B., J. E. Stiglitz and A. Weiss (1984) ‘Informational Imperfections in the Capital
Markets and Macro-economic Fluctuations’, American Economic Review, vol. 74, no. 1,
pp. 194–199.
Hansen, A. (1941) Fiscal Policy and Business Cycles (New York: W.W. Norton).
Harris, J. R. and M. P. Todaro (1970) ‘Migration, Unemployment and Development: A
Two-Sector Analysis’, American Economic Review, vol. 60, no. 1, pp. 126–142.
Hornbeck, R. (2011) ‘The Enduring Impact of the American Dust Bowl: Short and Long-
run Adjustments to Environmental Catastrophe’, forthcoming, American Economic
Review.
Keynes, J. M. (1936) The General Theory of Employment, Interest and Money (Cambridge:
Harcourt, Brace and Company).
Korinek, A. (2011) ‘Systemic Risk-taking: Amplification Effects, Externalities, and Regula-
tory Responses’, working paper, Department of Economics, University of Maryland.
Krugman, P. (2009) The Return of Depression Economics and the Crisis of 2008 (New York:
W.W. Norton).
League of Nations (1931) The Agricultural Crisis (Geneva: League of Nations).
Madsen, J. (2001) ‘Agricultural Crises and the International Transmission of the Great
Depression’, The Journal of Economic History, vol. 61, no. 2, pp. 327–365.
Majluf, N. and S. Myers (1984) ‘Corporate Financing and Investment Decisions When
Firms Have Information That Investors Do Not Have’, Journal of Financial Economics,
vol. 13, no. 2, pp. 187–221.
Rajan, R. (2010) Fault Lines: How Hidden Fractures Still Threaten the World Economy
Reinhart, C. and K. Rogoff (2009) This Time Is Different: Eight Centuries of Financial Folly
Robertson, D. H. (1956) Essays in Monetary Theory (London: Staples Press Limited).
Romer, C. D. (1992) ‘What Ended the Great Depression?’, Journal of Economic History, vol.
52, no. 4, pp. 757–784.
Sah, R. and J. E. Stiglitz (1992) Peasants versus City-dwellers: Taxation and the Burden of
Economic Development (Oxford: Clarendon Press).
Shiller, R. (2008) The Subprime Solution: How Today’s Global Financial Crisis Happened and
What to Do about It (Princeton: Princeton University Press).
Stiglitz, J. E. (1969) ‘Rural-urban Migration, Surplus Labor and the Relationship between
Urban and Rural Wages’, East African Economic Review, vol. 1–2, December, pp. 1–27.
Stiglitz, J. E. (1974) ‘Alternative Theories of Wage Determination and Unemployment
in LDCs: The Labor Turnover Model’, Quarterly Journal of Economics, vol. 88, no. 2,
pp. 194–227.
Stiglitz, J. E. (1988) ‘Money, Credit, and Business Fluctuations’, Economic Record, vol. 64,
no. 187, pp. 62–72.
Stiglitz, J. E. (1999) ‘Toward a General Theory of Wage and Price Rigidities and Economic
Fluctuations’, American Economic Review, vol. 89. no. 2, pp. 75–80.
Stiglitz, J. E. (2003) The Roaring Nineties (New York: W.W. Norton & Company).
Stiglitz, J. E. (2011) ‘Rethinking Macroeconomics: What Failed and How to Repair It’,
Journal of the European Economic Association, vol. 9, no. 4, pp. 591–645.
Stiglitz, J. E. and A. Weiss (1981) ‘Credit Rationing in Markets with Imperfect Information,
American Economic Review, vol. 71, no. 3, pp. 393–410.
Swanson, J. and S. Williamson (1972) ‘Estimates of National Product and Income for the
United States Economy, 1919–1941’, Explorations in Economic History, vol. 10, no. 1,
pp. 53–73.
Temin, P. (1991) Lessons from the Great Depression (Cambridge, MA: MIT Press).
Temin, P. (2010) ‘The Great Recession and the Great Depression’, NBER working paper no.
15645 (Washington, DC: NBER).
Timoshenko, V. P. (1933) World Agriculture and the Depression (Ann Arbor: University of
Michigan Press).
United Nations (2009) Report of the Commission of Experts of the President of the United
Nations General Assembly on Reforms of the International Monetary and Financial System,
also known as The Stiglitz Commission (New York: United Nations). Published as The
Stiglitz Report (New York: The New Press), 2010.
United States Bureau of the Census, 1975, Historical Statistics of the United States Colonial
Times to 1970 (Washington, DC: US Dept. of Commerce, Bureau of the Census).
Zandi, M. (2011) ‘To Shore Up the Recovery, Help Housing’, Special Report, Moody’s
Analytics, 25 May.
5
Capital Flows, Crises, and Externalities
Anton Korinek∗
University of Maryland, USA
1 Introduction
Emerging economies frequently experience episodes of large capital inflows. In

the mid-2000s for example, global financial markets were flush with liquid-
ity. Many emerging economies had better short-term growth prospects than
advanced countries and became an attractive destination for global investors.
Large capital inflows, or ‘capital flow bonanzas’ in the terminology of Reinhart
and Reinhart (2008), pushed up real exchange rates and inflated asset prices
in the countries affected. The ensuing rise in purchasing power and in the
value of domestic assets that could serve as collateral fueled a large increase
in indebtedness.
During the global financial crisis of 2008–09, the flows went into reverse as
global investors retrenched. And unsurprisingly, the countries that had expe-
rienced the largest inflows during the boom were hit hardest in the bust (IMF
2009). Massive deleveraging plunged country after country into crisis, leading
to severe downward pressure on exchange rates and asset prices. Output and
employment in the affected countries declined precipitously, in some instances
such as in Eastern Europe by up to 25 per cent, imposing massive social costs.
Figure 5.1 depicts the effects of large episodes of capital inflows on the prob-
ability of experiencing a financial crisis, based on calculations performed in
Korinek (2011b).1 The unconditional probability of a country experiencing a
crisis during the period was 5.8 per cent, as illustrated by the flat horizontal
line. Experiencing a capital inflow bonanza significantly raises the probability
of experiencing a crisis t years later, up to 8.3 per cent, as indicated by the solid
grey line and the dashed 95 per cent confidence intervals. A Granger causality
test shows that capital flow bonanzas Granger-cause financial crises at the 0.1
per cent significance level.
Policy-makers in emerging economies are justifiably worried when they expe-
rience large capital inflows. A growing chorus of academics, perhaps most
98
Capital Flows, Crises, and Externalities 99
12%
10%
8%
6%
4%
2%
0%
1 2 3 4 5 6 7 8 9
Figure 5.1 Conditional probability of crisis after capital flow bonanza
famously reflected in Stiglitz (2002), has argued that capital flows to emerg-
ing markets should therefore be regulated. Country after country, from Brazil
to Indonesia, Colombia, Korea, Peru, Taiwan, and Thailand, has followed their
advice since the financial crisis. In a notable reversal on earlier policies, the IMF
has given its blessing to capital controls under certain circumstances (see Ostry
et al. 2010). The traditional economic literature, as reflected for example in Fis-
cher (1998), argued that, based on the standard welfare theorems, free capital
flows ensured the most efficient global allocation of capital possible.
A growing body of literature, however, including Korinek (2009, 2010, 2011c)
and Jeanne and Korinek (2010a), make the welfare-theoretic case for regulating
capital flows as a form of Pigouvian taxation based on the notion that such flows
impose externalities on the recipient countries. Just as environmental pollu-
tion produces externalities that reduce societal wellbeing if unregulated, capital
inflows to emerging markets produce externalities that make such economies
more prone to financial instability and crises. By implication policy-makers can
achieve a Pareto-improvement by regulating and discouraging the use of risky
forms of external finance, in particular of foreign currency-denominated debts.
The economic rationale for such capital controls derives from the notion
that most modern financial crises involve private sector balance sheets. This
underlines the importance of a crucial category of market imperfections: when
international investors provide finance, they require that their loans are either
explicitly secured by collateral or implicitly by strong balance sheets of their
borrowers. However, the value of most of a country’s collateral and the health
of private sector balance sheets depend on exchange rates and asset prices: they
improve in good times when exchange rates appreciate; they deteriorate in
100 A. Korinek
bad times when exchange rates depreciate, but when access to finance is most
needed.
When an emerging economy is hit by a sufficiently strong adverse shock,
its exchange rate depreciates, the value of its domestic collateral declines, its
balance sheets deteriorate, and international investors become reluctant to roll
over their debts. The resulting capital outflows depreciate the exchange rate
even further and trigger an adverse feedback cycle of declining collateral values,
capital outflows, and falling exchange rates, as illustrated in Figure 5.2.
This gives rise to pecuniary externalities because each individual borrower
rationally takes market prices, such as the exchange rate, as given, but a planner
internalizes that changing the behavior of all agents will affect macroeconomic
aggregates and by implication market prices. In particular, inducing private
agents to take on less external finance and less risky forms of finance in good
times implies that they owe less in adverse states of nature and that the feedback
loop in Figure 5.2 is mitigated: exchange rates depreciate by less and balance
sheet constraints are loosened.
One interpretation of such regulation is that financial stability in the economy
is a public good, and that a planner who imposes prudential capital controls
induces agents in the private sector to internalize their effects on financial
stability. An alternative interpretation is that private agents face a prisoners’
dilemma – if they could all agree to use less external finance or less risky
financing instruments, the economy as a whole would become more stable and
everybody would be better off. This creates a natural role for policy intervention.
In a world where financial markets are complete and unconstrained, pecu-
niary externalities do not matter because the marginal rates of substitution
of all agents are equated and the wealth transfers that arise from changes in
Capital
outflows
Declining
collateral
Falling
exchange rates
Figure 5.2 Balance sheet crises and financial amplification in emerging economies
relative prices are Pareto-efficient – this was one of the fundamental insights of
the Arrow-Debreu framework. However, when an economy is subject to bind-
ing financial constraints, then pecuniary externalities do generally matter. If
prices move in a way that reallocates wealth from less constrained agents to
more constrained agents, a Pareto-improvement can be achieved. This is the
welfare-theoretic foundation of our results.
Our theory of externalities based on balance sheet effects also provides a clear
framework for how to determine the optimal magnitude of policy measures. The
reason why capital inflows expose an economy to financial fragility is that they
may reverse precisely when an economy is experiencing financial difficulty and
is subject to the described feedback loop.
Different forms of capital inflows result in different payoff characteristics in
the event of a crisis with different probabilities of future capital outflows, which
in turn leads to different externalities. Optimal capital controls should aim to
precisely offset these externalities.
If an emerging economy takes on dollar debts and subsequently experiences
a financial crisis, the exchange rate depreciates and the domestic value of the
debt increases sharply, implying that dollar debt imposes a large negative exter-
nality. CPI-indexed debt protects borrowers against the risk of exchange rate
fluctuations, imposing smaller externalities. Local currency debts and port-
folio investments play an insurance role, since the value of the local currency
and equity markets tend to go down during crises. Finally, non-financial foreign
direct investment often stays in the country when a financial crisis hits; in those
instances it does not impose any externalities.
More generally, optimal policy measures on capital inflows should be regularly
adjusted for changes in the financial vulnerability of the economy (see Jeanne
and Korinek 2010b). The externalities of foreign capital rise during booms when
leverage increases and financial imbalances build up. After a crisis has occurred
and economies have delevered, new capital inflows create smaller externali-
ties, justifying a zero tax in bad times when a country seeks to attract more
capital. Optimal capital flow regulation should therefore be strongly procycli-
cal. In a calibration to the case of Indonesia, we find that a tax on dollar
debt between 0 per cent and 30 per cent, with an average of 1.5 per cent, is
indicated.
The maturity structure of debt flows also plays a crucial role: international
creditors often refuse to roll over short-term debt when financial conditions
in an emerging economy deteriorate, creating a large risk of instability. On the
other hand, long-term loans cannot be recalled before their maturity date. Long-
term bonds that trade in secondary markets are somewhere in-between: they
can be sold by international investors in the event of a crisis, leading to capital
outflows and financial amplification. However, in such situations long-term
bond prices typically fall sharply, which gives them an equity-like characteristic
102 A. Korinek
and implies that the resulting capital outflows will be smaller than in the case
of short-term debt that is repatriated at par value.
The remainder of this chapter summarizes the findings of an active recent
literature on the externalities arising from balance sheet crises and on capital
controls to regulate them. We illustrate the basic arguments of this literature
in a simple analytic model based on Korinek (2010, 2011c) and discuss a range
of issues that arise when imposing capital controls. We conclude by pointing
toward future research directions.
2 An analytic illustration
2.1 Setup of benchmark model

The model that we develop in this section is based on Korinek (2010, 2011c) and
allows us to illustrate our arguments analytically. It captures a simple version
of balance sheet effects and financial amplification to illustrate the pecuniary
externalities that lead private agents to engage in excessive borrowing. In two
extensions, we will also exemplify how the same externality affects the riskiness
as well as the maturity of liabilities taken on by private agents.
Assume a small open endowment economy with three time periods t = 0, 1
and 2 that is inhabited by a representative agent. There is a tradable consump-
tion good in each period. In addition, we introduce a non-tradable good in
period 1 and denote its relative price by p, which constitutes a measure of the
real exchange rate. The utility function of the representative agent is
U = log(cTσ ,0 ) + log(c1 ) + cT ,2 where c1 = (cT ,1 )σ (cN,1 )1−σ (1)
The variables cT ,t and cN,t represent tradable and non-tradable consumption in

a given period t, and c1 is a consumption index that combines tradable and
non-tradable consumption in Cobb-Douglas fashion with expenditure shares σ
and 1 − σ .
The consumer obtains no endowment in period 0, endowments of tradable

and non-tradable goods yT ,1 , yN,1 in period 1, which we normalize to (σ , 1 − σ ),
and an endowment yT ,2 of tradable goods in period 2. For simplicity we assume
that there is no discounting and that the world gross interest rate is 1.
The only way to consume in period 0 is to borrow, which we denote by
d0 = cT ,0 . In period 1, the consumer chooses how much to consume in both
tradable and non-tradable goods, and how much debt d1 to carry into the fol-
lowing period. In the final period the consumer repays his debt and consumes
the remainder. Given the relative price of non-tradable goods p, the budget
constraints of the consumer are
cT ,0 = d0 (2)
cT ,1 + pcN,1 + d0 = yT ,1 + pyN,1 + d1 (3)

cT ,2 + d1 = yT ,2 (4)
Financial constraint
We capture the possibility of balance sheet effects and financial amplification
σ
by assuming that period 1 borrowing is constrained by a fraction κ < 1−σ of the
income of the representative agent,

d1 ≤ κ yT ,1 + pyN,1 (5)
A microfoundation for this constraint is that consumers may threaten to default

after taking on their debts and that they can renegotiate their debts at the end
of period 1. If they default, international lenders can seize at most a fraction
κ of the income of consumers in that period, convert all non-tradable goods
into tradable goods at the prevailing market price p, and repatriate what they
receive. If consumers have all the bargaining power, they could renegotiate their
debt down to the level indicated by the constraint, and lenders would never be
willing to lend more than this level. The amount yT ,1 + pyN,1 can be interpreted
as the international collateral of domestic borrowers. A decline in the exchange
rate p reduces the international collateral and by implication the borrowing
capacity of domestic consumers, which captures the notion of balance sheet
effects in our model.2
2.2 Model solution

We solve the model through backward induction. Assume first that the rep-
resentative consumer enters period 1 with an amount of tradable goods
m = yT ,1 − d0 , which captures the consumer’s endowment net of the debt d0 car-
ried into the period. We denote the utility of the consumer in the remaining two
periods as
σ 1−σ
V m; yT ,1 = max log cT ,1 cN,1 + cT ,2 s.t. (3), (4) and (5) (6)
where market clearing requires that cN,1 = yN,1 = 1 − σ for non-tradable goods
and cT ,1 = m + d1 for tradable goods. Assigning the shadow prices μ and λ to
the period 1 budget constraint and borrowing constraint of the consumer, his
first-order conditions are
σ
FOC cT ,1 : =μ
cT ,1
1−σ
FOC cN,1 : = μp
cN,1

FOC d1 : 1+λ = μ
104 A. Korinek
We combine the first two optimality conditions and impose market clearing for
non-tradable goods to obtain
cT ,1
p= (7)
σ
The real exchange rate is an increasing function of tradable consumption cT ,1 ,
which we can loosely interpret as a measure of aggregate demand in period 1
since non-tradable consumption is constant. As the consumer wants to spend
fixed shares of his consumption on tradable and non-tradable goods, any
increase in tradable consumption is matched by a parallel increase in the price
of non-tradable goods to keep the expenditure shares on the two goods constant
and ensure market-clearing.
Loose financial constraint

For sufficiently low levels of initial debt d0 the financial constraint on the repre-
sentative consumer will be loose, so λ = 0 and μ = 1. Then the consumer chooses
to consume cT ,1 = σ = yT ,1 and borrow d1 = d0 , and the exchange rate is p = 1.
This allocation satisfies the borrowing constraint if d0 ≤ κ. For unconstrained
levels of initial debt, the economy therefore achieves the first-best allocation
denoted by fb.
We substitute these allocations in equation (6) to express the utility of the
consumer in the first-best allocation as a function of the economy’s period 1
holdings of tradable goods m and yT ,1 .

V fb m; yT ,1 = v fb + m
for an appropriate constant v fb . The derivative of this function with respect to

m captures the marginal valuation of holding liquid tradable goods, to which
we will henceforth refer as the marginal value of liquidity
fb
Vm (·) = 1
Binding financial constraint

If d0 > κ, then the financial constraint on the consumer is binding. The
levels of borrowing and tradable consumption are then determined by the
binding constraint

d1 = κ yT ,1 + pyN,1 (8)
cT ,1 = m + d1 (9)
Equation (8) reflects the balance sheet effects of depreciations: a lower exchange
rate p reduces how much individual agents can borrow. Equation (9) captures
that lower borrowing d1 reduces the consumption of domestic agents when the
financial constraint is binding.
A constrained consumer with d0 > κ recognizes that his utility, given period
1 liquid tradable resources m = yT ,1 − d0 , is

V con m; yT ,1 = v con + σ log m + d1 − d1 (10)
for an appropriate constant v con . The consumer’s marginal value of liquidity

under binding constraints is
con σ
Vm (·) = (11)
cT ,1
fb
Since cT ,1 < σ under binding constraints, observe that Vm
con > V . The marginal
m
value of liquidity is higher (and, conversely, debt repayments are more costly)
when the constraint is binding than in the first-best allocation when the
constraint is loose.
In general equilibrium, the real exchange rate is given by p = cT ,1 /σ , which
implies
cT ,1
d1 = κ yT ,1 + yN,1 · (12)
σ
Solving the two equations (9) and (12) in d1 and cT ,1 , we obtain
σ yT ,1 + myN,1
d1 = κ ·
σ − κyN,1
m + κyT ,1
cT ,1 = σ · (13)
σ − κyN,1
Since d0 > κ when the constraint is binding, the fractions in both terms are
less than 1; therefore the constrained levels of borrowing d1 and consumption
cT ,1 are less than the unconstrained levels, which are given by d0 > κ and σ
respectively.
When the financial constraint is binding, our model exhibits financial ampli-
fication. Assume an exogenous change dm in the net liquid tradable resources of
the consumer in period 1. Under binding financial constraints, such an increase
not only allows for an increase in consumption by dm but also appreciates the
exchange rate by dm/σ , which in turn relaxes the financial constraint by dm·κ/σ ,
allows for a further increase in consumption by dm · κ/σ , and so forth. In total,
the response of consumption under binding constraints is
dcT ,1 κ
κ 2 σ
= 1+ + + ··· = >1
dm σ σ σ −κ
It is easily verified that the right-hand side of this equation can be obtained either
by calculating the sum of the geometric series that is listed in the equation,
or by taking the derivative of equation (13) with respect to m. Under bind-
ing constraints, an increase in m therefore leads to an amplified increase in
consumption.
106 A. Korinek
Solution of the period 0 problem

Using the function V (·), we formulate the consumer’s problem as

max σ log d0 + V yT ,1 − d0 ; yT ,1 (14)
and obtain the first-order condition

σ
= Vm (·)
cT ,0
Substituting the consumer’s marginal valuation of liquidity (11), we obtain that
the consumer’s optimality condition implies cT ,0 = cT ,1 .
If κ ≥ σ , then the period 1 Euler equation implies that consumers find it
optimal to consume and borrow
cT ,0 = cT ,1 = d0 = d1 = σ
The financial constraint is loose and the first-best equilibrium in the economy
can be implemented.
If κ < σ , then this allocation is not feasible and the decentralized equilib-
rium is characterized by a binding financial constraint. Following the period 0
optimality condition and equation (13), period 0 borrowing satisfies
(1 + κ)yT ,1 − d0
d0 = cT ,0 = cT ,1 = σ ·
σ − κyN,1
Solving explicitly for d0 and comparing the resulting equilibrium quantities we

find
σ (1 + κ)yT ,1
d1 < κ < d0 = cT ,0 = cT ,1 = <σ
2σ − κyN,1
2.3 The social planner’s solution

We now introduce a constrained social planner into our model and compare
the allocation she chooses with the decentralized equilibrium. We assume that
the planner optimizes the welfare of domestic consumers subject to the same
constraints as decentralized consumers, including the borrowing constraint, the
budget constraints and constraints on the market structure. (Since international
lenders are indifferent about their level of lending, maximizing the welfare of
domestic consumers amounts to maximizing global welfare.)
Naturally, the notion of a constrained social planner is different from a first-
best planner who is just bound by the economy’s resource constraints and can
ignore the borrowing constraints and budget constraints of individual agents.
But constrained planning problems better capture the situation of a regulatory
authority in the real world that has no other powers but to coordinate the actions
of individual agents. Establishing whether an economy is constrained efficient
is therefore the most basic test for whether there is a role for policy intervention.
In evaluating the borrowing constraint of consumers, the planner internalizes

that the exchange rate at which non-tradable collateral is valued is determined
by the marginal rate of substitution between tradable and non-tradable goods,
as given by the exchange rate equation (7).
Backward induction
As before we proceed by backward induction. We first solve for the period 1
and 2 equilibrium of the constrained planner and then determine the optimal
period 0 allocation and compare it to that of decentralized agents.
For any given level of debt d0 carried into period 1, the social planner and
decentralized agents choose the same allocations. If d0 ≤ κ so that the financial
constraint in the economy is loose, this is is easy to see since both the planner
and decentralized agents implement the first-best equilibrium.
If d0 > κ and the financial constraint in the economy is binding, the planner
is subject to the same constraint as decentralized agents and has no choice but
to borrow and consume the maximum possible in period 1, which is given by
equations (12) and (13).
However, even though their real allocations coincide, the two value liquidity
in period 1 differently if the financial constraint is binding. The reason why we
care about the valuation of liquidity in period 1 is that this variable is instru-
mental in determining the period 0 borrowing choices of decentralized agents
and the social planner. We formalize our finding as follows:
Lemma 1 (Valuation of Liquidity) If the period 0 debt level is sufficiently low

d0 ≤ κ, then the financial constraint in period 1 is loose and both the planner and
fb
decentralized agents perceive the marginal value of liquidity as Vm =1. Otherwise the
financial constraint in period 1 is binding and the planner values liquidity more highly
sp
than decentralized agents,Vm > Vm con .
Proof If d0 ≤ κ and the financial constraint is loose, the valuation of liquidity

is trivial.
If the financial constraint is binding, decentralized agents perceive the

marginal benefit of a unit of liquidity as Vm con = σ/c
T ,1 . In taking the deriva-
tive of the value function (10), consumers take the borrowing limit d1 as given
since all the variables on the right-hand side of equation (8) are exogenous for
price-takers. In general equilibrium, we can substitute for cT ,1 from equation
(13) to re-write this expression as
con σ − κyN,1
Vm (·) =
m + κyT ,1
By contrast, a constrained social planner internalizes that the exchange rate p

in (8) is endogenous to the aggregate period 1 tradable resources m = yT − d0 in
108 A. Korinek
the economy. She recognizes that the borrowing limit d1 is given by equation
(12) and the value function of consumers is
σ yT ,1 + myN,1
V sp m; yT ,1 = y sp + σ log m + κyT ,1 − κ ·
σ − κyN,1
for an appropriately chosen constant v sp . Taking the derivative of the value func-
tion with respect to m, we obtain the marginal benefit of liquidity as perceived
by the social planner
sp σ κyN,1
Vm (·) = −
m + κyT ,1 σ − κyN,1
Comparing the two marginal valuations of liquidity we find

sp con
Vm > Vm
(Recall that the constraint binds when d0 > κ and that we normalized yT ,1 +
yN,1 = 1.)
During episodes of financial amplification, decentralized agents only recog-
nize the private benefits of additional liquidity and take the tightness of the
financial constraint, as captured by Vm con , as given. A constrained social planner
coordinates the actions of decentralized agents and internalizes the social ben-
sp
efits of additional liquidity as captured by Vm . She recognizes that additional
liquidity m across the economy raises aggregate demand, which appreciates the
exchange rate and leads to positive financial amplification effects. We depict
the discrepancy between the private valuation of liquidity Vm con and the social
sp
valuation Vm in Figure 5.3.
One way of putting this result is that a healthy balance sheet, that is, holding
liquidity m when financial constraints are binding, is a public good. A planner
who internalizes this effect ensures the socially optimal provision of a public
good.
Solution of the period 0 problem

In solving for the period 0 problem, the social planner proceeds in the same
way as decentralized agents and solves the optimization problem expressed in
equation (14). He also obtains a first-order equation that follows along the same
lines,
σ
= Vm (·)
cT ,0
However, the different valuation of liquidity that we observed in lemma 1

implies that the planner chooses a different real allocation:
Proposition 1 (Excessive borrowing) If the financial constraint is relatively loose

( κ ≥ σ ), then a constrained social planner and decentralized agents can both implement
Marginal
valuation of
liquidity
Social valuation
Private valuation
Ex
te
rn
al
ity
Liquid net worth

Balance sheet effects
Figure 5.3 Private and social valuation of liquidity
the first-best equilibrium in the economy. If the financial constraint is relatively tight
( κ < σ ), a constrained social planner takes on less debt than decentralized agents.
fb
Proof If κ ≥ σ then the economy is unconstrained in period 1 so Vm = 1 and
d0 = σ . The allocations of decentralized agents and the planner coincide.
On the other hand, if κ < σ then the economy is constrained. It follows from
con < V sp . The period 0 Euler
lemma 1 that for any constrained level of debt, Vm m
sp
equation then implies that d0 > d0 . In other words, a social planner would
con
borrow less than private agents in period 0.
Graphical interpretation
Figure 5.4 illustrates the constrained social planner’s intervention graphically.
The left panel illustrates equilibrium in the period 0 market for debt in which
d0 is determined. The right panel depicts the period 1 market for debt in a
constrained equilibrium in which d1 is determined by the solid vertical lines.
In each panel, the horizontal axis captures the amount of debt, and the vertical
axis depicts the corresponding marginal rate of substitution between the current
and next period, that is, the price at which an agent would be willing to shift
a marginal unit of consumption between the two periods. We can interpret
the downward-sloping line representing the marginal rate of substitution of
the emerging market agent as the demand D for debt, and the flat horizontal
line representing the (constant) marginal rate of substitution of international
110 A. Korinek
MRS period 0/period 1 MRS period 1/period 2
1 S 1 S
Dcon D
Dsp
dsp
0
dcon
0
d0 dcon
1
dsp
1
d1
Figure 5.4 Constrained planner’s second-best intervention
lenders as the supply S of debt. The area between the two lines represents the
surplus of emerging-market consumers from borrowing. We denote variables in
the constrained decentralized equilibrium and in the social planner’s allocation
by the superscripts con and sp respectively.
Since we assumed that the financial constraint in period 1 is binding, observe
that each choice of debt d0 in period 0 determines a specific level of consump-
tion, the real exchange rate and the borrowing limit in period 1. However,
decentralized agents take the real exchange rate and therefore the period 1 bor-
rowing limit as given when they determine their period 0 borrowing – they
simply choose d0con such that their marginal rate of substitution equals that of
their lenders, which we assumed to be 1. They end up constrained at d1con in
period 1.
sp
A planner recognizes that marginally reducing period 0 borrowing to d0 cre-
ates a second-order welfare loss, illustrated by the shaded Harberger triangle
in the left panel of the figure. In the following period, lower debt d0 enables
higher consumption, pushes up the exchange rate and relaxes the borrowing
sp
limit in period 1 to d1 . This has a first-order benefit on consumer welfare, as
illustrated by the shaded trapezoid in the right panel of the figure.
Capital controls as Pigouvian taxation

The constrained planner’s equilibrium can be implemented by imposing a
Pigouvian tax τ on debt inflows that closes the wedge between the private valu-
ation Vmcon and the social valuation V sp of the cost of debt. If a tax on borrowing
m
d0 is imposed and the revenue is rebated lump sum, the Euler equation of private
consumers becomes σ (1 − τ )/d0 = Vm con (·). This condition replicates the planner’s
sp
constrained optimal intertemporal allocation σ/d0 = Vm (·) if the Pigouvian tax
is set such that
con
Vm
1−τ = sp (15)
Vm
sp con are evaluated at the planner’s allocation.
where the derivatives Vm and Vm
sp
Since Vm > Vm > 1, the tax is strictly positive but sufficiently small that it does
con
not discourage borrowing to the point where the constraint is loosened. In other
words, a planner would impose prudential capital controls so as to reduce the
magnitude of crises, but would not attempt to completely avoid them.
Excessive borrowing in a stochastic world

In our benchmark model we assumed that the world is deterministic and that
the tightness of the financial constraint depends on initial parameters. By doing
so we could capture the essence of the problem in the simplest possible way.
It is straightforward to extend our model of excessive borrowing such that
binding financial constraints occur on a probabilistic basis. One way of doing
so is to assume that the period 1 endowment yT ,1 is subject to a stochastic shock
λ then the constraint would be tighter the lower the endowment. Another way
of introducing uncertainty is to assume that the collateralizability parameter κ is
subject to ‘credit shocks’ that directly affect the borrowing capacity of emerging-
market agents. In both instances, it is easy to see that our findings on excessive
borrowing remain valid.
2.4 Risky financing decisions

Let us next extend our simple benchmark model to allow for different types
of capital flows. Since the marginal valuation of liquidity, that is, the pricing
kernel, of private agents is distorted so that they undervalue the social benefit of
liquidity in constrained crisis states, we find more generally that private agents
will choose a liability structure that leaves them excessively exposed to binding
constraints, even if they have access to state-contingent insurance instruments.
Emerging economies generally face a risk-return trade-off in their financing
decisions: instruments such as dollar debt are cheaper – they are available at
comparatively low interest rates – but they impose significant risk on borrowers
in case the exchange rate depreciates. On the other hand, financial instru-
ments that involve more risk-sharing, such as local currency debt, equity or
FDI, require a higher expected return that compensates international investors
for the additional risk.
Private agents choose their liability structure according to a private risk/return
trade-off, but fail to internalize that a risky private balance sheet also imposes
social costs. Put differently, they do not have proper incentives to take precau-
tions against financial amplification effects and buy too little crisis insurance
compared to a constrained planner.
112 A. Korinek
In our analytic model, we introduce two states of the world in period 1

and allow consumers to interact with risk-averse international investors and
make a state-contingent financing/insurance decision in period 0. Assume that
consumers value consumption according to the utility function
U = cT ,0 + E[u (c1 ) + cT ,2 ]
and that period 1 output can take on two realizations yT ,1 ∈ {y L , y H } with

probabilities p and 1 − p.
Consumers sell state-contingent securities d0L and d0H to international

investors who buy them at prices p (1 + ρ) and 1 − p each. In other words,
investors are paying the expected value for payoffs in the high state of nature,
but they are averse to the low state of nature and are willing to pay a premium
(1 + ρ) for payoffs in that state.
The maximization problem of domestic consumers is

max[p (1 + ρ) d0L + 1 − p d0H ] + Ei {V (y i − d0i ; y i )}
where the expectation is taken over i ∈ {L, H}. The resulting first-order condition
on d0L is
1 + ρ = Vm (mL ) (16)
Proposition 2 (Excessive risk-taking) A constrained social planner would com-

mit to smaller repayments d0L in the low state of nature than decentralized agents.
Proof According to equation (16), consumers choose a state-contingent repay-
ment that leaves them constrained in the low state of nature L, since the
cost of insurance against this state is greater than the marginal value of liq-
fb
uid resources in unconstrained states Vm = 1. This first-order condition pins
down a unique level of d0L since the value function V is strictly concave in that
region. By contrast, the first-order condition on repayments in the high state d0H
fb
is Vm mH = 1, which implies that the economy will be unconstrained in the
high state. Per lemma 1, the planner values liquidity more highly than decen-
tralized agents in the constrained state L, but the two value liquidity equally
in the unconstrained state H. By substituting the derivative of the planner’s
value function in the first-order condition (16), it can be seen that the planner
would promise lower repayments in the low state of nature, that is, the plan-
ner’s financing choices leave the economy less exposed to binding constraints
and financial amplification.
The portfolio decision of consumers can be interpreted as a risk/return trade-
off: they sell claims on the low state of nature to the point where they incur
binding constraints (‘risk’) because foreign investors are willing to buy such
claims at a higher price (‘return’). The planner perceives the cost of binding
constraints higher and will therefore sell fewer claims d0L . If the low state of
Table 5.1 Externalities imposed by different financial instruments in

Indonesia, 1997–98
Asset category Real gross Externality Optimal

return in 1998 tax
Dollar debt 218% 30.7% 1.54%

GDP-indexed dollar debt 190% 26.8% 1.34%
CPI-indexed rupiah debt 100% 14.1% 0.71%
Rupiah debt 63% 8.9% 0.44%
Stock market index 44% 6.2% 0.31%
nature materializes, the planner is responsible for smaller repayments, there

is less amplification, and consumption cT ,1 declines less severely than in the
decentralized equilibrium. In this sense a planner takes on less risk in her
financing decisions or buys more insurance against adverse states of nature than
decentralized agents.
The planner’s equilibrium can be implemented via Pigouvian taxes on the
payoffs of securities that pay in the low state of nature d0L . Consumers and the
planner value payoffs in the unconstrained high state d0H equally.
We can view real-world securities as different combinations of state-

contingent payoffs d0L , d0H . For example, foreign-currency-denominated debt
would correspond to a pair d0L > d0H – consumers who borrow in foreign cur-
rency have to repay more in low states of nature when a country’s exchange
rate depreciates. CPI-indexed local currency debt (real debt) would correspond
to d0L = d0H . On the other hand, non-indexed local currency debt would corre-
spond to d0L < d0H since it entails lower repayments in low states than in high
states of nature. FDI may be viewed as a contract in which d0L ≈ 0 since profits
are only repatriated in good times. The different weights on payoffs in the high
state and on externality-rife payoffs in the low state is what is responsible for
the pecking order of externalities in Table 5.1.
2.5 Short-term debt

Another manifestation of excessive risk-taking is that decentralized agents take
on too much short-term debt. Long-term debt insures emerging economies
against rollover risk during systemic crises, that is, against the risk that interest
rates rise or credit is rationed precisely when the country most needs funding.
During such rollover crises, financial amplification effects occur and give rise to
externalities. Individual market participants do not internalize this and take on
too little long-term debt as insurance against rollover risk.
We introduce a long-term bond into our benchmark model and assume that
domestic agents finance their period 0 consumption by issuing short-term debt
114 A. Korinek
d0 at gross interest rate 1 and long-term debt d0LT to be repaid in period 2 at an

interest rate 1 + ξ , where ξ > 0 reflects an exogenous term premium
cT ,0 = d0 + d0LT
For simplicity, we assume that the consumer’s long-term debt does not affect
the renegotiation problem at time 1 since it does not need to be rolled over. It
follows that the consume’s optimization problem is

max σ log(d0 + d0LT ) + V yT ,1 − d0 ; yT ,1 − (1 + ξ ) d0LT
The first-order conditions on d0 and d0LT imply

σ
= Vm (·) = 1 + ξ
cT ,0
Proposition 3 (Excessive short-term debt) A constrained planner would take on

less short-term debt d0 and more long-term debt d0LT than decentralized agents.
Proof The second equality pins down a unique level of short-term debt d0
since the value function V is strictly concave in the constrained region where
Vm > 1. Lemma 1 implies that the social planner would contract a smaller level
of d0 than decentralized agents. The first equality then pins down period 0 con-
sumption and, via the period 0 budget constraint, the level of long-term debt.
Since period 0 consumption cT ,0 is identical in the allocations of the decentral-
ized equilibrium and the planner, a smaller level of short-term debt implies that
the planner takes on a higher level of long-term debt than decentralized agents.
The planner’s equilibrium can again be implemented by imposing a Pigouvian

tax τ on short-term debt as given by equation (15).
3 Quantifying the externalities of capital flows
Korinek (2010) develops a sufficient statistics approach to estimate the magni-

tude of externalities imposed by different forms of financial instruments during
the 1997–98 crisis in Indonesia. He finds that a marginal outflow of one dollar
during the crisis imposed an externality of 14 cents on other borrowers. The first
column of Table 5.1 reports the gross return of different financial instruments.
The second column multiplies this by 14 per cent to obtain the marginal exter-
nality of each type of instrument under the assumption of a one-year maturity.
The third column determines the optimal tax rate if crises occur on average
every 20 years.
In the table, different forms of capital flows are ranked according to a peck-
ing order of decreasing externalities: dollar debt is one of the most dangerous
forms of finance, since the local currency typically depreciates during crises,
which inflates the value of dollar liabilities just when domestic agents are least
able to service their debt. The real gross return on dollar debt is reported as
218 per cent. CPI-indexed debt contracts or rupiah debt impose considerably
smaller externalities as they avoid such adverse valuation effects. Investments
in the stock market allow for a considerable degree of risk-sharing with foreign-
ers, which reduces the externalities even more. However, they are still associated
with externalities, since international investors often sell stocks during financial
crises, which leads to capital outflows and pressure on the exchange rate.
These theoretical predictions about the riskiness of different forms of finance
closely mirror the empirical findings on the effects of different forms of financial
liabilities on stability and growth (see, for example, Mauro et al. 2007).
Optimal policy measures on capital inflows should also be regularly adjusted
for changes in the financial vulnerability of the economy (see Jeanne and
Korinek 2010b). The externalities of foreign capital rise during booms when
leverage increases and financial imbalances build up. After a crisis has occurred
and economies have delevered, new capital inflows create smaller external-
ities, justifying a zero tax in bad times when a country seeks to attract
more capital. Optimal capital flow regulation should therefore be strongly
procyclical.
4 Conclusion
Building on a growing recent literature (Korinek 2010, 2011c), this chapter has
argued that there are externalities associated with balance sheet crises in emerg-
ing economies and has developed a simple model of a small open emerging
economy to illustrate the point. Furthermore, we have discussed that these
externalities can be readily calibrated and may justify capital controls of the
order of magnitude observed in the real world. However, there are a number of
questions on which further research is warranted. Without being exhaustive, let
us list a number of important challenges.
If we extend our focus beyond small open economies, capital controls have
spillover effects on other countries. As we show in our ongoing research (Korinek
2011a), they are still desirable from a global welfare perspective, but there may
be scope for policy coordination if such controls create distortions that can be
lessened by international cooperation.
Secondly, prudential capital controls are closely related to macroprudential
regulation. As we discussed in this chapter, capital controls may be the first
instrument of choice when policy-makers are concerned about balance sheet
effects arising from exchange rate volatility. On the other hand, macropruden-
tial regulation of debt, which does not discriminate based on the residency of
creditors, may be the optimal instrument to mitigate booms and busts in asset
prices that lead to balance sheet effects (see, for example, Jeanne and Korinek
2010b).
116 A. Korinek
Finally, like every form of regulation, capital controls create incentives for
circumvention. An important research agenda is to study how best to impose
robust controls that are effective in offsetting externalities while minimizing the
distortions arising from attempts at circumvention.
Notes
* The author would like to thank Julien Bengui, Olivier Jeanne, Nobuhiro Kiyotaki, Mar-
cus Miller, Carmen Reinhart and Joseph Stiglitz as well as participants of the 2011
IEA Meetings for helpful comments and suggestions. For contact information visit
www.korinek.com
1. The figure is based on IMF IFS data from 1980 to 2009. An episode of large capital inflows
is defined as a realization of the current account in its top quintile, as in Reinhart and
Reinhart (2008). Financial crises capture all currency crises according to the definition
of Frankel and Rose (1996) and banking crises according to the definition of Reinhart
and Rogoff (2009).
2. See Korinek (2011c) for an extensive discussion of alternative assumptions that would
lead to financial amplification effects that are similar to the ones discussed in our
framework.
References
Fischer, S. (1998) ‘Capital Account Liberalization and the Role of the IMF’, in S. Fis-
cher (ed.), Should the IMF Pursue Capital-Account Convertibility?, International Finance
Section, Department of Economics, Princeton University.
Frankel, J. A. and A. K. Rose (1996) ‘Currency Crashes in Emerging Markets: An Empirical
Treatment’, Journal of International Economics, vol. 41, no. 3, pp. 351–366.
IMF (2009) ‘Global Financial Stability Report: Responding to the Financial Crisis and
Measuring Systemic Risks’, technical report (Washington, DC: International Monetary
Fund).
Jeanne, O. and A. Korinek (2010a) ‘Excessive Volatility in Capital Flows: A Pigouvian
Taxation Approach’, American Economic Review, vol. 100, no. 2, pp. 403–407.
Jeanne, O. and A. Korinek (2010b) ‘Managing Credit Booms and Busts: A Pigouvian
Taxation Approach’, NBER Working Paper no. 16377 (Washington, DC: NBER).
Korinek, A. (2009) ‘Excessive Dollar Borrowing in Emerging Markets: Balance Sheet Effects
and Macroeconomic Externalities’, mimeo, University of Maryland.
Korinek, A. (2010) ‘Regulating Capital Flows to Emerging Markets: An Externality View’,
mimeo, University of Maryland.
Korinek, A. (2011a) ‘Capital Controls and Currency Wars’, mimeo, University of Maryland.
Korinek, A. (2011b) ‘Hot Money and Serial Financial Crises’, IMF Economic Review, vol. 59,
no. 2, pp. 306–339.
Korinek, A. (2011c) ‘The New Economics of Prudential Capital Controls: A Research
Agenda’, IMF Economic Review, vol. 59, no. 3, pp. 523–561.
Mauro, P., J. D. Ostry, G. Dell’Ariccia, J. di Giovanni, J. A. Faria, A. Kose, M. Schindler
and M. E. Terrones (2007) ‘Reaping the Benefits of Financial Globalization’, IMF
staff discussion paper (Washington, DC: International Monetary Fund), available at:
www.imf.org/external/np/res/docs/2007/0607.pdf
Ostry, J. D., A. R. Ghosh, K. Habermeier, M. Chamon, M. S. Qureshi and D. B.

Reinhardt (2010) ‘Capital Inflows: The Role of Controls’, IMF Staff Position Note, 10/04
(Washington, DC: International Monetary Fund).
Reinhart, C. M. and V. R. Reinhart (2008) ‘Capital Flow Bonanzas: An Encompassing View
of the Past and Present’, NBER Working Paper no. 14321 (Washington, DC: NBER).
Reinhart, C. M. and K. S. Rogoff (2009) This Time Is Different: Eight Centuries of Financial
Folly (Princeton: Princeton University Press).
Stiglitz, J. E. (2002) Globalization and Its Discontents (New York and London: W.W. Norton).
6
Liquidity Shocks and Asset Prices
in the Business Cycle
Shouyong Shi∗
University of Toronto, Canada
1 Introduction
Liquidity in the asset market has attracted increasing attention in public debate
and academic research, partly because of the recession in 2008 and 2009 in
the United States. At the onset of that recession, liquidity suddenly dried up
in the asset market. In particular, the market for collateralized debt obligations
almost shut down as major financial institutions either had or were perceived
to have insufficient funds to meet their contractual obligations. To prevent a
complete collapse of the financial market, the US government injected a large
amount of liquid assets into the market through various lending facilities which
ranged from short-term lending to outright purchases of private equity by the
government.
The large shortfall in liquidity in the 2008–2009 recession was likely caused by
the deterioration of economic fundamentals and, in particular, by the realiza-
tion that asset-backed securities had much higher default risks than previously
thought. However, the episode raises the general concern that a shortfall in
liquidity may affect the performance of the economy even if it is not caused by
changes in economic fundamentals. Non-fundamental events (such as sunspots)
may induce panics in the market and result in a shortage of liquidity. In this
chapter I treat the changes in asset market liquidity as exogenous shocks and
examine their importance for the business cycle. More specifically, I ask: Are the
effects of liquidity shocks on macro variables consistent with the ones observed
in the business cycle in the US? This question is clearly relevant for policy inter-
vention regarding liquidity because such intervention is justified only if the
answer to the question is affirmative.
For concreteness, I evaluate a hypothesis formulated by Kiyotaki and Moore
(2008) who argue that changes in the liquidity in the equity market play an
important role in the business cycle. The hypothesis places two equity-market
118
Liquidity Shocks and Asset Prices in the Business Cycle 119
frictions at the center. One is that a firm can issue equity on at most a fraction
θ ∈ (0, 1) of investment. The other friction is that only a fraction φ ∈ (0, 1) of the
existing equity can be sold in any given period. Kiyotaki and Moore (2008, the
KM model henceforth) interpret unanticipated changes in φ as liquidity shocks
in the equity market. They conjecture that a negative shock to equity liquid-
ity causes equity price to fall, which reduces firms’ ability to finance investment
through the equity market. As a result, investment falls, output falls and a reces-
sion starts. Similarly, an unanticipated increase in φ raises equity price, relaxes
the financing constraint and induces an economic boom.
In Shi (2011), I have constructed a tractable macro model to incorporate the
two frictions above. This model provided straightforward aggregation and led
to a natural definition of the recursive competitive equilibrium. After calibrat-
ing the model to the US data and computing the equilibrium, I have illustrated
that a large and persistent negative liquidity shock can generate large and per-
sistent reductions in aggregate investment, employment, and output. However,
contrary to the conjecture in the KM model, these large reductions in macro
quantities are not associated with a reduction in equity price. On the contrary,
a negative liquidity shock generates an equity price boom. I have explained intu-
itively why this counterfactual response of equity price arises in the model and
demonstrated its robustness to a wide range of extensions and modifications
of the model. The counterfactual response of equity price to liquidity shocks
indicates that liquidity shocks in the equity market are not the primary cause
of the business cycle. I have discussed some resolutions to the problem, all of
which require a negative liquidity shock to be accompanied by direct or indirect
reductions in productivity.
The current chapter is a partial summary of Shi (2011). Here, I focus on the
steady state and only informally describe the results on dynamics. The details
of the model, the calibration, the dynamic analysis, and the robustness checks
are relegated to the companion paper.
Financial frictions have been the focus of business cycle research for quite
some time (see Williamson 1987; Bernanke and Gertler 1989). The main char-
acter in the play in this literature is net worth of entrepreneurs and/or financial
intermediaries that determines the amount of borrowing. As net worth changes
procyclically over the business cycle, it generates the financial multiplier. In con-
trast to this literature, which focuses on debt financing, the KM model focuses
on the frictions in the equity market and abstracts from debt finance. So do I in
the main analysis in this chapter. I will explain later that adding debt finance is
unlikely to change the central result of the model. At the end of subsection 3.3,
I will discuss how this chapter is related to some recent papers that also explore
the KM model.
120 S. Shi
2 A macro model with equity financing
The economy is populated by a continuum of households, with measure one.

Each household has a unit measure of members who are identical at the begin-
ning of each period and whose role during the period is determined by a shock.
With probability π ∈ (0, 1), a member will be an entrepreneur and, with prob-
ability 1 − π, the member will be a worker in the period. These shocks are iid
across the members and time. All members follow the household’s instructions
that maximize the household’s intertemporal utility:
∞

E0 β t πu(cte ) + (1 − π)[U (ctw ) − h(t )] , where β ∈ (0, 1).
t=0
The variable c is consumption and a worker’s labor supply, where the super-
script e indicates an entrepreneur and the superscript w a worker. The functions
(u, U , h) are assumed to have standard properties. The expectation is taken over
aggregate shocks (A, φ) which will be described below. Lumping a large number
of members into a representative household simplifies aggregation (see Shi 1997
for a similar construct).
Consider an arbitrary period t, drop the index t, and indicate period t ± j with
the subscript ±j. At the beginning of the period, aggregate shocks (A, φ) are
realized and all members of a household are identical. The household pools the
assets, divides the assets among the members and gives the members instruc-
tions on what to do in the period. In this stage, the household holds (physical)
capital k, equity claims s, and liquid assets b. Capital resides in the household
and will be rented later to firms to produce consumption goods. Equity claims
are a diversified portfolio of shares of the household’s own capital and other
households’ capital. Liquid assets include government bonds. Because the mem-
bers are identical in this stage, the household divides all assets evenly among
the members. The household also gives each member the instructions on the
choices in the period, which are contingent on whether the member will be an
entrepreneur or a worker. For an entrepreneur, the household instructs him to
consume an amount c e , invest i, and hold a portfolio of equity and liquid assets
e ) at the end of the period. For a worker, the household instructs him
(se+1 , b+1
to consume an amount c w , supply labor , and hold a portfolio (sw +1 , b+1 ) at
w
the end of the period. After receiving these instructions, the members go to the
market and will remain separated from each other until the beginning of the
next period.
The next stage in the period is production. In this stage, each member receives
a shock whose realization determines whether the individual is an entrepreneur
or a worker in the period. If a member is an entrepreneur, he has no labor
endowment in the period but has an investment project that can transform any
amount i ≥ 0 of consumption goods into i units of capital. If a member is a
worker, he is endowed with one unit of labor from which he supplies units to
a perfectly competitive sector that produces consumption goods. The firms in
this sector rent capital from the households and hire labor to produce consump-
tion goods according to y = A F(kd , d ), where the superscript d indicates the
demand for productive factors and F has constant returns to scale. Total factor
productivity A follows a Markov process. Individuals receive their income from
the firms producing consumption goods. After production, a fraction (1 − σ ) of
existing capital depreciates, where σ ∈ (0, 1).
The third stage in the period is investment. The markets for assets and goods
are open. An entrepreneur undertakes investment, subject to the financing
constraint. In the final stage of the period, a worker consumes c w and holds
a portfolio (sw +1 , b+1 ), while an entrepreneur consumes c and holds a port-
w e
folio (s+1 , b+1 ). Then, individuals return to their households, arriving at the
e e
beginning of the next period.

An entrepreneur must rely on his own resource to finance investment and
other expenditures, because he is separated from other members of the house-
hold during the period. An entrepreneur needs to invest i, consume c e , and pay
a lump-sum tax τ . The following resources are available to him. The first is the
return to equity claims, rs, where r is the rental rate of capital. The second is the
receipt from selling liquid assets. The entrepreneur holds the amount b of liquid
assets, which mature in the period and can be redeemed for a value b. Suppose
that the entrepreneur will hold an amount b+1 e of new liquid assets at the end
of the period, which can be zero, and let pb be the price of new liquid assets.
Then, the net receipt to the entrepreneur from liquid assets is (b − pb b+1 e ). The
third type of resource is the receipt from selling new and existing equity. An
entrepreneur holds s shares of equity when entering the investment stage of the
period. After obtaining the return-to-equity claims, a fraction (1 − σ ) of capital
depreciates and so an entrepreneur’s holdings of existing equity are σ s. In addi-
tion, there are i shares of new equity corresponding to new investment, which
are either retained by the entrepreneur’s household or sold to other households.
Let q be the market price of a share of equity.1
Given that the entrepreneur will hold se+1 shares of equity at the end of the
period, the receipt from selling equity in the period is q(i + σ s − se+1 ). Thus, an
entrepreneur’s resource constraint is:
rs + q(i + σ s − se+1 ) + (b − pb b+1

e
) − τ ≥ i + ce . (2.1)
The amount of investment is constrained by financial frictions. In particu-

lar, the equity market has two liquidity frictions. The first is an equity-issuing
bound: an entrepreneur can issue new equity on at most a fraction θ ∈ (0, 1)
of investment. The second is a limit on equity resaleability: an individual can
re-sell at most a fraction φ ∈ (0, 1) of the existing equity in his portfolio. The
KM paper explains why these constraints are important features of the equity
122 S. Shi
market. Although it is interesting to endogenize θ and φ, they take θ and φ as

exogenous elements of the model. So do I here. Specifically, I fix θ and assume
that φ follows a Markov process. Shocks to φ are interpreted as shocks to equity
liquidity. With these two frictions in the equity market, an entrepreneur must
hold onto (1 − θ )i shares of new equity and (1 − φ)σ s shares of existing equity.
Thus, the two frictions in the equity market amount to the following constraint
on an entrepreneur’s end-of-period holdings of shares:
se+1 ≥ (1 − θ )i + (1 − φ)σ s. (2.2)
For this constraint to be effective, an entrepreneur must face a tight borrow-

ing limit; otherwise, the entrepreneur can simply borrow enough to finance
e ≥ 0.
investment. I set this borrowing limit to be zero; that is, b+1
e
Let me consolidate (2.1) with (2.2) by eliminating s+1 . Doing so yields
e
(r + φσ q)s + (b − pb b+1 ) − τ ≥ c e + (1 − θ q)i. (2.3)
This constraint reveals that the effective cost of a unit of investment to an

entrepreneur is 1 − θ q, because the entrepreneur can issue θ shares of new equity
per investment to raise the fund θq. Thus, the amount (1−θq) can be interpreted
as the ‘downpayment’ on the investment. This downpayment must be financed
by the return to capital and the receipts from selling assets. Moreover, because
the marginal utility of consumption is strictly positive, (2.1) holds with equal-
ity. Thus, the liquidity constraint (2.1) is binding if and only if the consolidated
constraint (2.3) is binding. For this reason, I refer to (2.3) as the equity liquidity
constraint.
Similarly, a worker faces a resource constraint in the period analogous to
(1), except that the worker has labor income and no investment project. This
constraint is:
rs + w + q(σ s − sw w w
+1 ) + (b − pb b+1 ) − τ ≥ c . (2.4)
A worker also faces the constraint on borrowing, b+1w ≥ 0, and on asset holdings,
s+1 ≥ (1 − φ)σ s, but these constraints are not binding in the equilibrium.
w
The government supplies a fixed amount of liquid assets per household, B ≥ 0,

and spends a fixed amount of consumption goods per household, g ≥ 0. The
lump-sum tax τ is collected to satisfy the government budget constraint in each
period: g = τ + (pb − 1)B.
Now let me formulate a household’s decisions as a dynamic programming
problem. At the beginning of a period, the aggregate state of the economy is
(K, Z), where K is the capital stock per household and Z = (A, φ) is the realiza-
tions of exogenous shocks to total factor productivity and equity resaleability. I
omit the amount of equity per household from the list of aggregate state vari-
ables because it is equal to the aggregate capital stock. Also, I omit B from this
list because it is a constant. A household’s individual state variables consist

of equity claims, s, and liquid assets, b. Denote the household’s value func-
e ) for an
tion as v(s, b; K, Z). In the period, the household chooses (i, c e , se+1 , b+1
entrepreneur and (, c , s+1 , b+1 ) for a worker. Rather than using (c , sw
w w w w
+1 , b+1 )
w
as part of the household’s choices, I use the corresponding average quantity per
member as the choices, in addition to the choices for each entrepreneur. For this
purpose, denote average consumption per member as c = πc e + (1 − π)c w and
the average holdings of the two assets per member as s+1 = πse+1 + (1 − π)sw +1
and b+1 = π b+1 e + (1 − π)bw , respectively. Also, I combine a worker’s resource
+1
constraint, (2.4), with an entrepreneur’s resource constraint, (2.1), to obtain the
following resource constraint on the household:
(r + σ q)s − qs+1 + (1 − π)w + (q − 1)π i + (b − pb b+1 ) − τ ≥ c. (2.5)
The household’s choices (i, c e , se+1 , b+1

e , , c, s , b ) solve:
+1 +1

v(s, b; K, Z) = max πu(c e ) + (1 − π) U (c w ) − h() + βEv(s+1 , b+1 ; K+1 , Z+1 )
(2.6)
subject to (2.3), (2.5), and the following constraints:
i ≥ 0,c e ≥ 0,se+1 ≥ 0,b+1

e
≥ 0, (2.7)
c e ≤ c/π ,se+1 ≤ s+1 /π,b+1
e
≤ b+1 /π. (2.8)
The expectation in the objective function is taken over next period’s aggregate
state (K+1 , Z+1 ). I have already derived (2.3) and (2.5) above. The non-negativity
constraints in (2.7) ensure that an entrepreneur cannot borrow. The constraints
in (2.8) come from similar non-negativity constraints on a worker. In the above
problem, I have suppressed the dependence of asset prices, (q, pb ), and factor
prices, (w, r), on the aggregate state (K, Z). The household’s optimal choice
e , , c, s , b ) can be expressed as a policy function x(s, b; K, Z).
x ∈ (i, c e , se+1 , b+1 +1 +1
With dynamic programming, it is natural to define an equilibrium in this
economy as a recursive equilibrium. A recursive competitive equilibrium consists
of asset and factor price functions (q, pb , r, w)(K, Z), a household’s policy func-
e , , c, s , b )(s, b; K, Z), the value function v(s, b; K, Z), the
tions (i, c e , se+1 , b+1 +1 +1
demand for factors by final-goods producers (kd , d ), and the law of motion of
the aggregate capital stock that meet the following requirements: i) a house-
hold’s value and policy functions solve a household’s optimization problem in

(2.6); ii) factor demands satisfy the optimality conditions, r = AF1 (kd , d ) and

w = AF2 (kd , d ), where the subscripts indicate partial derivatives; iii) prices clear
the markets:
goods: c(s, d; K, Z) + π i(s, d; K, Z) + g = AF(kd , d ), (2.9)
labor : d = (1 − π)(s, d; K, Z), (2.10)

124 S. Shi
capital : kd = K = s, (2.11)
liquid assets: b+1 (s, b; K, Z) = b = B, (2.12)
equity : s+1 (s, b; K, Z) = σ s + πi(s, b; K, Z); (2.13)
and iv) the dynamics of K are consistent with the aggregation of (2.13):
K+1 = σ K + πi(K, B; K, Z). (2.14)
Most of the requirements in the above definition are self-explanatory. The

equality K = s in (2.11) comes from the facts that all households are identical
and they own all capital in the form of equity. The equalities s = K and b = B
are used in (2.14). An equilibrium can be determined numerically by iterating
on the asset price functions q(K, Z) and pb (K, Z) (see Shi 2011).
3 Equilibrium effects of a liquidity shock
In this section, I derive the optimality conditions of a household’s choices and

analyze the effects of equity liquidity on the equilibrium, both in the long run
and the short run.
3.1 Optimality conditions

Let λe π U (c w ) be the Lagrangian multiplier of the equity liquidity constraint
e )
(2.3) in the household’s decision problem. The optimal choices of (i, se+1 , b+1
yield the following conditions:
q − 1 ≤ (1 − θ q)λe and i ≥ 0, (3.15)

e
λ [se+1 − (1 − θ )i − φσ s] = λe b+1
e
= 0. (3.16)
The two inequalities in (3.15) hold with complementary slackness. Intuitively,

optimal investment is positive if the cost (in terms of utility) of the down-

payment on each unit of investment, (1 − θ q)λe U (c w ), is equal to the benefit
w
of issuing equity on investment, (q − 1)U (c ). The two equalities in (3.16)
require that, if the equity liquidity constraint binds, then an entrepreneur’s
holdings of assets at the end of the period should be at the minimum, which is
se+1 = (1 − θ )i + φσ s for equity claims and b+1
e = 0 for liquid assets.

The optimal satisfies the standard condition: h ()/U (c w ) = w. The optimal

c satisfies: u (c e ) = U (c w ) 1 + λe , which states intuitively that a marginal unit
e
of the resource is more valuable to an entrepreneur than to a worker if and only if

an entrepreneur’s liquidity constraint is binding. Moreover, the optimality con-
ditions for (s+1 , b+1 ) and the envelope conditions for (s, b) imply the following
pricing equations for the assets:

w
U (c+1 ) e

q = βE r+1 + σ q+1 + πλ+1 r+1 + φ+1 σ q+1 , (3.17)
U (c w )
w
U (c+1 ) e

pb = βE 1 + πλ +1 . (3.18)
U (c w )
Note that equity market frictions affect the pricing equations directly through
φ+1 and λe+1 .
3.2 Effects of equity liquidity in the steady state

A non-stochastic steady state is one in which the exogenous state is constant
at Z = Z ∗ = (A∗ , φ ∗ ), and all endogenous variables are constant over time, as
indicated by the superscript ∗ . Let me examine the steady-state effect of an unan-
ticipated permanent reduction in equity liquidity, φ ∗ . Intuitively, this reduction
in liquidity tightens the equity liquidity constraint, thus increasing the shadow
price λe∗ . It is also intuitive that the tightened constraint on liquidity reduces
long-run investment and output. Although this intuitive result indicates that
the frictions in the equity market can be important for determining macro vari-
ables, not all the effects are as expected. The purpose of this subsection is to
illustrate that, contrary to what one may expect, a reduction in the equity
market liquidity is likely to create a long-run equity price boom.
To illustrate this puzzling result, note that investment in the steady state is
strictly positive because it is equal to i∗ = (1 − σ )K ∗ /π > 0. By (3.15), the shadow
price of the equity liquidity constraint in the steady state is
q∗ − 1
λe∗ = . (3.19)
1 − θ q∗
The asset-pricing equations, (3.17) and (3.18), yield the following steady-state
relations:
(β −1 − σ )q∗ − r ∗
λe∗ = , (3.20)
π(r ∗ + φ ∗ σ q∗ )
pb∗ = β(1 + π λe∗ ). (3.21)
Let me focus on the interesting case where the equity liquidity constraint is
binding in the steady state, i.e., λe∗ > 0. By (3.19), this is equivalent to requiring
q∗ ∈ (1, 1/θ ).
It is instructive to first examine how (λe∗ , q∗ , pb∗ ) respond to φ ∗ in the partial
equilibrium where r ∗ is fixed. Given r ∗ , (3.19) and (3.20) solve for (λe∗ , q∗ ). Let
Liquidity(q∗ ) denote the right-hand side of (3.19), and Pricing(q∗ ) the right-hand
side of (3.20). Figure 6.1 depicts these two functions and labels the solution
to (3.19) and (3.20) by point E.2 Figure 6.1 also depicts the partial-equilibrium
126 S. Shi
λe
E’
E
Pricing (q)
Liquidity (q)
1 1/θ q
Figure 6.1 Steady-state effect of reducing φ ∗ : partial equilibrium
effect of a reduction in equity liquidity, φ ∗ . Because (3.19) does not depend on

φ ∗ , a reduction in φ ∗ does not change the curve Liquidity(q) in Figure 6.1. In
contrast, a reduction in φ ∗ increases Pricing(q) for all q, and so it shifts up the
curve Pricing(q) in Figure 6.1. As a result, (λe∗ , q∗ ) increase from point E to point

E in Figure 6.1. By (3.21), pb∗ also increases with the reduction in φ ∗ .
Therefore, a reduction in equity liquidity creates an asset price boom for any
fixed r ∗ , which is opposite to the conjecture in the KM model and opposite
to what is observed in the US business cycle. To explain this seemingly puz-
zling result, note that when equity becomes less liquid, the financing constraint
on new investment becomes binding more severely. In this sense, the implicit
cost of new investment increases. For investment to be optimal, the price of
equity must rise to increase the amount of resource raised by each issue of equity
and to reduce the downpayment on new investment. The price of liquid assets
also rises because the liquidity provided by liquid assets is more valuable in
this case.
This partial equilibrium effect of φ ∗ is likely to carry over to the general equi-
librium in the steady state. In the general equilibrium, the rental rate of capital
responds to the reduction in equity liquidity. Note that the function Liquidity(q)
does not depend on r ∗ and that the function Pricing(q) decreases in r ∗ for all q.
It is intuitive that a reduction in φ ∗ increases the rental rate by reducing invest-
ment and the capital stock in the steady state. This increase in r ∗ attenuates the
upward shift of the curve Pricing(q) in Figure 6.1. However, to reverse the upward
shift in Pricing(q), the increase in r ∗ would have to be so large that the equity
liquidity constraint becomes less binding than before. Since this is unlikely to
occur, the reduction in φ ∗ will increase (λe∗ , q∗ , pb∗ ) as well as r ∗ in the new steady
state.
3.3 Dynamic effects of a liquidity shock

The analysis of the steady state is instructive, but it may not necessarily reflect
the effects of a liquidity shock on the transitional path to the steady state. To
find the dynamic effect, I have calibrated the model in Shi (2011) and assumed
that φ follows a first-order auto-regressive process. After computing the dynamic
equilibrium, I have obtained two main findings. First, a large and persistent
negative shock to equity liquidity can generate large and persistent reductions
in investment, employment and output. This is intuitive, because equity is an
important source of financing investment in the model. Second, the negative
shock generates a large and persistent asset price boom along the transition path
to the steady state.
To explain why a negative liquidity shock generates an asset price boom on
the transitional path, let me focus on the case i > 0 and write the corresponding
optimality condition of investment, (3.15), as follows:
q − 1 = (1 − θ q)λe . (3.8)
There are only two endogenous variables in this equation, (q, λe ). The marginal
benefit of investment, given by (q − 1), is a strictly increasing function of q. The
downpayment on investment, (1 − θ q), is a strictly decreasing function of q.
When there is a negative shock to liquidity, the implicit cost of raising funds to
finance the downpayment of investment increases. That is, the equity liquidity
constraint (2.3) becomes tighter and its shadow price λe increases. The higher
λe reduces the net marginal benefit of investment for any given equity price. To
restore the balance between the marginal benefit and cost of investment, equity
price must increase. As liquid resources become more scarce, the price of liquid
assets, pb , also increases.
This argument is quite general, because it only requires the negative liquidity
shock to tighten the liquidity constraint, which is what the shock is supposed
to do. With this generality, the argument can survive a wide range of exten-
sions/modifications of the model and the liquidity shock. For example, if θ
falls concurrently with φ, then the downpayment on each unit of investment
increases for any given equity price, in which case the equity price has to increase
even further in order to restore the optimality condition (3.8). In Shi (2011),
I have discussed how introducing nominal wage/price rigidity and habit per-
sistence in consumption can exacerbate the problem by making equity price
increase further. I have also explained why introducing adjustment costs in
investment is unlikely to resolve the problem.
Introducing debt finance into the model is also unlikely to change the qual-
itative response of equity price to a liquidity shock. To see why, suppose that
a fraction d of each unit of investment is financed by debt. In this case, an
entrepreneur needs to use the cash flow to finance (1 − d − θq) fraction of
128 S. Shi
each unit of investment. The condition for optimal investment is modified as

q − 1 = (1 − d − θq)λe . Using microdata, Covas and den Haan (2011) have docu-
mented that debt financing as a ratio of output is procyclical. This means that
d falls during recessions and rises during boom times. Supposing that liquidity
shocks are an important cause of the business cycles, then a negative liquid-
ity shock raises λe and reduces d at the same time. Again, equity price must
rise in order to restore the optimality condition for investment. In fact, adding
debt financing strengthens the positive response of equity price to a negative
liquidity shock when debt finance is procyclical.
For equity price to fall after a negative liquidity shock, as is typically seen
in recessions, the equity liquidity constraint must become less tight. To gen-
erate this paradoxical outcome, there must be other concurrent shocks that
sufficiently reduce the need for investment in the economy. Some candidates
of such concurrent shocks are reductions in the investment opportunity π and
reductions in total factor productivity A. In Shi (2011), I have computed the
dynamic response of the equilibrium to a combination of negative shocks to φ
and A. The results indicate that equity price can indeed fall after the shocks, as
investment, employment and output do.
Other authors have explored the KM model and some have independently
discovered the puzzling response of equity price to liquidity shocks. Calibrating
variations of the KM model to the US data, Nezafat and Slavik (2010) show that
a negative shock to θ increases equity price, and Ajello (2010) shows that a neg-
ative shock to φ increases equity price. However, these authors do not focus on
the puzzling response of equity price. Instead, Nezafat and Slavik (2010) focus
on the importance of shocks to θ in explaining the volatility of asset prices, and
Ajello (2010) on the importance of shocks to the intermediation cost in explain-
ing the volatility of investment and output. Another closely related paper is that
by Del Negro et al. (2011), who quantitatively evaluate the non-standard mone-
tary policy intervention in the 2008–2009 recession. Moreover, Del Negro et al.
(2011) are able to generate a negative response of equity price to a negative liq-
uidity shock, by introducing an interaction between the monetary policy rule
and the expectation on the zero lower bound on the nominal interest rate. Ajello
(2010) finds that an unexpected increase in the intermediation cost can induce
equity price to fall, because a shock to intermediation cost affects not only liq-
uidity but also the quality distribution of new investment projects undertaken
in the equilibrium. Both Ajello (2010) and Del Negro et al. (2011) incorporate
a range of elements into the KM model, such as wage/price rigidity, adjustment
costs in investment, and habit persistence in consumption. These elements are
intended to be realistic for addressing the issues in the two papers, but they cloud
the picture of how liquidity shocks affect equity price. I simplify the KM model
rather than complicate it. The simplified model is easy for aggregation and for
exploring the robustness of the response of equity price to liquidity shocks.3
4 Conclusion
Firms rely partly on the equity market to finance investment. If the equity mar-
ket is frictional, then aggregate investment and output depend on the liquidity
of the equity market. A popular hypothesis is that liquidity shocks in the equity
market are an important cause of the business cycle. To evaluate this hypoth-
esis, I have constructed a tractable macro model to incorporate the frictions
that reduce firms’ ability to issue new equity and to sell existing equity. I have
formulated the recursive equilibrium of this economy and analyzed how the
equilibrium responds to liquidity shocks in the equity market. A main result is
that a negative liquidity shock generates an equity price boom, which is opposite
to what is observed in recessions. This response of equity price occurs as long as
a negative liquidity shock tightens firms’ financing constraints on investment.
The counterfactual response of equity price to liquidity shocks indicates that
shocks to the asset market liquidity alone are not the primary driving force of
the business cycle.
A recommendation of this analysis is that policy should not be designed just
to pump liquidity into the asset market whenever there is a fall in liquidity
during a recession; instead, policy-makers should first find the causes of the
shortfall in liquidity during a recession and then see whether there is need to
use policy to correct these causes. If a shortfall in liquidity is driven by purely
non-fundamental events, then supplying liquidity seems a good policy. If a
shortfall in liquidity is generated by a deterioration in the quality of investment,
supplying liquidity to the market does not seem a good idea because it acts as
a subsidy to low-quality investment. If a shortfall in liquidity is generated by
an increase in the intermediation cost, then the corrective policy should be to
subsidize intermediation rather than inject liquidity directly into the market.
On the theoretical side, it is important to explicitly model why asset market
liquidity fluctuates and how it interacts with productivity. The tractable model
in this chapter seems useful for these endeavors.
Notes
* This paper is a partial summary of the paper (Shi 2011) that was presented at the
Canadian Economic Association meeting (Ottawa, 2011), the International Economic
Association meeting (Beijing, 2011), the Canon Institute for Global Studies (Tokyo,
2011), the Asian Meeting of the Econometric Society (Seoul, 2011), and the Chicago
Federal Reserve Bank conference on money, banking and payments (Chicago, 2011). I
am grateful to Nobu Kiyotaki for many conversations on the topic and to Andrea Ajello
for comments. Li Li provided excellent research assistance. I gratefully acknowledge
financial support from the Canada Research Chair, the Bank of Canada Fellowship and
the Social Sciences and Humanities Research Council of Canada. The view expressed
here is my own and does not reflect the view of the Bank of Canada.
130 S. Shi
1. As in the KM model, I assume that claims on the household’s own capital and other
households’ capital have the same liquidity. This assumption simplifies the analysis
because it implies that the two subsets of claims have the same price.
2. For any given r ∗ > 0, there exists a unique solution to (3.19) and (3.20) that satisfies
q∗ ∈ (1, 1/θ ) if and only if r ∗ < β −1 − σ . I maintain this condition here.
3. Del Negro et al. (2011) have revised their paper to adopt the construct of large
households from my model to simplify aggregation.
References
Ajello, A. (2010) ‘Financial Intermediation, Investment Dynamics and Business Cycle’,

Job market paper, Northwestern University.
Bernanke, B. S. and M. Gertler (1989) ‘Agency Costs, Net Worth, and Business Fluctua-
tions’, American Economic Review, vol. 79, no. 1, pp. 14–31.
Covas, F. and W. den Haan (2011) ‘The Cyclical Behavior of Debt and Equity Finance’,
Del Negro, M., G. Eggertsson, A. Ferrero and N. Kiyotaki (2011) ‘The Great Escape? A Quan-
titative Evaluation of the Fed’s Non-Standard Policies’, Manuscript, Federal Reserve
Bank of New York.
Kiyotaki, N. and J. Moore (2008) ‘Liquidity, Business Cycles, and Monetary Policy’,
manuscript, Princeton University.
Nezafat, P. and C. Slavik (2010) ‘Asset Prices and Business Cycles with Finan-
cial Frictions’, Social Science Research Network working paper no 1571754, at:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1571754
Shi, S. (1997) ‘A Divisible Search Model of Fiat Money’, 1997, Econometrica, vol. 65,
no. 1, pp. 75–102.
Shi, S. (2011) ‘Liquidity, Assets and Business Cycles’, manuscript, University of Toronto.
Williamson, S. (1987) ‘Financial Intermediation, Business Failures, and Real Business
Cycles’, Journal of Political Economy, vol. 95, no. 6, pp. 1196–1216.
7
Whither Capitalism? Financial
Externalities and Crisis
Marcus Miller∗
Lei Zhang
University of Warwick, UK
Introduction
The Pareto-efficiency of competitive economic equilibrium is, of course, a cen-

tral feature of the Arrow-Debreu paradigm. But in 1986 two papers appeared
concerning the welfare inefficiency of competitive equilibria. Geanakoplos and
Polemarchakis showed that ‘missing markets’ implied the possibility of Pareto-
improving interventions; while Greenwald and Stiglitz demonstrated that
missing markets and asymmetric information implied that competitive market
prices could generate ‘pecuniary externalities’ – with market prices generat-
ing side-effects conceptually similar to technological externalities (such as the
productive interactions of Silicon Valley or the negative effects of industrial
pollution).
Historical events were soon to lead to a much greater reliance on market
forces worldwide, however. The break-up of the USSR signaled the collapse of
the Communist challenge to market-oriented models of economic development
and encouraged a shift from managed capitalism to market capitalism – with
the administrations of Prime Minister Thatcher and President Reagan in the
vanguard. In the UK, for instance, there was widespread privatization of nation-
alized industries and the stock of public housing; and, in financial services, the
Big Bang of 1986 signaled the opening up of the City of London to the forces of
global competition. Likewise in the USA – where government intervention was
increasingly seen as ‘part of the problem not the solution’ – there was pressure to
deregulate financial services; and the success of Mr Greenspan in handling the
US stock market break on Black Monday in 1987 encouraged the belief that self-
regulation plus adept interest-rate management could head off financial crisis.
With New York and London being world-leading financial centres competing
for global business, this triggered a regulatory ‘race to the bottom’ – with reliance
on self-regulation on one side of the Atlantic matched by ‘light touch’ regulation
on the other.1 Over time this led to the repeal of the Glass-Steagall legislation
131
132 M. Miller and L. Zhang
in the USA, and to such a rapid expansion of UK banking that at the peak it was
host to an industry with a balance sheet more than five times local GDP!
What then of the early warnings from 1986? They seemed to be largely
forgotten. Even when severe financial crisis erupted in East Asia in 1997–98,
this was widely seen as a symptom of nascent capitalism – of poorly regulated
banks, connected lending and excessive foreign currency exposure – to be solved
by upgrading financial regulation to the exemplary standards of the leading
economies in the West. The IMF did put to one side its plans for increased
deregulation of the capital account;2 but faith in the efficacy of lightly regulated
markets in advanced economies was largely unshaken. Indeed the assumption
of financial market efficiency was to become the hallmark of macroeconomic
models used by central banks to steer the economy in the time of Great Moder-
ation. Even when markets departed from fundamentals, as in the US ‘high-tech’
bubble which characterized the early years of the 21st century, interest rate pol-
icy on its own seemed adequate for handling the consequences of the asset price
correction (Greenspan 2002).
But the financial crisis in North Atlantic economies in 2007–08 – and the
threat it posed of collapse for the Western financial system and a possible repeat
of the Great Depression – has forced a reconsideration of the consensus, with
Alan Greenspan himself acknowledging that his faith in the efficiency of market
forces had been misplaced.
Do financial crises provide concrete examples of financial externalities in
action? A recent empirical study by Majnoni and Powell (2011) using quar-
terly data for 139 corporate issues from the period 1999 to 2006 suggests that –
at least for emerging markets – they have. They test the hypothesis that corpo-
rate spreads will normally be determined by firm, country, and international
financial characteristics; but in addition they will rise at times of crisis due to
endogenous risk or amplification effects. Their empirical results show an ampli-
fication of shocks during crisis times depending on the size of the credit market
before the crisis. For banking crises in particular, the weakness of the banking
system amplifies shocks by increasing the cost of capital for non-financial firms.3
What is the nature of these externalities? How can the rules and structure of
banking world-wide be reshaped to limit them? How likely are these reforms
to be effected? These are the issues to be explored in this chapter. In the first
section we look at fire-sale externalities and the under-provision of liquidity; in
the second section we look at the risk-shifting due to limited liability; finally in
section three at the risk of contagion posed by the network feature of banking.
In conclusion, we note how vividly the shock to the Western economies –
now mired in recession with the prospect of years of slow growth to come –
contrasts with the success of managed capitalism of India and China both in
Whither Capitalism? Financial Externalities and Crisis 133
avoiding these crises and in maintaining enviable rates of economic growth. The
capacity of an economic system to limit pecuniary externalities may, it seems,
be an important determinant of capitalist development.
1 Fire-sales and the under-provision of liquidity

1.1 The ‘financial accelerator’ as pecuniary externality
Even without financial intermediaries, a credit-constrained market economy –
where collateral is used to handle repudiation risk – can exhibit liquidity crises
and collapsing asset prices. In the model of Kiyotaki and Moore (1997), for
example, productive small business entrepreneurs wish to raise outside finance
to acquire fixed capital assets but face an agency problem because the ‘human
capital’ used in the business is inalienable. Recourse is had to the issuance of
debt backed by physical collateral, priced to reflect its productivity outside the
entrepreneurial sector (that is, in the hands of ‘deep pocket’ lenders). In the face
of uncorrelated, idiosyncratic productivity shocks, agents adversely affected can
sell capital and pay down debt without affecting asset prices. But in the face of
an adverse macroeconomic shock to entrepreneurial productivity, the borrowing
constraint can lead to ‘fire-sales’ which affect the price of the collateral and
trigger yet further sales; that is, there is a pecuniary externality.
This is in sharp contrast to the ‘first best’ economy where all agents are
unconstrained in the credit market, and prices and production are unaffected
by net worth. How this externality can impact on the allocation of fixed
assets in the model of Kiyotaki and Moore (1997) can be seen schematically in
Figure 7.1.
From equilibrium at E, where debt-financed small business (SB) entrepreneurs
hold a stock k∗ of fixed assets at price q∗ , the immediate impact of an adverse
productivity shock is indicated by the ‘initial condition’ labelled DD – a schedule
for their disposal of fixed assets, k, as needed to match the fall in net worth due
to a one-period drop in productivity. This schedule can be interpreted as an
unexpected need for liquidity on their part (Krishnamurthy 2010). From this
perspective, asset prices have to fall until, at point X, the balance-sheet-driven
‘demand for liquidity’ by small businesses (measured to the left from k∗ to DD)
is matched by the ‘supply of liquidity’ by the residual buyers who have no
balance sheet problems (the agents with ‘deep pockets’) whose take-up of assets
is measured from k* to the schedule labelled SS.
As the figure shows, the impact on asset holding has two components. The
distance EA indicates how far small businesses need to contract their holdings
at a constant price, q*, as they dispose of assets to reduce their borrowing in line
with the fall in net worth;4 the second component, AX, indicates the need for
Asset Price
qt D' Initial
B'
conditions
SC
Bursting asset bubble D
B S
A
q*
θ E
qx
X
S
D' Temporary productivity shock

D
Insolvency Solvency SB Asset Holdings

kc k* kt
Figure 7.1 The financial accelerator as pecuniary externality
Source: Miller and Stiglitz (2010).
further disposals due to the adverse net worth effects of asset prices falling in
the face of concerted selling by small businesses to residual buyers with declin-
ing marginal productivity – net worth effects that are exacerbated by expected
persistence. In the absence of fresh shocks, the system will gradually return to
equilibrium along the stable path5 SS. Thus the pecuniary externality acts as a
‘financial accelerator’ that takes short-run equilibrium from A to point X on SS.
Like Gai et al. (2008), Korinek (2011) modifies this framework so that the
borrowing is done by financial intermediaries, risk-neutral bankers who raise
finance from households and invest in risky projects; and he shows how the
externality involved can be thought of in terms of their undervaluation of liq-
uidity. Banks who think that in adverse conditions they can sell assets fail to
realize that with correlated shocks these sales will help push prices down. A social
planner would anticipate the fall and take on less risk, as Korinek explains:
A planner internalizes the fact that a decline in asset prices leads to financial
amplification since it reduces the amount of liquidity that bankers can raise
from their sales of each unit of the assets. This pecuniary externality reduces
the efficiency of the distribution of capital. By contrast, decentralized bankers
take asset prices as given since they realize that the behaviour of an atomistic
agent has only an infinitesimal effect on asset prices.
Central bankers and regulators have not generally been acting like social plan-
ners, it seems. According to Majnoni and Powell (2011) ‘policy makers in the
developed world (albeit with notable exceptions) allowed financial institutions
to push leverage up to unprecedented limits under a shared optimism regarding
the capacity of capital markets to supply an almost infinite amount of liquidity’.
The difference between the private valuation and the planner’s social valua-
tion of liquidity, as shown in Figure 7.2, is defined as the pecuniary externality
(which falls to zero in unconstrained states). For social efficiency, Korinek pro-
poses a state-contingent, proportional tax on risk-taking that brings the private
cost in line with the social cost. This is a metaphor for macro-prudential regu-
lation as ‘it closely captures what BIS defines as the macro-prudential approach
to regulation: it is designed to limit system-wide financial distress that stems
from the correlated exposure of financial institutions and to avoid the resulting
real losses in the economy’ (2011: 26). He also proposes taxation on complex
securities such as CDS swaps ‘which [are] likely to require large payouts precisely
in times of financial turmoil’ (ibid.: 27).
While these policy measures are expressed in terms of taxes on externalities,
Korinek argues that they are broadly equivalent to capital adequacy require-
ments ‘which have tax-like effects because bank capital is costly’ (ibid.: 28).
Such capital requirements could be reduced if CDS swaps can be arranged ‘that
Valuation of
liquidity μω
Social valuation
Private valuation
ex
te
rn
al
ity
Productivity shock Aω
1
binding financing
constraints
Figure 7.2 Private and social valuation of liquidity
Source: Korinek (2011).

shift systemic risk to agents outside the financial system who are not subject to
financial constraint’. (A deal between US banks and Chinese sovereign wealth
funds to deliver liquidity in the crisis, for example?)
This welfare perspective seems to match that of Stiglitz (2010) when he argues
that:
The financial sector has imposed huge externalities on the rest of society.
America’s financial industry polluted the world with toxic mortgages, and,
in line with the well established ‘polluter pays’ principle, taxes should be
imposed on it. Besides, well-designed taxes on the financial sector might help
alleviate problems caused by excessive leverage and banks that are too big to
fail. Taxes on speculative activity might encourage banks to focus greater
attention on performing their key societal role of providing credit.
1.2 Risk of insolvency especially after collapse of an asset bubble

In discussing the amplification of shocks through balance sheets and asset prices,
it is customary to assume that the ‘overshooting’ will not be severe enough to
render the illiquid agents insolvent (Kiyotaki and Moore 1997; Krishnamurthy
2010; Korinek 2011). This is a rashly optimistic assumption in the aftermath
of an asset bubble – especially if highly levered institutions are involved – as
is suggested in Figure 7.1 where the disposal schedule D’D’ associated with an
asset bubble fails to intersect SS to the right of SC, the Solvency Constraint.
In Japan, for example, Koo (2008: 14–15) reports that deleveraging made
many firms technically insolvent after the bubble burst. In the recent North
Atlantic crisis, Lehman Brothers famously went bankrupt and Fanny Mae and
Freddie Mac were nationalized soon afterwards – as were two mortgage banks
in the UK. In both countries rescue plans for the financial sector also included
substantial injections of equity capital for financial institutions.6
How the existence of limited liability may induce firms to run the risk of
insolvency – and how this can lead to an asset bubble – is the subject of the next
section.
2 Risk-shifting and under-provision of capital
In Casino Capitalism, Hans Werner Sinn notes how the limited liability corpora-
tion was and is crucial for the mobilization of savings to fund risky investment,
as the limitation of liability is needed to convince the small shareholder to par-
ticipate. While this corporate form may be ‘capitalism’s secret of success’, it can
be misused by taking on massive leverage, which generates negative externalities
in the form of excessive risk-taking whose downside is borne by the creditors.
The case of US investment banks is cited as a case in point:
Investment banks, until well into the 1970s, were all organized as partner-
ships, and as such offered their market partners the unlimited private liability
of their owners. But they evolved eventually into corporations in order to

limit their liability to their equity capital. [. . .] For the five big American
investment banks in 2006, before the financial crisis erupted, business vol-
ume was leveraged by a factor of 22 to 33 of what would have been possible
if only equity had been lent. (Sinn 2010: 76–77)
If a business volume of 100 units of money is only backed by three to four-and

a-half units of equity, it can easily happen that in times of crises the losses
eat up the equity and lead to bankruptcy. Even worse was the fact that the
low level of equity combined with limited liability induced the stockhold-
ers to demand ever riskier business models in order to increase their profits.
[. . .] If some of the losses are borne by the creditors, it pays to take the risk.
(ibid.: 79)
It is the negative externality that bankers impose on their creditors and possi-
bly the taxpayer that induces excessive risk-taking, producing private profits
and social losses. (ibid.: 81)
This same logic is explored in a piece written about emerging markets by Allen
and Gale (2000) (see also Allen and Gale 2007, Chapter 9). But they go further
and show how the incentive for highly levered institutions to shift risk on to
their creditors can have the effect of raising asset prices above the fundamen-
tal level,7 that is, create a ‘bubble’, adding further to the vulnerability of the
financial system.
In this section, risk-shifting incentives are presented in the context of a simple
two bloc global model, designed to show how financial externalities, weak reg-
ulation and missing markets can expose the global economy to the risk of crisis.
First we discuss how limited liability and weak regulation of financial interme-
diaries can lead to an asset price bubble in the Home country; then we look at
how the precautionary motive in the Foreign country (which saves because of
an absence of insurance) can further inflate the asset price in the Home country.
2.1 Limited liability and excessive leverage

To formalize the argument we adapt the Allen and Gale (2000) model to study
asset pricing in the Home country (which, in the light of Sinn’s comments, could
be thought of as the US). Assume there are two periods and there are two assets
available in the Home country: a safe asset with variable supply and a risky one
in fixed supply of 1. The gross return on the safe asset in period 2 is R (R > 1) and
that on the risky asset is RH with probability γ and RL with probability 1 − γ .
The investment in the risky asset has to be done through financial interme-
diation (for example, risk-neutral banks) which is perfectly competitive. In the
absence of adverse incentives in the intermediation sector, the equilibrium price
of the risky asset in period 1 must satisfy:
RH RL
γ + (1 − γ ) =R (1)
PF PF
where PF represents the fundamental price of the risky asset, that is,
γ RH + (1 − γ )RL
PF = (2)
R
For each unit investment in the risky asset, an intermediary is required to finance
a fraction k by issuing equity and borrows the rest from the market at a cost of
R. (So 1/k indicates the leverage of the intermediary.) Let k be set by a regu-
latory authority, where a low value of k indicates weak regulation of financial
intermediaries. Assume specifically that
γ (RH − RL )
k≤ (3)
PF
that is, k is set too low to prevent risk-shifting behavior on the part of the
financial intermediaries.
If all intermediaries are protected by limited liability, then perfect competition
implies

RH
γ − R + (1 − γ )(−k) = 0 (4)
P
where P indicates equilibrium price for the risky asset with financial intermedi-
ation. For simplicity, we assume that the cost of the intermediary’s own capital
is R. So, the first term on the left hand side of (4) represents the payoff to the
intermediary in the good state, and the second the payoff in the bad state. Note
that, given (3), the realization of the bad state implies that the debt will not be
paid in full since RL /P ≤ (1 − k)R. In this case, the liability will be taken over by
an insuring agency, and the intermediary will be closed down and lose its own
capital, k.
Solving for P in (4) yields
γ RH
P= (5)
γ R + (1 − γ )k
and together with (3) this implies
P ≥ PF
that is, weak regulation leads to asset price bubbles.

Let τ (R, k) = P/PF represent the relative size of the bubble, then from (2) and
(5), one finds
τR ≡ ∂τ (R, k)/∂R < 0 (6)
and
τk ≡ ∂τ (R, k)/∂k < 0 (7)

that is, a higher interest rate R reduces the demand for the risky asset, damp-
ening its price rise; while weaker regulation (lower k) increases intermediaries’
incentive to shift risk and so pushes up the price of the risky asset.
As long as households are not aware of the risk-shifting incentives that exist
in the financial intermediaries, they will treat the bubble as if it is an increase
in their real wealth, as shown in Laibson and Mollerstrom (2010). In the next
section, we look at the global impact of this agency problem.
2.2 Missing insurance markets and precautionary saving

Let the global exchange economy consist of two countries – Home and Foreign –
and last two periods. Only non-state-contingent assets can be traded between
Home and Foreign. Each country is populated by a continuum of ex-ante identi-
cal consumers with preferences over consumption in periods 1 and 2, C1 and C2
respectively (with * indicates variables for the Foreign country), given by stan-
dard additively separable utility function which has a Constant Relative Risk
Aversion.
Assume there is no uncertainty in the first period and both countries are
endowed with 1 unit of the single tradable commodity. To reflect a higher inci-
dence of macroeconomic shocks in the Foreign country, let two possible states
of nature exist in the Foreign country, in period 2, with probability of 1 − π and
π respectively, where π is the probability of a low state. Specifically we assume
the state-dependent aggregate growth rate of its endowments is ĝ in the high
state and ĝ − in the low state, giving Foreign’s period 2 endowment as 1 + ĝ
and 1 + ĝ − respectively.
Assume also that consumers in the Foreign country are identical ex-ante but,
in period 2, they are divided into two groups. The first group, with a measure
of 1 − λ, is unaffected by the aggregate shock and enjoys the same endowment
in either state. The second group, with a measure of λ, suffers the full brunt of
the aggregate shock in the low state, so their endowment is disproportionally
reduced to 1 + ĝ − /λ in the low state.8 This formulation is based on Mankiw’s
(1986) model of the concentration of aggregate shocks, and is intended to cap-
ture the lack of risk-pooling among residents of Foreign country due to the
absence of a social safety net or the presence of private information and moral
hazard considerations.
Given the real global gross interest rate of R, the resulting allocation of con-
sumption in the Foreign country implies its period 1 consumption C1∗ (R, λ) has
the properties that: ∂C1∗ (R, λ)/∂R < 0 and ∂C1∗ (R, λ)/∂λ > 0. The first property is
because a rise in R, through both income and substitution effects, decreases
period 1 consumption. The second is because higher risk concentration leads
to higher precautionary saving in the Foreign country, reducing its period 1
consumption, as discussed in Miller et al. (2011).
Let the period 1 endowment of a consumer in the Home country be 1, so

period 2 resources will consist in an endowment of 1 + g and the returns on
the investment he/she made through the financial intermediary in period 1.
If consumers in the Home country are not aware of the incentive problem in
financial intermediaries, they would treat an increase in P as an increase in their
wealth. So the ex-ante wealth for the Home consumers is
WH = 1 + (1 + g)R−1 + ξ(τ )
where ξ (τ ) > 0 positive captures the wealth effect of the bubble and C1 (R, τ )
defined earlier measures’ over-valuation.
For a given real global gross interest rate R, the optimal allocation of con-
sumption in the Home country implies period 1 consumption C1 (R, τ ) has the
properties: ∂C1 (R, τ )/∂R < 0 and ∂C1 (R, τ )/∂τ > 0. The reason for the first prop-
erty is the same as that for the Foreign’s period 1 consumption. The second
property simply reflects the fact that asset price bubbles have a positive wealth
effect.
To complete the model, we introduce the market-clearing condition for period
1 to determine the equilibrium global interest rate R:
C1∗ (R, λ) + C1 (R, τ ) = 2 (8)
Using the properties of C1 (R, τ ) and C1∗ (R, λ) above, the equilibrium condition
(8) implies the following comparative statics for the global interest rate:
∂R ∂C1∗ (R, λ)/∂λ

=− ∗ >0 (9)
∂λ ∂C1 (R, λ)/∂R + ∂C1 (R, τ )/∂R + τR ∂C1 (R, τ )/∂τ
∂R τk ∂C1 (R, τ )/∂τ
=− ∗ <0 (10)
∂k ∂C1 (R, λ)/∂R + ∂C1 (R, τ )/∂R + τR ∂C1 (R, τ )/∂τ
How excessive leverage and risk concentration can affect the pattern of con-
sumption and global interest rates is illustrated with the aid of Figure 7.3, where
the horizontal axis measures period 1 consumption and the vertical the global
real interest rates. (Note that Home’s consumption is measured from point OH
and that of the Foreign country measured from OF .) In the absence of an asset
price bubble (τ = 1) and the risk concentration (λ = 1), the equilibrium is at point
A where the two demand schedules of the Home and the Foreign intersect. Note
that allocation at point A is only ‘constrained efficient’ because the global asset
markets are incomplete: full efficiency would require Arrow securities or GDP
linked bonds.
Limited liability, weak regulation and excessive leverage leading to an asset
bubble shift Home’s demand schedule up to C1 (R, τ > 1) reflecting the increased
wealth perceived by Home consumers, as shown in Laibson and Mollerstrom
(2010). The presence of risk concentration, on the other hand, shifts Foreign’s
C1*(R, λ = 1)
C1(R, τ > 1)
C1(R, τ = 1)
A B
C1(R, λ < 1)
OH OF
Figure 7.3 Risk concentration, excessive leverage, and global imbalances
demand schedule down to C1∗ (R, λ > 1) due to precautionary savings. Global
market-clearing equilibrium with risk concentration and excessive leverage is
given by point B where the two revised demand schedules intersect. With the
appropriate choice of τ and λ, substantial global imbalances will emerge with
little changes in real interest rates.
By assumption, both externalities and missing markets play a role in defin-
ing this equilibrium – a situation of excessive, bubble-driven consumption in
one country and high precautionary savings in the other. Welfare-improving
policy interventions would involve tightening regulation in the Home country
(increasing k) and providing a social safety net in the Foreign country (increas-
ing λ). With λ = 1 and k above the critical value shown in (3) above, equilibrium
would be at A. (Welfare improvement from A would require completing the
global asset market, for example by the issuance of GDP bonds.)
3 Network externalities, contagion, and circuit-breakers
Liquidity and solvency problems have been studied in previous sections without
considering the pattern of interconnectedness between agents – by assuming,
so to speak, a representative bank. This could be defended from a reductionist
perspective – why look at structure unless you have to? It has been found how-
ever that industry structure is key for contagion: research at the Bank of England
and the FRBNY using stochastic network theory shows that different structures
can lead to very different propagation mechanisms.
This is what we study in this section, beginning with a discussion of the spe-
cial nature of banks – how it arises from asymmetric information and missing
markets; and how it leads to institutional arrangements that call for a structural
analysis. This is followed by a simplified model of the banking industry where
risk-pooling encourages individual banks to consolidate into banking groups,
but the risk of contagion inside any group sets a limit to efficient group size. If
the activities that generate contagion can be hived off outside the banking indus-
try, however, group size can expand indefinitely. This could be interpreted as an
argument for a Glass-Steagall type of separation of commercial and investment
banking. It may also provide a rationale for the partial separation recommended
by the UK Vickers Commission where investment and commercial banking
activities can remain within the same corporate entity but are separated by a
‘ring-fence’.
3.1 Liquidity provision

Taking the Diamond and Dybvig (1983) model as a benchmark, the problem
facing agents wishing to invest long-term subject to random liquidity needs
is imperfect markets for liquidity insurance. The tension between long-term
investment and need for cash can be reduced by trading in the non-contingent
spot markets – what Allen and Gale call the market solution. But, as they show,
banks can achieve Pareto improvement by pooling liquidity risks using the law
of large numbers. This banking solution is however subject to a coordination
problem in that fears of depositor flight can lead to a bank run. Even without
bank runs, there is an obvious risk that the law of large numbers – while it applies
across banks – does not apply to any individual bank or those in a region. Hence
the emergence of interbank markets to pool liquidity.
But, as the subprime crisis has made clear, these interbank markets – where
short-term lending is not collateralized – are themselves prone to seizure when
fear of counterparty insolvency stalks the streets. In the crisis of 2008–10, inter-
bank lending effectively gravitated onto the balance sheet of central banks:
following the collapse of Lehman Brothers, the liabilities of the US Federal
Reserve promptly doubled while the balance sheet of the Bank of England
expanded three and half times.
So banks that pool liquidity risk among their customers and interbank markets
which economize on system-wide liquidity are both prone to collapse. Network
gains are subject to coordination failure.
3.2 Contagious insolvency

Similar issues arise with respect to insolvency. In their stochastic network model
of UK banks, Gai and Kapadia (2010) and Gai et al. (2010) find that an intercon-
nected financial system of a large number of banks can fail if and when a shock
affects a ‘super-spreader’ and is then dissipated widely around the system.
Modeling how network structure can amplify shocks may be relevant for the
analysis of the financial crisis, but in the view of Jon Danielsson (2010) a realis-
tic ‘endogenous risk model is beyond our abilities’. In these circumstances, he
argues that supervisors should focus their attention on where the risk is created
rather than trying to measure it; and that the most important factor is reso-
lution – that is, the closure of systematically important financial institutions
which have gambled and face failure.
Another approach, as shown in Stiglitz (2011), is to simplify the structure suf-
ficiently so as to achieve analytical tractability. Using the electricity grid system
as a metaphor, Stiglitz’s analysis involves production uncertainty where the risk
of productivity shocks is insured by sharing costs between all the players. The
result is a stylized model of endogenous risk, where the gains of connectivity
have to be balanced against the risk of propagating large shocks around the sys-
tem. By analogy with the electricity grid, the principal policy recommendation
is the implementation of ‘circuit-breakers’ designed to limit the propagation of
large shocks.
To throw light on structural measures currently being taken to reduce the
vulnerability of banking – both in terms of resolution procedures and of ‘ring-
fencing’ – a simplified version of the Stiglitz model is provided in the Annex
to this chapter, with a trade-off between sharing small risks and avoiding large
ones. Forming an interconnected group allows for risk-pooling, but it exposes
group members to contagion from a large shock hitting any group member, so
there is a limit to optimal group size. Given a ‘circuit-breaker’ which prevents
any contagion from a large shock, however, there is no limit to the size of the
group.
In an open economy context, Stiglitz (2011) suggests that such circuit-breakers
could be interpreted as restrictions on capital flows. Here we suggest that a
circuit-breaker be interpreted as a metaphor for structural changes designed
to limit the damage an insolvent Systemically Important Financial Institu-
tion (SIFI) can impose on the wider banking system. This could include both
resolution procedures, and ‘ring-fencing’ proposals. In the UK, for example,
the Banking Act 2009 created a Special Resolution Regime (SRR),9 giving the
authorities a framework for dealing promptly with distressed banks and build-
ing societies; and the Vickers Commission has recommended a ‘ring-fence’ for
the retail banking operations of big universal banks so that it will be possible
to let the riskier investment banking arms to fail without imperiling household
savings and small business lending.
4 Financial regulation: the state of play
What is being done to check the impact of externalities in the financial sector
since the crisis of 2008–10? The steps being taken involve, first, the regulation
of individual bank portfolios in the form of rules governing capital adequacy
and liquidity holdings; second, changes to the structure of the industry; and
finally, macro-prudential interventions across the industry which varies with
the business cycle.
A compact summary of the current state of play regarding Basel III on the first
two of these is provided by Barrell and Davis (2011):
On Capital:
The new regulations, which are basically complete, will raise common equity
from the previous minimum of 1 per cent of risk-weighted assets to at least
4.5 per cent, and Tier 1 as a whole to 6 per cent. A conservation buffer of 2.5
per cent of risk-weighted assets must also be built up with common equity,
and if this is exhausted in a crisis then the bank will be wound up. A mini-
mum ratio of capital to total (unadjusted) assets of 3 per cent must be held.
There is provision for a countercyclical capital buffer of up to 2.5 per cent
of risk-weighted assets, which is to be imposed at the discretion of the reg-
ulators. The regulation of subsidiaries and capital market activities has been
substantially tightened, including the introduction of stress-related bench-
marks for trading book capital and counterparty credit risk.
On Liquidity:
Two new regulations for liquidity risk are being introduced: first, a liquid-
ity coverage ratio (LCR) enforcing sufficient liquid assets to offset net cash
outflows during a 30-day period of stress; second, a net stable funding ratio
(NSFR) which seeks to ensure a degree of maturity matching over a one-year
horizon, including allowance for off-balance sheet commitments.
Table 7.1 Calibration of the capital framework: capital requirements and buffers
Common equity Tier I capital Total capital

(after deductions)
Minimum 4.5 6.0 8.0

Conservation 2.5
Minimum + conservation buffer 7.0 8.5 10.5
Countercyclical buffer range 0−2.5
Source: Barrel and Davis (2011: F5).
Notes: All numbers in per cent.

13
Capital framework (a)
8
3
Per cent
–2 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
Liquidity standards (b)
–7
–12
Minimum requirement Capital conservation buffer Countercyclical buffer

Leverage ratio LCR NSFR
Figure 7.4 Timeline for the implementation of Basel III
Source: Cechetti et al. (2011: R35).
4.1 The timetable

Given the virulence of the crisis, plans for implementation of these reforms are
extraordinarily protracted. As can be seen from the graphic from Cechetti et al.
(2011) (Figure 7.4), the build-up of the equity and the implementation of the
liquidity standards is spread over a decade.
4.2 Implications for bank leverage

Academic economists see the need for much tougher regulation than do the
guardians of financial stability housed in Basel, as may be seen from the follow-
ing table where the recommendation of Admati et al. (2010) in their evidence to
the ICB – that capital be 15 per cent on total assets implying leverage in single
figures – is shown in the last line.
By contrast, the minimum capital requirement shown at the top of the table
is the 7 per cent baseline ratio of equity to risk weighted assets (RWA) of Basel III.
As indicated in column 3, with risk-weighted assets at half the balance sheet, the
leverage for banks at the Basel III minimum could rise alarmingly close to 30.
The second line of the table shows what Barrel et al. (2010) calculate as the
level of capital necessary to reduce banking risk to acceptable levels – double the
Basel requirements; and this is followed by what Miles et al. (2010) estimate is
needed.
Of crucial importance for the regulation of banking in the UK are the pro-
posals of the Independent Commission for Banking established by the current
Coalition government and chaired by Sir John Vickers. In their Final Report, the
Commissioners specifically recommend a ‘ring-fence’ to insulate vital banking
services on which households and SMEs depend from problems elsewhere in
Table 7.2 Regulatory measures for capital, liquidity, and structural separation
Minimum Capital Requirement Liquidity Other

% RWA % Total Leverage
Assets (assuming
50% RWA)
Basel III 4.5 + 2.5 = 7 3.5 28 LCR, Resolution

+ Cyclical NSFR
Buffer (see text)
‘Double Basel’ 14 7 14
Barrell and Davis
(2011)
Miles et al. (2010) 19 9.5 11
ICB Final Report 20 10 10 Ring-fencing
(2011)
(Retail banking)*
Admati et al. n.a. 15 4
(2010)
Memo item
Hedge Funds 30 3.33
Alessandri and
Haldane (2009)
the financial system. (Such a ‘ring-fence’ is similar to imposing ex-ante circuit-

breakers, limiting the spread of systemic risks ex-post.) Ring-fenced banks are
required to have very substantial capital backing, roughly 10 per cent of equity
and 10 per cent of ‘Cocos’, bonds that convert to equity depending on market
conditions.
In addition, ring-fenced banks are explicitly banned from providing a wide
range of activities.
The list of these Prohibited Services (given in Chapter 3, para 3.39 of the
Report) includes:
• any service which would result in a trading book asset;

• any service which would result in a requirement to hold regulatory capital against
market risk;
• any service which results in an exposure to a non-ring-fenced bank or a non-bank
financial organisation, except those associated with the provision of payments
services where the regulator has deemed this appropriate;
• the purchase or origination of derivatives or other contracts which would result in
a requirement to hold regulatory capital against counterparty credit risk; and
• services relating to secondary markets activity including the purchase of loans or
securities.
Broadly speaking, these prohibitions would mean that wholesale/investment

banking divisions of existing banks could not be placed within the ring-fence;10
consequently, in quantitative terms, ring-fenced banks would only be allowed
to hold ‘between a sixth and a third of the total assets currently held by the UK
banking sector’ (ICB 2010: para. 3.40).
As the authors of the Report observe, in something of an understatement:
A ring-fence of this kind would also have the benefit that ring-fenced banks
would be more straightforward than some existing banking structures and
thus easier to manage, monitor and regulate. (ICB 2010: para. 3.40).
5 Conclusion
Given the pernicious externalities considered above – and the accumulating

evidence that they matter greatly in practice – the plans under Basel III look
seriously inadequate, especially when compared with the decisive steps taken
in the USA by the Roosevelt administration in the 1930s.
In Whither Socialism, Joseph Stiglitz suggested that Communist planning
could not succeed because the information required for the top-down allo-
cation of resources by planners was simply not available. Could information
issues prove the Achilles heel of market liberalism in the West? What if banks
use opaque financial products and limited liability to take on excessive risk –
collecting the upside and shifting the downside to the taxpayer via bail-outs in
times of crisis? Again and again!
The various recommendations reported in Table 7.2 above suggest that it will
take a good deal more than Basel III to ‘save Western capitalism from the capital-
ists’. Could the type of structural separation for UK banking recommended the
Vickers Commission – the ring-fencing of retail activities with explicit prohibi-
tion of risky products and activities therein – turn out to be an effective update
of Glass-Steagall for our times?
Annex
Optimal risk-sharing network with systemic risks
Consider a very simple case of financial network formation. The motivation
for connecting each node to an existing component is to insure against small
idiosyncratic shocks. The whole network, however, may be hit by some ‘macro
shock’ which leads to the failure of large fraction of nodes. The interaction
between these two effects may limit the extent of connectivity. In what follows,
we first show the existence of such limit and then illustrate how by isolating
the node hit by the large shock can increase the connectivity.
Figure 7.A1 A partially connected network
As shown by Gai and Kapadia (2010), whether solvency shocks on a small

fraction of nodes can lead to systemic risk depends on: 1) where the shocks
hit; and 2) the network structure (degree distribution). To make the problem
analytically tractable, we assume a very simple network structure: that there are
a fraction of nodes which are completely connected, and the rest are isolated
(as shown in Figure 7.A1). If a solvency shock (a macro shock) hits an isolated
node, only that node fails. If the solvency shock hits a node in the component,
the whole component fails.
Assume N ex-ante identical nodes, each faced with two types of shock:
a) xi = ε, small shock, could be shared by the network.

b) a large shock T occurs with probability p. So an individual node will get hit
with probability p/N.
The shocks, ε, are iid random variables with bounded support and Eεi = 0.
The type of shock b) is the solvency shock described above. To have clear-cut
results, we further assume that shocks a) and b) are independent.
To look at the incentive to form connections, we compare the expected utility
for two types of typical nodes: isolated and in a component.
I An isolated node
The expected utility at the node is given by:

N −1 1
U1 = (1 − p)Eu(εi ) + pEu(εi ) + u(T )
N N

N −1 p
= 1−p+ p Eu(εi ) + u(T ) (1)
N N
where u(.) is a standard increasing and strictly concave utility function, and E
is an expectation operator. Here, we assume that Eu(εi ) > u(T ).
II A node in a component
In a partially connected network, consider an completely connected component
of size n. The expected utility of a node in the component is:

εi N −n εi n
UC (n) = (1 − p)Eu +p Eu + u(T )
n N n N

N −n εi np
= 1−p+ p Eu + u(T )
N n N

np εi np
= 1− Eu + u(T ) (2)
N n N
In the completely connected component of size n, the idiosyncratic shocks are

evenly shared. (It is clear that if n = 1 the above becomes (1), the unconnected
case.)
Differentiating the above expected utility w.r.t. n yields

∂UC (n) p εi np 1 εi εi
= u(T ) − Eu + 1− − E u . (3)
∂n n n N n n n
The first term on the RHS of (3) represents the adverse effect of solvency shock
on the component: the larger is the component, the more severe will be this
adverse effect. The second term represents the positive effect of smoothing the
idiosyncratic shock: this effect declines with the increase in n because of the con-
cavity of the utility function. A trivial case is when Eu(εi ) >> u(T ): the negative
effect dominates, no connection is formed.
For some reasonable utility functions (or u(T ) not too small), it could be that
the positive effect dominates if n is small while the negative effect dominates if
n is large. In this case, the optimal network would be the one which maximizes
(2), as illustrated in Figure 7.A2, so there is a limit to the optimal size of a group.
Note that in a naturally formed network, an isolated node can make con-
nections and a connected node can break its connections, so the network will
have isolated components of size n*. In this case, the probability of systemic risk is
n∗ p/N.
III Circuit-breaker (withdrawal of node)

In Gai and Kapadia (2010), the policy of isolating the node hit by a large shock
was not examined, maybe because of difficulties associated with substantial
cross-holdings of assets among banks. Side-stepping such issues, let us assume
that the large shock is clearly identifiable. Assume that if a connected node
is hit by the large shock, its connections to other nodes in a component are
severed; then the expected utility of a member in an n-group with such kind of
Uc (n)
1
n* n
Figure 7.A2 Optimal network
Ub (n)
1
N n
Figure 7.A3 Expected utility of a node with circuit-breaker
‘circuit-breaker’ is:

εi N −n εi n−1 εi 1
Ub (n) = (1 − p)Eu +p Eu + Eu + u(T )
n n n N n−1 N

N −n n−1 εi p
≈ 1−p+ p+ p Eu + u(T )
N N n N

εi p
= (1 − p/N)Eu + u(T ) (4)
n N
The above is an increasing function of n, so the optimal size of a group is N.
The idea that by separating systemically important nodes could allow the
network to benefit fully its risk-sharing function is intuitive with the aid of
the simple structure considered. The real financial networks are rarely com-
pletely connected: they usually exhibit small-world properties with fat-tail
degree distribution and a high clustering coefficient. These imply that there are
some important financial hubs which are highly interconnected – the so-called
‘super-spreaders’. Haldane (2009) and Haldane and May (2011) have argued that
it is crucial to identify such ‘super-spreaders’ and to impose appropriate regu-
latory measures (such as higher capital buffers) to reduce their adverse effect
on the stability of the whole financial system. In a similar vein, Stiglitz (2011)
suggested, in the context of global financial integration, the use of ‘circuit-
breakers’ (through, for example, the use of capital controls) to separate the
infected component from the rest of the system.
As discussed in the Conclusion, the Independent Commission on Banking, in
its Final Report (2011), advocates a structural approach to banking regulation,
by ‘ring-fencing’ commercial banks from their investment arms, and subjecting
them to limits on risk assets and different capital adequacy requirements. Could
the ‘circuit-breaker’ used in the simple model above be a metaphor for such
ex-ante structural separation?
Notes
* For comments and suggestions, we are grateful to Peter Hammond, Paulo Santos
Monteiro and discussions at the IEA World Congress, Beijing, July 2011.
1. Substantial losses suffered by US banks on Latin American lending in the 1980s did,
however, lead to cooperation in searching for internationally accepted baselines for
prudential regulation. The result was the Basel Accord of 1988, setting a minimum
capital requirement of 8 per cent of total risk-weighted assets on individual banks,
which led to a substantial recapitalization of the international banking sector.
2. Just before the East Asian crisis, the IMF had been planning to change the Articles of
Agreement so as to remove the sovereign right of members to impose capital controls;
see Fischer (2004).
3. For advanced economies, Barrell et al. (2010) show how the probability of the crisis
can be explained by inadequate levels of capital and liquidity.
4. This would have little welfare significance if the user cost of capital – its productivity
in the hands of the residual buyers – were constant.
5. In their discussion of amplification through balance sheets and asset prices, Kiyotaki
and Moore (1997) assume that the ‘overshooting’ will not be severe enough to render
the illiquid agents insolvent.
6. Further details are available in Alessandri and Haldane (2009), Miller and Stiglitz
(2010), and Sinn (2010).
7. Prices that would be established without leverage.
8. The lower is the value of λ, the higher is the degree of risk concentration.
9. By, for example, transferring all or part of a bank to a private sector purchaser, or to a
‘bridge bank’ subsidiary of the Bank of England, or into temporary public ownership;
or sending it to be wound up.
10. With such a firewall, indeed, especially with improved resolution procedures, it is
intended that ‘the investment arm could, in extreme circumstances, be liquidated
efficiently and at no public expense while preserving the retail activities uninfected
by bad investment banking assets’ (ICB 2010: para 4.21).
References
Admati, A. et al. (2010) ‘Much more bank equity is needed and is not socially costly’,
evidence submitted to the Independent Commission on Banking. Available on the ICB
website at: http://bankingcommission.independent.gov.uk/
Alessandri, P. and A. G. Haldane (2009) ‘Banking on the State’, available at:
www.bankofengland.co.uk/publications/speeches/2009/speech409.pdf
Allen, F. and D. Gale (2000) ‘Bubbles and Crises’, Economic Journal, vol. 110, pp.
236–255.
Allen, F. and D. Gale (2007) Understanding Financial Crises (Oxford: Oxford University
Press).
Barrell, R. and E. P. Davis (2011) ‘Financial Regulation’, National Institute Economic Review,
No. 216 (April).
Barrell, R. and E. P. Davis and D. Karim (2010) ‘Calibrating Macroprudencial Policy’,
National Institute Discussion Paper No. 354 (London: NIESR).
Cecchetti, S. G., D. Domanski and P. von Goetz (2011) ‘New Regulation and the New
World of Global Banking’, National Institute Economic Review, No. 216 (April).
Danielsson, Jon (2010) ‘Systemic Risk Models: Issues and Prospects’, Talk presented at
Macroprudential Policy: Issues and Challenges in FMG, LSE, 2 November.
Diamond, D. W. and P. H. Dybvig (1983) ‘Bank Runs, Deposit Insurance, and Liquidity’,
Journal of Political Economy, vol. 91, no. 3, pp. 400–419.
Fischer, S. (2004) ‘Capital Account Liberalization and the Role of the IMF’, Chapter 5 of
S. Fischer, IMF Essays from a Time of Crisis (Cambridge, MA: MIT Press).
Gai, P. and S. Kapadia (2010) ‘Contagion in Financial Networks’, Proceedings of the Royal
Society A, vol. 466, no. 2120, pp. 2401–2423.
Gai, P., A. Haldane and S. Kapadia (2010) ‘Complexity, Concentration and Contagion –
Outline’, paper presented at the EEA Conference, Glasgow, August.
Gai, P., S. Kapadia, S. Millard and A. Perez (2008) ‘Financial Innovation, Macroeconomic
Stability and Systemic Crises’, Economic Journal, 118, pp. 401–426.
Geanakoplos, J. D. and H. M. Polemarchakis (1986) ‘Existence, Regularity, and Constrained
Suboptimality of Competitive Allocations When the Asset Market Is Incomplete’, Yale
University Cowles Foundation Paper, 652.
Greenspan, Alan (2002) ‘World Finance and Risk Management’, speech at Lancaster House
London, September. Available at: www.hm-treasury.gov.uk
Greenwald, B. and J. E. Stiglitz (1986) ‘Externalities in Economies with Imperfect Infor-
mation and Incomplete Markets’, Quarterly Journal of Economics, vol. 101, no. 2, pp.
229–264.
Haldane, A. G. (2009) ‘Rethinking the Financial Network’, speech delivered at the Financial
Student Association, Amsterdam, April. Available at: www.bankofengland.co.uk/pub-
lications/speeches/2009/speech386.pdf
Haldane, A. G. and R. M. May (2011) ‘Systemic Risk in Banking Ecosystems’, Nature, 469,
20, January, pp. 351–355.
ICB (Independent Commission on Banking) (2010) Issues Paper, Call for Evidence (London:
ICB).
ICB (Independent Commission on Banking) (2011) Final Report (London: ICB).
Koo, R. (2008) The Holy Grail of Macroeconomics – Lessons from Japan’s Great Recession
(Hoboken: John Wiley & Sons).
Korinek, A. (2011) ‘Systemic Risk-taking: Amplication Effects, Externalities, and Regula-
tory Responses’, mimeo, University of Maryland.
Krishnamurthy, A. (2010) ‘Amplification Mechanisms in Liquidity Crises’, American

Economic Journal – Macroeconomics, vol. 2, no. 3, pp. 1–30.
Kiyotaki, N. and J. Moore (1997) ‘Credit Cycle’, Journal of Political Economy, vol. 105, no.
2 (April), pp. 211–248.
Laibson, D. and J. Mollerstrom (2010) ‘Capital Flows, Consumption Booms and Asset
Bubbles: A Behavioural Alternative to the Savings Glut Hypothesis’, Economic Journal,
vol. 120, no. 544, pp. 353–373.
Majnoni, G. and A. Powell (2011) ‘On Endogenous Risk, the Amplification Effects of
Financial Systems and Macro Prudential Policies’, mimeo, IADB.
Mankiw, N. G. (1986) ‘The Equity Premium and the Concentration of Aggregate Shocks’,
Journal of Financial Economics, vol. 17, no. 1, pp. 211–219.
Miles, D., J. Yang and G. Marcheggiano (2011) ‘Optimal Bank Capital’, DP no. 31, External
MPC Unit, Bank of England.
Miller, M., P. Santos Monteiro and L. Zhang (2011) ‘Eastern Caution, Western Exuberance
and Global Imbalance’, Journal of Globalization and Development, vol. 2, no. 1, Article 1.
Miller, M. and J. E. Stiglitz (2010) ‘Leverage and Asset Bubbles: Averting Armageddon with
Chapter 11?’, Economic Journal, vol. 120, no. 544, pp 500–518.
Sinn, H. W. (2010) Casino Capitalism (Oxford: Oxford University Press).
Stiglitz, J. E. (2011) ‘Contagion, Liberalization, and the Optimal Structure of Globaliza-
tion’, Journal of Globalization and Development, vol. 1, no. 2, Article 2.
Stiglitz, J. E. (1994) Whither Socialism? (Cambridge, MA: MIT Press).
Stiglitz, J. E. (2010) ‘The Dangers of Deficit Reduction’, Project Syndicate, 5 March . See
also: www.project-syndicate.org/commentary/the-dangers-of-deficit-reduction.
Wolf, W. (2010) ‘In the Quest of Systemic Externalities: A Review of the Literature’, CESifo
Economic Studies, vol. 56, no. 1, pp. 96–111.
Vickers, J. (2011) Opening remarks at the ICB Interim Report Publication, 11 April.
8
Bank Lending and Credit
Supply Shocks
Simon Gilchrist
Boston University and NBER, USA
Egon Zakrajšek
Division of Monetary Affairs, US Federal Reserve Board
1 Introduction
The turmoil that raged in the global financial markets during the 2007–09 crisis
left a significant imprint on bank lending over the past several years. The succes-
sive waves of turbulence that ripped through the financial system during that
period exerted substantial pressure on both the asset and liability sides of banks’
balance sheets, and banks, at the height of the crisis in the latter part of 2008,
faced funding markets that were largely illiquid and secondary markets that were
essentially closed to sales of certain types of loans and securities. Together with
the slowdown in economic activity that emerged at the end of 2007 and accel-
erated appreciably in latter part of 2008, these financial disruptions led banks
to become significantly more cautious in the extension of credit and to take
steps to bolster their capital and liquidity positions. Moreover, the persistent
tightness of credit conditions faced by many borrowers, in combination with
generally weak demand for bank-intermediated credit, have continued to affect
lending during the sluggish recovery. Indeed, two years after the official end of
the recession, core loans outstanding – the sum of bank loans to households and
nonfinancial businesses – remain 13 per cent below the level reached during the
cyclical peak in December 2007.1
The 2007–09 financial crisis began to unfold as rising delinquencies on sub-
prime mortgages in the first half of 2007, triggered by the end of the housing
boom in the United States, started to lead to large losses on related structured
credit products. At that time, banks, in addition to their mounting concerns
about actual and potential credit losses, recognized that they might need to take
a large volume of assets on to their balance sheets, given their existing commit-
ments to customers and the heightened reluctance of investors to purchase an
increasing number of securitized products. The recognition that the ongoing
154
Bank Lending and Credit Supply Shocks 155
turmoil in financial markets could lead to substantially larger than anticipated

calls on their funding abilities and investors’ concerns about valuation prac-
tices for opaque assets arguably led banks to start adopting a significantly
more defensive lending posture in an effort to conserve their balance sheet
capacity.
The fully fledged global nature of the crisis burst to the forefront on
15 September 2008, when Lehman Brothers – with its borrowing capacity
severely curtailed by a lack of collateral – filed for bankruptcy. Investor anxiety
about financial institutions escalated sharply, and market participants became
extraordinarily skittish and pulled back from risk-taking even further. Amid
cascading effects of these financial disruptions, which included the US gov-
ernment’s rescue of Fannie Mae, Freddie Mac, and AIG, a run on money
market mutual funds, and the failure of Washington Mutual (a large US thrift),
asset values dropped sharply, business and consumer confidence slumped, and
aggregate credit conditions tightened further. Bank lending to households and
businesses, which had held up reasonably well through the early stages of the
crisis, slumped, and the US economy found itself in the so-called ‘adverse feed-
back loop’, a precarious situation in which deteriorating economic and financial
conditions become mutually reinforcing.
In this chapter, we examine empirically the linkages between credit sup-
ply conditions, bank lending, and economic activity. We begin by providing
a historical context for the behavior of the major categories of bank loans to
households and businesses during the 2007–09 financial crisis. We document
that prior to the escalation of financial market turmoil in the autumn of 2008,
the provision of credit to businesses and households by commercial banks –
with the exception of home mortgage loans – appeared to have been well-
maintained relative to the stage of the business cycle. However, bank lending
exhibited a sharp slowdown at the end of 2008, and loans on banks’ balance
sheets have continued to run off throughout the nascent recovery. From a pure
timing perspective, therefore, the contraction in bank lending is a coincident, if
not lagging, indicator of business cycle dynamics. As such, one may be tempted
to conclude that the contraction in lending in the latter stages of the downturn
was a symptom rather than a causal mechanism during the crisis.
To explore this issue, we consider the effects of credit supply shocks on bank
lending and economic activity. The notion that the banking system may serve
as a propagation mechanism for, or a source of, macroeconomic shocks is, of
course, not new. In the 1960s, for example, Brunner and Meltzer (1963) criti-
cized small-scale macroeconomic models for not including multiple measures
of credit prices and quantities (including those of bank loans), while in his sem-
inal work on the Great Depression, Bernanke (1983) argued that the widespread
bank failures during the early 1930s helped exacerbate the depth and length of
the ensuing economic contraction.2 The issue, however, remains contentious,
156 A. Gilchrist and E. Zakrajšek
because any empirical investigation of the role that banks – or financial interme-
diaries more generally – play in economic fluctuations is complicated by severe
endogeneity problems, as a plethora of financial shocks that can affect the sup-
ply of bank loans is also likely to have independent effects on economic activity.3
To tackle this difficult identification problem, we employ the approach used
recently by Gilchrist and Zakrajšek (2012), who use a large panel of secondary
market prices of bonds issued by US nonfinancial corporations to decompose
the associated credit spreads into two components: 1) a component capturing
the usual countercyclical movements in expected defaults; and 2) a compo-
nent representing the cyclical changes in the relationship between default risk
and credit spreads – the so-called excess bond premium. As shown by Gilchrist
and Zakrajšek (2012), movements in the excess bond premium appear to reflect
shifts in the risk attitudes of financial intermediaries, the marginal investors
pricing corporate debt claims. As such, fluctuations in the excess bond premium
may provide an especially timely indicator of cyclical changes in credit supply
conditions, both within the corporate cash market and in the market for bank-
intermediated credit. Indeed, our results indicate that shocks to the excess bond
premium – a measure of disruptions to the credit intermediation process – cause
a pronounced contraction in economic activity, a decline in nominal interest
rates, a sharp fall in equity valuations, and an eventual decline in bank lending.
Thus, the fact that a runoff in bank loans during the 2007–09 recession did not
materialize until the latter stages of the recession appears to be a general feature
of credit market disruptions.
To analyze in more detail the apparently sluggish response of bank lending to
credit supply shocks, we examine the joint dynamics of business loans outstand-
ing and unused business loan commitments, an especially cyclically sensitive
component of bank-intermediated credit. Our focus on unused commitments is
motivated by the fact that the banking system provides credit to businesses (and
households) in two important ways: by originating new loans (on balance sheet)
and by providing lines of credit (off balance sheet). This distinction is crucial for
understanding the cyclical dynamics of bank lending, because unused loan com-
mitments, which represent a significant source of off-balance-sheet credit risk,
started to contract immediately with the onset of the crisis in mid-2007, while
business loans outstanding on banks’ balance sheet expanded briskly during the
first year of the recession.
A part of the contraction in unused commitment occurred as banks, in
response to mounting capital and liquidity pressures, started to reduce their
off-balance-sheet credit exposures by cutting their customers’ existing lines of
credit, a move that was entirely coincident with the overall deterioration in
financial market conditions as measured by a rise in the excess bond premium.4
Indeed, unused business loan commitments plummeted at a 30 per cent annual
rate during the height of the crisis in the fourth quarter of 2008, precisely when
the excess bond premium reached its historical peak. Business loans on bank
balance sheets, by contrast, were still expanding at this point of the cycle; see,
for example, Chari et al. (2008) and Ivashina and Scharfstein (2010).
While both the extent and severity of financial market disruptions during the
2007–09 crisis were unprecedented by postwar standards, our results indicate
that such behavior represents a typical response of bank lending to credit sup-
ply shock – that is, in the initial phase of the cyclical downturn, the capacity of
businesses to borrow from the banking sector shrinks primarily through reduc-
tions in unused commitments and only eventually through a reduction in loans
outstanding. All told, the empirical evidence presented in this chapter indicates
that loan quantities are highly sensitive to changes in credit supply conditions
once one takes into account the differences in cyclical dynamics of unused loan
commitments versus loans outstanding.
2 Credit aggregates during the 2007–09 financial crisis
In this section, we place the behavior of credit aggregates during the recent cri-
sis into a historical context. We begin by documenting the cyclical properties
of total private nonfinancial debt and its main components. We then focus on
credit intermediated by commercial banks and provide an explicit comparison
of lending patterns at banks with the aggregate credit flows. A common story
that emerges from these different cuts of the data is that, with the exception of
home mortgage lending, other forms of credit outstanding to households and
businesses increased appreciably during the early stages of the recession. How-
ever, the extension of credit came to an abrupt end in late 2008, following the
sharp escalation of financial market turmoil sparked by the collapse of Lehman
Brothers.
Our data on debt stocks outstanding come from the Federal Reserve’s Z.1 Statis-
tical Release, ‘Flow of Funds Accounts of the United States’ and cover the period
from 1952:Q1 to 2010:Q4. As noted above, we begin by analyzing the behavior
of total private nonfinancial debt and its main components: home mortgages,
consumer credit (credit card, auto, student, and other consumer loans), and debt
extended to nonfinancial businesses (bank and non-bank loans, commercial
paper, and corporate bonds). We normalized each credit category by nominal
GDP, a transformation that facilitates the comparison of credit cycles across
recessions that differ appreciably in their severity and duration. The resulting
credit ratios, however, still contain pronounced secular trends, reflecting the
myriad of structural changes that took place in credit markets over the past six
decades. To abstract from these developments, we regressed the logarithm of
each credit-to-GDP ratio on a constant and linear and quadratic time trends; for
each NBER-dated recession since 1952, we then normalized the detrended series
to equal zero at its respective business cycle peak.
The solid black lines in Figure 8.1 depict the average behavior of these credit
aggregates around NBER-dated business cycle peaks, calculated using data for
all recessions since 1953, excluding the 2007–09 downturn; the shaded band in
each panel represents the corresponding range of outcomes, while the solid line
shows the behavior of each series during the period surrounding the 2007–09
financial crisis. As shown in the upper left panel, total private nonfinancial debt
outstanding continued to expand – relative to the cycle – over the first five quar-
ters following the NBER-dated peak in 2007:Q4. The expansion in private debt
outstanding owed importantly to a significant increase in debt extended to non-
financial business and, to a lesser extent, a rise in consumer credit. In contrast,
Private nonfinancial debt Home mortgages

Per cent change from business cycle peak Per cent change from business cycle peak
5 20
15
10
0
5
0
–5
–5
–10
2007–09 recession –10
Average –15
Range
–20
–15 –25
–4 –2 0 2 4 6 8 10 12 –4 –2 0 2 4 6 8 10 12
Quarter to and from business cycle peak Quarter to and from business cycle peak
Consumer credit Nonfinancial business debt

15 20
10
10
5
0
0
–5
–10
–10
–15
–20
–20
–25 –30
–4 –2 0 2 4 6 8 10 12 –4 –2 0 2 4 6 8 10 12
Figure 8.1 Cyclical dynamics of private nonfinancial debt and its main components
Per cent
100
Quarterly
Home mortgages
Commercial mortgages
Consumer credit 80
Nonfinancial business credit
60
40
20
0
1952 1956 1960 1964 1968 1972 1976 1980 1984 1989 1992 1996 2000 2004 2008
Figure 8.2 The relative importance of bank lending to households and businesses
home mortgage debt started to decline immediately with the onset of the reces-
sion, with the runoff intensifying noticeably precisely at the stage of the cycle
when the expansion of consumer credit and nonfinancial business debt came
to an end. On the cyclically adjusted basis, the contraction in home mortgage
debt outstanding during the 2007–09 recession is unprecedented by postwar
standards, while the behavior of consumer credit and nonfinancial business
debt is largely within the historical norms.
To help understand the role of the commercial banking sector in the credit
allocation process, Figure 8.2 plots the four major components of bank loans
outstanding, expressed as a percentage of total debt outstanding in that cate-
gory. According to this metric, the relative importance of banks in the direct
provision of credit to households and businesses has clearly changed signifi-
cantly over time.5 First, the role of the banking sector in the provision of credit
to nonfinancial businesses has diminished steadily over the past six decades, a
pattern reflecting primarily the deepening of, and firms’ greater access to, capital
markets, as well as the rise of non-bank financial intermediaries.6
Second, banks appear to have offset this loss of business, in part, by signifi-
cantly stepping up commercial real estate lending, as the share of commercial
mortgages outstanding on banks’ balance sheets has risen from about 20 per
cent to more than 50 per cent over the same period. The importance of banks
in the provision of consumer credit has also declined noticeably since the late
1970s, due mainly to the increased importance of the consumer finance indus-
try. In contrast, the share of home mortgages on the books of commercial banks
has remained fairly stable at around 18 per cent over the past six decades.
Figure 8.3 depicts the cyclical dynamics of these four aggregate bank loan
categories.7 Focusing on the most recent patterns reveals that bank lending
Home mortgages Commercial mortgages

20 20
15
2007–09 recession
Average
10 10
Range
5
0
0
–5
–10
–10
–15 –20
–20
–25 –30
–4 –2 0 2 4 6 8 10 12 –4 –2 0 2 4 6 8 10 12
Consumer credit Nonfinancial business credit

15 20
10
10
5
0
0
–5
–10
–10
–15
–20
–20
–25 –30
–4 –2 0 2 4 6 8 10 12 –4 –2 0 2 4 6 8 10 12
Figure 8.3 Cyclical dynamics of household and business loans at commercial banks
to households and businesses generally conformed to the aggregate credit

trends shown in Figure 8.1. One exception appears to be the behavior of
bank-intermediated consumer credit, which during the recent crisis diverged
noticeably from historical norms. Compared with total consumer credit, con-
sumer loans on banks’ books increased – on the cyclically adjusted basis –
significantly more over the first five quarters of the downturn. Nevertheless, the
subsequent contraction in bank-intermediate consumer credit has been quite
severe and has persisted well into the initial recovery stage of the cycle, a pattern
that is also evident in the behavior of total consumer credit.
On the business side, the exclusion of arms-length finance associated with

the commercial paper and corporate bond markets serves only to highlight the
role that the banking sector serves as the provider of business credit in the
form of credit lines. Compared with total nonfinancial business credit, bank
lending to nonfinancial firms through C&I loans expanded faster and by a
greater percentage in the early stages of the 2007–09 recession.8 Bank lending to
businesses through commercial real estate loans also increased robustly on the
cyclically adjusted basis during the first year of the recession before contracting
significantly in early 2009. In general, the runoff in business loans outstanding
during the most recent downturn was considerably more severe and occurred
at a significantly later stage of the cycle, compared with an average postwar
recession.
3 Credit supply shocks and bank lending
In this section, we analyze the effects of credit supply shocks on bank lend-
ing and economic activity. As discussed above, we rely on the excess bond
premium – an indicator of financial market stress developed by Gilchrist and
Zakrajšek (2012) (GZ hereafter) – to measure changes in credit supply conditions.
Our choice is motivated by an emergent literature that stresses the importance
of balance sheet conditions of financial intermediaries – including those of com-
mercial banks – for the joint determination of asset prices and macroeconomic
aggregates.
3.1 The excess bond premium

Before presenting our main results, we briefly outline the empirical methodol-
ogy underlying the construction of the excess bond premium. Specifically, GZ
employ a large panel of corporate bonds issued by nonfinancial firms to decom-
pose the associated credit spreads into two components: a default-risk compo-
nent capturing the usual countercyclical movements in expected defaults, and
a non-default-risk component that captures the cyclical fluctuations in the rela-
tionship between default risk and credit spreads. The GZ decomposition of credit
spreads is based on the regression model of the following type:

ln Sit [k] = −βDDit + θ Xit [k] + it [k], (1)
where Si [k] denotes the credit spread on bond k (issued by firm i); DDit is the
distance-to-default for firm i; Xit [k] is a vector of bond-specific characteristics that
controls for the optionality features embedded in most corporate securities as
well as for potential term and liquidity premiums; and it [k] is a ‘pricing error’.
The key feature of the GZ approach is that the firm-specific credit risk is cap-
tured by the distance-to-default (DD), a market-based indicator of default risk
based on the option-theoretic framework developed in the seminal work of Mer-

ton (1974). In this contingent-claims approach to corporate credit risk, it is
assumed that a firm has just issued a single zero-coupon bond of face value F
that matures at date T . Rational stockholders will default at date T only if the
total value of the firm VT < F; by assumption, the rights of the bondholders
are activated only at the maturity date, as stockholders will maintain control of
the firm even if the value of the firm Vs < F for some s < T . Under the assump-
tions of the model, the probability of default – that is, Pr[VT < F] – depends on
the distance-to-default, a volatility-adjusted measure of leverage inferred from
equity valuations and the firm’s observed capital structure.
Using the estimated parameters of the credit spread model (1), GZ define the
excess bond premium in month t by the following linear decomposition:
EBPt = S̄t −
S̄t ,
where S̄t denotes the average credit spread in month t and S̄t is its predicted
counterpart. Figure 8.4 shows the estimated monthly excess bond premium
from January 1973 to December 2010. According to the results reported by GZ,
the majority of the well-documented information content of credit spreads for
future economic activity is attributable to movements in the excess bond pre-
mium – that is, to deviations in the pricing of corporate debt claims relative to
the expected default risk of the issuer. Moreover, GZ show that shocks to the
Percentage points
3
Monthly
–1
–2
1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009
Figure 8.4 The excess bond premium

excess bond premium that are orthogonal to the current macroeconomic con-
ditions cause economically and statistically significant declines in economic
activity and inflation, as well as in risk-free rates and broad equity prices.
Importantly, GZ also show that fluctuations in their excess bond premium
are closely related to the financial condition of broker-dealers, highly lever-
aged financial intermediaries that play a key role in most financial markets.9
Taken together, the evidence presented by GZ is consistent with the notion that
deviations in the pricing of long-term corporate bonds relative to the expected
default risk of the underlying issuer reflect shifts in the effective risk-aversion of
the financial sector. Increases in risk-aversion, in turn, lead to a contraction in
the supply of credit, both through the corporate bond market and the broader
commercial banking sector.
3.2 The excess bond premium and bank lending

Following GZ, we employ an identified vector autoregression (VAR) to analyze
the near- and longer-term effects of financial disturbances – as measured by
shocks to the excess bond premium – on bank lending. Our first set of results
is based on monthly data, a data frequency that allows us to be more precise
about the causal mechanism at work. In particular, we show that in a monthly
VAR, orthogonalized innovations to the excess bond premium have a negligible
contemporaneous effect on financial asset prices. This finding implies that our
identification strategy is unlikely to be associated with reverse causation, in
which broad declines in asset valuation cause movements in the excess bond
premium, and we interpret these baseline results as providing strong validation
to our overall approach of identifying credit supply shocks.
To examine the macroeconomic consequences of credit market disturbances,
we augment an otherwise standard VAR with the excess bond premium and bank
lending variables. Specifically, our VAR includes the following endogenous vari-
ables: 1) the change in the unemployment rate; 2) the log-difference in the index
of manufacturing industrial production; 3) inflation as measured by the log-
difference of the price deflator for personal consumption expenditures (PCE);
4) the excess bond premium; 5) the log-difference of real loans outstanding
to households; 6) the log-difference of real loans outstanding to businesses; 7)
the log-difference of the price-dividend ratio; 8) the 10-year (nominal) Treasury
yield; and 9) the effective (nominal) federal funds rate.10
The first three variables provide monthly measures of economic activity
and inflation. They are traditionally viewed as ‘slow-moving’ variables and
are ordered before the excess bond premium when performing the Cholesky
decomposition used to compute orthogonalized innovations to the excess bond
premium. The block after the excess bond premium includes bank lending and
asset prices, all of which viewed are as ‘fast-moving’ variables and thus are
allowed to react contemporaneously to innovations in the excess bond pre-

mium. According to our identification scheme, therefore, shocks to the excess
bond premium may engender an immediate response in stock market valua-
tions – as measured by the price-dividend ratio – as well in nominal interest
rates and bank lending. Importantly, this ordering also allows for the possibility
that innovations to the excess bond premium may cause an immediate shift in
the stance of monetary policy – as measured by the movements in the federal
funds rate – to perceived disruptions in the credit intermediation process.
Figure 8.5 depicts the impulse responses of these variables to an orthogonal-
ized shock to the excess bond premium, based on the VAR(6) model estimated
Unemployment rate Industrial production PCE prices

Percentage points Percentage points Percentage points
0.4 0.5 0.10
0.05
0.3 0.0
0.00
0.2 –0.5 –0.05
0.1 –1.0 –0.10

–0.15
0.0 –1.5
–0.20
–0.1 –2.0 –0.25
0 6 12 18 24 30 36 0 6 12 18 24 30 36 0 6 12 18 24 30 36
Months after the shock Months after the shock Months after the shock
Excess bond premium Real business loans Real household loans

0.30 2 2
0.25
1 1
0.20
0.15 0 0
0.10
0.05 –1 –1
0.00
–2 –2
–0.05
–0.10 –3 –3
0 6 12 18 24 30 36 0 6 12 18 24 30 36 0 6 12 18 24 30 36
Price-dividend ratio 10-year Treasury yield Federal funds rate

1 0.05 0.1
0
0.00 0.0
–1
–2 –0.05 –0.1
–3 –0.10 –0.2
–4
–0.15 –0.3
–5
–6 –0.20 –0.4
0 6 12 18 24 30 36 0 6 12 18 24 30 36 0 6 12 18 24 30 36
Figure 8.5 Credit supply shocks, economic activity, and bank lending
over the period from January 1986 to December 2010.11 An unanticipated

increase of one standard deviation in the excess bond premium – about 30 basis
points – causes a significant and protracted slowdown in economic activity. In
economic terms, the implications of this relatively modest financial disruption
are substantial: the unemployment rate increases a little less than one-quarter
of a percentage point, while industrial production falls a full percentage point
relative to trend growth, with the peak decline in economic activity occurring
about two years after the initial shock.
The economic downturn is accompanied by a substantial fall in broad equity
valuations, as evidenced by the decline in the price-dividend ratio. In this sense,
the results in Figure 8.5 are consistent with the notion that turmoil in the
corporate bond market – owing to an increase in risk-aversion of financial inter-
mediaries – leads to stresses in asset markets more generally, though with a bit
of a delay. Finally, the combination of growing economic slack and apprecia-
ble disinflation in the wake of the financial shock elicits a significant easing of
monetary policy and a decline in longer-term yields. However, it is important
to note that these forward-looking asset prices do not react contemporaneously
to our financial disturbance – indeed, the price-dividend ratio does not begin
to decline about until four months after the impact of the shock to the excess
bond premium.12
On the credit side, the shock to the excess bond premium has no imme-
diate effect on bank lending to either households or businesses. In the first
six months after the shock, both household and business loans outstanding
increase slightly. About one year after the initial financial disruption, however,
business loans outstanding begin to contract and bottom out more than one
percentage point below their trend growth. The effect on household lending
over this period, by contrast, is statistically and economically indistinguishable
from zero. All told, a shock to the excess bond premium leads to a substantial
decline in business loans outstanding, though the contraction in business lend-
ing occurs with a significant lag. In the next section, we provide further insight
into this delayed reaction by examining the behavior of loans outstanding versus
unused loan commitments.
3.3 Loans outstanding vs loan commitments

A key issue in the analysis of credit cycles is the degree to which firms and
households are financing their activities by borrowing through newly issued
loans versus drawing on their existing lines of credit.13 Similarly, from the
credit-supply perspective, banks may curtail their credit exposures in the ini-
tial phases of a cyclical downturn primarily by reducing the amount of unused
commitments. To the extent that banks’ initial reaction to financial disturbances
is to cut back on unused commitments, this may help explain the sluggish
response of loans outstanding to credit supply shocks. In fact, this hypothesis
(a) Core loans outstanding and used commitments
$ Trillions
7
Quarterly
Core unused commitments 6
Core loans outstanding
5
0
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010
(b) Composition of unused commitments
$ Trillions
7
Quarterly
HELOCs 6
Credit cards
Business 5
4
3
2
1
0
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010
Figure 8.6 Core loans and unused commitments at commercial banks
is consistent with the evidence presented by Morgan (1998), who shows that
changes in loans outstanding not made under commitment are more sensitive
to changes in the stance of monetary policy than changes in loans made under
a previous commitment.
The banks’ unique role as a provider of credit in the form credit lines is
illustrated in Figure 8.6. According to the Call Report data, core loans outstand-
ing exceeded the corresponding unused commitments by a substantial margin
during the early 1990s.14 Over time, however, banks’ off-balance-sheet credit
exposures have grown more rapidly, and by the most recent business cycle peak
at the end of 2007, core unused commitments totaled close to seven trillion
dollars, substantially more than about five trillion dollars of core loans out-
standing. As shown in the bottom panel, credit card commitments accounted
for the majority of this off-balance-sheet exposure, followed closely by business
credit lines.15
Per cent
25
Quarterly
20
15
10
–5
Business loans –10

Business unused commitments
Business lending capacity –15
–20
–25
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Figure 8.7 Growth in business lending capacity at commercial banks
In the latter half of 2007, this sizable off-balance-sheet credit exposure pre-
sented banks with a major risk in light of escalating financial market strains and
an emerging slowdown in economic activity. The bulk of the contraction in
core unused commitments during that period was accounted for by a reduction
in business credit lines, the most cyclically sensitive component of bank-
intermediated credit. Given the relative importance of banks’ commitments
to fund business loans, we define a broader measure of credit intermediation
by commercial banks in this segment of the market – business lending capac-
ity, which attempts to capture the full potential of businesses to borrow from
the banking sector over time, as measured by the sum of business loans out-
standing and corresponding commitments to fund such loans. The black line
in Figure 8.7 depicts the (annualized) quarterly growth rate of business lending
capacity, while the shaded portions of the vertical bars represent the quarterly
growth contributions of business loans outstanding and corresponding unused
commitments.
According to Figure 8.7, cyclical fluctuations in business lending capacity are
driven importantly by changes in unused commitments, a pattern that was
especially pronounced during the most recent crisis. Indeed, the data shown
in the figure reveal a more general pattern: a reduction in business lending
capacity at an early stage of a downturn is due entirely to a decline in unused
commitments rather than business loans outstanding. During the recent crisis,
in particular, the growth of unused commitments stepped down immediately
with the emergence of financial market turmoil in the summer of 2007, and
unused commitments continued to contract well after the official end of the
Percentage points Net per cent

4 100
Monthly
Excess bond premium (left scale)

3 Change in C&I lending standards (right scale) 75
2 50
1 25
0 0
–1 –25
–2 –50
–3 –75
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010
Figure 8.8 The excess bond premium and banks’ willingness to lend
recession in the spring of 2009. Business loans outstanding, in contrast, began

to decline four quarters after the onset of the recession. Lastly, note that unused
commitments typically begin to expand several quarters before a resumption in
the growth of business loans outstanding.
An obvious question to consider next is the extent to which the contrac-
tion in unused commitments in the early stages of the business cycle is due to
demand or supply factors. On the supply side, banks may reduce unused com-
mitments by canceling (or reducing) their customers’ existing lines of credit and
by restricting the supply of new commitments. On the demand side, firms may
draw on their existing commitments to satisfy countercyclical liquidity needs
associated with the financing of inventories and trade payables. As we have
argued above, we view movements in the excess bond premium as a timely
and reliable indicator of changes in risk attitudes in the financial intermedi-
ary sector, with an increase in risk-aversion leading to a tightening of lending
standards and an eventual reduction in credit availability. The link between
the excess bond premium and banks’ willingness to lend is made explicit in
Figure 8.8, which plots the excess bond premium against the change in lending
standards on C&I loans obtained from the Federal Reserve Board’s Senior Loan
Officer Opinion Survey on Bank Lending Practices (SLOOS).16 According to the
figure, these two indicators contain essentially the same information regarding
the availability of business credit, though there is some evidence that changes
in lending standards precede movements in the excess bond premium by a few
months.
Percentage points Per cent

3 30
Monthly
2 20
1 10
0 0
–1 –10
–2 Excess bond premium (left scale) –20

Change in business unused commitments
(right scale)
–3 Change in business loans (right scale) –30
–4 –40
2005 2006 2007 2008 2009 2010
Figure 8.9 The excess bond premium and business lending, 2005–2010
In Figure 8.9, we focus on the recent events by plotting the excess bond
premium against the growth of unused commitments and business loans out-
standing over the 2005–10 period. This figure highlights the fact that the
contraction in unused commitments is entirely coincident with the rise in
the excess bond premium that began in the second half of 2007. Moreover,
unused commitments plummeted as the excess bond premium shot up during
the height of the financial market turmoil in late 2008. To the extent that fluctu-
ations in the excess bond premium reflect changes in credit supply conditions,
this coincident behavior strongly suggests that adverse credit supply dynamics
played an important role during the most recent economic downturn.
To investigate this issue more systematically, we estimate a VAR in which
shocks to the excess bond premium have a separate effect on the two compo-
nents of business lending capacity. Because data on unused commitments are
available only from 1990 onwards, we consider a relatively parsimonious speci-
fication that includes the log-difference of real GDP, the excess bond premium,
the log-difference of real business loans outstanding, the log-difference of real
unused business loan commitments, the real 10-year Treasury yield, and the real
federal funds rate.17 As credit supply shocks, we again consider orthogonalized
innovations to the excess bond premium, using a recursive ordering scheme in
which the excess bond premium is ordered after output growth but before all
other endogenous variables of the system.
The impulse responses of the endogenous variables to the excess bond pre-
mium shock based on the VAR(2) model are shown in Figure 8.10. The effects
of a financial shock on the real economy and interest rates are substantial and
Real GDP Excess bond premium

Percentage points Percentage points
0.2 0.4
0.0
0.3
–0.2
0.2
–0.4
–0.6 0.1
–0.8
0.0
–1.0
–0.1
–1.2
–1.4 –0.2
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Quarters after the stock Quarters after the stock
Real business loans Real business unused commitments

2 2
0 0
–2 –2
–4 –4
–6 –6
–8 –8
–10 –10
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Real 10-year Treasury yield Real federal funds rate

0.05 0.2
0.00
0.0
–0.05
–0.10 –0.2
–0.15
–0.20 –0.4
–0.25
–0.6
–0.30
–0.35 –0.8
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Figure 8.10 Credit supply shocks and business lending capacity
similar to those shown in Figure 8.5 and are entirely in line with the quarterly
VAR results reported by Gilchrist and Zakrajšek (2012). Also consistent with
the results reported in Figure 8.5 is the fact that business loans outstanding
exhibit no immediate response to a shock to the excess bond premium, though
business loans eventually fall about three percentage points relative to trend
growth. This is an appreciably larger decline than the one documented earlier,
and it primarily reflects the limited time series range of the data, a time frame
in which financial disturbances associated with the past three recessions played
an especially prominent role. In contrast to the response of business loans out-

standing, unused commitments fall almost a full percentage point upon the
impact of the financial shock and eventually decline more than five percentage
points relative to trend growth.
In summary, our results indicate that both business loans outstanding and
corresponding unused loan commitments are highly responsive to changes in
credit supply conditions, both during the most recent financial crisis and, as
evidenced by the VAR results, more generally in response to credit supply shocks,
as measured by the innovations to the excess bond premium. According to
our results, the initial effect of a credit supply shock manifests itself in a sharp
reduction in unused loan commitments and only with an appreciable lag does
it lead to a reduction in loans outstanding. The immediate decline in unused
commitments in response to an adverse credit supply shock implies a substantial
reduction in the capacity of businesses to borrow from the commercial banking
sector, a development that contributes significantly to an ensuing economic
downturn.
4 Conclusion
The 2007–09 financial crisis highlighted the importance of the health of finan-
cial intermediaries for macroeconomic outcomes. This chapter examined one
facet of this relationship, namely the link between financial market stress and
bank lending. Our results indicate that financial market strains – as measured
by a rise in the excess bond premium – preceded the decline in bank lend-
ing by a significant margin and that fluctuations in the excess bond premium
provided a reliable and timely gauge of changes in credit supply conditions
during the crisis. In spite of the early deterioration in credit supply condi-
tions, core loans outstanding increased noticeably during the initial phase of the
crisis.
To help elucidate the link between financial distress and bank lending – both
during the financial crisis and as a general response to adverse credit supply
shocks – we analyzed the joint dynamics of business loans outstanding and the
corresponding unused commitments. According to our results, banks’ initial
reaction to disruptions in financial markets is to reduce their off-balance-sheet
credit exposures – by cutting their customers’ unused loan commitments –
while the decline in business loans outstanding occurs with a substantial delay.
These differential dynamics between on- and off-balance-sheet credit exposures
have important implications for policy-makers who must rely on the available
information on credit prices and quantities to gauge the severity of credit mar-
ket disruptions and their potential effects on the macroeconomy. Our results
also suggest that macroprudential policy should take into account banks’ off-
balance-sheet exposures created by unused commitments when assessing the
overall risk of the financial sector.
Notes
∗ This paper was prepared for the International Economic Association World Congress
held July 4–8, 2011, in Beijing China. We thank Bill Bassett, Hesna Genay, and
Gretchen Weinbach for helpful discussions. Michael Levere and Ben Rump provided
outstanding research assistance. The views expressed in this paper are solely the
responsibility of the authors and should not be interpreted as reflecting the views
of the Board of Governors of the Federal Reserve System or of anyone else associated
with the Federal Reserve System.
1. Based on the Federal Reserve’s H.8 Statistical Release, ‘Assets and Liabilities of
Commercial Banks in the United States’, 29 July 2011.
2. Empirical studies documenting the macroeconomic effects of adverse shocks to bank
loan supply include, among others, Bernanke and Lown (1991), Peek and Rosengren
(1995, 1997, 2000), Ashcraft (2005), Lown and Morgan (2006), and Bassett et al.
(2012).
3. An example of such a shock is an unanticipated change in the stance of monetary
policy. A large body of research has focused on whether monetary policy might have
effects on real activity through the market for bank loans: if banks were not able
to readily substitute other sources of funding for deposits, then changes in the fed-
eral funds rate – which affect banks’ opportunity cost of issuing certain kinds of
deposits – would influence the price and supply of bank loans. In turn, this change
in credit market conditions would affect investment and consumption decisions of
bank-dependent borrowers. However, monetary policy shocks also affect consump-
tion and investment through their influence on other interest rates. Thus, parsing the
marginal effect of monetary policy shocks on economic activity through the market
for bank loans requires additional identifying assumptions, which, in practice, are
difficult to come by; see, for example, Bernanke and Blinder (1988) and Kashyap and
Stein (1994, 2000).
4. A portion of the decline in unused commitments, of course, reflected drawdowns
on the existing lines by businesses, which, at the same time, boosts the amount of
loans outstanding on banks’ balance sheets. The existing data on credit flows through
the banking sector, however, are inadequate to parse out these two effects; for a
more thorough discussion of the difficulties surrounding the measurement of the
underlying credit intermediated by banks during cyclical downturns, see Bassett et al.
(2011).
Note: The panels of Figure 8.2 depict the behavior of the major categories of credit
outstanding to households and nonfinancial businesses around NBER-dated business
cycle peaks. Each category of credit is normalized by nominal GDP; the logarithm of
each credit ratio was detrended using linear and quadratic time trends. For each credit
category, the average cyclical component (the black lines) and the range of cyclical
components (the shaded bands) are based on data for recessions designated by the
NBER since 1953, excluding the 2007–09 downturn.
Note: Figure 8.2 depicts the major components of loans outstanding at commercial
banks, expressed as a share of total debt outstanding in that category. Shaded vertical
bars represent NBER-dated recessions.
5. For a more thorough and complete discussion of these issues, see Boyd and Gertler
(1994).
6. Non-bank lenders active in the commercial space include business factors (non-
depository institutions that specialize in receivable-based financing and inventory
finance), credit card companies, finance and leasing groups, and private and pub-
lic syndicates. These lenders offer a range of business loans and some of them even
specialize in lending to specific industries.
7. Although our analysis is focused on the commercial banking sector, we note that the
general cyclical patterns of these four loan categories at banks are very similar to those
at all depository institutions.
8. As shown by Gertler and Gilchrist (1993) the expansion in C&I loans outstanding dur-
ing the initial stages of an economic downturn is driven primarily by differences in
borrowing between small and large firms. Because of a strong countercyclical demand
for short-term credit to finance inventory accumulation in the first few quarters fol-
lowing a business cycle turning point, firms, in general, would like to increase their
borrowing to smooth the effects of declining cash flows. However, only firms with rel-
atively unimpeded access to credit markets – typically large firms – are able to obtain
the desired funds at prevailing market rates. In such circumstances, large ‘high-quality’
borrowers tap the commercial paper market (Calomiris et al. 1995) or draw down their
lines of credit (Morgan 1988).
Note: The panels of Figure 8.3 depict the behavior of the major categories of
loans outstanding – held on the books of commercial banks – to households and
nonfinancial businesses around NBER-dated business cycle peaks. Each loan category
is normalized by nominal GDP; the logarithm of each loan ratio was detrended using
linear and quadratic time trends. For each loan category, the average cyclical com-
ponent (the black lines) and the range of cyclical components (the shaded bands)
are based on data for recessions designated by the NBER since 1953, excluding the
2007–09 downturn.
Note: Figure 8.3 depicts the estimated excess bond premium (see Gilchrist and
Zakrajšek 2012 for details). The shaded vertical bars represent the NBER-dated
recessions.
9. According to Adrian and Shin (2010), broker-dealers are financial institutions that
buy and sell securities for a fee, hold an inventory of securities for resale, and differ
from other types of institutional investors by their active procyclical management
of leverage. As documented by Adrian and Shin (2010), expansions in broker-dealer
assets are associated with increases in leverage as broker-dealers take advantage of
greater balance sheet capacity; conversely, contractions in their assets are associated
with deleveraging of their balance sheets.
10. Business loans include C&I loans and business loans secured by commercial real estate;
household loans include residential mortgages, credit card loans, and other consumer
loans. All series were obtained from the Federal Reserve’s H.8 Statistical Release. The
household and business loans outstanding were deflated by the PCE price deflator.
Note: The panels of Figure 8.5 depict the impulse responses to a 1 standard devi-
ation orthogonalized shock to the excess bond premium (see text for details). The
responses of the change in the unemployment rate, the log-difference of industrial
production, PCE price inflation, the log-difference of the price-dividend ratio, the
log-difference of real business loans, and the log-difference of real household loans
have been accumulated. Shaded bands denote 95 per cent confidence intervals based
on 1,000 bootstrap replications.
11. The starting date of the estimation period is dictated by the availability of the monthly
bank lending data.
12. As a robustness check, we also included macroeconomic uncertainty – as proxied by
either the option implied (that is, the VXO index) or realized stock market volatility –
into the ‘fast-moving’ block of our VAR specification. The inclusion of this variable
had no effect on the impulse responses reported in Figure 8.5.
13. Another option, of course, is to rely on internal liquidity. As shown by Acharya et al.
(2009), firms with greater exposure to aggregate risk find it more costly to obtain
credit lines from banks and, as a result, tend to rely more heavily on cash reserves to
manage their future liquidity needs.
Note: The black line in the top panel of Figure 8.6 depicts the dollar amount of
core unused commitments, and the dotted line depicts the dollar amount of core
loans outstanding at US commercial banks. Core loan categories include C&I, real
estate, and consumer loans. The bottom panel depicts the composition of unused
commitments. All series are deflated by the GDP price deflator (2005 = 100). Shaded
vertical bars represent NBER-dated recessions.
14. Date on unused commitment were added to Call Reports in 1990:Q2.
15. It is important to note that what we label as ‘business lines’ is recorded in Call Reports
prior to 2010 as ‘other’ unused commitments. More detailed data available since 2010
suggest that credit lines to businesses – both financial and nonfinancial – account for
the vast majority of this category, which indicates that these data provide a useful
proxy for unused credit lines to businesses.
Note: The black line in Figure 8.7 depicts the seasonally adjusted (annualized) quar-
terly growth business lending capacity at US commercial banks; lending capacity is
defined as the sum of business loans outstanding and corresponding unused commit-
ments. All series are deflated by the GDP price deflator (2005 = 100). Shaded vertical
bars represent NBER-dated recessions.
16. The SLOOS is usually conducted four times per year by the Federal Reserve Board,
and up to 60 banks participate in each survey. Banks are asked to report whether they
have changed their credit standards over the past three months on the major cate-
gories of loans to businesses and households. The series plotted is the net percentage
of banks that reported tightening their credit standards on C&I loans to large and
middle-market firms. Reported net per cent equals the per cent of banks that reported
tightening their standards minus the per cent that reported easing their standards.
The SLOOS data are plotted in such a way that the change in C&I credit standards
over the survey period is aligned with the quarter in which the reported change took
place. For the full text of the questions and more information on the survey, see
www.federalreserve.gov/boarddocs/SnLoanSurvey/.
Note: The black line in Figure 8.9 depicts the excess bond premium. The short-dash
line with overlaid dots depicts the (annualized) quarterly growth of unused business
commitments, while the long-dash line with overlaid dots depicts the (annualized)
quarterly growth of business loans outstanding. The shaded vertical bar denotes the
2007–09 NBER-dated recession.
17. By specifying the VAR in ‘real’ terms, we abstract from inflationary dynamics, which
are not the focus of our analysis. Both the real federal funds rate and the real
10-year Treasury yield are defined as the corresponding nominal rate less average
CPI inflation over the next ten years, as reported by the Survey of Professional
Forecasters.
Note: The panels of Figure 8.10 depict the impulse responses to a 1 standard devi-
ation orthogonalized shock to the excess bond premium (see text for details). The
responses of the log-difference of real GDP, the log-difference of real business loans,
and the log-difference of real unused commitments have been accumulated. Shaded
bands denote 95 per cent confidence intervals based on 1,000 bootstrap replications.
References
Acharya, V. V., H. Almeida and M. Campello (2009) ‘Aggregate Risk and the Choice
Between Cash and Credit Lines’, Working Paper, Department of Finance, University
of Illinois at Urbana-Champaign.
Adrian, T. and H. S. Shin (2010) ‘Liquidity and Leverage’, Journal of Financial Intermediation,
vol. 19, no. 3, pp. 418–437.
Ashcraft, A. B. (2005) ‘Are Banks Really Special? New Evidence from the FDIC-
Induced Failure of Healthy Banks’, American Economic Review, vol. 95, no. 5,
pp. 1712–1730.
Bassett, W. F., M. B. Chosak, J. C. Driscoll and E. Zakrajšek (2012) ‘Changes in Bank
Lending Standards and the Macroeconomy’, Finance and Economics Discussion Series
Paper 2012–24, Federal Reserve Board.
Bassett, W. F., S. Gilchrist, G. C. Weinbach and E. Zakrajšek (2011) ‘Improving Our Ability
to Monitor Bank Lending’ forthcoming in M. K. Brunnermeier and A. Krishnamurthy
(eds), Systemic Risk and Macro Modeling (Washington, DC: NBER). Available at SSRN at:
http://ssrn.com/abstract=1943992.
Bernanke, B. S. (1983) ‘Nonmonetary Effects of the Financial Crisis in Propagation of the
Great Depression’, American Economic Review, vol. 73, no. 3, pp. 257–276.
Bernanke, B. S. and A. S. Blinder (1988) ‘Credit, Money, and Aggregate Demand’, American
Economic Review, vol. 78, no. 2, pp. 435–439.
Bernanke, B. S. and C. S. Lown (1991) ‘The Credit Crunch’, Brookings Papers on Economic
Activity, vol. 22, no. 2, pp. 205–239.
Boyd, J. H. and M. Gertler (1994) ‘Are Banks Dead? Or Are the Reports Greatly
Exagerrated?’, Federal Reserve Bank of Minneapolis Quarterly Review, 18 (Sum), pp. 2–23.
Brunner, K. and A. H. Meltzer (1963) ‘The Place of Financial Intermediaries in the
Transmission of Monetary Policy’, American Economic Review, vol. 53, no. 2,
pp. 372–382.
Calomiris, C. W., C. P. Himmelberg and P. Wachtel (1995) ‘Commercial Paper, Corpo-
rate Finance, and the Business Cycle: A Microeconomic Perspective’, Carnegie-Rochester
Conference Series on Public Policy, vol. 42, no. 1, pp. 203–250.
Chari, V. V., L. Christiano and P. J. Kehoe (2008) ‘Facts and Myths About the Financial
Crisis of 2008’, Federal Reserve Bank of Minneapolis Working Paper No. 666.
Gertler, M. and S. Gilchrist (1993) ‘The Role of Credit Market Imperfections in the
Monetary Transmission Mechanism: Arguments and Evidence’, Scandinavian Journal
of Economics, vol. 95, no. 1, pp. 43–64.
Gilchrist, S. and E. Zakrajšek (2012) ‘Credit Spreads and Business Cycle Fluctuations’,
NBER Working Paper No. 17021, forthcoming, American Economic Review, vol. 102,
no. 4, pp. 1692–1720.
Ivashina, V. and D. Scharfstein (2010) ‘Bank Lending During the Financial Crisis of 2008’,
Journal of Financial Economics, vol. 97, no. 3, pp. 319–338.
Kashyap, A. K. and J. C. Stein (1994) ‘Monetary Policy and Bank Lending’, in N. G. Mankiw
(ed.), Monetary Policy (Chicago: University of Chicago Press), pp. 221–262.
Kashyap, A. K. and J. C. Stein (2000) ‘What Do a Million Observations on Banks Say

about the Transmission of Monetary Policy?’, American Economic Review, vol. 90, no. 3,
pp. 407–428.
Lown, C. S. and D. P. Morgan (2006) ‘The Credit Cycle and the Business Cycle: New
Findings from the Loan Officer Opinion Survey’, Journal of Money, Credit, and Banking,
vol. 38, no. 6, pp. 1575–1597.
Merton, R. C. (1974) ‘On the Pricing of Corporate Debt: The Risk Structure of Interest
Rates’, Journal of Finance, vol. 29, no. 2, pp. 449–470.
Morgan, D. P. (1998) ‘The Credit Effects of Monetary Policy: Evidence From Using
Loan Commitments’, Journal of Money, Credit, and Banking, vol. 30, no. 1,
pp. 102–118.
Peek, J. and E. S. Rosengren (1995) ‘The Capital Crunch: Neither a Borrower nor a Lender
Be’, Journal of Money, Credit, and Banking, vol. 27, no. 3, pp. 625–638.
Peek, J. and E. S. Rosengren (1997) ‘The International Transmission of Financial Shocks’,
Peek, J. and E. S. Rosengren (2000) ‘Collateral Damage: Effects of the Japanese Bank Crisis
on Real Activity in the United States’, American Economic Review, vol. 90, no. 1, pp.
30–45.
9
A Mechanism Design Approach
to Financial Frictions∗
Nobuhiro Kiyotaki
1 Introduction
In a thought-provoking article ‘Can a “Credit Crunch” Be Efficient?’ Edward

Green and Soo Nam Oh use a mechanism design approach to present a model of
financial intermediation in which phenomena qualitatively resembling a ‘credit
crunch’ occur but are efficient. In this short chapter, we extend and modify the
model of Green and Oh in order to examine how different environments of
private information and limited commitment generate different financial fric-
tions. Following a tradition of mechanism design, which considers the market
structure as an equilibrium outcome of the underlying environment, we ask
questions such as: Which markets are open? Which contracts are used? Which
institutions arise? We find that the model of Green and Oh is a useful benchmark
to explain the recent literature on the mechanism design approach to financial
frictions.
2 Green and Oh’s model
There is a continuum of agents with population size being normalized to one.

All agents live for three dates, dates 0, 1, and 2. There is one homogeneous
goods at each date. At date 0, everyone is identical and signs a contract. At
date 1, an idiosyncratic income shock realizes, and each agent receives either a
high income (or endowment) yh with probability πh , or receives a low income
yl with probability πl , where 0 < yl < yh and πh + πl = 1. We assume that exactly
a fraction πh of agents receives the high income and that a fraction πl receives
the low income so that there is no uncertainty about the aggregate income
ȳ = πh yh + πl yl at date 1. At date 2, everyone receives an identical income of z,
where
0 < yl < z < ȳ < yh .
177
178 N. Kiyotaki
The agent enjoys utility of consuming goods at dates 1 and 2, and his or her
preferences at date 0 are determined by the expected utility

EU = πi [u(c1i ) + βu(c2i )], (1)
i=h,l
where cti denotes date t consumption of the agent whose date 1 income is
yi (i = h, l). β ∈ (0, 1) denotes the common discount factor. We assume u(c) satis-
fies the usual regularity conditions: u (c) > 0, u (c) < 0, u (0) = ∞ and u (∞) = 0.
Agents can store goods which allow them to transform x units of goods stored
at date 1 into (1/β) · x units of goods at date 2.
2.1 Public information and full commitment

Before introducing the financial frictions, let us examine the economy in which
all the information about individual income and storage is public information
and the individuals can fully commit to obey the contract (that is, there is no
limitation on enforcing contracts). Then the competitive economy corresponds
to a solution of the planner’s problem in which consumption and the storage
(c1h , c2h , c1l , c2l , x) are chosen to maximize the expected utility of a typical agent
(1) subject to the resource constraints:
πh c1h + πl c1l = ȳ − x,
1
πh c2h + πl c2l = z + x.
β
When the storage is non-negative, the resource constraints can be combined as
πh (c1h + βc2h ) + πl (c1l + βc2l ) = ȳ + βz, (2)
and the solution is

1
c1h = c1l = c2h = c2l = ȳ + βz ,
1+β
β
x= ȳ − z > 0.
1+β
Thus, if there is no friction of information and commitment, individual con-

sumption does not depend upon idiosyncratic income shock, and depends
only upon aggregate income, because agents can perfectly insure against the
idiosyncratic income risks. With perfect risk-sharing, the marginal utility of
consumption is equalized across agents. When all agents have identical prefer-
ences, consumption is equalized too. Moreover, everyone’s consumption should
be smoothed over time through storage (whose rate of return is equal to the time
preference rate 1/β).
A Mechanism Design Approach to Financial Frictions 179
The solution of this planner’s problem can be considered as an outcome of a

competitive economy in which intermediaries compete with each other to offer
contracts for state-contingent net transfers. Under Bertrand-style competition
between intermediaries, the contract offered by all intermediaries in equilibrium
would be the contract which maximizes the expected utility of the customers
subject to the resource constraints. This frictionless competitive economy –
the so-called Arrow-Debreu economy – serves as a benchmark. However, the
prediction appears to contradict the observation that individual consumption
significantly depends upon the individual’s income (in addition to the aggre-
gate income per capita) in household panel data. (See, for example, Altug and
Miller 1990; Cochrane 1991; Mace 1991). The next sections depart from such a
frictionless economy in order to explain the household data.
2.2 Private information of individual income

Green and Oh (1991), which is based upon a classic paper of Green (1987),
considers private information about individual income as a key friction to
explain phenomena like ‘credit’. If individual income is private information,
the perfect risk-sharing acheived in the Arrow-Debreu economy is no longer
compatible with the incentive constraints since everyone would claim to have
earned the low income to receive a positive transfer. Thus when individual
income is private information, the allocation rule has to satisfy an incentive con-
straint which requires that high-income agents not pretend to be a low-income
agent:
u(c1h ) + βu(c2h ) ≥ u(yh + c1l − yl ) + βu(c2l ). (3)
The left hand side (LHS) denotes the utility when the high-income agent tells
the truth to the intermediary about her income. The right hand side (RHS)
denotes instead the utility when the high-income agent misrepresents herself
as a low-income agent. By pretending to be a low-income agent, she receives
the transfer c1l − yl and consumes yh + c1l − yl at date 1, and pays z − cl and
consumes z − (z − cl ) at date 2. To derive this, we have assumed that the storage
was public information so that she cannot use storage privately in order to adjust
consumption across dates.
Thus, in this environment, the optimal contract (c1h , c2h , c1l , c2l ) can be found
by maximizing the expected utility subject to the resource constraint and the
incentive constraint (3). Hence, the associated Lagrangian is
L = πh [u(c1h ) + βu(c2h )] + πl [u(c1l ) + βu(c2l )]

+ λ[ȳ + βz − πh (c1h + βc2h ) − πl (c1l + βc2l )]
+ μh [u(c1h ) + βu(c2h ) − u(yh + c1l − yl ) − βu(c2l )]
180 N. Kiyotaki
where λ and μh are the Lagrangian multipliers associated to the resource con-
straint and the incentive constraint. The first order conditions of such a problem
require

μ μ
1 + h u (c1h ) = λ = 1 + h u (c2h )
πh πh

μh μh
u (c1l ) − u (yh + c1l − yl ) = λ = 1 − u (c2l ).
πl πl
Then, since u (yh + c1l − yl ) < u (c1l ), we obtain that
c2l < c1l < c1h = c2h .
In this environment, a ‘credit’-like arrangement arises as an endogenous

outcome of the private information about the individual income. In order to
prevent the high-income agent from pretending to be a low-income agent, the
intermediary cannot transfer a positive amount to the low-income agent in both
periods. Those who consume more than their income at present (like borrowers)
have to pay and consume less than their income in future. Thus, agents can-
not be fully insured, since low-income agents consume less than high-income
agents.
Moreover, although consumption is smooth over time for the high-income
agent, consumption of the low-income agent is skewed towards date 1. This
is because skewing the consumption of low-income agents towards date 1
reduces the gains of the high-income agent from misrepresenting herself as a
low-income agent, and in turn, relaxes the incentive constraint.
Figure 9.1 illustrates the allocation when individual income is private infor-
mation. The horizontal axis measures the date 1 income and consumption,
while the vertical axis measures those of date 2. Under public information and
full commitment, the competitive economy achieves the first best allocation in
which consumption is equalized across agents and across time at point C∗ in the
figure. When the individual income is private information, the consumption of
the low-income agent is at point Cl , in which consumption is larger than income
at date 1 but it is smaller than income at date 2 (in order to repay the ‘debt’).
Moreover, consumption at date 1 is larger than consumption at date 2 (c1l > c2l )
for the low-income agent. Consumption of the high-income agent is balanced
at point Ch . Point Ch is consumption if the high-income agent pretends to be a
low income agent to receive the net transfer
Ch = Yh + (Cl − Yl ),
where capital letters denote vectors such as Cl = (c1l , c2l ). Point Ch and point Ch
are on the same indifference curve as the incentive constraint is binding for the
high-income agent.
c2, y2
45-degree line
Ch
c2h
c2* C*
Yh
z
Yl
c2l
Cl Ch'
c1, y1
y1l c1l c1* c1h y1h
Figure 9.1 The Green-Oh model
When we compare the present value of consumption and income, we learn

that there is a transfer from the high-income agents to the low-income agents.
Both consumption allocations Ch and Cl are closer to the first best allocation
C∗ than income Yh and Yl in terms of the present value. But consumption is
not equalized across agents, because insurance must be imperfect with private
information.
3 Private information of individual income and

storage (Cole-Kocherlakota)
Cole and Kocherlakota (2001) consider an environment in which not only

individual income but also storage (saving) are private information. Then the
high-income agent who pretends to be a low-income agent can use storage pri-
vately in order to smooth consumption. Let V (W) be date 1 value function,
when the agent chooses consumption and storage in order to maximize the
utility subject to the constraint that the present value of consumption is equal
to W, that is,
V (W) = Max[u(c1 ) + βu(c2 )],

c1 ,c2
subject to W = c1 + βc2 .
By the envelope theorem, the value function satisfies V (W) = u (c1 ) = u (c2 ).
The incentive constraint of the high income agent has to be modified to take
182 N. Kiyotaki
into account the additional friction as
u(c1h ) + βu(c2h ) ≥ V (yh − yl + c1l + βc2l ). (4)
The LHS denotes the utility of the high-income agent who tells the truth. In
the RHS, the utility of the high-income agent who misrepresents herself to be
a low-income agent is the function of her wealth, which is the sum of the date
1 income gap (that the high-income agent hides) and of the present value of
consumption of the low-income agent.
In this setup, the optimal contract (c1h , c2h , c1l , c2l ) maximizes the expected
utility subject to the resource constraint and the incentive constraint (4). Using
the Lagrangian

+ μh [u(c1h ) + βu(c2h ) − V (yh − yl + c1l + βc2l )],
the first order conditions can be arranged as

μ μ
1 + h u (c1h ) = λ = 1 + h u (c2h )
πh πh
μh
λ+ V (yh − yl + c1l + βc2l ) = u (c1l ) = u (c2l ).
πl
Thus we have
1
c1h = c2h = y + βz ,
1+β h
1
c1l = c2l = y + βz .
1+β l
The present value of consumption of each agent is equal to the present value
of his or her income here. There is no insurance nor transfer of wealth across
agents. Because the high-income agent can announce the income which maxi-
mizes the present value of net transfer from the intermediary, the present value
of net transfer must be zero for both high-income and low-income agents. The
intermediary cannot cross-subsidize agents, because all agents would choose to
receive the subsidy and not to pay to the intermediary. On the other hand,
consumption is smooth over time. Therefore, Cole and Kocherlakota (2001)
provide a mechanism design foundation of the ‘permanent income theory of
consumption’ of Friedman (1967) – individual consumption reacts to idiosyn-
cratic income shocks, even though agents can smooth consumption over time.
This result is valuable because the optimal contract becomes simpler and more
‘realistic’ when private information about storage is added to a model with
private information about income.
4 Limited commitment
In a decentralized market economy, people often do not keep their promises

and the intermediary cannot enforce contracts completely. This is a problem of
limited commitment on the side of the agents and limited contract enforcement
on the side of the intermediary. Suppose that the income is paid directly to the
individual who cannot commit to pay a large fraction of income in future. It is
not easy for the intermediary to enforce the individual to pay a large amount
which contracts specify. In this setup the financial friction can arise endoge-
nously. In fact, such limitation in commitment and enforcement can generate
financial friction even if the individual income is public information.
To be more specific, suppose that the agent will not pay more than θ ∈ (0, 1)
fraction of present and future income, and that nobody can take away any
fraction of the individual’s storage. Then consumption of the low-income agent
cannot be smaller than 1 − θ fraction of income at date 2:
c2l ≥ (1 − θ )z, (5)
because the intermediary cannot force her to pay more than θz. At date 1, the
high-income agent will not give up more than θ fraction of the wealth. The
incentive constraint for the high-income agent to follow to her intermediary’s
specified net transfer requires
u(c1h ) + βu(c2h ) ≥ V ((1 − θ )(yh + βz)). (6)
The optimal contract (c1h , c2h , c1l , c2l ) maximizes the expected utility of a typical
agent subject to the resource constraint and the two incentive constraints (5,
6). If θ is sufficiently small, both incentive constraints are binding and thus we
have:
1
c1h = c2h = (1 − θ )(yh + βz)
1+β
c2l = (1 − θ )z
θ (ȳ + βz)
c1l = (1 − θ )yl + . (7)
πl
The RHS of the first equation is permanent income of the high-income agent
which no intermediary can take away – Holmstrom and Tirole (1999) call it
‘non-pledgeable’ income. The second equation says consumption of the low-
income agent is equal to her non-pledgeable income at date 2. In the RHS of
the last equation, the first term is the non-pledgeable income of the low-income
agent. The numerator of the second term is the fraction of the aggregate wealth
which the intermediary can reallocate across agents – Holmstrom and Tirole
call it ‘pledgeable’ wealth.1 Thus the optimal contract under such a severe lim-
itation of contract enforcement requires the intermediary to allocate all the
184 N. Kiyotaki
pledgeable wealth to the most needy agents, that is, the low-income agents at
date 1. Here, unlike the previous examples of the private information economy,
the low-income agent faces a binding contract enforcement constraint (loosely
speaking, the bankruptcy constraint). A distinctive characteristic of an economy
with limited commitment (but without private information) is that, although
the resource base which contracts can reallocate is limited, there is no further
restriction on how contracts reallocate the pledgeable wealth.
5 Limited commitment and private information
What happens if the individual income and storage are private information
and the individual cannot commit to pay a large amount to the intermediary
in future? Formally, the optimal contract (c1h , c2h , c1l , c2l ) would maximize the
expected utility subject to the resource constraint and the three incentive con-
straints (4,5,6). As argued above, with the private information about individual
income and storage, the intermediary cannot cross-subsidize agents, and the
present value of net transfer must be zero for all agents. Thus (6) is not binding.
Then, using the Lagrangian

+ μh [u(c1h ) + βu(c2h ) − V (yh − yl + c1l + βc2l )]
+ μl [c2l − (1 − θ )z]
the first order conditions can be arranged as

μ μ
1 + h u (c1h ) = λ = 1 + h u (c2h )
πh πh
μh
u (c1l ) = λ + V (yh − yl + c1l + βc2l )
πl
μ μ
u (c2l ) = λ − l + h V (yh − yl + c1l + βc2l ).
βπl πl
Then when θ is small enough, we have
1
c1h = c2h = y + βz ,
1+β h
c1l = yl + βθz, (8)
c2l = (1 − θ )z.
Thus, as in permanent income theory, the present value of the individual con-
sumption is equal to the individual income – there is no insurance. Moreover,
the low-income agent faces a binding borrowing constraint at date 1. He can
borrow only against θ fraction of pledgeable future income. His date 1 con-
sumption is equal to his current income and the present value of his pledgeable
future income in (8). Therefore limitations in commitment and private informa-
tion of income and saving lead to a permanent income theory of consumption
with borrowing constraint – arguably the most common contract we observe
in data.
In Figure 9.2, the points Yh and Yl show income of the high-income agent and
the low-income agent. If the individual income and storage are private informa-
tion but the individual can commit to pay in future, then the consumption of
the high-income agent is Ch = (c1h , c2h ) on the 45-degree line and the present val-
ues of consumption and income are equal (the line Yh Ch has a slope of −(1/β)).
The consumption of the low-income agent is Cl = (c1l , c2l ) on the 45-degree line,
and again the present values of consumption and income are the same. This is
a simplified version of Cole and Kocherlakota (2001).
If, in addition to the private information of the individual income and storage,
the individual cannot commit to pay more than θ fraction of future income,
then the consumption of the low-income agent becomes Cl = (c1l , (1 − θ)z). The
low-income agent wants to borrow as much as c1l − yl , but can only borrow
up to c1l − y at date 1 because he can commit to pay only θz at date 2. The
l
consumption of the high-income agent is unchanged at Ch . The high-income
agent is not contained in her borrowing, because she lends to the intermediary
instead of borrowing at date 1.2
c2, y2 45-degree line
c2h Ch
Yl
Yh
z
(1-theta)z Cl'
c2l Cl
c1, y1
yl c1I' c1l c1h y1h
Figure 9.2 Hidden storage and limited commitment

186 N. Kiyotaki
6 Concluding remark
In this chapter, we illustrate how ‘credit’-like arrangements arise endogenously

and take a particular form in response to both private information of individual
income and storage, and limited commitment. So far we have ignored the prob-
lem of the incentive constraint of the intermediary. But how do we know
that the intermediaries are trustworthy? What happens if the intermediary
has private information or limited commitment? There is some literature on
this.3 Given that financial intermediaries experienced significant financing con-
straints during the recent financial crisis, we expect that active research will
take place in the near future which accounts for incentive constraints of the
intermediary in the tradition of mechanism design and general equilibrium
literature.
Notes
∗ This paper was originally prepared for the International Economic Association 2011
World Congress held on 4–8 July 2011 in Beijing. I would like to thank Benjamin Moll
and Francesco Nava for thoughtful comments.
1. Kiyotaki and Moore (1997) consider an economy in which, instead of a fraction of
future income, fixed assets such as real estate become the pledgeable wealth (collateral),
exploring the interaction between the collateral value and aggregate production.
2. See Townsend (1989), for example, for the early literature on the related topic.
Ljungqvist and Sargent (2004) explain recent developments of optimal contract
literature in infinite horizon frameworks.
3. There is a vast literature on incentive constraint of the financial intermediaries from
the perspective of microeconomics of banking; see Freixas and Rochet (1998). For a
more mechanism design and/or general equilibrium tradition, see, for example, Krasa
and Villamil (1992), Holmstrom and Tirole (1997), Gertler and Karadi (2011), Gertler
and Kiyotaki (2010) and Gertler et al. (2011).
References
Altug, S. and R. Miller (1990) ‘Household Choices in Equilibrium’, Econometrica, vol. 58,
no. 3, pp. 543–570.
Cochrane, J. (1991) ‘A Simple Test of Consumption Insurance’, Journal of Political Economy,
vol. 99, no. 5, pp. 957–976.
Cole, H. and N. Kocherlakota (2001) ‘Efficient Allocations with Hidden Income and
Hidden Storage’, Review of Economic Studies, vol. 68, no. 3, pp. 523–542.
Freixas, X. and J.-C. Rochet (2008) Microeconomics of Banking, Second Edition (Cambridge,
MA: MIT Press).
Friedman, M. (1967) A Theory of Consumption Function (Princeton: Princeton University
Press).
Gertler, M. and P. Karadi (2011) ‘A Model of Unconventional Monetary Policy’, Journal of
Monetary Economics, vol. 58, no. 1, pp. 17–34.
Gertler, M. and N. Kiyotaki (2010) ‘Financial Intermediation and Credit Policy in Business
Cycle Analysis’, in B. Friedman and M. Woodford (eds), Handbook of Monetary Economics
(Amsterdam: Elsevier).
Gertler, M., N. Kiyotaki and A. Queralto (2011) ‘Financial Crises, Bank Risk Exposure and
Government Financial Policy’, Journal of Monetary Economics, forthcoming.
Green, E. (1987) ‘Lending and the Smoothing of Uninsurable Income’, in E. Prescott
and N. Wallace (eds), Contractual Arrangements for Intertemporal Trade (Minneapolis:
University of Minnesota Press), pp. 3–25.
Green, E., and S.-N. Oh (1991) ‘Can a “Credit Crunch” Be Efficient?’, Federal Reserve Bank
of Minneapolis Quarterly Review, vol. 15, no. 4, pp. 3–17.
Holmstrom, B. and J. Tirole (1997) ‘Financial Intermediation, Loanable Funds and the
Real Sector’, Quarterly Journal of Economics, vol. 112, no. 3, pp. 663–691.
Holmstrom, B. and J. Tirole (1998) ‘Private and Public Supply of Liquidity’, Journal of
Political Economy, vol. 106, no.1, pp. 1–40.
Kiyotaki, N. and J. Moore (1997) ‘Credit Cycles’, Journal of Political Economy, vol. 105,
no. 2, pp. 211–248.
Krasa, S. and A. Villamil (1992) ‘Monitoring the Monitor: An Incentive Structure for a
Financial Intermediary’, Journal of Economic Theory, vol. 57, no. 1, pp. 197–221.
Ljungqvist, L. and T. Sargent (2004) Recursive Macroeconomic Theory: Second Edition
(Cambridge, MA: MIT Press).
Mace, B. (1991) ‘Full Insurance in the Presence of Aggregate Uncertainty’, Journal of Political
Economy, vol. 99, no. 5, pp. 928–956.
Townsend, R. (1989) ‘Currency and Credit in a Private Information Economy’, Journal of
Political Economy, vol. 97, no. 6, pp. 1323–1344.
Part III
Behavior of Financial Institutions
and Prudential Regulations
10
Systemic Risk and Macroprudential
Regulation
Franklin Allen
Elena Carletti
European University Institute, Italy
1 Introduction
During the recent crisis microprudential regulation of the banking system

turned out to be unable to maintain financial stability largely because it did
not recognize the problem of systemic risk. This chapter discusses in detail
the sources of systemic risk, their importance for financial stability and the
macroprudential policies that are necessary to address them.
Systemic risk arises from the following sources:
1. Common exposure to asset price bubbles, particularly real estate bubbles;

2. Liquidity provision and mispricing of assets;
3. Multiple equilibria and panics;
4. Contagion;
5. Sovereign default;
6. Currency mismatches in the banking system.
We discuss each of these in turn in Section 2. Section 3 then considers the

macroprudential policies that can be used to counter these systemic risks.
2 Systemic risk
2.1 Real estate bubbles
Herring and Wachter (1999) and Reinhart and Rogoff (2009) provide persuasive
evidence that collapses in real estate prices, either residential or commercial
or both, are one of the major causes of financial crises. In many cases these
collapses occur after bubbles in real estate prices that are often created by loose
monetary policy and excessive availability of credit. When the bubble bursts,
the financial sector and the real economy are adversely affected.
191
192 F. Allen and E. Carletti
The current crisis provides a good example of this. Allen and Carletti (2009)
argue that the main cause of the crisis was that there was a bubble in real estate in
the US and in a number of other countries such as Spain and Ireland. When the
bubble burst in the US, many financial institutions experienced severe problems
because of the collapse in the securitized mortgage market. Problems then spread
to the real economy. Figure 10.1 shows the movement in property prices in the
US, Spain, and Ireland. It can be seen that in all three countries house prices
rose significantly and then dropped.
The real estate bubble in these countries was the result of loose monetary
policy and global imbalances that led to excessive credit availability. Central
banks, in particular in the US, set very low interest rates during the period 2003–
04 to avoid a recession after the tech bubble in 2000 and the 9/11 terrorist
attacks in 2001 at a time when house prices were already rising quite fast. As
argued by Taylor (2008), these levels of interest rates were much lower than in
previous US recessions relative to the economic indicators at the time captured
by the ‘Taylor rule’. In such an environment of low interest rates, people in
the US started to borrow and buy houses to benefit from their growing prices.
Unlike stock prices where returns follow random walks, returns on housing are
positively serially correlated, as found by Case and Shiller (1989), Englund et al.
(1998), and Glaeser and Gyourko (2007). This means that by lowering interest
rates significantly below the current rate of house price appreciation, the Fed
effectively created a profitable opportunity to buy property. Other public policies
such as the tax deductibility of interest rates on mortgages contributed further
to the housing boom.
500.00
450.00
400.00
350.00
300.00
250.00 Ireland
200.00 Spain
150.00 USA
100.00
50.00
0.00
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
Figure 10.1 Housing prices in Ireland, Spain, and the US

Systemic Risk and Macroprudential Regulation 193
As Figure 10.1 showed, Spain and Ireland also had very large run-ups in prop-
erty prices. According to Taylor (2008), these countries also had loose monetary
policies relative to the Taylor rule. Spain, which had one of the largest deviations
from the rule, also had the biggest housing boom as measured by the change
in housing investment as a share of GDP. Other countries in the Eurozone such
as Germany did not have a housing boom. Their inflation rates and other eco-
nomic indicators were such that for them the European Central Bank’s interest
rates did not correspond to a loose monetary policy.
As Allen and Gale (2000a, 2007) have argued, asset price bubbles are also
caused by growth in credit. During the recent crisis, credit expanded rapidly
in the countries with a loose monetary policy due to the presence of global
imbalances. Several Asian countries had accumulated large amounts of foreign
exchange reserves since the late 1990s. Figure 10.2 shows that much of this
acquisition of reserves was by Asian countries. The reserves in Latin American
and Central and Eastern European countries did not increase significantly. There
are a number of reasons behind this accumulation. Allen and Carletti (2009)
argue that the Asian countries affected by the crisis of 1997 started accumulat-
ing reserves in response to the tough conditions that the International Monetary
Fund imposed on them in exchange for financial assistance. The motivations
for the reserve accumulation of China, which is the largest holder, are probably
more complex than this. Besides precautionary reasons, China started accu-
mulating reserves to avoid allowing its currency to strengthen and damage its
exports as well as to increase its political power. The accumulated reserves were
mostly invested internationally. Much of it was invested in US dollars in debt
securities such as Treasuries, and Fannie and Freddie mortgage-backed securi-
ties. The large supply of debt in the US helped to drive down lending standards
4,000
3,000
2,000
1,000
0
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
Asia Latin Americal Central & Eastern Europe
Figure 10.2 A comparison of foreign exchange reserves in different regions
Source: IMF website. Asia comprises the six East Asian countries: China, Hong Kong, Japan, Singapore,
South Korea, Taiwan-China.
to ensure that there was enough demand for debt from house-buyers and other
borrowers. However, funds did not only flow to the US, Spain, and Ireland also
ran large current account deficits.
The burst of a bubble has a clear effect on the stability of the financial sector,
as documented in Herring and Wachter (1999) and Reinhart and Rogoff (2009).
In the current crisis, for example, the sudden drop in asset prices starting in the
summer of 2007 triggered by the fall in real estate prices and the large volatility
that followed worsened the balance sheets of financial institutions significantly
and froze several financial markets, including the normally stable interbank
market.
The financial crisis then spread to the real sector. The burst of a bubble can,
however, also create direct damaging effects on the real economy. In the current
crisis, for example, the burst of the property bubble in Spain led to a doubling
of unemployment without the financial sector being much affected, at least
initially. This was thanks to strict financial regulation and the use of some
macroprudential instruments such as countercyclical loan loss reserve ratios.
The fact that the burst of a bubble can affect both the financial and the real
sector significantly underlines the importance of preventing bubbles.
2.2 Liquidity provision and mispricing of assets

Another important source of systemic risk is inefficient liquidity provision and
the mispricing of assets. Asset pricing theory in financial economics relies on the
assumption of fully rational agents and perfect and complete markets. Under
these assumptions, assets are always correctly priced at their fundamental values.
Agents understand risks perfectly; financial institutions and liquidity do not play
any role and crises should not occur. The recent crisis, however, has shown the
flaws of these theories in practice. Financial markets, including money markets,
can work badly, and financial institutions and their role as liquidity creators can
be at center stage.
The theories explaining the role of liquidity in creating systemic risk (see, for
example, Allen and Gale 2004, 2007; Allen and Carletti 2006, 2008a; Allen,
Carletti and Gale 2009) combine the functioning of financial institutions and
markets in a model of liquidity. Financial intermediaries provide liquidity insur-
ance to consumers against their individual liquidity shocks. Markets allow
financial intermediaries (and hence their depositors) to share aggregate risks.
If financial markets are complete, the financial system provides liquidity effi-
ciently in that it ensures that banks’ liquidity shocks are hedged. By contrast,
in the plausible case where markets are incomplete, banks cannot hedge com-
pletely against shocks and the financial system stops providing an efficient level
of liquidity. This can generate mispricing of assets and even the prices of safe
assets can fall below their fundamental values. The reason is that with incom-
plete markets, liquidity provision is achieved by selling assets when liquidity is
required. If liquidity is scarce, asset prices are determined by the available liq-
uidity or in other words by the ‘cash-in-the-market’. Some financial institutions
must hold extra liquidity that allows them to buy up low-price assets when liq-
uidity is scarce. However, holding liquidity is costly as it prevents investment
in the more profitable long asset. With incomplete markets, the suppliers of liq-
uidity must be compensated for the cost of holding liquidity across states rather
than state by state as with complete markets. To do this, asset prices must be low
in the states where banks need more liquidity. This is inefficient as banks in need
of liquidity have to sell at a low price to the providers of liquidity. There is nega-
tive insurance and suboptimal risk-sharing. If asset prices are sufficiently low the
banks in need of liquidity will go bankrupt and this can lead to large deadweight
costs. The asset price volatility is costly because depositors are risk-averse and
their consumption varies across banks with high and low idiosyncratic liquidity
risk. This leaves scope for central bank intervention to improve welfare.
To summarize, liquidity is costly to hold and asset price volatility is neces-
sary to provide incentives to agents to hold it. But price volatility can cause
crises. When prices fall to low enough levels, financial institutions can go into
bankruptcy. There is a market failure that provides the justification for central
bank operations and other kinds of intervention to improve the allocation of
resources and to avoid crises (Allen, Carletti and Gale 2009).
2.3 Multiple equilibria and panics

The importance of panics in the current crisis is unclear. However, they remain
an important source of systemic risk and thus deserve some attention as one
of the macroeconomic aspects of systemic risk. Two distinct theories for the
origins of banking panics have been proposed in the academic literature (see
Allen, Babus and Carletti 2009 for a survey). One line of argument maintains
that panics are undesirable events caused by random deposit withdrawals unre-
lated to changes in the real economy. In the seminal work by Bryant (1980)
and Diamond and Dybvig (1983), for example, panics are self-fulfilling events.
Agents have uncertain needs for consumption and long-term investments are
costly to liquidate. They deposit their endowment in a bank in exchange for a
demand deposit contract which insures them against their liquidity needs. If all
depositors believe that the other depositors withdraw their funds only accord-
ing to their consumption needs, then the good equilibrium arises in which the
bank can satisfy all depositors’ demands without liquidating any of the long-
term assets. If, however, depositors believe that other depositors will withdraw
prematurely, then all agents find it rational to redeem their claims and a panic
occurs.
The second set of theories suggests that banking crises are not random events
but a natural outgrowth of the business cycle. The idea is that an economic
downturn will reduce the value of bank assets, raising the possibility that banks
are unable to meet their commitments. If depositors receive information about

the impending downturn in the cycle, they will anticipate financial difficulties
in the banking sector and try to withdraw their funds prematurely, as argued by
Gorton (1988). This attempt will precipitate the crisis. In this case crises are a
response of depositors to the arrival of sufficiently negative information on the
unfolding economic circumstances.
The empirical evidence is mixed. Friedman and Schwartz (1963) argue that
the systemic risk and financial instability in the US in the late 18th and early
19th century were panic-based, as evidenced by the absence of downturns in
the relevant macroeconomic time series prior to the crises. In contrast, Gorton
(1988), Calomiris and Gorton (1991), and Calomiris and Mason (2003) provide
a wide range of evidence that many of the crises that occurred in the US in that
period were fundamental-based.
We have only touched on some highlights of the literature on banking crises
and panics here. More complete surveys are provided by Gorton and Winton
(2003), Allen and Gale (2007), and Freixas and Rochet (2008).
2.4 Contagion
One source of systemic risk that does appear to have been important during
the recent financial crisis is contagion. This refers to the possibility that the
distress of one financial institution spreads to others in the financial system,
thus leading ultimately to a systemic crisis. Central banks often use the risk of
contagion to justify intervention, especially when the financial institution in
distress is big or occupies a key position in particular markets. This is the origin
of the term ‘too big to fail’. The recent crisis abounds with examples of this.
For example, Bernanke (2008) argues that the takeover of Bear Stearns by JP
Morgan arranged by the Federal Reserve Bank in March 2008 was justified by
the likelihood that its failure would lead to a whole chain reaction where many
other financial institutions would have gone bankrupt. There would have been
contagion through the network of derivative contracts that Bear Stearns was
part of.
When Lehman Brothers failed in September 2008, it was presumably expected
by the Federal Reserve that its failure would not generate contagion. In fact there
was contagion but it was quite complex. The problem spread first to the money
market funds and the government had to intervene rapidly by providing a guar-
antee of all money market mutual funds. In addition, the failure of Lehman led
to a loss of confidence in many financial firms as investors feared that other
financial institutions might also be allowed to fail. The volumes in many impor-
tant financial markets fell significantly and there was a large spillover into the
real economy. World trade collapsed and in trade-based economies such as Ger-
many and Japan GDP fell significantly in the fourth quarter of 2008 and the
first quarter of 2010. This dramatic fall in GDP in many countries underlines
the importance of the process of contagion.
Despite its importance, our understanding of the effects of contagion risk is
still limited. The academic literature has provided a few explanations of the
mechanisms at play, but much work is still needed. The literature on conta-
gion takes a number of approaches (see Allen, Babus and Carletti 2009 for a
survey). In looking for contagious effects via direct linkages, early research by
Allen and Gale (2000b) studied how the banking system responds to conta-
gion when banks are connected under different network structures. It can be
shown that incomplete networks are more prone to contagion than complete
structures. Following research focused on network externalities created from
individual bank risk and some others applied network techniques to the study
of contagion in financial systems. The main result in this theoretical literature is
that greater connectivity reduces the likelihood of widespread default. However,
shocks may have a significantly larger impact on the financial system when they
occur.
Wagner (2010), Ibragimov et al. (2011), and Allen, Babus and Carletti (2012)
consider a second type of contagion where systemic risk arises from common
asset exposures. Diversification is privately beneficial but increases the likeli-
hood of systemic risk as portfolios become more similar. The use of short-term
debt can lead to a further significant increase in systemic risk.
The recent crisis has shown that asset price volatility can also be an important
form of contagion. Prices drop when banks have to sell their assets in response
to a negative shock. These sales depress asset prices and may have negative
spillovers on other banks in various ways. For example, low asset prices may
affect the solvency of other financial institutions that need to raise liquidity in
the market by selling assets (see, for example, Allen and Carletti 2006). Because
of the low prices, banks in need of liquidity raise little money from the asset
sale. This forces banks to sell larger quantities of assets with the consequence of
further deterioration in asset prices. A similar mechanism holds when mark-to-
market accounting is in use (Allen and Carletti 2008a). In an integrated financial
system this form of contagion can easily transmit shocks internationally, in par-
ticular through large institutions operating cross-border or on global financial
markets.
The key issue is how likely contagion is in practice. A substantial strand of
empirical literature is based on the first type of approach where financial insti-
tutions have direct linkages resulting from the mutual claims they have on one
another. Most of these papers use balance sheet information to estimate bilat-
eral credit relationships for different banking systems and estimate the stability
of the interbank market by simulating the breakdown of a single bank. This
literature is surveyed in Upper (2011). Overall the main finding in the empirical
literature is that contagion is unlikely. However, there are a number of reasons
for caution in accepting this result and concluding that policy-makers need not
worry about contagion between banks. The first is that they do not model price
effects of bankruptcy. Cifuentes et al. (2005) have argued that these price effects
are the main transmission mechanism for contagion. As Upper (2011) points
out, they also rely on the initial shock being confined to a single bank. If there
is an initial shock that affects several banks simultaneously, then this can also
lead to contagion being more likely.
2.5 Sovereign default

The problems of Greece, Ireland, and Portugal have underlined the problem of
sovereign default. Until recently, there had been no credit risk for sovereign debt
in developed countries. The introduction of the euro led to a significant inte-
gration in the European bond market. The spread on the sovereign debt of the
different euro countries had decreased significantly over the last decade. This
reflected the idea that the monetary union across countries together with the
fiscal rules of the Stability Pact would suffice to guarantee a greater fiscal harmo-
nization across Europe and thus the solvency of all euro countries. During 2010
and the first part of 2011 it became clear that the architecture embedded in the
Maastricht Treaty is not sufficient to achieve the predefined goals and that there
can be credit risk in sovereign debt. This is a serious problem in its own right
but also a critical problem because of its effect on the stability of the banking
system.
The problems that started in Europe in the spring of 2010 show clearly that
a sovereign debt crisis and the stability of the financial system are closely inter-
linked. The relation works both ways: The euro crisis puts pressure on the
financial system and the financial crisis in Europe puts pressure on the euro.
Moreover, the financial system and the Eurozone share important features like
the lack of resolution procedures and burden-sharing, the reliance on volun-
tary cooperation among Member States and the presence of imbalances. There
is no resolution procedure in Europe for a sovereign default. Cooperation and
collaboration among European countries to bail out a government in distress is
voluntary in that there are neither clear rules nor guidelines on how this should
be pursued. Allen et al. (2011) contains an analysis of sovereign default in the
Eurozone from an economic and legal perspective.
The sovereign debt crisis has put pressure on banks’ balance sheets through
different channels. For example, it has increased the cost of funding for finan-
cial institutions since it has increased the risk of their assets. Perceived as being
riskier, financial institutions holding large proportions of sovereign debt issued
by countries in distress may have to pay higher interest rates and have more
difficulty in raising funds in the wholesale markets. There may be potential
pressure on these financial institutions to raise capital and liquidity hold-
ings. Finally, financial institutions operating in the country in distress may
also suffer if the country experiences capital outflows and asset substitution
and if the large amount of sovereign debt leads to a crowding out of private
investments.
2.6 Currency mismatches in the banking system

One of the major problems in the 1997 Asian Financial Crisis was currency
mismatch. Banks and firms in Korea, Thailand, and the other countries had
borrowed in foreign currencies, particularly dollars. When the crisis hit, the
banks and firms found that they were unable to borrow. Central banks did not
have enough foreign exchange reserves and were unable to borrow in the mar-
kets. As a result a number had to turn to the IMF. Despite being one of the most
successful economies in the world in the preceding decades, South Korea was
forced by the IMF to raise interest rates to maintain the exchange rate, and to
cut government expenditure. Given that Korean firms used significant amounts
of trade credit, the rise in interest rates was very damaging for them and many
thousands of firms were driven into bankruptcy. Unemployment went from
around three per cent to nine per cent and there was a long recession. It was
this experience that impressed upon the Koreans that they must accumulate
sufficient reserves going forward. Allen and Hong (2011) argue that this led to
the large increase in reserves shown in Figure 10.2.
During the current crisis the major central banks agreed on foreign exchange
swaps and this made a considerable difference in easing the international aspects
of the crisis compared to 1997. Allen and Moessner (2010) describe the problems
raised by banks lending in a low interest rate foreign currency and funding these
loans in various ways.
The foreign currencies that were typically used to make loans were the US dol-
lar, the Japanese yen, and the Swiss franc. These were funded in two ways. The
first was the international wholesale deposit market. The second was to take
deposits in domestic currency and then use the foreign exchange swap mar-
ket to exchange these into the required foreign currency. The largest currency
specific liquidity shortage was $400 billion in US dollars in the Eurozone. The
second largest was $90 billion worth of yen shortfall in the UK, the next largest
$70 billion worth of euros in the US, and after that $30 billion worth of Swiss
francs in the Eurozone.
As the crisis progressed banks found it more and more difficult to fund
these shortfalls. The international wholesale deposit market dried up for many
banks and became difficult for many others. This forced funding of foreign-
denominated loans using domestic currency funding. Since longer maturities of
domestic funding were also often unavailable much of this funding was at short
maturities so the foreign currency mismatch was exacerbated by a maturity mis-
match. The volatility in the foreign exchange markets meant these mismatches
created a large amount of systemic risk for many banks.
3 Macroprudential regulation
The discussion above has highlighted a number of sources of systemic risk in

the financial sector. We next discuss macroprudential regulatory measures and
policies that could be put in place to deal with the identified sources of systemic
risk. What is most important is that the new macroprudential regulation deals
with systemic risk and no longer only with the risk of failure of single finan-
cial institutions. The current crisis has clearly shown that the microprudential
approach to financial regulation does not suffice to prevent systemic crisis.
In this section we consider policies to deal with the six types of systemic risk
identified above. The policies listed above are a useful start in trying to eliminate
systemic risks. However, a more comprehensive approach is needed. Systemic
risk is a complex and pervasive phenomenon that needs to be dealt with in
many diverse ways.
3.1 Dealing with asset price bubbles

As highlighted above, asset price bubbles seem to be one of the major causes
of financial crises. In order to avoid future crises it is of the utmost importance
to try to be able to predict them and prevent their emergence. In an impor-
tant early paper, Borio and Lowe (2002) argue that while it is difficult to predict
asset price bubbles and in particular property bubbles, it is not impossible. They
provide evidence that rapid credit growth combined with large increases in real
estate prices can lead to financial instability. In low-inflation environments they
suggest that inflationary pressures can first appear in asset prices rather than in
the prices of goods and services. They argue that in such cases it may be appro-
priate to use monetary policy to prick asset bubbles and to preserve financial
and monetary stability.
Bubbles, in particular real estate bubbles, seem to be caused by loose monetary
policy and excessive credit supply. One way to prevent them is then through
interest rate policy. In particular, very low interest rates at a time when property
prices are surging should be avoided. Once property prices have started rising,
the question is whether interest rates should be raised to prick them. It may be
possible and desirable to do this in economies with a high degree of homogene-
ity as in small countries like Sweden or possibly the UK. However, doing this
may be difficult for political reasons. In particular when such policies are first
introduced, it may be difficult to explain why it is worth causing a recession to
burst a property bubble.
The problem is more complicated in heterogeneous economies like the US,
China and the Eurozone. Different regions within these economies differ in
terms of economic fundamentals and the rate of property price increases. Using
interest rates to prick bubbles will not be so desirable because this will adversely
affect the areas that do not have bubbles. The recent events in the Eurozone
constitute a clear example. The interest rate policy followed by the European
Central Bank was correct for countries like Germany where there was no bubble
but it was inappropriate for Spain, where it contributed to the creation of the
property bubble. A tighter policy may have been effective for preventing the
bubble in Spain but this would have been at the cost of a recession or at least
slower growth in some of the other countries.
When interest rates cannot be used, it may be better to use other forms of
macroprudential regulation to prevent bubbles. One example would be limits
on loan-to-value ratios that would be lowered as property prices increase at a
faster pace. This can be effective for residential property but may be difficult to
enforce for commercial property. The reason is that firms may be able to use
pyramids of companies that effectively increase leverage. Another measure is
to have property transfer taxes that are greater the higher the rate of property
price increases. Another, perhaps more direct, measure is to impose restrictions
on real estate lending in certain regions.
Crowe et al. (2011) consider the effects of these types of macroprudential
measures to eliminate real estate booms. They have been tried in several Asian
countries including Hong Kong, Korea, and Singapore. They appear to be
effective in the short term but not in the medium and long term.
Saying that monetary policy should not be used to prick bubbles in larger
economies or in monetary unions where countries have varying economic con-
ditions does not imply that monetary policy should not be constrained. Loose
monetary policy is arguably one of the main causes for the emergence of bubbles,
as the recent crisis has shown. One of the most important macroprudential mea-
sures should be constraining monetary policy so that it does not trigger bubbles.
Excessively low levels of interest rates should not be implemented, particularly
when real estate prices are already rising. A possible way to do this is to set up a
check and balance mechanism. The idea behind this is to introduce some form
of accountability for central banks. Another, more drastic, way to do it is to give
central banks the clear mandate to prevent asset bubbles.
As discussed in the previous section, the second major cause of bubbles is
excessive credit. During the recent crisis excessive credit emerged because of
large global imbalances. To prevent bubbles in the future, it is important to
solve this problem. While it is individually advantageous for countries to self-
insure by accumulating reserves, this is an inefficient mechanism from a global
perspective.
As argued above and in Allen and Hong (2011), the accumulation of reserves
by the Asian countries was at least partly a response to the harsh policies that
the IMF imposed on a number of countries during the Asian Crisis in the late
1990s. Part of the problem was the fact that East Asian countries were not well
represented in the senior staff of the IMF. It is therefore important to reform the
governance structure of the IMF and of the other international organizations to
ensure that the Asian countries receive equal treatment when they need financial
help. This would reduce the need of these countries to accumulate reserves as a
self-insurance mechanism.
To reduce the large accumulation of reserves by China, other measures are
necessary, however. For example, senior Chinese officials have proposed having
a global currency to replace the dollar. This has the advantage that reserves can
be created initially without large transfers of resources and the attendant risk of
a crisis. All countries could be allocated enough reserves in the event of a crisis
so that they can survive shocks. The problem is that an international institution
like the IMF would need to implement the currency. There would then be again
the issue of whether all countries, and in particular the Asian ones, are properly
represented in the governance process of this institution.
A more likely medium-term scenario is that the Chinese Rmb becomes fully
convertible and joins the US dollar and the euro as the third major reserve
currency. With three reserve currencies there would be more scope for diversifi-
cation of risks and China itself would have little need of reserves. This is perhaps
one of the most practical solutions to the global imbalances problem. The Chi-
nese have already taken some steps in this direction. They have started to allow
the settlement of trade in Rmb. They have also allowed the issue of Rmb bonds
by Western companies such as McDonald’s in Hong Kong. Of course, the most
important aspect of being a reserve currency is full convertibility of the Rmb.
That is still some way off and this is the sense in which this solution to the
global imbalances problem is a medium-term one.
3.2 Policies with regard to mispricing of assets

As discussed above, one of the major problems in the current crisis is that many
securitized securities appear to have been mispriced. One explanation is that this
is due to cash-in-the-market-pricing and limits to arbitrage. Another is that it is
due to asymmetric information and in particular to adverse selection. The prob-
lem is to design policies that allow this problem to be corrected. This was the
origin of the TARP program in the US. The idea was that by buying large amounts
of toxic assets the Treasury could restore the functioning of the market. In prac-
tice they were unable to actually implement the program. This type of direct
intervention seems problematic to say the least. Clearly there are significant
political economy issues. There are also issues of whether such a scheme could
actually restore the market to proper functioning. This is one of the most impor-
tant areas of macroprudential policy-making. As yet no convincing solutions
have been suggested.
Given the lack of an immediate solution to this problem, what should gov-
ernments do? One of the major problems is that recent reforms have ensured
that financial institutions mark their assets to market. In normal times this
is undoubtedly the best system. Financial institutions have traditionally used
historic cost accounting for many of their assets. This system has the disadvan-
tage that it allows institutions to hide falls in asset values for significant periods
of time. A good example is the S&L crisis in the US in the 1980s. This kind of
episode encouraged the move to mark-to-market accounting by the IASB and
US FASB (see, for example, Allen and Carletti 2008a and Plantin et al. 2008). The
divergence between asset prices, particularly those of securitized products, and
apparent fundamentals in the current crisis meant that mark-to-market account-
ing came under severe criticism from financial institutions and was relaxed by
the FASB under political pressure from Congress.
How should the advantages and disadvantages of mark-to-market accounting
be balanced? As long as markets are efficient, mark-to-market accounting domi-
nates. However, if during times of crisis they cease to be efficient, market prices
do not provide a good guide for regulators and investors. The key issue then
becomes how to identify whether financial markets are working properly or not.
Allen and Carletti (2008b) suggest that when market prices and model-based
prices diverge significantly (more than 2 per cent, say), financial institutions
should publish both. If regulators and investors see many financial institutions
independently publishing different valuations they can deduce that financial
markets may no longer be efficient and can act accordingly.
3.3 Panics
As argued previously, it is unclear whether panics have played an important role
in the current crisis. In contrast, panics were thought to be the main cause of
crises in the past. Starting with the influential work of Friedman and Schwarz
(1963) it was widely believed that the crises that occurred in the US in the latter
part of the 19th century up until the early 1930s were mostly caused by panics.
The introduction of deposit insurance for retail depositors represented one sim-
ple way to stop them. The idea is that if people know that the government will
cover any losses, it becomes rational for everybody to leave their money in the
banking system. This eliminates runs stemming from panics.
This system has been effective in eliminating runs until recently. However,
deposit insurance is only for small depositors. It does not cover large deposits
or wholesale funding that, as shown in the recent crisis, constitute the majority
of funding for many financial institutions. As a result, deposit insurance alone
is no longer adequate for solving the problem of panics.
A simple possibility is to extend deposit insurance and guarantee all forms of
short-term debt. While this solution can be effective in preventing panics, it has
the clear drawback of generating moral hazard. If banks have access to low-cost
funds guaranteed by the government, they have an incentive to take significant
risks. A better solution to prevent risk-taking may be to remove deposit insurance
and deal with the problem of panic runs through lender-of-last-resort policies. If
depositors know that the central bank will provide the needed liquidity if they
attempt to withdraw early, they won’t withdraw and there won’t be a run.
The other significant problem with deposit insurance and short-term guar-
antees is that if there are other types of systemic risk in addition to panics,
then they can be extremely costly to implement. An example is Ireland where
the blanket bank debt guarantees in September 2008 effectively bankrupted the
country and drove the government to seek funds from the European Financial
Stability Fund.
3.4 Contagion
As argued above, contagion is arguably one of the most important problems of
systemic risk. Despite its importance, it is not yet fully understood how con-
tagion can occur and there has been very little work done so far on how to
stop it. The channels for contagion are multiple, ranging from direct linkages
among banks on the interbank market or the payment system to common asset
exposure.
There may be the need for several policies and regulations that address the
different channels and types of contagion. Capital regulation has been the main
tool for regulating banks in recent years. This has been coordinated internation-
ally through the Basel agreements. It is the main tool for ensuring stability in
the international financial system. The traditional justification in the academic
literature for capital regulation has been that it is required to offset the moral
hazard arising from deposit insurance (for examples of exceptions, see Hellman
et al. 2000). Because banks have access to low-cost funds guaranteed by the
government, they have an incentive to take significant risks. If the risks pay off
they receive the upside, while if they do not the losses are borne by the gov-
ernment. Capital regulation is needed to offset the incentives for banks to take
risks, otherwise shareholders will lose significantly. Moreover, capital acts as a
buffer to absorb losses thus making banks more resilient to shocks and losses
and perhaps most importantly reducing the risk of contagion.
There is a longstanding debate on how much capital banks should hold. The
recent crisis and the current discussions behind the proposal for a new regulatory
framework have highlighted the difficulties embodied in these proposals. The
starting point of the discussion is that capital is a more costly form of funding
than debt so that, if unregulated, banks minimize the use of capital. Thus there
is the need for a regulation that forces banks to hold minimum levels. The
same argument is typically assumed in the academic literature (see, for example,
Gorton and Winton 2003).
Modeling the cost of equity finance for financial institutions is one of the
major problems in designing capital regulation. The first issue is whether equity
is in fact more costly than debt. If that is so, the second issue is whether equity
is more costly only in the financial industry or also in all other industries. It
is the case that financial institutions hold approximately 10 per cent of capital
while industrial companies operate with 30-40 per cent equity. Understanding
the reasons for this large difference in capital structures is of crucial importance
to designing capital regulation appropriately.
One simple answer as to why capital is privately more costly is that in many
countries debt interest is tax deductible at the corporate level but dividends are
not. It is not clear why this is so and whether this should be the case. There does
not seem to be any good public policy rationale for having this deductibility.
It seems to have arisen as an historical accident. When the corporate income
tax was introduced interest was regarded as a cost of doing business in the same
way that paying wages to workers was a cost. However, from a modern corpo-
rate finance perspective, this is not the correct way to think about it. Equity
and debt are just alternative ways of financing the firm. If tax deductibility is
behind a desire to use debt rather than equity, then the simple solution is to
remove it. If without deductibility financial institutions are willing or can be
induced through regulation to use more equity, then financial stability would
be considerably enhanced.
Other possible rationales for the high cost of equity are agency problems
within the firm. According to this rationale, the cost of equity does not pro-
vide the correct incentives to shareholders or managers to provide the right
monitoring. High leverage is needed to ensure this. There is little empirical evi-
dence that this is in fact a severe problem in the banking sector. For example,
leverage in private equity and venture capital firms where the agency problem
seems much greater is typically less than in banks.
A final point concerns the reason why financial institutions hold so little cap-
ital relative to other industries. The tax deductibility argument cannot explain
this difference, as deductability applies to all industries. A more plausible expla-
nation is that debt in the financial industry is implicitly subsidized through
government guarantees and bailouts. If this is why financial institutions rely
so much on debt, then it is necessary to remove guarantees and create credible
enforcement mechanisms, that is, proper resolution procedures. Without this,
banks will continue minimizing the amount of capital they hold and society
will bear the costs of this through increased financial instability.
In the current debates on capital regulation two main proposals have been
put forth. The first one concerns countercyclical capital regulation. The second
concerns the use of hybrid instruments in the form of contingent convertible
debt (CoCos).
One of the most widely suggested macroprudential policies is countercyclical
capital regulation. The idea is that during normal times banks and other finan-
cial institutions can accumulate capital reserves and buffers that will allow them
to survive serious shocks to the financial system. These measures are related to
the countercyclical loan reserves that have been implemented by the Bank of
Spain for some time. Spanish banks did accumulate loan reserves in the period
before the crisis and this helped them to weather the crisis better than they oth-
erwise would have done. This experience suggests that countercyclical capital
ratios may be helpful. However, they did not prevent the credit boom and the
bubble in property prices in Spain, so not too much reliance should be placed
on them.
It has been widely suggested that banks should issue convertible debt that
can be converted into equity in the event of a crisis. The issue of this kind of
security by the Royal Bank of Scotland and Lloyds in the UK and Unicredit in
Italy is an example. The idea is that these CoCos have two main advantages.
First, it is not necessary for banks to raise capital in difficult times as it would
already be available. Second, contingent capital allows the sharing of losses
with debtholders. This would also play a disciplinary role and would induce
bank managers to behave more prudently.
Another way to stabilize markets and avoid contagion is to have a combina-
tion of public and private financial institutions. This is the case for example in
Chile, where Banco Estado is a publicly owned commercial bank that competes
with private sector banks. In times of crisis, such a bank can expand and help
stabilize the market as all market participants know that it is backed by the state
and will not fail. That is what many central banks have effectively been doing by
buying large quantities of commercial paper. These central banks have become
like large commercial banks. But the officials in charge of central banks do not
usually have much expertise in running a commercial bank or know much about
credit risk. It would be better to have expertise in the public sector that would
allow the state to perform commercial banking functions during times of crisis.
These state institutions would act as firebreaks and limit the damage that can
be caused by contagion.
3.5 Sovereign default

The Greek sovereign debt crisis in 2010 underlined problems with the design
of the Eurozone. The Stability Pact contained rules on the amount of current
public deficits. The possibility that a country would go into default was not
even contemplated in the architecture of the Eurozone. When the Greek crisis
emerged, there were no guidelines or regulations that could be used. In the end
the way the EU and the Eurozone dealt with the problem of sovereign default
was to set up a bailout fund. This obviously creates moral hazard in that it
changes the incentives of governments when dealing with fiscal excesses. In
addition there is the question of how sustainable the bailout mechanism is in
terms of political economy. If Greece and any other countries do default or
restructure their debt, Germany will pay a large share of the cost. This opens
up the question of the extent to which German voters are willing to subsidize
defaulting countries, or for that matter citizens of other countries that end up
making large contributions. In such a scenario there may be strong demands to

pull out of the Eurozone to avoid having to make large transfers. This political
economy problem is a very serious one. As the perception that a country might
default increases, investors will have to make a judgment as to how this political
process will play out. This makes the effect of the guarantees quite uncertain.
The Greek crisis has led to calls for the reform of the Growth and Stability Pact.
As discussed above, this is designed to prevent governments from running up
large debts and then not fulfilling their obligations. One issue is how meaningful
the Pact is anyway because of the fact that once France and Germany flouted the
Pact and were not penalized, it became clear the rules were unenforceable except
for the small countries. However, perhaps a more serious problem with the Pact
is that two countries that are among those that came under the highest pressure
in the crisis, namely Ireland and Spain, were actually doing extremely well in
terms of the Pact criteria as they were running surpluses and had low levels of
public debt. The reason they had a problem is because they ran a large current
account deficit and imported large amounts of capital that fueled the property
bubble. This raises the question of whether the Pact needs to be reformed to
include restrictions on capital inflows that are invested in real estate.
The real problem with sovereign default is the contagion that will follow such
an event. For example, if Greece does default then there is likely to be a severe
problem with instability among Greek banks and among many other coun-
tries’ banks. Under the current arrangements there would be great uncertainty
as to how any resolution would proceed and this in itself would considerably
exacerbate the problems in the markets. After Argentina’s default in 2001, for
example, it took a great deal of time to determine how much creditors would
receive.
A better solution to the problem of sovereign default is perhaps a bankruptcy
mechanism of some kind that would avoid the need for a bailout. This would
remove a great deal of uncertainty and thus could be expedited. There could also
be the equivalent of corporate debtor-in-possession finance for governments.
If Greece does default then the first thing the Greek government could then
do is to issue senior bonds to save its banking system. One way that such a
mechanism could work is for the country to declare default. A team from the
ECB and European Commission would carry out an assessment of what the best
repayment structure would be. If the country did not agree with the team’s
assessment then its alternative would be to withdraw from the Eurozone.
Another issue raised by the question of sovereign default is what risk weight-
ings should be attached to sovereign debt. Banking regulation should recognize
that the market apparently views the debt of many countries as a worse credit
risk than the debt of many corporations. One way to achieve this is to assign
risk weightings and diversification requirements on sovereign debt. This should
be done in conjunction with the introduction of a bankruptcy mechanism.
One of the major problems in the Eurozone that has been highlighted by the
problems of Greece, Ireland, and Portugal is that it is very difficult for countries
to adjust their economies after a severe shock. For example, in Greece wage
levels seem to be at too high a level. Should there be some mechanism within
the Eurozone to deal with this wage rigidity? One possibility would be to allow
temporary exit from the Eurozone and then re-entry. Such exit and re-entry
might considerably help the adjustment process in such countries.
3.6 Currency mismatches

As discussed earlier, one of the major problems in the 1997 Asian Financial
Crisis was that the banks couldn’t obtain foreign currencies. The main rea-
son for currency mismatches not being such a problem in the current crisis
was the introduction of central bank currency swaps. Allen and Moessner
(2010) document how the swap system worked. There were four overlapping
networks:
• The Fed network to supply US dollars;

• The ECB network to supply euros;
• The Swiss franc network;
• The Latin American and Asian networks.
These swap networks involved considerable overlap (as shown in Graph 7.1 in
Allen and Moessner 2010). As they were organized between central banks, the
credit risk borne was sovereign rather than commercial. The receiving central
bank would then pass on the foreign currency to firms and financial institutions
so that these bore the commercial credit risk. Some of the swaps between central
banks were collateralized with the currency of the counterparty central bank.
These swaps considerably eased foreign exchange problems during the crisis and
are widely regarded as having been a success.
References
Allen, F., A. Babus and E. Carletti (2009) ‘Financial Crises: Theory and Evidence’, Annual
Review of Financial Economics, vol. 1, pp. 97–116.
Allen, F., A. Babus and E. Carletti (2012) ‘Asset Commonality, Debt Maturity and Systemic
Risk’, Wharton Financial Institutions Center Working Paper 10-30, Journal of Financial
Economics, vol. 104, no. 3, pp. 519–534.
Allen, F. and E. Carletti (2006) ‘Credit Risk Transfer and Contagion’, Journal of Monetary
Allen, F. and E. Carletti (2008a) ‘Mark-to-Market Accounting and Liquidity Pricing’, Journal
of Accounting and Economics, vol. 45, no. 2–3, pp. 358–378.
Allen, F. and E. Carletti (2008b) ‘Should Financial Institutions Mark to Market?’, Bank of
France Financial Stability Review, vol. 12, October, pp. 1–6.
Allen, F. and E. Carletti (2009) ‘An Overview of the Crisis: Causes, Consequences and
Solutions’, International Review of Finance, vol. 10, no. 1, pp. 1–27.
Allen, F., E. Carletti and G. Corsetti (2011) Life in the Eurozone With or Without Sovereign
Default? (Philadelphia: FIC Press), at: http://finance.wharton.upenn.edu/FIC/FICPress.
Allen, F., E. Carletti and D. Gale (2009) ‘Interbank Market Liquidity and Central Bank
Intervention’, Journal of Monetary Economics, vol. 56, no. 3, pp. 639–652.
Allen, F. and D. Gale (2000a) ‘Bubbles and Crises’, Economic Journal, vol. 110,
pp. 236–255.
Allen, F. and D. Gale (2000b) ‘Financial Contagion’, Journal of Political Economy, vol. 108,
no. 1, pp. 1–33.
Allen, F. and D. Gale (2004) ‘Asset Price Bubbles and Monetary Policy’, in M. Desai and
Y. Said (eds), Global Governance and Financial Crises (New York and London: Routledge),
pp. 19–42.
Allen, F. and D. Gale (2007) Understanding Financial Crises, Clarendon Lecture Series in
Finance (Oxford: Oxford University Press).
Allen, F. and J. Hong (2011) ‘Why Are There Large Foreign Exchange Reserves? The Case
of South Korea’, Korean Social Science Journal, vol. 38, no. 2, pp. 1–33.
Allen, W. and R. Moessner (2010) ‘Central Bank Co-operation and International Liquidity
in the Financial Crisis of 2008–9’, Bank for International Settlements Working Paper
310 (Basel: BIS).
Bernanke, B. (2008) ‘Opening Remarks’, Jackson Hole Symposium, Maintaining Stability in
a Changing Financial System, Federal Reserve Bank of Kansas City, pp. 1–12.
Borio, C. and P. Lowe (2002) ‘Asset Prices, Financial and Monetary Stability: Exploring the
Nexus’, Bank for International Settlements Working Paper 114 (Basel: BIS).
Bryant, J. (1980) ‘A Model of Reserves, Bank Runs, and Deposit Insurance’, Journal of
Banking and Finance, vol. 4, no. 4, pp. 335–344.
Calomiris, C. and G. Gorton (1991) ‘The Origins of Banking Panics, Models, Facts, and
Bank Regulation’, in R. G. Hubbard (ed.), Financial Markets and Financial Crises (Chicago:
University of Chicago Press).
Calomiris, C. and J. Mason (2003) ‘Fundamentals, Panics and Bank Distress during the
Depression’, American Economic Review, vol. 93, no. 5, pp. 1615–1647.
Case, K. and R. Shiller (1989) ‘The Efficiency of the Market for Single-Family Homes’,
Cifuentes, R., G. Ferrucci and H. Shin (2005) ‘Liquidity Risk and Contagion’, Journal of the
European Economic Association, vol. 3, nos 2–3, pp. 556–566.
Crowe, C., G. Dell’Ariccia, D. Igan and P. Rabanal (2011) ‘How to Deal with Real Estate
Booms: Lessons from Country Experiences’, IMF Working Paper 11/91 (Washington,
DC: International Monetary Fund).
Diamond, D. and P. Dybvig (1983) ‘Bank Runs, Deposit Insurance, and Liquidity’, Journal
of Political Economy, vol. 91, no. 3, pp. 401–419.
Englund, P., J. Quigley and C. Redfearn (1998) ‘Improved Price Indexes for Real Estate:
Measuring the Course of Swedish Housing Prices’, Journal of Urban Economics, vol. 44,
no. 2, pp. 171–196.
Freixas, X. and J.-C. Rochet (2008) Microeconomics of Banking, second edition (Cambridge,
MA: MIT Press).
Friedman, M. and A. Schwartz (1963) A Monetary History of the United States, 1867–1960
Glaeser, E. and J. Gyourko (2007) ‘Housing Dynamics’, working paper, Wharton School,
University of Pennsylvania.
Gorton, G. (1988) ‘Banking Panics and Business Cycles’, Oxford Economic Papers, vol. 40,
no. 4, pp. 751–781.
Gorton, G. and A. Winton (2003) ‘Financial Intermediation’, in G. Constantinides, M.
Harris and R. Stulz (eds), Handbook of the Economics of Finance, Volume 1A (Amsterdam:
North-Holland), pp. 431–552.
Hellmann, T., K. Murdock and J. Stiglitz (2000) ‘Liberalization, Moral Hazard in Banking,
and Prudential Regulation: Are Capital Requirements Enough?’, The American Economic
Review, vol. 90, no. 1, pp. 147–165.
Herring, R. and S. Wachter (1999) ‘Real Estate Booms and Banking Busts: An International
Perspective’, Wharton Financial Institutions Center Working Paper 99-27, University
of Pennsylvania.
Ibragimov, R., D. Jaffee and J. Walden (2011) ‘Diversification Disasters’, Journal of Financial
Plantin, G., H. Sapra and H. Shin (2008) ‘Marking-to-market: Panacea or Pandora’s Box?’,
Journal of Accounting Research, vol. 46, no. 2, pp. 435–460.
Reinhart, C. and K. Rogoff (2009) This Time Is Different: Eight Centuries of Financial Folly
(Oxford and Princeton: Princeton University Press).
Taylor, J. (2008) ‘The Financial Crisis and the Policy Responses: An Empirical Analysis of
What Went Wrong’, working paper, Stanford University.
Upper, C. (2011) ‘Simulation Methods to Assess the Danger of Contagion in Interbank
Markets’, Journal of Financial Stability, vol. 7, no. 3, pp. 111–125.
Wagner, W. (2010) ‘Diversification at Financial Institutions and Systemic Crises’, Journal
of Financial Intermediation, vol. 19, no. 3, pp. 333–354.
11
Filling the Gaps – the Vienna Initiative
and the Role of International Financial
Institutions in Crisis Management
and Resolution
Erik Berglof
European Bank for Reconstruction and Development
1 Introduction
Global financial integration has proceeded ahead of supporting governance

arrangements. As a result large gaps have developed in regulation and supervi-
sion, and in crisis management and resolution. The incompleteness of the global
financial architecture became acutely clear in the financial crisis. Large vulner-
abilities had been allowed to develop, and the mechanisms in place for dealing
with the crisis, at least its cross-border aspects, proved grossly inadequate.
No world region illustrates the gap between the extent of integration and the
supporting architecture better than Emerging Europe. During the 1990s foreign
banks had rapidly expanded their networks of subsidiaries across the region to
the extent that they in many countries controlled more than 80 per cent of the
banking system (see Figure 11.1). The cross-border financial architecture was
rudimentary at best with little or no authority on regulation and supervision at
the EU level, and most powers vested in the home authorities of the large banks.
In this institutional vacuum, credit expanded rapidly and large exposures built
up in the period leading up to the crisis. When Lehman Brothers collapsed and
the financial crisis broke out in full, there was no EU-wide framework for crisis
management and resolution. Moreover, the banking networks extended much
beyond EU member states.
In the wake of the global crisis, ambitious attempts have been made to fill
these gaps, not the least in Europe. On the initiative of the G20 a wave of regula-
tory reform has swept the industry worldwide and new cross-border supervisory
institutions have been created in the European Union. In parallel, host coun-
tries have reasserted their authority over regulation and supervision, also within
the EU, reinforcing the fragmentation of the global architecture.
211
212 E. Berglof
100
90
80
70
60
50
40
30
20
10
0
Bulgaria
Hungary
Croatia
BiH
Romania
Albania
Serbia
Georgia
FYROM
Slovakia
Poland
Kyrgyz
Figure 11.1 Share of foreign banks owned by Eurozone-based groups
Source: EBRD (2011).
Filling the remaining gaps in the global architecture will not be easy, given
the complex and evolving nature of the financial sector and the lack of legit-
imacy of supranational arrangements. Despite all the efforts of policy-makers,
significant incompleteness is likely to remain. The question is what can be done
to design institutions and procedures to facilitate coordination where existing
formal frameworks fail.
This chapter exploits the experience of the global financial crisis to understand
the potential role of the international financial institutions (IFIs) in this regard.
In particular, we look at a coordination exercise in Europe, the so-called Vienna
Initiative, and how some international financial institutions contributed to its
design and implementation. We examine the evidence on the effectiveness of
the initiative and discuss its wider implications.
The experience from the Vienna Initiative suggests that the international
financial institutions indeed can play a constructive role in helping coordinate
crisis management and resolution, and possibly also in preventing vulnera-
bilities from building. By facilitating coordination between home and host
countries, and among home and host countries, as well as between public
authorities and the banks, and among the banks, the IFIs contributed to pre-
venting a traditional emerging market crisis with sharp currency depreciations
and the collapse of banking systems. If this is a meaningful role IFIs could play,
this would have implications for how they are designed and incentivized.
The chapter starts by describing the situation in the financial sector of
Emerging Europe at the time of the collapse of Lehman Brothers with large
vulnerabilities and the lack of an effective framework of regulation and super-
vision. It then proceeds to discuss the Vienna Initiative and its impact. Finally,
we examine what lessons we can learn from the role played by the IFIs in the
initiative and how to design these institutions.
The Vienna Initiative and the Role of the International Financial Institutions 213
2 Emerging Europe after Lehman Brothers
During the period leading up to the crisis massive amounts of capital flowed
from the advanced economies in Western Europe to the transforming economies
in Eastern Europe. Unlike most other emerging market regions, financial inte-
gration was strongly correlated with growth in Eastern Europe, at least until
2004 (EBRD 2009). Indeed, the evidence suggests that financial integration sup-
ported growth in the region. However, when the crisis hit large vulnerabilities
were unveiled. It was clear that the existing regulatory and supervisory frame-
works, whether national or regional, had not been able to contain the risks.
Private sector debt had been allowed to increase to sometimes unsustainable
levels and much of this debt was in foreign currency, mostly euros but also
more ‘exotic’ currencies like Swiss francs and Japanese yen.
The main engine of financial integration and financial development had been
the foreign bank groups. In many countries they controlled more than 80 per
cent of the banking sector, and most of the foreign subsidiaries were systemic in
that their collapse would threaten the entire banking system of the country in
which they operated. The extent of functional integration differed across banks,
but some banks, such as, for example, the Swedish SEB, had gone very far in
this regard. The predominant banking business model in the larger cross-border
banking groups serving Emerging Europe tended towards centralized liquidity
management and branches (rather than subsidiaries which had been the domi-
nant mode since the foreign banks began their expansion into the region). The
cross-border banking groups were typically active in many countries, in some
cases with branches or subsidiaries in 20 or so countries in Emerging Europe, and
sometimes with separate subsidiaries in other lines of activity, like leasing and
insurance. The extent of reliance on funding from the parent differed greatly
across subsidiaries, but many of them had quite limited local funding. Parent-
bank funding was very cheap and thus an important source of competitive
advantage.
When the crisis hit, non-bank finance like equity and bond finance quickly
disappeared from Emerging Europe, as did effectively wholesale funding for
parents as well as subsidiaries (EBRD 2009). Many banks active in the region,
not only foreign banks, had been heavily dependent on wholesale funding due
to the limited deposit base. Syndicated bank finance also vanished, and non-
strategic banks, that is, banks without large retail franchises, radically reduced
their volumes in the region. The strategic cross-border banks did also deleverage,
but on a smaller scale and over a longer period. The combined result of all these
developments was a significant credit contraction, particularly in the relatively
early stages of the crisis. The tightening was caused by low equity ratios, low Tier-
1 capital ratio, and losses on financial assets (Popov and Udell 2010). Foreign
banks also transmitted a larger portion of similar financial shocks to the real
214 E. Berglof
economy than did domestic banks. High-risk firms and firms with fewer tangible
assets were more affected by the credit tightening.
Cross-border banking had been allowed to expand without a supporting
framework for crisis management and resolution at the European level. The
trend had been towards home country regulation and supervision with less and
less influence for host authorities. In many ways, this was an understandable
development given the increasing size and complexity of these bank groups,
but it left host countries with little comfort that their interests would be taken
into account. As with the general area of financial regulation and supervision,
the European Commission had been given very limited powers in cross-border
banking. The so-called Lamfallusy process was primarily based on intergovern-
mental approximation of regulatory and supervisory frameworks. Memoranda
of understanding had also been signed between home and host country supervi-
sors within the European Union, but in the crisis these documents proved to be
of little consequence. Moreover, despite the economic and political significance
of the large financial groups in the region, contacts between the groups and
the relevant authorities were limited. Finally, and probably most importantly,
there were no common frameworks for cross-border resolution or agreement on
burden-sharing between home and host countries. Many countries did not even
have national, let alone mutually compatible, resolution schemes at the time
the crisis broke out.
Despite considerable progress on the regulatory and supervisory agenda since
the crisis first hit the global financial architecture will remain severely incom-
plete. Moreover, the global financial system is constantly transforming itself in
response to changes in the demand and supply of capital and financial inno-
vation, and as a result of changes in national frameworks. For example, the
trend towards centralized liquidity management and branches rather than sub-
sidiaries has been reversed or at least halted as a result of the crisis. Similarly, the
pre-crisis trend towards ‘home country rule’, where home country authorities
were supposed to regulate and supervise the parent banks, has been replaced
by a rapid increase in host country intervention. This trend is likely to further
increase the gaps in the global architecture as harmonization fails and national
jurisdictions come into conflict with each other.
3 The Vienna Initiative
The Vienna Initiative was conceived in the autumn of 2008 when withdrawals
by multinational banks from emerging Europe threatened the stability of the
region and ultimately the entire European banking system. The rapid credit
growth during the pre-crisis period had left the private sector in many coun-
tries highly leveraged and heavily exposed to currency depreciations. During
the early autumn there were a number of informal, mostly bilateral, contacts
between a few pan-European banks and the international financial institutions,

but it was clear that a more systemic approach was necessary.
In November 2008, five Western European banks with a large presence in
Emerging Europe sent a letter to the European Commission calling for a quick
and coordinated response to the problems in the banking systems of Emerging
Europe. There was no immediate response to the letter, and when asked to lead
a coordination effort the European Commission declared that it had little or
no competence in the area. Moreover, the problem extended beyond the Euro-
pean Union, and even to countries that had very little prospect of becoming
EU member states in the foreseeable future. The political leadership of the rel-
evant directorate of the EU also initially refused to see the seriousness of the
situation.
Instead, in December 2008 the Austrian government collaborated with the
international financial institutions to organize the hosting of a first meeting in
January 2009 of what was to become the Vienna Initiative, initially only with
representatives of the public sector institutions. It was conceived of as an infor-
mal coordination platform to assure continued commitment of parent banks to
their subsidiaries. The platform was then extended to the banks themselves. It
eventually involved most multinational banks with a strong presence in Emerg-
ing Europe, their home and host country supervisors, fiscal authorities, and
the international financial institutions.1 Altogether 17 banking groups covering
over 60 per cent of all banking assets in the region participated in the initiative.
As part of the Vienna Initiative, and as a complement to the coordination
efforts, the EBRD, EIB, and the World Bank Group – three investing interna-
tional financial institutions – on 27 February 2009 launched the ‘Joint IFI Action
Plan’ with the objective ‘to support banking sector stability and lending to the
real economy in crisis-hit Central and Eastern Europe’.2 A funding package of
d24.5 billion was put together. By end-September 2009, d16.3 billion of IFI sup-
port – both senior loans and subordinated debt (tier 2 capital) – had been made
available to banks (Vienna Initiative 2009: 3).
The commitments under the Vienna Initiative were embedded in some of the
IMF and EU stabilization programmes introduced in a large number of countries
when external financing gaps were opening up as foreign currency debt matured.
A significant part of this debt had been issued by Western European banks active
in the region. Insufficient roll-over of this debt would have undermined the suc-
cess of IMF-EU balance of payments stabilization programs. The international
institutions were also concerned not to substitute commercial funding with pub-
lic sector money, that is, allowing IMF-EU contributions to be used to effectively
bail out the private banks and their shareholders. Indeed, debt roll-overs by com-
mercial banks were made conditional on finalizing some of the macroeconomic
support packages. Local authorities also offered some regulatory forbearance in
exchange for the banks’ commitments to maintain exposures.
216 E. Berglof
In five countries – Bosnia and Herzegovina, Hungary, Latvia, Romania, and

Serbia – a total of 17 parent banks pledged through so-called ‘commitment
letters’3 to largely maintain their exposures and to recapitalize subsidiaries as
long as these countries kept their IMF-EU programs on track.4 The exact nature
of the commitments differed across countries. For example, in the case of Latvia,
Swedish banks were to roll over at least 80 per cent of their lending to the coun-
try, most of which was to their own subsidiaries.5 In the case of Hungary, banks
promised to ensure a ‘prudent capitalization of their subsidiaries’ and to main-
tain at least 95 per cent of their September 2008 exposure. In Romania, pledges
were most concrete as banks promised to ‘increase the minimum capital ade-
quacy ratio for each subsidiary from 8 to 10 percent’ and to fully maintain their
exposure for the duration of the IMF program. These commitments were closely
monitored and some of them reaffirmed later in 2009.6 As the crisis subsided,
the pressure to maintain exposures was reduced and in some cases roll-over
commitments were lowered by early 2010. The commitments were allowed to
lapse in the spring of 2011.
At the time, worries were expressed that the focus of the commitment letters
on five core countries could lead multinational banks to support their lend-
ing in these countries by withdrawing funds from less affected countries (such
as Poland and the Czech Republic). Also countries without EU/IMF programs
feared that their foreign bank subsidiaries would be drained in order to meet
commitments in countries with such programs. For example, the Lithuanian
authorities criticized the Latvian agreement for not ensuring maintained expo-
sures to Latvia at the expense of Lithuania and Estonia, the two other significant
countries in the Eastern European portfolios of the Swedish banks. Such nega-
tive spillovers would have contributed to the cross-border transmission of the
crisis. To some extent these concerns were alleviated by a number of informal
agreements between banks and the EBRD that extended the informal commit-
ments of EBRD-supported banks to the transition region as a whole. Moreover,
in September 2009 and March 2010 so-called ‘horizontal meetings’ were held
with 17 and 20 multinational banking groups, respectively, as well as the rele-
vant national and international authorities. Here, the focus was more on lending
to the region rather than the five program countries alone. In the case of the
Latvian agreement, an explicit reference was made to the need to also maintain
exposures to Estonia and Lithuania.
An important concern in particular at the beginning of the crisis was that
home country support programs for the banks would impose restrictions on the
use of funds for their subsidiaries abroad. Had such restrictions been imposed,
host country authorities would most likely have retaliated by ring-fencing or
even nationalizing subsidiaries. In March 2009, an Emergency Summit of EU
leaders confirmed that bank support packages at the national level should not
lead to any restrictions on banks’ Eastern European subsidiaries.7
The international financial institutions participating in the Vienna Initia-

tive also lobbied the European Commission to ensure that any compensatory
measures requested in response to the state aid provided to the home coun-
try support programs would not undermine the efforts to contain deleveraging
in the region. The directorate responsible for enforcement of state aid proce-
dures had to go through a steep learning curve in establishing the difference
between aid to manufacturing firms in normal times and support to systemically
important financial institutions in a global financial crisis. Initial determina-
tions in some of the cases were also challenged to make sure that necessary
downsizing would not discriminate against the banks’ presence in Emerging
Europe.
4 Empirical evidence
Judging from the macro evidence, the Vienna Initiative appears to have been
uniquely successful in avoiding a traditional emerging market crisis. Even
though Emerging Europe was the hardest hit region globally capital outflows
in the year after the Lehman collapse were the smallest among major world
regions.8 The feared large-scale, uncoordinated withdrawal of banks from
Emerging Europe did not materialize, and no foreign subsidiaries failed. There
were no system-wide bank failures and no currency collapse, with the pos-
sible exception of the Ukranian hryvnia. Fiscal packages from home bank
governments were allowed to support subsidiaries abroad, and host countries
implemented responsible policies, strengthening deposit insurance schemes
and allowed foreign subsidiaries and domestic banks equal access to liquidity
support.9
40
30
20
Per cent
10
0
–10
–20
–30
–40
–50
Em. Asia excl. Lat. Am. excl. CEB & SEE RUS, UKR, BRA, IND, CHN
CHN, IND BRA KAZ
Peak outflows (08Q4 & 09Q1)

Total flows (08Q3–10Q1)
Recovery flows (09Q3–10Q1)
Figure 11.2 Capital flows 2008–2010 from major world regions
Source: Bank of International Settlements.

218 E. Berglof
Attributing all this to the Vienna Initiative is probably excessive. At least

until recently there was no strong empirical support for the contribution of
the Vienna Initiative to this outcome. Some country-level evidence is provided
by Cetorelli and Goldberg (2011). They analyze how adverse liquidity shocks
on main developed country banking systems affect emerging markets and find
three separate channels: 1) direct, cross-border lending by foreign banks; 2) local
lending by foreign banks’ affiliates in emerging markets; and 3) loan supply to
domestic banks (via a decline in interbank, cross-border lending). Their data
show that the decline in domestic bank lending was somewhat mitigated in the
case of Vienna Initiative countries.
Vallee and Pawlovsky (2011) exploit the difference in timing between the
coordination part of the Vienna Initiative and the Joint Action Plan, with the
latter happening two months after the first announcement of the Vienna Ini-
tiative. They find that the Joint Action Plan had a significant impact on market
perceptions of the crisis, whereas the coordination as such seemed to leave little
mark.
Using a more direct approach, de Haas et al. (2011) find, based on a compre-
hensive bank-level dataset with detailed data on ownership, that foreign bank
subsidiaries overall reduced their credit supply earlier and faster than domes-
tic banks. In their study home country state support packages had no impact
on lending by multinational banks’ subsidiaries in Emerging Europe, but for-
eign banks that took part in the Vienna Initiative were somewhat more stable
lenders compared to banks that did not participate. Finally, there is no evidence
of Vienna Initiative banks retrenching from non-Vienna Initiative countries in
order to maintain exposures to countries where they signed commitment letters.
If anything, participation in the Vienna Initiative had positive, not negative,
spillover effects to other countries in emerging Europe.
5 Conclusions
The Vienna Initiative was a coordination effort to address the collective action
problem associated with the large exposures of Western European banks in
Emerging Europe. By managing the process of deleveraging in the interest of
the banks as a collective, large dead-weight losses from individual institutions
running to the exit could be avoided. Moreover, the Initiative aspired to prevent
negative spillovers from national crisis measures in home and host countries of
these institutions. The evidence suggests that it was effective in achieving these
objectives. International financial institutions played an important role in ini-
tiating and managing this coordination among private players and between
authorities in different jurisdictions, and between private and public actors.
The International Monetary Fund, of course, is an integral part of the global
financial architecture with its responsibility for financial stability. During the
crisis the IMF, together with the European Commission, played an important
role by integrating the commitments under the Vienna Initiative in their joint
programs. They both had strong incentives to ensure that private actors were
also bound by the programs. If they were not, the official sector could be seen
as simply allowing banks to withdraw funds from the program countries. The
World Bank also contributed to the Vienna Initiative, both in its role as a repos-
itory for regulatory experience and as a source of budget support to help shore
up troubled banking systems.
In a spirit similar to ours, Pistor (2012) discusses the Vienna Initiative as
an ‘alternative, coordinated multi-stakeholder governance framework’ in a
world where financial institutions operate globally but regulation has remained
national at the core. A critical prerequisite for such a regime is a coordinating
agent, or ‘anchor tenant’, in this case the EBRD, that is deeply vested in the
stability of transnational financial systems, but does not directly compete with
market actors or regulators.
As emphasized by Pistor, the international financial institutions often have
strong incentives to act when there are shortcomings in cross-border gover-
nance as they tend to internalize the costs of breakdown. For example, the
EBRD post-Lehman had very significant exposures to the entire financial sector
in Emerging Europe. Actions by home countries or individual host countries
of banks, or a breakdown of coordination among private banks, would have
important spillovers across the EBRD’s region of operations. The fact that one
third or so of its financial sector exposure in the region was in the form of equity
further incentivized the EBRD to intervene. The IMF and the World Bank also
internalize spillovers across program countries, for example, between the Greek
program and the various arrangements they have with other countries in South-
east Europe. However, as their reach is global their incentives are not as strongly
aligned with those of Emerging Europe. The European Union has a higher stake
in the success of efforts to stabilize the banking systems of Emerging Europe,
but many countries in the region are not members of the European Union.
The various international financial institutions play different roles and as
a result their financial linkages to individual economies are different. These
variations can lead to excessive competition and lack of coordination in crisis
situations. In the response to the global financial crisis an important element
was the Joint Action Plan of the World Bank, the European Investment Bank and
the EBRD. They jointly committed d24.5 billion to support the European banks
active in Emerging Europe. Even though individual transactions were negoti-
ated separately and the total amount was small compared to aggregate needs,
the signal of the plan and collaboration on due diligence, among other tasks,
most likely contributed to the stabilization of the situation in the banking sector.
An important aspect of the global financial crisis was the significance of the
private sector debt. Much of the debt that had accumulated was in the private
220 E. Berglof
sector and the large private banks were critical to any solution. Without private
sector participation coordination would most likely not have been effective.
International financial institutions, particularly those like the EBRD and the IFC
that regularly work with private investors, can play an important role in facili-
tating coordination within the private sector and between the private banks and
the authorities. Playing this role is much more difficult for the rest of the World
Bank Group and, in particular, for the IMF. Moreover, whereas the latter insti-
tutions can address spillovers across program countries, the EBRD and the IFC
can more directly address bank-driven spillovers from non-program countries
to program countries.
6 Epilogue
The roles of the various international financial institutions depend on the nature
of the crisis and differ across different stages of a crisis. As the global financial
crisis transitioned from a private sector debt crisis into a sovereign debt cri-
sis, the coordination challenges changed in nature. Despite reinforcements,
particularly at the level of the European Union, the institutional framework
supporting cross-border banking appeared even weaker at this phase of the
crisis. Many sovereigns were fiscally weakened and could no longer credibly
support their banks or provide stimulus to their economies. What had essen-
tially been a liquidity problem of the subsidiaries of the large banking groups
now became an issue of capitalization at the core of the parent banks. The cap-
italization challenge was reinforced by new capital requirements under Basel
III and anticipated regulation on systemically important financial institutions,
so-called SIFIs. In 2008–09 the fundamental model of cross-border banking
was viewed as sustainable, indeed worth encouraging, but by the end of 2011
the future of cross-border banking looked much more uncertain, in Emerging
Europe and globally. In particular, there were many challenges to the model
based on significant intermediation of capital flows from capital-rich Western
Europe to capital-poor Eastern Europe. Moreover, there was a perception that
banking itself was going through a fundamental transformation towards a model
with much less risk (cross-border banking being an important source of risk) and
lower returns.
This phase of the crisis posed a renewed threat to the banking systems of
Emerging Europe. Given the weak markets for bank equity globally the main
option for many large cross-border bank groups to meet capital requirements
was asset disposal. For banks which could not meet the new ratios even with
asset disposals governments would have to step in, and for those where govern-
ment resources were not adequate additional capital would have to come from
the newly created European Financial Stability Facility. To the extent that such
state aid was provided to a particular bank it would be asked to substantially
restructure its balance sheets, likely to primarily affect non-core assets, many
of them in Emerging Europe. To mitigate the short-term impact of deleverag-
ing, the European Central Bank was providing essentially unlimited liquidity
support and short to medium-term lending at very favorable rates to banks that
could provide collateral, generously defined.
As in 2008–09 the immediate threat at the end of 2011 was a rapid, even
chaotic deleveraging in Emerging Europe. These concerns were reinforced by
desperate policy measures by individual countries to protect sovereign ratings.
In particular, the Austrian bank supervisor on 21 November announced that the
capital requirements for Austrian banks active in Central and Eastern Europe
would be raised. The supervisor also imposed requirements on the ratio of local
funding of new lending in subsidiaries. These measures provoked massive criti-
cism from host countries who felt they had been singled out and not properly
consulted. They were later challenged by the European Commission. In addition
to the public sector actions, a number of key banks in the region announced
major changes in their priorities, shifting the balance in favor of home markets
away from Emerging Europe.10
It was against this background that the international financial institutions
behind the Vienna Initiative and the Austrian government agreed to meet in
January 2012 to discuss a Vienna Initiative 2.0. The public and private sector
participants had continued in several working groups to discuss a number of
issues arising out of the crisis, but the crisis management and resolution mech-
anism itself had been in hibernation after the exposure commitments were
phased out in the spring of 2011. At the time of writing, the Vienna Initia-
tive 2.0 was envisioned initially to be limited to the public sector parties with
the objective to strengthen coordination between home and host countries,
but also among host countries and among home countries. Once sufficient
agreement had been reached on the public sector side, the private banks would
once again be invited. An important new objective was to support the fledgling
institutions at the EU level, the European Banking Authority and the Euro-
pean Systemic Risk Board, at a time when national interests were increasingly
asserting themselves. However, it was also recognzed that these institutions nat-
urally would focus on cross-border banking within the European Union itself,
and that additional arrangements might be necessary to supplement the new
institutions.
The Initiative was re-launched in full recognition that the situation was dif-
ferent from 2008–09 and that neither the realistic end result nor the measures
to get there would necessarily be the same. While the fundamental risk of
uncoordinated deleveraging was very much the same, the parties involved rec-
ognized that the headwind would be much stronger this time. Both national
governments and the international financial institutions were much more
resource-constrained. Cross-border banking and indeed banking itself may well
222 E. Berglof
look very different once the latest phase of the crisis has worked its way through
the system.
Notes
1. The initiative was later formalized as the European Bank Coordination Initiative with
the IMF and the European Commission as co-chairs, and the EBRD as a member of the
steering committee. The description of the initiative relies heavily on de Haas et al.
(2011).
2. For details see www.ebrd.com/pages/news/press/2009/090227.shtml.
3. These letters were signed by the CEOs of the banks. While they were not legally bind-
ing, breaking the commitments would have led to reputational losses. The authorities
also used a combination of moral suasion and regulatory forbearance to ensure that
commitments were respected.
4. Commitment letters were signed for Romania and Serbia in March 2009, Hungary in
May 2009, Bosnia and Herzegovina in June 2009. Ukraine had an IMF program but
no commitment letters were signed.
5. In the case of Latvia the banks actually never signed the letters, but in practise they
largely behaved as if they had.
6. In the case of Romania, parent banks ultimately did not maintain full exposures.
With the exception of three banking groups, parent bank financing declined before
the commitments were reaffirmed (see IMF 2010).
7. In fact, the leaders agreed that state aid to a bank in one member state could not be
prevented from benefitting a subsidiary in another member state. This stipulation, to
be enforced by the DG Competition of the European Commission, in practise meant
that funds could not be prevented from going to any subsidiary even if it was in a
jurisdiction outside the European Union.
8. However, outflows from Emerging Europe continued over a longer period of time. See
Figure 11.2.
9. Some informal ring-fencing in the form of extraordinary procedures to control cap-
ital outflows was implemented, but on the whole host country reactions were very
restrained given the magnitude of the shock.
10. These banks included Commerzbank, Unicredit Bank and KBC.
References
Cetorelli, N. and L. Goldberg (2011) ‘Global Banks and International Shock Transmission:
Evidence from the Crisis’, Staff Reports 446, Federal Reserve Bank of New York.
De Haas, R., Y. Korniyenko, E. Loukoianova and A. Pivovarsky (2011) ‘Foreign Banks
During the Crisis: Sinners or Saints?’, EBRD Working Paper (London: European Bank
for Reconstruction and Development).
EBRD (2009) Transition Report: Transition in Crisis? (London: European Bank for Reconstruc-
tion and Development), at: www.ebrd.com/downloads/research/transition/TR09.pdf
EBRD (2011) Transition Report: Crisis and Transition: The People’s Perspective (London:
European Bank for Reconstruction and Development), at: www.ebrd.com/downloads/
research/transition/tr11.pdf
IMF (2010) Romania: Letter of Intent and Technical Memorandum of Understanding
(Washington, DC: International Monetary Fund), February.
Pistor, K. (2012) ‘Governing Interdependent Financial Systems: Lessons from the Vienna
Initiative’, Journal of Globalization and Development, vol. 2, no. 2, forthcoming.
Popov. A. and G. Udell (2010) ‘Cross-border Banking and the International Transmission
of Financial Distress during the Crisis of 2007–2008’, European Central Bank Discussion
Paper, series 1203, at: www.ecb.int/pub/pdf/scpwps/ecbwp1203.pdf
Vallee, S. and B. Pawlovski (2011) ‘Safety Nets and Financial Flows to Emerging Europe
during the 2008/2009 Financial Crisis’, Bruegel Working Paper (Brussels: Bruegel),
Vienna Initiative (2009) Progress Report on the Joint IFI Action Plan, EBRD, EIB, World Bank,
5 October.
12
Some Recent Progresses on Financial
Structure and Development∗
Justin Yifu Lin
Peking University, China
Lixin Colin Xu
World Bank
I Introduction
Financial structure differs greatly across countries. In bank-based financial sys-

tems such as in Germany, Japan (until a decade ago), and India, banks offer
the main financial services in mobilizing savings, allocating capital, monitoring
corporate managers, and providing risk management services.1 In market-based
systems such as in the UK, the US, and Malaysia, both stock markets and banks
play important roles in all financial services. What explains financial structure?
Does the combination of institutions and markets that constitutes the financial
system have any impact on economic development? These questions have fasci-
nated economists for decades. One of the earliest attempts to address these ques-
tions was Goldsmith (1969), who 40 years ago tried to document the change of
financial structure over time and to assess the impacts of financial development
on economic development. He states that ‘one of the most important problems
in the field of finance, if not the single most important one, almost everyone
would agree, is the effect that financial structure and development have on eco-
nomic growth’. With data from 35 countries for the pre-1964 period, he finds
positive correlation between financial development and economic growth. But
he could not go far on financial structure due to data constraints: he could only
rely on careful comparisons of Germany and the United Kingdom. Obviously it
is hard to extend the conclusions from case studies to the rest of the world.
Since the era of Goldsmith, there has been great progress in research on finan-
cial structure. Having collected comprehensive cross-country data on financial
structure themselves (along with their co-authors) and using this new data set,
Demirguc-Kunt and Levine (2001) find that financial systems become more com-
plex as countries become richer with both banks and markets getting larger,
more active, and more efficient. But in general, the structure becomes more
224
Recent Progresses on Financial Structure and Development 225
market-based in higher-income countries. They also find strong and consistent

evidence that what matters for economic development is the level of financial
development, and that the relative mix of banks and stock markets does not
matter much (Beck et al. 2001).
This conclusion as to the irrelevance of financial structure for development
does face significant challenges. Several authors argue theoretically that finan-
cial structure likely should matter a great deal. After all, economic development
increases the demand for the services provided by securities markets relative
to services provided by banks (Allen and Gale 2000; Boyd and Smith 1998).
Moreover, banks and stock markets exhibit distinct effectiveness in delivering
corporate governance and investor protection (Stulz 2001), and may have dif-
ferent effectiveness under different developmental stages (Gerschenkron 1962;
Aoki and Patrick 1994; Hellman et al. 1996). In particular, banks are better
at reducing the market frictions related to financing standardized, shorter-run,
lower-risk, and well-collateralized projects, while security markets are better in
financing more innovative, longer-run, and higher-risk projects that rely more
on intangible inputs such as human capital (Allen and Gale 2000). Furthermore,
the fact that stock markets become significantly more active and important as
economies develop also conjures up the notion that stock markets may become
more useful as income levels rise.
In this chapter, we summarize some recent progress, both theoretical and
empirical, that suggests that financial structure matters for economic develop-
ment, that banks and stock markets play different roles in countries at different
development stages, and that there might be an optimal financial structure
associated with each development stage. We also offer evidence that the actual
financial structure in a country may deviate from its optimal state due to politics.
In the rest of this chapter, we first summarize conventional wisdom and find-
ings about financial structure. We then proceed to discuss some new ideas on
this topic, along with empirical support for these new ideas.
II Traditional views and empirical results
While the literature on the relative merits of banks versus markets is large, it can
be summarized by four views (Beck et al. 2001; Levine 2002; Stulz 2001). The
first is the financial structure-irrelevancy view. In a perfect capital market with risk-
neutral agents, the interest rate determines which investment opportunities are
worth taking up, and all investment opportunities yielding positive net return
(after capital costs) will be taken (Stulz 2001). If there is imperfect capital mobil-
ity, that is, capital flows across borders may be hindered by country-specific
risks, then what matters for job creation, firm growth, and efficient allocation
of resources is a financial system that can provide efficient financial services
and that provides sufficient access to finance, and the mixture of banks and
226 J. Y. Lin and L. C. Xu
markets does not matter. According to this view, only financial depth but not
its structure matters for economic performance.
The second view, a particular version of the financial structure irrelevance
view, is the law and finance view, which argues that the primary determinant of
the soundness of the financial system is the legal system (La Porta et al. 1998,
2000). In particular, this view holds that what is relevant for growth is not
financial structure, that is, whether it is bank-based or market-based. Rather, it
is the overall financial development, which is in turn determined by the legal
system and law origins. The legal system may affect external finance because
good legal protection would increase investors’ confidence that the returns from
investments (managed by firms) would be (partially) returned to them – either
by banks or by investors directly – and as a result, they are more likely to provide
such funds to firm managers (La Porta et al. 2000; Stulz 2001).
Underlying the financial structure irrelevancy view are strong assumptions
which may not hold in reality. When savings can be invested inefficiently due
to differences in financial structure – when the financial system fails to direct
savings to its more efficient uses – financial structure becomes important (Stulz
2001). Two key market imperfections destroy perfect financial markets (Stulz
2001): managers have information advantage over investors about the firm’s
activities (‘hidden information’), and managers’ actions cannot be observed by
investors (‘hidden action’). Hidden information and hidden action allow man-
agers to have discretion to pursue their own objectives. And, managers cannot
credibly commit to return investment returns to investors, who in turn may
fail to finance projects that may have positive returns in a perfect-information
world. With these two issues in mind, financial structure leads to real conse-
quences when it changes information and transaction costs, affects the cost of
capital, and alters the incentives and monitoring of management.
The third view is the bank-based view, which emphasizes the positive role
of banks in mobilizing resources, identifying good projects, monitoring man-
agers, and managing risks, and highlights the shortcomings of the stock market
(Beck et al. 2001). One of the pioneers in research on financial structure, Ger-
schenkron, suggests that banks are better than markets in the early stage of
economic development when the institutional environment cannot support
market activities effectively (Gerschenkron 1962). The reason is that even in
countries with fragile legal and accounting systems and frail institutions, pow-
erful banks can rely on self-enforcing, forcing firms to reveal information and
pay their debts, thereby helping industrial growth (Rajan and Zingales 1999).
Moreover, banks may be better at providing external finance to new firms requir-
ing staged financing than markets: banks can more credibly commit to making
additional funding available as the project proceeds, while markets find it more
difficult to make credible, long-term commitments. Banks may also provide bet-
ter corporate governance in countries at the early stage of development (Aoki
and Patrick 1994). In contrast, a good stock market quickly and fully reveals
information in public markets, which decreases the incentives for investors to
acquire information (Stiglitz 1985). Great market development may thus impede
incentives for identifying innovative projects, and thereby hinder efficient
resource allocation. Moreover, liquid markets also lead to a myopic investment
sentiment – all investors need to do is to watch stock prices without having to
monitor firm managers actively, which hinders corporate control (Bhide 1993).
In contrast, the fourth view, the market-based view, regards stock markets as cru-
cial in promoting economic success (Beck et al. 2001). Markets allow investors
to diversify and manage risks more effectively, thereby encouraging greater sup-
ply of external finance. Market-based systems also facilitate competition, which
induces stronger incentives for R&D and growth. Thus, market-based systems
may be especially effective in promoting innovative and more R&D-based indus-
tries (Allen and Gale 2000). Liquid stock markets also allow investors to build
and seek large stakes, therefore enabling hostile takeovers to discipline shirking
or incompetent managers (Stulz 2001). This market-based view also emphasizes
the negative roles played by banks. By spending expensive resources on infor-
mation about firms, banks can extract large rents from firms, which reduces the
incentives for firms to undertake high-risk, high-return projects since firms lose
a large share of the rents to the banks (Rajan 1992). Moreover, because of the
nature of the debt contracts – banks do not benefit from high returns but may
be harmed by high risks – banks prefer to finance safe and low-return projects,
retarding innovation and growth. Moreover, powerful banks may collude with
firm managers to prevent entry by other investors, which reduces competition
and effective corporate control and therefore growth.
Demirguc-Kunt and Levine (2001) use the new cross-country database on
financial structure to document how financial structure evolves with economic
development. They characterize financial structure by ratios of banking sec-
tor development (measured in terms of size, activity, and efficiency) relative to
stock market development (similarly defined), and a higher ratio means a more
bank-based structure, and classify countries into bank-based or market-based
countries. Demirguc-Kunt and Levine find that banks, nonbanks, stock markets,
and bond markets are larger, more active, and more efficient in richer countries,
confirming the findings of Goldsmith (1969) with a smaller sample of coun-
tries in earlier periods. Thus financial systems on average are more developed in
richer countries. In addition, stock markets in higher-income countries tend to
be more active and efficient relative to banks. Furthermore, financial structure
is more market-oriented in countries with a common law tradition (as distinct
from a civil law tradition), strong protection of minority shareholder rights,
good accounting systems, low levels of corruption, and no explicit deposit insur-
ance. This is consistent with theories that argue that higher information costs
and worse legal protection of property rights tend to favor banks over markets
(Allen and Gale 2000; Stulz 2001).
Beck et al. (2001) provide comprehensive evidence that financial structure
does not matter, but financial depth matters. The paper combines the new
cross-country database of financial structure with both firm-level and cross-
country industry-level data. Relying on evidence about financial structure and
economic performance at three levels (pure cross-country comparisons; cross-
industry, cross-country methods; and firm-level data across many countries),
they obtain consistent results. They find no evidence that financial structure
helps explain country economic performance: ‘Countries do not grow faster,
financially dependent industries do not expand at higher rates, new firms are
not created more easily, firms’ access to external finance is not easier, and firms
do not grow faster in either market-based or bank-based systems.’ In contrast,
distinguishing countries by overall financial development does help explain

cross-country differences in economic performance. Measures of bank devel-
opment and market development are strongly linked to economic growth.
More specifically, the data indicate that economies grow faster, industries
depending heavily on external finance expand at faster rates, new firms form
more easily, firms’ access to external financing is easier, and firms grow
more rapidly in economies with a higher level of overall financial-sector
development.
They also find that the part of financial development explained by the legal
system consistently explains firm, industry, and national economic success,
consistent with the law and finance view of financial structure.
III New waves of theoretical arguments
Does financial structure really not matter for development? Several doubts have
emerged over time. One source of doubt comes from the hunch that there is
often no one-size-fits-all recipe for development (Kremer 1993; Hellman et al.
1997). Countries often differ in areas with the largest reform payoffs, and
there are often development ‘bottlenecks’, which conjures up the image of
the famous failure of the space shuttle Challenger: with thousands of compo-
nents, it ‘exploded because it was launched at a temperature that caused one
of those components, the O-rings, to malfunction’ (Kremer 1993). Consistent
with this notion of country-specific and development-stage-specific bottlenecks,
some research has found policy complementarity in various contexts. In particu-
lar, Xu (2011) summarizes evidence that suggests that the effects of the business
environment on development tend to be heterogeneous depending on the stage
of development, and that in particular bad infrastructure and labor inflexibility
in India tend to be key bottlenecks due to their negative indirect effects.
Another source of doubt is theory on financial structure. Relying on the com-

parative experience of Germany, Japan, the United Kingdom, and the United
States, Allen and Gale (2000) examine whether financial structure matters.
They conclude that since banks and markets offer distinct financial services,
economies at different stages of development require distinct mixtures of these
financial services to work efficiently.2 They conjecture that as a country becomes
richer, it will require different mixtures of financial services (that is, banks and
stock markets) (Boyd and Smith 1998), and that when a country’s actual finan-
cial structure differs from the appropriate mixture of banks and markets, the
economy will not obtain the appropriate mix of financial services, hurting
economic growth.
Lin et al. (2011) argue that financial structure has to matter for development
from a different angle. The key reason is that an efficient financial structure
must reflect the demand of the real economy. Fundamentally, factor endow-
ments (labor, capital, and natural resources) determine industrial structure,
which in turn needs the support of a certain development-stage-specific finan-
cial structure. In particular, at each stage of development, a country has a specific
combination of factor endowments. That combination determines factor prices,
which in turn determine the optimal industry structure, its associated risk nature
and distribution of firm size (Lin 2009). Since enterprises that operate in differ-
ent industries differ in terms of size, risk, and financing needs, the demands
of the real economy for financial services at some development stages can be
systematically different from those of the same economy at other stages. When
the characteristics of financial structure match those of an economy’s indus-
trial structure, the financial system can perform its fundamental functions most
efficiently and thus contribute to sustainable and inclusive development. There-
fore, there is an appropriate financial structure for an economy at each stage of
development.
With respect to developing economies, the key characteristic of their endow-
ment structures is the relative abundance of unskilled labor (and scarcity of
capital). Labor-intensive industries and the labor-intensive sections of capital-
intensive industries have comparative advantages and thus should dominate
these economies. Since the experiences from developed economies can be
mimicked, the industries, products, and technologies that are appropriate in
developing economies are relatively mature. With respect to firm size, firms in
labor-intensive industries are usually smaller, especially in terms of capital, rela-
tive to firms in capital-intensive industries. The efficiency of the financial system
in developing countries therefore depends on its ability to serve the financing
needs of labor-intensive, small, and mature businesses. Since those firms also
tend to be more opaque due to their lack of standard financial information,
screening firms and monitoring firm managers become the major concerns for
providers of external funds to these firms. In such an economic environment,
banks, especially small local banks, have more strengths than stock markets,
due to their superior abilities to harness local information, assess ‘soft’ infor-
mation regarding creditworthiness, and engage in long-term relationships with
borrowers. In addition, banks are particularly attractive to firms in low-income
countries because banks represent lower costs of capital to firms in such coun-
tries. First, when borrowing from no more than a few banks, firms do not need
to have public information such as financial statements and external auditing
ready for the lenders, thus saving precious capital in low-income countries.
Moreover, low-income countries often do not have the legal and institutional
framework to provide public information inexpensively, reflecting the path-
dependent nature of institutional development (Aoki and Kim 1995), and the
lack of sufficient demand for such institutional infrastructure. Second, interest
rate payments for loans tend to be lower than returns to shares in the stock
market due to lower risks associated with bank loans, a fact that again saves
precious capital from the perspective of firms in developing countries. Thus, if
there are no distortions, the financial systems in these economies are likely to
be characterized by the dominance of banks.
It is likely that small regional banks play an especially significant role to serve
efficiently small firms in developing countries. Recent evidence suggests that
there is a match between bank size and the size of firms that these banks serve.
Large firms tend to shy away from small businesses but rather focus on large
businesses, while small banks tend to target small businesses. Large banks can
save transaction costs if making loans largely to large businesses – since making
a loan largely involves the same procedure and filling in a similar amount of
forms, making large loans to large businesses would lower the unit costs of loans
for large banks. Serving small firms is thus left to small banks in developing
countries.3
In contrast, for developed countries, the key characteristic of their endowment
structure is the relative abundance of skilled labor and capital. The compara-
tive advantage of these countries is then capital-intensive industries. Firms in
such industries tend to be large, demanding more external financing. Since
these countries are already at or near the technological frontier, firms there
would spend much more resources on R&D and innovations, and firms will
bear higher risks for technological innovation and product innovation.4 With
larger firm size, firms can afford the (more or less) fixed costs of providing stan-
dard financial information to the market, and specialized financial agencies can
make sufficient money and become viable in providing specialized financial
and auditing information. Thus with standard financial information available,
stock markets, bond markets, and big banks become the main finance providers
to these capital-intensive firms.
Moreover, there are arguments that stock markets would be better suited
to richer countries. For firms with new technologies or innovative projects,
investors do not have much information and often have diverse opinions about
the prospects of these new technologies. Decentralized stock markets allow peo-
ple to agree to disagree about the future prospects of these firms, and these firms,
as a result, are more likely to get funded (Allen and Gale 1999, 2000). Further-
more, stock markets can take advantage of the standard financial information –
information available only in richer countries – to reduce the information asym-
metry between the managers of a firm and the external investors, which allows
investors to make more informed decisions about what firms to invest in and
in which firms they are more likely to have safer returns. With very high risks
for innovative and capital-intensive firms, venture capital is often involved in
the early stage of these innovative firms, but stock markets remain crucial by
providing exit options for venture capital and financing further development
of these high-tech businesses. Banks can also offer such staged investment once
venture capital has identified good projects as demonstrated by good initial
returns. Thus, for rich countries, the appropriate financial structure is likely
characterized by a large and active stock market, augmented with many large
banks.5
As a result, for a country at a certain stage of its economic development,
some specific financial structures will be more efficient in mobilizing and
allocating capital. In other words, there is a certain appropriate financial
structure in a specific stage of development, in which the composition and
relative importance of available financial arrangements can most efficiently
allocate financial resources to viable firms in the competitive sectors of the
optimal industrial structure, which is in turn determined by its endowment
structure. The appropriate financial structure for developing countries tends
to feature a stronger role for banks (especially relatively small banks) than
stock markets, while the opposite is true for developed countries. Moreover,
the optimal financial structure is dynamic. As endowment structure changes
with physical and human capital accumulation, the appropriate industrial and
financial structure would change accordingly. There is therefore no unique
financial structure that fits all countries. For future reference, we call this
view of optimal financial structure specific to each development stage the ‘new
structural view’.
IV Some new empirical results
Several recent papers offer supporting evidence that financial structure matters
in various ways for economic development. In one of the earliest and most
systematic papers that addresses the issue of financial structure and economic
performance, Carlin and Mayer (2003), uses a sample of industry-level panel
data from 14 countries from 1970 to 1995 in the OECD along with data from
four countries in earlier stages of development (Korea, Mexico, Portugal, and
Greece) to examine the determinants of firm growth and investment in physi-

cal and R&D capital. The two samples differ significantly in income per capita:
The four less-rich countries had a GDP per capita of $2200 to $6300 in 1970,
compared to $7300 to $15,000 in the 14 OECD countries. The particular ques-
tions that they are interested in are how the effects of financial structure (that is,
financial disclosure, bank concentration, and ownership concentration) depend
on industry characteristics (that is, equity reliance and bank reliance, among
others), and whether the effects differ by the stage of development (that is,
OECD countries versus middle-income countries such as Korea, Mexico, Por-
tugal, and Greece). They use the Rajan-Zingales-type regressions in which the
industry performance can be affected by industry characteristics interacted with
country-level financial structure variables, controlling for country and indus-
try dummies to hold constant all country and industry-specific time-invariant
factors.
For developed countries, accounting disclosure (or financial infrastructure) is
associated with faster growth of industries that are equity- and skill-dependent.
Such industries also have higher R&D intensity. Information disclosure is found
to matter more than the size of the stock market or the size of the banking
system, implying the key importance of information disclosure in explaining
the finance-growth link or the finance-innovation link.6 Bank concentration is
associated with slower growth and lower R&D intensity in equity-dependent
industries and of R&D intensity in skill-dependent industries. In general, the
interaction between a country’s financial structure and industry characteristics
is important for R&D but not for fixed investment.
The behavior of those middle-income countries differs dramatically from
that of OECD countries. The Chow-test suggests that the behavior of the
low and high-income countries differs significantly. In those four middle-
income countries (Korea, Mexico, Portugal, and Greece), bank concentration
(rather than competition) and accounting standards are associated with higher
growth of bank-dependent industries, and lower growth of skill-dependent
industries.7 This is not so for developed countries. In developed countries,
stock market development is found to be irrelevant for bank-dependent indus-
tries. In the poorer four countries, there is a negative relationship of growth
in bank-dependent industries with stock market development, but a posi-
tive relationship with the size and concentration of banking systems, along
with information disclosure. Thus, reasonable bank concentration and infor-
mation disclosure turn out to be useful for bank-dependent industries. This
is consistent with Huang and Xu (1999), who argue that single-bank and
concentrated-creditor systems favor industries with lower uncertainty and imi-
tative investments, while a dispersed-creditor system is more efficient for
R&D-intensive industries featuring young firms and high uncertainty. Similarly,
Gerschenkron (1962) suggests that in developing countries featuring imitative
and mature economies, a concentrated bank system may work better, while
in developed countries featuring high uncertainty and more innovation, bank
competition is preferred (Gerschenkron 1962).
Shifting from industry-level cross-country panel data to a much larger cross-
country panel data, Demirguc-Kunt et al. (2011) provide perhaps the most
dramatic evidence on financial structure and economic development. Noting
that the past literature has rarely been successful in identifying the impor-
tance of financial structure in a cross-country setting, they explore whether
deviations from an optimal financial structure are associated with the speed of
development.8 They use data from 72 countries from 1980 to 2008 to reassess its
role in economic development. More specifically, they assess whether the sensi-
tivity of economic development to increases in bank development and increases
in securities market development changes during the process of economic devel-
opment, and whether each level of economic development is associated with
an optimal financial structure. Financial structure here is measured as private
credit (over GDP) over security market capitalization (over GDP) and some of
its variants.
Demirguc-Kunt et al. use quantile regressions to assess how the sensitivities of
economic activity to banks and securities market development evolve as coun-
tries grow. The quantile regressions provide information on how the associations
between economic development and both bank and securities market develop-
ment change as countries grow richer. In contrast, conventional cross-country
studies tend to focus on the association between economic development and
financial structure for the ‘average’ country. The reliance on quantile regres-
sion, which implicitly insists that the effects of financial structure have distinct
effects for countries at different income levels, proves to be the key for finding
that financial structure matters.
A measure of optimal financial structure at each level of development is con-
structed by regressing a measure of financial structure over GDP per capita for
the sample of OECD countries, while also controlling for key institutional, geo-
graphic, and structural traits of those countries. The maintained hypothesis is
that conditional on these traits, the OECD countries provide information on
how the optimal financial structure varies with economic development. Next,
the authors use the coefficients from the OECD regression to compute the esti-
mated optimal financial structure for each country in each year. They then
compute a ‘financial structure gap’ which is equal to the natural logarithm of the
absolute value of the difference between the actual and the estimated optimal
financial structure.
They find that as economies develop, both banks and markets become larger
relative to the size of the overall economy. More importantly, as countries
become richer, the sensitivity of economic development to changes in bank
development decreases, while the sensitivity of economic development to
changes in securities market development increases. Thus the relative demand

for the services provided by stock markets increases as an economy develops,
and these services differ from what banks provide, as suggested by Boyd and
Smith (1998) and Allen and Gale (2000). Boyd and Smith suggest that in devel-
oped countries, monitoring is relatively more expensive than capital costs, so
technologies that involve relatively lower monitoring are more appropriate –
that is, equity rather than debt. This explains why equity markets become more
important as countries develop.
They find some support for the notion that there is an appropriate financial
structure for countries at distinct stages of development. In particular, devia-
tions in an economy’s actual financial structure from its estimated optimal one
(that is, the size of the financial development gap) are associated with reduced
economic output. Even when controlling for the level of bank development,
securities market development, a standard set of controls, and country fixed
effects, there is a robust and negative relationship between the financial struc-
ture gap and economic activity. They also look at whether it matters if the
non-optimal financial structure is due to too much bank orientation or too
much market orientation, and find it not to matter. Magnitudes of the effects of
deviations from the optimal financial structure are non-trivial: a one-SD increase
in the financial structure gap is associated with a drop in log real GDP per capita
of 0.06, that is, a 6 per cent reduction in economic activity. Further controlling
for country and period fixed effects and some standard controls, the magnitude
drops by 50 per cent but remains significant. The magnitude, at face value, is
interesting: it is certainly non-trivial, but the magnitude is not overwhelmingly
important.
It is useful to point out that this chapter does not deal with the potential
endogeneity of the financial structure. If financial structure does respond to
income level, as the theory would suggest, then there may be bias for the esti-
mated effects of financial structure. Moreover, it is not clear that using OECD
countries to infer optimal financial structure is completely convincing. After all,
OECD countries are all rich, and it is unclear that such a collection of countries
can form the base to infer optimal financial structure in much poorer countries.
Still, the paper offers plausible empirical support for the new structural view of
financial structure, and is a nice first step toward disentangling the mystery of
the impact of financial structure.
Another clue about the effects of financial structure comes from a large firm-
level data set. Cull and Xu (2011) use firm-level data across 89 countries, looking
at how labor growth rates of firms vary with their country’s financial structure.
An important advantage of combining firm-level data and cross-country indi-
cators of financial structure is that we are able to examine how various types
of firms may be affected differently by financial structure. Such heterogeneity
across firm size is a key characteristic to distinguish the efficiency-based view
(that is, the new structural view) and the political-economy-based approach to
understand the evolution of financial structure.
Cull and Xu regress firm-level labor growth rates on country-level measures
of bank development and stock market development (after controlling for basic
firm and country characteristics). They are concerned about the potential endo-
geneity of financial structure in the labor growth equation for two reasons. First,
there might be omitted variables that are correlated with both financial structure
and labor growth rates. Such variables might, for instance, include any non-
finance business environment variables (Xu 2011). Second, causality might go
both ways, from finance to firm growth or vice versa. They thus resort to instru-
mental variables to deal with such issues. In particular, they consider potential
instrumental variables including natural resource dependence, the level of trust
in a society, cereal plantation patterns, settler mortality, and so on, and choose a
subset of these potential instrumental variables that are both related to financial
structure yet pass the over-identifying restrictions test. Beside the instrumen-
tal variable approach, they also use the Rajan-Zingales difference-in-difference
approach to examine whether firms in industries that rely more heavily on
external finance benefit more in terms of firm growth from financial develop-
ment at the country level, holding constant both country and industry fixed
effects, therefore controlling for all country- and industry-specific factors. This
approach significantly reduces the extent of omitted variable bias.
Relating firm growth to firm and country characteristics and financial struc-
ture, and taking into account the potential endogeneity of financial structure,
Cull and Xu find that labor growth is swifter in low-income countries that have
a higher level of private credit/GDP, and the growth-spurring effects of bank-
ing development are especially pronounced in industries that heavily rely on
external finance. In high-income countries, labor growth rates are increasing in
the level of stock market capitalization. Both patterns are consistent with pre-
dictions from the new structural view and some earlier theoretical conjectures
(Allen and Gale 2000; Boyd and Smith 1998; Lin et al. 2011).
A further clue about the effects of financial structure emerges from exam-
ining the impact of financial structure on poverty. Financial structure might
affect poverty because entrepreneurs have trouble obtaining finance due to
information asymmetry between them and investors – they know more about
the prospects of the projects than banks and atomistic investors in the stock
market. A number of researchers argue that banks are better able to reduce
this information asymmetry problem than stock markets. One reason is that
banks form long-term relationships with borrowers, and can benefit from the
value of the information obtained from this long-term relationship. In contrast,
well-established stock markets quickly and publicly reveal information, thereby
reducing the incentives for individual investors to acquire information. Banks
therefore may have better capacity to reduce the information asymmetry issue
and make external financing possible. Moreover, since stock markets rely more
strongly on the legal and accounting framework to safeguard necessary returns
to investors, the effects of stock markets may depend on institutions to a greater
extent, whereas banks can more effectively force firms and households to honor
their contracts than stock markets (Gerschenkron 1962; Boyd and Smith 1998),
and are therefore especially important in poorer countries with weak contract
enforcement.
Based on the above logic, Kpodar and Singh (2011), using data from 47 devel-
oping countries from 1984 to 2008, show that financial deepening through
banks is associated with reduced poverty levels, while market-based measures
of financial development are associated with higher incidence of poverty in
this sample. In addition, the interaction between institutional quality and the
size-based measures of the importance of stock markets relative to banks is nega-
tive and significant in their regressions, indicating that as institutions improve,
the positive link between market-based financial development and poverty
incidence phases out and even reverses after some threshold of institutional
quality is reached. Conversely, the results suggest that in weak institutional
environments bank-based financial systems tend to reduce poverty more than
market-based ones. The authors have dealt with the endogeneity of financial
structure for poverty by using the system GMM framework, which controls for
country fixed effects, and allows financial structure and other variables to be
endogenous and predetermined.
V The deviation from the optimal financial structure
Besides derived demand based on industrial structure that originates from the
endowment structure, there are other determinants of financial structure, which
cause the actual financial structure to deviate from the optimal one. Earlier
research has shown that financial structure is significantly and robustly related
to law and legal origins (La Porta et al. 1998, 2000; Demirguc-Kunt and Levine
2001); here we focus on several other factors such as the role of belief and ideas
and the role of politics, which have emerged as potentially important in recent
studies.
The first factor stems from the belief of government leaders.9 In most devel-
oping countries, the government plays a very important role in defining the
structure of the economy. And the belief of the top government leaders will
naturally shape the country’s financial structure. A case in point is financial
repression that is widely observed in many developing countries. Countries
featuring financial repression tend to adopt policies restricting entry into the
banking sector, controlling interest rates, and intervening in the allocation of
bank loans. As a result, a few big banks tend to dominate the banking land-
scape, and capital tends to flow to large firms. Small businesses, which have
comparative advantage in these economies, get little access to credit and have to
make do with internal capital or resort to informal channels for external finance.
Why do countries adopt such obviously inefficient policies? Inappropriate
development strategies adopted by the government are likely the main driving
force leading to these repressive policies and distorted financial systems.10 If the
government’s priority is to promote industries that are inconsistent with the
comparative advantages endogenously determined by the economy’s endow-
ment structure, it has to use distortional policies so as to channel scarce resources
into the priority sectors. As a result, government interventions and consequent
repression of the financial system are inevitable.11 Due to inertia of institutional
change, such distorted policies can have prolonged influence on the evolution
of the financial system.
A good example of this practice is China. In the 1950s, the factor endowments
in the Chinese economy were characterized by extreme scarcity of capital and
enormous abundance of labor. The government, however, decided to adopt
an ambitious comparative-advantage-defying development strategy in which
establishment and development of heavy industries took the first priority. In
order to push the development of heavy industries, which are very capital-
intensive, the government had to deliberately distort prices of various products
and production factors including labor, capital, foreign exchange, and oth-
ers, replace the market mechanism with a government planning system so as
to control the allocation of production factors, nationalize private businesses,
and collectivize agricultural production with the People’s Communes. In this
centrally planned economic regime, banks were closed or merged into the Peo-
ple’s Bank of China, which became the only financial institution in the whole
economy until the end of the 1970s. After the reform and opening in the late
1970s, the government adopted a dual-track approach to the transition: on
the one hand, some transitory protections and subsidies have been provided
to firms in the old priority sectors, and, on the other hand, entry to the sec-
tors that are consistent with the economy’s comparative advantages and were
repressed in the old strategy has been liberalized. As part of the economic reform,
four big state-owned banks were established in the early 1980s. A dozen joint-
stock commercial banks were also set up in the late 1980s and early 1990s. But
interest rates are still under control of the state and domestic entry into the
banking sector is rigidly restricted by the government. The market share of the
four big state-owned banks has slowly declined, but they still hold a dominant
position in the banking system today. Because of this serious mismatch of finan-
cial structure with optimal industrial structure, labor-intensive small businesses
have very limited access to formal financial credit; this reduces job creation and
contributes to widening inequality of income distribution in China.
A second factor behind the deviation of the actual financial structure from
its optimal one is due to the belief of many policy advisors regarding financial
liberalization and the possibility of leapfrogging in financial development. As a

policy prescription to correct financial repression, financial liberalization has
been generously prescribed by theorists and exercised by many developing
countries. While those repressive policies should be removed, some new, less
noticeable policy distortions may be introduced in the process of financial lib-
eralization. It is not rare that developing countries are advised to establish and
develop a financial system similar to that in the advanced economies. The US
and UK financial systems, where financial markets are highly active, are often
taken as the best model that developing countries should follow. This model is
often justified by the supposed superiority of financial markets. As a result, some
low-income small economies are eager to develop stock markets, consolidate
small banks into large banks and repress the development of local banks.
However, as the new structural view has argued – and with supporting evi-
dence emerging – the optimal financial structure for poor countries is likely
to be systemically different from that for advanced economies. Thus imitating
the financial model of advanced economies will not lead to improved effi-
ciency of the financial system nor generate better economic performance in
poor countries. Such imitation may even result in destructive consequences
such as financial crises. Such policy advice is also inconsistent with the growth
experience of those successful economies in their industrialization periods.
For instance, in the British Industrial Revolution, industrial enterprises were
typically very small and mainly internally financed at both the start-up and
expansion stages. In the case of external finance, personal contacts played a
crucial role. The role of the banking system in financing long-term investment
in industrial sectors was insignificant. The British banks were typically small and
locally based with a limited number of offices until at least the mid-19th century.
The bank merger movement in England did not develop until the 1860s, with
the peak of merger activity occurring in the late 1880s and early 1890s. When it
comes to the role of capital markets, history shows that capital markets started to
play an important role in financing industrial sectors only at the end of the 19th
century. In the United States before 1890, industrial firms were numerous, small,
and closely owned. Industrial securities, except in the coal and textile industries,
were almost unknown. A capital market for industrial preferred stocks devel-
oped in the 1887–1904 merger wave. And public markets for common stocks
developed even later. Therefore, while financial markets are a prominent part
of the current financial systems of the UK and the US, this was not the case at
the early stages of economic development in these countries. According to Cull
et al. (2006), during the 19th and early 20th centuries, a variety of local financial
institutions emerged to meet the needs of small and medium-sized firms in the
economies of the North Atlantic Core. These financial intermediaries were able
to tap into local information networks and so extend credit to firms that were
too young or small to get funds from large, financial center banks.
The third factor causing deviation of reality from optimal financial structure
is politics, as argued by Calomiris and Haber (2011) in the case of bank crisis –
in which case the financial structure is clearly not optimal. Many under-banked
economies repeatedly supplied credit imprudently: once a crisis was over, banks
appeared to continue misallocating scarce credit to firms and households that
were default-prone. Why? Calomiris and Haber rely on reasoning rooted in polit-
ical economy that can explain the prevalence of fragile banking systems that
allocate credit narrowly. The key reason is that government actors face inher-
ent conflicts of interest when it comes to the operation of the banking system,
and those conflicts can lead to banking instability and undersupply of credit.
Specifically, governments regulate and supervise banks to limit risk-taking but
they also rely on banks as a source of risky public finance (by borrowing from
them and taxing them). In addition, while governments enforce contracts that
discipline bank borrowers, they also depend on bank debtors for votes or politi-
cal support. Finally, governments distribute losses among creditors when a bank
fails, but they also must depend on the largest creditor group – bank depositors –
for their political fate. These conflicts of interest imply that regulatory policies
toward banks often reflect the interest of the political coalitions that support
the government.
This political-economic framework turns out to be very useful for understand-
ing banking structure in a series of historical case studies (Scotland, England,
the United States, Canada, Mexico, and Brazil). Indeed, formation of viable
political coalitions under different types of government dictated the evolution
of banking structure. Adapting the conceptual framework to the historical case
studies leads to a number of conclusions. Foremost, the nature of the coalitions
that generate barriers to entry in banking varies across types of political regimes.
In autocracy, it is easier to create a stable coalition in favor of tight entry restric-
tions, in part because potential borrowers from banks do not have a voice in
the political process. Autocracies therefore tend to create banking systems that
allocate credit narrowly to the government and to enterprises owned by an elite
class of government-selected bankers. In the meantime, the narrow allocation
of credit under authoritarian regimes has not resulted in greater banking sector
stability: in times of economic strife, bank insiders and the government expro-
priate firms and households that are either loosely or not at all affiliated with
the coalition (that is, minority shareholders and depositors). In times of extreme
difficulty, the autocrat can (and has) expropriated bank insiders.
Mass suffrage, by giving voice to mass economic actors, makes it harder to
sustain a banking system that allocates credit narrowly to an elite group. It
does not, however, necessarily guarantee banking stability. Bank borrowers can
vote for representatives that expand the supply of credit, improve the terms
on which the credit is offered, and then forgive those debts when they prove
difficult to repay. This was largely the story of the US subprime crisis. Under
any type of political system, banking systems are fragile. Therefore, only a small
share of countries have been able to enjoy stable banking along with broad credit
supply, because this outcome requires political institutions that allow for mass
suffrage, but also limit the authority and discretion of the parties in control of
the government.
To shed light on whether real financial structure tends to deviate from the
optimal financial structure, Cull and Xu (2011) examine what types of firms ben-
efit more from private credit market development. In particular, they allow the
private credit variable (that is, private credit over GDP) to interact with firm char-
acteristics such as firm size and capital intensity in the labor growth equation,
estimated at the firm level. The authors find no evidence that small-scale firms
in low-income countries benefit most from private credit market development.
Rather, the labor growth rates of large and capital-intensive firms increase more
with the level of private credit market development. This suggests that the actual
financial structure likely deviates from the optimal financial structure. In par-
ticular, the likely scenario is that banks in poor developing countries tend to
lend mostly to large and capital-intensive firms, allowing a small segment of
elite firms to grow faster. Such a scenario could be due to an over-concentrated
banking structure dominated by large banks, which in turn lend largely only
to large firms (Lin et al. 2011); or due to the financial restraint policy of the
government to increase the franchise value of banks to prevent their oppor-
tunistic behavior (Hellmann et al. 1996); or due to a political coalition between
political and banking insiders that restricts entry into the banking sector, result-
ing in a bank sector dominated by large banks, which lend largely to affiliated
inside firms that tend to be large and capital-intensive (Calomiris and Haber
2011).
VI Conclusions
What explains the vast variations across countries in financial structure? Does
financial structure have any impact on economic development? There has been
some evolution on these questions. The traditional theoretical views tend to
argue that financial structure does not matter, and when it matters, either banks
or markets are superior. The traditional empirical consensus tends to imply that
it is financial depth, not financial structure, that determines aggregate economic
performance.
Several researchers have recently argued that financial services are endoge-
nous to industrial structure which in turn depends on a country’s relative
endowment structure, and optimal financial structure should be specific to the
particular development stage. And some recent findings seem to support this
view. In particular, while both banks and stock markets become larger and more
active over time, stock markets become relatively more important. Moreover, as
economies become richer, the sensitivity of economic development to changes
in bank development decreases, while the sensitivity of economic development
to changes in stock market development increases; thus the relative demand
for the service provided by stock market increases. In addition, deviation of a
country from its optimal financial structure is found to be negatively and sig-
nificantly related to a lower income level. Firm-level evidence also shows that
bank development has particularly strong effects in relatively poor countries,
especially in those industries heavily relying on external finance, while stock
market development has particularly strong effects in relatively rich countries.
Banks (relative to stock markets) are also found to be relatively better at reducing
poverty in developing countries, especially in institutionally weak countries. On
the other hand, there is no evidence that small firms in developing countries
benefit more from bank development, due to the deviation of actual financial
structure from the appropriate one.
The findings have important implications. First, appropriate financial struc-
ture changes – becoming more market-oriented – as economies develop. Second,
new evidence suggests that indeed different financial structures may be better at
promoting economic activity at different stages of a country’s economic devel-
opment. These findings support financial structure as an independent financial
policy consideration. And, if the appropriate mixture changes as an economy
develops then this suggests the desirability of appropriately adjusting finan-
cial policies and institutions as countries develop. Third, politics, legal origins,
and beliefs of government leaders may cause the actual financial structure in a
country to deviate from its appropriate structure, resulting in some efficiency
and welfare losses to the economy. Improving the understanding of what the
optimal is and the efficiency/welfare loss due to the deviation from the optimal,
therefore, may mitigate the impact of political and other belief-related factors
in the determination of a country’s actual financial structure.
Notes
∗ This chapter has benefitted from discussions with Wendy Carlin, Robert Cull, and
Asli Demirguc-Kunt, and the discussions at the World Bank Conference on Financial
Structure held in Washington, DC in June 17, 2011. Masahiko Aoki provided helpful
comments. The views expressed do not implicate the World Bank and the countries
that it represents.
1. Financial structure does evolve. Japan, for instance, can no longer be classified as a
bank-oriented economy (Aoki 2010).
2. See also Aoki and Patrick (1994), which examines the merits and demerits of the
banking system, and use country studies of Japan, Korea, India, and Mexico to shed
light on their arguments.
3. However, Hellmann et al. (1996) suggest that banks should not be too small and
should have some market power in poor developing countries. Banks face agency costs
in the tendency toward gambling and looting when taxpayers and the government
tend to bail out banks in financial distress so that banks benefit from the upside
but do not hurt enough from the downside of their investments. To prevent this,
the government can increase banks’ franchise value by engage in financial restraint,
that is, limiting bank entry and setting deposit rate ceilings, as is often observed in
many East Asian countries, so that banks have incentives to refrain from short-term
gambling and looting to preserve banks’ franchise values. This consideration suggests
that while banks should be small in poor developing countries to best serve small
firms, they should not be so small that they don’t have proper long-term incentives.
4. Technological innovation risks are those related to successfully developing new prod-
ucts, while product innovation risks concern those related to successfully getting the
new product accepted by the market.
5. There would also numerous small banks offering services to small labor-intensive
firms in the non-tradable sectors.
6. Relatedly, Rajan and Zingales (2001) and Claessen and Laeven (2003) suggest that, as
credit markets and accounting standards develop, equity-financed industries tend to
employ less fixed assets, and use more intangible assets due to their increasing ease
to be financed. They do not have to distort asset holding toward fixed capital just for
the sake of being easily collaterizable.
7. Similarly, Cetorelli and Gambera (2001) find that industries relying on external
finance grow faster under a concentrated banking system.
8. It is important to point out that there are important empirical works, often case-
study-based, that offer suggestive evidences that financial structure matters. See, for
instance, Goldsmith (1969), Aoki and Patrick (1994), Aoki and Kim (1995), Allen and
Gale (2000), among others.
9. The next four paragraphs draw heavily from Lin et al. (2011). See also references
therein.
10. See Lin (2009) for detailed discussion of development strategy and its impact on the
development of financial institutions.
11. An alternative explanation is provided by Hellmann et al. (1996) as noted in
footnote 4.
References
Allen, F. and D. Gale (1999) ‘Diversity of Opinion and Financing of New Technologies’,
Journal of Financial Intermediation, vol. 8, nos 1–2, pp. 68–89.
Allen, F. and D. Gale (2000) Comparing Financial Systems (Cambridge, MA: MIT Press).
Aoki, M. and H. Patrick (1994) The Japanese Main Bank System (Oxford: Oxford University
Press).
Aoki, M. and H. Kim (1995) Corporate Governance in Transitional Economies (Washington,
DC: World Bank).
Aoki, M., H. Kim and M. Okuno-Fujiwara (1996) The Role of Government in East Asian
Development (Oxford: Oxford University Press).
Beck, T., A. Demirguc-Kunt, R. Levine and V. Maksimovic (2001) ‘Financial Structure and
Economic Development: Firms, Industry, and Country Evidence’, in A. Demirguc-Kunt
and R. Levine (eds), Financial Structure and Economic Growth: A Cross-country Comparison
of Banks, Markets, and Development (Cambridge, MA: MIT Press).
Bhide, A. (1993) ‘The Hidden Costs of Stock Market Liquidity’, Journal of Financial
Boyd, J. H. and B. D. Smith (1998) ‘The Evolution of Debt and Equity Markets in Economic
Development’, Economic Theory, vol. 12, no. 3, pp. 519–560.
Calomiris, C. and S. Haber (2011) ‘Fragile Banks, Durable Bargains: Why Banking Is All
about Politics and Always Had Been’, mimeo, Stanford University.
Carlin, W. and C. Mayer (2003) ‘Finance, Investment, and Growth’, Journal of Financial
Cetorelli, N. and M. Gambera (2001) ‘Banking Market Structure, Financial Dependence
and Growth: International Evidence from Industry Data’, Journal of Finance, vol. 56,
no. 2, pp. 617–648.
Claessen, S. and L. Laeven (2003) ‘Financial Development, Property Rights and Growth’,
Journal of Finance, vol. 58, no. 6, pp. 2401–2436.
Cull, R. and L. C. Xu (2011) ‘Firm Growth and Finance: Are Some Financial Institutions
Better Suited to Early Stages of Development than Others?’, mimeo, World Bank.
Cull, R., L. E. Davis, N. R. Lamoreaux and J. Rosenthal (2006) ‘Historical Financing of
Small- and Medium-sized Enterprises’, Journal of Banking and Finance, vol. 30, no. 11,
pp. 3017–3042.
Demirguc-Kunt, A., E. Feyen and R. Levine (2011) ‘Optimal Financial Structures and
Development: The Evolving Importance of Banks and Markets’, mimeo, World Bank.
Demirguc-Kunt, A. and R. Levine (2001) ‘Bank-based and Market-based Financial Systems:
Cross-country Comparisons’, in A. Demirguc-Kunt and R. Levine (eds), Financial Struc-
ture and Economic Growth: A Cross-country Comparison of Banks, Markets, and Development
Gerschenkron, A. (1962) Economic Backwardness in Historical Perspective, a Book of Essays
(Cambridge, MA: Harvard University Press).
Goldsmith, R. W. (1969) Financial Structure and Development (New Haven: Yale University
Press).
Hellmann, T., K. Murdock and J. Stiglitz (1996) ‘Financial Restraint and the Market Enhan-
ceing View’, in M. Aoki, H. Kim and M. Okuno-Fujiwara (eds), The Role of Government
in East Asian Development (Oxford: Oxford University Press).
Huang, H. and C. Xu (1999) ‘Institutions, Innovations and Growth’, American Economic
Review, vol. 89, no. 2, pp. 438–443.
Kpodar, K. and R. Singh (2011) ‘Does Financial Structure Matter for Poverty? Evidence
from Developing Countries’, mimeo, IMF.
Kremer, M. (1993) ‘The O-Ring Theory of Economic Development’, Quarterly Journal of
La Porta, R., F. Lopez-de-Silanes, A. Shleifer and R. W. Vishny (1998) ‘Law and Finance’.
Journal of Political Economy, vol. 106, no. 6, pp. 1113–1155.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer and R. W. Vishny (2000) ‘Investor Protection
and Corporate Governance’. Journal of Financial Economics, vol. 58, nos 1–2, pp. 3–27.
Levine, R. (2002) ‘Bank-Based or Market-Based Financial Systems: Which Is Better?’, Journal
of Financial Intermediation, vol. 11, no. 4, pp. 1–30.
Lin, J. Y. (2009) Economic Development and Transition: Thought, Strategy, and Viability (New
York: Cambridge University Press).
Lin, J. Y., X. Sun annd Y. Jiang (2011) ‘Toward a Theory of Optimal Financial Structure’,
mimeo, World Bank.
Rajan, R. G. (1992) ‘Insiders and Outsiders: The Choice between Informed and Arm’s
Length Debt’, Journal of Finance, vol. 47, no. 4, pp. 1367–1400.
Rajan, R. G. and L. Zingales (1998) ‘Financial Dependence and Growth’, American Economic
Review, vol. 88, no. 3, pp. 559–586.
Rajan, R. G. and L. Zingales (1999) ‘Financial System, Industrial Structure, and Growth’,
mimeo, School of Business, University of Chicago.
Rajan, R., and L. Zingales (2001) ‘Financial Systems, Industrial Structure, and Growth’,
Oxford Review of Economic Policy, vol. 17, no. 4, pp. 467–482.
Stiglitz, J. E. (1985) ‘Information and Economic Analysis: A Perspective’, Economic Journal,
95, Supplement: Conference Papers, pp. 21–41.
Stulz, R. (2000) ’Does Financial Structure Matter for Economic Growth? A Corporate
Finance Perspective’, in A. Demirguc-Kunt and R. Levine (eds), Financial Structure
and Economic Growth: A Cross-country Comparison of Banks, Markets, and Development
Xu, L. C. (2011) ‘The Effects of Business Environments on Development: A Survey of New
Firm-level Evidence’, World Bank Research Observer, vol. 26, no. 2, pp. 310–340.
13
The Race to Zero
Andrew G. Haldane*
Bank of England, UK
1 Introduction
Stock prices can go down as well as up. Never in financial history has this adage
been more apt than on 6 May 2010. Then, the so-called ‘Flash Crash’ sent shock-
waves through global equity markets. The Dow Jones experienced its largest ever
intraday point fall, losing $1 trillion of market value in the space of half an hour.
History is full of such fat-tailed falls in stocks. Was this just another to add to
the list, perhaps compressed into a smaller time window?
No. This one was different. For a time, equity prices of some of the world’s
biggest companies were in freefall. They appeared to be in a race to zero. Peak
to trough, Accenture shares fell by over 99 per cent, from $40 to $0.01. At
precisely the same time, shares in Sotheby’s rose three thousand-fold, from $34
to $99,999.99. These tails were not just fatter and faster. They wagged up as well
as down.
The Flash Crash left market participants, regulators and academics agog. More
than two years on, they remain agog. There has been no shortage of potential
explanations. These are as varied as they are many: from fat fingers to fat tails;
from block trades to blocked lines; from high-speed traders to low-level abuse.
From this mixed bag, only one clear explanation emerges: that there is no clear
explanation. To a first approximation, we remain unsure quite what caused the
Flash Crash or whether it could recur.1
That conclusion sits uneasily on the shoulders. Asset markets rely on accurate
pricing of risk. And financial regulation relies on an accurate reading of markets.
Whether trading assets or regulating exchanges, ignorance is rarely bliss. It is
this uncertainty, rather than the Flash Crash itself, which makes this an issue of
potential systemic importance.
In many respects, this uncertainty should come as no surprise. Driven by a
potent cocktail of technology and regulation, trading in financial markets has
evolved dramatically during the course of this century. Platforms for trading
245
246 A. G. Haldane
equities have proliferated and fragmented. And the speed limit for trading has
gone through the roof. Technologists now believe the sky is the limit.
This rapidly changing topology of trading raises some big questions for risk
management. There are good reasons, theoretically and empirically, to believe
that while this evolution in trading may have brought benefits such as a
reduction in transaction costs, it may also have increased abnormalities in the
distribution of risk and return in the financial system. Such abnormalities hall-
marked the Flash Crash. This chapter considers some of the evidence on these
abnormalities and their impact on systemic risk.
Regulation has thin-sliced trading. And technology has thin-sliced time.
Among traders, as among stocks on 6 May, there is a race to zero. Yet it is
unclear that this race will have a winner. If it raises systemic risk, it is possible
capital markets could be the loser. To avoid that, a redesign of mechanisms for
securing capital market stability may be needed.
2 The topology of trading
During the course of this century, financial market trading has undergone a
transformation. This has been driven in part by technology and in part by
regulation. The key words are structure and speed. Both the structure of the
markets in which participants operate and the behavior of those participants
has undergone a phase shift. That highly adaptive topology of trading has made
understanding markets a more hazardous science than a decade ago.
Figure 13.1 plots equity market capitalization relative to nominal GDP in
the United States, Europe, and Asia through this century. On the face of it,
it paints a rather unexciting picture. Equity market values relative to GDP in
the US are roughly where they started the century. In Asia there is evidence of
some deepening of equity markets relative to the whole economy but it is pretty
modest.
Measures of equity market capitalization to GDP have often been used as prox-
ies for the contribution of financial development to economic growth (Arestis
and Demetriades 1997). These effects are typically found to be significant. By
that metric, the contribution of equity markets to economic growth in the US,
Europe, and Asia has been static, at best mildly positive, during the course of
this century.
Yet that picture of apparent stasis in equity markets conceals a maelstrom of
activity beneath the surface. To see this, Figure 13.2 plots stock market turnover
in the US, Europe, and Asia over the same period. It shows a dramatic rise,
especially in the world’s most mature equity market, the United States. Equity
market turnover in the US has risen nearly fourfold in the space of a decade.
Put differently, at the end of the Second World War, the average US share was
held by the average investor for around four years. By the start of this century,
The Race to Zero 247
North America
Ratio
Europe and Russia 1.8
Asia 1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
2001 02 03 04 05 06 07 08 09
Figure 13.1 Market capitalization by region, relative to GDP
Source: World Federation of Exchanges.
North America Ratio

Europe and Russia 7.0
Asia
6.0
5.0
4.0
3.0
2.0
1.0
0.0
2001 02 03 04 05 06 07 08 09
Figure 13.2 Equity market turnover by region
Source: World Federation of Exchanges.
that had fallen to around eight months. And by 2008, it had fallen to around
two months.
What explains this story? Regulation is part of it. Two important, and almost
simultaneous, regulatory developments on either side of the Atlantic changed
248 A. G. Haldane
fundamentally the trading landscape: in the US, Regulation NMS (National Mar-
ket System) in 2005; and in Europe, MiFID (Markets in Financial Instruments
Directive) in 2004. Though different in detail, these regulatory initiatives had
similar objectives: to boost competition and choice in financial market trading
by attracting new entrants.
Central exchanges for the trading of securities evolved from the coffee houses
of Amsterdam, London, New York, and Paris in the 17th century. From those
foundations emerged the physical exchanges which dominated the period from
the 18th right through to the 20th centuries. Central trading exchanges main-
tained their near-monopoly status for 300 years. In the space of a few years that
monopoly has been decisively broken.
A diverse and distributed patchwork of exchanges and multilateral trading
platforms has emerged in its place. These offer investors a range of execution
characteristics, such as speed, cost, and transparency, typically electronically.
Equity market trading structures have fragmented. This has gone furthest in the
US, where trading is now split across more than half a dozen exchanges, mul-
tilateral trading platforms and ‘dark pools’ of anonymous trading (Figures 13.3
and 13.4). Having accounted for around 80 per cent of trading volume in NYSE-
listed securities in 2005, the trading share of the NYSE had fallen to around 24
per cent by February 2011.
A similar pattern is evident across Europe. In the UK, the market share of the
London Stock Exchange has fallen from two-thirds in June 2008 to less than
one third today (Figures 13.5 and 13.6). The same pattern is found in Germany
and France. In Asia, there is as yet less fragmentation.
16%
0%
0%
80%
2%
2%
NYSE NYSE/Arca
NASDAQ BATS
DirectEdge Other inc. dark and OTC
Figure 13.3 NYSE-listed trading volume shares, January 2005
Source: Fidessa.
24%
33%
11%
9%
15%
9%
NYSE NYSE/Arca
NASDAQ BATS
DirectEdge Other inc. dark and OTC
Figure 13.4 NYSE-listed trading volume shares, February 2011
Source: Fidessa.
25%
66%
0.3%
0.1%
9%
LSE Chi-X Dark SI OTC
Figure 13.5 FTSE 100 trading volume shares, June 2008
Source: Fidessa.
Accompanying this structural evolution in trading has been a technologi-

cal revolution. Electronic trading is not new. The first electronic exchange
(NASDAQ) is already over forty years old. But advances in computing power
have shifted decisively the frontier of electronic, and in particular algorithmic,
trading over the past few years. That frontier is defined by speed.
250 A. G. Haldane
29%
45%
13%
4%
3%
2% 0% 4%
0%
LSE Chi-X Bats Europe
Turquoise Nyse Arca Equiduct
Dark SI OTC
Figure 13.6 FTSE 100 trading volume shares, June 2011
Source: Fidessa.
The average speed of order execution on the US NYSE has fallen from around
20 seconds a decade ago to around one second today. These days, the lexicon
of financial markets is dominated by talk of High-Frequency Trading (HFT). It
is not just talk. As recently as 2005, HFT accounted for less than a fifth of US
equity market turnover by volume. Today, it accounts for between two-thirds
and three-quarters.
The picture is similar, if less dramatic, in Europe. Since 2005, HFT has risen
from a tiny share to represent over 35 per cent of the equity market. In Asia
and in emerging markets, it is growing fast from a lower base. What is true
across countries is also true across markets. HFT is assuming an ever-increasing
role in debt and foreign exchange markets. In some futures markets, it already
accounts for almost half of turnover. In the space of a few years, HFT has risen
from relative obscurity to absolute hegemony, at least in some markets.
HFT itself is far from monolithic, comprising a range of strategies.2 Some
involve high-speed liquidity provision, which is akin to market-making. Others
involve statistical arbitrage, using trading algorithms to detect and exploit pric-
ing anomalies between stocks or markets. Because these anomalies tend to be
eliminated quickly, HFT algorithms have to be highly adaptive, not least to keep
pace with the evolution of new algorithms. The half-life of an HFT algorithm
can often be measured in weeks.
One variant of these arbitrage strategies exploits pricing differences between
common securities quoted on competing trading platforms. For that reason,
HFT firms tend to have their tentacles spread across multiple trading venues,
arbitraging tiny differences in price (Figure 13.7). These strategies have grown
up as a direct response to the fragmentation of trading infrastructures. In other
words, HFT is at least in part the (possibly unplanned) progeny of regulators
pursuing competitive ends.
over 20 liquidity venues
up to 20 liquidity venues
2 liquidity venues
1 liquidity venue
0 5 10 15 20 25 30 35
Per cent
Figure 13.7 Access to multiple trading venues by European HFTs
Notes: Survey amongst 171 high frequency traders.
Source: Automated Trader.
The ascent of HFT goes a long way towards explaining the rise in equity market
turnover in the major equity markets and in particular the rise in number, and
fall in the average size, of trades executed. Put differently, the trading behavior
of HFT has contributed to the downward fall in the average duration of stock
holdings. HFT holding periods lie in a narrow time range. The upper bound
is perhaps around one day. The lower bound is a perpetual downward motion
machine, as computing capacity compresses the timeline for trading.
A decade ago, execution times on some electronic trading platforms dipped
decisively below the one second barrier. As recently as a few years ago, trade
execution times reached ’blink speed’ – as fast as the blink of an eye. At the time
that seemed eye-watering, at around 300–400 milliseconds or less than a third
of a second. But more recently the speed limit has shifted from milliseconds
to microseconds – millionths of a second. Several trading platforms now offer
trade execution measured in micro-seconds (Table 13.1).
As of today, the lower limit for trade execution appears to be around 10
microseconds. This means it would in principle be possible to execute around
40,000 back-to-back trades in the blink of an eye. If supermarkets ran HFT pro-
grammes, the average household could complete its shopping for a lifetime in
under a second. Imagine.
It is clear from these trends that trading technologists are involved in an arms
race. And it is far from over. The new trading frontier is nanoseconds – billionths
of a second. And the twinkle in technologists’ (unblinking) eye is picoseconds –
trillionths of a second. HFT firms talk of a ‘race to zero’. This is the promised
252 A. G. Haldane
Table 13.1 Round-trip latencies for selected exchanges, April 2010
Exchange Average round-trip latency (microseconds)
Nasdaq OMX INET 250

Bats Europe 270
Chi-X Europe 400
NYSE Arca 900
LSE Tradelect 2000
TSE Assowhead 5000
Source: http://mondivisione.com via AFM 2010.
land of zero ‘latency’ where trading converges on its natural (Planck’s) limit, the
speed of light.3
The race to zero seems like a million miles from the European coffee shop
conversations of the 17th century and the noisy trading floors of the 18th, 19th,
and 20th centuries. The dawn of electronic trading coincided with the dusk for
floor trading. Physical proximity lost its allure. As soon as computers processed
faster than traders talked, the time was up for human interactions on physical
exchanges. Trading became virtual, distance a dinosaur.
Or so it seemed. Latterly, however, HFT is turning that logic on its head. The
race to zero has encouraged traders to eke out the last picosecond. And one way
to do that is by limiting physical distance. The shorter the cable to the matching
engine of the trading exchange, the faster the trade. Every 100 miles might add a
millisecond to estimated execution times. For HFT, that is the difference between
the tortoise and the hare.
The recognition of that has led to a phenomenon known as ‘co-location’.
HFT firms have begun to relocate their servers as close as physically possible to
the trade-matching engine. That allows them to eke a pico over their (non co-
located) competitors. For a price, a number of exchanges now offer co-located
services, with a perimeter strictly enforced, including the NYSE, Euronext, and
the London Stock Exchange.
This has added a new dimension to the ‘adverse selection’ problem in
economics – of uninformed traders suffering at the hands of the informed.
Being informed used to mean being smarter than the average bear about the
path of future fundamentals – profits, interest rates, order flow, and the like.
Adverse selection risk meant someone having a better informed view on these
fundamentals.
Adverse selection risk today has taken on a different shape. In a high-speed, co-
located world, being informed means seeing and acting on market prices sooner
than competitors. Today, it pays to be faster than the average bear, not smarter.
To be uninformed is to be slow. These uninformed traders face a fundamental
uncertainty: they may not be able to observe the market price at which their
trades will be executed. This is driving through the rear-view mirror, stock-
picking based on yesterday’s prices.
Co-location speeds up the clock. But it has also had the effect of turning
it back. Location matters once more. The race to zero has become a physical,
as well as a virtual, one. Distance matters more than ever. The dinosaur has
been resurrected, this time in high definition. In some ways, it is the ultimate
contradiction.
In sum, through this century changes in the structure of trading, and in the
behavior of traders, have gone hand in hand. Liberalization and innovation
have delivered fragmentation of structure and transformation of speed. Both
structure and speed have experienced a high-velocity revolution. So what impact
has this race to zero had on market dynamics?
3 From microstructure to macrostructure
This is difficult detective work. But there are theoretical clues and empirical
fingerprints. The theoretical clues come from a literature that flourished after
the stock market crash of 1987 – the so-called market microstructure literature
(for example, O’Hara 2004). This departs from the notion of frictionless trading
and information-efficient prices. In its place it introduces frictions in the price-
setting process, which arise from characteristics of market participants (such
as their trading speed) and of the trading infrastructure (such as its degree of
fragmentation).
Frictions in pricing arise from the process of matching buyers and sellers. Here,
the role of market-makers is key. The market-maker faces two types of problem.
One is an inventory-management problem – how much stock to hold and at
what price to buy and sell. The market-maker earns a bid-ask spread in return
for solving this problem since they bear the risk that their inventory loses value
(Stoll 1978).
Market-makers face a second, information-management problem. This arises
from the possibility of trading with someone better informed about true prices
than themselves – an adverse selection risk. Again, the market-maker earns a
bid-ask spread to protect against this informational risk (Glosten and Milgrom
1985).
The bid-ask spread, then, is the market-makers’ insurance premium. It pro-
vides protection against risks from a depreciating or mis-priced inventory. As
such, it also proxies the ‘liquidity’ of the market – that is, its ability to absorb,
buy and sell orders and execute them without an impact on price. A wider
bid-ask spread implies greater risk in the sense of the market’s ability to absorb
volume without affecting prices.
254 A. G. Haldane
This basic framework can be used to assess the impact of the changing trad-
ing topology on systemic risk, moving from analyzing market microstructure to
market macrostructure. Take the effects of fragmentation. That heightens compe-
tition among trading platforms, offering differing execution options and wider
access to participants. This would tend to attract liquidity providers, including
cross-market arbitraging HFT. As new liquidity-providers enter the market, exe-
cution certainty and price efficiency might be expected to improve. Inventory
and information risk would thereby fall and, with it, bid-ask spreads.
Some of the early empirical fingerprints suggest just such a pattern. For exam-
ple, Brogaard (2010) analyzes the effects of HFT on 26 NASDAQ-listed stocks.
HFT is estimated to have reduced the price impact of a 100-share trade by $0.022.
For a 1000-share trade, the price impact is reduced by $0.083. In other words,
HFT boosts the market’s absorptive capacity. Consistent with that, Hendershott
et al. (2010) and Hasbrouck and Saar (2011) find evidence of algorithmic trading
and HFT having narrowed bid-ask spreads.
Figure 13.8 plots a measure of bid-ask spreads on UK equities over the past
decade, normalizing them by a measure of market volatility to strip out volatility
spikes. It confirms the pattern from earlier studies. Bid-ask spreads have fallen
by an order of magnitude since 2004, from around 0.023 to 0.002 percentage
Percentage points
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0
99 00 01 02 03 04 05 06 07 08 09 10 11
Figure 13.8 Median end-of-day bid-ask spread for largest 20 stocks in FTSE 100 as a pro-
portion of realized volatility(a)
Source: Thomson Reuters Datastream.
(a) 22-day average. Largest stocks by market capitalization. Selected stocks updated annually.
points. On this metric, market liquidity and efficiency appear to have improved.
HFT has greased the wheels of modern finance.
But bid-ask spreads can sometimes conceal as much as they reveal. For exam-
ple, by normalizing on volatility, Figure 13.8 airbrushes out what might be most
interesting: normalizing volatility might normalize abnormality. It risks falling
foul of what sociologists call ‘normalization of deviance’ – that is, ignoring small
changes which might later culminate in an extreme event.4
So is there any evidence of increasing abnormality in market prices over the
past few years? Measures of market volatility and correlation are two plausible
metrics (see Brogaard 2010 and Zhang 2010). Figure 13.9 plots the volatility of,
and correlation between, components of the S&P 500 since 1990. In general,
the relationship between volatility and correlation is positive. Higher volatility
increases the degree of co-movement between stocks.
Now consider how this volatility/correlation nexus has changed. This can be
seen from the difference between the mass of grey dots (covering the period 1990
to 2004) and black dots (covering the period 2005 to 2010) in Figure 13.9. Two
things have happened since 2005, coincident with the emergence of trading
platform fragmentation and HFT.
First, both volatility and correlation have been somewhat higher. Volatil-
ity is around 10 percentage points higher than in the earlier sample, while
correlation is around 8 percentage points higher. Second, the slope of the
volatility/correlation curve is steeper. Any rise in volatility now has a more pro-
nounced cross-market effect than in the past. Another way of making the same
point is to plot measures of excess correlation’ – measured market correlation
in excess of volatility. This is currently at historic highs (Figure 13.10).
S&P 500 1Y Correlation

70%
1990–2004
60%
2005–2010
50%
40%
30%
20%
10%
0%
0% 10% 20% 30% 40% 50%
S&P 500 1Y volatility
Figure 13.9 Volatility and correlation of S&P 500
Source: JP Morgan and Bank of England calculations.

256 A. G. Haldane
Per cent Excess correlation (LHS) Per cent

20 60
1yr correlation (LHS)
15
1yr volatility (RHS) 50
10
40
5
+
0 30
-
5
20
10
10
15
20 0
90 92 94 96 98 00 02 04 06 08
Figure 13.10 Excess volatility
Source: JPMorgan Chase.
Taken together, this evidence points towards market volatility being both
higher and propagating further than in the past. Intraday evidence on volatili-
ties and correlations appears to tell a broadly similar tale. Overnight and intraday
correlations have risen in tandem (Lehalle et al. 2010). And intra-day volatility
has risen most in those markets open to HFT (Lehalle and Burgot 2010).
Coincidence does not of course imply causality. Factors other than HFT may
explain these patterns. Event studies provide one way of untangling this knit-
ting. Recent evidence from the Flash Crash pinpoints the particular role played
by HFT using transaction-level data. The official report on the Flash Crash, while
not blaming HFT firms for starting the cascade, assigns them an important role
in propagating it. For example, many HFT firms significantly scaled back liquid-
ity and overall HFT firms were net sellers of stock (CFTC-SEC (2010): Kirilenko
et al. 2011).
Taken together, this evidence suggests something important. Far from solving
the liquidity problem in situations of stress, HFT firms appear to have added to
it. And far from mitigating market stress, HFT appears to have amplified it. HFT
liquidity, evident in sharply lower peacetime bid-ask spreads, may be illusory. In
wartime, it disappears. This disappearing act, and the resulting liquidity void, is
widely believed to have amplified the price discontinuities evident during the
Flash Crash (see, for example, Jarrow and Protter 2011; Cvitanic and Kirilenko
2010). HFT liquidity proved fickle under stress, as flood turned to drought.
In some respects, this may sound like old news. For example, an evapo-
ration of liquidity, amplified by algorithmic trading, lay at the heart of the
1987 stock market crash. And it is also well-known that stock prices exhibit
non-normalities, with the distribution of asset price changes fatter-tailed and
more persistent than implied by the efficient markets hypothesis at frequencies

of years and months, perhaps weeks and days (see, for example, Gopikrishnan
et al. 1999; Bouchaud et al. 2009). But these abnormalities were thought to dis-
appear at higher frequencies, such as hours and minutes. Over shorter intervals,
efficient market pricing restored itself.
Recent studies point, however, to a changing pattern. Non-normal patterns
in prices have begun to appear at much higher frequencies. A recent study by
Smith (2010) suggests that, since around 2005, stock price returns have begun
to exhibit fat-tailed persistence at 15-minute intervals. Given the timing, these
non-normalities are attributed to the role of HFT in financial markets.
The measure of stock price abnormality used by Smith is the so-called ‘Hurst’
coefficient (Blackledge 2008). The Hurst coefficient is named after the English
civil engineer H. E. Hurst. It was constructed by plotting data on the irregular
flooding patterns of the Nile delta over the period 622–1469 AD. Hurst found
that flooding exhibited a persistent pattern. Large floods were not only frequent,
but came in clumps. They had a long memory.
The Hurst coefficient summarizes this behavior in a single number. For exam-
ple, a measured Hurst equal to 0.5 is consistent with the random walk model
familiar from efficient markets theory. A Hurst coefficient above 0.5 implies
fatter tails and longer memories. In his study, Smith finds that the Hurst coef-
ficient among a selection of stocks has risen steadily above 0.5 since 2005. In
other words, the advent of HFT has seen price dynamics mirror the fat-tailed
persistence of the Nile flood plains.
To illustrate, Figure 13.11 plots the path of three simulated price series with
Hurst coefficients of 0.5, 0.7 and 0.9. A higher Hurst coefficient radically alters
H = 0.5
H = 0.7 40
H = 0.9
20
+
0
–
20
40
60
80
100
Time
Figure 13.11 Simulated price series under different Hurst coefficients
Source: Bank of England.

258 A. G. Haldane
the probability of sharp dislocations in prices. It also prolongs these disloca-

tions. Prices become de-anchored and drift; their variance grows over time
and is unbounded. If this long-memory property of prices is emerging at ever-
higher frequencies, it might provide an important clue to how HFT affects
systemic risk.
4 A sketch model of market macrostructure
To see that, consider a sketch model of market-making. This builds on an ana-

lytical insight which is already more than 40 years old. It owes its existence to
the late Benoit Mandelbrot, the French-American mathematician and architect
of fractal geometry. Mandelbrot found that a great many real-world topologies
exhibited a fractal pattern. By this he meant that the basic pattern repeated
itself, whatever the scale at which it was observed. They were ‘self-similar’. Self-
similarity appears to be present throughout the physical world, from coastlines
to cauliflowers, from snowflakes to lightning bolts, from mountain ranges to
river deltas (see Peters 1994).
One of Mandelbrot’s earliest applications of fractal geometry was to stock
prices. In a 1967 paper, he argued that stock prices could best be understood
by distinguishing between two measuring rods: clock time and volume time
(Mandelbrot and Taylor 1967; see also Clark 1973). While empirical studies
typically used the first measuring rod (days, hours, seconds, milliseconds), stock
prices were better understood by using the second.
Mandelbrot’s explanation was relatively simple. If trading cannot occur
within a given time window, price movements can only reflect random pieces of
news – economic, financial, political. So, consistent with efficient market the-
ory, price changes would be drawn from a normal distribution with a fat middle
and thin tails when measured in clock time. They were a random walk.
But as soon as trading is possible within a period, this game changes. Strategic,
interactive behavior among participants enters the equation. Volumes come and
go. Traders enter and exit. Algorithms die or adapt. Behavior within that time
interval may then no longer be random noise. Rather, trading volumes will
exhibit persistence and fat tails. This will then be mirrored in prices.5 So when
measured in clock time, price changes will have thinner middles and fatter tails,
just like a cauliflower, a coastline, or a cosmos.6
Subsequent studies have shown that this clock time/volume time distinction
helps explain equity price dynamics, especially at times of market stress. For
example, Easley et al. (2011) show that the distribution of price changes during
the Flash Crash was highly non-normal in clock time, with fat tails and persis-
tence. But in volume time, normal service – indeed, near-normality – resumed.
This fractal lens can be used to explain why market liquidity can evaporate
in situations of market stress, amplifying small events across time, assets, and
markets. Fractal geometry tells us that what might start off as a snowflake has
the potential to snowball.
4.1 Behavior of high frequency traders

HFT has had three key effects on markets. First, it has meant ever-larger volumes
of trading have been compressed into ever-smaller chunks of time. Second, it has
meant strategic behavior among traders is occurring at ever-higher frequencies.
Third, it is not just that the speed of strategic interaction has changed but also
its nature. Yesterday, interaction was human-to-human. Today, it is machine-to-
machine, algorithm-to-algorithm. For algorithms with the lifespan of a ladybird,
this makes for rapid evolutionary adaptation.
Cramming ever-larger volumes of strategic, adaptive trading into ever-smaller
time intervals would, following Mandelbrot, tend to increase abnormalities in
prices when measured in clock time. It will make for fatter, more persistent
tails at ever-higher frequencies. That is what we appear, increasingly, to find in
financial market prices in practise, whether in volatility and correlation or in
fat tails and persistence.
This change in price dynamics will in turn influence market-making behavior.
Consider the problem facing an HFT market-maker. They face inventory risk
from market fluctuations and information risk from adverse selection. Pricing
these risks means forming a guess about the future path of prices. The greater
the potential range of future prices, the larger the insurance premium they will
demand.
The future price range (maximum – minimum) for a price series at some future
date T , R(T ), can be written generically as:7
R(T ) = kσT T H . (1)
where κ is a constant and σT is the standard deviation of the process up to time

T . H is our old friend the Hurst coefficient, dredged up from the Nile; it can be
thought to summarize the degree of fat-tailedness and persistence in prices.
When the holding period is short (say T = 1), as with HFT, the future price
range is the volatility of the series magnified by the Hurst coefficient. In other
words, the fatter and more persistent the tails (H), the greater the market risk
and the wider the bid-ask spread of the HFT.
This has implications for the dynamics of bid-ask spreads, and hence liquidity,
among HFT firms. During a market crash, the volatility of prices (σ ) is likely to
spike. From equation (1), fractality heightens the risk-sensitivity of HFT bid-
ask spreads to such a volatility event. In other words, liquidity under stress is
likely to prove less resilient. This is because one extreme event, one flood or
drought on the Nile, is more likely to be followed by a second, a third and a
fourth. Reorganizing that greater risk, market makers’ insurance premium will
rise accordingly.
260 A. G. Haldane
This is the HFT inventory problem. But the information problem for HFT
market-makers in situations of stress is in many ways even more acute. Price
dynamics are the fruits of trader interaction or, more accurately, algorithmic
interaction. These interactions will be close to impossible for an individual
trader to observe or understand. This algorithmic risk is not new. In 2003, a
US trading firm became insolvent in 16 seconds when an employee inadver-
tently turned an algorithm on. It took the company 47 minutes to realise it had
gone bust (Clark 2010).
Since then, things have stepped up several gears. For a 14-second period dur-
ing the Flash Crash, algorithmic interactions caused 27,000 contracts of the
S&P 500 E-mini futures contracts to change hands. Yet, in net terms, only 200
contracts were purchased. HFT algorithms were automatically offloading con-
tracts in a frenetic, and in net terms fruitless, game of pass-the-parcel. The result
was a magnification of the fat tail in stock prices due to fire-sale forced machine
selling (CFTC-SEC 2010).
These algorithmic interactions, and the uncertainty they create, will mag-
nify the effect on spreads of a market event. Pricing becomes near-impossible
and with it the making of markets. During the Flash Crash, Accenture shares
traded at 1 cent, and Sotheby’s at $99,999.99, because these were the lowest
and highest quotes admissible by HFT market-makers consistent with fulfilling
their obligations. Bid-ask spreads did not just widen, they ballooned. Liquidity
entered a void. That trades were executed at these ‘stub quotes’ demonstrated
algorithms were running on autopilot with liquidity spent. Prices were not just
information inefficient; they were dislocated to the point where they had no
information content whatsoever.
4.2 Behavior of low frequency traders

If the story stopped here the ending would be an unhappy, but perhaps not
critical, one. After all, the Flash Crash was done and dusted within an hour. But
the framework developed suggests these effects need not, in general will not,
be transient. To the contrary, these effects might actually magnify. To see why,
consider now the behavior of low frequency traders (LFT).
They face the same set of risks as the HFT market-maker – inventory and infor-
mation risk. But persistence and fat tails in short-term price movement amplify
these risks. From equation (1), the greater the holding period, T, the greater the
potential dispersion in future prices. Intuitively, with fat-tailed persistence, a
large shock is both more likely and its effects will linger longer. Slower trading
increases the distance prices can travel once de-anchored.
Figure 13.12 illustrates this. The distribution of returns observed at a low
frequency has far greater variance and kurtosis than that at a higher frequency.
For example, the variance of prices faced by a LFT firm (trading at T = 50) is
around nine times greater than for a HFT firm (trading at T = 1), for H = 0.9.8
T=1
T = 20
T = 50
– +
Figure 13.12 Distribution of simulated returns at different time horizons with H = 0.9(a)
Source: Bank of England
(a) Returns at different time horizons have been scaled by the inverse of the square root of time,
1/sqrt(T).
Given these price dynamics, HFT aggravates the market-making problem for
LFT firms by magnifying the market risk they face.9 And it is not just that LFT
firms are slower to execute. In situations of stress, they may not even be able
to see the prices at which they can trade. Co-located traders may have executed
many thousands of trades before LFT firms have executed their own. LFT firms
face intrinsic uncertainty about prices. When volumes and price movements are
large, LFT market-making is driving in the dark, stock-picking with a safety-pin.
During the Flash Crash, many traders suffered just this problem. Message
traffic resulted in delays in disseminating quotes for over 1000 stocks. These
delays lasted for up to 35 seconds. As a result, discrepancies emerged between
the prices of common stocks trading on different exchanges (Figure 13.13). Faced
with such uncertainty, a number of market participants paused or halted trading.
The equilibrating force of long-term investors went missing. Bargain-hunting
shoppers simply had no price list.
The combined effects of these inventory and information problems is to widen
the bid-ask spreads LFT market-makers charge. Greater execution risk and uncer-
tainty calls for a larger insurance premium. This, too, may have an adverse
feedback effect on financial market pricing. That is because it is likely to render
uncompetitive LFT firms relative to HFT firms able to charge tighter spreads.
Market-making will increasingly congregate around HFT firms proffering these
lower spreads.
If the way to make money is to make markets, and the way to market markets
is to make haste, the result is likely to be a race – an arms race to zero latency.
262 A. G. Haldane
Price difference ($)

0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
14:41 14:43 14:45 14:47 14:49
Time
Figure 13.13 Difference in maximum and minimum price of GE shares across different
exchanges on 6 May 2010
Notes: Difference in bid prices across NASDAQ, NYSE, and Pacific Exchange.
Source: NANEX.
Competitive forces will generate incentives to break the speed barrier, as this is
the passport to lower spreads which is in turn the passport to making markets.
This arms race to zero is precisely what has played out in financial markets over
the past few years.
Arms races rarely have a winner. This one may be no exception. In the trading
sphere, there is a risk the individually optimizing actions of participants generate
an outcome for the system which benefits no-one – a latter-day ‘tragedy of the
commons’ (Hardin 1968). How so? Because speed increases the risk of feasts and
famines in market liquidity. HFT contributes to the feast through lower bid-ask
spreads. But they also contribute to the famine if their liquidity provision is
fickle in situations of stress.
In these situations, backstop sources of longer-term liquidity ought to ride to
the rescue. But HFT has also affected this outside option. LFT market-making has
been squeezed out by competitive pressures from HFT. And those LFT market-
makers that remain are at an acute informational disadvantage in situations of
stress. The result is a potentially double liquidity void.
4.3 Behavior across assets and markets

So far the story has been confined to behavior of one stock trading on a single
exchange. But the changing structure of markets and trading mean it is unlikely
that any stock-specific price dislocation will be localized. The new topology of
trading makes contagion more of a potential bogeyman now than in the past.
One source of contagion is between stock prices and derivatives written on

these stocks. HFT activity means that arbitrage opportunities between these
markets are likely to be quickly removed. Or, put differently, price dislocations in
the cash market are likely to be transmitted instantly to futures markets and vice-
versa. That was the story of the Flash Crash, with order imbalances propagated
through the futures market due to pass-the-parcel dynamics before ricocheting
back to affect prices in the cash market (CFTC-SEC 2010).
A second channel of contagion is between different exchanges and trading
platforms. Here, too, HFT has changed the landscape. Because HFT strategies
have emerged to arbitrage differences, price transmission across exchanges and
platforms is near-instantaneous. It also has the potential to be more disruptive.
Liquidity on these exchanges is no longer pooled and centralized. Instead it
is distributed and localized, increasing the potential for illiquidity premia to
emerge in periods of stress.
A third potential contagion channel is across stocks. HFT algorithms tend
to amplify cross-stock correlation in the face of a rise in volatility due to their
greater use of algorithmic trend-following and arbitrage strategies. That is con-
sistent with the evidence in Figure 13.9, with heightened correlation per unit
of volatility relative to the past.
Taken together, these contagion channels suggest that fat-tailed persistence in
individual stocks could quickly be magnified to wider classes of asset, exchange,
and market. The micro would transmute to the macro. This is very much
in the spirit of Mandelbrot’s fractal story. Structures exhibiting self-similarity
magnify micro behavior to the macro level. Micro-level abnormalities manifest
themselves as system-wide instabilities.
In many respects, this is an unsurprising conclusion. The emergence of HFT
in fragmented trading platforms makes for a financial marketplace exhibiting
complexity and contagion. As the race for zero has gathered pace, those two
features are likely to have grown in importance. The trading system has become
increasingly complex and tightly knit. Following the work of Charles Perrow,
these are the two essential ingredients in the recipe for systemic failure (Perrow
1999; see also Harford 2011).
5 Market macrostructure and public policy
The rapidly adapting topology of trading poses a challenging set of questions

for policy-makers, particularly those safeguarding the resilience of financial
markets. How is this changing market macrostructure best understood and
monitored? And how, if at all, can it be modified to bolster systemic resilience?
This is unlikely to be an easy regulatory task. For the first time in financial
history, machines can execute trades far faster than humans can intervene. That
gap is set to widen. In some respects the 2010 Flash Crash and the 1987 stock
264 A. G. Haldane
market crash have common genes – algorithmic amplification of stress. But

they differ in one critical respect. Regulatory intervention could feasibly have
forestalled the 1987 crash. By the time of the Flash Crash, regulators might have
blinked – literally, blinked – and missed their chance.
Nonetheless, the experience of managing large-scale, non-financial systems
suggests risk-reducing regulatory intervention is possible. Drawing on that
experience, two sets of policy tool are worth developing further.
5.1 Mapping market macrostructure

Many securities regulators collect transactional data for the markets they
oversee, typically to help them detect instances of market abuse. But the
transactional data collected internationally is about to be transformed. In
future, a much larger array of over-the-counter transactions will be cleared
through central counterparties (CCPs). And transactions in a number of non-
centrally cleared markets will be recorded and warehoused in so-called trade
repositories.10
This presents a real opportunity. Prospectively over the next decade, securities
regulators internationally will have at their disposal a database which is far richer
in its frequency and scope than anything previously. For the first time ever,
many of the raw ingredients will exist to map price and quantity dynamics
within, and across, markets in close to real time.
To date, relatively little thought has been given to how best to put these data to
work. They will certainly be useful as an archaeological site, allowing a detailed
forensic sifting of the genealogy of historical market events. As with the Flash
Crash, this may enable regulatory detectives to solve the whodunit: who did
what when and, perhaps, why. It will allow DNA sampling of the victim.
An altogether bigger prize would be to put these data to work before the fact,
identifying the next victim pre-autopsy. That could mean using transactions
data to help detect early warnings of systemic fault-lines and stresses. This is a
potentially massive analytical and technical challenge.
The technical challenge is certainly surmountable. Advances in computer
power over the past decade mean that storing and processing huge volumes
of data poses no technological barrier. The answer lies in the clouds – conve-
niently enough, since they too exhibit fractal properties. If we can search and
track the World Wide Web in close to real time, we can certainly do the same
for its financial sub-component.
The analytical challenge is altogether greater. In essence, it is to find summary
measures of billions of transactions data which are informative about impend-
ing market stress. There is serious needle-in-haystack risk with such an endeavor.
And experience in economics and finance of finding robust early warning indi-
cators is mixed. But here again, the market microstructure literature offers some
tantalizing clues.
For example, Easley et al. (2011b) have suggested that measures of ‘order
imbalance’ may provide early warning signs of liquidity voids and price disloca-
tions. Their measure of imbalance follows closely in Mandelbrot’s footprints. It
uses a volume-based metric of the proportion or orders from informed traders.
Any imbalance towards informed traders causes potential liquidity problems
down the line as a result of adverse selection risk. Easley et al. show that their
imbalance measure rose sharply ahead of the Flash Crash, contributing to an
eventual evaporation of liquidity.
A more ambitious approach still would be to develop a system-wide model of
financial market interaction. Cliff (2010) describes the trading infrastructure as
an example of a ‘socio-technical system of systems’. These involve a complex
fusion between technology and human behavior. This interaction increases the
system’s vulnerability to catastrophic failure. He proposes a ‘test rig’ for such
systems, using simulation techniques to unearth potential systemic fault-lines.
These techniques have already been applied to other large socio-technical
systems, such as defense, space and weather systems. In each case, there have
been some successes. The lessons from these exercises seem to be twofold. First,
that although technology may pose a problem, it may also provide the solution.
Second, that even if it ain’t broke, there is a compelling strong case for fixing it.
Not to do so today runs too great a risk of catastrophic failure tomorrow. We do
not need to await a second Flash Crash to establish it was no fluke. To wait is to
normalize deviance.
5.2 Modifying market macrostructure

Regulators in the US and Europe are in the process of reviewing regulatory
rules for trading. In the US, some changes to market rules have already been
implemented, while others are the subject of ongoing research (European Com-
mission 2010; CFTC-SEC 2011). In Europe, a review is underway of MiFID with
an early set of proposals tabled (CFTC-SEC 2011). It is interesting to assess those
proposals using the framework developed here.
i) Market-making guidelines: In principle, a commitment by market-makers to

provide liquidity, whatever the state of the market, would go to the heart
of potential price discontinuity problems. Market-making commitments
would not forestall the arms race. But they would lessen the chances of
liquidity droughts and associated fat tails and persistence in prices. They
would, in effect, lower the impact of H. Perhaps for that reason, there have
been proposals in both the US and Europe for such a set of market-making
commitments (European Commission 2010).
The difficulty appears, first, in specifying these commitments in a precise
enough fashion and, relatedly and just as importantly, enforcing them. In
a sense, even the market-makers offering their stub quotes on 6 May were
266 A. G. Haldane
fulfilling a market-making commitment, albeit a paper-thin one. If hard

law commitments are too difficult to define or enforce, an alternative may
be a set of soft law guidelines. A number of electronic broking platforms,
notably in foreign exchange markets, have codes or rules of conduct – for
example, around price-making and price-taking which control the extent
to which any one firm can steal a technological march on others. If these
codes were extended across trading platforms and assets, perhaps in time
market-making behavior might adapt.
ii) Circuit-breakers: Circuit-breakers already exist on US and European
exchanges. Indeed, circuit-breakers played an important role in stalling
the Flash Crash. In the face of pass-the-parcel algorithmic dynamics, the
Chicago Mercantile Exchange imposed an automatic five-second pause on
trading in its S&P 500 E-mini futures contracts (MacKenzie and Demos
2011). It worked, providing time for human traders to take stock – and,
as importantly, buy stock.
The rationale for such rules is well-understood. They temporarily impose a
minimum execution time on trading, T. By calling a halt to trading, circuit-
breakers provide a means of establishing a level informational playing field
for all traders, long and short, fast and slow. In other words, they help solve
the asymmetric information (risk and uncertainty) problem across traders of
different types. The changing topology of trading, both speed and structure,
has made these informational frictions more acute. So the case for circuit-
breakers is perhaps stronger now than in the past.
Circuit-breakers are also a means of heading off suicidal spiralling, when
algorithms join hands and chase each other over a cliff edge, as during
the Flash Crash. Because they are automatic, circuit-breakers allow time for
human intervention to catch up with, and forestall, machine (summary)
execution. They close the gap between man and machine, if only temporar-
ily. Because HFT has widened this gap, the case for deploying circuit-breakers
is stronger now than in the past.
Historically, circuit-breaking rules have been security or exchange-specific.
But in a world of multiple exchanges, and increasingly correlated securities,
rules of this type may not be restrictive enough. Contagion across securities
and exchanges has become the rule. Recognizing that, the US authorities
have recently revised their circuit-breaking rules with all trading in a speci-
fied set of securities and futures now halted, irrespective of where trading is
taking place (CFTC-SEC 2010).
Although the pattern of trading fragmentation in Europe is similar to
the US, current MiFID proposals do not envisage an automatic cross-market
trading halt. But in a world of location-free, tightly arbitraged trading, cross-
market circuit-breakers might become increasingly important. Indeed, these
rules may potentially need to cross continents, as well as countries and
platforms, if price dislocations are not to be redirected.
iii) Resting rules: Circuit-breakers are an ex-post, state-contingent intervention

rule, specifying a minimum T for a temporary period. A more ambitious
proposal still would be to impose a minimum T, or resting period for trades,
at all times. Minimum resting periods are an ex-ante, non-state contingent
intervention rule. They tackle the arms race at source by imposing a speed
limit on trading. Though mooted in both the US and Europe, they have not
been implemented in either trading jurisdiction.
In some respects, it is easy to see why. From equation (1), imposing a
minimum T will tend to result in a higher average bid-ask spread in all
states. By increasing the per period transaction cost, the imposition of a
minimum resting period would tend to widen bid-ask spreads and damage
market liquidity in peacetime. HFT would be constrained in its offering to
the liquidity feast.
That is of course only one side of the coin. Setting a minimum T would also
tend to reduce the risk of liquidity drought. While raising the average bid-ask
spread, it might also lower its variability at times of stress. Liquidity would
on average be more expensive but also more resilient. So in determining
whether there is a role for minimum resting periods, this trade-off between
market efficiency and stability is key.
In calibrating this trade-off, a judgement would need to be made on the
social value of split-second trading and liquidity provision and whether
this more than counterbalances the greater market uncertainty it poten-
tially engenders. At times, the efficiency of financial markets and their
systemic resilience need to be traded off. This may be one such moment.
Historically, the regulatory skew has been heavily towards the efficiency
objective. Given today’s trading topology, it may be time for that to
change.
6 Conclusion
The Flash Crash was a near miss. It taught us something important, if uncom-
fortable, about our state of knowledge of modern financial markets. Not just that
it was imperfect, but that these imperfections may magnify, sending systemic
shockwaves. Technology allows us to thin-slice time. But thinner technological
slices may make for fatter market tails. Flash Crashes, like car crashes, may be
more severe the greater the velocity.
Physical catastrophes alert us to the costs of ignoring these events, of normal-
izing deviance. There is nothing normal about recent deviations in financial
markets. The race to zero may have contributed to those abnormalities, adding
liquidity during a monsoon and absorbing it during a drought. This fattens tail
risk. Understanding and correcting those tail events is a systemic issue. It may
call for new rules of the road for trading. Grit in the wheels, like grit on the
roads, could help forestall the next crash.
268 A. G. Haldane
Notes
∗ Disclaimer: The views in this chapter are not necessarily those of the Bank of England
or the Financial Policy Committee.
1. For a regulatory perspective see CFTC-SEC (2010), for a market perspective see
Automated Trader (2010) and for an academic perspective see Easley et al. (2011b).
2. Mackenzie (2011) provides a brilliant recent account of the role of HFT, in particular
in the context of the Flash Crash. HFT is a sub-set of a broader class of algorithmic
trading strategies. See also Aldridge (2010).
3. ‘Latency’ refers to the time it takes from sending an order to it being executed.
4. This term has its origins in work by Diane Vaughan on NASA’s decision-making in
the run-up to the space shuttle Challenger disaster in 1986, where repeated oversight
of small problems culminated in a big problem (Vaughan 1996). It has since been
found in a much broader range of phenomena, where small cognitive biases have
had disastrous physical consequences (Cliff 2010; Harford 2011).
5. This finding can be given a variety of behavioral interpretations, including persistence
in gaps in the limit order book (Bouchard et al. 2009). Empirical support for this
hypothesis is found in Easley et al. (2011a).
6. Andrew Lo’s ‘adaptive market hypothesis’ is a more recent manifestation of essentially
the same story.
7. Equation (1) holds asymptotically in expectation for large T and for a broad range of
statistical processes (see Peters 1994 and Qian and Rasheed 2004).
8. Even once an adjustment has been made for the degree of time scaling associated with
a standard Brownian motion.
9. Indeed, with fractal price dynamics the variability of prices will potentially grow
without bound over time.
10. In the United States, the Office of Financial Research, created under the Dodd Frank
Act, is charged with collecting and analyzing data submitted by US firms.
References
Aldridge, I. (2010) High-Frequency Trading: A Practical Guide to Algorithmic Strategies and

Trading Systems (Hoboken: Wiley Trading).
Arestis, P. and P. Demetriades (1997) ‘Financial Development and Economic Growth:
Assessing the Evidence’, The Economic Journal, vol. 107, no. 442, pp. 783–799.
Automated Trader (2010) ‘What Just Happened?’, vol. 18, no. 3, available at:
www.automatedtrader.net/articles/risk/50335/what-just-happened.
Bouchaud, J., J. D. Farmery and F. Lill (2009) ‘How Markets Slowly Digest Changes in
Supply and Demand’, Santa Fe Institute Working Paper, at: www.santafe.edu/∼jdf/
papers/MarketsSlowlyDigest.pdf
Blackledge, J. (2008) ‘Application of the Fractal Market Hypothesis for Macroeconomic
Time Series Analysis’, ISAST Transactions on Electronics and Signal Processing, vol. 1, no. 2,
pp. 89–110.
Brogaard, J. (2010) ‘High Frequency Trading and Its Impact on Market Quality’, mimeo.
CFTC-SEC (2010) ‘Findings Regarding the Market Events of May 6, 2010’, Report of the
staffs of the CFTC and SEC to the joint advisory committee on emerging regulatory
issues, 30 September.
CFTC-SEC Advisory Committee on Emerging Regulatory Issues (2011) ‘Recom-
mendations Regarding Regulation to the Market Events of May 6, 2010’,
available at: www.cftc.gov/ucm/groups/public/@aboutcftc/documents/file/jacreport_

021811.pdf
Clark, C. (2010) ‘Controlling Risk in a Lightning-speed Trading Environment’, Federal
Reserve Bank of Chicago Financial Markets Group Policy Discussion Paper Series PDP
2010–11.
Clark, P. K. (1973) ‘A Subordinated Stochastic Process Model with Finite Variance for
Speculative Prices’, Econometrica, vol. 41, no. 1, pp. 135–155.
Cliff, D. (2010) ‘The Flash Crash of May 6th 2010: Why Technology Failures?’, speech given
at the Royal Society of London, available at: www.cs.bris.ac.uk/home/dc/cliff_wtf_
transcript_final.pdf
Cvitanic, J. and A. Kirilenko (2010) ‘High Frequency Traders and Asset Prices,’ SSRN
eLibrary, at: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1569067
Easley D., M. Lopez de Prado and M. O’Hara (2011a) ‘Flow Toxicity and Volatility in a
High Frequency World’, working paper.
Easley D., M. Lopez de Prado and M. O’Hara (2011b) ‘The Microstructure of the “Flash
Crash”: Flow Toxicity, Liquidity Crashes, and the Probability of Informed Trading’,
Journal of Portfolio Management, vol. 37, no. 2, pp. 118–128.
European Commission (2010) ‘Review of the Markets in Financial Instruments Directive
(MiFID)’, available at: http://ec.europa.eu/internal_market/consultations/docs/2010/
mifid/consultation_paper_en.pdf
Glosten, L. and J. Milgrom (1985) ‘Bid and Ask Transaction Prices in a Specialist Market
with Heterogeneously Informed Traders’, Journal of Financial Economics, vol. 14, no. 1,
pp. 71–100.
Gopikrishnan, P., V. Plerou, L. Amaral, M. Meyer and H. E. Stanley (1999) ‘Scaling of
the Distribution of Fluctuations of Financial Market Indices’, Physical Review E, vol. 60,
pp. 5305–5316.
Hardin, G. (1968) ‘The Tragedy of the Commons’, Science, vol. 162, no. 3859,
pp. 1243–1248.
Harford, T. (2011) Adapt: Why Success Always Starts with Failure (London: Little, Brown).
Hasbrouck, J. and G. Saar (2011) ‘Low-latency Trading’, Johnson School Research Paper
Series No. 35-2010 (New York: Samuel Curtis Johnson Graduate School of Management
at Cornell University).
Hendershott, T., C. Jones and A. Menkveld (2010) ‘Does Algorithmic Trading Improve
Liquidity?’, Journal of Finance, vol. 66, no. 1, pp. 1–33.
Jarrow, R.A. and P. Protter (2011) ‘A Dysfunctional Role of High Frequency Trading in
Electronic Markets’, Johnson School Research Paper Series No. 08-2011 (New York:
Samuel Curtis Johnson Graduate School of Management at Cornell University).
Kirilenko, A, A. S. Kyle, M. Samadi and T. Tuzun (2011) ‘The Flash Crash: The
Impact of High Frequency Trading on an Electronic Market’, SSRN eLibrary, at:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1686004
Lehalle, C, and R. Burgot (2010) ‘Navigating liquidity 4’, Cheuvreux Credit Agricole Group,
April, available at: www.cheuvreux.com/pdf/Navigating_Liquidity_september2009.pdf
Lehalle, C., R. Burgot, M. Lasnier and S. Pelin (2010) ‘Navigating Liquidity 5’, Cheuvreux
Credit Agricole Group, December, available at: www.cheuvreux.com/pdf/Navigating
Liquidity5_December2010.pdf
Mackenzie, D. (2011) ‘How to Make Money in Microseconds’, London Review of
Books, available at: www.lrb.co.uk/v33/n10/donald-mackenzie/how-to-make-money-
in-microseconds
Mandelbrot, B. B. and H. W. Taylor (1967) ‘On the Distribution of Stock Price Differences’,
Operations Research, vol. 15, no. 6, pp. 1057–1062.
270 A. G. Haldane
O’Hara, M. (2004) Market Microstructure Theory (Oxlford: Blackwell Publishing).

Perrow, C. (1999) Normal Accidents: Living with High Risk Technologies (Princeton: Princeton
University Press).
Peters, E. (1994) Fractal Market Analysis: Applying Chaos Theory to Investment and Economics
(Hoboken: Wiley Finance).
Qian, B. and K. Rasheed (2004) ‘Hurst Exponent and Financial Market Predictabil-
ity’, IASTED conference on ‘Financial Engineering and Applications’ (FEA 2004),
pp. 203–209.
SEC (2010) ‘SEC Approves Rules Expanding Stock-by-Stock Circuit Breakers and Clarify-
ing Process for Breaking Erroneous Trades’, available at: www.sec.gov/news/press/2010/
2010-167.htm.
Smith, R. (2010) ‘Is HFT Inducing Changes in Market Microstructure and Dynamics’,
working paper, available at: http://arxiv.org/abs/1006.5490
Stoll, H. (1978) ‘The Supply of Dealer Services in Securities Markets’, Journal of Finance,
vol. 33, no. 4, pp. 1133–1151.
Vaughan, D. (1996) The Challenger Launch Decision: Risky Technology, Culture, and Deviance
at NASA (Chicago: University of Chicago Press).
Zhang, F. (2010) ‘High-Frequency Trading, Stock Volatility, and Price Discovery’, mimeo,
available at: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1691679
14
A Model of Private Equity Fund
Compensation∗
Wonho Wilson Choi
Korea Advanced Institute of Science and Technology, South Korea
Andrew Metrick
Yale School of Management, USA
Ayako Yasuda
University of California at Davis, USA
1 Introduction
Private equity funds are typically organized as limited partnerships, with pri-
vate equity firms serving as general partners (GPs) of the funds and investors
providing capital as limited partners (LPs). These partnerships usually last for
ten years, and partnership agreements (investor contracts) signed at the funds’
inceptions clearly define the expected GP compensation. Since the payments
to GPs can account for a significant portion of the total cash flows of the
fund, the fund fee structure is a critical determinant of the expected net fund
returns that the LPs receive. Metrick and Yasuda (2010a) estimate the expected
present value of the compensation to GPs as a function of the fee structure
specified in investor contracts, but do not consider the fair-value test (FVT)
scheme, which is a commonly used carried interest scheme in practice.1 In this
chapter, we evaluate the present value of the FVT carried interest scheme by
extending the simulation model developed in Metrick and Yasuda (2010a), and
compare the relative values of the FVT carry scheme to other benchmark carry
schemes.
The FVT carried interest scheme allows early carry payments before the fund’s
carry basis has been returned to investors if certain conditions are met. The FVT
scheme is almost always accompanied by a clawback provision (see Section 2.6
for definitions); thus, the final nominal amount of carry for the fund’s lifetime is
unchanged whether the fund uses an FVT scheme or a more conservative carry
timing scheme, holding all other fund terms (such as carry % level) equal. In
other words, the main impact of the FVT scheme derives from the time value of
money.
271
272 W. W Choi, A. Metrick and A. Yasuda
The conditions for the FVT scheme are twofold. First, upon any exit, the cost
bases of all exited or written-off companies to date must be returned to LPs
before any distribution to GPs. In addition, the distribution to GPs is made only
if the sum of the fair values of all un-exited (that is, remaining) companies under
management at the time of the exit equals or exceeds a threshold value, defined
as a multiple of the total cost bases of un-exited investments with the most typ-
ical multiple being 1.2 (120%). The fair values of remaining investments cannot
be easily marked to market since these private equity investments are illiquid
by nature; in practice, estimate values that are reported by GPs are used. Since
GPs are thought to possess an information advantage over LPs as insiders, the
information asymmetry between them gives rise to a potential agency problem
when GPs use self-reported portfolio values to calculate their carried interest. We
investigate whether GPs are tempted to inflate the portfolio values of un-exited
companies by examining the effects of inflated values on the expected PV of GP
compensation.
In our analyses, we extend the model employed in Metrick and Yasuda (2010a)
by mapping the exit timing and exit values of portfolio investments as well as
the interim values of un-exited investments into the timing and amount of
GP carry according to the FVT carry scheme. We obtain detailed information
on the terms and conditions for fair-value tests used in practice from a large
anonymous investor who also provided other information for the analyses in
Metrick and Yasuda (2010a). We match the parameter values of our FVT model
to the values most commonly used in these actual funds. We then compare the
expected GP compensation of the fund with an FVT carry scheme to those of
two other benchmark funds.
Our findings generally indicate that the FVT carry scheme is substantially
more valuable to the fund managers than other commonly observed (and more
conservative) carry schemes, but interestingly, conditional on having an FVT
carry scheme, fund managers’ incremental gains from inflating the reported
values of the funds’ un-exited portfolio companies would be negligible.
The remainder of the chapter is organized as follows. In section 2, we describe a
model of private equity fund compensation in a risk-neutral pricing framework.
In section 3, we report the model outputs as a function of various input values.
We conclude in section 4.
2 A model of private equity fund compensation
Payments to GPs running private equity funds consist of management fees and
carried interest for venture capital (VC) funds; for buyout (BO) funds, there are
additional fees called transaction fees and monitoring fees. While management
fees are based on the cost bases of fund portfolio investments (and/or the fund
size), the amount of carried interest (= carry) received by GPs is based in general
A Model of Private Equity Fund Compensation 273
on the timing and exit values of portfolio companies and thus is sensitive to
fund performance. In the FVT carry scheme, the timing and amount of GP
carry also depend on the interim values of un-exited portfolio companies.
In this section, we describe a risk-neutral valuation method for the estimation
of the PV of carry starting with the determination of the initial investment value
of a portfolio company. We then specify the dynamics of the company value
during the holding period, the stochastic exit time point, and the values of
exited and all other un-exited investments in the fund portfolio at every exit
time point. We finally apply various functions that correspond to specific profit
sharing rules by mapping the exit (and interim) values of portfolio companies
to the amount of GP carry.
2.1 Risk-neutral valuation

The estimation of the present value of GP carry for a VC/PE fund is complicated
because appropriate discount rates are hard to estimate empirically. Since invest-
ments are illiquid and individual project returns are not fully realized until the
end of the fund life – usually ten years – it is not easy to measure risk (‘beta’) at
the fund level using standard time-series correlations with the market and other
factor returns. Many of the studies that employ fund-level cash flow data make
an effective assumption that market beta for the asset class is equal to one.2 In
this analysis, we take a risk-neutral valuation approach and build a simulation
model to overcome this data problem while matching parameter values of the
simulation model to those that are supported by empirical evidence wherever
estimates are available.3
2.2 Initial investment values

Since GPs receive a stream of semi-fixed compensation through management
fees and these fees come out of committed capital, the investment capital that
can be used for investments is always less than the committed capital that is
provided by LPs. Since a minimum necessary condition for any type of equi-
librium should state that at least the committed capital be returned to LPs in
expectation, GPs must somehow create values to reconcile the gap between the
investment capital and the committed capital. For example, the value creation
may come from the possibility that GPs make a lucrative purchase at a low price,
and/or from the possibility that GPs has a special skill to improve the value of
the firm over time.
We assume a fixed amount of value creation by GPs in each investment, fol-
lowing Metrick and Yasuda (2010a). We set this value such that a fund with $100
of committed capital would have a total initial value of investments at $106.71.
This number is chosen so that the expected value to LPs is exactly equal to com-
mitted capital for our benchmark VC fund (Fund I as described in Section 2.6).
That is, for every $100 in committed capital, the LPs pay some amount in man-
agement fees and the GPs then create value (after which the portfolio is worth
$106.71 in present value) and take out another expected amount in carried
interest, after which exactly $100 in expectation is left over for the LPs. Then,
given this initial investment value, we simulate the value paths for individ-
ual investments by assuming stochastic processes as described in the following
section.
2.3 The dynamics of the value of a portfolio company

Let Xti be the market value of portfolio company i at time t. It is assumed to
follow a Gaussian diffusion process in a risk neutral world of the following form:
dXti

= rdt + σ 1 − ρ 2 dWti + ρdWtF (1)
Xti
where r is the risk-free rate and σ is the volatility of the investment. Note that
Wti and WtF are standard Brownian motions, which are mutually independent
where Wti is specific to portfolio company i and WtF is common across portfolio
companies. By assuming the diffusion process as such, ρ captures the correlation
between the values of a portfolio company and the common factor. We further
j j
assume that Wti and Wt (i = j) are uncorrelated so that corr(d ln Xti , d ln Xt ) = ρ 2 .4
It is important to note that the process is not for the intrinsic value of a com-
pany, but for its market value. The intrinsic value of an illiquid asset is generally
different from its market value that would be appraised once it becomes tradable.
However, the carry distributions to GPs could occur only when a fund makes
any exit after which the exited company becomes less illiquid. For this reason,
we assume that the proceeds from an exited company at any exit are equiva-
lent to its market value while ignoring some frictions.5 It is also important to
note that this assumption makes our risk-neutral valuation method consistent.
Although the interim values of un-exited companies under management might
not be close to the market values, the interim values are not correlated with the
exited values in our model, so the assumption of the market value for un-exited
companies is not inconsistent with the risk-neutral valuation.
2.4 Random investment duration and random exit time

Random investment duration
Let di be the investment duration for portfolio company i. We specify di as a ran-
dom variable that follows an exponential distribution with the instantaneous
hazard rate of λ as follows:
f (di ) = λe−λdi (di ≥ 0) (2)
This distributional assumption is based on the observation that, in practice,

neither LPs nor GPs control the exit timing; rather, exit opportunities arrive
more or less exogenously. Furthermore, di is assumed to be independent of the

company value. While this second assumption is certainly false, it is compu-
tationally expensive to handle these correlations on large portfolios, and in
robustness checks using small portfolios we have not found any clear pattern
between correlation structures and expected carried interest.
For the baseline model, we use the exit rate of 20% (an inverse of 5) since the
average holding period for early VC investments is about five years (see Metrick
and Yasuda 2010b).
Random exit time

Private equity funds may invest in a portfolio company at any time during the
fund’s investment period, which typically lasts for five years starting from its
inception. Denote the time point of an investment in a portfolio company i by
si . In our simulation model, si is deterministic and is set to match the average
investment pace that is empirically observed in the data (see Section 2.7 for
details). Then, the exit time point ti for portfolio company i is the sum of si and
di , which is again a random variable.
2.5 Fair value and exit value

The fair value of a managed portfolio is the sum of the fair values of individual
portfolio companies under management. At time t, portfolio company i is under
management if and only if si ≤ t < ti . Given the diffusion process (1), the fair
value of portfolio company i (FVti ) follows a log-normal distribution:
For i ∈ {i|si ≤ t < ti },

i i i i σ2 2
FVt = Xt where ln Xt ∼ N ln XS + r − (t − si ), σ (t − si ) (3)
i 2
Similarly, the exit value of portfolio company i (EVti ) at its exit ti follows a
i
log-normal distribution:
For i ∈ {i|t = ti },

i i i i σ2 2
FVt = Xt where ln Xt ∼ N ln XS + r − (ti − si ), σ (ti − si ) (4)
i i i i 2
2.6 Mapping to carry amount

In our analysis, we evaluate the PV of carry for a fund with the FVT carry scheme
along with two other funds with commonly used carry rules (benchmark funds)
and compare them. Following Metrick and Yasuda (2010a), we choose the most
typical carry scheme for VC and BO funds, respectively (Fund I), and also the
simple carry scheme that does not allow early carry timing (Fund II) as the
benchmark funds. Fund III is the FVT fund, as described below:
Fund I: The fund with no or 8% hurdle, contributed capital returned first,

with clawback
The carry rule for this fund requires that, upon any exit, LPs must have received
cumulative exit distributions equal to the contributed capital (= cost bases of
all investments made to date + cumulative management fees paid to date), plus
any hurdle return (if any), before any distribution of carried interests to GPs
is allowed. Note that the calculation of contributed capital is at the aggregate
fund level, not at the deal level. Thus this amount starts small at the beginning
of a fund’s life, and eventually converges to the committed capital at the end
of its life for a fully invested fund. This carry rule with no hurdle is the most
popular carry rule for VC funds; for BO funds, this carry rule with an 8% hurdle
rate with a (100%) catch-up is most commonly employed.6 The main reason
for its popularity is that it allows GPs opportunities to earn carry early on in
the fund’s life. However, since the contributed capital changes over time, it is
possible for a fund with this carry rule to overpay carry to GPs (for example, if
the fund does well early on and then falters). The clawback provision requires
GPs in such instances to return the overpaid portion of their carry at the end of
the fund life.
Fund II: The fund with committed capital returned first

The fund with this carry rule requires that, upon any exit, LPs must have received
cumulative exit distributions equal to the committed capital before any distri-
bution of carried interest to GPs is allowed. Note that, this carry rule employs a
stricter notion of ‘fund profitability’ than Fund I, and as a result carry timing is
delayed. It rules out possibilities of carry overpayment to GPs by ensuring that
the LPs get paid the entire carry basis (= committed capital) before any carry is
distributed to GPs. While it is the least GP-friendly type of carry rule, it is found
in about a quarter of the VC fund data used in Metrick and Yasuda (2010a), and
is a useful benchmark case.
Fund III: The fund with a fair-value test, with clawback

The fund with this carry rule requires that, upon any exit, i) LPs must have
first received the cost basis of all exited (and written-off) companies to date
plus prorated management fees and ii) the fair-value test (FVT) is met before
any distribution of carried interest to GPs is allowed. The fair-value test requires
that the fair value (= estimated, reported value) of the remaining fund portfolio
companies equals or exceeds a preset threshold amount. The threshold amount
is calculated as a fixed percentage of the cost basis of all un-exited investments,
and a typical percentage is in the neighborhood of 120%. If the first criterion
is met but there is a small deficit between the fair value of the remaining fund
portfolio and the threshold value, the remaining exit value can be used to pay
down the deficit so that the FVT is met, and any leftover exit value can then be
split between GPs and LPs according to the carry level (for example, 20:80 for a
20% carry). Like Fund I, this type of carry rule is designed to allow GPs to earn
carry early in the fund’s life. Consequently, this carry rule is also susceptible
to potential carry overpayment. If the GPs are found to be overpaid carry at
the end of the fund’s life, the clawback provision requires that GPs return the
overpaid portion of the carry payment to LPs.
Note that, according to a survey on fees and carried interest (Dow Jones 2007),
the majority of respondent funds require the return of only a portion of con-
tributed capital before carry kicks in, suggesting that the first part of the FVT
scheme is commonly practiced. Furthermore, the second part of the FVT scheme
(the fair-value test) is also employed by about a fifth of the survey respondent
funds (21.2% of VC funds and 14.0% of BO funds). However, the same survey
also indicates that there are concerns among LPs that ‘GPs who tie the timing
of carried interest to [fair-value] tests might have an incentive to report higher
valuations than other GPs’. To the best of our knowledge, the effects of having
this type of carry rule on i) the value of GP compensation and ii) GP incentives
to inflate the value of unexited company portfolios have not been examined
before. Our chapter sheds light on both of these questions.
2.7 Simulation
Assessing the present value of a GP carry scheme is analogous to pricing a basket
call option. Although a basket option can be priced approximately in a closed
form,7 the evaluation of a GP carry scheme is more complicated because: 1) the
number of assets in the portfolio changes over time; and 2) the strike price also
fluctuates during the fund life for some of the carry schemes. Thus, we use the
Monte Carlo simulation method and compute the PV of carry numerically. To
analyze the GP carry as a function of the value paths of portfolio companies,
we further parameterize the baseline model as follows:
1) The fund makes a predetermined number of investments with equal sizes.

PE funds often have covenant provisions that prohibit GPs from investing a
large portion of the fund’s capital in a single investment, thereby ensuring
that the fund capital is diversified across investments (Metrick and Yasuda
2010b). The number of investments in a fund is set to match the median
value of the fund sample used in Metrick and Yasuda (2010a) – 25 for VC
funds and 11 for BO funds.
2) Investments are made at the beginning of each calendar year during the
first five years. The investment pace is set to match the empirically observed
average pace used in Metrick and Yasuda (2010a) – 8, 6, 7, 3, 1 for VC funds,
and 3, 3, 3, 1, 1 for BO funds.
3) The annual exit probability (20% ) is set to match the inverse of the average
holding period (five years), as in Metrick and Yasuda (2010a).
4) Any remaining investments not yet exited are (forced to be) liquidated at
the end of the 12th year from its inception. This cutoff date is based on
the observation that a fund commonly lasts for 10 years and there is fre-
quently a provision in the fund partnership agreement that allows up to two
consecutive one-year extensions on the fund’s life subject to LP approval.
5) For the BO fund model, we extend the VC fund model with additional struc-
tures pertaining to i) leverage, ii) transaction fees, and iii) monitoring fees.
Each individual BO fund investment is leveraged with 2:1 leverage ratio; thus,
the transaction price for each investment is three times the equity investment
in the firm by the BO fund. Entry transaction fees are charged to the portfolio
company at the time of the initial investment by the BO fund and the fees
are then split 50:50 between LPs and GPs. We set the entry transaction fees to
match the empirical average of 1.37% of firm value, as in Metrick and Yasuda
(2010a). Monitoring fees are set to be 2% of EBITDA, or 0.4% of firm value
per year for a firm with an EBITDA multiple of five, with a five-year contract.
These fees are assessed (as 0.4% *5 years = 2% of firm value) at exit, and then
split 80:20 between LPs and GPs. Note that leverage has direct impacts on the
transaction and monitoring fees, since these fees are charged as percentages
of the total firm value, as opposed to just the equity value. Furthermore, both
transaction fees and monitoring fees paid to LPs are used to pay down the
carry basis; thus, these fees affect the timing and amounts of carry for BO
funds, and thus are integral parts of our simulation model.
Under these assumptions, we make 10,000 Monte Carlo simulations and obtain
the average of GP carry.
3 Model outputs
3.1 Baseline results

We first report the results of our baseline model, using the parameter values as
described in the first vertical panel of Table 14.1. The parameter values for the
baseline VC (BO) model are: 20% exit probability, 20% carry level, $100 carry
basis (= committed capital), 90% (60% ) total volatilities, 50% (20% ) pairwise
correlation, and 120% fair-value threshold level (plus 2:1 leverage ratio for the
BO model). Table 14.2 presents the simulation results. Panel A presents the PV
of carry per $100 of committed capital for VC funds; Panel B presents the results
for BO funds. For the VC fund with a 120% fair-value test and clawback, the PV
of carry is $9.42. This compares quite favorably to the two benchmark VC funds:
the fund with contributed capital returned first, with clawback, has the expected
PV of carry of $8.67, while the fund with committed capital returned first has
the PV of carry of only $8.61. Thus, our baseline model analysis indicates that
the FVT scheme is quite GP friendly for VC funds.
Table 14.1 Parameter values for the simulation model
Baseline model Variation considered
Panel A: Venture Capital Funds

Exit probability 20% 10%, 30%
Carry level 20% 25%, 30%
Carry basis $100 $82 (investment capital)
Total volatility of an investment 90% 60%, 120%
Pairwise correlation 50% 30%, 70%
Fair-value threshold level 120% 112%, 125%, 130%
Inflated value of un-exited investments 100% 125%, 150%
Panel B: Buyout Funds
Exit probability 20% 10%, 30%
Carry level 20% 25%, 30%
Carry basis $100 $88 (investment capital)
Total volatility of an investment 60% 30%, 90%
Pairwise correlation 20% 10%, 50%
Fair-value threshold level 120% 112%, 125%, 130%
Inflated value of un-exited investments 100% 125%, 150%
Leverage 2:1 1:1, 4:1
This table describes i) the default parameter values used in the baseline simulation model and ii)
variations considered for sensitivity analysis. Panel A presents the parameter values chosen for the VC
model; Panel B presents the values for the BO model. In the baseline model, a VC (BO) fund makes
25 (11) investments of equal sizes at the pace of 8, 6, 7, 3, and 1 (3, 3, 3, 1, and 1) investment(s) at
the beginning of each of the first five years, respectively. The investment pace follows the empirically
observed average investment pace as discussed in Metrick and Yasuda (2010a). From the time of the
investment, each portfolio company is assumed to have the instantaneous hazard rate (= death rate,
or exit probability) of 20%, independently with respect to any other portfolio companies. The market
value of portfolio company i at time t , Xti , is assumed to follow dXti /Xti = rdt + σ ( 1 − ρ 2 dWti + ρdWtF )
where the default risk-free rate (r ) is 5%, the volatility (σ ) is 90% (60% for BO), and the pairwise
correlation (= ρ 2 ) is 50% (20% for BO). For a given carry scheme, the default carry level is 20%, the
carry basis is $100, the threshold level for the fair-value test is 120%, and the reported value of un-
exited investments is 100% of the actual value (that is privately observed/assessed by GPs). For the
baseline BO model, the leverage ratio of 2:1 is also assumed. While the carry level and basis determine
the nominal amount of carry that GPs are entitled to, the fair-value threshold level and the ratio of
reported to actual values of un-exited investments determine the carry timing
The BO model results are qualitatively similar. The PV of carry for the BO
fund with a 120% FVT is $6.18. The low value in comparison to the VC fund
is due to the differences in underlying parameter values, in particular the lower
volatility (60% vs 90% ) for individual investments and also the lower pairwise
correlation (20% vs 50% ) between investments. More importantly, the relative
GP friendliness of the FVT scheme remains unchanged: the BO fund with con-
tributed capital plus 8% hurdle returned first, with clawback, has the expected
PV of carry of $5.04, which is a lot lower than the FVT fund expected carry of
$6.18. The 8% hurdle delays the carry timing, which hurts the PV of GP carry
and makes this fund term less GP-friendly than the other two. The BO fund with
Table 14.2 Baseline model results
Most comon (Fund I) No early carry (Fund II) FVT (Fund III)
$8.67 $8.61 $9.42

$5.04 $5.60 $6.18
This table presents the simulation results of calculating the expected PV of carry for the three repre-
sentative funds. Panel A presents the PV of carry per $100 of committed capital for VC funds; Panel B
presents the results for BO funds. The parameter values for the baseline VC (BO) model are: 20% exit
probability, 20% carry level, $100 carry basis (= committed capital), 90% (60% ) total volatilities, 50%
(20% ) pairwise correlation, and 120% fair-value threshold level (plus 2:1 leverage ratio for BO). For
the VC model, the most common fund (Fund I) requires that, upon any exit, LPs must have received
cumulative exit distributions equal to the contributed capital (= cost bases of all investments made to
date + cumulative management fees paid to date) before any distribution of carried interests to GPs is
allowed. For the BO model, the most common fund (Fund I) requires that, upon any exit, LPs must
have received cumulative exit distributions equal to the contributed capital (= cost bases of all invest-
ments made to date + cumulative management fees paid to date), plus 8% hurdle return, before any
distribution of carried interests to GPs is allowed. GPs then catch up with LPs with the catch-up rate
of 100%. For both the VC and BO model, the ‘no early carry’ fund (Fund II) requires that, upon any
exit, LPs must have received cumulative exit distributions equal to the committed capital before any
distribution of carried interests to GPs is allowed. For both the VC and BO model, the FVT fund (Fund
III) requires that, upon any exit, i) LPs must have first received the cost bases of all exited (and written
off) companies to date plus prorated management fees and ii) the fair-value test (FVT) is met before
any distribution of carried interests to GPs is allowed. The fair-value test requires that the fair value (=
estimated, reported value) of the remaining fund portfolio companies equals or exceeds 120% of the
cost bases of all un-exited investments. If the first criterion is met but there is a small deficit between
the fair value of the remaining fund portfolio and the threshold value, the remaining exit value can
be used to pay down the deficit so that the FVT is met, and any leftover exit value can then be split
20:80 between GPs and LPs.
committed capital returned first has the expected PV of carry of $5.53, which is
also significantly smaller than the FVT fund carry.
3.2 Effect of early timing advantage of the FVT scheme on PV of carry

The baseline analysis indicates that, while the FVT scheme is the most GP-
friendly for both the VC and BO funds, it particularly favors GPs of funds
with high-volatility investment portfolios, such as early-stage VC funds. This
is because high volatility makes it more likely that GPs earn early carry; con-
ditional on getting carry early, the high volatility also makes such carry larger
in expectation. Note, however, that the FVT fund we examine in the analysis
above, as well as virtually all funds observed in practice, come with the claw-
back provision, so that the nominal amount of carry net of clawback at the
end of the fund’s life is the same across all three funds. The clawback provision
requires GPs to return any excess carry payment at the end of fund life, when
all three funds are required to have returned to LPs exit distributions equal
to committed capital (because the contributed capital converges to committed
Table 14.3 The effect of carry timing rules on PV of carry
Risk-free Rate
5% 4% 3% 2% 1% 0%
Fund I: with no hurdle, contributed capital returned first with clawback

Present value of GP carry before clawback 8.77 8.75 8.72 8.69 8.66 8.63
Present value of the clawback 0.10 0.10 0.11 0.12 0.13 0.14
Present value of GP carry (net of clawback) 8.67 8.64 8.61 8.57 8.53 8.49
Fund II: with no early carry
Present value of GP carry 8.61 8.59 8.57 8.55 8.52 8.49
Fund III: with a 120% threshold fair-value test, with clawback
Present value of GP carry before clawback 10.25 10.21 10.15 10.09 10.03 9.97
Present value of the clawback 0.84 0.94 1.05 1.18 1.32 1.48
Present alue of GP carry (net of clawback) 9.42 9.27 9.10 8.91 8.71 8.49
This table presents the simulation results for the PVs of carried interest (in $, per $100 of committed
capital) as functions of carry timing rules and the level of the risk-free rate. PVs of GP carry are calculated
for three different fund terms: ‘Fund I: with no hurdle, contributed capital returned first with clawback’
is a fund whose VC GPs are entitled to carry after returning the contributed capital to LP, subject to
clawback. ‘Fund II: with no early carry’ is a fund whose GPs must return all of carry basis before they are
entitled to carry, thus ruling out any necessity for clawback. ‘Fund III: with a 120% threshold fair-value
test, with clawback’ is a fund whose GPs are entitled to carry after returning the cost basis of all exited
(or written-off) investments and meeting the 120% fair-value test criteria for un-exited investments.
The risk-free rates vary from 0% to 5% by 1% increments.
capital for fully invested, completed funds). Thus, the difference in PV of carry
across three funds derives entirely from the time value of money, or the discount
rate. In other words, the FVT scheme is GP-friendly because of its carry timing
advantage, not because it entitles GPs to more carry in expectation.
To illustrate this point, we simulate and present the VC model results with
different values of risk-free rate in Table 14.3. For the two funds with early carry
possibilities and clawback, we further break down the results into i) the PV of
carry before clawback, ii) the clawback amount, and iii) the PV of carry after
clawback. With 0% risk-free rate (as shown in the last column), the amounts
of GP carries net of clawback are identical across the three funds ($8.49), since
the excess early carry is exactly offset by the clawback amount. Note that the
FVT fund has a clawback amount ($1.48) that is ten times as large as the fund
with contributed capital returned first ($0.14). When the risk-free rate (which
is the discount rate in the risk-neutral world) is positive, this large early carry
gives the FVT carry scheme (Fund III) a larger PV of carry than the other two.
Furthermore, while the fund with the contributed capital returned first (Fund
II) also earns a larger PV of carry than Fund I when the risk-free rate is positive,
the impact of increasing risk-free rates is more pronounced for the FVT fund.
Thus the present value of GP carry in Fund III is more sensitive to risk-free rate
increases than that in Fund I.
3.3 Effect of inflation of un-exited investment values

Next we investigate whether GPs are tempted to inflate the portfolio values of
un-exited companies. As mentioned earlier, there are concerns among LPs that
‘GPs who tie the timing of carried interest to [fair-value] tests might have an
incentive to report higher valuations than other GPs’. Yet it is also plausible
that having some kind of FVT is better than not requiring any FVT threshold at
all (effectively setting the FVT threshold at 0%) even in the presence of asymmet-
ric information. Table 14.4 examines the effects of inflated values of un-exited
portfolios on the GP compensation. The benchmark case of 100% is the case
in which the market values are accurately appraised and self-reported by GPs.
Relative to the benchmark case, the present value of GP carry before clawback
increases slightly with the inflated level of 125% from $10.25 to $10.40, and
increases further but only marginally with the level of 150% from $10.40 to
$10.48. However, note that the clawback condition kicks in and offsets much
of these increases. As expected, the amount of clawback is bigger with a higher
inflated level for un-exited investments. With this offset, the PV of carry net
Table 14.4 The effect of inflated (reported) values of un-exited investments on the PVs
of carry
Inflation Level of
Un-exited Investments
100% 125% 150%
Panel A: venture capital funds

Fund III: with a fair-value test, with clawback
Present value of GP carry before clawback 10.25 10.40 10.48
Present value of the clawback 0.84 0.93 0.98
Present value of GP carry (net of clawback) 9.42 9.48 9.51
Panel B: buyout funds
Fund III: with a fair-value test, with clawback
Present value of GP carry before clawback 6.61 6.85 7.01
Present value of the clawback 0.43 0.57 0.67
Present value of GP carry (net of clawback) 6.18 6.28 6.34
This table presents the simulation results for the PVs of GP carry as a function of the ratio of reported to
actual values (that are privately observed/assessed by GPs) of un-exited investments. The actual port-
folio values of un-exited investments are generated by the stochastic process as described in Equation
(1). For the baseline model, the reported value is 100% of the actual value (no inflation). For the results
in the last two columns, the reported values are assumed to be inflated by 25% and 50%, respectively,
from the actual (privately observed) values.
of clawback is affected only moderately when GPs inflate the value of their un-
exited investments, where the minor increases (from $9.42 at 100% to $9.51 at
150%) come from the time value of early carry. These increases amount to less
than 1% of the total PV of carry. The results are qualitatively similar for BO funds
(presented in Panel B), though we note that the increases are proportionately
larger; the increase from $6.18 (at 100%) to $6.34 (150%) represents 2.6% of
the total PV of carry ($6.18). Thus, we find that, conditional on having the
FVT carry scheme, GPs make only negligible amounts of gains by inflating the
values of their un-exited portfolios, suggesting that the LPs’ concerns are not
warranted.
3.4 Effect of other parameter values

We examine the effects of altering various parameter values on the present value
of GP compensation across the three funds to investigate whether these effects
are more or less substantial for Fund III that applies the FVT. We examine the
effects on carry of perturbing six (seven for BO) model parameter values – exit
probabilities, carry levels, carry basis, total volatilities of companies, pairwise
correlations between portfolio companies, and fair-value threshold levels (plus
leverage for BO funds).
Table 14.5 presents the effects of altering parameter values on PV of carry
for the three funds. The parameter values for the baseline VC (BO) model are:
20% exit probability, 20% carry level, $100 carry basis (= committed capital),
90% (60%) total volatility, 50% (20%) pairwise correlation, and 120% fair-value
threshold level (plus 2:1 leverage for BO funds). We perturb these values as
described in the second vertical panel of Table 14.1. Panel A of Table 14.5
presents the results for VC funds; Panel B presents the results for BO funds.
The exit probabilities have negative effects on the present values of GP carry in
a concave way across the three levels of 10%, 20%, and 30%. The carry levels
have positive effects in a concave way across the three levels of 20%, 25%, and
30%, but the concavity is quite negligible. When the carry basis changes to the
investment capital of $82 ($88 for BO funds),8 the magnitude and the percent-
age of the increases in the value is the largest for Fund III, and the smallest for
Fund II.
As expected, increases in either the volatility of an individual company or the
pairwise correlation lead to higher compensation to GPs. However, given the
levels of GP carry, the effects of either volatility or pairwise correlation are about
the same across the three funds for VC; for the BO, the impact is larger for Fund
I, because this fund has to meet the 8% hurdle rate. Also as expected, the FVT
threshold level has a negative effect on the present value of GP carry. Finally,
the leverage has a negative effect on the PV of carry for BO funds. Though this
may be surprising, note that the total expected compensation to GPs, which
includes transaction fees and monitoring fees, rises with higher leverage.9 Both
Table 14.5 Sensitivity analysis
Most common No early carry FVT

(Fund I) (Fund II) (Fund III)

Baseline model 8.67 8.61 9.42
10% exit probability 11.43 11.41 12.17
25% carry level 10.84 10.76 11.77
30% carry level 13.00 12.92 14.13
Investment capital basis 9.74 9.65 10.60
60% volatility 6.78 6.76 7.55
120% volatility 9.55 9.45 10.24
30% pairwise correlation 7.93 7.88 8.66
112% fair-value threshold level 9.44
Baseline model 5.04 5.60 6.18
25% carry level 6.27 7.00 7.72
30% carry level 7.49 8.40 9.27
Investment capital basis 5.98 6.49 7.23
30% volatility 2.65 4.01 4.54
90% volatility 7.20 7.40 7.98
1:1 leverage 5.16 5.73 6.31
4:1 leverage 4.80 5.35 5.91
This table presents the effects of altering the parameter values of the simulation model on the estimated
PV of carry. Fund I for the VC model (BO model) is a fund with no hurdle (8% hurdle), contributed
capital returned first with clawback. Fund II is a fund with committed capital returned first. Fund III is a
fund with a fair-value test and with clawback. The baseline model refers to the model results reported in
Table 2. ‘10% exit probability’ refers to an altered model that is the same as the baseline model, except
that the exit probability is set to 10% (instead of 20%). ‘30% exit probability’ is similarly defined. ‘25%
carry level’ refers to an altered model that is the same as the baseline model except that the carry level
is set to 25%. ‘30% carry level’ is similarly defined. ‘Investment capital basis’ refers to an altered model
that is the same as the baseline model except that the carry basis is investment capital (which is set to
$82 ($88 for BO) per $100 of committed capital). ‘60% volatility’ refers to an altered model that is the
same as the baseline model except that the annual volatility of individual investments is set to 60%.
‘120% volatility’ is similarly defined. ‘30% pairwise correlation’ is an altered model that is the same as
the baseline model except that the pairwise correlation between individual investments is set to 30%.
‘70% pairwise correlation’ is similarly defined. ‘112% fair-value test threshold’ is an altered model that
is the same as the baseline model except that the threshold level for the fair-value test is set to 112%.
‘125% fair-value test threshold’ and ‘130% fair-value test threshold’ are similarly defined.
the transaction fees and monitoring fees are assessed on the total firm value,
which become larger relative to the size of the BO fund’s equity investment
when leverage is higher. Entry transaction fees reduce the initial value of equity
investments, while the exit monitoring fees reduce the amount of exit value to
be split between LPs and GPs. Both of these effects reduce the amount of carry
that GPs receive in expectation, while sharply increasing the transaction and
monitoring fees that GPs and LPs share. The impact of leverage on PV of carry
is quantitatively similar across the three funds examined.
4 Conclusion
This chapter has analyzed the economics of private equity fund compensation.
We have evaluated the effect of using a fair-value test GP carry scheme on the
present value of GP carried interest relative to other carry schemes. We find that,
while the use of the fair-value test has a significantly positive effect on the PV of
carry relative to other commonly used carry schemes, GPs gain only a marginal
increase in their expected PV of carry by reporting inflated values for the un-
exited (and therefore illiquid) investments remaining in their fund portfolios.
Our findings suggest that the fair-value test scheme is a favorable compensation
scheme for GPs, but should not induce GPs to significantly misreport portfolio
values.
Notes
∗ This paper was previously titled as ‘Expected Carried Interest for Private Equity Funds’.
All errors and omissions are our own.
1. Dow Jones (2007).
2. See Section 4.1 of Metrick and Yasuda (2011) and the citations therein.
3. See Section 2.2 of Metrick and Yasuda (2010a) for more detailed discussions.
4. This correlation structure in a stochastic process is widely used in credit risk manage-
ment and commonly known as one-factor Gaussian copula (see, for example, Briys and
de Varenne 1997, Duffie and Singleton 2003, Hull 2007 and Schonbucher 2003).
5. A majority of exits are made through IPOs or sales to other companies. While the
proceeds from an exit may be different from the market value, for the purpose of our
analysis we ignore these differences. In the case of IPOs, the difference may come from
the total direct costs (see Lee et al. 1996) and the underpricing of IPOs.
6. A hurdle rate (also known as preferred return) is quite popular among BO funds; it is
less popular among VC funds. The catch-up feature is almost always present in funds
with hurdle rate. This feature allows GPs to receive disproportionate amounts of exit
distributions after the fund distributes the required hurdle returns to LPs until GPs
‘catch up’ with LPs. With the catch-up feature in place, the hurdle return affects carry
timing but not the final carry amount as long as the overall fund return is equal to or
above the hurdle rate; if the fund return is below the hurdle rate, then the carry amount
is also affected. See Metrick and Yasuda (2010a, 2010b) for more detailed explanations
and examples.
7. A basket option is an option on a portfolio of assets with a predetermined strike price.

A basket option can be priced only approximately in a closed form. See Gentle (1993)
and Huynh (1994) for lognormal approximations and Milevsky and Posner (1998) for
a reciprocal Gamma approximation.
8. The investment capital level is set to match the empirically observed average fund
terms, as in Metrick and Yasuda (2010a). The BO fund has larger investment capital
than the VC fund because BO funds on average charge lower management fees.
9. See Metrick and Yasuda (2010a), Figure 4 (p. 2334).
References
Briys, E. and F. de Varenne (1997) ‘Valuing Risky Fixed Rate Debt: An Extension’, Journal
of Financial and Quantitative Analysis, vol. 32, no. 2, pp. 239–248.
Dow Jones (2007) Private Equity Partnership Terms and Conditions, Fifth Edition, see:
www.dowjones.com/info/private-equity-terms-and-conditions.asp
Duffie, D. J. and K. J. Singleton (2003) Credit Risk (Princeton: Princeton University Press).
Gentle, D. (1993) ‘Basket Weaving’, Risk, vol. 6, no. 6, pp. 51–52.
Hull, J. C. (2007) Risk Management and Financial Institutions (Harlow: Pearson-Prentice
Hall).
Huynh, C. B. (1994) ‘Back to Baskets’, Risk, vol. 7, no. 5, pp. 5–61.
Lee, I., S. Lochhead, J, Ritter and Q. Zhao (1996) ‘The Costs of Raising Capital’, Journal of
Financial Research, vol. 19, no. 1, pp. 59–74.
Metrick, A. and A. Yasuda (2010a) ‘The Economics of Private Equity Funds’, Review of
Financial Studies, vol. 23, no. 6, pp. 2303–2341.
Metrick, A. and A. Yasuda (2010b) Venture Capital and the Finance of Innovation (Hoboken:
John Wiley and Sons).
Metrick, A. and A. Yasuda (2011) ‘Venture Capital and Other Private Equity: A Survey’,
European Financial Management, vol. 17, no. 4, pp. 619–654.
Milevsky, M. A. and S. E. Posner (1998) ‘A Closed-form Approximation for Valuing Basket
Options’, Journal of Derivatives, vol. 5, no. 4, pp. 54–61.
Schonbucher, P. J. (2003) Credit Derivatives Pricing Models (New York : Wiley and Sons).
Part IV
Taxation in a Globalized Economy
15
How Should Income from
Multinationals Be Taxed?
Roger Gordon∗
UCSD, USA, and CKGSB, China
What do optimal tax models imply about how multinationals should be taxed?
How can we best explain the difference between actual policies and the optimal
policies implied by existing theories? These questions are becoming increasingly
salient, given the growing importance of multinationals in the global economy.
The tax treatment of multinationals is one of the most esoteric parts of the
tax law. Any discussion must necessarily deal with the interactions of the tax
laws among different countries. It must take into account provisions in various
bilateral tax treaties, and also take into account OECD rules restricting available
options for the design of the tax law.
Unavoidable as well is a discussion of problems with tax enforcement. Mea-
suring the overall income of a multinational is much more difficult than for
a purely domestic firm, given the government’s difficulties in monitoring the
earnings of subsidiaries located abroad. In addition, the government needs to
identify not only how much the multinational earned in total but also where
this income was earned, since the statutory tax treatment depends on the source
of this income. Firms can easily engage in transfer pricing to manipulate the
reported location of their income, and governments cannot easily double-check
all reported transactions prices within a multinational.
While most of the existing literature focuses on the tax treatment of cross-
border capital investments by firms, this chapter focuses as well on the tax
treatment of the above-normal return to multinationals arising from past
entrepreneurial efforts. As seen below, existing tax rules come close to what the
theory recommends for the tax treatment of entrepreneurial income, but dif-
fer dramatically from what the theory recommends for the taxation of income
from cross-border investments.
We can then reconcile many aspects of the existing tax treatment of cross-
border income flows with theoretical forecasts if countries do not desire to tax
the return to savings, only to impose a uniform tax on the return to labor
effort. The theory focuses on the pressure both home and host governments
289
290 R. Gordon
face due to income-shifting between the personal and corporate tax base by
employees of any firm, and the income-shifting across countries undertaken by
multinationals. Given this income-shifting, host-country governments have an
incentive to tax the income of foreign subsidiaries located in the country, to
discourage income-shifting by the subsidiary’s employees. Home-country gov-
ernments have an incentive to impose a sufficient tax at repatriation to avoid
any tax avoidance by multinationals through transfer pricing. Given the tax
already imposed by the host country on the earnings of a foreign subsidiary, the
home-country tax rate sufficient to deter income-shifting turns out to be entirely
consistent with OECD rules governing worldwide taxation, with full domestic
taxation of foreign-source income at repatriation and a credit for corporate taxes
paid abroad.
The outline of this chapter is as follows. Section 1 provides an initial devel-
opment of the theory in a closed economy setting when there is potential
income-shifting by entrepreneurs. Section 2 then examines policy in an open
economy, providing a theoretical analysis of optimal tax policies in both home
and host countries towards cross-border activity by portfolio investors and by
multinationals.
Section 3 compares actual tax policies to those forecasted by the theory.
Actual policies include both a ‘territorial’ treatment which exempts foreign-
source income from domestic corporate taxes, and ‘worldwide taxation’ which
taxes foreign-source income in full at repatriation but with a credit for taxes paid
abroad. Most of the differences between the theoretical forecasts and ‘worldwide
taxation’ disappear if countries are not attempting to tax the return to the sav-
ings of domestic residents. Most of the remaining differences disappear when the
initial theory is expanded to include income-shifting not only by entrepreneurs
but also by other employees in a firm. The difference between the theoretical
forecasts and ‘territorial’ tax treatment remains stark, though. Finally, section 4
provides brief conclusions.
1 Optimal design of a corporate tax in a closed economy
In this section, we explore in a closed economy how the tax law can be designed
to impose a tax at some rate m on labor income and rate n on real income from
savings.
A key assumption driving our analysis is that the desired tax rate on income
from labor does not vary depending on the form or source of this income,
and similarly for income from savings. The next subsection discusses this
assumption in more detail.
The following subsection examines the tax treatment of non-corporate
income, recognizing that the observed income includes returns to both labor
and capital, with a different desired tax rate on each. The last subsection then
How Should Income from Multinationals Be Taxed? 291
considers the tax treatment of corporate income, recognizing that entrepreneurs

can easily shift their earnings between the personal and the corporate tax
bases.
Conditions justifying a uniform tax rate on income

from either labor or savings
A key assumption in our analysis is that the desired tax rate on labor income
is the same regardless of the source of the income. In particular, the desired
tax rate is the same whether the individual is an employee or a business owner.
Similarly, we assume that the desired tax rate on savings is the same whether
the individual invests in financial assets or in her own business. While natural,
these assumptions merit some justification.
One initial justification is the argument by Diamond and Mirrlees (1971) that
in the presence of optimal proportional taxes on factor supplies and consump-
tion bundles, the optimal tax structure should avoid any distortions to how
goods are produced. Any such production distortions not only change the rela-
tive prices of different factors or different consumer goods, as can also be done
using differential factor taxes or differential retail sales tax rates, but in addi-
tion they add extra costs due to the resulting inefficiencies in production. The
second-best tax structure should avoid such production inefficiencies.
Saez (2002) provides a more general context for such an analysis, allowing
for nonlinear tax schedules and equity considerations. He finds no grounds for
making use of any tax distortions beyond a nonlinear tax on observed labor
income if the following two conditions are satisfied: 1) an individual’s cho-
sen amount of some activity conveys no additional information about his/her
marginal utility of income beyond what is known based on his/her observed
labor income; and 2) any tax on some activity affects labor supply simply due
to the resulting changes in the individual’s real wage rate.1
We assume from now on that under the optimal tax structure an individual’s
choice between working as an employee or becoming self-employed, or between
investing in financial assets or instead investing in one’s own firm, satisfies
the above two conditions. In particular, we assume that each of these choices
conveys no additional information about the individual’s marginal utility of
income.2 Nor does the elasticity of labor supply or savings vary across these
choices. As we proceed with the discussion, we will make the same assumption
with regard to the choices to set up a corporate vs a non-corporate business,
whether to undertake business activity at home or abroad, and whether to invest
one’s savings in domestic vs foreign financial securities.
Given these assumptions, there are neither equity nor efficiency grounds for
any differential tax treatment of particular investments or particular occupa-
tional choices.
292 R. Gordon
Tax treatment of non-corporate business income

How can the tax law be designed to impose a uniform tax rate on an individual’s
labor income, and a separate tax rate on income from savings, when an indi-
vidual owns a non-corporate business whose observed income includes both a
return to labor and a return to savings invested in the firm?
One possible approach, mirroring the current tax treatment is the following:
non-corporate income net of depreciation deductions can be taxed at rate m.
The tax rate on the labor income component of this income is then the same
as the rate the individual faces on any wage and salary income, avoiding any
distortions to the type of employment.
The effective tax rate on capital invested in a non-corporate firm then depends
on the chosen depreciation provisions. These provisions can be designed to
yield any desired effective tax rate on new investment, even given that the
resulting income net of depreciation is taxed at rate m. For example, expensing
yields a zero effective tax rate on such business investments whereas economic
depreciation leads to an effective tax rate equal to rate m.
Distortions to the allocation of savings are then avoided if the effective tax
rate on investments in a non-corporate business is the same as the effective
tax rate n on the return to financial savings. In particular, denote the interest
rate by r, the exponential rate of economic depreciation by de , the exponential
depreciation rate allowed under the tax law by δ, and the equilibrium marginal
product of capital by fK . The investment breaks even given an opportunity cost
of r(1 − n) if
∞ ∞
V (1 − m δe−tδ e−r(1−n)t dt) = fK (1 − m)e−de t e−r(1−n)t dt (1)
0 0
Here, V is the cost of the investment. Denote the present value of these tax
depreciation deductions by z, implying that z = δ/[r(1 − n) + δ]. We then infer
from equation (1) that
(r(1 − n) + de )(1 − mz)
fK = (2)
1−m
To avoid tax distortions to portfolio choice, z must then be chosen to assure
that fK − de = r, implying that
de + r(1 − n/m)
z= (2a)
r(1 − n) + de
We then conclude that
(1 − n)[de + r(1 − n/m)]
δ= = δ(de , n, m) (3)
n/m − n
With n = m, this implies economic depreciation, while with n = 0 it implies
expensing. In all other cases, depreciation schedules must vary by personal tax
bracket to avoid introducing portfolio distortions. The resulting link between

personal tax rates and the resulting depreciation rate is captured by the function
δ(de , n, m).3
In such a setting, how should capital gains from non-corporate activity be
taxed? Capital gains can arise for various reasons. For one, whenever the rate of
tax depreciation differs from the rate of economic depreciation, the market price
and the tax basis for a used asset will differ, generating capital gains (or losses) at
realization. To avoid tax incentives to churn, selling capital in order to write-up
the basis to the current market value at the cost of paying capital gains taxes,
capital gains need to be taxed at a rate equal to g = mz. Formally, the individual
breaks even if the firm is sold after s years if the loss in future depreciation
deductions for the initial firm and the capital gains taxes paid by the selling
firm just offset the new depreciation deductions received by the acquiring firm:
∞ ∞
−δs −δt −r(1−n)t −de s −δs −de s
mVe δe e dt + gV (e − e ) = mVe δe−δt e−r(1−n)t dt (3a)
0 0
Simple algebra shows that equation (3a) holds if g = mz, given that this capital
gains tax is applied to the market value of the firm minus its tax basis, Ve−de s −
Ve−δs , with the tax paid at realization.
Capital gains and losses can also arise due to new information, for exam-
ple, learning that an entrepreneurial venture was successful. To avoid dis-
tortions to the decision whether or not to sell the firm, what capital gains
tax rate is appropriate? If the entrepreneur keeps the business, assume
she earns the wage she could get elsewhere in period t plus Rt , where
these above normal returns decay over time at some exponential rate dR .
The present value of these above normal returns, after tax, then equals
∞ −r(1−n)+dR dt = R(1 − m)/[r(1 − n) + d ]. If instead the entrepreneur
0 R(1 − m)e R
sells the firm, the equilibrium price V will satisfy R/V = (r + dR ), as long as the
tax law does not distort the buyer’s choice of investment. With a capital gains
tax rate of g and no basis, the sale yields (1 − g)V . Simple algebra shows that
the entrepreneur is indifferent to selling if g = mz when z satisfies equation (2a)
with a depreciation rate dR rather than de .
Since z varies depending on the depreciation rate of a particular asset, g should
vary depending on the depreciation rate of the asset as well.
If there is only one value for g, then capital gains on rapidly depreciating
assets are too lightly taxed, and vice versa. The compromise value for g, though,
certainly satisfies g < m when n > 0 (though g = m when n = 0).
Tax treatment of corporate business income in a closed economy

Given this tax treatment of non-corporate businesses in a closed economy, how
should corporate income be taxed to avoid introducing distortions to the form
of business activity?
294 R. Gordon
What if the above firm incorporates? For simplicity, we ignore dividends, and
assume that the resulting income to the corporation is taxable solely as capital
gains at realization under the personal tax.4 We assume that the ‘effective’ capital
gains tax rate on accruing corporate income equals g e = ρg, where ρ < 1 due to the
deferral of tax payments until realization but ρ > 1 due to the lack of indexing
of capital gains for inflation.
When g e < m, due to deferral and a lower statutory tax rate,5 corporate
shares are treated more favorably under the personal income tax than are
non-corporate shares. This in itself creates a tax distortion encouraging the
entrepreneur to incorporate her firm, and then retain earnings rather than pay-
ing them out to the entrepreneur as wages. Taxes are deferred until shares in the
firm are ultimately sold, and then they are taxed at a lower statutory tax rate.6
A corporate tax can serve to minimize these avoidance opportunities. To do so,
the corporate tax should be designed so that the effective combined corporate
and personal capital gains tax liabilities are the same as the taxes due had the
firm been non-corporate. This surtax should be applied to corporate retained
earnings, which are otherwise untaxed under the personal tax.7
Tax distortions to the form of organization (and the resulting production
inefficiencies) can then be avoided if
(1 − m) = (1 − τ )(1 − g e ) (4a)
and if
δ c = δ, (4b)
where τ is the corporate tax rate and δ c is the depreciation rate used under
the corporate tax. Given these tax provisions, all returns to labor effort
(entrepreneurial income) are taxed at effective tax rate m, and all income from
savings is taxed at rate n, regardless of the organizational form of the firm.
Given any value of τ , however, equation (4a) holds for only one value of
(1 − m)/(1 − g e ). The choice of a corporate tax schedule then must trade off cases
where the resulting distortions go in different directions. The optimal compro-
mise rate would be below the top personal tax rate, but likely not by much since
most of the income-shifting opportunities are available to entrepreneurs in the
top personal tax brackets.8 Those in the highest personal tax brackets then face
an artificial incentive to prefer the corporate form, and conversely for those in
lower personal tax brackets.
An addition problem is that τ varies across firms due to any progressive rate
structure under the corporate tax schedule as well as due to incomplete loss
offset arising from the inability to make full use of loss carry-backs and the drop
in the present value of deductions due to loss carry-forwards.
The resulting distortions, due to variation in both m and τ , are at the heart
of several parallel literatures dealing with corporate decisions. For one, these
rate differences are the focus in discussions of the choice between corporate
and non-corporate forms of business. They are central to discussions of taxes
and use of debt vs equity finance. They also enter into discussions of forms
of compensation, given that employees can convert wage income into capital
gains through becoming self-employed or receiving compensation in the form
of underpriced corporate equity.
While some distortions remain, existing tax structures largely seem consistent
with the above theory in their tax treatment of domestic activity.
2 Optimal design of a corporate tax in an open economy
The above discussion, and much of the past literature on the effects of corporate
vs non-corporate tax rates, focuses on purely domestic firms. Yet with glob-
alization, an increasing fraction of GDP is produced by multinationals. How
would the above results be extended to an open economy? We consider sev-
eral specific situations in turn: outbound portfolio investment (FPI), inbound
FPI, outbound FDI undertaken by domestic multinationals, and inbound FDI.
Throughout, we draw on our key assumption that the optimal tax structure
should avoid production inefficiencies, so avoid any distortions to how the
labor supply or savings of domestic residents are allocated across firms or across
locations.
2.1 Taxation of outbound portfolio investments

If the domestic corporate tax rate has been chosen so that (1−m) = (1−g e )(1−τ ),
then individuals face no tax distortion when they choose whether to be an
employee, or become an entrepreneur setting up either a non-corporate or a
corporate business. Similarly, investors face no tax distortions to the form of
their savings if δ nc = δ c = δ(de , n, m): they face a tax rate of n whether they invest
their savings in bonds, a non-corporate business or corporate equity.
What if they invest some of their savings in foreign bonds? Individuals face
no domestic tax distortions to which bonds they invest in as long as the effec-
tive tax rate on the resulting real interest income is the same. This occurs if
the real income from foreign bonds is taxable at rate n under the domestic per-
sonal income tax each year. Uncovered interest parity implies that the pre-tax
interest rates will differ due to expected changes in the exchange rate between
the two currencies. Neutrality would therefore require that accruing capital
gains or losses on the bond principal should be included in taxable income each
year.9
What about purchases of foreign equity? To avoid distorting the type of equity
individuals invest in, the country’s effective tax rate must be the same on both
foreign and domestic equity. The return to both forms of equity is taxed at
296 R. Gordon
the capital gains tax rate under the personal tax. For investments in domes-
tic equity, there is an additional corporate tax on the retained earnings of the
firms, to ensure that the combined corporate plus capital gains tax rate results
in an overall tax rate of n. To ensure the same effective tax rate on purchases of
foreign equity, a corporate surtax would also need to be imposed at accrual on
the retained earnings attributable to the shares owned by domestic investors in
foreign equity.
There are several important obstacles to such policies, however. For one, these
policies imply that any publicly traded firm faces taxation by the government
in each country of residence of some of its shareholders. The definition of tax-
able income in general will vary by country, if only due to variation in what is
taxed under the personal vs the corporate tax in each country. The result is a
substantial administrative burden on any publicly traded firm.
An additional administrative problem is that non-resident governments have
no access to the financial records needed to audit the tax base of foreign firms
whose shares are partly owned by domestic investors.
A third administrative problem is monitoring the foreign investments of
domestic residents. Foreign firms are under no legal obligation to report the
earnings of their shareholders to any government other than the government
in the country in which the firm is located. A (partial) solution to these prob-
lems has been bilateral tax treaties. Through such treaties most countries have
agreed to some degree of sharing of information about such cross-border income
flows. However, each country faces a financial interest not to follow through
with such information-sharing, since this information-sharing makes its firms
a less attractive purchase for non-resident investors. Feasible tax treaties must
be mutually beneficial. When cross-border investments are not of comparable
size, side payments between governments may be needed to reach agreement
on such a treaty. For example, within the EU, countries reporting cross-border
interest payments to the investors’ home country get to keep 75 per cent of the
resulting personal tax revenue.
The dominant problem, though, is that, by international tax conventions, a
country does not have ‘tax nexus’ to impose a corporate tax on the profits of
a foreign firm, even if the firm has domestic shareholders, unless the firm is a
subsidiary of a multinational based in the home country. Such a restriction is
a natural way to avoid arbitrary taxes on economic activity in other countries.
However, this restriction prevents countries from imposing neutral taxation on
outbound FPI, leading to excess FPI.10
2.2 Taxation of inbound portfolio investment

From a domestic perspective, how should inbound portfolio investment be
taxed? If the country is a price-taker in the world capital market, then the
Diamond-Mirrlees (1971) result implies that the income accruing to foreign
portfolio owners should not be taxed.11 This implies to begin with that this
income should not be subject to withholding taxes. However, in addition it
implies that the investment should not be subject to domestic corporate income
taxes either. Corporate taxation of foreign capital invested in the domestic econ-
omy discourages gains from trade in capital, a distortion that should not be part
of an optimal tax structure in a small open economy.
To eliminate domestic tax on inbound portfolio investments, corporate
investments should be expensed to the degree that the return accrues to foreign
shareholders, and depreciated at rate δ(de , n, m) to the degree that the return
accrues to domestic shareholders.
Note that this tax exemption creates a strong financial incentive for domestic
residents to hide information about their country of residence when investing in
domestic shares. Partly, this can occur through chains of ownership, requiring
complicated tracing rules to back out the residence of the underlying sharehold-
ers. Even sophisticated tracing rules can be thwarted by routing the investment
through a holding company in a country that keeps confidential the identity of
the underlying investor.
One alternative for avoiding such evasion would be to exempt income from
domestic corporate taxes solely on those shares coming through countries that
agree to report the residence of the underlying investors. This is done currently
under the withholding tax, whereby there is a high default withholding tax rate
that is reduced through bilateral tax treaties. An important provision in these
treaties allowing such a reduction is information-sharing about the earnings
abroad of domestic residents.12
2.3 Taxation of outbound FDI

In order to assure uniform taxation of the labor income of domestic residents
at rate m and capital income at rate n, the domestic government would want
to impose a corporate tax each year at the rate satisfying equation (4a) on the
resulting foreign-source income, and to allow depreciation deductions as under
domestic tax law at rate δ(de , n, m). With this tax treatment, any entrepreneurial
income retained within the foreign subsidiary becomes taxable at the domestic
corporate tax rate and then taxable again as capital gains, implying an effective
tax rate equal to m.
One complication with such a tax on foreign-source entrepreneurial income
based on the residence of the entrepreneur is that the tax creates an incentive
on the entrepreneur to change residence. The main issue is tax deferral. Individ-
uals may come up with a very profitable idea, but the resulting taxable income
shows up gradually over many years. Emigration can then reduce taxes due on
all future taxable income generated by past effort and ideas. In fact, the same sit-
uation arises whenever individuals emigrate with large unrealized capital gains
on shares they own.
298 R. Gordon
To neutralize such incentives created by tax deferral of capital gains, a solution

would be constructive realization of these capital gains at the date of emigration.
The equivalent solution for entrepreneurs would require a constructive sale by
the firm or the entrepreneur of any intellectual property, with full taxation at
rate m of the resulting proceeds.
2.4 Taxation of inbound FDI

To what degree do the above arguments change when the foreign investor in the
domestic economy is a foreign multinational rather than a foreign individual?
If portfolio investment and FDI are perfect substitutes for foreign investors,
then they should be taxed equally. Productive efficiency then argues for a zero
net tax on the earnings of such foreign subsidiaries operating in the domestic
economy.
Foreign multinationals inevitably have at least some domestic shareholders.
The optimal tax treatment then would allow depreciation at rate δ(de , n, m) to
the degree the firm has domestic shareholders, and allow expensing for the
remaining fraction of new investments.
A key complication, however, is that the multinational as a whole may well
earn profits above the normal rate of return to capital, due to the value of
the entrepreneurial ideas generating demand for the firm’s products. These
entrepreneurial profits would not show up in the subsidiary’s tax base under
arm’s length pricing, being fully offset by royalty payments to the parent firm
for use of the ideas/technology generating these higher returns.13
Nonetheless, the host country may try to impose a tax on these above-normal
returns to the foreign subsidiary by restricting deductions for royalty payments,
or imposing a withholding tax on these payments. In order to assess the incen-
tives faced by the host-country government, consider alternative sources of
above-normal returns. One source of above-normal returns could be monopoly
profits on the sale of a unique product to domestic customers. Here, the opti-
mal VAT or retail sales tax would be positive, to transfer some of the monopoly
profits to the domestic economy. However, there are no grounds for the tax to
vary depending on whether the goods are produced in the domestic economy
through a foreign subsidiary or instead are imported from abroad.
Another source of above-normal returns could be valuable domestic infras-
tructure or a valuable legal system, such as protection of intellectual property,
which together make the subsidiary more profitable than it would be elsewhere.
If all firms benefit equally from these aspects of the domestic economy, then
domestic land prices and/or domestic wage rates will be bid up to the point
that all firms break even, including foreign subsidiaries. However, the benefits
may matter more for some firms, for example, those with important intellectual
property, and these firms are more likely to be multinationals.
To the extent that a country is not a price-taker in the market for attracting
profitable subsidiaries of foreign multinationals, the incidence of any corpo-
rate tax on these firms should fall in part on the firms, rather than entirely on
domestic factors (workers) employed by these firms. What is the evidence? Two
papers, by Mathur and Hassett (2006) and by Arulampalam (2007), both exam-
ine the incidence of corporate tax, and find that it is largely shifted to domestic
workers through a lower wage rate, consistent with full shifting of the benefits
from domestic infrastructure through higher wage rates to domestic workers.
Given standard errors, though, estimated coefficients cannot rule out some
fraction of the burden falling on firms rather than on their workers. In this
case, the optimal tax on the foreign subsidiary would be positive. In particular,
consider a tax at rate τ on the firms’ pure profits, defined to equal revenue
minus labor and capital costs but with any deduction for royalty payments
disallowed:14 = pQ − wL − rK. Assume that output prices and the interest rate
are set in the world market and do not change in response to taxes on the
pure profits of these foreign subsidiaries. The tax then either falls on workers
through a fall in their wage rate w or it falls on the firm’s shareholders. Let
the fraction of the tax falling on workers be denoted by α. Assume that the
government adjusts personal income tax schedules to leave the net-of-tax wage
rate unaffected, implying no resulting change in labor supply. The only impact
of the tax on the domestic economy is therefore on tax revenue, and tax rates
should be set to maximize tax revenue.
Tax revenue equals τ + T (wL), where T (wL) is the personal income tax
schedule. The value of τ that maximizes tax revenue, given the simultaneous
adjustment in the personal tax schedule to leave workers unaffected on net,
satisfies:
∂ ∂w ∂ ∂w
+τ +τ |w − L =0 (5)
∂w ∂τ ∂τ ∂τ
Given our assumption that workers bear α per cent of the tax, we know that
∂w/∂τ L = −α/(1 −τ ). Also, let ε = [(1 − τ )/]∂/∂(1 − τ )|w denote the elasticity
of the tax base with respect to the fraction of profits kept by the firm, holding
the wage rate fixed.
Substituting, we find that τ/(1 − τ ) = (1 − α)/ε. If the incidence of the tax falls
entirely on workers, then foreign subsidiaries should be exempt from domestic
taxation. As an alternative example, if workers bear 80 per cent of the tax,
and ε = .4, implying only a moderate responsiveness of firms to the tax, then
the equation implies τ = .33. In contrast, if ε = 1, then the optimal corporate
rate satisfies τ = .167. With a statutory tax rate around τ = .33, the first case
implies denying any deductions for royalty payments while the second implies
restricting them to about half of an arm’s length price.
This taxation of the above-normal profits of foreign subsidiaries that locate
in the country in part is serving as an optimal tariff, being used whenever the
300 R. Gordon
country is not a price-taker in the market for subsidiaries (when ε < ∞). In addi-
tion, however, these taxes help internalize a positive externality to foreign firms
in response to domestic investments in infrastructure and in the design of better
legal codes. Only with such taxes would the host country share in the benefits
to foreign residents resulting from improvements in domestic infrastructure.
Even if the government can impose effective restrictions on royalty deduc-
tions, however, firms have available many other ways to shift profits between
countries, including the location of debt finance and transfer pricing. Grubert
(1998) in fact estimates that foreign subsidiaries in the US have very low tax-
able income under US tax law, in part due to royalty payments, but also in part
due to heavy use of debt finance, and presumably in part due to use of transfer
pricing. The firm’s flexibility in the allocation of its taxable profits across loca-
tions, even with a restriction on royalty payments, should be sufficient that ε
should be very high, leading to little opportunity to gain from the taxation of
inbound FDI.
2.5 Optimal policies compared with various proposed neutrality

rules for taxing capital income
The existing academic literature focuses on a variety of neutrality conditions
that it argues the optimal policies should satisfy. To what degree would the
optimal policies satisfy each of the proposed criteria?
One such criterion is ‘capital import neutrality’, or CIN. Under this criterion,
capital invested in a country should face the same tax rate, regardless of the
country of residence of either the firm undertaking the investment or the indi-
vidual investor financing the investment. This condition would not hold under
the optimal policies described above. Capital invested in a country would face
no taxation from the host-country government, but the resulting income accru-
ing to the individuals owning the return to this investment would be taxed at
a rate varying according to the country of residence of these individuals.
Another commonly discussed criterion is ‘capital-export neutrality’, or CEN.
According to this criterion, capital investments undertaken by residents in a
country should face the same tax rate regardless of the location of the invest-
ment. This condition does hold under the optimal policies described above.
Capital investments would face the same domestic tax rate n, regardless of the
country in which the investment occurs, while the host country would impose
no net tax on capital investments in the country, given expensing.
A third criterion that has been introduced into the discussion by Desai and
Hines (2003) is ‘capital-ownership neutrality’, or CON. Under this criterion,
the tax rate faced on the return to the capital invested in a firm should be
unaffected by the identity of the individual investor, or her country of residence.
CON is clearly violated due to differences in n both within a country due to
progressive rate schedules and across countries due to differences in optimal tax
rates.
Except for CEN, therefore, these neutrality conditions do not help identify
tax structures that approximate optimal tax policies.
3 Actual tax treatment of cross-border activity
The actual tax treatment differs extensively from the optimal tax treatment
forecast above. The next subsection describes these differences. Under OECD
regulations, countries can either make use of ‘worldwide taxation’ of the
earnings of domestic multinationals, or instead can use a ‘territorial’ tax
system.
The second subsection explores possible omissions from the above model that
help reconcile the forecasts from the theory with use of ‘worldwide taxation’. In
particular, many of the differences disappear if n ≈ 0, implying no desired taxa-
tion of the return to savings. As seen below, most of the remaining differences
disappear if the model allows employees as well as entrepreneurs to shift their
earnings between the personal and the corporate tax base.
3.1 Description of the key differences between optimal

and actual tax provisions
Under OECD regulations, member countries can make use of either of two
alternative tax systems. Under the first, a ‘territorial’ tax system, no domestic
corporate taxes are collected on the foreign-source earnings of domestic multi-
nationals or on the financial returns abroad to domestic portfolio investors.
Under the second, ‘worldwide taxation’, domestic multinationals can be taxed
in full on their foreign-source earnings (typically at repatriation), but with a
credit for any corporate or withholding taxes paid abroad.15
These actual tax laws differ in a variety of ways from the tax law forecasted
above. For one, under existing law inbound FDI and FPI are both subject in
full to domestic corporate taxation, and may face additional withholding taxes
on cross-border payments. The above model, in contrast, forecasts no net tax-
ation of income accruing to portfolio investments, and at best low taxation of
inbound FDI.
Under ‘worldwide taxation’, outbound FPI faces accrual taxation under the
personal tax on the resulting financial income to individual shareholders, as
forecasted by the theory. However, no attempt is made to impose domestic cor-
porate taxes on the underlying income accruing abroad on equity investments.
Also, most host countries impose some withholding taxes on payouts to foreign
portfolio investors, contrary to the forecasts from the above model.
Consistent with the above theory, under ‘worldwide taxation’ domestic
multinationals owe domestic corporate taxes on their foreign source earnings.
302 R. Gordon
However, in virtually all countries using ‘worldwide taxation’, this tax is not
assessed at accrual but only when the resulting profits are repatriated. To the
extent that there is a tax on the return to savings, this deferral of tax payment
lowers the present value of the resulting liabilities.
In most major countries other than the US, foreign-source earnings of
home-country multinationals are exempt from domestic taxation, receiving a
‘territorial’ tax treatment. Here, the differences from the tax structure forecast
from the theory are particularly stark.
Another difference from the tax policies forecast above is the availability of
tax credits for any taxes paid abroad, up to the amount of taxes due in the home
country. Of course, the above theory forecasts no taxes due abroad that could
quality for such credits, except perhaps for low taxes on FDI abroad by domestic
multinationals.
3.2 Optimal and actual tax provisions when n ≈ 0

While in general the differences between the optimal and actual tax provisions
seem stark, these differences largely disappear if n ≈ 0, implying no desired tax
on income from savings.
With n ≈ 0, all investment qualifies for expensing and not just that fraction
owned by foreign shareholders. With n ≈ 0, there is also no pressure to impose
corporate tax on portfolio investments abroad.
There would still be grounds, however, for imposing a domestic corporate
tax on the profits generated abroad by FDI undertaken by domestic multina-
tionals since here there are likely to be above-normal profits, representing a
return to past entrepreneurial efforts by domestic entrepreneurs. When n ≈ 0,
however, the form that this tax takes becomes much more flexible. While it
remains appropriate to tax the above-normal profits each year at accrual, it is
equivalent in present value simply to tax all repatriated profits. To avoid in the
process introducing a net tax on the return to savings invested in such firms,
the law could allow an immediate deduction for all funds sent abroad, which
compensates in present value for the taxes due on all funds generated from these
investments that are later repatriated. This difference between accrual taxation
and taxation at repatriation is equivalent to the difference between taxation
each year of wage income and the taxation on receipt of pension benefits.16
Current tax law does not allow an immediate deduction for funds sent abroad,
instead exempting from tax the eventual return of capital. This difference mat-
ters only to the degree that the earnings of the foreign subsidiary represents
a return to invested capital rather than a return to the ideas generating the
demand for the multinational’s product.
The argument that existing tax structures broadly correspond to the optimal
policies discussed above then rests on two premises. First that n ≈ 0 and second
that the foreign-source earnings of a multinational are only secondarily a return

to cross-border investments by the multinational.
To what degree does the existing tax law differ from what would exist with
n ≈ 0? Gordon et al. (2004b) examined in particular how much tax revenue
would have been lost in the US in 1995 if statutory tax rates were left unchanged
but the tax base were adjusted to eliminate any taxes on income from savings
and on capital investments in the US.17 The paper found only a slight fall in
tax revenue from this change, suggesting effectively no tax revenue on net was
being collected on the return to savings and investment in the US. Gordon et
al. (2004b) then show that this lack of revenue from taxes on the return to sav-
ings does indeed suggest under reasonable assumptions that savings incentives
on average are largely undistorted by the existing tax law, even though this is
certainly not the case for every individual and every form of investment.
3.3 Remaining puzzles, and possible explanations

Even if we can rationalize various features of actual tax laws by assuming that n ≈
0, we are left with at least three puzzles: first, why have some countries chosen
to exempt foreign-source income of their domestic multinationals from tax at
repatriation (a ‘territorial’ treatment)? Second, why do countries grant credits
for corporate and withholding taxes collected by host-country governments on
the income accruing to home country firms and home-country residents? And
third, why do host countries tax foreign subsidiaries at the same rate as applies
to domestic firms in the host country?
Of course, a quick answer is that OECD conventions require either a ‘territorial’
treatment or use of crediting. The question is then why these policy rules have
remained in force for so many years, an outcome that seems implausible if
countries view these rules as having significant costs.
Begin with the question why countries may adopt a ‘territorial’ treatment for
their domestic multinationals. Doing so exempts the foreign-source earnings of
their domestic entrepreneurs from tax, and enables these entrepreneurs to avoid
tax on their domestic-source activity through use of transfer pricing to make
these earnings appear to be foreign-source. Allowing domestic entrepreneurs to
avoid tax through shifting their earnings abroad creates a tax distortion favor-
ing entrepreneurial activity and favoring opening up a foreign subsidiary. Due
to information spillovers, economists commonly argue that there is too little
entrepreneurial activity. Whether this tax avoidance opportunity is the best
means of encouraging more entrepreneurial activity can easily be questioned,
but it does serve this role.18 There may be additional informational spillovers
generated when firms become multinationals, since these multinationals pick
up ideas abroad that then filter through to other domestic firms. Allowing tax
avoidance linked to having a foreign subsidiary encourages firms to become
multinationals. The US may have many other policies in use to encourage
304 R. Gordon
entrepreneurship compared with those in place in territorial countries, less-

ening the value of making use of this instrument to encourage entrepreneurial
activity.19
How can we best explain the granting of credits for taxes paid abroad
on income accruing to domestic residents? The key addition to the above
analysis is the assumption that all employees can potentially engage in income-
shifting between the corporate and the personal tax base, and not just the
entrepreneur.
In particular, assume that a firm’s domestic production function is f (K, L)
while its production function abroad is f ∗ (Ks∗ , L∗s ), where both production
functions have decreasing returns to scale.20
The objective function of the firm is to maximize the firm’s overall net-of-
tax profits. The foreign subsidiary is assumed to be a price-taker in the labor
market, and must provide each worker with a net-of-tax income equal to w∗n .
It can pay workers, though, either through taxable wages, w∗ , or through non-
wage compensation ω∗ , taxable at an effective capital-gains tax rate g ∗ . While
wage compensation is taxable to the individual and tax deductible to the firm,
we assume that non-wage compensation is not tax deductible.21 Total com-
pensation must be sufficient to provide workers a net-of-tax wage rate of w∗n :
(w∗ (1 − m∗ ) + ω∗ (1 − g ∗ )) = w∗n .
Use of non-wage compensation is assumed to come with a cost c(fL∗ − w∗ )L∗s ,
where c(.) is a convex non-negative function with a minimum value of zero
at w∗ = fL∗ . We assume as well that w∗ ≤ fL∗ : non-wage compensation must be
non-negative.22
To avoid the equivalent income-shifting by the employees of the parent firm,
we assume that the domestic government has set τ based on equation (4a).
The multinational will also charge its subsidiary royalty payments S for use
of the firm’s technology. Arm’s length pricing would give the parent firm the
full return to the technology it developed, leaving the subsidiary with zero net
profits (after deducting capital as well as labor expenses). The firm, though, can
choose the royalty payment to charge the subsidiary, but any deviation from
arm’s length pricing is assumed to generate real costs equal to σ (S∗ − S), where
σ (.) is a non-negative convex function with a minimum at zero costs when S
equals the arm’s length price S∗ . The choice for S can be at most the arm’s length
price, since any higher S leaves the subsidiary with negative profits.23
The after-tax income to the firm then equals
(1 − g e )[(1− τ )[f (K, L) − rK − wL + S − σ (S∗ − S)]

(6)
+(1 − τs ){(1 − τs∗ )[f ∗ (Ks∗ , L∗s ) − rKs∗ − w∗ L∗s − c(fL∗ − w∗ )L∗s − S] − ω∗ L∗s }]
Here, profits of the foreign subsidiary are subject to tax by both the host country
(τs∗ ) and the home country (τs ), with each rate potentially different from the
corporate tax rate that applies to purely domestic firms in each country. The host
country taxes the accruing income, whereas the home country taxes repatriated
profits.24 For convenience, let τsa ≡ τs∗ + (1 − τs∗ )τs .
The firm’s optimal choices for capital and labor satisfy the following first-order
condition:
fK − r = fK∗ − r = 0, and (7a)

(1 − τ )(fL − w) = (1 − τs∗ )(fL∗ − w∗ − c ∗ ) − ω∗ =0 (7b)
Since non-wage compensation must be non-negative, we conclude that fL∗ −

w∗ − c ∗ ≥ 0.
The first-order condition for the firm’s choice of royalty payment to charge
the subsidiary equals
τ − τsa
σ = (8)
1−τ
with S = S∗ when τsa ≥ τ .
Consider next the firm’s optimal choices for taxable compensation w∗ . The
first-order condition here is
(1 − τs∗ )(1 − g ∗ ) − (1 − m∗ )
c = . (9)
(1 − τs∗ )(1 − g ∗ )
Tax distortions to forms of compensation exist to the extent that the tax term
is positive. (If the tax term is negative, then all compensation takes the form of
wage payments.)
Consider the Nash equilibrium policy choices of each government, taking as
given the policy choices of the other government. Our focus is on the choices
for τs by the domestic government and for τs∗ by the foreign government.
The domestic government is assumed to choose its tax rates to maximize
the sum of the after-tax worldwide profits of domestic multinationals plus tax
revenue: workers and capital owners get the going returns, regardless of these
choices. The domestic government is then choosing its tax rates to maximize:
f (K, L) − rK − wL − σ (S∗ − S)
(10)
+(1 − τs∗ )[f ∗ (Ks∗ , L∗s ) − rKs∗ − c(fL∗ − w∗ )L∗s ] − ω∗ L∗s + τs∗ S
Domestic tax rates enter implicitly in this equation through their impact on the
value of S chosen by domestic multinationals.
The first-order condition for τs is simply:
∂S
(σ + τs∗ ) =0 (11)
∂τs
Through use of τs , the domestic government gains from discouraging domestic
multinationals from shifting their profits abroad. Income-shifting is eliminated
306 R. Gordon
when τsa ≥ τ . At such rates, there is no longer any income-shifting from the
parent firm, reducing the subsidiary’s taxable profits to zero.
What would be the objective of the host country, given this policy choice
by the home country? Certainly it gains from extra tax revenue. Since workers
in the subsidiary are simply paid their opportunity cost, they break even by
working for the subsidiary. Any decreased demand for labor by the subsidiary,
though, in principle causes a fall in the equilibrium wn and/or a fall in L∗s . For
simplicity, we assume that purely domestic firms have constant returns to scale
and are price-takers in the international market, so that their labor demand will
expand to ensure full employment at the original wage rate faced by domestic
firms of w∗n /(1 − m∗ ). With an unchanging net wage rate, aggregate labor supply
L∗ is unchanged and overall labor demand is unchanged. Any changes in labor
demand by the subsidiary are simply offset by changes in labor demand by
domestic firms. With unchanging factor and output prices for residents in the
country, the host-country government is choosing τs∗ to maximize tax revenue,
of which the relevant components are :
m∗ w∗n ∗
τs∗ [f ∗ (Ks∗ , L∗s ) − rKs∗ − (w∗ + c ∗ )L∗s − S∗ ] + (m∗ w∗ + g ∗ ω∗ )L∗s + (L − L∗s ) (12)
1 − m∗
The first-order condition for τs∗ takes the general form:
∂L∗ ∂w∗
∗ + A ∗ + B ∗ = 0, (13)
∂τs ∂τs
where ∗ is the corporate tax base. The third term is zero if τs∗ + g ∗ (1 − τs∗ ) = m∗ ,
since then there is no income-shifting by domestic workers: labor costs as a
result are fully deductible and c = ω∗ = 0. When labor costs are fully deductible,
∗ = 0. Finally, at these tax rates, A = 0 since labor income faces the same tax
rate m∗ in both sectors. At the optimum, we then infer that τs∗ + g ∗ (1 − τs∗ ) = m∗ .
Given this optimal host-country rate, to achieve τsa = τ we then infer that the
home country will set
τ − τs∗
τs = (14)
1 − τs∗
Remarkably, this optimal tax rate replicates the tax rate that arises under existing
OECD rules governing use of worldwide taxation. Under worldwide taxation,
domestic taxes are imposed on the foreign-source earnings needed to finance
observed repatriations. If repatriations equal R, domestic taxes are owed on
R/(1 − τs∗ ). A credit must then be given for the foreign taxes, τs∗ R/(1 − τs∗ ), paid
on this income. Under worldwide taxation, the tax rate τs on repatriated profits
then satisfies equation (14).25
Given that OECD rules replicate optimal policies, it is not surprising that they
have survived for as long as they have.
3.4 Remaining pressures and possible responses

This taxation at repatriation rather than at accrual leads to neutral incentives,
though only if the corporate and personal tax rates are constant over time. If
rates change, then the firm has an incentive to time its repatriations for years
in which the tax rate is low. Any fluctuations in a firm’s marginal tax rate then
generate tax distortions to the timing of repatriations. This opportunity to shift
taxable income over time to take advantage of fluctuations in marginal tax rates
introduces a tax advantage to investments abroad relative to investments at
home, and to entrepreneurial activity relative to wage and salary income. For
example, the one-year reduction in the corporate tax rate faced on profits repa-
triated to the US in 2005 created an ex-post reduction in the effective tax rate
on entrepreneurial income generated abroad in previous years, and increased
the ex-ante expectation of more such opportunities in the future. The resulting
incentive to postpone repatriations in the hopes of additional such holidays in
the future creates efficiency costs from this use of a tax at repatriation instead
of taxation each year on accruing income abroad.
Even if tax holidays are avoided, tax schedules inevitably change over time
due to broader changes in tax structures. The option to postpone repatriation
remains valuable, resulting in at least some reduction in the effective tax rate
on the returns to entrepreneurial ideas pursued abroad, or on domestic earnings
shifted abroad.
A second source of pressure arises from the tax incentive to shift the loca-
tion of the headquarters of a multinational. An implicit assumption in the
above argument is that a multinational’s headquarters is located in the coun-
try where the entrepreneur lives. This is a natural assumption, since the firm
would have originated there, and since the entrepreneur gains from having
the headquarters nearby. However, it remains feasible for the entrepreneur to
relocate the headquarters to another country. While the entrepreneur is still
subject to personal taxes on the capital gains on the firm’s shares, by shifting
the location of the headquarters to a country with a ‘territorial’ tax system, the
entrepreneur can avoid any supplementary corporate taxes on the firm’s foreign-
source income, can generate a net tax subsidy on past cross-border investments
to the extent that these investments qualified for prior tax deductions, and
can shift future domestic-source earnings abroad to low-tax countries with no
offsetting subsequent domestic corporate taxation.
What responses are available in the country of origin of a firm’s headquar-
ters operation, where presumably the entrepreneur resides? The gain to the firm
from shifting its headquarters is the avoidance of domestic corporate surtaxes on
all its earnings abroad, past and future. To neutralize at least part of this incen-
tive to shift the location of a firm’s headquarters, the domestic country could
require effective repatriation of the value of all operations abroad when this
shift in location occurs, implying immediate domestic corporate taxation on all
308 R. Gordon
past entrepreneurial earnings abroad. With this policy, the entrepreneur faces a
combined corporate plus capital gains tax rate equivalent to full personal tax-
ation on wage income for all past earnings abroad. However, the entrepreneur
faces only capital gains taxation on future earnings abroad, accruing following
the shift in the location of the headquarters. Even with effective repatriation of
past foreign-source earnings when a firm moves its headquarters, tax rules still
remain vulnerable.
What alternatives exist? One alternative would require that all securities
owned by an individual in firms in which the individual has personal links
(direct or indirect) be held in a registered account. With a registered account,
all funds are deductible when they are invested in the fund, and all withdrawals
are fully taxable. No tax is due within the fund as earnings accrue.
Under this tax treatment, the normal rate of return on savings held in the fund
is tax exempt, assuming the personal tax rate at withdrawal is the same as the
rate applying to deductions for the initial investments in the fund. However, any
above-normal return, in particular any return to entrepreneurial effort, would
be fully taxable. The same would be true when taxing employees who receive
equity rather than wage and salary compensation.
With this tax provision, entrepreneurial income is already taxed in full under
the personal tax. As a result, there would be no need for a corporate tax to correct
for any under-taxation of the return to entrepreneurship. Since the corporate
tax is not needed as a backstop for taxing the return to savings when n = 0,
the tax could be eliminated with this use of registered accounts for sources of
income where income-shifting is feasible.
4 Conclusions
How should multinationals be taxed? How are they taxed? To the extent that
the answers differ, how do we best explain the difference?
In this chapter, we first derived the optimal tax treatment of foreign
subsidiaries by both home and host-country governments, and found that
the home-country government should subject the profits of these foreign
subsidiaries to domestic corporate taxation at accrual, while the host-country
government would (under certain assumptions) exempt this income from tax.
In practice, some home countries do tax the foreign-source income of their
multinationals, but only at repatriation and with a credit for any taxes paid
abroad. In sharper contrast to the theoretical forecasts, most countries exempt
foreign-source income from domestic corporate taxes. In addition, host-country
governments normally tax these profits at the same rate that applies to their
domestic firms.
Part of the reason for these observed policies could be OECD guidelines. But
the question is then why these guidelines survive, if according to the theory
they are contrary to the policies that would be in each country’s interests.
We argue that the forecasted policies can be reconciled with observed poli-
cies if two conditions exist. First, countries do not aim to tax the income their
residents receive on their savings. Without a tax on the return to savings, defer-
ring the tax on the foreign-source earnings of domestic multinationals until
repatriation per se creates no problems.
The second condition is that income-shifting by employees of firms between
the corporate and the personal tax base is an important consideration in the
design of the tax law in the host country. Such income-shifting creates pres-
sures on the host country to tax the income of foreign subsidiaries located in
the country at the same rate that applies to domestic firms, to forestall such
income-shifting. Given such taxation in the host country, we find that the opti-
mal tax policy in the home country is to ensure that the combined host-country
and home-country tax rate on the income of the subsidiary is the same as applies
to the income of domestic firms in the home country: this tax rate is just suffi-
cient to discourage income-shifting by domestic multinationals. The resulting
optimal tax rate replicates OECD rules, with home-country taxes applying to
profits before host-country taxes, but with a credit against home-country taxes
for those taxes paid in the host country.
A remaining puzzle is why many countries choose not to tax the foreign-
source income of their domestic multinationals. Here, the best answer we could
come up with is that this favorable tax treatment leads to a lighter effective
tax rate on entrepreneurial income. Entrepreneurial income accruing through
foreign subsidiaries becomes exempt from domestic taxes while income-
shifting from the parent to foreign subsidiaries can exempt domestic-source
entrepreneurial income from tax as well. Perhaps the explanation for these poli-
cies is a desire to encourage entrepreneurship, and also to encourage domestic
firms to become multinationals, in order to pick up valuable ideas abroad that
then become available more broadly in the domestic economy. There are much
more cost-effective ways of encouraging entrepreneurial activity, however.
Notes
∗ This chapter was originally written for the 16th World Congress of the International
Economic Association, held in Beijing from 4 to 8 July 2011. I would like to thank Jay
Wilson for comments on an earlier draft. Part of this paper was written while visiting
CKGSB. I would like to thank CKGSB for its hospitality and financial support.
1. The latter condition is just a generalization of the result in Corlett and Hague (1953)
that commodity taxes should be used as a supplement to labor income taxes only to
the degree that goods vary in their cross-price elasticities with leisure.
2. However, see Gordon (2004) and Gordon and Kopczuk (2010) for examples where
particular portfolio decisions might well convey information about an individual’s
underlying ability, even given observed labor income.
3. If the depreciation rate allowed under the tax law does not vary with the individual’s
tax bracket, then the choice of rate must trade off offsetting distortions to portfolio
choice in different tax brackets.
310 R. Gordon
4. The key complication in introducing dividends is the need to include some explana-
tion for why dividends are paid, in spite of the tax disadvantage of doing so relative to
share repurchases. For a review of alternative theories, and the many inconsistencies
between the forecasts from these theories and stylized facts about firm behavior, see
Gordon and Dietz (2008).
5. We argued above that the capital gains rate should satisfy g = mz, implying g < m.
6. We assume, though, that only the entrepreneur is in a position to engage in such
income-shifting, and not other employees. We return to this issue below.
7. In particular, all payouts from the firm that are fully taxable under the personal tax
(such as wages, rents, royalties, and interest payments), or fully taxable to other firms
(such as payments for inputs), should be allowed as deductions from the corporate
tax base.
8. For example, Gordon and Slemrod (2000) find that reported corporate income
responds much more to the top personal tax rates than to tax rates in lower brackets.
9. Any deviations from neutrality open up arbitrage possibilities, with investors going
short in bonds with a high taxable interest rate and long in bonds with a low taxable
interest rate.
10. Home bias, though, may limit the resulting misallocations.
11. When a country is not a price-taker in the international capital market, it can make
use of the tax law to take advantage of its market power. One reason why a country
might not be a price-taker is that foreign investors are attracted to domestic securities
for portfolio diversification or hedging reasons. Given that the resulting market power
seems very small, however, optimal tax rates will be very small as well.
12. Evasion can still potentially occur, though, through routing savings first through a
country that hides the identity of the investor and then through a country with an
information-sharing treaty with the domestic government. Tax would then need to
be imposed whenever the country of residence of the underlying investor cannot be
verified.
13. Neutral taxation by the firm’s home country induces the firm to use arm’s length
pricing as long as there are any real costs of deviating from arm’s length pricing.
14. Even if the firm earns above-normal profits, it should still be a price-taker in the
market for capital, implying that the optimal tax rate on capital investments by the
subsidiary remains equal to zero.
15. Similarly, domestic portfolio investors can be taxed on their foreign-source earnings
with a credit for withholding taxes collected abroad.
16. With the tax based simply on funds leaving or entering the country, there is no need to
deal with the auditing or taxation of foreign corporations, or to document the foreign
portfolios of domestic investors, as long as cross-border payments can be identified
and traced to their recipient.
17. In particular, the paper examined the impact on tax revenue of replacing depreciation
deductions on any new investment with expensing, eliminating all taxes (both cor-
porate and personal) on interest, dividend, and capital gains income, and eliminating
any deductions (corporate and personal) for interest payments.
18. Gordon and Cullen (2006) argue in contrast that increasing the tax savings on business
losses is a much more effective means of encouraging entrepreneurship than reduc-
ing the tax rate on the most successful business outcomes, since given risk aversion
potential losses are more salient than the largest potential profits.
19. For example, venture capital funding is much better developed in the US. There are
fewer labor market restrictions that hinder entry and exit decisions, or hiring and
firing decisions. Bankruptcy rules in the US are more favorable to the debtor, as
emphasized in Fan and White (2003).
20. Implicit in this specification is the assumption that the foreign subsidiary can earn
these pre-tax profits only by locating in this particular host country. If the location
of these facilities is more flexible, then there are additional pressures to keep host-
country tax rates low.
21. An example of such compensation in the US is incentive stock options in the firm,
which by statute receive this tax treatment. Another example would be equity com-
pensation in a closely held firm. By statute, workers are taxed on the market value of
this compensation, and the firm can take this market value as a tax deduction. How-
ever, when the firm is closely held, the firm has great discretion in asserting a market
value of this compensation for tax purposes. If τ < m, the firm has the incentive to
claim that the shares have no value. Other types of compensation are untaxable to
the worker but still deductible expenses for the firm, such as a fancy office. Qualita-
tively, results will be the same if the firm’s deductions are not lost with the alternative
compensation.
22. In equilibrium, any higher deduction for wage payments will leave the firm with
negative taxable income, eliminating any potential tax savings through the larger
deduction.
23. We assume here that the host country cannot obstruct these royalty deductions,
contrary to the discussion above.
24. With a tax on repatriated funds, non-deductible labor expenses are still a cost of
business, thereby reducing the amount of repatriated profits. Rather than giving the
parent firm an immediate deduction for capital invested in the subsidiary and then
a full tax on all repatriated earnings from the investment, for convenience in the
analysis we allow the opportunity cost of the capital as a deduction each year: both
approaches yield no net tax on the return to capital invested in the subsidiary.
25. In fact, the theory forecasts that foreign subsidiaries will have no net profits to be
repatriated, consistent with the evidence in Hines and Hubbard (1990). Receipts of
foreign-source earnings will instead take the form of royalty payments.
References
Arulampalam, W., G. Maffini and M. P. Devereux (2007) ‘The Incidence of Corporate

Income Tax on Wages’, Oxford University Centre for Business Taxation Working Paper
07/07.
Corlett, W. J. and D. C. Hague (1953) ‘Complementarity and the Excess Burden of
Taxation’, Review of Economic Studies, vol. 21, no. 1, pp. 21–30.
Cullen, J. and R. Gordon (2006) ‘Tax Reform and Entrepreneurial Activity’, Tax Policy and
the Economy, vol. 20, pp. 41–72.
Desai, M. and J. Hines (2003) ‘Evaluating International Tax Reform’, National Tax Journal,
vol. 56, no. 3, pp. 487–502.
Diamond, P. and J. Mirrlees (1971) ‘Optimal Taxation and Public Production I: Production
Efficiency’, American Economic Review, vol. 61, no. 1, pp. 8–27.
Fan, W. and M. J. White (2003) ‘Personal Bankruptcy and the Level of Entrepreneurial
Activity’, Journal of Law and Economics, vol. 46, no. 2, pp. 545–567.
Gordon, R. (2004) ‘Taxation of Interest Income’, International Tax and Public Finance, vol.
11, no. 1, pp. 5–15.
312 R. Gordon
Gordon, R. and M. Dietz (2008) ‘Dividends and Taxes’, in A. Auerbach and D. Shaviro (eds),
Institutional Foundations of Public Finance: Economic and Legal Perspectives (Cambridge,
MA: Harvard University Press), pp. 204–224.
Gordon, R., L. Kalambokidis and J. Slemrod (2004a) ‘A New Summary Measure of the
Effective Tax Rate on Investment’, in Peter Birch Sörensen (ed.), Measuring the Tax Burden
on Capital and Labor (Cambridge, MA: MIT Press), pp. 99–128.
Gordon, R., L. Kalambokidis and J. Slemrod (2004b) ‘Do We Now Collect Any Revenue
from Taxing Capital Income?’, Journal of Public Economics, vol. 88, no. 5, pp. 981–1009.
Gordon, R. and W. Kopczuk (2010) ‘The Choice of Personal Income Tax Base’, mimeo.
Gordon, R. and J. Slemrod (2000) ‘Are “Real” Responses to Taxes Simply Income Shifting
Between Corporate and Personal Tax Bases?’, in Joel Slemrod (ed.), Does Atlas Shrug?
The Economic Consequences of Taxing the Rich (New York: Russell Sage Foundation), pp.
240-80.
Grubert, H. (1998) ‘Taxes and the Division of Foreign Operating Income among Royalties,
Interest, Dividends, and Retained Earnings’, Journal of Public Economics, vol. 68, no. 1,
pp. 269–290.
Hines, J. R., Jr and R. G. Hubbard (1990) ‘Coming Home to America: Dividend Repatri-
ations by US Multinationals’, in A. Razin and J. Slemrod (eds), Taxation in the Global
Economy (Chicago: University of Chicago Press), pp. 161–208.
Mathur, A. and K. Hassett (2006) ‘Taxes and Wages’, American Enterprise Institute Working
Paper # 138.
Saez, E. (2002) ‘The Desirability of Commodity Taxation under Non-linear Income Tax-
ation and Heterogeneous Tastes’, Journal of Public Economics, vol. 83, no. 2, pp.
217–230.
16
Taxing Multinationals in a World
with International Mergers and
Acquisitions: Should the Home
Country Exempt Foreign Income?∗
John Douglas Wilson
Michigan State University, USA
1 Introduction
The taxation of foreign-source income has come under increasing attack by aca-
demic researchers and policy-makers. According to the traditional view, national
welfare maximization requires that a capital-exporting country tax income from
capital invested at home and abroad at the same rate, with a deduction for
taxes paid to foreign governments. The basic idea is that a lower tax rate on
foreign-source income will cause an inefficiently large outflow of investment,
as domestic capital owners seek to escape the higher tax rate at home.1 But Desai
and Hines (2003, 2004) have argued that this close relation between investment
at home and investment abroad is not observed, and that foreign-source income
should not be taxed because the extra tax borne by domestic investors distorts
their decisions to buy and sell foreign companies. Their view of the world is
described by the following passage:
… modern scholars view FDI as arising from differential capabilities, and
consequently differential productivity, among firms, and the extension of
intangible assets across borders. This intuition squares well with empirical
FDI patterns, which include the fact that most of the world’s FDI represents
investment from one high-income country into another, and the fact that
a very high fraction of such investment takes the form of acquiring exist-
ing businesses. Consequently, most FDI represents transfers of control and
ownership, and need not involve transfers of net savings. (2004: 956)
In other words, whereas the focus of the traditional view has been on ‘capital
export neutrality’, the new view elevates ‘ownership neutrality’ to center stage,
and argues that foreign-source income should be exempt from taxes levied by
the home government.
313
314 J. D. Wilson
In a recent paper, Becker and Fuest (2010) present a formal model intended
to capture this view of the world. In particular, only M&A investment is con-
sidered, and multinationals are free to invest at home or abroad, constrained
only by a limited number of target firms for which an acquisition results in
an adequate productivity improvement. Though these acquisitions are funded
by selling equity to domestic investors, the availability of funds is effectively
unlimited, because these investors can borrow and lend at an internationally
determined interest rate. Thus additional FDI need not reduce investment at
home. Using this model, Becker and Fuest show that the only fully efficient
tax system is one where FDI is not taxed. They also show that the desirability
of an exemption system goes away if the number of acquisitions available to
a multinational is limited, representing ‘managerial constraints’. But the latter
model seems to violate the Desai-Hines view that FDI arises from ‘… the exten-
sion of intangible assets across borders’, given that such an extension can be
accomplished at little or no cost.
While most international tax specialists would now agree that taxing FDI
creates a host of deadweight losses, it is important to recognize that most admin-
istratively feasible tax instruments distort economic decision-making in some
way. The issue then is whether the distortions created by taxing foreign-source
income are so large that such taxes should be abandoned, or whether there
remains an important role for such taxes at part of a country’s optimal tax
system. The Becker-Fuest framework is not suitable for addressing this issue,
because taxes on investment are not needed to raise revenue in their model;
there is effectively a distortion-free tax instrument for meeting any government
revenue needs.
In this chapter, I start with a rather general model of investment at home
and abroad by a ‘small’ home country, including both M&A investment and
greenfield investment. The home country’s tax instruments include taxes on
the domestic and foreign income earned by home firms, income from portfo-
lio investment, and wage income. The model is formulated to emphasize the
issue of whether the tax system violates the ‘ownership neutrality’ condition
emphasized by Desai and Hines. But it is sufficiently general to encompass both
the traditional and Desai-Hines views as special cases (or the two Becker-Fuest
models).
Using this model, I argue that there remains an important role for taxing for-
eign income generated by multinationals. Following the current US practice, the
model assumes that these taxes are collected on repatriated income. The model
recognizes opportunities available to multinationals for deducting the costs of
their foreign investments, in which case the tax system effectively lowers these
costs and thereby reduces or eliminates the disincentive effects of taxes. But the
analysis also emphasizes the potential impacts that foreign subsidiaries have
on the productivities of the multinational’s subsidiaries located at home. For
Taxing Multinationals in a World with International Mergers and Acquisitions 315
a simplified version of the model, taxing foreign income leads to productivity

improvements in the home country, which supports the case for relatively high
taxes on repatriated foreign income.
The next section describes the model, and Section 3 shows how taxes affect
the equilibrium conditions. Section 4 contains a key result about the impact
of a repatriation tax on investment abroad. The model is then specialized in
Section 5, allowing for a more specific result about how foreign-source income
should be taxed. The income-shifting activities of multinationals are briefly
discussed in Section 6. Section 7 concludes.
2 The model
Consider a small open economy, called ‘home’, which is initially populated by a

fixed number of domestic firms. The economy lasts two periods, and a represen-
tative consumer possesses the utility function, u(c1 , c2 , L), where c1 and c2 are
consumption in periods 1 and 2, and L is labor, which is supplied in period 2.
This consumer starts period 1 with an endowment of income, I, plus ownership
of a fixed number of ‘domestic firms’ and a fixed number of ‘multinationals’. In
the first period, the consumer divides income between consumption and saving.
Saving includes purchases of equity or debt from domestic firms and the multi-
nationals. In addition, the consumer may borrow and lend at an internationally
determined interest rate, r, which I refer to as portfolio investment. So far, the
model is similar to that of Becker and Fuest (2010), except for the labor-leisure
decision. The after-tax return on portfolio investment will uniquely determine
total saving if we follow them by assuming away income effects, but this is not
necessary for the main insights
Becker and Fuest model domestic and foreign firms as generating exogenous
amounts of income from their ownership of assets. Multinationals purchase
these firms because they can increase their income flows. I also assume that firms
posses ‘firm-specific assets’, but I then treat capital and labor as separate inputs
that are chosen to maximize profits. Capital is internationally mobile, whereas
the home labor is supplied by the representative consumer. Labor abroad is
available at a fixed foreign wage.
Domestic firms use a constant-returns technology to produce output in the
second period from the three factors. The technology is represented by the pro-
duction function relating output net of the capital investment, f D (K D , LD ), to
the inputs of capital, K D , and labor, LD . The firm-specific assets are suppressed
as an argument because they are fixed. Decreasing returns prevail in capital and
labor alone.2
The multinationals invest at home or abroad. I normalize the number of
multinationals to equal one, but the ‘representative multinational’ is treated
as a price-taker. It possesses its own firm-specific assets, but engages in M&A
316 J. D. Wilson
investment and greenfield investment. M&A investment consists of purchas-

ing existing domestic and foreign firms.3 The purchase of an existing firm
provides the multinational with the firm’s firm-specific assets, and the multi-
national then uses its own existing firm-specific assets to achieve productivity
improvements.4 After these purchases occur, the multinational and remaining
independent domestic firms choose their capital investment levels. All invest-
ment occurs in period 1. In period 2, labor is hired and output is produced and
consumed.
Greenfield investment is done by directly investing in capital and combining
it with labor and the multinational’s firm-specific assets, rather than combining
these assets with those of another firm.
A major issue is the extent to which the multinational’s firm-specific assets
may be shared among subsidiaries without diminishing their effectiveness. In
Becker and Fuest’s first model of M&A investment, only a fixed number of firms
may be purchased, because of the constraints imposed by ‘managerial capacity’.
In their second model, ownership advantage is a ‘public good’, so any number of
firms may be purchased, though the number is limited by a scarcity of attractive
target firms. In the current chapter, I consider a single model that encompasses
these two cases and those in between. In particular, let N ij denote the number
of subsidiaries obtained in country i(i = H, F) through type-j investment (j = A
for M&A; j = G for greenfield), and let the non-superscripted symbol, N, denote
the vector, (N HA , N FA , N HG , N FG ). For mathematical simplicity, N ij is treated as
a continuous variable, but the analysis could be redone with discrete numbers
of subsidiaries by stating the relevant optimality conditions using inequalities,
rather than equalities, at the cost of added complexity. The total output pro-
duced by the N ij subsidiaries is a denoted f ij (K ij , Lij , N), where the arguments
K ij and Lij denote the total capital and labor used by the N ij subsidiaries. This
function exhibits decreasing returns to scale in capital and labor. Throughout
this chapter, these output functions are defined as output net of the original
capital investment.
The additional output produced by adding another type-j subsidiary in coun-
try i (j = A or G) is obtained by taking the total derivative of f ij (K ij , Lij , N) with
respect to N ij , including the accompanying changes in K ij and Lij . Note that
these changes in total factor demands do not necessarily represent the factor
usage by the marginal subsidiary, because the addition of another subsidiary
may cause other subsidiaries in the same country to alter their factor usage, in
part because the number of subsidiaries sharing the multinational’s firm-specific
assets has increased. Because of this sharing aspect, we expect increases in N ij to
have a diminishing effect on total (i, j)-output. The partial derivative, ∂f ij /∂N ij ,
which gives the change in total (i, j)-output from another (i, j)-subsidiary, hold-
ing fixed total factor usage, can be expected to be positive at low levels of N ij ,
since re-allocating labor among a larger number of subsidiaries means that the
initial subsidiaries have less labor and capital and therefore higher marginal
products. But eventually this derivative will drop below zero if the marginal
subsidiary becomes sufficiently unproductive. On the other hand, it could stay
positive if there exists agglomeration economies, where adding another sub-
sidiary to a country increases the productivities of those already there. But then
other forces would need to limit the equilibrium number of subsidiaries, such as
their negative impact on subsidiary productivity in the other country. To allow
for different cases, I will generally leave open the impact of N on output, except
to assume this impact is sufficiently well-behaved not to create problems for
the existence of an equilibrium. I consider only the case where not all domestic
firms are purchased.
Different assumptions about the ownership advantage are reflected by the
dependence of f ij on N. The case where the ownership advantage is a public
good is captured by assuming that f ij does not depend on the number of firms
not in country i or not type j. This case is consistent with the view emphasized by
Desai and Hines (2003, 2004) that investment in one location does not reduce
investment in other locations. For the case where managerial capacity fixes the
number of acquisitions, we could assume that N HA + N FA is exogenously fixed, or
we could assume that managerial capacity fixes the total number of subsidiaries,
both greenfield and M&A. Although trade in intermediate inputs associated
with vertical FDI is not explicitly modeled, the benefits of vertical FDI could be
captured by assuming that f ij increases with N kA or N kG , k = i.
In the first period, the multinational first chooses which firms to purchase,
and how many subsidiaries in each location to create through greenfield invest-
ment. Then all firms choose their capital investment levels. To finance their
investments, firms may issue shares or sell debt. In the absence of income effects
in the saving decision, these investments do not alter the home residents’ total
saving, since it is determined by the after-tax return on portfolio investment.
Rather, equity investments merely create an offsetting reduction in the resi-
dents’ portfolio investment. Note too that the purchase of a domestic firm does
not represent a net equity investment, but is rather a transfer of assets among
home residents.
The tax instruments available to the home government are a wage tax, a tax on
residents’ interest income from portfolio investments, denoted τ , a tax on the
taxable domestic profits earned by domestic firms and the multinational, t H , and
a tax on foreign income remitted by the multinational to the home country, t F .
The assumption that only remitted foreign income is taxed is retained to follow
US practice and emphasize the tax breaks available to home subsidiaries located
abroad under this system. I sometimes refer to this tax as a ‘remittance tax’.
Taxes on dividends are omitted, because they do not add additional insights.5
Note finally that taxes paid to foreign governments are treated as tax deductible,
since the traditional analysis shows a deduction system to be optimal. Such taxes
318 J. D. Wilson
do not explicitly appear in the algebraic treatment of the model, but they can
be included by defining the foreign production functions and cost terms as net
of these taxes.
Assuming a standard corporate income tax, where debt is tax deductible but
equity is not, the cost of capital will depend on the method of finance. The
required return on equity is (1 − τ )r, which is the return investors could obtain
in portfolio investment. Thus, (1 − τ )r is the cost of funds in the case of equity
investment. In contrast, the required return on debt is r, but the tax deduction
implies a cost of funds equal to (1 − t)r for a firm facing corporate tax rate t.
However, there are well-known agency costs associated with debt financing,
plus real costs associated with default. As a result, firms use a combination of
debt and equity financing. Since the current chapter is concerned primarily with
whether the income from foreign direct investment should be taxed, we will
treat the taxes on domestic sources of income, along with deductibility rules, as
fixed. In addition, the tax τ on all portfolio investments is treated as exogenous.
I then assume that the cost of funds on domestic investments can be written
(1 − bH )r where bH < t H , reflecting less than full deductibility of capital costs.6
For the subsequent analysis, the critical implication is that there is inefficient
underinvestment at home.
In the case of FDI, a common practice is to use foreign debt as the source of
funding, suggesting that the net costs of foreign debt are relatively low.7 Also not
captured in this 2-period model are the benefits of deferral under a repatriation
system, whereby tax payments can be delayed by not repatriating the income.
In light of the tax benefits inherent in a repatriation system I will emphasize the
special case of a ‘corner solution’ in the financing problem, where investment
costs are completely financed using local debt, so that the cost of funds is
(1 − bF )r = (1 − t F )r. (1)
This equality is referred to throughout the chapter as ‘full deductibility’ of

financing costs, recognizing that it includes equity financing when t F = τ .
After-tax profits for a type-(i, j) subsidiary are
MaxK ij ,Lij (1 − t i )[f ij (K ij , Lij , N) − wi Lij ] − (1 − bi )rK ij , (2)
where wi is the wage rate in country i. The analysis abstracts from differences
between foreign countries and so does not consider tax planning activities
designed to take advantage of tax differences between different foreign coun-
tries. Later, I discuss tax planning involving income-shifting between the home
country and abroad as a whole, but for now I assume that the taxable profits
in country i are given by output minus wage costs and tax-deductible costs
associated with capital investments in that country. When income-shifting
is introduced, there will be a role for the multinational’s parent company in
charging subsidiaries royalty payments for use of the firm-specific assets.
3 Equilibrium conditions
This section describes how taxes affect the equilibrium conditions. First, the
conditions for capital and labor usage are standard:
ij
(1 − t i )fK = (1 − bi )r; (3)
(1 − t H )fKD = (1 − bH )r; (4)
ij
fL = wi ; (5)
fLD =w .D
(6)
For greenfield investment, subsidiaries are set up to the point where the
additional after-tax value of the output from another firm equals zero:

(1 − t H )fiG
Hl
+ (1 − t F )fiG
Fl
= 0, (7)
l
where the subscript ij denotes a derivative with respect to N ij . Note here that
we can ignore any increases in labor and capital usage associated with another
firm, because profit maximization tells us that the cost of these inputs equals
the output obtained from them at the margin. This is an envelope theorem
argument. At the margin, the firm is indifferent between employing more labor
and capital, or undertaking production in the new subsidiary by drawing labor
and capital away from other subsidiaries.8 In other words, we could amend (7)
by adding the revenue generated by new capital, but then subtracting the cost
of the new capital, but these terms would cancel out. If greenfield investment
does lead to greater capital usage, then a rise in the cost of funds will discourage
this form of investment.9
A critical insight from (7) is that the number of subsidiaries created in country
i depends not only on the tax there, but also on the tax in the other country k,
provided subsidiaries in i affect productivity in k. Neutrality requires not only
the full deductibility of capital costs (i.e., bH = t H ), but also t H = t F .
Next, consider the condition for equilibrium purchases of domestic firms.
Competition drives the sales price of these firms down to the original owners’
reservation prices, P H , which is discounted future profits, minus any tax on the
revenue of the sale. Recognizing the taxation of capital gains under standard
corporate tax rules, let us treat P H as a capital gain and assume that it is taxed.10
Then P H is given by

1 − t H f D − wH LD − 1 − bH rK D
PH = (8)
1 − t H (1 + (1 − τ )r)
where the after-tax return on portfolio investment, (1 − τ )r, represents the dis-
count rate. In particular, the numerator is the future income that the original
320 J. D. Wilson
owners can obtain by not selling, whereas P H (1 − t H )(1 + (1 − τ )r) is the future
income that they can obtain by selling, paying the capital gains tax, and using
the proceeds for portfolio investments. The multinational is willing to buy the
firm if the discounted future profits are at least as high as P H . But with the seller
paying a capital gains tax, the buyer gets to deduct t H pH as a capital loss at the
end of period 2, since the firm has become worthless in this 2-period model.
Taking into account this deduction, the buyer equates the discounted future
profits from the purchase of the firm to P H . This condition may be written

l 1−t
H f Hl + 1 − t F f Fl
HA HA 1 − t H f D − wH LD − 1 − bH rK D
= (9)
1 − t H + 1 − bH r 1 − t H (1 + (1 − τ )r)
where the discount rate for the multinational buyer is the cost of finance,
(1 − bH )r. In particular, the multinational equates the after-tax income gen-

erated by the purchase, l [(1 − t H )fHA Hl + (1 − t F )f Fl ], to the price and financing
HA
costs, net of the depreciation deduction: (1 − t H + (1 − bH )r)P H .
It is clear that the tax system will generally distort the multinational’s deci-
sion to purchase domestic firms if t H = t F or financing costs are less than fully
deductible, as it does with greenfield investment. But now a distortion remains
even if if t H = t F and there is full deductibility. In this case (9) becomes
Hl
D
l fHA + fHA
Fl
f − wH LD − rK D
= (10)
1+r 1 + (1 − τ )r
Acquisitions are discouraged by the tax on portfolio investment because this tax
lowers the return that the seller receives on the income obtained from selling
the firm, whereas the multinational is willing to purchase the marginal firm if
it receives the return r on the purchase price. This is an important difference
between greenfield and M&A investment. For greenfield investment, efficiency
requires that the before-tax return on capital investment equals the interna-
tionally determined return r. For M&A investment, efficiency requires that the
acquirer receive the same return that the domestic seller gets on the proceeds
from the sale, which is the after-tax return (1 − τ )r. But the acquirer receives the
same return r on both capital investments and acquisition costs, so both con-
ditions cannot be met. Efficiency requires that the tax on portfolio investment
be eliminated, so both portfolio investment and M&A investment generate the
same return, r.
This inefficiency could also be eliminated if we replaced the tax deduction
for debt and introduced expensing of all investment costs, meaning these costs
could be deducted at the time they are incurred. With expensing, the capital cost
term 1 − bH in the numerators of the right side of (9) would become 1 − t H , but
the discount rate used by the acquirer would become (1 − τ )r, since there would
now be an incentive to finance investments with equity. With equal discount
rates for the acquirer and target firm, M&A investment would be efficient. But
the topic of this chapter is whether it is beneficial to tax foreign income, given a
system of domestic taxes that roughly conforms to the features of the US system.
Although the US system allows for accelerated depreciation, it does not typically
go all the way to expensing.
Turning to foreign M&A investments, the price of a firm is exogenously set
at P F from the viewpoint of the home country. As in the case of domestic M&A
investments, the acquirer will receive a capital-loss deduction with a tax saving
of t F P F in period 2. Thus, second period costs are [(1 − t F ) + (1 − bF )r]P F , so firms
are purchased to the point where

(1 − t H )fFA
Hl
+ (1 − t F )fFA
Fl
= [(1 − t F ) + (1 − bF )r]P F (11)
l
The tax is now neutral if t H = t F and financing costs are fully tax deductible,
i.e., bF = t F .
Actually, a common practice is to use foreign debt to fund foreign acquisitions.
In this case, a cash outflow of (1+ r)P F is incurred in the second period, reducing
second-period remittances to the home country by (1 + r)P F . Thus, there is no
difference between local funding of foreign acquisitions and funding at home,
except to the extent that the decision impacts the cost of funds. I retain this tax
treatment of P F throughout the chapter.11
To conclude, we are so far not seeing a major argument for why the prevalence
of foreign M&A investments justifies taxing foreign-source income particularly
lightly.
4 Investment linkages
Desai and Hines (2003, 2004) emphasize the lack of a negative relation between
investment at home and investment abroad, and suggest that there may even
be a positive relation, with foreign acquisitions leading to more investment at
home.12 In the current model, such interactions are captured by a positive value
of the derivative of home output with respect to the number of firms acquired
Hl > 0. Then (11) shows that if firms are able to take full advantage of
abroad, fFA
the deductibility of local debt, increasing the foreign tax t F above t H actually
increases foreign acquisitions, because all of the costs of these acquisitions are
deductible, but the higher t F does not apply to all of the resulting increases in
output, which includes additional output at home. I summarize as follows:
Proposition 1 Assume that the financing costs for FDI are fully tax deductible (i.e.,
bF = t F ). Then increasing the repatriation tax increases (reduces) foreign M&A invest-
ment if fFA Hl > (<)0. For foreign M&A investment to be efficient with f Hl = 0, all
FA
corporate income must face the same tax rate: t H = t F .
322 J. D. Wilson
This result emphasizes the importance of the linkages between additional

acquisitions abroad and the productivity of home subsidiaries. If firms take
advantage of the generous provisions associated with a repatriation system,
then the tax system has the potential to reduce after-tax costs so much that
M&A investment is actually encouraged.
5 International linkages and optimal FDI taxation
Desai and Hines (2004) write:
The national welfare considerations that form the basis of NON [national
ownership neutrality] suggest, much as is evident in practice, that countries
should want to exempt foreign income from taxation. This policy prescrip-
tion stems from the observation that outbound foreign investment need
not be accompanied by reduced domestic investment in a world of shifting
ownership patterns.
But we have seen that taxing foreign-source income can raise M&A investment
abroad in cases where it improves productivity at home, presumably leading
to more investment at home (Proposition 1). In this section, I use a simplified
version of the model to show that raising the tax on foreign-source income leads
to a welfare-improving rise in investment at home, regardless of whether home
productivity is positively or negatively related to the number of subsidiaries operating
abroad.
The assumptions of the simplified model are as follows:
Model S
a) The multinational has a fixed number of home subsidiaries, and subsidiaries abroad
are created only through M&A investment, using fully tax-deductible financing.
b) Production functions for all firms at home are homothetic in labor and capital, and
total output for subsidiaries at home takes the form g(N F )h(K HS , LHS ), where N F
is the number of foreign subsidiaries.
c) At equilibrium prices, home subsidiaries have higher capital-labor ratios than the
independent domestic firms.
d) There are no income effects in labor supply.
I now prove:
Proposition 2 For model S, an increase in t F increases total investment in the home

country, holding fixed the other tax rates, if an increase in the number of subsidiaries
abroad has a non-zero effect on the productivity of subsidiaries at home, regardless of
the sign of this effect.
Proof Assume first that the derivative of home subsidiary outputs with respect
to the number of foreign subsidies, fFH , is positive. The condition for the
equilibrium number of foreign subsidiaries is
(1 − t F )fFH + (1 − t F )fFF − (1 − t F )(1 + r)P F = 0 (12)
where the subscript F denotes a derivative with respect to N F . An increase in t F

has no effect on labor and capital usage abroad, given N F , since there is full cost
deductibility. But with fFH > 0, the rise in t F turns the left side of (12) positive.
By second-order conditions, equilibrium is restored by increasing N F .
In the home country, the productivity improvement increases the total
demand for labor, causing the before- and after-tax wage rates to rise (given
the fixed tax rate). As a result, the supply of labor expands. In addition, the
productivity improvement increases the share of labor used by subsidiaries,
which have the relatively high capital-labor ratio (by assumption c) of Model
S). Holding fixed all firms’ capital-labor ratios, these effects increase total capital
investment in the home country. But the capital-labor ratios all rise, because
the higher wage rate implies that capital has become cheaper (using the homo-
theticity assumption, and holding fixed t H ). Combining these effects, we may
conclude that total investment rises in the home country.
Suppose now that fFH is negative. Then (12) implies that a rise in t F lowers
F
N , but this change now raises productivity in the home country. Repeating the
previous arguments, we may conclude that investment expands. Q.E.D.
This result tells us that raising the repatriation tax leads to efficiency improve-
ments at home, if investment is initially distorted by less than fully tax-
deductible financing. What is surprising here is that these efficiency improve-
ments arise regardless of whether greater M&A investment abroad directly raises
or lowers productivity at home. If productivity rises, then we see that the
rise in t F leads to greater M&A investment, which occurs because investment
costs abroad are tax deductible. In this case, the number of subsidiaries rises
abroad, leading to higher productivity at home and desirable increases in home
investment. But if more foreign subsidiaries lowers home productivity, then the
rise in t F lowers M&A investment abroad, which also enhances productivity
at home.
This investment benefit of a higher repatriation tax clearly improves welfare.13
In fact, it strengthens the case for taxing foreign-source income at a higher
rate than the tax on domestic firms. If we start with t F = t H and optimize
the other taxes, then raising t F above t H will actually have a negative dead-
weight loss if investment at home is initially distorted, because it will offset this
distortion.
Of course, the model omits features that work against this result. With less
than full deductibility of investment costs for FDI, we would need to trade off
the distorting effects of a repatriation tax on investment abroad against the
potentially beneficial effects of this tax on investment at home. While we might
324 J. D. Wilson
no longer find that t F exceeds t H at the optimum, this would still be the case
if there were sufficient, though not full, deductibility of financing costs. The
case for some taxation of repatriations seems strong under the generous tax
deductions afforded firms under the repatriation system.
6 Income-shifting
A popular tax-avoidance activity is the use of transfer prices to shift profits from
high-tax locations to low-tax locations. In fact, the Desai-Hines view that FDI
arises ‘from differential capabilities, and consequently differential productivity,
among firms, and the extension of intangible assets across borders’ suggests
that tax avoidance through transfer price manipulation is rather easy, given the
implied public good nature of the source of profits.
Assume that foreign governments tax the profits of the home multinational’s
foreign subsidiaries. The home government applies a repatriation tax at the same
rate as the tax levied on firms operating in the home country, with a deduction
for foreign taxes paid. To avoid the combined tax burden of the home and
foreign taxes, the parent company of the home multinational can charge royalty
fees for the profit-generating services provided to the subsidiaries, reducing their
taxable income to zero. These fees are then taxed at the home country rate alone,
implying that all income from the firm-specific assets is taxed at the single home
rate. But then only the foreign government is hurt by this income-shifting,
because home-country tax payments do not change. If the repatriation tax were
higher than the tax on firms in the home country, then the difference would
be eliminated by this income-shifting activity, which would hurt the home
country if the repatriation tax was initially optimized under the assumption of
no income shifting. If it were lower but not so low to make income-shifting
unprofitable, then income-shifting would increase tax payments of the home
government at the expense of the foreign government.
Thus, income-shifting limits the ability of the home government to tax
foreign-source income more heavily or somewhat more lightly that domestic
income, but it does not appear to fundamentally alter the arguments in favor
of a repatriation tax.
On the other hand, shifting from a deduction system to a system of tax credits
for taxes paid to foreign governments will alter incentives to shift income, and in
some cases, can justify the use of tax credits. See Gordon (2012) for an argument
along these lines.
7 Concluding remarks
Recent arguments for exempting foreign income from taxation argue that a
large portion of FDI involves M&A investment. Since this type of investment
does not involve a reallocation of capital from home to abroad, tax rate dif-
ferences between domestic and foreign income are no longer distortionary, and
foreign income should be exempted from taxation to avoid distorting ownership
patterns.
This chapter has questioned this reasoning by constructing a model that seems
roughly consistent with the frameworks used by the exemption proponents. In
fact, investment abroad may even lead to more investment at home. But this
positive linkage can actually strengthen the case for sizable taxes on foreign
income. If a firm recognizes that its foreign investment will help its domestic
operations, then this added benefit is subject to the domestic tax, not the foreign
tax, and so raising the foreign tax does not necessarily reduce investment abroad,
and will actually increase it when investment costs are tax deductible.
Along with emphasizing investment linkages, this chapter has emphasized
that foreign taxes lower the costs of investing abroad, given the tax deduc-
tions available to firms operating abroad. From an optimal tax perspective,
these deductions should be chosen to turn the tax on foreign income into a
tax on economic rents. Devereux (2008: 716) observes that taxing only eco-
nomic rents would not achieve production efficiency in a world where a firm
is choosing between different locations for discrete investments, unless statu-
tory tax rates were completely harmonized. But from the viewpoint of home
country welfare, harmonization would imply a residence-based tax system with
a deduction for taxes paid to foreign governments. In contrast, Devereux argues
elsewhere in his paper that ‘there is no convincing argument for taxing the
returns from outbound direct investment …’ (716–717), but he observes that
this conclusion assumes that financing can be obtained with inflows of portfolio
investment, so that firms do not choose between different locations, but rather
consider projects separately on their own merits. This is the framework consid-
ered in this chapter, but my analysis emphasizes the deductibility of investment
costs and the interdependence between a firm’s operations in different countries.
As described above, this interdependence implies that tax differences between
locations do matter. Moreover, if the taxation of foreign income is designed
to tax only economic rents, then a case can be made for setting this tax at a
relatively high level, particularly if the taxation of domestic firms continues to
distort investment decisions. An exemption system instead fails to tax any rents
earned abroad, and the emphasis that proponents of the exemption system place
on ownership advantages suggests that these rents are sizable.
This argument in favor of taxing foreign income must of course be qualified
by the recognition that the model omits important complications. In partic-
ular, the model does not distinguish between different foreign locations and
therefore does not capture the wasteful tax avoidance activities that involve
the shifting of taxable income between countries with different tax rates.14 I
have focused on a deduction system, since the purpose of this chapter has been
326 J. D. Wilson
to re-evaluate the traditional argument in favor of such a system, in light of

the recent emphasis on M&A investment. But the use of tax credits for taxes
paid to foreign governments creates its special forms of wasteful tax-avoidance
behavior. Income-shifting activities have been only briefly considered here, and
they are part of the general problem on how to define income earned in differ-
ent locations for the purpose of taxation. Firms in my model are either owned
by home residents or foreign residents, avoiding the thorny problems for tax
design associated with the ownership of firms by residents from different coun-
tries. Once this international ownership is recognized, it becomes difficult to
determine which firms should be taxed by which countries, but this problem
exists under both source- and residence-based tax systems.
I have not considered the taxation of inbound investment, but it could be
added to the model by assuming that foreign multinationals set up subsidiaries
in the home country. In the case of M&A investment, these multinationals
would purchase domestic firms. Then the assumption that the home country is
a small price-taker would imply that these subsidiaries must earn an exogenous
‘reservation profit’, so that taxing them causes the wage to drop to maintain this
profit. The sales prices of domestic firms would again equal the discounted prof-
its that the initial owners could receive by not selling. But these prices would
depend on the tax burden avoided by the owners as a result of selling the firm
(recall equation (8)). In general, these sales prices will not reflect the social value
of selling the firm, but in general we cannot say whether inbound investment
should be taxed or subsidized without considering special cases.15 While taxing
inbound investment will certainly have an effect on the optimal taxation of out-
bound investment when there are the types of productivity linkages described
previously, there seems to be no simple relation between the two taxes.16 This
issue deserves further research.
The model has also assumed perfectly competitive markets, whereas modern
discussions of mergers and acquisitions in the industrial organization literature
typically involve imperfect competition. If we were to model FDI in models
with imperfect competition, then we would expect the pattern of FDI to be
inefficient, in which case taxes that would distort investment in the case of
efficient competitive markets may now partially counteract the inefficiencies
associated with imperfect competition. But there seem to be no general results
about whether international mergers are desirable or not. For example, see the
analysis by Lommerud et al. (2006) of national and international mergers in a
model with union bargaining. In some cases international mergers are desirable,
but not in others.
To conclude, the prevalence of M&A investment does not appear to provide
a justification for abandoning the traditional arguments in favor of taxing for-
eign income. But this conclusion does not mean that the current US system for
levying such taxes is not in urgent need of reform.
Notes
∗ Prepared for the 16th World Congress of the International Economics Association.
Comments by Roger Gordon on an earlier draft are appreciated.
1. See Fuest et al. (2005) for a thorough review of the key results from the international
tax literature. Devereux (2008) provides a review of the results concerning outbound
investment, including a clear discussion of the argument in favor of the exemption
method.
2. I could assume domestic firms differ in their production characteristics, but this is a
needless complication.
3. Foreign purchases of domestic firms are ignored until the concluding section.
4. At the cost of additional notation, I could allow domestic and foreign firms to start
the period with endowments of capital, which become part of the assets acquired
in an acquisition. Firms would then have an opportunity to increase or sell off their
capital inputs prior to production.
5. Consider the introduction of a tax on dividends, at rate t d . Equity investors must
receive the same after-tax return that they could obtain on interest income, (1 − τ )r.
But now firms must pay (1 − τ )r/(1 − t d ) for investors to receive this return. Thus,
if equity financing is used, domestic investment is further distorted by the dividend
tax. But no such distortions occur in the case of FDI that is locally funded by issuing
debt.
6. To focus on differences between home investment and FDI, I simplify the discussion
by assuming that domestic firms and the multinational’s home subsidiaries face the
same cost of finance,
7. Alternatively, if the multinational’s foreign subsidiaries have available retained earn-
ings, the US government would, with some exceptions, apply the repatriation tax to
the returns obtained from using these earnings for portfolio investments (Subpart F
rules). Thus, using retained earnings for M&A and greenfield investments would delay
the repatriation tax to the second period, resulting in a further reduction in the cost
of finance.
8. In a model with a discrete number of subsidiaries, this indifference condition would
not hold because marginal products of labor and capital in the other firms would vary
with changes in their input usage.
9. In some cases, additional greenfield investment could generate cost savings that take
the form of less overall capital usage.
10. Following Becker and Fuest (2011), the purchase price received by the seller, P H , is
assumed not to be taxed. But the results would be unchanged if the seller paid a tax
and the purchaser received a depreciation deduction for the fall in the value of the
firm to zero at the end of period 2.
11. If the foreign government treats interest on debt as tax deductible, then r should be
defined net of foreign taxes.
12. ‘Indeed, it is conceivable that greater outbound FDI is associated with greater domestic
investment, either by home country firms undertaking the FDI or by unrelated foreign
investors’ (2004: 956).
13. A full welfare analysis requires an examination of the wage increases from the pro-
ductivity improvement, but these increases represent an indirect way of effectively
taxing economic rents and are therefore welfare improving.
14. For example, see Altshuler and Grubert (2003).
15. Becker and Fuest (2011) present a model where a source-based tax is fully efficient,
but they drop the small-country assumption by assuming a two-country model with a
328 J. D. Wilson
home multinational that faces a fixed set of potential foreign acquisitions, each with
a unique productivity. In particular, each country’s domestic firms are essentially
a unique set of fixed assets, which earn economic rents, so a source-based tax is
essentially an efficient tax on economic rents.
16. Slemrod et al. (1997) show that a rise in the tax on inbound investment lowers the
optimal tax paid by residents on outbound investment, a phenomenon called the
‘seesaw principle’. But they work with a simple partial equilibrium model, in which
outbound investment is subject to increasing cost. Devereux (2004) obtains this prin-
ciple in a two-sector model with both direct investment and portfolio investment,
but does not consider M&A investment.
References
Altshuler, R. and H. Grubert (2003) ‘Repatriation Taxes, Repatriation Strategies and

Multinational Financial Policy’, Journal of Public Economics, vol. 87, no. 1, pp. 73–107.
Becker, J. and C. Fuest (2010) ‘Taxing Foreign Profits with International Mergers and
Acquisitions’, International Economic Review, vol. 51, no. 1, pp. 171–186.
Becker, J. and C. Fuest (2011) ‘Source Versus Residence Based Taxation with International
Mergers and Acquisitions’, Journal of Public Economics, vol. 95, nos 1–2, pp. 28–40.
Desai, M. A. and J. R. Hines (2003) ‘Evaluating International Tax Reform’, National Tax
Journal, vol. 56, no. 3, pp. 487–502.
Desai, M. A. and J. R. Hines (2004) ‘Old Rules and New Realities: Corporate Tax Policy in
a Global Setting’, National Tax Journal, vol. 57, no. 4, pp. 937–960.
Devereux, M. P. (2004) ‘Some Optimal Tax Rules for International Portfolio and Direct
Investment’, Finanzarchiv, vol. 60, no. 1, pp. 1–23.
Devereux, M. P. (2008) ‘Taxation of Outbound Direct Investment: Economic Principles
and Tax Policy Considerations’, Oxford Review of Economic Policy, vol. 24, no. 4, pp.
698–719.
Fuest, C., B. Huber and J. Mintz (2005) ‘Capital Mobility and Tax Competition’, Foundations
and Trends in Microeconomics, vol. 1, no. 1, pp. 1–62.
Gordon, R. (2012) ‘How Should Income from Multinationals be Taxed?’, in M. Aoki,
T. Kuran and G. Roland (eds), Approaches to the Evolving World Economy: The Proceedings
of the 16th World Congress of the International Economic Association, Vol. III (Basingstoke:
Palgrave Macmillan)
Lommerud, K. E., O. R. Straume and L. Sorgard (2006) ‘National Versus International
Mergers in Unionized Oligopoly’, Rand Journal of Economics, vol. 37, no. 1, pp. 212–233.
Slemrod, J., C. Hansen and S. Procter (1997) ‘The Seesaw Principle in International Tax
Policy’, Journal of Public Economics, vol. 65, no. 2, pp. 163–176.
Index
Actual Household Final Consumption 46 bank failures 155, 217

Adjusted Households’ Disposable rural areas 91
Income 46 bank lending 6, 154–76
adverse feedback loop 100, 155, 261 business 159, 160, 167, 169
adverse shocks 100, 172 commercial real estate 159
aggregate demand 4, 65, 68, 69, 85–6, 87, credit supply shocks 161–71
91, 92, 104, 108 developing countries 230
agricultural sector 71–2 household 159, 160
collapse 62, 67–9 loans outstanding vs loan commitments
global demand 81–2 165–71
income decline 89–90 bank lending rates 64
mobility constraints 74–7 banking crises 67, 85, 88, 91, 116, 132,
prices 75–6 195–6, 239
productivity growth 72–3, 76–7, 78 Basel Accord 145–6, 147, 151, 204, 220
surplus labor 72–3, 74, 75 basket options 277, 286
wages 71, 74–5, 76 Bear Stearns 196
AIG 155 Becker-Fuest framework 314, 315
Albania, foreign bank ownership 212 Better Life Initiative 20
Arrow-Debreu economy 6, 101, 131, 179 Beyond the Market 26, 29
Asia Bhutan 22–3
capital flows 217 bid-ask spread 253, 254–5, 260
equity market capitalization 247 Big Bang 131
equity market turnover 247 borrowing 19, 87, 90, 119, 134, 230, 239
Financial Crisis (1997) 199, 208 business 165, 173
foreign exchange reserves 193 constraints 6, 103, 104, 105–7, 122, 133,
asset mispricing 184–5
macroprudential regulation 202–3 cost of 91
systemic risk 194–5 excessive 102, 108–9, 111
asset prices 4, 18, 19, 26, 61, 99, 118–30, household 165
132, 134, 161, 163 taxation of 110
falling 134–5, 194 borrowing limit 107–8, 110, 122
inflation 98, 137 Bosnia and Herzegovina
and liquidity 195 commitment to Vienna Initiative 216
volatility of 115, 133, 195, 197 foreign bank ownership 212
asset price bubbles 7, 89, 136, 137, 138, Brazil, capital flows 217
140, 191 broker-dealers 163, 173
macroprudential regulation 200–2 Bulgaria, foreign bank ownership 212
systemic risk 193 Bureau of Economic Analysis 26, 29–30
business cycles 5, 26, 62
backward induction 103, 107–8 asset prices 118–30
balance sheet crises 100 household and business loans 160
Banco Estado (Chile) 206 liquidity shocks 118–30
banks private nonfinancial debt 158
developing countries 230, 241–2 real models 85, 86
financial role 226–7 and social wellbeing 21–2
329
330 Index
business lending 159, 160, 167 recursive 119, 123

and excess bond premium 169 welfare inefficiency 131
business lending capacity 167 consumer credit 157–60
and credit supply shocks 170 contagion 132, 141–4, 263–4
buyout funds 272, 279, 280, 282, 284 asset price volatility 197
macroprudential regulation 204–6
capital sovereign default 207
financial regulation 144, 147 systemic risk 196–8
under-provision 136–41 contagious insolvency 143–4
capital controls 99–100, 101 contingent convertible debt (CoCos)
as Pigouvian taxation 110–11 205, 206
capital flows 98–117, 217 corporate business income taxation
bonanzas 98, 99 293–5
cross-border 214, 225 corporate taxation 9–10, 289–312
externalities see externalities closed economy 290–5
inflows 98–9, 101 corporate business income 293–5
outflows 100 income from labor/savings 291
capital regulation 204 non-corporate business income
countercyclical 205–6 292–3
capital-export neutrality 9, 300, 313 cross-border activity 301–8
capital-import neutrality 300 optimal vs actual tax provisions
capital-labor ratio 323 301–3
capital-ownership neutrality 300–1 pressures and responses 307–8
carbon emissions 18 puzzles 303–6
cash-in-the-market 195, 202 foreign income exemption 313–28
central counterparties 264 open economy 295–301
central trading exchanges 248 capital income 300–1
Central/Eastern Europe inbound FDI 298–300
foreign bank ownership 212 inbound portfolio investment 296–7
foreign exchange reserves 193 outbound FDI 297–8
Chile, Banco Estado 206 outbound portfolio investments
China 295–6
capital flows 217 countercyclical capital regulation 144,
currency reserves 202 145, 161, 205–6
financial repression 237 credit aggregates 157–61
circuit-breakers 141–4, 149–51, 266 credit expansion 193, 201
closed economy corporate tax 290–5 credit supply shocks 161–71
corporate business income 293–5 and business lending capacity 170
income from labor/savings 291 excess bond premium 156, 161–5, 168
non-corporate business 292–3 loans outstanding vs loan commitments
co-location 252–3 165–71
Commission on Measurement of Croatia, foreign bank ownership 212
Economic Performance and Social cross-border activity 301–8
Progress 2, 17 optimal vs actual tax provisions 301–3
recommendations of 40–1 pressures and responses 307–8
commitment letters 216, 218, 222 puzzles 303–6
commitment to pay cross-border banking 214
full 178–9 cross-country regressions 16–17
limited 183–5 currency mismatching
Common International Classification of macroprudential regulation 208
Ecosystem Services (CICES) 52 systemic risk 199
competitive equilibria 71 Cutler, David 30
Index 331
Danielsson, Jon 143 emissions

debt aggregates 52
CoCos 205, 206 carbon 18
external 88 employment 50
private nonfinancial 158 urban 75–6
short-term 101, 102, 113–14, 197, 203–4 endowment structures 229–31
debt flows 101 developed countries 230–1
deep downturns 89 developing economies 229–30
see also financial crisis; Great environment 51–2
Depression; Great Recession Environmental Protection Expenditure
deleveraging 4, 64–5, 87, 98 Account 52
Emerging Europe 221 equity liquidity 119, 124
and insolvency 136 constraint 5, 122, 124, 128
deposit insurance 203–4 negative shocks 119, 127
Desai-Hines framework 314, 317 steady state 125–6
developed countries equity markets 240–1
accounting disclosure 232 bid-ask spread 253, 254–5, 260
endowment structures 230–1 capitalization 246–7
stock market development 232 correlation 255
developing countries developed countries 230–1, 232
banking sector 230, 241–2 financial role 227
endowment structures 229–30 fragmentation 254
optimal financial structure 237 macrostructure see market
development see economic development macrostructure
Directors General of the National microstructure–macrostructure
Statistical Institutes, Sofia transition 253–8
Memorandum 42 regulation 247–8
distance-to-default 161–2 turnover 247
domestic material consumption 52 volatility 255–6
domestic material input 52 see also trading
equity prices
economic development 8 boom 33, 119, 125, 127–8
bottlenecks 228 collapse 31, 119, 128, 245–70
endowment structures 229–31 and liquidity shocks 119, 128, 129
and financial liberalization 237–8 equity resaleability 121, 122
and financial structure 224–44 equity-issuing bound 121
holistic approach 22–3 Europe
economic growth equity market capitalization 247
expanded measures 28–9 Market in Financial Instruments
human capital 28–9 Directive 248
innovation and intangibles 28 European Bank Coordination Initiative
and spending power 26–7 222
economic performance measures 45–7 European Bank for Reconstruction and
consumption 46 Development (EBRD) 7, 219
extension of national accounts 46–7 European Banking Authority 221
GDP see GDP European Central Bank 193, 221
income 46 interest rate policy 201
education 49–50 European Commission 40
efficiency wages 62, 63, 69, 77–8, 79, 91, GDP and Beyond: Measuring Progress in a
92, 93, 94 Changing World 41
electronic trading 249–52 European Financial Stability Facility
co-location 252–3 204, 220
332 Index
European Framework for Integrated behavior of 6–9

Environmental and Economic capital regulation 204–5
Accounting for Forests (IEEAF) 52 combined public/private 206
European Investment Bank 219 international see international financial
European Statistical System 41–2 institutions
European Systemic Risk Board 221 systemically important (SIFIs) 143, 217,
European Union 211 220
cross-border banking 214 financial liberalization 237–8
effect of Lehman Brothers collapse financial markets
213–14 interaction 265
Eurozone 200–1 recapitalization 84–5, 88
bailout fund 206–7 see also equity markets
exit and re-entry 208 financial regulation 6–9, 144–6, 147
foreign bank ownership 212 and bank leverage 145–6
sovereign debt crisis 198–9 capital 144, 147
Stability Pact 198, 206 liquidity 144, 147
excess bond premium 156, 161–3 prohibited services 146
and bank lending 163–5 timetable 145
and banks’ willingness to lend 168 financial repression 236–7
and business lending 169 financial structure
exit value 272, 273, 275, 276, and development 8, 224–44
280, 285 bank-based view 226–7
external debt 88 bottlenecks 228
externalities 5, 101–14 empirical data 231–6
benchmark model 102–3 endowment structures 229–31
fire-sale 132 irrelevancy view 225–6
model solution 103–6 law and finance view 226
network 141–4 market-based view 227
pecuniary 131 optimal 233–4
financial accelerator 133–6 deviation from 236–40
insolvency risk 136 and poverty 235–6
quantification of 114–15 financial structure gap 233, 234
risky financing solutions 111–13 financing decisions, risky 111–13
short-term debt 113–14 fire-sales 133–6
social planner solution 106–11 externalities 132
firm-specific assets 315
fair-value test (FVT) 9, 271–2, 275, fiscal policy 79, 91
280–2 Flash Crash 245, 256, 260, 264, 267
Fannie Mae 136, 155, 193 see also equity markets; trading
FDI 313 foreign direct investment see FDI
inbound 298–300 foreign equity purchases 295–6
outbound 297–8 foreign exchange reserves 193, 199
taxation 322–4 foreign income
Federal Reserve Bank 196 income-shifting 324
Federal Reserve Board 33 investment linkages 321–4
financial accelerator 133–6 taxation of 313–28
financial business sector leverage 34 fractal geometry, application to market
financial crisis (2008–09) 4, 13, 98, 155 behavior 258–63
credit aggregates 157–61 France 20
see also Great Recession Conseil d’Analyse Économique 42
financial frictions 177–87 Freddie Mac 136, 155, 193
financial institutions FVT see fair-value test
Index 333
GDP 2–3, 13–17 Physical Component Summary measure

cross-country regressions 16–17 49
as economic performance measure 45–6 healthcare costs 29–30
effect of Lehman Brothers collapse high-frequency trading 8, 249–52, 259–60
196–7 and market volatility 255–6
and equity market capitalization 246–7 price impact 254
and equity price growth 33 round-trip latency 252
and housing stock value 32 zero latency 252
and income 47 home ownership 21
labor market affecting 15 household
as measure of wellbeing 14 asset value 32
per capita 39 consumption 46
GDP and Beyond: Measuring Progress in a deleveraging 64–5
Changing World 41 lending 159, 160
Georgia, foreign bank ownership 212 housing market 3
Germany 20 collapse 31
Council of Economic Experts 42 Human Development Index 38, 42
Gilchrist-Zakrajs̆ek indicator see excess Human Development Report 42
bond premium Hungary
Glass-Steagall law 131–2 commitment to Vienna Initiative 216
Global Project on Measuring the Progress of foreign bank ownership 212
Societies 40 hurdle rate 276, 283, 285
globalization 9, 47, 62, 83, 85, 295 Hurst coefficient 257–8, 259
GNP 36, 92
Granger causality test 98 income 46
Great Depression 4, 25, 62, 67–78, 155 Adjusted Households’ Disposable
agricultural sector collapse 62, 67–9 Income 46
price falls 64 and GDP 47
recovery 70 individual 179–82
Great Recession 4, 61–97 personal 27
explanations for 63–7 and housing stock value 31, 32
government interventions 79–81 taxation 9, 295, 299
fiscal policy 79, 91 Replacement Rate 46
migration subsidies 80–1 income tax 9, 295, 299, 317
wage adjustments 79–80 income-shifting 290, 294, 304, 309, 315,
models for 67–78 318
efficiency wages and induced by multinationals 324
unemployment 77–8 elimination of 305–6
free mobility 70–3 incomplete markets 194–5
mobility constraints 74–7 Independent Commission for Banking
Greece 19 145, 151
sovereign debt crisis 206–7 India, capital flows 217
Green, Edward 177–87 Indonesia, externalities 113
Greenspan, Alan 131, 132 inequality 14, 85
gross domestic product see GDP information spillovers 303
gross national happiness 22, 26 innovation 28, 163–4, 169, 214, 227, 230,
see also Bhutan 253
gross national product see GNP risks 252
insecurity 15, 18, 40, 51
health 49 insolvency 134
Mental Component Summary measure contagious 143–4
49 risk of 136
334 Index
intangibles 25, 28 Kiyotaki-Moore model 119, 128, 133

International Adult Literacy Survey 50 Kuznets, Simon 25
international financial institutions 211–23 Kyrgyz, foreign bank ownership 212
Joint IFI Action Plan 215
Vienna Initiative 214–20 Labor Force Survey 50
International Monetary Fund 7, 193, 199, labor market
201–2, 218–19 and GDP 15
interpersonal relationships 51 growth in 235
investment linkages 321–2 skilled labor 230
and optimal FDI taxation 322–4 surplus labor 72–3, 74, 75
Ireland unskilled labor 229–30
bank debt guarantees 204 labor market institutions 15
real estate bubbles 192–3 labor mobility 93
irrational exuberance 33 constrained 74–7, 93
Istanbul Declaration 40 free 70–3
Istat see Italian National Institute of migration subsidies 80–1
Statistics Lamfallusy process 214
Italian National Institute of Statistics 38 Latin America
economic performance measures 45–7 capital flows 217
consumption 46 foreign exchange reserves 193
extension of national accounts 46–7 Latvia, commitment to Vienna Initiative
GDP 45–6 216
income 46 legal system, financial system effects 226
environmental statistics 51–2 Lehman Brothers collapse 136, 142, 155,
quality of life measures 48–51 157, 196, 211
education 49–50 effect on Europe 213–14
health 49 effect on GDP 196–7
insecurity 51 lending see bank lending
interpersonal relationships 51 leverage
material living standards 48–9 excessive 137–9
personal activities and work 50 and financial regulation 145–6
political voice and governance 50 Lewis, William Arthur 91
Italy 20 limited liability 5, 132, 136, 137–9, 140
National Council for Economics and liquidity
Labour 52–3 countercyclical 168
social wellbeing 38–57 equity 119, 124
Istat statistics 45–52 constraint 5, 122, 124, 128
national consultation 52–5 negative shocks 119, 127
sustainability framework 43–5 steady state 125–6
Unicredit 206 financial regulation 144, 147
provision of 142
Japan 13–14 undervaluing of 5
financial crisis 66 valuation 107–8, 109
Joint IFI Action Plan 215 private 135
JP Morgan 196 public 135
liquidity provision 194–5
Kazakhstan, capital flows 217 liquidity shocks 5, 118–30
Kennedy, Robert F. 25 dynamic effects 127–8
Keynes, John Maynard 95 equilibrium effects 124–8
Keynesian liquidity trap 64 equity liquidity in steady state 125–6
Keynesian policies 4, 67, 70, 84, 85, 89 optimality conditions 124–5
impact of 79 Kiyotaki-Moore model 119, 128
Index 335
macro model with equity financing migration subsidies 80–1

120–4 Millennium Development Goals 38
liquidity trap 64 missing insurance markets 139–41
Lloyds Bank 206 monetary policy 31, 64, 128, 164, 165,
loans see bank lending 172, 191–3, 200, 201
long-run crises 61–97 ‘Monitoring Economic Performance,
low-frequency trading 260–2 Quality of Life and Sustainability’
42–3
Maastricht Treaty 198 mortgages 34, 49, 65, 136, 154, 159, 192,
Macedonia, foreign bank ownership 212 193
macroeconomics 1, 3–4, 63, 88 multi-factor productivity 28
macroprudential regulation 200–8 multinationals, taxation of see corporate
asset mispricing 202–3 taxation
asset price bubbles 200–2 multiple equilibria 7, 195–6
contagion 204–6
currency mismatching 208 NASDAQ 249–50
panics 203–4 round-trip latency 252
sovereign default 206–8 National Council for Economics and
Mandelbrot, Benôit 258 Labor 39
manufacturing sector 70–1 National Market System 248
labor migration to 75 National Science Foundation 28
productivity gains 62, 66–7 net domestic product (NDP) 45
wages 75–6 network externalities 141–4
mark-to-market accounting 203 New Deal 69
market failures 4, 16, 195 non-bank lenders 173
Market in Financial Instruments Directive non-corporate business income taxation
248 292–3
market macrostructure 258–63 nonfinancial business credit 159
behavior across assets/markets 262–3 Nordhaus, William 2
high-frequency trading 259–60 normalization of deviance 255
low-frequency trading 260–2 North America
mapping of 264–5 equity market capitalization 247
modification of 265–7 see also USA
and public policy 263–7
see also equity markets; trading OECD 10, 14, 15, 38
market microstructure 253, 264 Better Life Initiative 20
market-based systems 227 Istanbul Declaration 40
market-makers 253, 259, 260, 261, 265 OECD countries, financial structure 232–3
market-making guidelines 265–6 off-balance-sheet credit exposure 156–7,
Markov process 121, 122 166–7, 171
material living standards 48–9 Oh, Soo Nam 177–87
mechanism design approach 177–87 open economy corporate tax 295–301
limited commitment 183–4 capital income 300–1
and private information 184–5 inbound FDI 298–300
private information inbound portfolio investment 296–7
individual income 179–81 outbound FDI 297–8
individual income and storage 181–2 outbound portfolio investments 295–6
public information and full order imbalance 263, 265
commitment 178–9 Organization for Economic Co-operation
Mental Component Summary measure 49 and Development see OECD
mergers and acquisitions 313–28 Organization of the Islamic Conference 40
microeconomics 3, 186 ownership neutrality 313, 314
336 Index
panics limited partners 271, 273–4

macroprudential regulation 203–4 mapping to carry amount 275–7
systemic risk 195–6 model outputs 278–85
Pareto efficiency 3, 101, 131 portfolio company valuation 274
pecuniary externalities 131 random exit time 275
financial accelerator 133–6 random investment duration 274–5
insolvency risk 136 risk-neutral valuation 273
see also externalities sensitivity analysis 285
Perrow, Charles 263 simulation 277–8
personal activities 40, 50 private information
personal consumption expenditure 163 individual income 179–81
personal income 27 individual income and storage 181–2
and housing stock value 31, 32 limited commitment 184–5
tax 9, 295, 299, 317 private nonfinancial debt 158
Physical Component Summary measure production 120–1
49 productivity growth 62, 66–7
physical trade balance 52 and unemployment 72–3, 74, 75, 85
Pigouvian taxation 5, 99, 113, 114 prohibited services 146
capital controls as 110–11 public goods 100, 108, 316, 317, 324
Poland, foreign bank ownership 212 public information 178–9
political participation 50
pollutants 52 quality of life 20
carbon emissions 18 education 49–50
portfolio investments health 49
company valuation 274 insecurity 51
compensation schemes see private interpersonal relationships 51
equity fund compensation material living standards 48–9
exit value 272, 273, 275, 276, 280, 285 personal activities and work 50
fair-value test 9, 271–2, 275, 280–2 political voice and governance 50
taxation 310, 317
inbound investments 296–7 Reagan, Ronald 131
outbound investments 295–6 real business cycle models 85, 86
see also corporate taxation real estate 186
poverty 48 investment 64, 87
and financial structure 235–6 loans 161, 173, 174, 201
indicators 49 value 32, 91
reduction of 8, 241 real estate bubbles 2, 7, 13, 84, 191–4,
precautionary saving 139–41 200, 202
preferred return see hurdle rate recursive competitive equilibrium 119,
pricing 123
agricultural sector 75–6 remittance tax 317
asset prices see asset prices repatriation, tax at 307
frictions in 253 Replacement Rate 46
prisoners’ dilemma 100 Resource Use and Management
private equity fund compensation 8–9, Expenditure Account 52
271–86 resting rules 267
exit value 272, 273, 275, 276, 280, 285 ring-fencing 145–6, 147
FVT 9, 271–2, 275, 280–2 risk-weighted assets 144, 145, 151
general partners 271, 272, 273–4 risk-sharing network 147–51
inflation of un-exited investment values circuit-breaker 149–51
282–3 isolated node 148
initial investment values 273–4 node in component 149
Index 337
risk-shifting 132, 136–41 unemployment 13

limited liability and excessive leverage Sofia Memorandum 42
137–9 South Korea 199
missing insurance markets and sovereign default
precautionary saving 139–41 contagion 207
risky financing solutions 111–13 macroprudential regulation 206–8
Romania systemic risk 198–9
commitment to Vienna Initiative 216 Spain 13
foreign bank ownership 212 countercyclical loan reserves 205–6
Royal Bank of Scotland 206 real estate bubbles 192–3
Rubini, Noriel 31 unemployment 194
Russia Special Resolution Regimes (SRR) 143
capital flows 217 spending power and economic growth
equity market capitalization 247 26–7
Stability Pact 198
Sarkozy, Nicolas 40 Stiglitz, Joseph 146
savings 181–2 Stiglitz-Sen-Fitoussi Report 2, 26, 29, 46
precautionary 139–41 stock markets see equity markets
tax on 291 super-spreaders 143, 151
savings rates 87 sustainability 17–19, 43–5
and household asset values 32 measures of 30–5
precautionary saving 139–41 progress indicators 19, 43–5
zero 65 sustainable investment 35–6
sectoral imbalances 61–97 systemic risk 7, 191–9
seesaw principle 328 asset mispricing 194–5
Senior Loan Officer Opinion Survey on contagion 196–8
Bank Lending Practices (SLOOS) currency mismatching 199
168, 174 liquidity provision 194–5
Serbia multiple equilibria and panics 195–6
commitment to Vienna Initiative 216 real estate bubbles 191–4
foreign bank ownership 212 sovereign default 198–9
SERIEE 52 Systemically Important Financial
shocks Institutions (SIFIs) 143, 220
adverse 100, 172
credit supply 161–71 TARP 61, 63, 84, 202
liquidity 118–30 tax holidays 307
short-term debt 101, 102, 113–14, 197 tax neutrality 295, 296, 300–1, 307, 310
guarantees 203–4 taxation
Sinn, Hans Werner 136–7 on borrowing 110
Slovakia, foreign bank ownership 212 corporate see corporate taxation
social capital 22, 47, 51 FDI 322–4
social planner 106–11 foreign income 313–28
social wellbeing 13–24 income tax 9, 295, 299, 317
assessment of 20–1 Pigouvian 5, 99, 110–11, 113, 114
and business cycle 21–2 savings 291
inequality 14 territorial system 301, 303
Italy 38–57 worldwide 301–2
Istat statistics 45–52 Taylor rule 193
national consultation 52–5 territorial tax system 301, 303
sustainability framework 43–5 Thatcher, Margaret 131
metrics and policies 15–17 TIPS 87
sustainability 17–19, 43–5 Tobin, James 2
338 Index
‘too big to fail’ 196 urban employment, and rural prices 75–6
total material consumption 52 USA 13
total material requirement 52 economic growth
toxic assets 202 expanded measures 28–9
see also TARP and spending power 26–7
trade repositories 264 healthcare costs 29–30
trading 246–53 National Market System 248
adverse selection risk 252–3 real estate bubbles 192–3
electronic 249–52 sustainability measures 30–5
high-frequency see high-frequency tangible investment 35–6
trading
low-frequency 260–2
vector autoregression 163, 169, 174
multiple venues 251
venture capital funds 272, 284
topology 246–53
hurdle rate 276, 283, 285
volume shares 248–9, 250
Vickers Commission 143
see also equity markets
Vickers, John 145
Treasury Inflation Protected Securities see
Vienna Initiative 7, 214–20
TIPS
commitment letters 216
Troubled Asset Relief Program see TARP
success of 217–18
Tunisia 14
UK 20 wage tax see income tax

Industrial Revolution 238 wages
Ukraine, capital flows 217 adjustments in 79–80
unemployment 21 agricultural sector 71, 74–5, 76
induced 77–8 efficiency 62, 63, 69, 77–81, 91, 92, 93,
and productivity growth 72–3, 74, 75, 94
85 rigidity in 88
and wage demands 80 Washington Mutual 155
Unicredit 206 Well-Being Index 26
United Nations 40 World Bank 40, 219
Development Programme 40, 42 worldwide taxation 301–2

The Global Macro Economy and Finance (PDFDrive)

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

The Global Macro Economy and Finance (PDFDrive)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Global Macro Economy and Finance (PDFDrive)

Uploaded by

Copyright:

Available Formats

The Global Macro Economy and Finance

This is IEA conference volume no. 150-III

List of Tables vii

List of Figures viii

Part I Beyond GDP

2 The Role of Statistics in the United States’ Economic Future 25

3 Measuring Equitable and Sustainable Wellbeing in Italy 38

Part II Structural Imbalances, Financial

5 Capital Flows, Crises and Externalities 98

6 Liquidity Shocks and Asset Prices in the Business Cycle 118

7 Whither Capitalism? Financial Externalities and Crisis 131

8 Bank Lending and Credit Supply Shocks 154

9 A Mechanism Design Approach to Financial Frictions 177

Part III Behavior of Financial

12 Some Recent Progresses on Financial Structure and

13 The Race to Zero 245

14 A Model of Private Equity Fund Compensation 271

Part IV Taxation in a Globalized Economy

16 Taxing Multinationals in a World with International

3.1 Goals of progress 44

2.1 Reconciling measures of income 27

Franklin Allen is Nippon Life Professor of Finance, Professor of Economics and

the American Economic Review. He is a Fellow of the Econometric Society and

as a member of the Board of Editors of the American Economic Review, and as

The International Economic Association (IEA), the association of national

evolving environmental, demographic and global market conditions. Will these

of development is now about to end. Simply put, improvements in the qual-

* Agarwal, Bina, India (Delhi University)

* Roland, Gérard, Belgium (University of California, Berkeley)

Cai, Hongbin (Peking University)

July 27, 2012

Structural imbalances, ﬁnancial frictions and externalities

of tight ﬁnancial constraint, a constrained social planner takes on less debt

Behavior of ﬁnancial institutions and prudential regulations

in most countries. Much of this regulation was designed to prevent ﬁnancial

Corporate taxation in a globalized economy

With globalization, there is an increasing amount of cross-border activity by

Unless the return to the creative efforts of domestic entrepreneurs accruing

1 Metrics and policies

Many of the advocates of unfettered markets – who see any interven-

2 Use and misuse of the concept of sustainability

We can do better. The Commission on the Measurement of Economic Perfor-

countries are currently looking for sustainability indicators in order to assess

4 Wellbeing and the business cycle

This broader and redeﬁned perspective on measurement is relevant not only

discrepancies between the ﬂuctuation of output and that of wellbeing. It may

5 Beyond GDP: the experience of Bhutan

Barro, Robert (1995) ‘Determinants of Economic Growth: A Cross-country Empirical

The need for better statistics

The national accounts were developed to address the paucity of comprehensive

Distribution of growth and spending power

Figure 2.1 Reconciling measures of income

Source: BEA, IRS Statistics of Income.

Expanded measures of the sources of growth

Although researchers ranging from Dennison and Madison to Jorgenson and

The increasing role of innovation and intangibles in

of Jorgenson-Fraumeni estimates of human capital (based on the discounted