Reinhart, Carmen M. - Goldstein, Morris Arthur - Forecasting Financial Crises - Early Warning Signals For Emerging Markets

ASSESSING
FINANCIAL
VULNERABILITY
An Early Warning System
for Emerging Markets
ASSESSING
FINANCIAL
VULNERABILITY
An Early Warning System
for Emerging Markets
Morris Goldstein
Graciela L. Kaminsky
Carmen M. Reinhart
INSTITUTE FOR INTERNATIONAL ECONOMICS

Washington, DC
June 2000
Morris Goldstein, Dennis Weatherstone INSTITUTE FOR INTERNATIONAL
Senior Fellow, has held several senior staff ECONOMICS
positions at the International Monetary 11 Dupont Circle, NW
Fund (1970-94), including deputy director Washington, DC 20036-1207
of its research department (1987-94). He (202) 328-9000 FAX: (202) 328-5432
has written extensively on international http://www.iie.com
economic policy and on international
capital markets. He is author of The Asian C. Fred Bergsten, Director
Financial Crisis: Causes, Cures, and Systemic Brigitte Coulton, Director of Publications
Implications (1998), The Case for an and Web Development
International Banking Standard (1997), The Brett Kitchen, Marketing Director
Exchange Rate System and the IMF: A
Modest Agenda (1995), coeditor of Private Printing and typesetting by
Capital Flows to Emerging Markets after the Automated Graphic Systems
Mexican Crisis (1996), and project director
of Safeguarding Prosperity in a Global Copyright © 2000 by the Institute for
Financial System: The Future International International Economics. All rights
Financial Architecture (1999) for the Council reserved. No part of this book may be
on Foreign Relations Independent Task reproduced or utilized in any form or by
Force on International Financial any means, electronic or mechanical,
Architecture. including photocopying, recording, or by
information storage or retrieval system,
Graciela Kaminsky, visiting fellow, is a without permission from the Institute.
professor of economics and international
relations at George Washington For reprints/permission to photocopy
University. She was assistant professor of please contact the APS customer service
economics at the University of California, department at CCC Academic Permissions
San Diego (1985-92) and a staff economist Service, 27 Congress Street, Salem, MA
at the Board of Governors of the Federal 01970.
Reserve System (1992-98) before joining
George Washington University. She has Printed in the United States of America
been a consultant and visiting scholar at 02 01 00 54321
the International Monetary Fund and the
World Bank, and has published
Library of Congress Cataloging-in-
extensively on issues in open economy
macroeconomics. In the last few years, her Publication Data
areas of research have been on financial
crises, contagion, and herding behavior. Goldstein, Morris, 1944-
Assessing financial vulnerability : an
Carmen Reinhart, visiting fellow, is a early warning system for emerging
professor at the University of Maryland in markets / Morris Goldstein, Graciela L.
the School of Public Affairs and the Kaminsky, Carmen M. Reinhart.
Department of Economics. She is a p. cm.
research associate at the National Bureau Includes bibliographical references and
of Economic Research. She was vice index.
president at the investment bank Bear ISBN 0-88132-237-7
Stearns for several years before joining the 1. Bank examination. 2. Foreign
research department at the International exchange. 3. Financial Crises.
Monetary Fund in 1988. Her work on I. Kaminsky, Graciela Laura.
various topics in macroeconomics and II. Reinhart, Carmen M. III. Title.
international finance and trade—including
capital flows to emerging markers, capital HG1707.5 .G65 2000
controls, inflation stabilization, currency 332⬘.0917⬘4--dc21 00-038291
and banking crises, and contagion—has
been published in leading scholarly
journals and featured in the financial press. ISBN 0-88132-237-7
The views expressed in this publication are those of the authors. This publication is
part of the overall program of the Institute, as endorsed by its Board of Directors, but
does not necessarily reflect the views of individual members of the Board or the
Advisory Committee.
Institute for International Economics | http://www.iie.com

Contents
Preface ix
Acknowledgments xv
1 Introduction 1
Organization of the book 9
2 Methodology 11
General Guidelines 11
Putting the Signals Approach to Work 18
3 Empirical Results 33
The Monthly Indicators: Robustness Check 33
The Annual Indicators: What Works? 38
Do the Indicators Flash Early Enough? 40
Microeconomic Indicators: Selective Evidence 42
4 Rating the Rating Agencies 45

Do Sovereign Credit Ratings Predict Crises? 45
Why Do Credit Ratings Fail to Anticipate Crises? 49
Do Financial Markets Anticipate Crises? 52
5 An Assessment of Vulnerability: Out-of-Sample Results 55

Vulnerability and Signals 56
A Composite Indicator and Crises Probabilities 64
6 Contagion 73
Defining Contagion 74
Theories of Contagion and Their Implications 75
Empirical Studies 76

Trade and Financial Clusters and a Composite Contagion Index 77
What the Index Reveals about Three Recent Crisis Episodes 79
7 The Aftermath of Crises 85

The Recovery Process 85
Some Caveats 89
8 Summary of Results and Concluding Remarks 95

Summary of Findings 95
Would the Publication of the Indicators Erode Their Early
Warning Role? 109
Do the Better Performing Indicators Carry Policy Implications? 110
Appendix A: Data and Definitions 111
References 115
Index 121
Tables
Table 1.1 Emerging Asia: real GDP growth forecasts, 1996-98 3
Table 1.2 Rating agencies’ performance before the Asian crisis:
Moody’s and Standard & Poor’s long-term debt ratings,
1996-97 4
Table 2.1 Currency crisis starting dates 22
Table 2.2 Banking crisis starting dates 24
Table 2.3 Selected leading indicators of banking and currency crises 26
Table 2.4 Optimal thresholds 29
Table 2.5 Examples of country-specific thresholds: currency crises 30
Table 3.1 Ranking the monthly indicators: banking crises 34
Table 3.2 Ranking the monthly indicators: currency crises 35
Table 3.3 Annual indicators: banking crises 38
Table 3.4 Annual indicators: currency crises 39
Table 3.5 Short-term debt: selected countries, June 1997 40
Table 3.6 How leading are the signals? 41
Table 3.7 Microeconomic indicators: banking crises 42
Table 4.1 Comparison of Institutional Investor sovereign ratings with
indicators of economic fundamentals 46
Table 4.2 Do ratings predict banking crises? 48
Table 4.3 Do ratings predict currency crises? 48
Table 4.4 Do ratings predict banking crises for emerging markets? 50
Table 4.5 Do ratings predict currency crises for emerging markets? 50
Table 4.6 Rating agencies’ actions on the eve and aftermath of the Asian
crisis, June-December 1997 51
vi

Table 4.7 Do financial crises help predict credit rating downgrades?
(Institutional Investor) 52
Table 4.8 Do financial crises predict credit rating downgrades? (Moody’s) 52
Table 5.1 Signals of currency crises, June 1996-June 1997 57
Table 5.2 Borderline signals of currency crises, June 1996-June 1997 59
Table 5.3 Signals of banking crises, June 1996-June 1997 60
Table 5.4 Borderline signals of banking crises, June 1996-June 1997 61
Table 5.5 Weighting the signals for currency and banking crises in
emerging markets, June 1996-June 1997 62
Table 5.6 Vulnerability to financial crises in emerging markets: alternative
measures, June 1996-June 1997 63
Table 5.7 Composite indicator and conditional probabilities of
financial crises 66
Table 5.8 Scoring the forecasts: quadratic probability scores 67
Table 6.1 Crises that showed few signals, 1970-97 74
Table 6.2 Conditional probabilities and noise-to-signal ratios for
financial and trade clusters 77
Table 6.3 Countries sharing financial and trade clusters with original
crisis country or region 80
Table 6.4 Contagion vulnerability index 81
Table 6.5 Characteristics of affected countries in Asian and
Mexican episodes 82
Table 6.6 Asia and Latin America: added power of Thai crisis in
explaining probability of contagion in bank cluster, July 1997 83
Table 7.1 Length of recovery from financial crises 86
Table 7.2 Time elapsed from beginning of banking crises to their peaks 87
Table 7.3 Comparison of inflation and growth before and after
currency crises 88
Table 7.4 The wake of devaluations: a review of the literature 90
Table 7.5 Comparison of severity of crises by region and period, 1970-97 92
Table 8.1 Currency and banking crises: best performing indicators 97
Table 8.2 Country rankings of vulnerability to currency crises for
two periods 99
Table 8.3 Country rankings of vulnerability to banking crises for
two periods 101
Figures
Figure 2.1 Mexico: real exchange rate, 1970-96 31
Figure 5.1 Probability of currency crises for four Southeast Asian
countries, 1990-97 69
vii

Preface
The last decade has been disrupted by a series of currency, banking, and
debt crises. Hence the research agenda of the Institute for International
Economics over this period has given high priority to the analysis of,
and policy prescriptions for, crisis prevention and management. Wendy
Dobson, in Economic Policy Coordination (1991), and C. Fred Bergsten and
C. Randall Henning, in Global Economic Leadership and the Group of Seven
(1996), explored ways to improve G-7 policy coordination. Morris Gold-
stein, in The Exchange Rate System and the IMF (1995), examined the
appropriate design of G-3 currency arrangements and the role of the
International Monetary Fund. John Williamson, in What Role for Currency
Boards? (1995), analyzed the pros and cons of currency boards. Guillermo
Calvo, et al., in Private Capital Flows to Emerging Economies after the Mexican
Crisis (1996), drew lessons from the Mexican peso crisis. Banking crises
in emerging economies—and how to reduce their incidence and severity
by adopting a voluntary international banking standard—were the objects
of Morris Goldstein’s 1997 study, The Case for an International Banking
Standard, which led to subsequent official adoption of the Basel Core
Principles of Effective Banking Supervision.
The Asian financial crisis provided a powerful additional stimulus to the
Institute’s work on crisis prevention and management. Morris Goldstein
outlined the origins of the Asian crisis as well as the key policy responses
needed to overcome it in The Asian Financial Crisis: Causes, Cures and
Systemic Implications (1998). Marc Noland, et al., in Global Economic Effects
of the Asian Currency Devaluations (1998), used a general equilibrium model
to assess the implications for the global economy of devaluations in the
ix

Asian crisis economies. Adam Posen, in Restoring Japan’s Economic Growth
(1998), completed the Institute’s Asian trilogy by identifying the macroeco-
nomic and financial sector policies necessary to pull Japan’s economy out
of its long period of stagnation and banking distress. In addition, the
possibility of a currency crisis in the largest country of all was addressed
by Catherine Mann in Is the U.S. Trade Deficit Sustainable? (1999).
Two recent Institute books have been at the center of the ongoing debate
on the reform of the international financial architecture. Barry Eichengreen
offered a practical agenda for reform that has subsequently become the
definitive statement of the case for ‘‘moderate’’ reform of the system in
Toward A New International Financial Architecture: A Practical Post-Asia
Agenda (1999). Shortly thereafter, the Institute published a Council on
Foreign Relations (CFR) task force report on the future architecture, Safe-
guarding Prosperity in a Global Financial System (1999). The CFR report,
representing the consensus view of 29 distinguished experts and authored
by Morris Goldstein, put forth an integrated set of recommendations that
has generated considerable bipartisan support for reform and has been
compared favorably to other recent reports (such as that of the Interna-
tional Financial Institutions Advisory Commission chaired by Alan
Meltzer).
This new study by Dennis Weatherstone Senior Fellow Morris Goldstein
and Visiting Fellows Graciela Kaminsky and Carmen Reinhart, employs
a comprehensive battery of empirical tests to identify the best leading
indicators of currency and banking crises in emerging economies. Using
the ‘‘signals’’ methodology, pioneered by Kaminsky and Reinhart, the
authors show that there is a systemic pattern of abnormalities leading up
to most currency and banking crises. They demonstrate that an ‘‘early
warning system’’ based on such leading indicators performs quite well,
not only in tracking currency and banking crises in emerging economies
over the 1970-95 sample period but also in anticipating most of the coun-
tries affected by the Asian crisis. In an advance on existing empirical
work, Goldstein, Kaminsky, and Reinhart address both own-country and
cross-country ‘‘contagion’’ elements of crisis vulnerability and compare
their better-performing leading indicators to market indicators of future
crises (i.e., interest rate spreads and sovereign credit-ratings). We hope
and expect that their analysis will contribute to the creation of a more
effective system of crisis prevention in the future, and commend it to the
several international groups that are seeking to improve the monetary
system in that regard.
The Institute for International Economics is a private nonprofit institu-
tion for the study and discussion of international economic policy. Its
purpose is to analyze important issues in that area and develop and
communicate practical new approaches for dealing with them. The Insti-
tute is completely nonpartisan.

The Institute is funded largely by philanthropic foundations. Major
institutional grants are now being received from the William M. Keck, Jr.
Foundation and the Starr Foundation. A number of other foundations
and private corporations contribute to the highly diversified financial
resources of the Institute. About 26 percent of the Institute’s resources in
our latest fiscal year were provided by contributors outside the United
States, including about 11 percent from Japan. The Rockefeller Brothers
Fund provided generous financial support to this project.
The Board of Directors bears overall responsibility for the Institute and
gives general guidance and approval to its research program—including
the identification of topics that are likely to become important over the
medium run (one to three years), and which should be addressed by the
Institute. The Director, working closely with the staff and outside Advi-
sory Committee, is responsible for the development of particular projects
and makes the final decision to publish an individual study.
The Institute hopes that its studies and other activities will contribute
to building a stronger foundation for international economic policy
around the world. We invite readers of these publications to let us know
how they think we can best accomplish this objective.
C. Fred Bergsten
Director
June 2000
xi

INSTITUTE FOR INTERNATIONAL ECONOMICS
11 Dupont Circle, NW, Washington, DC 20036-1207
(202) 328-9000 Fax: (202) 328-5432
C. Fred Bergsten, Director

BOARD OF DIRECTORS ADVISORY COMMITTEE
*Peter G. Peterson, Chairman Richard N. Cooper, Chairman
*Anthony M. Solomon, Chairman,
Executive Committee Robert Baldwin
Barry P. Bosworth
Leszek Balcerowicz
Susan M. Collins
Conrad Black
Wendy Dobson
W. Michael Blumenthal
Juergen B. Donges
Chen Yuan
Rudiger Dornbusch
Jon S. Corzine
Gerhard Fels
George David
Isaiah Frank
Miguel de la Madrid
Jeffrey A. Frankel
*Jessica Einhorn
Jacob A. Frenkel
George M. C. Fisher
Stephan Haggard
Maurice R. Greenberg
David D. Hale
*Carla A. Hills
Dale E. Hathaway
Nobuyuki Idei
Nurul Islam
W. M. Keck II
John Jackson
Nigel Lawson
Peter B. Kenen
Lee Kuan Yew
Lawrence B. Krause
Donald F. McHenry
Anne O. Krueger
Minoru Murofushi
Paul R. Krugman
Suliman S. Olayan
Roger M. Kubarych
Paul H. O’Neill
Jessica T. Mathews
I. G. Patel
Rachel McCulloch
Karl Otto Pöhl
Isamu Miyazaki
*Joseph E. Robert, Jr.
Michael Mussa
David Rockefeller
Sylvia Ostry
Renato Ruggiero
Tommaso Padoa-Schioppa
Stephan Schmidheiny
Jacques J. Polak
Jean Claude Trichet
Dani Rodrik
Laura D’Andrea Tyson
Jeffrey D. Sachs
Paul A. Volcker
Joseph E. Stiglitz
*Dennis Weatherstone
Alan Wm. Wolff
Edward E. Whitacre, Jr.
Robert B. Zoellick
Marina v.N. Whitman
Lynn R. Williams
Peter K. C. Woo
Ex officio
*C. Fred Bergsten
Richard N. Cooper
Honorary Directors
Alan Greenspan
Reginald H. Jones
Frank E. Loy
George P. Shultz
*Member of the Executive Committee

Acknowledgments
The authors wish to thank Richard Cantor for sharing his database on
credit ratings and Andrew Berg, Michael Bordo, Stijn Claessens, Richard
Cooper, Susan Collins, Hali Edison, Barry Eichengreen, Cathy Mann, Peter
Montiel, Adam Posen, Vincent R. Reinhart, Ewoud Schuitemaker, Harry
Stordel, Holger Wolf, and participants at the Institute for International
Economics seminar for useful comments and suggestions. We are particu-
larly indebted to C. Fred Bergsten, not only for his extensive suggestions
on earlier drafts, but also for his unwavering confidence in the project.
Ian Anderson, Trond Augdal, Mark Giancola, and Neil Luna provided
superb research assistance.
xv

1
Introduction
This study analyzes and provides empirical tests of early warning indica-
tors of banking and currency crises in emerging economies. The aim is
to identify key empirical regularities in the run-up to banking and cur-
rency crises that would enable officials and private market participants
to recognize vulnerability to financial crises at an earlier stage. This, in
turn, should make it easier to motivate the corrective policy actions that
would prevent such crises from actually taking place. Interest in identify-
ing early warning indicators of financial crises has soared of late, stoked
primarily by two factors: the high cost to countries in the throes of crisis
and an increasing awareness of the insufficiency of the most closely
watched market indicators.
There is increasing recognition that banking and currency crises can
be extremely costly to the countries in which they originate. In addition,
these crises often spill over via a variety of channels to increase the
vulnerability of other countries to financial crisis.
According to one recent study, there have been more than 65 develop-
ing-country episodes during 1980-95, when the banking system’s capital
was completely or nearly exhausted;1 the public-sector bailout costs of
resolving banking crises in developing countries during this period have
1. See Caprio and Klingebiel (1996b). Other identifications of banking crises over this period
can be found in Demirgüç-Kunt and Detragiache (1998), Eichengreen and Rose (1998), IMF
(1998c), Kaminsky and Reinhart (1999), and Lindgren et al. (1996).

been estimated at around $250 billion.2 In more than a dozen of these
banking crises, the public-sector resolution costs amounted to 10 percent
or more of the country’s GDP.3 In the latest additions to the list of severe
banking crises, the cost of bank recapitalization for the countries most
affected in the ongoing Asian financial crisis is expected to be huge—on
the order of 58 percent of GDP for Indonesia, 30 percent for Thailand, 16
percent for South Korea, and 10 percent of GDP for Malaysia (World
Bank 2000).
In addition to the enormous fiscal costs, banking crises exacerbate
declines in economic activity, prevent national saving from flowing to its
most productive use, limit the room for maneuver in the conduct of
domestic monetary policy, and increase the chances of a currency crisis
as well (Lindgren, Garcia, and Saal 1996; Goldstein and Turner 1996).
Illustrative of the magnitude of output losses, an International Monetary
Fund study (IMF 1998c), drawing on a sample of 31 developing countries,
reported that it typically takes almost three years for output growth to
return to trend after the outbreak of a banking crisis and that the cumula-
tive output loss averaged 12 percent.4 Probably the main reason Mexican
authorities did not make more aggressive use of interest rate policy after
the assassination of presidential candidate Luis Donaldo Colosio in March
1994 is that bad loan problems in the banking system had by then already
become serious, and they were worried that recourse to higher interest
rates would push Mexican banks over the edge. Yet failure to increase
domestic interest rates in the face of international investors’ rising concern
contributed to a rapid decline in international reserves and helped to
transform a banking problem into a currency and debt crisis (Calvo and
Goldstein 1996). This pattern in the timing of the banking and currency
crises is not unique to the Mexican case. Drawing on a broader sample
of banking and currency crises in emerging economies, there is evidence
that the onset of a banking crisis typically precedes a currency crash
(Kaminsky and Reinhart 1999; IMF 1998a).5
2. This figure is net of the estimated amount of loans that were eventually repaid. See
Honohan (1997).
3. See Goldstein (1997) for a list of these severe banking crises. For comparison, the public-
sector tab for the US saving and loan crisis is typically estimated at about 2 to 3 percent of
US GDP.
4. In chapter 7, we present our own estimates of how long it takes growth rates of real
output to recover after banking and/or currency crises.
5. In chapter 3, we provide further evidence that the presence of a banking crisis is one of
the better leading indicators of a currency crisis in emerging economies. At the same time,
the evidence also suggests that a currency crash aggravates the problems in the banking
sector, as the peak of a banking crisis most often occurs following the collapse of the
currency. The dating of currency and banking crises is discussed in detail in chapter 2.
2 ASSESSING FINANCIAL VULNERABILITY

Table 1.1 Emerging Asia: real GDP growth forecasts, 1996-98
1998 Actual
Country 1996 1997 (as of May 1997) 1998
Indonesia 7.8 4.9 7.5 ⳮ13.7
Thailand 6.4 ⳮ1.3 7.0 ⳮ8.0
South Korea 7.1 5.0 6.3 ⳮ5.8
Malaysia 8.6 7.8 7.9 ⳮ6.7
The Philippines 5.7 5.1 6.4 ⳮ0.5
Hong Kong 4.9 5.3 5.5 ⳮ5.1
Source: International Monetary Fund, World Economic Outlook.
Although the contagion of financial disturbances usually runs from

large countries to smaller ones, the Asian financial crisis has shown that
severe financial-sector difficulties in even a relatively small economy
(namely Thailand) can have wide-ranging spillover effects if it acts as a
‘‘wake-up call’’ for investors to reassess country risk and if a set of other
economies have vulnerabilities similar to those in the economy first
affected.6
The costs of currency crises have likewise been shown to be significant
both in terms of reserve losses and output declines (see chapter 7). During
the Exchange Rate Mechanism (ERM) crises of the fall of 1992 and summer
of 1993, about $150 billion to $200 billion was spent on official exchange-
market intervention in a fruitless effort to stave off the devaluation and/or
floating of ERM currencies. Mexico’s peso crisis was accompanied in 1995
by a decline in real GDP of 6 percent—its deepest recession in 60 years.
In emerging Asia, consensus forecasts for 1998 growth issued just before
the crisis (that is, in May-June 1997) generally stood in the 6 to 8 percent
range. As indicated in table 1.1, these forecasts were subject to unprece-
dented downward revisions in the midst of the currency, banking, and
debt crises enveloping these economies. The IMF (1998c) estimates that
emerging economies suffer, on average, an 8 percent cumulative loss in
real output (relative to trend) during a severe currency crisis. And like
banking crises, currency crises seem contagious. One recent study found
that a currency crisis elsewhere in the world increases the probability of a
speculative attack by an economically and statistically significant amount
even after controlling for economic and political fundamentals in the
country concerned (Eichengreen, Rose, and Wyplosz 1996; see also Calvo
and Reinhart 1996; Kaminsky and Reinhart 2000).
The more costly it is to clean up after a financial crisis, the greater the
returns to designing a well-functioning early warning system.
6. See Calvo and Reinhart (1996) and Goldstein (1998a). Kaminsky and Reinhart (2000)
provide an analysis of contagion in the Asian crisis that stresses the financial links among
these countries—including the sudden withdrawal of funds by a common commercial bank
lender or mutual fund investor. See also chapter 6.
INTRODUCTION 3

4
Institute for International Economics
Table 1.2 Rating agencies’ performance before the Asian crisis: Moody’s and Standard & Poor’s long-term debt
ratings,a 1996-97
15 January 1996 2 December 1996 24 June 1997 12 December 1997
Rating Outlook Rating Outlook Rating Outlook Rating Outlook
Moody’s foreign currency debt
Indonesia Baa3 Baa3 Baa3 Baa3
Malaysia A1 A1 A1 A1
|
Mexico Ba2 Ba2 Ba2 Ba2

http://www.iie.com
The Philippines Ba2 Ba2 Ba2 Ba2

South Korea A1 A1 stable Baa2 negative
Thailand A2 A2 A2 Baa1 negative
Standard & Poor’s October 1997

Indonesia
Foreign currency debt BBB stable BBB stable BBB stable BBB negative
Domestic currency debt AⳭ AⳭ Aⳮ negative
Malaysia
Foreign currency debt AⳭ stable AⳭ stable AⳭ positive AⳭ negative
Domestic currency debt AAⳭ AAⳭ AAⳭ AAⳭ negative
Mexico
Foreign currency debt BB negative BB BB
Domestic currency debt BBBⳭ BBBⳭ stable BBBⳭ positive
The Philippines
Foreign currency debt BB positive BB positive BBⳭ positive BBⳭ stable

Domestic currency debt BBBⳭ BBBⳭ Aⳮ Aⳮ stable
South Korea
Foreign currency debt AAⳮ stable AAⳮ stable
Domestic currency debt
Thailand
Foreign currency debt A stable A stable A stable BBB negative
Domestic currency debt AA AA A negative
a. From highest to lowest, Moody’s rating system includes Aaa, Aa1, Aa2, Aa3, A1, A2, A3, Baa1, Baa2, Baa3, Ba1, Ba2, and Ba3, and Standard &
Poor’s runs AAA, AAⳭ, AA, AAⳮ, AⳭ, A, Aⳮ, BBBⳭ, BBB, BBBⳮ, BBⳭ, BB, and BBⳮ.
Source: Radelet and Sachs (1998).

|
http://www.iie.com
5
The second reason for the increased interest in early warning indicators
of financial crises is that there is accumulating evidence that two of the
most closely watched ‘‘market indicators’’ of default and currency risks—
namely, interest rate spreads and changes in credit ratings—frequently
do not provide much advance warning of currency and banking crises
(see chapter 4).
Empirical studies of the 1992-93 ERM crisis have typically concluded
that market measures of currency risk did not raise the specter of signifi-
cant devaluations of the weaker ERM currencies before the fact (Rose and
Svensson 1994). Another study, encompassing a larger number of crisis
episodes, similarly concluded that the currency forecasts culled from
survey data are useless in anticipating the crises (Goldfajn and Valdés
1998). In the run-up to the Mexican crisis, market signals were again
muted or inconsistent. More specifically, measures of default risk on
tesobonos (dollar indexed, Mexican government securities) jumped up
sharply in April 1994 (after the Colosio assassination) but stayed roughly
constant between then and the outbreak of the crisis (Leiderman and
Thorne 1996; Obstfeld and Rogoff 1995). From April 1994 on, market
expectations of currency depreciation on the peso usually were beyond
the government’s announced rate; nevertheless, this measure of currency
risk fluctuated markedly. The gap between market expectations and the
official rate was widest in summer of 1994, but the attack came with most
ferocity only in late December (Obstfeld and Rogoff 1995; Leiderman and
Thorne 1996; Rosenberg 1998).
The evidence now available suggests that the performance of interest
spreads and credit ratings was likewise disappointing in the run-up to
the Asian financial crisis. Examining interest rate spreads on three-month
offshore securities, one study found that these spreads gave no warning
of impending difficulties (i.e., were either flat or declining) for Indonesia,
Malaysia, and the Philippines and produced only intermittent signals for
Thailand (Eschweiler 1997b). A recent analysis of spreads using local
interest rates for South Korea, Thailand, and Malaysia similarly found
little indication of growing crisis vulnerability (Rosenberg 1998).
Sovereign credit ratings (on long-term, foreign-currency debt) issued
by the two largest international ratings firms were even less prescient in
the Asian crisis (see chapter 4, as well as Radelet and Sachs 1998; Goldstein
1998c).7 As shown in table 1.2, there were almost no downgrades for the
7. In a recent report, Moody’s (1998) argues that its rating record in the East Asian crisis
was better than it appears at first sight from ratings changes alone. More specifically, the
report argues, inter alia, that Moody’s went into the crisis with lower ratings for the crisis
countries than the other major ratings agencies (i.e., Standard & Poor’s and Fitch-IBCA),
that it took ratings actions before its main competitors, that its low bank financial strength
ratings identified many of the banks that subsequently experienced stress in the crisis
countries, that changes in sovereign credit ratings led to a widening of yield spreads in the
crisis countries, and that one should examine the sovereign research reports—not just the

most severely affected countries in the 18-month run-up to the crisis. As
The Economist (13 December 1997, p. 68) put it, ‘‘[I]n country after country,
it has often been the case of too little, too late.’’ Looking at a larger sample
of cases, a recent study by the Organization for Economic Cooperation
and Development (OECD) was unable to find consistent support for the
proposition that sovereign credit ratings act more like a leading than a
lagging indicator of market prices (that is, of interest rate spreads; see
Larraı́n, Reisen, and von Maltzan 1997).
Furthermore, international organizations such as the IMF did not do
better than the rating agencies in anticipating several of the recent crises.
A recent external evaluation of IMF surveillance concludes:
We found that the Fund—in both bilateral and multilateral surveillance—largely

failed to identify the vulnerabilities of the countries that subsequently found
themselves at the center of the Asian financial crisis, except in the case of Thailand.
In particular, it failed until rather late in the day to address a number of systemic
issues. Moreover, to the extent that surveillance did identify these vulnerabilities,
the tone of published Fund documents—notably [the World Economic Outlook]—
was excessively bland prior to the December 1997 update of WEO [and the
International Capital Markets Report], after the crisis had erupted. (IMF 1999, 56)
There are, of course, several reasons interest rate spreads or changes

in sovereign credit ratings may not anticipate financial crises well.8 For
one thing, market participants may not have timely, accurate, and compre-
hensive information on the borrower’s creditworthiness. Several recent
examples underscore the point (see also Goldstein 1998a; Corsetti, Pesenti,
and Roubini 1998; BIS 1998). Thailand’s commitments in the forward
exchange market and South Korea’s lending of international reserves to
commercial banks meant that official figures on gross international
reserves gave a misleading (i.e., overoptimistic) view of each country’s net
usable reserves. Similarly, external foreign-currency denominated debt of
Indonesian corporations, along with nonperforming bank loans in South
Korea, Thailand, Malaysia, and Indonesia, turned out to be considerably
larger than precrisis published official data suggested. Ceteris paribus,
one could conjecture that if the true size of liquid assets and liabilities
were known at an earlier stage, interest rate spreads would have been
higher and credit ratings would have been lower than actually observed
before the Asian crisis; this in turn could well have moderated the sharp
ratings—in looking for early warning signals. At the same time, the report acknowledges
that the firm is studying several potential enhancements to its analytical methodology to
help improve the predictive power of its sovereign ratings.
8. It is sometimes also argued that even when credit rating agencies or international financial
organizations (such as the IMF) conclude that crisis vulnerability is high, they will be
reluctant to go ‘‘public’’ with a downgrade or a warning for fear of being accused of
precipitating the crisis.
INTRODUCTION 7

change in market sentiment that was associated with the ‘‘news’’ of the
lower-than-expected net worth of Asian debtors.
The other reason market prices may not signal impending crises is that
market participants strongly expect the official sector—be it national or
international—to bail out a troubled borrower.9 In such cases, interest
rate spreads will reflect the creditworthiness of the guarantor—not that
of the borrower. Again, it is not difficult to find recent examples where
such expectations could well have impaired market signals. In Asian
emerging economies, several authors have argued that implicit and
explicit guarantees of financial institutions’ liabilities were important in
motivating the large net private capital inflows into the region in the
1990s. Others have emphasized that the disciplined fiscal positions of
these countries may have convinced investors that, should banks and
finance companies experience strains, governments would have the
resources to honor their guarantees.10
In the case of the Mexican peso crisis, it has similarly been argued that,
after the United States had agreed to the North American Free Trade
Agreement, or NAFTA, it would have been very costly for it to stand
by while Mexico either devalued the peso or defaulted on its external
obligations and that expectations of a US bailout blunted the operation of
early warning signals (Leiderman and Thorne 1996; Calvo and Goldstein
1996). Looking eastward, investments in Russian and Ukrainian govern-
ment securities have in recent years sometimes been known on Wall Street
as ‘‘the moral hazard play’’—reflecting the expectation that geopolitical
factors and security concerns would lead to a bailout of troubled borrow-
ers. Suffice it to say that the size and frequency of IMF-led international
financial rescue packages—including commitments of nearly $50 billion
for Mexico in 1994-95; over $120 billion for Thailand, Indonesia, and South
Korea in 1997-98; over $25 billion for Russia and Ukraine in 1998; and
another $42 billion for Brazil late that year—illustrate that market expecta-
tions of official bailouts cannot be dismissed lightly.
If interest rate spreads and sovereign credit ratings only give advance
warning of financial crises once in a while increased interest attaches to
the question of whether there are other early warning indicators that
would do a better job and if so, what they might be. This is a key question
for this book.
9. Michael P. Dooley has stressed this point in several papers (see Dooley 1997, for instance).
10. See Krugman (1998), Dooley (1997), and Calomiris (1997) on the role of expected national
and international bailouts in motivating capital flows and/or banking crises. Zhang (1999),
on the other hand, tests for such ‘‘moral hazard’’ effects in private capital flows to emerging
markets and finds no evidence for it. Claessens and Glaessner (1997) highlight the link
between fiscal positions and the wherewithal to honor explicit and implicit guarantees in
the financial sector. The Council on Foreign Relations (1999) offers a set of proposals on how
the moral hazard associated with international financial rescue packages might be reduced.

Organization of the Book
Chapter 2 takes up the leading methodological issues surrounding the
forecasting of crisis vulnerability, including the choice of sample countries,
the definition of currency and banking crises, the selection of leading
indicators, the specification of the early warning window, and the signals
approach to calculating optimal thresholds for indicators and the probabil-
ity of a crisis.
Chapter 3 presents the main empirical results for the in-sample estima-
tion (1970-95), with a focus on the best-performing monthly and annual
indicators, on a comparison of credit ratings and interest rate spreads
with indicators of economic fundamentals, and on the ability of the signals
approach to predict accurately previous currency and banking crises. In
chapter 4, we offer some preliminary results on the track record of rating
agencies in forecasting currency and banking crises.
In chapter 5, we use two overlapping out-of-sample periods (namely,
January 1996 through June 1997 and January 1996 through December 1997)
to project which emerging economies were recently the most vulnerable to
currency and banking crises. This exercise also permits us to gauge the
performance of the model in anticipating the Asian financial crisis. In
chapter 6, we analyze the contagion of financial crises across countries,
with particular emphasis on how fundamentals-based contagion is influ-
enced by trade and financial sector links. Chapter 7 examines data on the
aftermath of crises in order to assess how long it usually takes before
recovery from financial crises takes hold. Finally, chapter 8 summarizes
our main results and contains some brief concluding remarks, along with
suggestions for how the leading-indicator analysis of currency and bank-
ing crises in emerging economies might be improved.
INTRODUCTION 9

2
Methodology
Our approach to identifying early warning indicators of financial crises in

emerging economies reflects a number of decisions about the appropriate
methodology for conducting such an empirical exercise. Key elements of
our thinking are summarized in the following guidelines.
General Guidelines
First, finding a systematic pattern in the origin of financial crises means
looking beyond the last prominent crisis (or group of crises) to a larger
sample. Otherwise there is a risk either that there will be too many
potential explanations to discriminate between important and less impor-
tant factors or that generalizations and lessons will be drawn that do not
necessarily apply across a wider body of experience.1 We try to guard
against these risks by looking at a sample of 87 currency crises and 29
banking crises that occurred in a sample of 25 emerging economies and
smaller industrial countries over 1970-95.2
Several examples help to illustrate the point. Consider the last two
major financial crises of the 1990s: the 1994-95 Mexican peso crisis and
1. One can also view ‘‘early warning indicators’’ as a way to discipline or check more
‘‘subjective’’ and ‘‘idiosyncratic’’ assessments of crisis probabilities for particular econo-
mies—just as more comprehensive, subjective assessments can act as a check on the quality
of early warning indicator projections.
2. Our out-of-sample analysis spans 1996-97. Our criteria for defining a currency and a
banking crisis is described later in this chapter.
11

the 1997-99 Asian financial crisis. Was the peso crisis primarily driven by
Mexico’s large current account deficit (equal to almost 8 percent of its
GDP in 1994) and by the overvaluation of the peso’s real exchange rate,
or by the maturity and composition of Mexico’s external borrowing (too
short term and too dependent on portfolio flows), or by the uses to which
that foreign borrowing was put (too much for consumption and not
enough for investment), or by the already-weakened state of the banking
system (the share of nonperforming loans doubled between mid-1990 and
mid-1994), or by bad luck (in the form of unfortunate domestic political
developments and an upward turn in US international interest rates)? Or
was it driven by failure to correct fast enough earlier slippages in monetary
and fiscal policies in the face of market nervousness, or by a growing
imbalance between the stock of liquid foreign-currency denominated lia-
bilities and the stock of international reserves, or by an expectation on
the part of Mexico’s creditors that the US government would step in to
bail out holders of tesobonos?3
Analogously, was the Asian financial crisis due to the credit boom
experienced by the ASEAN-4 economies (Thailand, Indonesia, Malaysia,
and the Philippines), or a concentration of credit in real estate and equities,
or large maturity and currency mismatches in the composition of external
borrowing, or easy global liquidity conditions, or capital account liberal-
ization cum weak financial sector supervision? Was it the relatively large
current account deficits and real exchange rate overvaluations in the
run-up to the crisis, a deteriorating quality of investment, increasing
competition from China, global overproduction in certain industries
important to the crisis countries, or contagion from Thailand?4 There
are simply too many likely suspects to draw generalizations from two
episodes—even if they are important ones. To tell, for example, whether
a credit boom is a better leading indicator of currency crises than are,
say, current account deficits, we need to run a horse race across a larger
number of currency crises.5
Equally, but operating in the opposite direction, there is a risk of ‘‘jump-
ing the gun’’ by generalizing prematurely about the relative importance
of particular indicators from a relatively small set of prominent crises.
One example is credit booms—that is, expansions of bank credit that are
large relative to the growth of the economy. These have been shown to
3. See Leiderman and Thorne (1996) and Calvo and Goldstein (1996) for an analysis of the
Mexican crisis.
4. These alternative explanations of the Asian crisis are discussed in BIS (1998), Corsetti,
Pesenti, and Roubini (1998), Goldstein (1998a), Radelet and Sachs (1998), IMF (1997), and
World Bank (1998).
5. Some of these explanations, of course, are not mutually exclusive. For example, large
current account deficits may be the outcome of financial liberalization and its attendant
credit booms.

forerun banking crises in Japan, in several Scandinavian countries, and
in Latin America (Gavin and Hausman 1996). Yet when we compare
credit booms as a leading indicator of banking crises to other indicators
across a larger group of emerging economies and smaller industrial coun-
tries, we find that credit booms are outperformed by a variety of other
indicators. Put in other words, credit booms have been a very good leading
indicator in some prominent banking crises but are not, on average, the
best leading indicator in emerging economies more generally. Again, it
is helpful to have recourse to a larger sample of crises (in this study nearly
30) to sort out competing hypotheses.
The second guideline is to pay equal attention to banking crises and
currency crises. To this point, most of the existing literature on leading
indicators of financial crises relates exclusively to currency crises.6 Yet
the costs of banking crises in developing countries appear to be greater
than those of currency crises. Furthermore, banking crises appear to be
one of the more important factors in generating currency crises, and the
determinants and leading indicators of banking crises should be amenable
to the same type of quantitative analysis as currency crises are.7
Some policymakers have argued that, looking forward, the emphasis
in surveillance efforts should be directed to banking sector problems
rather than currency crises. The underlying assumption supporting that
view is that as more countries adopt regimes of managed floating, cur-
rency crises become a relic of the past. We believe this view to be overly
optimistic. It is noteworthy that among all the Asian countries that had
major currency crises in 1997-98 only Thailand had an ‘‘explicit pegged
exchange rate’’ policy. Indonesia, Malaysia, and South Korea were all
declared managed floaters, while the Philippines in principle (but not in
practice) had a freely floating exchange rate. Among emerging markets,
there is widespread ‘‘fear of floating,’’ and many of the countries that are
classified as floaters have implicit pegs, leaving them vulnerable to the
types of currency crises we study in this book.8
6. See Kaminsky, Lizondo, and Reinhart (1998) for a review of this literature. Among the
relatively few studies that include or concentrate on banking crises in emerging economies,
we would highlight Caprio and Klingebiel (1996a, 1996b), Demirgüç-Kunt and Detragiache
(1998), Eichengreen and Rose (1998), Furnam and Stiglitz (1998), Honohan (1997), Gavin
and Hausman (1996), Goldstein (1997), Goldstein and Turner (1996), Kaminsky (1998),
Kaminsky and Reinhart (1998, 2000), Rojas-Suarez (1998), Rojas-Suarez and Weisbrod (1995),
and Sundararajan and Baliño (1991).
7. Both Kaminsky and Reinhart (1998) and the IMF (1998c) conclude that the output costs
of banking crises in emerging economies typically exceed those for currency crises and that
these costs are greater still during what Kaminsky and Reinhart (1999) dubbed ‘‘twin crises’’
(that is, episodes when the country is undergoing simultaneous banking and currency
crises). We provide further empirical evidence on this issue in chapter 7.
8. See Calvo and Reinhart (2000) and Reinhart (2000) for a fuller discussion of this issue.
METHODOLOGY 13

We analyze banking and currency crises separately, as well as exploring
the interactions among them. As it turns out, several of the early warning
indicators that show the best performance for currency crises also work
well in anticipating banking crises. At the same time, there are enough
differences regarding the early warning process and in the aftermath of
crises to justify treating each in its own right.
A third feature of our approach—and one that differentiates our work
from that of many other researchers—is that we employ monthly data
to analyze banking crises as well as currency crises.9 Use of monthly
(as opposed to annual data) involves a trade-off. On the minus side,
because monthly data on the requisite variables are available for a smaller
number of countries than would be the case for annual data, the decision
to go with higher frequency data may result in a smaller sample. Yet
monthly data permit us to learn much more about the timing of early
warning indicators, including differences among indicators in the first
arrival and persistence of signals. Indeed, many of the annual indicators
that have been used in other empirical studies are only publicly available
with a substantial lag, which makes them plausible for a retrospective
assessment of the symptoms of crises but ill-suited for the task of provid-
ing an early warning. Hence, we conclude that the advantages of monthly
data seemed to outweigh the disadvantages.10 In the end, we were able
to assemble monthly data for about two-thirds of our indicator variables;
for the remaining third, we had to settle for annual data.
A fourth element of our approach was to include a relatively wide
array of potential early warning indicators. We based this decision on
a review of broad, recurring themes in the theoretical literature on financial
crises. These themes encompass
䡲 asymmetric information and ‘‘bank run’’ stories that stress liquidity/

currency mismatches and shocks that induce borrowers to run to liquid-
ity or quality,
䡲 inherent instability and bandwagon theories that emphasize excessive
credit creation and unsound finance during the expansion phase of the
business cycle,
䡲 ‘‘premature’’ financial liberalization stories that focus on the perils of
liberalization when banking supervision is weak and when an extensive
9. For example, the studies of banking crises in emerging markets by Caprio and Klingebiel
(1996a, 1996b), Goldstein and Turner (1996), Honohan (1995), and Sundararajan and Baliño
(1991) are primarily qualitative, while the studies by Demirgüç-Kunt and Detragiache (1997),
Eichengreen and Rose (1998), and the IMF (1998c) use annual data for their quantitative
investigation of the determinants of banking crises.
10. Private-sector ‘‘early warning’’ analyses likewise seem to be moving in the direction of
using monthly data. See Ades, Masih, and Tenegauzer (1998) and Kumar, Perraudin, and
Zinni (1998).

network of explicit and implicit government guarantees produces an
asymmetric payoff for increased risk taking,
䡲 first- and second-generation models of the vulnerability of fixed
exchange rates to speculative attacks,11 and
䡲 interactions of various kinds between currency and banking crises.
In operational terms, this eclectic view of the origins of financial crises

translates into a set of 25 leading indicator variables that span the real
and monetary sectors of the economy, that contain elements of both the
current and capital accounts of the balance of payments, that include
market variables designed to capture expectations of future events, and
that attempt to proxy certain structural changes in the economy (e.g.,
financial liberalization) that could affect vulnerability to a crisis.
Once a set of potential leading indicators or determinants of banking
and currency crises has been selected, a way has to be found both to
identify the better performing ones among them and to calculate the
probability of a crisis. In most of the existing empirical crisis literature,
this is done by estimating a multivariate logit or probit regression model
in which the dependent variable (in each year or month) takes the value
of one if that period is classified as a crisis and the value of zero if there
is no crisis. When such a regression is fitted on a pooled set of country
data (i.e., a pooled cross-section of time series), the statistical significance
of the estimated regression coefficients should reveal which indicators are
‘‘significant’’ and which are not, and the predicted value of the dependent
variable should identify which periods or countries carry a higher or
lower probability of a crisis.
A fifth characteristic of our approach is that we use a technique other
than regression to evaluate individual indicators and to assess crisis
vulnerability across countries and over time. Specifically, we adopt the
nonparametric ‘‘signals’’ approach pioneered by Kaminsky and Reinhart
(1999).12 The basic premise of this approach is that the economy behaves
differently on the eve of financial crises and that this aberrant behavior
has a recurrent systemic pattern. For example, currency crises are usually
preceded by an overvaluation of the currency; banking crises tend to
follow sharp declines in asset prices. The signals approach is given diag-
nostic and predictive content by specifying what is meant by an ‘‘early’’
warning, by defining an ‘‘optimal threshold’’ for each indicator, and by
choosing one or more diagnostic statistics that measure the probability
of experiencing a crisis.
11. First-generation models stress poor fundamentals as the cause of the currency crises,
while second-generation models focus on shifts in market expectations and self-fulfilling
speculative attacks. See Flood and Marion (1999) for a recent survey of this literature.
12. This approach is described in detail in Kaminsky, Lizondo, and Reinhart (1998).
METHODOLOGY 15

By requiring the specification of an explicit early warning window, the
signals approach forces one to be quite specific about the timing of early
warnings. This is not the case for all other approaches. For example, it
has been argued that an asymmetric-information approach to financial
crises implies that the spread between low- and high-quality bonds will
be a good indicator of whether an economy is experiencing a true financial
crisis—but there is no presumption that this interest rate spread should
be a leading rather than a contemporaneous indicator (Mishkin 1996).
Furthermore, the indicator methodology takes a comprehensive approach
to the use of information without imposing too many a priori restrictions
that are difficult to justify.
Finally, we use the signals to rank the probability of crises both across
countries and over time. We do so by calculating the weighted number
of indicators that have reached their optimal thresholds (that is, are ‘‘flash-
ing’’), where the weights (represented by the inverse of the individual
noise-to-signal ratios) capture the relative forecasting track record of the
individual indicators.13 Indicators with good track records receive greater
weight in the forecast than those with poorer ones. Ceteris paribus, the
greater the incidence of flashing indicators, the higher the presumed
probability of a banking or currency crisis. For example, if in mid-1997
we were to find that 18 of 25 indicators were flashing for Thailand versus
only 5 of 25 for Brazil, we would conclude that Thailand was more
vulnerable to a crisis than Brazil. Analogously, if only 10 of 25 indicators
were flashing for Thailand in mid-1993, we would conclude that Thailand
was less vulnerable in mid-1993 than it was in mid-1997. Thus we can
calculate the likelihood of a crisis on the basis of how many indicators
are signaling. Furthermore, as will be shown in chapter 5, we can attach
a greater weight to the signals of the more reliable indicators. Owing to
these features, the signals approach makes it easy computationally to
monitor crisis vulnerability. In contrast, the regression-based approaches
require estimation of the entire model to calculate crisis probabilities. In
addition, because these regression-based models are nonlinear, it becomes
difficult to calculate the contribution of individual indicators to crisis
probabilities in cases where the variables are far away from their means.14
13. While this is one of many potential ‘‘composite’’ indicators (i.e., ways of combining the
information in the individual indicators), Kaminsky (1998) provides evidence that this
weighting scheme shows better in-sample and out-of-sample performance than three alterna-
tives. Also, see chapter 5. One can equivalently evaluate the performance of individual
indicators by comparing their conditional probabilities of signaling a crisis.
14. Of course, ease of application is only one of many criteria for choosing among competing
crisis-forecasting methodologies. For example, the signals approach also carries the disad-
vantage that is less amenable to statistical tests of significance. In addition, some of the
restrictions it imposes (e.g., that indicators send a signal only when they reach a threshold)
may leave out valuable information.

Guideline number six is to employ out-of-sample tests to help gauge
the usefulness of leading indicators. The in-sample performance of a
model may convey a misleading sense of optimism about how well it
will perform out of sample. A good case in point is the experience of the
1970s with structural models of exchange rate determination for the major
currencies. While these models fit well in sample, subsequent research
indicated that their out-of-sample performance was no better—and often
worse—than that of ‘‘naive’’ models (such as using the spot rate or the
forward rate to predict the next period’s exchange rate; see Meese and
Rogoff 1983). In this study, we use data from 1970-95 to calculate our
optimal thresholds for the indicators, but we save data from 1996 through
the end of 1997 to assess the out-of-sample performance of the signals
approach, including the ability to identify the countries most affected
during the Asian financial crisis.
Our seventh and last guideline is to beware of the limitations of this
kind of analysis. Because these exercises concentrate on the macroeco-
nomic environment, they cannot capture political triggers and exogenous
events—the Danish referendum on the European Economic and Monetary
Union (EMU) in 1992, the Colosio assassination in 1994, or the debacle
over Suharto in 1997-98, for instance—which often influence the timing
of speculative attacks. In addition, because high-frequency data are not
available on most of the institutional characteristics of national banking
systems—ranging from the extent of ‘‘connected’’ and government-
directed lending to the adequacy of bank capital and banking supervi-
sion—such exercises cannot be expected to capture some of these longer-
term origins of banking crises.15 Also, because we are not dealing with
structural economic models but rather with loose, reduced-form relation-
ships, such leading-indicator exercises do not generate much information
on why or how the indicators affect the probability of a crisis. For example,
a finding that exchange rate overvaluation typically precedes a currency
crisis does not tell us whether the exchange rate overvaluation results
from an exchange rate-based inflation stabilization program or from a
surge of private capital inflows.
Nor is the early warning study of financial crises immune from the
‘‘Lucas critique’’: that is, if a reliable set of early warning indicators were
identified empirically, it is possible that policymakers would henceforth
behave differently when these indicators were flashing than they did in the
past, thereby transforming these variables into early warning indicators of
corrective policy action rather than of financial crisis. While this feedback
effect of the indicators on crisis prevention has apparently not yet been
strong enough to impair their predictive content, there is no guarantee
15. Indeed, for many countries, detailed data on the state of the banks may not even be
available annually.
METHODOLOGY 17

that this feedback effect will not be stronger in the future (particularly if
the empirical evidence in favor of robust early warning indicators was
subsequently viewed as more persuasive).
Much like the leading-indicator analysis of business cycles, we are
engaging here in a mechanical exercise—albeit one that we think is inter-
esting on a number of fronts. Moreover, this research is still in its infancy,
with many of the key empirical contributions coming only in the last two
to three years. In areas such as the modeling of contagion and alternative
approaches to out-of-sample forecasting, too few ‘‘horse races’’ have been
run to know which approaches work best. For all of these reasons, we see
the leading-indicator analysis of financial crises in emerging economies as
one among a number of analytical tools and not as a stand-alone, sure-
fire system for predicting where the next crisis will take place. That being
said, we also argue that this approach shows promising signs of generating
real value added and that it appears particularly useful as a first screen
for gauging the ordinal differences in vulnerability to crises both across
countries and over time. A family of estimated conditional crisis probabili-
ties will provide the basis of this ordinal ranking across countries at a
point in time or for a given country over time.
Putting the Signals Approach to Work
The signals approach described above was first used to analyze the
performance of macroeconomic and financial indicators around ‘‘twin
crises’’ (i.e., the joint occurrences of currency and banking crises) in Kamin-
sky and Reinhart (1999). We focus on a sample of 25 countries over 1970
to 1995. The out-of-sample performance of the signals approach will be
assessed using data for January 1996 through December 1997. These are
the countries in our sample:
䡲 Africa: South Africa

䡲 Asia: Indonesia, Malaysia, the Philippines, South Korea, Thailand
䡲 Europe and the Middle East: Czech Republic, Denmark, Egypt, Fin-
land, Greece, Israel, Norway, Spain, Sweden, Turkey
䡲 Latin America: Argentina, Bolivia, Brazil, Chile, Colombia, Mexico,
Peru, Uruguay, Venezuela
The basic premise of the signals approach is that the economy behaves
differently on the eve of financial crises and that this aberrant behavior
has a recurrent systematic pattern. This ‘‘anomalous’’ pattern, in turn, is
manifested in the evolution of a broad array of economic and financial
indicators. The empirical evidence provides ample support for this prem-

ise.16 To implement the signals approach, we need to clarify a minimum
number of two key concepts which will be used throughout the analysis.
Currency Crisis
A currency crisis is defined as a situation in which an attack on the

currency leads to substantial reserve losses, or to a sharp depreciation of
the currency—if the speculative attack is ultimately successful—or to
both. This definition of currency crisis has the advantage of being compre-
hensive enough to capture not only speculative attacks on fixed exchange
rates (e.g., Thailand’s experience before 2 July 1997) but also attacks that
force a large devaluation beyond the established rules of a crawling-peg
regime or an exchange rate band (e.g., Indonesia’s widening of the band
before its floatation of the rupiah on 14 August 1997.) Since reserve losses
also count, the index also captures unsuccessful speculative attacks (e.g.,
Argentina’s reserve losses in the wake of the Mexican 1994 peso crisis.)
We constructed an index of currency market turbulence as a weighted
average of exchange rate changes and reserve changes.17 Interest rates
were excluded, as many emerging markets in our sample had interest
rate controls through much of the sample. The index, I, is a weighted
average of the rate of change of the exchange rate, ⌬e/e, and of reserves,
⌬R/R, with weights such that the two components of the index have
equal sample volatilities:
I ⳱ (⌬e/e) ⳮ (␴e /␴R) * (⌬R/R) (2.1)
where ␴e is the standard deviation of the rate of change of the exchange

rate and ␴R is the standard deviation of the rate of change of reserves.
Since changes in the exchange rate enter with a positive weight and
changes in reserves have a negative weight attached, readings of this
index that were three standard deviations or more above the mean were
cataloged as crises.18
For countries in the sample that had hyperinflation, the construction
of the index of currency market turbulence was modified. While a 100
percent devaluation may be traumatic for a country with low to moderate
inflation, a devaluation of that magnitude is commonplace during hyper-
inflation. A single index for the countries that had hyperinflation episodes
would miss sizable devaluations and reserve losses in the moderate infla-
16. See Kaminsky, Lizondo, and Reinhart (1998) for a survey of this literature.
17. This index is in the spirit of that used by Eichengreen, Rose, and Wyplosz (1996), who
also included interest rate increases in their measure of turbulence.
18. Of course, for a study of market turbulence as well as crisis, one may wish to consider
readings in this index that are two standard deviations away from the mean.
METHODOLOGY 19

tion periods because the high-inflation episodes would distort the historic
mean. To avoid this, we divided the sample according to whether inflation
in the previous six months was higher than 150 percent and then con-
structed an index for each subsample.19
As noted in earlier studies that use the signals approach, the dates of
currency crises derived from this index map well onto the dates that
would be obtained if one were to define crises by relying exclusively on
events, such as the closing of the exchange markets or a change in the
exchange rate regime.
Banking Crises
Our dating of banking crises stresses events. This is because on the banking
side there are no time series comparable to international reserves and the
exchange rate. For instance, in the banking panics of an earlier era large
withdrawals of bank deposits could be used to date the crisis. In the wake
of deposit insurance, however, bank deposits ceased to be useful for
dating banking crises. As Japan’s banking crisis highlights, many modern
financial crises stem from the asset side of the balance sheet, not from
deposit withdrawals. Hence the performance of bank stocks relative to the
overall equity market could be an indicator. Yet in many of the developing
countries an important share of the banks are not traded publicly. Large
increases in bankruptcies or nonperforming loans could also be used to
mark the onset of the crisis. Indicators of business failures and nonper-
forming loans are, however, usually available only at low frequencies, if
at all; the latter are also made less informative by banks’ desire to hide
their problems for as long as possible.
Given these data limitations, we mark the beginning of a banking crisis
by two types of events: bank runs that lead to the closure, merging, or
takeover by the public sector of one or more financial institutions (as in
Venezuela in 1993); and if there are no runs, the closure, merging, takeover,
or large-scale government assistance of an important financial institution
(or group of institutions) that marks the start of a string of similar out-
comes for other financial institutions (as in Thailand in 1997). We rely on
existing studies of banking crises and on the financial press; according
to these studies the fragility of the banking sector was widespread during
these periods.
Our approach to dating the onset of the banking crises is not without
drawbacks. It could date the crises ‘‘too late’’ because the financial prob-
lems usually begin well before a bank is finally closed or merged. It could
also date crises ‘‘too early’’ because the worst of crisis may come later.
19. Similar results are obtained by looking at significant departures in inflation from a 6-
and 12-month moving average.

To address this issue we also indicate when the banking crisis hits its
peak, defined as the period with the heaviest government intervention
and/or bank closures.
Identifying the end of a banking crisis is one of the more difficult
unresolved problems in the empirical crisis literature—that is, there is no
consensus on what the criteria ought to be for declaring the crisis to be
over (e.g., resumption of normal bank lending behavior, or a marked
decrease in the share of nonperforming loans, or an end to bank closures
and large-scale government assistance). In our discussion of the aftermath
of crises in chapter 7, however, the end of a banking crisis is understood
to be its resolution (i.e., the end of heavy government financial interven-
tion), not when bank balance sheets cease to deteriorate.
Other empirical studies on banking crises have focused on annual data
and provide no information on the month or quarter in which banking
sector problems surface. Hence it is not possible to compare the exact
dates with our own analysis. We can, however, compare the dating of
the year of the crisis. In most cases, our dates for the beginning of crises
correspond with those found in other studies, but there are several
instances where our starting date is a year earlier than theirs. Tables 2.1
and 2.2 list the currency and banking crisis dates, respectively, for the 25
countries in our sample.
The Indicators
In addition to the 15 early warning indicators originally considered in

Kaminsky and Reinhart (1999), we evaluate the ability of nine additional
indicators that figure prominently in both the theoretical literature on
banking and currency crises and in the popular discussion of these events.
The indicators used in Kaminsky and Reinhart (1999) were international
reserves (in US dollars), imports (in US dollars), exports (in US dollars),
the terms of trade (defined as the unit value of exports over the unit value
of imports), deviations of the real exchange rate from trend (in percentage
terms),20 the differential between foreign (US or German) and domestic
real interest rates on deposits (monthly rates, deflated using consumer
prices and measured in percentage points), ‘‘excess’’ real M1 balances,
the money multiplier (of M2), the ratio of domestic credit to GDP, the
real interest rate on deposits (monthly rates, deflated using consumer
prices and measured in percentage points), the ratio of (nominal) lending
20. The real exchange rate is defined on a bilateral basis with respect to the German mark
for the European countries in the sample and with respect to the US dollar for all other
countries. The real exchange rate index is defined such that an increase in the index denotes
a real depreciation.
METHODOLOGY 21

Table 2.1 Currency crisis starting dates
Country Currency crisis
Argentina June 1975
February 1981*
July 1982
September 1986*
April 1989
February 1990
Bolivia November 1982
November 1983
September 1985
Brazil February 1983
November 1986*
July 1989
November 1990
October 1991
Chile December 1971
August 1972
October 1973
December 1974
January 1976
August 1982*
September 1984
Colombia March 1983*
February 1985*
Czech Republic May 1997
Denmark May 1971
June 1973
November 1979
August 1993
Egypt January 1979
August 1989
June 1990
Finland June 1973
October 1982
November 1991*
September 1992*
Greece May 1976
November 1980
July 1984
Indonesia November 1978
April 1983
September 1986
August 1997
Israel November 1974
November 1977
October 1983*
July 1984
Malaysia July 1975
August 1997*
(continued next page)

Table 2.1 (continued)
Country Currency crisis
Mexico September 1976
February 1982*
December 1982*
December 1994*
Norway June 1973
February 1978
May 1986*
December 1992
Peru June 1976
October 1987
The Philippines February 1970
October 1983*
June 1984
July 1997*
South Africa September 1975
July 1981
July 1984
May 1996
South Korea June 1971
December 1974
January 1980
October 1997
Spain February 1976
July 1977*
December 1982
February 1986
September 1992
May 1993
Sweden August 1977
September 1981
October 1982
November 1992*
Thailand November 1978*
July 1981
November 1984
July 1997*
Turkey August 1970
January 1980
March 1994*
Uruguay December 1971*
October 1982*
Venezuela February 1984
December 1986
March 1989
May 1994*
December 1995
* ⳱ twin crises
METHODOLOGY 23

Table 2.2 Banking crisis starting dates
K & R (1999) and G, K, & R C&K IMF (1996
Country (beginning) (1996) and 1998a & b)
Argentina March 1980 1980 1980
May 1985 1985 1985
1989
December 1994 1995 1995
Bolivia October 1987 1986 n.a.
Brazil November 1985 1990
December 1994 1994 1994
Chile 1976
September 1981 1981
Colombia July 1982 1982 1982
April 1998
Czech Republic 1994 n.a. n.a.
Denmark March 1987 n.a. 1988
Egypt January 1980 1980 1981
January 1990 1990 1990
Finland September 1991 1991 1991
Greece 1991 n.a. n.a.
Indonesia November 1992 1994 1992
1997
Israel October 1983 1977 1983
Malaysia July 1985 1985 1985
September 1997
(continued next page)
to deposit interest rates,21 the stock of commercial banks’ deposits (in

nominal terms), the ratio of broad money (converted into foreign currency)
to gross international reserves, an index of output, and an index of equity
prices (in US dollars). All these series are monthly. For greater detail, see
the appendix. The links between particular early warning indicators and
underlying theories of exchange rate and banking crises are discussed in
some detail in earlier papers (e.g., Kaminsky and Reinhart 1999).
Turning to the nine ‘‘new’’ indicators introduced here, four of them
are expressed as a share of GDP. These are the current account balance,
short-term capital inflows, foreign direct investment, and the overall bud-
21. This definition of the spread between lending and deposit rates is preferable to using
merely the difference between nominal lending and deposit rates because inflation affects
this difference and thus the measure would be distorted in the periods of high inflation.
An alternative would have been to use the difference between real lending and deposit rates.

Table 2.2 (continued)
K & R (1999) and G, K, & R C&K IMF (1996
Country (beginning) (1996) and 1998a & b)
Mexico September 1982 1981 1982
October 1992 1995 1994
Norway November 1988 1987 1987
Peru March 1983 n.a. 1983
Philippines January 1981 1981 1981
July 1997
South Africa December 1977 1977 1980
South Korea January 1986 n.a. 1983
July 1997 1997
Spain November 1978 1977 1977
Sweden November 1991 1991 1990
Thailand March 1979 1983 1983
May 1996 1997
Turkey 1982
January 1991 1992 1991
1994 1994
Uruguay March 1971
March 1981 1981 1981
Venezuela 1980 1980
October 1993 1994 1993
n.a. ⳱ not applicable

K & R ⳱ Kaminsky and Reinhart (1999)
G, K, & R ⳱ Goldstein, Kaminsky, and Reinhart
C & K ⳱ Caprio and Klingebiel (1996b)
get deficit. In addition, we look at the growth rates in the following

variables (the first three as shares in GDP and the fourth as a share of
investment): general government consumption, central bank credit to the
public sector, net credit to the public sector, and the current account
balance. The latter measure of the current account was motivated by the
view, particularly popular in the wake of the 1994-95 Mexican peso crisis,
that large current account deficits are more of a concern if they stem from
low saving as opposed to high levels of investment. Recent events in
Asia—a region noted for its exceptionally high levels of domestic saving
and its even higher levels of investment—have led to a reassessment of
that view. We also look at two measures of sovereign credit ratings. As
most of the new indicators are not available at monthly or quarterly
frequencies, annual data were used.
Table 2.3 provides a list of the indicators we examine in this book, their
periodicity, and the transformation used. In chapter 4, we examine the
METHODOLOGY 25

Table 2.3 Selected leading indicators of banking and currency crises
Indicator Transformation Data frequency
Real output 12-month growth rate Monthly
Equity prices 12-month growth rate Monthly
International reserves 12-month growth rate Monthly
Domestic/foreign real interest rate Level Monthly
differential
Excess real M1 balances Level Monthly
M2/ international reserves 12-month growth rate Monthly
Bank deposits 12-month growth rate Monthly
M2 multiplier 12-month growth rate Monthly
Domestic credit/GDP 12-month growth rate Monthly
Real interest rate on deposits Level Monthly
Ratio of lending interest rate to deposit Level Monthly
interest rate
Real exchange rate Deviation from trend Monthly
Exports 12-month growth rate Monthly
Imports 12-month growth rate Monthly
Terms of trade 12-month growth rate Monthly
Moody’s sovereign credit ratings 1-month change Monthly
Institutional Investor sovereign credit ratings Semiannual change Semiannual
General government consumption/GDP Annual growth rate Annual
Overall budget deficit/GDP Level Annual
Net credit to the public sector/GDP Level Annual
Central bank credit to public sector/GDP Level Annual
Short-term capital inflows/GDP Level Annual
Foreign direct investment/GDP Level Annual
Current account imbalance/GDP Level Annual
Current account imbalance/investment Level Annual
track record of sovereign credit ratings when it comes to ‘‘predicting’’

financial crises. Specifically, we examine the performance of the Institu-
tional Investor and Moody’s ratings.
As noted, in most cases we focus on 12-month changes in the variables.
This transformation has several appealing features. First, it eliminates
the nonstationarity problem of the variables in levels. It also makes the
indicators more comparable across countries and across time. Some of
the indicators have a strong seasonal pattern, which the 12-month transfor-
mation corrects for. For some indicators, such as equity prices, one could
contemplate using a measure of under- or overvaluation. However, the
empirical performance of most asset pricing models is not strong enough
to justify such an exercise.
For the monthly variables (with the exception of the deviation of the
real exchange rate from trend, the ‘‘excess’’ of real M1 balances, and the
three variables based on interest rates), the indicator on a given month
was defined as the percentage change in the level of the variable with
respect to its level a year earlier. This filter has several attractive features:
it reduces the ‘‘noisiness’’ of working with monthly data, it facilitates
cross-country comparisons, and it ensures the variables are stationary
with well-defined moments.

Turning to credit ratings, Institutional Investor constructs an index that
rises with increasing country creditworthiness and ranges from 0 to 100;
this index is published twice a year and is released in March and Septem-
ber.22 Hence we work with the six-month percentage change in this rating
index. For Moody’s Investor services, monthly changes in the sovereign
ratings are used. A downgrade takes on the value of minus one; no change
in the rating takes on a value of zero, and an upgrade takes on the value
of one. Since Moody’s ratings take on values from 1 to 16, we also worked
with changes in the ratings that took into account the magnitude of the
change. This issue will be discussed in greater detail in chapter 4.
The Signaling Window

Let us call a signal (yet to be precisely defined) a departure from ‘‘normal’’
behavior in an indicator.23 For example, an unusually large decline in
exports or output may signal a future currency or banking crisis. If an
indicator sends a signal that is followed by a crisis within a plausible
time frame we call it a good signal. If the signal is not followed by a crisis
within that interval, we call it a false signal, or noise. The signaling
window for currency crises is set a priori at 24 months preceding the crisis.
If, for instance, an unusually large decline in exports were to occur 28
months before the crisis, the signal would fall outside the signaling win-
dow and would be labeled a false alarm.
Alternative signaling windows (18 months and 12 months) were consid-
ered as part of our sensitivity analysis. While the results for the 18-month
window yielded similar results to those reported in this book, the 12-
month window proved to be too restrictive. Specifically, several of the
indicators we use here, including real exchange rates and credit cycles,
signaled relatively early (consistent with a protracted cycle), and the
shorter 12-month window penalized those early signals by labeling them
as false alarms.
For banking crises, we employ a different signaling window. Namely,
any signal given in the 12 months preceding the beginning of the crisis
or the 12 months following the beginning of the crisis is labeled a good
signal. The more protracted nature of banking crises and the high inci-
dence of denial by both bankers and policymakers that there are problems
in the banking sector motivate the more forgiving signaling window for
banking crises.
22. Since there are two readings of this index per year, in a typical year, say 1995, we would
have the percentage change in the rating from September 1994 to March 1995, from March
1995 to September 1995, and the change from September 1995 to March 1996.
23. Of course, normal behavior may change over time, hence, this approach, like other
commonly used alternatives (such as logit or probit) is not free from Lucas-critique limita-
tions. For further discussion of this issue, see Kaminsky and Reinhart (1999).
METHODOLOGY 27

The Threshold
Suppose we wish to test the null or maintained hypothesis that the econ-
omy is in a ‘‘state of tranquility’’ versus the alternative hypothesis that
a crisis will occur sometime in the next 24 months. Suppose that we
wish to test this hypothesis on an indicator-by-indicator basis. As in any
hypothesis test, this calls for selecting a threshold or critical value that
divides the probability distribution of that indicator into a region that is
considered normal or probable under the null hypothesis and a region
that is considered aberrant or unlikely under the null hypothesis—the
rejection region. If the observed outcome for a particular variable falls
into the rejection region, that variable is said to be sending a signal.
To select the optimal threshold for each indicator, we allowed the size
of the rejection region to oscillate between 1 percent and 20 percent. For
each choice, the noise-to-signal ratio was tabulated and the ‘‘optimal’’ set
of thresholds was defined as the one that minimized the noise-to-signal
ratio—that is, the ratio of false signals to good signals.24
Table 2.4 lists the thresholds for all the indicators for both currency
and banking crises. For instance, the threshold for short-term capital
flows as a percentage of GDP is 85 percent. This conveys two kinds of
information. First, it indicates that 15 percent of all the observations in
our sample (for this variable) are considered signals. Second, it highlights
that the rejection region is located at the upper tail of the frequency
distribution, meaning that a high ratio of short-term capital inflows to
GDP will lead to a rejection of the null hypothesis of tranquility in favor
of the alternative hypothesis that a crisis is brewing.
While the threshold or percentile that defines the size of the rejection
region is uniform across countries for each indicator, the corresponding
country-specific values are allowed to differ. Consider the following illus-
tration. There are two countries, one which has received little or no short-
term capital inflow (as a percentage of GDP) during the entire sample,
while the second received substantially larger amounts (also as a share
of GDP). The 85th percentile of the frequency distribution for the low
capital importer may be as small as a half a percent of GDP and any
increase beyond that would be considered a signal. Meanwhile, the coun-
try where the norm was a higher volume of capital inflows is likely to
have a higher critical value; hence only values above, say 3 percent of
GDP, would be considered signals.
24. For variables such as international reserves, exports, the terms of trade, deviations of
the real exchange rate from trend, commercial bank deposits, output, and the stock market
index, for which a decline in the indicator increases the probability of a crisis, the threshold
is below the mean of the indicator. For the other variables, the threshold is above the mean
of the indicator.

Table 2.4 Optimal thresholds (percentile)
Indicator Currency crisis Banking crisis
Bank deposits 15 20
Central bank credit to the public sector 90 90
Credit rating (Institutional Investor) 11 11
Current account balance/GDP 20 14
Current account balance/investment 15 10
Domestic credit/GDP 88 90
Interest rate differential 89 81
Excess M1 balances 89 88
Exports 10 10
Foreign direct investment/GDP 16 12
General government consumption/GDP 90 88
Imports 90 80
Lending-deposit interest rate ratio 88 87
M2 multiplier 89 90
M2/reserves 90 90
Net credit to the public sector/GDP 88 80
Output 10 14
Overall budget deficit/GDP 10 14
Real exchange ratea 10 10
Real interest rate 88 80
Reserves 10 20
Short-term capital inflows/GDP 85 89
Stock prices 15 10
Terms of trade 10 19
a. An increase in the index denotes a real depreciation.
Table 2.5 illustrates the ‘‘custom tailoring’’ of the optimal threshold by

showing the country-specific critical values for export growth and annual
stock returns for Malaysia, Mexico, and Sweden. A 25 percent decline in
stock prices would be considered a signal of a future currency crisis in
Malaysia and Sweden but not in Mexico, with the latter’s far greater
historical volatility.25
Figure 2.1 provides another illustration of the country-specific nature
of the optimal threshold calculations. It shows for the entire sample our
measure of the extent of overvaluation in the real exchange rate for Mexico.
The horizontal line is the country-specific threshold, and a reading below
this line (recall that a decline represents an appreciation) represents a
signal. The shaded areas are the 24 months before the crisis, or the signal-
ing window. Around 1982 the shaded area is wider due to the fact that
there was a ‘‘double dip,’’ with two crises registering. If the indicator
crossed the horizontal line and no crisis ensued in the following 24 months,
25. Indeed, as shown in Kaminsky and Reinhart (1998), the volatility pattern for these
three countries is representative of the broader historical regional pattern. The wild gyrations
in financial markets in Asia in 1997-99, however, may be unraveling those historic patterns.
METHODOLOGY 29

Table 2.5 Examples of country-specific thresholds: currency
crises
Critical value for exports Critical value for stock prices
Country (12-month percentage change) (12-month percentage change)
Malaysia ⳮ 9.05 ⳮ15.20
Mexico ⳮ13.10 ⳮ38.30
Sweden ⳮ11.25 ⳮ20.78
as it did in early 1992, it is counted as a false alarm. In the remainder of

this section we will define these concepts more precisely.
Signals, Noise, and Crises Probabilities
A concise summary of the possible outcomes is presented in the following

two-by-two matrix (for a currency crisis).
Crisis occurs in the No crisis occurs in the

following 24 months following 24 months
Signal A B
No signal C D
A perfect indicator would only have entries in cells A and D. Hence, with
this matrix we can define several useful concepts that we will use to
evaluate the performance of each indicator.
If one lacked any information on the performance of the indicators, it
is still possible to calculate, for a given sample, the unconditional probability
of crisis,
P(C) ⳱ (A Ⳮ C)/(A Ⳮ B Ⳮ C Ⳮ D) (2.2)
If an indicator sends a signal and that indicator has a reliable track record,
then it can be expected that the probability of a crisis, conditional on a signal,
P(C/S), is greater than the unconditional probability. Where
P(C 兩 S) ⳱ A/(A Ⳮ B) (2.3)
Formally,
P(C 兩 S) ⳮ P(C) ⬎ 0 (2.4)
The intuition is clear: if the indicator is not ‘‘noisy’’ (prone to sending

false alarms), then there are relatively few entries in cell B and P(C 兩 S)
⬇ 1. This is one of the criteria that we will use to rank the indicators in
the following chapters.

Figure 2.1 Mexico: real exchange rate, 1970-96
average of the sample ⳱ 100
|
http://www.iie.com
31
We can also define the noise-to-signal ratio, N/S, as
N/S ⳱ [B/(B Ⳮ D)]/[A/(A Ⳮ C)] (2.5)
It may be the case that an indicator has relatively few false alarms in its
track record. This could be the result of the indicator issuing signals
relatively rarely. In this case, there is also the danger that the indicator
misses the crisis altogether (it does not signal and there is a crisis). In this
case, we also wish to calculate for each indicator the proportion of crises
accurately called,
PC ⳱ C/(A Ⳮ C). (2.6)
In the next chapter, we employ these concepts to provide evidence on

the relative merits of a broad range of indicators in anticipating crises.

3
Empirical Results
The signals approach was applied to the indicators around the dates of
the 29 banking and the 87 currency crises. In what follows, we first
compare our results for the 15 monthly indicators to those presented in
Kaminsky and Reinhart (1999) and reproduced in table 3.1. In addition
to presenting our in-sample findings, this exercise allows us to gauge
robustness of the signals approach, since the results reported here are
derived from a larger sample of countries (25 versus 20.)1 Moreover, in
this chapter we report results for many of the indicators that have been
stressed in the financial press surrounding the coverage of the Asian crisis.
The Monthly Indicators: Robustness Check
Tables 3.1 and 3.2 summarize the in-sample performance of the monthly
indicators along the lines described in chapter 2 and presented in Kamin-
sky, Lizondo, and Reinhart (1998) and Kaminsky (1998). Table 3.1 covers
banking crises, and table 3.2 presents the results for currency crises. The
variables are shown in descending order based on their marginal predict-
ive power. For banking crises, for instance, the real exchange rate has the
greatest predictive power and imports the least. For each indicator, the
first column of the tables shows the noise-to-signal ratio. An indicator
with a noise-to-signal ratio of unity, such as those in the bottom of the
1. The five countries included here that were not a part of the Kaminsky and Reinhart
(1999) sample are the Czech Republic, Egypt, Greece, South Africa, and South Korea.
33

Table 3.1 Ranking the monthly indicators: banking crises
Percent Difference
Noise- of crises Rank in in rank
to- accurately P(C兩S)ⴑ Kaminsky (Ⳮ denotes an
Indicator signal called P(C兩S) P(C) (1998) improvement)
Real exchange 0.35 52 24.0 14.1 1 0
rate
Stock prices 0.46 76 23.4 11.2 3 0
M2 multiplier 0.46 63 18.3 9.0 4 0
Output 0.54 90 17.3 7.2 5 0
Exports 0.68 79 14.3 4.7 7 Ⳮ1
Real interest rate 0.68 96 16.8 4.2 6 ⳮ1
Real interest rate 0.73 100 15.6 3.7 8 0
differential
Bank deposits 0.73 64 12.9 3.1 9 0
M2/reserves 0.84 72 11.4 1.7 10 0
Excess real M1 0.88 44 11.0 1.2 13 Ⳮ2
balances
Domestic credit/ 0.89 46 10.9 1.1 11 ⳮ1
nominal GDP
Reserves 0.92 83 10.7 0.8 12 ⳮ1
Terms of trade 1.01 92 11.6 ⳮ0.1 14 0
Lending-deposit 1.48 56 8.3 ⳮ3.5 15 0
interest rate
Imports 1.75 64 6.0 ⳮ4.1 16 0
Sources: The authors and Kaminsky (1998).
tables, issues as many false alarms as good signals. The second column
shows the percent of crises (for which there were data for that indica-
tor) accurately called, while the third column lists the probability of a
crisis conditional on a signal from the indicator, P(C 兩 S). The fourth column
shows the difference between the conditional and unconditional probabili-
ties, P(C 兩 S) ⳮ P(C), the fifth column shows the ranking that the indicator
received in the previous signals approach analysis, and the last column
calculates the difference between its current and previous rank. Hence,
a Ⳮ3 in the last column would mean that the indicator moved up three
notches as the sample was enlarged, while a ⳮ2 would reflect a decline
in its ranking.
The indicators’ rankings based on their marginal predictive power are
shown under the heading P(C 兩 S) ⳮ P(C). The better the indicator, the
higher the probability of crisis conditioned on its signaling—that is, the
higher the P(C 兩 S)—and the bigger the gap between the conditional proba-
bility (P(C 兩 S) and the unconditional probability P(C). The unconditional
probability of a banking crisis (not shown) varies slightly from indicator to

Table 3.2 Ranking the monthly indicators: currency crises
Percent Difference in
Noise- of crises Ranking rank
to- accurately P(C兩S)ⴑ in K & R (Ⳮ denotes an
Indicator signal called P(C兩S) P(C) (1999) improvement)
Real exchange rate 0.22 58 62.1 35.2 1 0
Banking crisis 0.32 46.0 17.0 2 0
Stock prices 0.46 66 47.6 18.3 4 Ⳮ1
Exports 0.51 80 42.4 15.0 3 ⳮ1
M2/reserves 0.51 75 42.3 14.9 5 0
Output 0.57 71 43.0 12.5 6 0
Excess real M1 0.57 57 40.1 12.3 7 0
balances
Reserves 0.58 72 38.9 12.2 8 0
M2 multiplier 0.59 72 39.2 11.6 9 0
Domestic credit/ 0.68 57 35.6 8.3 10 0
nominal GDP
Terms of trade 0.74 77 35.4 6.5 11 0
Real interest rate 0.77 89 32.0 5.5 12 0
Imports 0.87 59 30.1 2.9 14 Ⳮ1
Real interest rate 1.00 86 26.1 ⳮ0.1 12 ⳮ1
differential
Lending-deposit 1.32 63 24.4 ⳮ4.8 16 Ⳮ1
interest rate
Bank deposits 1.32 43 22.3 ⳮ5.2 15 ⳮ1
K & R ⳱ Kaminsky and Reinhart (1999).
Sources: The authors and Kaminsky and Reinhart (1999).
indicator because of differences in data availability, since not all indicators

span the entire sample.2 For some indicators the sample is such that the
incidence of banking crises (i.e., their unconditional probability) is as low
as 9.8 percent or as high as 12 percent. For currency crises, the uncondi-
tional probability is clustered in the 27 to 29 percent range.
Several interesting features stand out from tables 3.1 and 3.2.
First, the ranking of the indicators appears to be quite robust across
sample selections, as shown in the last column of table 3.1. In other words,
the results from the 25-country sample closely match the results of the
20-country sample. For currency crises, none of the monthly indicators
changes in relative performance by more than one position as the sample
is enlarged, and for 10 of the indicators, there is no change at all. For
2. As shown in Kaminsky, Lizondo, and Reinhart (1998), the bigger the gap between the
conditional probability (P(C兩S) and the unconditional probability P(C), the lower the noise-
to-signal ratio.
EMPIRICAL RESULTS 35

banking crises, the maximum ranking change is two positions and 10 of
the monthly indicators show no change in their relative ranking. This is
a positive factor for the expected out-of-sample usefulness of the signals
approach. Specifically, it suggests that the indicators could well have a
similar relative predictive ability for countries that are not included in
the sample.3
Second, some of the most reliable indicators are the same for banking
and currency crises. Deviations of the real exchange rate from trend and
stock prices stand out in this regard. Close runners-up are output and
exports. A similar statement applies to the least useful indicators; imports
and the lending-deposit ratios, for example, do not have any predictive
ability for either type of crisis. Several of the low-scoring indicators also
carry the weakest or most ambiguous theoretical rationale.4
Third, there are some important differences in the ranking of indicators
between currency and banking crises. This suggests that currency and
banking sector vulnerability takes on different forms. A case in point is
the ratio of M2 (in dollars) to foreign exchange reserves, a variable stressed
by Calvo and Mendoza (1996) as capturing the extent of unbacked implicit
government liabilities. It does quite well (ranks fifth) among the 16 indica-
tors of currency crises, but it is far less useful when it comes to anticipating
banking crises. Similarly, the money multiplier, real interest rates, and
bank deposits are of little use when it comes to predicting currency crises
but do much better in predicting vulnerability to banking crises. This
result should not come as a surprise. Both the money multiplier and real
interest rates are strongly linked to financial liberalization, which itself
helps predict banking crises. As shown in Galbis (1993), real interest
rates tend to increase substantially in the wake of financial liberalization.
Furthermore, the steep reductions in reserve requirements that usually
accompany financial liberalization propel increases in the money multi-
3. We did not included the larger industrial countries (particularly the G-7 countries) in
our sample because they have characteristics (such as the ability to borrow in their own
currency, a relatively good external-debt servicing history, and high access to private capital
markets) that on a priori grounds would seem to make their crisis vulnerability different
from that of most emerging economies. In addition, data constraints, extremely large struc-
tural shifts over time, and difficulties associated with identifying a ‘‘normal’’ period led to
the decision to exclude China, Russia, and most of the transitional economies from the
sample. Finally, we excluded low-income developing countries from the sample because
we wanted to concentrate on emerging economies that had (in addition to the requisite
data availability) significant involvement with private international capital markets. In the
end, however, one can only tell whether our sample selection results in certain biases by
doing further robustness checks on alternative samples of countries.
4. For instance, lending-deposit interest rate spreads could widen in advance of a crisis
due to a deterioration in loan quality or a worsening in adverse selection problems. Alterna-
tively, it could be persuasively argued that ahead of financial crises, banks may be forced
to offer higher deposit rates, so as to stem capital flight.

plier. Bank runs and deposit withdrawals are at the heart of multiple-
equilibriums explanations of banking crises (Diamond and Dybvig 1983)
yet figure less prominently in explanations of currency crises.5
Lastly, banking crises are even more of a challenge to predict than
currency crises. For currency crises, the marginal predictive power of 12
of the 16 indicators (column five) is 5 percent or higher; for the real
exchange rate, marginal predictive power goes as high as 35 percent.
Indeed 9 of the 16 indicators have marginal predictive power in excess
of 10 percent. By way of contrast, for banking crisis 11 of the 15 indicators
have marginal predictive power of less than 5 percent, and even the top-
ranked macroeconomic indicators have marginal predictive power of less
than 15 percent. This relative inability of indicators to anticipate crises in
sample may be due to two factors. For one thing, for the earlier part of
the sample, banking crises were still relatively rare vis-à-vis currency
crises—there is a large discrepancy between the number of currency and
banking crises studied here. Detecting recurring patterns becomes more
difficult in the smaller sample of banking crises. Also, pinning down the
timing of a banking crisis requires a tricky judgment about when banking-
sector ‘‘distress’’ turns into a full-fledged crisis. As discussed in chapter 2,
the timing of currency crises is more straightforward.
The empirical evidence on the ‘‘predictability’’ of banking crises is
still limited to a handful of studies. Some have followed the approach
pioneered by Blanco and Garber (1986) for currency crises and have
attempted to model the probability of banking crises on the basis of
domestic and external fundamentals. These studies have encountered
some of the same problems highlighted in table 3.1—specifically, the
relatively poor predictive power of the models. Moreover, the results in
the studies sometimes conflict with one another. Eichengreen and Rose
(1998), for example, find that external conditions, specifically international
interest rates, play an important role in predicting banking crises. Real
exchange rate overvaluations, growth, and budget deficits have predictive
power in their regressions. The composition of external debt also seems
to matter. Other variables, including credit growth, they conclude, have
little or no predictive ability. In contrast, Demirgüç-Kunt and Detragiache
(1998) find no evidence in favor of budget deficits, while real interest
rates, credit growth, and M2/reserves figure prominently among their
significant regressors. Both studies do find, however, that slower economic
growth increases the probability of a banking crisis. In any case, it appears
that, to improve upon the ability to predict banking crises, we may need
to look beyond macroeconomic indicators—an issue that we take up later.
5. However, some recent models (Goldfajn and Valdés 1995) have highlighted the role of
bank runs in precipitating currency crises.

Table 3.3 Annual indicators: banking crises
Percent
of crises
Noise-to- accurately
Indicator signal called P(C兩S) P(C兩S)ⴑP(C)
Short-term capital inflows/GDP 0.38 43 36.8 18.5
Current account balance/ 0.38 38 36.1 18.4
investment
Overall budget deficit/GDP 0.47 52 26.9 12.1
Current account balance/GDP 0.50 33 29.3 12.1
Central bank credit to the public 0.52 23 23.8 7.6
sector/GDP
Net credit to the public sector/GDP 0.72 15 18.3 4.5
Foreign direct investment/GDP 1.05 24 15.6 ⳮ0.6
General government consumption/ 1.44 15 10.0 ⳮ3.8
GDP
The Annual Indicators: What Works?

Tables 3.3 and 3.4 present evidence on the performance of eight annual
indicators that have been prominent in recent discussions of the causes
of financial crises. The indicators include the fiscal variables stressed in
the Krugman (1979) model of a currency crisis as well as the short-term
debt exposure indicators stressed in recent theoretical and empirical expla-
nations of the Asian crisis (Calvo 1998; Calvo and Mendoza 1996; Gol-
dstein 1998b; Radelet and Sachs 1998). As before, the indicators are ranked
according to their marginal predictive power. The first column provides
information on the noise-to-signal ratio, the second column lists the per-
cent of crises accurately called, the third column provides information on
the probability of crisis conditional on signaling, while the last column
provides information on the marginal predictive power of the variable.
The top indicator for banking crises is the share of short-term capital
inflows to GDP. This is consistent with the results in Eichengreen and Rose
(1997) and supports the view that the banking sector becomes particularly
vulnerable during cycles of short-term capital inflows. Such short-term
inflows are more likely to be intermediated through the domestic banking
sector than other types of capital flows, such as foreign direct investment
(FDI) and portfolio flows. Indeed, the share of FDI/GDP does poorly as
a predictor of banking crises. Two of the fiscal variables—the budget
deficit and central bank credit to the public sector—do moderately well,

Table 3.4 Annual indicators: currency crises
Percent
of crises
Noise-to- accurately
Indicator signal called P(C兩S) P(C兩S)ⴑP(C)
Current account balance/GDP 0.41 56 43.2 19.5
Current account balance/ 0.49 31 39.0 15.1
investment
Overall budget deficit/GDP 0.58 22 36.4 11.5
Short-term capital inflows/ 0.59 29 35.2 10.9
GDP
General government 0.74 15 29.4 5.9
consumption/GDP
Net credit to the public 0.88 20 26.2 2.4
sector/GDP
Central bank credit to the 0.99 13 23.8 0.1
public sector/GDP
Foreign direct investment/ 1.00 24 21.7 0.1
GDP
while the third—government consumption—does poorly. Hence the role

of the public sector in fueling banking crises is somewhat mixed.
Without overinterpreting the results, it is interesting that the composi-
tion of the current account matters, in the sense that the current account
as a percentage of investment does better in predicting banking crises
than the current account as a share of GDP. It may be that investment is
more likely to be financed through the international issuance of bonds
and stocks or overseas loans, while consumption is more dependent on
local bank credit.
Turning to currency crises, the annual indicators that perform best are
those measuring current account imbalances. This finding is not represen-
tative of the broader empirical literature. As discussed in Kaminsky,
Lizondo, and Reinhart (1998), most of the studies that have attempted to
explain the k-period ahead probability of a currency crisis have had mixed
results regarding the current account, with most studies finding it
insignificant.
The various fiscal indicators do moderately well in anticipating currency
crises, lending some support to Krugman-type models. By contrast with
banking crises, the composition of capital inflows appears to have rela-
tively little to add to our understanding of what drives a currency crisis.
This result, however, may in part be due to the fact that a large share of

Table 3.5 Short-term debt: selected countries, June 1997 (percent)
Short-term Short-term
Country debt/total debt debt/reserves
Asia
Indonesia 24 160
Malaysia 39 55
Philippines 19 66
South Korea 67 300
Thailand 46 107
Latin America
Argentina 23 108
Brazil 23 69
Chile 25 44
Colombia 19 57
Mexico 16 126
Sources: Bank for International Settlements; International Financial Statistics; World Bank.
the currency crises (as opposed to the banking crises) took place in the
1970s in an environment of highly regulated internal and external financial
markets, where portfolio flows were negligible.
While our list of indicators is comprehensive, it is by no means exhaus-
tive. The Asian crisis in particular highlighted the importance of currency
and maturity mismatches in increasing vulnerability to currency and
banking crises. Table 3.5 presents an indicator of the imbalance between
liquid liabilities and liquid assets: namely, the ratio of short-term debt to
international reserves. All the emerging economies in this group with
debt-to-reserves levels in excess of 100 percent in mid-1997 have been
casualties of financial turmoil in recent years (even if not all the speculative
attacks ultimately succeeded, as in the case of Argentina.) This suggests
that variables such as short-term debt to reserves could be a valuable
addition to our list of leading indicators of crisis vulnerability.6
Do the Indicators Flash Early Enough?

The previous discussion has ranked the indicators according to their
ability to anticipate crises while producing few false alarms. Such criteria,
however, do not speak to the lead time of the signal. From the vantage
point of a policymaker or financial market participant who wants to
implement preemptive or risk-mitigating measures, it is not a matter of
6. See Calvo and Mendoza (1996) for an early discussion of this issue. We did not use the
ratio of short-term debt to reserves as an indicator in our tests because its relevance was
highlighted mainly by the Asian crisis and we did not want the out-of-sample tests to be
biased by its inclusion. In addition, the data were not available for the early part of our sample.

Table 3.6 How leading are the signals? (average number of months
from when the first signal is issued to the crisis month)
Indicator Currency crisis Banking crisis
Bank deposits 15 8
Beginning of banking crisis 19 n.a.
Domestic credit/GDP 12 7
Domestic-foreign interest rate differential 14 16
Excess M1 balances 15 6
Exports 15 16
Imports 16 11
Lending-deposit interest rate ratio 13 6
M2 multiplier 16 12
M2/reserves 13 14
Output 16 13
Real exchange rate 17 10
Real interest rate 17 16
Reserves 15 10
Stock prices 14 12
Terms of trade 15 18
indifference whether an indicator sends a signal well before the crisis

occurs or if the signal is given only when the crisis is imminent. Consider
for example, the Conference Board’s composite indices of business cycle
activity for the United States, which are published on a monthly basis. Both
financial market participants and policymakers alike find the leading-
indicator composite index more valuable than the coincident and lagging
indices. Market participants incorporate this information in their invest-
ment decisions, while policymakers give it weight in their policy reactions.
Over the years, US monetary policy has become increasingly forward-
looking and hence preemptive rather than reactive. One could argue that
this transition was facilitated by an improvement in our understanding
of the business cycle and early signs of its turning points.
In what follows, we tabulate for each of the monthly indicators the
average number of months before the crisis when the first signal occurs.
This, of course, does not preclude the indicator from giving signals
through the entire period immediately preceding the crisis. Indeed, for
the more reliable indicators, signals tend to become increasingly persistent
ahead of crises. For the low-frequency (annual) indicators, lead time is
not much of an issue since some of these are published with a considerable
lag and hence tend to be of less use from an early warning standpoint.
Table 3.6 presents the lead times for our monthly indicators—both for
currency and banking crises. In the case of currency crises, the most
striking observation is that, on average, all the indicators send the first
signal anywhere between a year and 18 months before the crisis erupts,
with banking-sector problems (our second-ranked indicator) offering the

Table 3.7 Microeconomic indicators: banking crises
Percentage of crises
Indicator accurately called Noise-to-signal
Bank lending-deposit interest rate spread 73 0.28
Interbank debt growth 80 0.35
Interest rate on deposits 80 0.47
Rate of growth on loans 58 0.72
Net profits to income 60 1.14
Operating costs to assets 40 1.59
Change in banks’ equity prices 7 2.00
Risk-weighted capital-to-asset ratio 7 2.86
Source: Rojas-Suarez (1998).
longest lead time—namely, 19 months. The average lead time for these
early signals is 15 months for currency crises. All the indicators considered
are therefore best regarded as leading rather than coincident, which is
consistent with the spirit of an ‘‘early warning system.’’ For banking
crises, there is a greater dispersion in the lead time across indicators, and
the average lead time is also lower (about 11 months). Furthermore, most
of the indicators signal at about the same time, thus the signaling is
cumulative and all the more compelling. Thus, on the basis of these
preliminary results, there does appear to be adequate lead time for pre-
emptive policy actions to avert crises.
Microeconomic Indicators: Selective Evidence

If, as the previous discussion suggests, banking crises are more difficult
to predict on the basis of macroeconomic indicators than currency crises,
it appears that the analysis of banking crises may benefit from including
a variety of microeconomic indicators of bank health. Gonzales-Hermosi-
llo et al (1997) and Rojas-Suarez (1998) provide some insights in this
direction. Rojas-Suarez uses bank-specific data from Colombia, Mexico,
and Venezuela and applies the ‘‘signals’’ methodology to this data to
glean which items in bank balance sheets are most useful in predicting
banking distress.
Her results are summarized in table 3.7. They do indeed suggest that
bank-specific information could make an important contribution in assess-
ing the vulnerability of the banking sector in emerging markets. More
‘‘traditional’’ indicators, such as liquidity ratios and bank capitalization,
turn out to be less useful indicators in Rojas-Suarez’s tests, in large part
because they are ‘‘noisy’’ and likely to send many false alarms while
missing many of the problem spots. At the other end, bank spreads and
the interest rate that banks offer on deposits appear to systematically
identify the weak banks.

One possible explanation for why interest rate spreads at the micro
level may be more useful indicators of banking crisis than aggregate
spreads is that the latter may reflect mainly cross-country differences in
the extent of banking competition. In contrast, micro spreads are more
likely to be more informative about a bank’s risk taking, as all banks
within a country are apt to face a more common competitive environment.
Goldstein (1998b) stresses bank exposure to the property sector as an
indicator in the context of banking crises. He notes that in many of the
affected Asian countries, estimates of the share of bank lending to the
property sector exceeded 25 percent. Banking sector external exposure,
measured in terms of foreign liabilities as a percentage of foreign assets,
also appears to be a worthy addition to the list of sectoral or microeco-
nomic indicators of banking-sector problems.

4
Rating the Rating Agencies and
the Markets
The discussion in the preceding chapters focused on the ability of a variety

of indicators to signal distress and to pinpoint the vulnerability of an
economy to banking or currency crises. In this chapter, we assess the
ability of sovereign credit ratings to anticipate such crises. In addition,
given the wave of sovereign credit ratings downgrades that have followed
the crisis in Asia, we investigate formally the extent to which credit ratings
are reactive. Along the way, we discuss a small but growing literature
that examines the extent to which financial markets anticipate crises.
Do Sovereign Credit Ratings Predict Crises?
We attempt to evaluate the predictive ability of sovereign credit ratings

using two approaches. First, we tabulate the descriptive statistics for the
ratings along the lines of the ‘‘signals’’ approach and compare how these
stack up to the other leading indicators we have analyzed. Second, we
follow the approach taken in much of the literature on currency and,
more recently, banking crises and estimate a probit model. Specifically,
we estimate a series of regressions where the dependent variable is a
crisis dummy that takes on the value of one if there is a crisis and zero
otherwise and where the explanatory variable is the credit ratings.
Our exercise is very much in the spirit of Larraı́n, Reisen, and von
Maltzan (1997), who, using Granger causality tests, assess whether credit
ratings lead or follow market sentiment as reflected in interest rate differ-
entials. These interest rate differentials reflect the ease or difficulty with
45

Table 4.1 Comparison of Institutional Investor sovereign ratings
with indicators of economic fundamentals
Percent of Difference between
crises conditional and
Noise-to- accurately unconditional
Type of crisis and indicator signal called probability
Currency crisis
Institutional Investor 1.05 31 5.4
sovereign rating
Average of the top five 0.45 70 19.1
monthly indicators
Average of the top three 0.49 36 15.4
annual indicators
Banking crisis
Institutional Investor 1.62 22 0.9
sovereign rating
Average of the top five 0.50 72 9.1
monthly indicators
Average of the top three 0.41 44 16.3
annual indicators
which sovereign countries can tap international financial markets. In their

analysis, they focus on the sovereign ratings of Moody’s and Standard &
Poor’s; in what follows, we examine the behavior around financial crises
of sovereign credit ratings issues by Moody’s Investor Service and Institu-
tional Investor (II).
The II sample begins in 1979 and runs through 1995. This gives us the
opportunity to study 50 currency crises and 22 banking crises. There are
20 countries in this sample, with 32 observations per country for a total
of 640 observations.1 For the Moody’s ratings, we have an unbalanced
panel.2 Here there are 21 currency crises and 7 banking crises. Because
the II database encompasses a more comprehensive sample of crises, we
will place more emphasis on these results.
Table 4.1 presents the basic descriptive statistics that we used in chapter
3 to gauge an indicator’s ability to anticipate crises: namely, the noise-to-
signal ratio, the percentage of crises accurately called, and the marginal
predictive power (i.e., the difference between the conditional and uncondi-
tional probabilities). We compare II sovereign ratings to averages for the
more reliable monthly and annual indicators of economic fundamentals.
1. The 20 countries are those in the Kaminsky and Reinhart (1999) sample: Argentina,
Bolivia, Brazil, Chile, Colombia, Denmark, Finland, Indonesia, Israel, Malaysia, Mexico,
Norway, Peru, the Philippines, Spain, Sweden, Thailand, Turkey, Uruguay, and Venezuela.
2. An unbalanced panel, in this case, refers to the fact that we do not have the same number
of observations for all the countries.

The basic story that emerges from table 4.1 is that the credit ratings
perform much worse for both currency and banking crises than do the
better indicators of economic fundamentals. The noise-to-signal ratio is
higher than one for both types of crises, suggesting a similar incidence
of good signals and false alarms. Hence, not surprisingly, the marginal
contribution to predicting a crisis is small relative to the top indicators;
for banking crises the marginal contribution is nil. Furthermore, the per-
centage of crises called is well below those of the top indicators. Indeed,
the II ratings compare unfavorably with even the worst indicators. For
example, consider the performance of the terms of trade ahead of banking
crises (shown in the last row of table 3.1). The terms of trade has a noise-
to-signal ratio of about one, making it almost as noisy as the credit rating.
Yet the terms of trade accurately called 92 percent of the crises in sample—
so while it sends many false alarms, it misses few crises. The II ratings,
on the other hand, score poorly on both counts as, in addition to being
noisy, they miss anywhere between two-thirds and three-quarters of the
crises, depending on which type of crisis we focus on.
Next we assess the predictive ability of ratings via probit estimation.
The dependent variable is a crisis dummy (banking and currency crises
are considered separately), and the independent variable is the change
in the credit rating in the preceding 12 months. The II ratings are allowed
to enter with a lag. The basic premise underpinning the simple postulated
model is as follows. If the credit rating agencies are using all available
information on the economic fundamentals to form their rating decisions,
then credit ratings should help predict crises because (as shown in the
preceding chapter) macroeconomic indicators have some predictive
power and the simple model should not be misspecified—that is, other
indicators should not be statistically significant, since that information
would already presumably be reflected in the ratings themselves. Thus
the state of the macroeconomic fundamentals should be captured in a
single indicator—the ratings.
Recent studies that have examined the determinants of credit ratings
do provide support for the basic premise that ratings are significantly
linked with selected economic fundamentals (Lee 1993; Cantor and Packer
1996a). For example, Cantor and Packer (1996a) find that per capita GDP,
inflation, the level of external debt, and indicators of default history and
of economic development are significant determinants of sovereign rat-
ings. The question we seek to answer is whether these are the ‘‘right’’ set
of fundamentals when it comes to predicting financial crises.
Table 4.2 presents the results of the probit estimation, using both the
II ratings and the Moody’s ratings as regressors. The results shown in
table 4.2 are based on the 12-month change in the ratings, but alternative
time horizons ranging from 6-month changes to 18- and 24-month changes
produced very similar results.3 The method of estimation corrected for
3. These results are not reported here but are available from the authors.
RATING THE RATING AGENCIES AND THE MARKETS 47

Table 4.2 Do ratings predict banking crises? (probit estimation with
robust standard errors)
Independent Standard Marginal Pseudo
variable Coefficient error effects Probability R2
12-month change in ⳮ0.921 1.672 ⳮ0.070 0.421 0.004
the Institutional
Investor rating
Moody’s rating
Table 4.3 Do ratings predict currency crises? (probit estimation with

robust standard errors)
the Institutional
Investor rating
12-month change in ⳮ0.22* 0.101 ⳮ0.009 0.013 0.021
Moody’s rating
*Denotes significance at the 5 percent level.
serial correlation and for heteroscedasticity in the residuals. For banking

crises, the coefficients of the credit ratings have the anticipated negative
sign—that is, an upgrade reduces the probability of a crisis. However,
neither of the two credit-rating variables is statistically significant, and
their marginal contribution to the probability of a banking crisis is very
small.
These results would, on the surface, be at odds with the findings of
Larraı́n, Reisen, and von Maltzan (1997), who find evidence that ratings
‘‘cause’’ interest rate spreads. Our interpretation, however, is that, while
ratings may systematically lead yield spreads (they present evidence of
two-way causality)—yield spreads are poor predictors of crises, as high-
lighted in tables 3.1 and 3.2. Hence the inability of ratings to explain crises
is not inconsistent with the ability to influence spreads. This issue will
be taken up later in this section.
The analogous exercise for currency crises is reported in table 4.3. Again,
the estimated coefficients of the ratings have the anticipated negative
sign. Only in the case of the Moody’s rating, however, is the coefficient
statistically significant at standard confidence levels. Even there, its mar-
ginal contribution to the probability of a currency crisis is quite small: a
one-unit downgrade in the Moody’s rating increases the probability of a
currency crisis by about 1 percent. The fact that the II ratings behave
differently from Moody’s in the probit regression is consistent with other

research on ratings performance. Cantor and Packer (1996b), for example,
provide extensive evidence of rating agency disagreements.
Why Do Credit Ratings Fail to Anticipate

Crises?
As discussed in chapter 1, credit ratings and interest rate spreads may
fail to anticipate a crisis either because lenders do not have access to
timely and comprehensive information on the creditworthiness of the
borrower or because lenders expect an official bailout of a troubled sover-
eign borrower. We now take up two related issues that could be associated
with the poor performance of credit ratings as predictors of financial crises.
The first one relates to the distinction between default and financial
crises. The credit rating agencies themselves often argue that sovereign
credit ratings are meant to provide an assessment of the likelihood of
sovereign default. Hence to the extent that a domestic banking crisis or
a currency crisis is decoupled from the probability of sovereign default,
credit ratings should not a priori be expected to predict currency or banking
crises. For example, the three Nordic countries included in our sample
had both currency and banking crises in the 1990s, yet default on sovereign
debt was never a likely event.
Whatever the merits of this argument for industrialized economies, it
looks less persuasive for developing countries and transition economies—
where many default episodes have been preceded by banking and/or
currency crises.4 Latin America’s experience in the early 1980s attests to
this pattern. Furthermore, had it not been for large-scale rescue packages
under the auspices of the International Monetary Fund (IMF), Mexico
in 1994-95 and Indonesia, South Korea, and Thailand in 1997-98 would
probably have been new additions to this list.
While more research on this default versus crisis distinction is war-
ranted, one simple implication is that the rating agencies should do better
in predicting currency and banking crises in developing countries (since
financial crises are more closely linked to the probability of sovereign
default there than in industrial countries). To examine this issue empiri-
cally, we reestimated our simple model of crises and ratings, excluding
industrial countries from the sample. The results, shown in tables 4.4 and
4.5, are not appreciably different from those for the full sample. For
banking crises, neither of the ratings variables are statistically significant.5
4. The IMF’s World Economic Outlook of April 2000 notes that sovereign risk and devaluation
tend to move together in the case of emerging economies.
5. We do not place much weight on the Moody’s results, as the number of banking crises
is very small. In future work, it would be interesting to test whether Moody’s bank financial
strength ratings (which are meant to capture the health of banks independent of the likelihood
of a government bailout) are better predictors of banking crises. However, as these ratings

Table 4.4 Do ratings predict banking crises for emerging markets?
(probit estimation with robust standard errors)
the Institutional
Investor rating
Moody’s rating
Table 4.5 Do ratings predict currency crises for emerging

markets? (probit estimation with robust standard errors)
the Institutional
Investor rating
12-month change in ⳮ0.34* 0.161 ⳮ0.026 0.011 0.041
Moody’s rating
*Denotes significance at the 5 percent level.
For currency crises, the II ratings remain insignificant, while Moody’s

ratings are significant but with a quite small marginal effect (below 3 per-
cent).
The second issue challenges the basic notion that credit ratings should
be expected to be leading indicators. Because rating agencies receive fees
from the borrowers they rate and because downgrades can subject the
agencies to charges of having precipitated a crisis, some have argued—
including in the Asian crisis—that credit ratings are apt to behave as
lagging indicators of crises, with downgrades coming on the heels of
crises. The anecdotal evidence surrounding the events in Asia seems to
point in this direction (table 4.6). Only in the case of Thailand did there
appear to be any substantive anticipatory action downgrades.
To examine this issue more formally, we test whether the presence of
a banking or currency crisis helps to predict credit rating downgrades.
In other words, our dependent and independent variables now switch
roles. For the II ratings, where we have available a continuous time series,
we regress the six-month change in the credit rating index on the financial
crisis dummy lagged by six months.6 The method of estimation is general-
were only introduced in 1995, the time series is not yet long enough to encompass many
banking crises.
6. We want to examine whether the rating changes follow immediately after the crisis, but
as the index is only published twice a year this ability to discriminate is not possible.

Table 4.6 Rating agencies’ actions on the eve and aftermath of the
Asian crises, June-December 1997
Country Date Events and action taken
Hong Kong 20-23 October Stock market plunges as speculators attack
Hong Kong dollar.
30 October Moody downgrades Hong Kong banks on
concerns about property exposure.
Malaysia 14 July Ringit falls as central bank abandons support.
18 August S&P cuts ratings from ‘‘positive’’ to ‘‘stable.’’
25 September S&P changes outlook to negative.
South Korea 8 June S&P and Moody change outlook from ‘‘stable’’ to
‘‘negative.’’
22 October Finance minister announces small-scale bank
bailout and government takeover of Kia Motors.
24 October S&P downgrades government debt.
27 October Moody downgrades government debt.
27 November Moody downgrades ratings.
3 December IMF pact.
10 December Moody downgrades ratings.
11 December S&P downgrades ratings.
Thailand September 1996 Moody downgrades short-term government debt.
February 1997 Moody puts government debt under review.
April Moody cuts ratings but still rates government
debt ‘‘investment grade.’’ S&P reaffirms rating.
June S&P reaffirms rating.
13 August Government seeks IMF bailout; S&P puts ratings
under review.
3 September S&P cuts rating to Aⳮ.
October Moody and S&P downgrade government debt.
4 November Prime minister resigns.
27 November Moody lowers ratings to near junk.
Source: Goldstein (1998a).
ized least squares, correcting for heteroskedasticity and serial correlation

in the residuals. For the Moody’s ratings, the dependent variable is three-
month changes in the rating, while the explanatory variable is the financial
crisis dummy lagged three months. The latter specification should allow
us to glean more precisely whether downgrades follow rapidly after crises
take place. For the Moody’s ratings, our dependent variable assumes the

Table 4.7 Do financial crises help predict credit rating
downgrades? (dependent variable ⳱ II 6-month changes in
sovereign rating, estimated using OLS with robust standard
errors)
Independent Standard Probability Pseudo
variable Coefficient error value R2
Banking crisis ⳮ0.009 0.019 0.61 0.01
dummy
Currency crisis ⳮ0.04* 0.014 0.005 0.06
dummy
Table 4.8 Do financial crises help predict credit rating

downgrades? (dependent variable ⳱ Moody’s 3-month
changes in sovereign rating, estimated using ordered probit)
Independent Standard Probability Pseudo
variable Coefficient error value R2
Banking crisis ⳮ0.11 0.90 0.901 0.001
dummy
Currency crisis ⳮ0.27* 0.14 0.048 0.02
dummy
* ⳱ Significant at the 5 percent level.
value of minus one, zero, or one depending on whether there was a

downgrade, no change, or an upgrade, respectively. We therefore estimate
the Moody’s ratings regression with an ordered probit technique.
The results of the estimation are summarized in tables 4.7 and 4.8. For
banking crises, the historical experience through 1995 does not support the
proposition that credit-rating agencies behaved in a ‘‘reactive’’ manner.
In contrast, our results suggest that currency crises help predict credit
downgrades for both the Institutional Investor and Moody’s ratings. That
is, as the explanatory variable increases (from zero to one when there is
a crisis), ratings fall. However, while the coefficients are significant at
standard confidence levels, their marginal predictive contribution is small.
For example, in the case of Moody’s ratings, a currency crisis increases
the likelihood of a downgrade by only 5 percent.
Our results are consistent with the findings of Larraı́n, Reisen, and
von Maltzan (1997), who find evidence of two-way causality between
sovereign ratings and market spreads. That is, not only do markets react
to changes in the ratings, but the ratings systematically react (with a lag)
to market sentiment.
Do Financial Markets Anticipate Crises?

The empirical tests presented here on sovereign credit ratings and finan-
cial crises need to be supplemented with tests on the ratings and larger

samples to determine whether our findings are robust. Nevertheless, we
do not find surprising the evidence suggesting that sovereign ratings fail
to anticipate banking and currency crises and are instead adjusted ex
post. Kaminsky and Reinhart (1999) show that domestic-foreign interest
rate differentials are not good predictors of crises, particularly currency
crises. This result is reenforced in tables 3.1 and 3.2. Similarly, Goldfajn
and Valdés (1998) use survey data on exchange rate expectations (culled
from the Financial Times) to test whether market expectations are adept
at foreseeing financial crises. Using a broad array of crises definitions and
approaches, their answer is a negative one and much along the lines of
what we have found for the sovereign ratings. We concluded tentatively
that if one is looking for early market signals of crises, it would be better
to focus on equity returns rather than market exchange rate expectations
and sovereign ratings.

5
An Assessment of Vulnerability:
Out-of-Sample Results
As emphasized in chapter 1, predicting the timing of currency and banking

crises is likely to remain an elusive task for academics, financial market
participants, and policymakers. Recent events, however, have highlighted
the importance of improving upon a system of early warnings. In this
chapter, we apply the signals approach to several out-of-sample exercises
using data for January 1996 through June 1997. Besides providing an
assessment of the model’s out-of-sample performance, this exercise may
shed light on why most analysts did not foresee the Asian crisis.
In the first exercise, we look at measures across countries of crisis
vulnerability (e.g., total number of signals, proportion of indicators signal-
ing, and the number of top indicators signaling). But this exercise does
not weigh the signals according to the relative track record of the indica-
tors issuing the signal, or it only does so in a very approximate way.
The second exercise extends the cross-country analysis by adjusting the
threshold for each indicator so as to include more borderline signals in
our measure of vulnerability. A third exercise weighs the indicators by
the inverse of their noise-to-signal ratio to generate a series of cross-
country vulnerability ratings for both currency and banking crises. In yet
a fourth exercise, we construct a composite indicator to map the time-
varying probability of crisis; we compare its in- and out-of-sample perfor-
mance to that of a naive forecast and the best of the univariate indicators.
Finally, our last exercise focuses on the time-series dimension by mapping
out the probability of crises for four Asian countries over the January
1996-December 1997 period.
Needless to say, such exercises are fraught with the traditional Type I
and Type II errors. Assume that the null hypothesis is that the economy
55

is in a state of ‘‘tranquility.’’ If a high proportion of indicators are flashing,
then one could reject that hypothesis in favor of the alternative—namely,
that a crisis is likely in the next 24 months. Yet even though a country
may be vulnerable, in the sense that a high proportion of variables are
signaling trouble, the crisis may be averted through either good luck,
good policies, or credible implicit bailout guarantees. This would be an
example of a Type II error (rejecting the null hypothesis when it is true).
A recent example of this case is Brazil, in which multiple signals were
flashing as early as 1997, but these warning signs did not culminate in a
full-fledged crisis until 1999. Alternatively, the crisis may occur without
much warning from the indicators; this is a Type I error (failing to reject
the null hypothesis when it is false). Borrowing a phrase from Sherlock
Holmes, such a situation can be regarded as ‘‘the dog that did not bark
in the night’’ and could be interpreted as evidence of contagion or multi-
plicity of equilibriums, an issue that we take up in chapter 6 and one that
is particularly relevant for understanding the Indonesian crisis.
Vulnerability and Signals
Table 5.1 shows how our 25 sample countries compare on vulnerability

to currency crises over the June 1996-June 1997 period, using several
simple measures of vulnerability. The first column shows the total number
of signals from among the 15 monthly indicators listed in table 3.1 that
‘‘flashed’’ during the period. The next column indicates how many of the
15 indicators sent signals, while the third data column lists the number
of ‘‘top five’’ indicators sending signals. (For banking crises, these are
real exchange rates, stock prices, the money multiplier, output, and
exports, and for currency crises, they are real exchange rates, stock prices,
exports, M2/reserves, and output.). The next set of columns give the
comparable information for the eight annual indicators. In this case, we
focus on the ‘‘top three’’ indicators. (For banking crises, the share in
GDP of short-term capital inflows, current account balance as a share of
investment, and the overall budget deficit as a share of GDP, and for
currency crises, they are the current account balance as a share of GDP,
the current account balance as a share of investment, and the overall
budget deficit as a share of GDP.) The last column gives the percentage
of the 23 indicators that are signaling. The reason to highlight the number
of top indicators signaling is that these are the indicators with the lowest
noise-to-signal ratios; hence a signal from these is more meaningful than
a signal from a less reliable indicator.
Table 5.1 provides this information for currency crises using the thresh-
olds reported in table 3.2. There is considerable cross-country variation,
with the lowest proportion of signals coming from Egypt and the highest

Table 5.1 Signals of currency crises, June 1996-June 1997
Monthly indicators Annual indicators Total
Number of Top Number of Top Percentage

Total indicators indicators Total indicators indicators of indicators
Country signals signaling signaling signals signaling signaling signaling
Argentina 35 3 1 2 2 0 22
Bolivia 33 6 2 0 0 0 26
Brazil 37 5 3 0 0 0 22
Chile 34 2 1 1 1 0 13
Colombia 27 5 1 3 3 0 35
Czech Republic 77 10 3 4 2 2 52
Denmark 21 3 1 1 1 0 17
Egypt 14 3 2 0 0 0 13
Finland 74 7 1 2 2 0 39
Greece 32 8 2 3 2 0 43
Indonesia 6 3 1 1 1 0 17
Israel 24 4 1 1 1 0 22
Malaysia 36 9 3 0 0 0 39
|
Mexico 11 2 0 2 2 0 17
http://www.iie.com
Norway 9 3 0 1 1 0 17
Peru 16 2 0 1 1 0 13
The Philippines 59 8 1 2 2 0 43
South Africa 42 8 3 1 1 0 39
South Korea 32 8 3 3 3 0 48
Spain 44 6 2 1 1 0 30
Sweden 55 5 1 1 1 0 26
Thailand 50 6 3 1 1 1 30
Turkey 22 4 1 3 3 0 30
Uruguay 58 5 0 1 1 0 26
Venezuela 18 5 2 1 1 0 26
57
from the Czech Republic, which indeed floated following a speculative
attack and substantial reserve losses in May 1997.
Table 5.2 repeats the same accounting exercise, but here we include
‘‘borderline’’ signals. Specifically, we enlarged the size of the rejection
region by 5 percent for all the indicators. For instance, instead of having
a 10 percent threshold for stock prices, we now have a 15 percent threshold.
This sensitivity analysis increases the likelihood of making a Type II error
(rejecting the null hypothesis of tranquility when you should not) while
reducing the probability of a Type I error (not rejecting when you should).
Including borderline signals does not seem to generate large shifts in the
most and least vulnerable groups. As shown in the last column in table
5.2, borderline signals do not alter the picture at all for some countries
(such as Argentina), but they do markedly increase the proportion of
indicators signaling, as well as the number of signals, for countries such
as South Korea (from 48 to 65 percent) and South Africa (from 39 to
52 percent).
Tables 5.3 and 5.4 report the results for banking crises using the original
thresholds and the ‘‘borderline’’ scenario, respectively. The country pro-
files that emerge are similar to those for currency crises; this may reflect
the fact that several of the indicators have common thresholds for currency
and banking crises.
While conveying useful information on vulnerability, the preceding
analysis does not fully discriminate between the more and less reliable
indicators. Kaminsky (1998) shows how to construct a ‘‘composite index’’
to gauge the probability of a crisis conditioned on multiple signals from
various indicators; the more reliable indicators receive a higher weight
in this composite index. This methodology and its out-of-sample results
are described in the remainder of this chapter.
In weighting individual indicators, a good argument can be made for
eliminating from our list of potential leading indicators those variables
that had a noise-to-signal ratio above unity; this is tantamount to stating
that their marginal forecasting ability, P(C 兩 S) ⳮ P(C), is zero or less.
Applying this criterion to banking crises, the lending-deposit ratio, the
terms of trade, government consumption growth, and FDI as a share of
GDP should be dropped. For currency crises, the excluded indicators are
the domestic-foreign interest rate differential, the lending-deposit ratio,
bank deposits, central bank credit to the public sector, and FDI as a share
of GDP. For the remaining indicators with noise-to-signal ratios below
unity, we weighed the signals by the inverse of the noise-to-signal ratios
reported in tables 3.1 through 3.4. For a currency crisis, suppose that both
the real exchange rate and imports are issuing a signal. Because the real
exchange rate has a very low noise-to-signal ratio (0.22), it would receive
a weight of 4.55 (i.e., 1/0.22); in contrast, with a relatively high noise-to-
signal ratio (0.87), imports would receive a weight of only 1.49 (i.e., 1/0.87).

Table 5.2 Borderline signals of currency crises, June 1996-June 1997

Argentina 35 3 0 2 2 0 22
Bolivia 33 8 2 0 0 0 35
Brazil 39 7 3 0 0 0 32
Chile 40 5 2 1 1 0 26
Colombia 49 7 3 3 3 0 43
Denmark 28 5 1 2 2 0 30
Egypt 22 3 2 1 1 0 17
Finland 86 8 1 3 2 1 43
Greece 41 8 3 3 3 0 48
Indonesia 9 4 3 1 1 0 22
Israel 37 6 3 1 1 0 30
Malaysia 40 9 3 0 0 0 39
|
Mexico 24 2 0 2 2 0 17
http://www.iie.com
Norway 31 7 2 1 1 0 35
Peru 26 5 1 2 2 0 30
South Africa 63 10 3 2 2 0 52
South Korea 63 11 3 5 4 1 65
Spain 55 7 2 1 1 0 35
Sweden 60 6 1 1 1 0 30
Thailand 54 6 3 1 1 1 30
Turkey 33 5 3 3 3 0 35
Uruguay 71 5 1 1 1 0 26
Venezuela 29 5 2 1 1 0 26
59
60
Table 5.3 Signals of banking crises, June 1996-June 1997


Argentina 36 4 0 1 1 1 22
Bolivia 42 8 2 0 0 0 35
Brazil 39 6 2 0 0 0 26
Chile 34 2 1 1 1 0 13
Colombia 38 5 2 3 3 1 35
Denmark 24 4 0 1 1 0 22
Egypt 18 3 0 0 0 0 13
Finland 77 7 1 3 2 1 39
Greece 39 8 3 2 2 1 43
Indonesia 10 3 2 1 1 1 17
Israel 32 6 3 1 1 0 30
Malaysia 42 9 3 0 0 0 39
|
Mexico 16 4 1 2 2 0 26
http://www.iie.com
Norway 30 8 2 1 1 0 39
Peru 19 5 1 1 1 0 26
South Africa 55 10 3 1 1 0 43
South Korea 42 10 4 3 3 1 57
Spain 51 7 1 1 1 0 35
Sweden 59 5 1 1 1 0 26
Thailand 53 6 2 1 1 1 30
Turkey 27 5 3 2 2 0 30
Uruguay 74 5 1 1 1 0 26
Venezuela 18 4 1 2 2 0 26
Table 5.4 Borderline signals of banking crises, June 1996-June 1997

Argentina 46 7 1 1 1 1 35
Bolivia 45 8 2 0 0 0 35
Brazil 44 9 3 0 0 0 39
Chile 43 7 3 1 1 0 35
Colombia 70 9 3 3 3 1 52
Denmark 29 6 1 1 1 0 30
Egypt 24 3 0 0 0 0 13
Finland 88 8 1 3 2 1 43
Greece 50 9 4 2 2 1 48
Indonesia 14 4 3 1 1 1 22
Israel 49 8 4 1 1 0 39
Malaysia 49 9 3 0 0 0 39
|
Mexico 30 5 1 2 2 0 30
http://www.iie.com
Norway 37 8 2 1 1 0 39
Peru 27 6 2 1 1 0 30
South Africa 68 10 3 1 1 0 48
South Korea 74 13 5 3 3 1 61
Spain 58 9 2 1 1 0 43
Sweden 66 8 2 1 1 0 39
Thailand 58 9 3 1 1 1 43
Turkey 35 6 3 2 2 0 35
Uruguay 86 6 1 1 1 0 30
Venezuela 34 6 2 2 2 0 35
61
Table 5.5 Weighting the signals for currency and banking crises in
emerging markets, June 1996-June 1997
Currency crises Banking crises
Country Weighted signals Rank Weighted signals Rank
Argentina 5.41 16 7.98 10
Bolivia 6.59 12 7.30 13
Brazil 7.57 10 6.08 14
Chile 5.90 15 5.74 16
Colombia 10.59 8 11.87 6
Czech Republic* 15.42 2 17.24 1
Egypt 6.02 14 8.33 9
Greece 14.27 6 14.15 3
Indonesia* 7.54 11 8.33 9
Israel 6.30 13 10.38 8
Malaysia* 12.46 7 7.74 12
Mexico 2.82 19 2.59 19
Peru 2.82 19 5.33 17
The Philippines* 14.40 5 11.52 7
South Africa 16.52 1 12.74 4
South Korea* 14.57 4 14.55 2
Thailand* 14.63 3 12.09 5
Turkey 8.21 9 7.87 11
Uruguay 4.40 18 4.88 18
Venezuela 5.28 17 6.02 15
Note: An asterisk (*) denotes the country had a currency crisis, a banking crisis, or both in
1997-98.
Formally, we construct the following composite indicator,

n
It ⳱ 兺 S /␻
j⳱1
j
t
j
(5.1)
In equation 5.1, it is assumed that there are n indicators. Each indicator

has a differentiated ability to forecast crises, and as before, this ability
can be summarized by the noise-to-signal ratio, here denoted by ␻ j. S jt is
a dummy variable that is equal to one if the univariate indicator, Sj crosses
its critical threshold and is thus signaling a crisis and is zero otherwise.
As before, the noise-to-signal ratio is calculated under the assumption that
an indicator issues a correct signal if a crisis occurs within the following 24
months. All other signals are considered false alarms.
If all 18 good indicators were sending signals, the maximum value that
this composite vulnerability index could score is 30.05 for banking crises
and 33.23 for currency crisis. This score is a simple sum of the inverse of the
noise-to-signal ratios for the good indicators that are retained. However, it
is seldom the case that every indicator signals. Table 5.5 presents the
composite score of the indicators that are signaling for the 20 emerging

Table 5.6 Vulnerability to financial crises in emerging markets:
alternative measures, June 1996-June 1997
Average
Average proportion of
proportion of top eight
indicators indicators Average of
signaling both signaling both ‘‘weighted’’
Country crises Rank crises Rank signals Rank
Argentina 29 11 11 5 6.69 14
Bolivia 35 8 22 4 6.94 12
Brazil 36 7 33 3 6.82 13
Chile 31 9 33 3 5.74 17
Colombia 48 4 44 2 11.23 7
Czech Republic* 57 2 56 1 16.33 1
Egypt 15 15 11 5 6.42 15
Greece 48 4 44 2 14.21 4
Indonesia* 22 14 44 2 7.93 11
Israel 35 8 44 2 8.34 9
Malaysia* 39 6 33 3 10.10 8
Peru 30 10 22 4 4.08 19
The Philippines* 46 5 33 3 12.96 6
Mexico 24 13 11 5 2.71 20
South Africa 50 3 33 3 14.63 2
South Korea* 63 1 56 1 14.56 3
Thailand* 35 8 44 2 13.36 5
Turkey 35 8 33 3 8.04 10
Uruguay 28 12 11 5 4.88 18
Venezuela 31 9 22 4 6.02 16
Note: An asterisk (*) denotes the country had a currency crisis, a banking crisis, or both in
1997-98.
economies in our sample; currency and banking crises are treated sepa-
rately. The first data column provides the relevant value of the index for
a currency crisis. The next column shows the country’s ordinal ranking
for the vulnerability index relative to the remaining 19 countries. South
Africa, the Czech Republic, and Thailand emerge as the most vulnerable
on the basis of the signals issued and the quality of those signals during
January 1996-June 1997.
For banking crises, the comparable exercise ranks the Czech Republic,
South Korea, and Greece as the most vulnerable. Perhaps not surprisingly,
near the bottom of the list are countries such as Mexico and Venezuela,
which are still recovering from their 1994-95 crises.
Thus far, we have treated banking and currency crises separately in
our vulnerability rankings. If one wanted to assess the ‘‘average’’ vulnera-
bility to both banking and currency crises, one may want to combine
the information contained in these two measures. Table 5.6 provides
information on the average proportion of indicators signaling banking
OUT-OF-SAMPLE RESULTS 63

and currency crises, the average proportion of the top eight indicators
(monthly and annual) that are signaling, and the average of the ‘‘weigh-
ted’’ indices reported in table 5.5 for currency and banking crises. The
table also ranks the countries, by these three criteria, depending on the
degree of ‘‘vulnerability.’’
Concentrating on the average of the ordinal rankings derived from the
weighted signals (last column of table 5.6), we can see that clustered at
the top of the list are several of the countries that have had or are still
undergoing financial crises; these countries are denoted by an asterisk.
This suggests a relatively encouraging out-of-sample performance for the
signals approach. The three measures of vulnerability provide similar
rankings for most of the ‘‘extreme’’ cases, such as the Czech Republic,
South Korea, Malaysia, and the Philippines among the countries that have
already had crises and South Africa, Colombia, and Greece among those
that have not. In the case of Greece, however, there was an orderly devalu-
ation, while in Colombia’s case there was both a devaluation (in August
1998) as well as serious banking sector difficulties. For countries such as
Thailand and to a lesser degree Indonesia, taking into account the ‘‘qual-
ity’’ of the indicator that is signaling considerably changes the overall
ranking.
The Composite Indicator and Crises

Probabilities
While the foregoing exercise allows us to assess the relative propensity

to crisis across countries at a point in time—like a snapshot—it does not
convey information on the dynamics of the process. To assess the extent
to which a country is becoming more or less vulnerable to crisis over
time, one would need a continuum of such snapshots. To do so, it is
convenient to link the composite index to the implied probability of crisis.
Once we construct this composite indicator, we can then proceed—as
we did with the individual indicators in chapters 2 and 3—to choose a
critical value for the composite indicator so that when the composite
indicator crosses this threshold, a crisis is deemed to be imminent.1 As
before, this critical threshold could be chosen so as to minimize the noise-
to-signal ratio of the composite indicator. Moreover, we could calculate
the probability of a crisis conditional on the composite indicator signaling
a crisis (i.e., crossing the critical threshold) as well as the odds of a crisis
when the composite indicator is not signaling. However, this procedure
would not give us an exhaustive reading of vulnerability as the crisis
approaches because it is dichotomous—that is, it will only provide two
1. Meaning, as in the individual indicators, in the subsequent 24 months.

types of information—namely, signal or no signal. We want also to intro-
duce shades of gray in crisis vulnerability.
The idea is to analyze the empirical distribution of the composite indica-
tor jointly with the occurrences of crises and to estimate probabilities of
crises conditional on different values of the composite indicator. We would
like to evaluate what the odds are of a crisis if none of the individual
indicators are signaling (i.e., when the composite indicator takes on a
value of zero) or when all the indicators are signaling (that is, when the
composite reaches its maximum value). But we would also like to evaluate
the intermediate scenarios, which depend on both how many and which
of the indicators are signaling. For example, we would like to calculate
the probability of a crisis conditioned on knowing that the value of the
indicator is in the 9 to 14 range, which as we saw from the cross-section
analysis earlier in this chapter was associated with a number of the recent
crises (table 5.5).
In practice, we can construct this set of probabilities using the informa-
tion on the value of the composite indicator for all the countries in the
sample together with the information on crises. Probabilities of crises are
estimated as follows:
P(C 兩 I ⬍ I t ⬍ I) ⳱ A/(A Ⳮ B) (5.2)
where I is the lower bound of the range we are interested in (9 in our

earlier example) and I is the upper bound of the range we are interested
in (14 in our example). As before, we have the following two-by-two
matrix,
Crisis occurs in the No crisis occurs in the

following 24 months following 24 months
I ⱕ It ⱕ I A B
It 僆
/ [I,I] C D
These probabilities will be estimated using all the information from all
the countries in the sample. Once we estimate these probabilities and use
the information on the number of signals being issued at any moment in
time, we can construct time-series probabilities of crisis for every country.
P tm denotes the probability of a crisis for country m in period t.
Once we construct these time series of crisis probabilities, we can also
evaluate the forecasting ability of the composite indicator and compare
its track record with that of other indicators, such as our top-ranked
univariate indicator, the real exchange rate. To conduct this horse race,
we follow Diebold and Rudebusch (1989) and employ the Quadratic
Probability Score (QPS) as our metric of goodness of fit. In particular, the
QPS evaluates the average closeness of the predicted probabilities and

Table 5.7 Composite indicator and conditional probabilities of
financial crises
Probability of a Probability of a
Value of indicator currency crisis banking crisis
0-1 0.10 0.03
1-2 0.22 0.05
2-3 0.18 0.06
3-4 0.21 0.09
4-5 0.27 0.12
5-7 0.33 0.13
7-9 0.46 0.16
9-12 0.65 0.27
12-15 0.74 0.37
Over 15 0.96 n.a.
Memorandum:
Unconditional Unconditional
probability of a probability of a
currency crisis banking crisis
0.29 0.10
Source: Kaminsky (1998).
the observed realizations, as measured by a dummy variable that takes

on a value of one when there is a crisis and zero otherwise.2
T
QPS k ⳱ 1/T 兺 2(P
t⳱1
k
t ⳮ Rt ) 2 (5.3)
where k ⳱ 1,2,3 refers to the indicator Pk, refers to the probability associ-
ated with that indicator, and Rt refers to the zero-one realizations. The
QPS ranges from zero to two, with a score of zero corresponding to
perfect accuracy.
Empirical Results
Table 5.7 reports the conditional probabilities of both currency and bank-
ing crises using the composite indicator. One column reports the likeli-
hood of currency crises. When almost none of the indicators are signaling
a future crisis, the composite indicator takes on values between zero and
two, and the probability of a currency crisis is only about 10 percent. The
probability of a currency crisis increases sharply and nonlinearly as signs
2. This approach has also been used to assess the ability of various indicators to anticipate
turning points in the business cycle (Diebold and Rudebusch 1989).

Table 5.8 Scoring the forecasts: quadratic probability scores
Tranquil Crisis Tranquil Crisis
Indicator times times times times
Naive forecast 0.173 1.008 0.024 1.620
Real exchange rate 0.115 0.979 0.018 1.589
Composite indicator 0.110 0.862 0.024 1.309
of vulnerability in the economy increase. Specifically, the probability of a

currency crisis reaches almost 100 percent when the composite indicator
takes on a value of 15 or above.3 The right column reports the same evidence
for banking crises. As with currency crises, the probabilities of a collapse of
the banking sector increase sharply as the economy deteriorates. However,
as we found with the univariate indicators, banking crises are harder to
anticipate. Even when nearly all the univariate indicators are signaling, the
probability of a banking crisis only climbs to about 40 percent.
Table 5.8 turns to the forecasting accuracy of the composite indicator. The
left side of the table looks at currency crises, while the right side examines
banking crises. The performance of the composite indicator is compared with
the performance of the real exchange rate—the best univariate indicator—as
well as to the naive forecast based on the unconditional probability of crisis.
The score statistics are reported separately for ‘‘crisis times’’ and for ‘‘tranquil
times;’’ this provides information on the performance of the leading indica-
tors across regimes. Recall that the closer the score in table 5.8 is to zero,
the more accurate is the forecast. The real exchange rate does significantly
better in anticipating currency crises than the unconditional forecast of cur-
rency crises. More important, the composite indicator performs better—in
terms of accuracy—than the real exchange rate, but the larger improvements
are obtained when forecasting in crisis times.
As shown on the right side of table 5.8, all indicators score worse
when predicting the onset of the banking crises—that is, the 24 months
bracketing the beginning of the banking crises. Again, the real exchange
rate does better than the unconditional forecast of banking crises in gen-
eral. For example, the quadratic probability score declines from 0.024 and
1.620 for the naive forecast of currency crises to 0.018 and 1.589 for the
real exchange rate forecast during tranquil and crisis times, respectively.
The composite indicator outperforms the real exchange rate in forecasting
the onset of a banking crisis but is outperformed by the real exchange
rate during tranquil times. This is explained by the fact that the real
exchange rate issues very few false alarms during tranquil periods.
3. Note we are not using the annual indicators in this exercise.

An Out-of-Sample Application to Southeast Asia
Using the information on the monthly value of the composite indicator
and on the conditional probabilities of crises, we can construct a time
series of probabilities of crises for our sample countries both in the sample
period (from January 1970 to December 1995) and out of it (from January
1996 to December 1997). As an illustration, figure 5.1 reports the time-
series probabilities of currency crises for four Southeast Asian economies
in the 1990s. The vertical lines in the figures represent the onset of a crisis.
With the exception of Indonesia, all the Southeast Asian countries
showed a severe state of distress, with about 65 percent of the indicators
flashing signals during the year preceding the crisis.4 The onset of these
crises occurred as the economies entered a marked slowdown in growth
after a prolonged boom in economic activity fueled by rapid credit cre-
ation.5 This dramatic surge in credit is explained, in large measure, by
heavy capital inflows and partly by the reform of the financial system;
financial liberalization was accompanied by large reductions in reserve
requirements. Overall, the explosive growth in these countries came to
an end with a real appreciation of the domestic currencies (which are, in
differing degrees, tied to the US dollar) and the corresponding loss of
export markets. It is noteworthy that during the latter part of this period,
there was a substantial appreciation of the dollar vis-à-vis the yen.
Short-term capital inflows to Thailand amounted to 7 to 10 percent of
GDP in each of the years 1994 through 1996, with the growth rate of
credit to the nonfinancial private sector amounting to more than 23 percent
over 1990-95. While output growth rates increased in the early 1990s to
almost 9 percent, fueled in part by easy credit, this rapid growth showed
signs of coming to an end with the real appreciation of the domestic
currency and the corresponding loss of export markets. The annual growth
rate of exports fell from a peak of 30 percent per year in 1994 to about 0
in 1996. Financial sector fragilities were also evident, with runs against
major banks starting to occur as early as May 1996. Finally, the sharp
increase in interest rates in 1997 to defend the baht put the nail in the
coffin of the already weak banking sector.6 Overall, 75 percent of the
indicators for which there are available data were exhibiting ‘‘anoma-
lous’’ behavior.
A boom-bust cycle in lending was also evident in the Philippines. As
in Thailand, the boom was fueled by capital inflows but also by a dramatic
4. For a more detailed exposition of the incidence of flashing indicators in the run-up to
the Asian crisis, see Kaminsky and Reinhart (1999).
5. This is at odds with the interpretation of these crises provided in Radelet and Sachs
(1998), who argue these crises are the byproduct of a financial panic.
6. It is noteworthy that finance companies had been receiving substantial assistance from
the central bank during this period.

Figure 5.1 Probability of currency crisis for four Southeast Asian
countries, 1990-97
Indonesia
1.0
0.8
0.6
0.4
0.2
0.0
1990 1991 1992 1993 1994 1995 1996 1997 1998
Malaysia
1.0
0.8
0.6
0.4
0.2
0.0
1990 1991 1992 1993 1994 1995 1996 1997 1998
The Philippines
1.0
0.8
0.6
0.4
0.2
0.0
1990 1991 1992 1993 1994 1995 1996 1997 1998
Thailand
1.0
0.8
0.6
0.4
0.2
0.0
1990 1991 1992 1993 1994 1995 1996 1997 1998
Note: Vertical lines indicate currency crisis date.


reduction in reserve requirements, accompanying financial liberalization.
Bank credit increased by 44 percent a year during 1995-96. As in Thailand,
rapidly expanding credit was an important contributor to the rally in
stock and real estate markets, with a fourfold increase in prices in both
markets. Foreign currency exposure increased in the Philippines in the
1990s via foreign borrowing to finance domestic lending. Foreign borrow-
ing was concentrated in short maturities. Consumer lending also increased
and fueled a surge in consumption, leading to a deterioration of the current
account. This deterioration in the external accounts was aggravated by
the real exchange rate appreciation of the domestic currency. The loss of
competitiveness anticipated a future decline in growth and also contrib-
uted to a substantial deterioration of the quality of banks’ assets, further
reducing the odds of survival of many individual financial institutions.
Overall, about 50 percent of the indicators were signaling the increased
vulnerability of the economy during the two years before the collapse of
the implicit peg in July 1997.7
Malaysia shared certain vulnerabilities with Thailand. It too was
affected by the slowdown in the region, though to a much smaller degree.
It too had large current account deficits during 1990-95, although in 1996
the outlook for the external sector improved somewhat, with the current
account to GDP ratio shrinking to -5.3 percent (in Thailand, the current
account to GDP ratio in 1996 was roughly -8.0 percent). Moreover, Malay-
sia, like Thailand, accumulated debt rapidly in the 1990s, with capital
inflows fueling a stock and real estate market boom and with asset prices
increasing about 300 percent in the early 1990s. Malaysia also suffered
from financial sector vulnerabilities (although not to the same extent as
Thailand) as a result of the high degree of leveraging in the economy.
Indeed, Malaysia had one of the highest ratios of credit-to-GDP in the
world, and the banks had a large exposure to the property and equity
markets. For Malaysia, about 60 percent of the indicators were showing
signs of distress at the onset of the crisis.
Indonesia looked somewhat different. While it too exhibited banking
fragilities and while short-term debt easily exceeded available foreign
exchange reserves (about 1.7 times the stock of the country’s reserves),8
the current account deficit did not deteriorate as fast (reaching only 3.5
percent of GDP in 1996), the slowdown in growth was not yet evident,
and the real exchange rate did not appreciate as much as in the other
7. The Philippines was classified as a managed float in the IMF’s exchange rate arrangements
classification. Yet even a relatively uninformed bystander could see the large-scale extent
of foreign exchange intervention before mid-1997, which kept the Philippine peso’s value
virtually unchanged against the dollar.
8. The beginning of the banking crisis in Indonesia can be dated to November 1992, when
a large bank (Bank Summa) collapsed and triggered runs on three smaller banks. Most
state-owned banks also experienced serious difficulties.

countries in the region. Relatively few indicators (less than 20 percent)
showed signs of strains in the economy in the months before the crisis.
Here, over and beyond all the political uncertainty, as we explain further
in chapter 6, a key factor seemed to be contagion from the flurry of
financial crises elsewhere in the region—particularly the liquidity squeeze
associated with the withdrawal of Japanese banks (the major lenders to
the region) in the wake of losses they suffered in the Thai crisis.9
To sum up, we have seen in this chapter that the signals approach can
draw coarse distinctions, both across countries and over time, in crisis
vulnerability during out-of-sample periods (in this case, 1996-97). The
approach does reasonably well in anticipating currency crises in most of
the Asian crisis countries. At this stage, the model performs much better
for currency crises than for banking crises. The evidence presented here
also indicates that it is worthwhile to work with a composite index, which
outperforms the best of the univariate indicators.
9. The reversal was, in fact, quite pronounced, from capital inflows in the region of $50
billion in 1996 to an outflow of $21 billion in 1997. See Kaminsky and Reinhart (2000) and
the next chapter for a discussion on world and regional financial links and their effects on
the probability of currency crises.

6
Contagion
As suggested earlier, in most cases, the leading indicators signaled ahead

of the 1997-98 currency and banking crises. The Indonesian case, however,
is an example of an episode where ‘‘the dog did not bark.’’ Despite the fact
that this country experienced a meltdown in its currency and a collapse in
its banking industry, Indonesia was firmly anchored near the bottom of
the list in table 5.5, as relatively few indicators gave advanced warning.
In a similar vein, although Argentina was the hardest-hit country during
the ‘‘tequila effects’’ that followed the Mexican financial crisis of 1994-
95, it too would not have been judged as vulnerable on the basis of the
fundamentals reviewed in the preceding chapters.
Of the 89 currency crises and nearly 30 banking crises in our sample,
only a handful of these occur with as few indicators flashing as was the
case for Indonesia (22 percent). As shown in table 6.1, less than 15 percent
of the currency and banking crises shared the Indonesian silence of signals.
Still, the Indonesian crisis suggests something is missing from our previ-
ous analysis. The most obvious candidate is cross-country contagion of
financial crises.1
The empirical evidence on contagion is still limited to relatively few
studies, but the weight of the empirical results suggests it is important.
To the extent that contagion or spillovers matter, being near the bottom
of the ‘‘vulnerability’’ list based on own-country fundamentals would not
preclude a country from having a crisis. In this chapter, we briefly review
1. Of course, the political turmoil at this time in Indonesia is likely to have contributed to
the meltdown of the currency and the economy.
73

Table 6.1 Crises that showed few signals, 1970-97
Number of crises Proportion of crises
that occurred with that occurred with
Type of crisis Number of five or fewer five or less
and sample crises indicators signaling indicators signaling
Banking, 1970-95 29 3 10.3
Currency, 1970-95 87 12 13.5
Banking, 1996-97 6 1 16.7
Currency, 1996-97 6 1 16.7
some of the theoretical underpinnings for contagion and then construct

a ‘‘contagion or spillover vulnerability index’’ that attempts to capture
trade and finance links among countries. We then explore the extent to
which crises probabilities increased for other emerging markets following
the Mexican crisis of 1994 and the Asian crisis of 1997, owing to trade
and financial links.
Most of the theoretical work on contagion has attempted to provide a
framework for understanding how shocks in one country are transmitted
elsewhere. Our review of this literature emphasizes its empirical implica-
tions in terms of defining contagion, delineating its channels of influence,
and testing for its presence.
Defining Contagion
Only one study that we are aware of examined the issue of contagion in
the context of Latin America’s debt crisis of the 1980s. Doukas (1989)
interprets contagion as the influence of ‘‘news’’ about the creditworthiness
of a sovereign borrower on the spreads charged to the other sovereign
borrowers, after controlling for country-specific macroeconomic funda-
mentals. Most other studies, such as Valdés (1997), define contagion as
excess comovement in asset returns across countries, be it for debt or
equity. This comovement is said to be excessive if it persists even after
common fundamentals, as well as idiosyncratic fundamental factors, have
been controlled for. A recent variant to this approach (as in Forbes and
Rigobon 1998) defines ‘‘shift-contagion’’ as an increase in excess comove-
ment of asset returns during crisis periods.
Eichengreen, Rose, and Wyplosz (1996) define contagion as a case where
knowing that there is a crisis elsewhere increases the probability of a
crisis at home, even when fundamentals have been properly taken into
account. This is the definition of contagion that we will explore in the
remainder of this chapter. These fundamentals could be country-specific,
along the lines analyzed in the preceding chapters, or they could be
external and common to all countries or a group of countries. Changes

in international interest rates are a plausible candidate for a common
shock. If international interest rates rise markedly, as they did in the early
1980s, and many countries have financial crises simultaneously, we would
not attribute the common timing of the crises to contagion—we would
place the blame, instead, on a common shock.
In the absence of a common shock, a crisis in one country can spread
to others via links in trade and finance. Some studies would not call this
contagion either but rather label it a spillover (e.g., Masson 1998). These
studies would reserve the term contagion for cases where a crisis spreads
from one country to another despite the absence of any trade or finance
link—possibly owing to shifts in sentiment and herding behavior on the
part of investors.
Since it is impossible to predict when such shifts in sentiment will take
place and which countries will be most affected by changes in financial
markets’ expectations, our focus in the empirical part of this chapter
will be on assessing countries’ vulnerability to a crisis elsewhere when
financial and trade links are evident.
Theories of Contagion and Their Implications

There are several explanations for why crises tend to be bunched or
clustered. Some recent models have revived Nurkse’s story of ‘‘competi-
tive devaluations.’’ This explanation emphasizes trade links, be they bilat-
eral or with a third party.2 Once one country has devalued, it is costly
(in terms of a loss of competitiveness and output) for other countries
(with strong trade links to the first country) to maintain their parities. In
this setting, subsequent devaluations reflect a policy choice, with a salu-
tary effect on output. In any case, an empirical implication of this story
of contagion is that we should either observe a high volume of trade
among the ‘‘synchronized’’ devaluers or competition in a common
third market.3
Calvo (1998) stresses the role of liquidity. A leveraged investor facing
margin calls needs to sell his or her asset holdings to an uninformed
counterpart. Because of information asymmetries, a ‘‘lemons problem’’
arises, and the asset can only be sold at a fire sale price. A variant of this
story can be told about an open-end fund portfolio manager who needs
to raise liquidity in anticipation of future redemptions. The strategy will
be not to sell the asset whose price has already collapsed but other assets
2. See Gerlach and Smets (1995) for a model that emphasizes bilateral trade and Corsetti
et al. (1998) for one in which emerging markets compete in a common third market.
3. As a story of fundamentals-based contagion, of course, this explanation does not speak
to the fact that central banks often go to great lengths to avoid the devaluation in the
first place.
CONTAGION 75

in the portfolio. In doing so, other asset prices are depressed, and the
original disturbance spreads across markets.
Yet another potentially important channel of transmission that has been
largely ignored in the contagion literature but that is stressed by Kaminsky
and Reinhart (2000) is the role of common lenders—in particular, commer-
cial banks. US banks had large loan exposure to Latin America in the
early 1980s, much in the way that Japanese banks did during the Asian
crisis of 1997. The need to rebalance the overall risk of the bank’s asset
portfolio and to recapitalize following the initial losses can lead to a
marked reversal in commercial bank credit, both in the original crisis
country and for others who rely heavily on the same lender.
Another family of contagion models has deemphasized the role of trade
in goods and services in favor of the role of trade in financial assets,
particularly in the presence of information asymmetries. Calvo and Men-
doza (2000) present a model where the fixed costs of gathering and pro-
cessing country-specific information give rise to herding behavior, even
when investors are rational. Kodres and Pritsker (1998) also present a
model with rational agents and information asymmetries. However, they
stress the role played by investors who engage in cross-market hedging
of macroeconomic risks.
In these financial contagion explanations, the channels of transmission
come from the global diversification of financial portfolios. Here, the
implication is that countries with more internationally traded financial
assets and more liquid markets are likely to be relatively vulnerable to
contagion. Small emerging economies with relatively illiquid financial
markets are likely to be underrepresented in international portfolios to
begin with and thus ought to be shielded from this type of contagion.
In addition, cross-market hedging usually requires a moderately high
correlation of asset returns. For our purposes, the key empirical implica-
tion is that countries whose asset returns exhibit a high degree of comove-
ment with the original crisis country (for example, Argentina with Mexico
in 1994-95 or Malaysia with Thailand in 1997-98) will be more vulnerable
to contagion via the cross-market hedges that were in place as the cri-
sis erupted.
Empirical Studies
Very few studies have attempted to run ‘‘horse races’’ among alternative
models of contagion. Eichengreen, Rose, and Wyplosz (1996) tested the
influence of bilateral trade links against similarities to the crisis country
in macroeconomic fundamentals. Glick and Rose (1998) examined the
trade issue further within a much broader country sample, while Wolf
(1997) attempted to explain pairwise correlations in stock returns by bilat-
eral trade and by common macroeconomic fundamentals. All studies

Table 6.2 Conditional probabilities and noise-to-signal ratios for
financial and trade clusters
Percentile of countries High Third-party Bilateral
sharing a cluster Bank correlation trade trade
Noise-to-signal ratio
25 to 50 0.90 0.58 1.54 2.34
50 and above 0.07 0.39 0.57 0.08
Weight in vulnerability
index
25 to 50 1.10 1.73 0.64 0.42
50 and above 14.08 2.57 1.75 12.5
Probability of a crisis
conditioned on crises
elsewhere in the cluster
minus unconditional
probability of crisis
25 to 50 ⳮ3.1 20.8 ⳮ6.3 ⳮ21.8
50 and above 52.0 47.1 30.7 47.3
Source: Based on Kaminsky and Reinhart (2000).
conclude that trade linkages play an important role in the propagation

of shocks. Because trade tends to be more intra- than interregional in
nature, Glick and Rose (1998) conclude that this helps explain why conta-
gion tends to be mainly regional rather than global. Kaminsky and Rein-
hart (1998b) also look at trade links (both bilateral and third-party) but
emphasize financial sector links. In an early paper on the subject, Frankel
and Schmukler (1996) find evidence of contagion in emerging market
mutual funds.
Trade and Financial Clusters and a Composite

Contagion Index
As shown in chapter 5, one can construct a composite index to gauge

the probability of a crisis conditioned on multiple signals from various
indicators (i.e., economic fundamentals); the more reliable indicators
receive greater weight in this composite index. This methodology can be
readily applied to construct a composite ‘‘contagion vulnerability index.’’
As in Kaminsky and Reinhart (2000), we consider four channels
through which shocks can be transmitted across borders: two channels
deal with the interlinkages in financial markets, be they through foreign
bank lending or globally diversified portfolios, and two deal with trade
in goods and services. Table 6.2 reports the noise-to-signal ratios and the
difference between the conditional probability of a crisis (conditioned on
CONTAGION 77

knowing there is a crisis elsewhere in that particular cluster) and the
unconditional probability of crisis.
Hence the four clusters—bilateral trade, third-party trade, common
bank lender, and high correlations—play the same role as the indicators.4
If a country shares a common cluster with the initial crisis country, it is
a signal; if a crisis occurs in the second country within the following 24
months, it is a good signal; if a crisis does not occur, it is a false alarm.
Hence for these possible linkages, the number of signals could range from
zero (no common clusters) to four, in which the country shares all four
clusters with the initial crisis country.
As when we weighted individual indicators, a good argument can be
made for eliminating potential leading indicators that had a noise-to-
signal ratio above unity (that is, those whose marginal forecasting ability
is zero or less). Applying this criterion to our results, we would focus on
the case in which more than 50 percent of the countries that share a
common cluster are experiencing a crisis. As shown in table 6.2, the
highest noise-to-signal ratio is 0.57, well below unity—but the track record
of the signals in each of the clusters is far from uniform. Thus we weight
the signals by the inverse of the noise-to-signal ratios reported in table
6.2 (see Kaminsky and Reinhart 2000 for details).
Formally, as we did in chapter 5 for the macroeconomic fundamentals,
we construct the following composite indicator:
n
It ⳱ 兺 S /␻
j⳱1
j
t
j
(6.1)
In equation 6.1 it is assumed that there are n indicators (i.e., clusters).

Each cluster has a differentiated ability to forecast crises, and as before,
this ability can be summarized by the noise-to-signal ratio, here denoted
by ␻ j. S jt is a dummy variable that is equal to one if the univariate indicator,
S j, crosses its critical threshold and is thus signaling a crisis and is zero
otherwise. As before, the noise-to-signal ratio is calculated under the
4. The countries are classified by bank clusters according to which financial center they
depend on the most (on the basis of the Bank for International Settlements data). For the
high-correlation asset returns cluster, we include countries that have a correlation that is
0.35 percent or higher in their daily stock returns. For the bilateral trade cluster, we include
countries for which either imports or exports to the second country are 15 percent or higher.
For the third-party trade cluster, we require countries to have a common third market and
similar commodity export structure. We focus on the top 10 to 15 goods that account for
40 percent or more of exports in the initial crisis country; we then see if those same goods
account for a significant share (20 percent or higher) of exports of the remaining countries.
For example, the top 14 Thai exports account for 46 percent of total exports; these same
goods account for 44 percent of Malay exports; hence Malaysia is in the same third-party
trade cluster. By contrast, those goods only account for 15 percent of Indonesia’s exports,
leaving Indonesia outside the third-party trade cluster.

assumption that an indicator issues a correct signal if a crisis occurs within
the following 24 months. All other signals are considered false alarms.
The maximum value that this composite contagion vulnerability index
could score is 30.9 if a country belonged to the same four clusters as the
crisis country. This score is a simple sum of the inverse of the noise-to-
signal ratio. Table 6.3 records a one if a country is in the same cluster as
the original crisis country in that episode and no entry otherwise.
What the Composite Contagion Vulnerability

Index Reveals about Three Recent Crisis
Episodes
We now consider, on the basis of the trade and financial sector linkages
discussed here, which countries would have been classified as vulnerable
to contagion during three recent episodes of currency crises in emerg-
ing markets.
The first of these episodes began with the devaluation of the Mexican
peso in December 1994. On the heels of the Mexican devaluation, Argen-
tina and Brazil were the countries to come under the greatest speculative
pressure. In a matter of a few weeks in early 1995, the central bank of
Argentina lost about 20 percent of its foreign exchange reserves and bank
deposits fell by about 18 percent as capital fled the country. Such a severe
outcome could hardly be attributed to trade linkages and competitive
devaluation pressures, as Argentina does not trade with Mexico on a
bilateral basis, nor does it compete with Mexican exports in a common
third market. 5 In the case of Brazil, the speculative attack was brief,
although the equity market sustained sharp losses. Both of these countries
record high vulnerability index scores following the Mexican devaluation.
While the effects on Asia of the Mexican crisis were relatively mild,
the country that encountered the most turbulence in the region was the
Philippines, which also registers a relatively high vulnerability score.
In the case of the Thai crisis, Malaysia shares both trade and finance
links with Thailand. For the other Asian countries, the potential channels
of transmission are fewer. As noted earlier, the Philippines is a part of
the same third-party trade cluster as Thailand, which receives a weight
of 1.75 (i.e., 1/0.57) in the composite index; it is also part of the Asian
high-correlation cluster, which receives a weight of 2.57 (i.e., 1/0.39) in
the index. Indonesia shares the same high-correlation cluster with Thai-
land, and it is a part of the Japanese bank cluster, which receives a weight
of 14.08 (i.e., 1/0.07). Hence, as shown in table 6.4, Indonesia and the
Philippines’ contagion vulnerability index scores are 16.65 and 4.32,
5. See Kaminsky and Reinhart (2000) for details on the pattern of trade.
CONTAGION 79

80
Table 6.3 Countries sharing financial and trade clusters with original crisis country or region
High-correlation Third-party trade Bilateral trade
Bank cluster cluster cluster cluster

Latin Latin Latin
Country Japan US Asia America Asia America America
Argentina 1 1 1
Bolivia
Brazil 1 1 1 1
Chile 1 1
Colombia 1 1
Denmark
Finland
Indonesia 1 1
Israel
Malaysia 1 1 1
Mexico 1 1 1
Norway
|
Peru 1
http://www.iie.com
The Philippines 1 1 1
South Korea 1 1
Spain
Sweden
Thailand 1 1 1
Turkey
Uruguay 1 1
Venezuela 1 1
Source: Kaminsky and Reinhart (2000).

Table 6.4 Contagion vulnerability index
Contagion vulnerability index
Mexican crisis Thai crisis Brazilian crisis
Country (December 1994) (July 1997) (January 1999)
Argentina 16.65 0 29.15
Bolivia 0 0 0
Brazil 18.4 0 n.a.
Chile 0 0 26.58
Colombia 12.5 0 15.83
Denmark 0 0 0
Finland 0 0 0
Indonesia 0 16.65 0
Israel 0 0 0
Malaysia 0 28.33 0
Mexico n.a. 0 18.4
Norway 0 0 0
Peru 2.57 0 2.57
The Philippines 14.08 4.32 14.08
South Korea 0 26.58 0
Spain 0 0 0
Sweden 0 0 0
Thailand 0 n.a. 0
Turkey 0 0 0
Uruguay 0 0 26.58
Venezuela 12.5 0 15.83
respectively. South Korea also borrowed heavily from Japanese banks.

Accordingly its exposure to Thailand came more from having a common
lender than from conventional competitive trade pressures.
The most recent of these emerging market crises was Brazil’s devalua-
tion of the real in early 1999. Not surprisingly, Argentina, which has both
trade (through Mercosur) and financial linkages with Brazil, shows the
highest vulnerability; other Mercosur countries come close in suit.
Table 6.5 provides additional details on some of the possible channels
through which the crisis may have spread during these episodes. To the
extent that there is herding behavior and investors lump together all
emerging markets—or perhaps only those in the infected region—that
would add yet another channel of transmission to those laid out in
table 6.5.
As regards the potential role of bilateral and third-party trade linkages,
Malaysia would be the country most closely linked with Thailand, with
South Korea and the Philippines exhibiting more moderate trade expo-
sure. Trade is certainly not the main culprit in explaining the vulnerability
of Argentina and Brazil following the Mexican devaluation or of Indonesia
following the Thai crisis.
CONTAGION 81

82
Table 6.5 Characteristics of affected countries in Asian and Mexican episodes of contagion
Level of trade with
Level of liquid common third

market/high Level of party in same
representation bilateral trade, commodities,
in mutual funds, percentage of percentage of
Affected Exchange Nature of Common High percentage of exports to exports competing
country rate regime contagion bank correlation emerging affected with top exports of
(onset month) at onset or spillover lender of returns market portfolio country affected country
Tequila crisis: 1994-95 First crisis: Mexico, December 1994

Argentina Currency Turbulence Yes High, 0.56 Moderate, 2.98 Low, 1.7 Low, 15.6
board
Brazil Peg Turbulence Yes Moderate, 0.36 High, 13.07 Low, 2.4 Low, 10.9
Asian flu: 1997-98 First crisis: Thailand, July 1997

|
Malaysia (July) Managed Crisis Yes High, 0.60 Moderate, 5.88 Moderate, 4.1 High, 44.4
http://www.iie.com
float
The Philippines Managed Crisis Yes High, 0.68 Low, 2.40 Moderate, 3.8 Low, 19.2
(July) float
Indonesia Narrow Crisis Yes High, 0.54 Moderate, 4.35 Low, 1.8 Low, 15.5
(August) band
Hong Kong Currency Turbulence No High, 15.33 Low, 1.0 Low
(October) board
South Korea Crawling Crisis Yes Low, 0.24 Moderate, 6.16 Low, 2.0 Moderate, 27.9
(November) band
Table 6.6 Asia and Latin America: added power of Thai crisis in
explaining probability of contagion in bank cluster,
July 1997
Probability of a crisis conditioned
on crises elsewhere in the cluster minus
Country unconditional probability of crisis
Asia
Indonesia 0.60
Malaysia 0.35
The Philippines 0.02
Latin America
Argentina 0.02
Chile 0.02
Mexico 0.02
Turning to financial links stemming from a common lender, exposure

to European and Japanese banks, which rapidly pulled out of the region
after the outbreak of the Thai crisis, was common to all the affected
countries except Hong Kong. Brazil and Argentina were in the same (US)
bank cluster as Mexico in 1994-95, but US banks were not as exposed to
Latin American borrowers as they were in the early 1980s, and portfolio
flows had replaced bank lending as the main source of funding for these
emerging economies.
Most of the affected Asian countries (except South Korea) had high
correlations of asset returns with Thailand, although none except Hong
Kong were home to relatively liquid markets. The same is true of stock
returns in Argentina, which had the highest correlation of asset returns
with Mexico of any country in the region. Here, it is hard to separate
cause and effect. A high correlation may reflect past contagion, but to the
extent that current cross-hedging strategies use such historical correlations
as a guide, it could be the vehicle for future contagion.
In sum, while this is a preliminary assessment of contagion channels,
it suggests that financial sector linkages, be it through banks or via interna-
tional capital markets, could have been influential in determining how
shocks were propagated in recent crises episodes, particularly for Argen-
tina, Brazil, and Indonesia.
In table 6.6 we take this analysis one step further. Specifically, the table
compares some of the larger emerging markets in Asia and in Latin
America at the onset of the Thai crisis (July 1997) based on how much
added explanatory power a crisis elsewhere added to the probability of
crisis at home. The numbers reported in the table are the simple difference
between the probability of a crisis conditioned on our composite index
of fundamentals, P(C 兩 F), and the probability of crisis conditioned on the
fundamentals and a crisis elsewhere related to a common lender, P(C 兩 F,
CONTAGION 83

CE). If knowing that a crisis elsewhere in the cluster helps predict a crisis
at home, then P(C 兩 F, CE) ⬎ P(C 兩 F). It is noteworthy that the conditional
probability of a crisis does not change much for those Latin American
countries and the Philippines that are not a part of the Japanese bank
cluster. For them, the contagion from the Thai crisis via this channel is
minimal. By way of contrast, for countries that are in the same bank cluster
as Thailand, the probability of crisis increases markedly, for Malaysia and
particularly for Indonesia. Malaysia’s crisis probabilities conditioned on
the fundamentals alone were well above Indonesia’s, as shown in figure
5.1. Hence, for Malaysia, the incremental explanatory power of the crisis-
elsewhere variable is smaller than for Indonesia.
To sum up, the empirical evidence contained in this chapter suggests
that the analysis of fundamentals stressed in the signals approach can be
strengthened by incorporating financial sector linkages, which increase
the vulnerability to contagion. While assessing the predictive ability of
the individual bank clusters is a useful exercise to discriminate among
competing explanations of contagion, countries that are linked in trade
are also often linked in finance. This implies that multiple channels of
contagion may be operating at once.

7
The Aftermath of Crises
The preceding chapters have focused on the antecedents of financial crises.

The emphasis has been on the indicators’ ability to anticipate crises and
to measure the extent of a country’s vulnerability. In this chapter, we
begin with the premise that, whether anticipated or not, financial crises
occur, and once they do, policymakers and market participants become
concerned about their consequences for economic activity. In light of
Asia’s recent woes, there was much speculation as to how long it would
take those economies to recover from such destabilizing shocks and what
the consequences for growth and inflation would be over the near and
medium term. In what follows, we review the historical experience of the
aftermath of currency and banking crises.
The Recovery Process
If we want to assess how our indicators behave following financial crises

and, in particular, how many months elapse before their behavior returns
to normal, we must define ‘‘normal.’’ One way to do that is to compare
‘‘tranquil’’ and ‘‘crisis’’ periods. We define periods of tranquility as the
periods that exclude the 24 months before and after currency crises. In
the case of banking crises, the 24 months before the banking crisis begins
and the 36 months following it are excluded from tranquil periods. For
each indicator, we tabulate its average behavior during tranquil periods.
We then compare the postcrisis behavior of the indicator to its average
in periods of tranquility.
85

Table 7.1 Length of recovery from financial crises (average number
of months for a variable to return to ‘‘normal’’ behavior)a
Indicator Banking crisis Currency crisis
Bank deposits 30 (below) 12 (above)
Domestic credit/GDPb 15 (above) 9 (above)
Exports 20 (below) 8 (below)
Excess M1 balances 9 (above) 8 (below)
Imports 29 (below) 18 (below)
Lending-deposit rate ratio 0 3 (above)
M2 multiplier 7 (above) 21 (below)
M2/reserves 15 (above) 7 (above)
Output 18 (below) 10 (below)
Real exchange rate 8 (below-overvalued) 23 (above-undervalued)
Real interest ratec 15 (above) 7 (below)
Real interest rate differential 15 (above) 7 (below)
Stock prices 30 (below) 13 (below)
Terms of trade 4 (below) 9 (below)
a. We note in parentheses whether the variable remained below or above the tranquil-
period norm.
b. Domestic credit as a share of GDP remains above normal levels largely as a result of
the decline in GDP following the crisis.
c. The disparity between the postcrisis behavior of real interest rates lies in the fact that a
large share of the currency crises occurred in the 1970s, when interest rates were controlled
and not very informative about market conditions.
Table 7.1 summarizes the results of that aftermath exercise for currency
and banking crises. The number given after each indicator is the average
number of months that it takes for that variable to reach its norm during
tranquil periods. In parentheses, we note whether the level or growth
rate of the variable remains above or below its norm in the postcrisis
period. Several findings merit special attention.
First, the deleterious effects of banking crises do linger longer than
currency crises’ effects. This is evident in several of the indicators. While
the 12-month change in output remains below its tranquil-period norm
for (on average) 10 months following the currency crash, it takes nearly
twice that amount of time to recover following the banking crisis. This
more sluggish recovery pattern is also evident in imports, which take
about 21⁄2 years to return to their norm. The weakness in asset prices,
captured here by stock prices that are below the norm, persist for 30
months on average for banking crises—more than twice the time it takes
to recover from a currency crash.
There are several explanations for banking crises’ more protracted
recovery periods. One concerns the special nature of the ‘‘twin’’ crises.
The bulk of the banking crises in this sample were accompanied by cur-
rency crises, and twin crises ought to have more severe effects on the
economy, as argued in Kaminsky and Reinhart (1999).

Table 7.2 Time from beginning of banking
crises to their peaks (months)
Descriptive statistic Number of months
Mean 19
Minimum 0
Maximum 53
Standard deviation 17
Source: based on Kaminsky and Reinhart (1999).
A second explanation, not mutually exclusive, is that a banking crisis

cuts off both external and domestic sources of funding for households
and firms, whereas a currency crisis only cuts off the former. In other
words, the credit crunch is more severe.
A third explanation derives from the distribution of crises across the
sample period. The currency crises are roughly evenly distributed
between the pre- and postliberalization periods, while the banking crises
are bunched in the 1980s and 1990s. To the extent that crises have become
more severe following deregulation, the slower pace of recovery in bank-
ing crises may reflect that. This is an issue we take up later.
A second finding highlighted in table 7.1 is that there are important
sectoral differences in the pace of recovery, depending also on the type
of crisis. For instance, following the devaluations that characterize the
bulk of currency crises, exports recover relatively quickly and ahead of
the rest of the economy at large. In contrast, following banking crises,
exports continue to sink for nearly two years. This may reflect a persistent
overvaluation, high real interest rates, or a ‘‘credit crunch’’ in the aftermath
of banking crisis.
Table 7.2 underscores the protracted nature of banking crises by show-
ing the average number of months elapsed from the beginning of the
crisis to its zenith for the 26 banking crises studied in the Kaminsky and
Reinhart (1999) sample. On average, it takes a little over a year and a half
for a banking crisis to ripen; in some instances it has taken over four
years. Often, financial sector problems do not begin with the major banks,
but rather with more risky finance companies. As the extent of leveraging
rises, households and firms become more vulnerable to adverse economic
or political shocks that lead to higher interest rates and lower asset values.
Eventually, defaults increase and problems spread to the banks. If there
are banks runs, such as in Venezuela in 1994, the spread to the larger
institutions may take less time.
The information in table 7.2 does not fully capture the length of time
that the economy may be weighed down by banking-sector problems,
since it does not cover information on the time elapsed between the crisis
peak and its ultimate resolution. Rojas-Suarez and Weisbrod (1996), who
examine the resolution of several banking crises in Latin America, high-
THE AFTERMATH OF CRISES 87

Table 7.3 Comparison of inflation and growth rates before and
after currency crises (percent)
average of
tⴑ1
Indicator and tⴑ2 t tⴐ1 tⴐ2 tⴐ3
Real GDP growth
All countries 3.3 1.0 1.8 3.1 2.9
Moderate-inflation 3.5 2.1 2.4 3.3 4.0
countriesa
High-inflation 3.0 ⳮ0.6 1.0 3.1 1.7
countries
Inflation
Moderate inflation 14.0 15.7 18.0 15.7 14.8
countries
High-inflation 270.9 732.8 394.8 707.4 964.7
countries
t ⳱ year of crisis
a. Moderate-inflation countries are those with inflation rates below 100 percent in all years
surrounding the crisis; high-inflation countries are those in which inflation exceeded 100
percent in at least one year.
light the sluggishness of the resolution process in many episodes. The

Japanese banking crisis, which has spanned most of the 1990s and is still
ongoing, is a recent example of the protracted nature of the recognition-
admission-resolution process.
We next focus in table 7.3 on the evolution of two of the most closely
watched macroeconomic indicators—growth and inflation—in the after-
math of currency crises. Instead of comparing tranquil versus crisis peri-
ods, we compare the immediate two precrisis years with the postcrisis
years.1 We distinguish between moderate-inflation and high-inflation
countries; the latter encompass mostly Latin American countries. The
numbers for ‘‘all countries’’ represent an average of the 89 currency crises
in our sample.
Perhaps the most interesting finding from table 7.3 is that it takes
between two and three years in currency crisis episodes for economic
growth to return to the precrisis average. Devaluations may be expansion-
ary in industrial countries—witness the sharp recovery in the United
Kingdom following its floatation of the pound during the European
exchange rate mechanism (ERM) crisis and the strong growth performance
of the Australian economy in 1998 coincident with a large depreciation
of the Australian dollar.
1. For a comparison of the recent crises with the historic norm, see Kaminsky and Rein-
hart (1998).

Yet, as shown in table 7.3, devaluations in the developing world are
most often associated with recessions. There are numerous theoretical
explanations for this finding (see Lizondo and Montiel 1989 for a survey
of the literature). Two are of particular interest. First, devaluations that
occur in the context of balance of payments crises are associated with
losses of confidence and increases in uncertainty that are damaging to
economic activity. It is usually the case that credibility problems are more
severe for developing countries than for their industrial counterparts.
Second, while industrial countries do not face a higher debt-servicing
costs following devaluation, as their debt is predominantly denominated
in their own currencies, developing-country debt is largely denominated
in US dollars or other foreign currencies. Hence a large devaluation will
have staggering implications for debt servicing burdens.
Furthermore, recessions following currency crises appear to be more
severe among the high-inflation countries. This may be because inflation
itself has adverse effects on growth (Fischer 1993) or because high-inflation
countries may be especially prone to losing their access to international
credit relative to their low-inflation counterparts. The evidence presented
in Cantor and Packer (1996a) does indeed show that private credit ratings
penalize high inflation.
In any event, most existing studies find devaluation episodes in emerg-
ing economies to be contractionary, with their negative impact diminish-
ing within two years, and table 7.3 supports these findings.2 In this connec-
tion, Morley (1992) concludes that the reason that earlier studies, which
are largely focused on devaluations during the 1950s and 1960s, find
milder recessions and even positive output consequences is that many of
those devaluation episodes occurred in the context of trade liberalization
and exchange market reform—not in the context of balance of pay-
ments crises.
Table 7.3 shows that inflation picks up in the two years following the
currency crisis in both moderate- and high-inflation countries. The
increase is far more dramatic for high-inflation countries, where inflation
remains at a substantially higher level following the crisis (usually because
of recurring devaluations at an accelerating rate). For the moderate-infla-
tion countries, inflation returns to its precrisis rate in about three years.
These patterns are consistent with those found by Borensztein and de
Gregorio (1998) in 19 devaluation episodes in low- and high-inflation
countries. The empirical studies surveyed in table 7.4 arrive at similar
conclusions.
Some Caveats
The preceding discussion has suggested a ‘‘representative time profile’’
for the recovery process in the wake of currency and banking crises. This
2. See also Kamin and Rogers (1997) for an interesting analysis of the case of Mexico.

90
Table 7.4 The wake of devaluations: a review of the literature

Study Sample Variable Results
Borensztein and 1982-94, 19 devaluation Inflation About one-quarter of the devaluation is offset by higher inflation
de Gregorio episodes, five of which after three months, about 60 percent after two years. Except for
(1998) are industrial countries Latin American cases, inflation returned to its pre-devaluation
level in three years or less.
Cooper (1971) 1951-70, 24 large GDP Devaluations are followed by either a recession or a reduction in
devaluations the rate of growth. These output effects were small, however.
Edwards (1986) 1965-80, 12 developing GDP Regressing income on the real exchange rate while controlling
countries for policy fundamentals, he finds a negative and significant
coefficient on the real exchange rate in the first year; this was
|
offset by positive coefficients later on. Long-run effect is neutral.

http://www.iie.com
Edwards (1989) 1962-82, 39 devaluations Inflation, GDP, current Inflation doubles, on average from about 8 to 16.7 percent one
greater than 15 percent in account as a share of year after the crisis; net foreign assets/money fall by about 5
24 countries GDP, change in net percent in the three years following the crisis.
foreign assets/money
Kamin (1988) 1953-83, 50 to 90 Inflation, GDP, exports, The trade balance does not change much the year following the
devaluations in excess of imports, export prices, devaluation; import and export growth increase. Capital inflows
15 percent import prices, capital and reserves are about the same at tⳭ1 as in the year of the
inflows, trade balance, devaluation. Inflation increases the year of the devaluation then
reserves declines. GDP growth falls the year of the devaluation then
recovers the following year.
Kiguel and Ghei 1950-90, 33 devaluations Real exchange rate, About 60 percent of the devaluation is not eroded by increases
(1993) in excess of 20 percent in inflation, GDP growth, in domestic prices. Inflation increases, on average by about
low-inflation countries exports/GDP, reserves/ 11⁄2 percentage points, between tⳭ3 and ⳮ1; growth
imports, parallel pre- increases by 1 percent in that same period; exports and
mium reserves/imports also rise between ⳮ1 and tⳭ3; the parallel

premium falls.
Krueger (1978) 1951-70, 22 large GDP As in Cooper (1971), devaluations are followed by either a
devaluations recession or a reduction in the rate of growth. These output
effects were small.
Morley (1992) 1974-83, 28 devaluations Capacity utilization After controlling for other fundamentals, the real exchange
in excess of 15 percent. rate is found to have a negative and significant effect on
capacity utilization for up to two years. He finds real
devaluations are associated with sharp declines in
investment.
|
http://www.iie.com
91
Table 7.5 Comparison of severity of crises by region and period,
1970-97
Banking crises (bailout cost as
Currency crises (indexa) share of GDP)
Latin Latin
Period America East Asia Other America East Asia Other
1970-94 48.1 14.0 9.0 21.6 2.8 7.3
1995-97 25.4 40.0 n.a. 8.3 15.0 n.a.
a. See text for description of index’s construction.
Source: Kaminsky and Reinhart (1998).
profile suggests that growth will return to normal within about two years
of the crisis and that the inflationary consequences of the devaluation
will abate within about three years. Yet this pattern would hardly describe
the protracted recovery process of many Latin American economies dur-
ing the 1980s, not even the relatively rapid recovery experienced by Chile.3
The speed at which the economy recovers from a financial crisis will
be heavily influenced by how policymakers respond to the crisis as well
as by external conditions. The high level of international real interest rates
in the 1980s (the highest levels since the 1930s) were hardly conducive
to speedy recovery. Moreover, as suggested in Kaminsky and Reinhart
(1998a), severe currency and banking crises are apt to be associated with
more delayed recoveries. This latter point is particularly relevant to the
recovery from the 1997-98 crises in Asian countries, which are significantly
more severe that the earlier crises in that region.
To analyze this issue formally, we measure the severity of currency
and banking crises, as in Kaminsky and Reinhart (1998). For banking
crises, the measure of severity is the cost of the banking bailout expressed
as a share of GDP. For currency crises, we construct an index that gives
equal weights to reserve losses and currency depreciation. This index is
centered on the month of the currency crisis, and it combines the percent-
age decline in foreign exchange reserves in the six months before the crisis,
since reserve losses typically occur before the central bank capitulates,
and the depreciation of the currency in the six months following the
abandonment of the existing exchange rate arrangement. This latter com-
ponent captures the magnitude of the currency meltdown.
Table 7.5 presents these measures of severity for the 76 currency crises
and 26 banking crises in the Kaminsky-Reinhart sample. For the 1970-94
sample, currency and banking crises were far more severe in Latin
America than elsewhere. The 1970-94 crises in East Asia, by contrast, were
relatively mild and not that different by these metrics from the crises in
3. Chile’s inflation rate was in single digits when it abandoned its crawling peg policy in 1982.

the European countries that largely represent the ‘‘other’’ group. This
divergence may also help explain the subpar performance of the high-
inflation countries during recovery (table 7.3). The picture that emerges
during 1995-97 is distinctly different.
Both in terms of this measure of the severity of the currency crisis as
well as the estimated costs of bailing out the banking sector, the severity
of the recent Asian crises surpasses that of their Latin American counter-
parts in the late 1990s and represents a significant departure from its
historic regional norm. In this sense, the V-shaped recoveries in Asia
have been less protracted than the history of past severe crises would
have suggested.

8
Summary of Results and Concluding
Remarks
In this book we have introduced a set of indicators that, on the basis of

both in-sample and out-of-sample tests, appear to be useful for gauging
vulnerability to currency and banking crises in emerging economies. The
indicators are not precise enough to make fine distinctions in crisis vulner-
ability across countries and over time, but they can draw some distinctions
between the most and least vulnerable groups of countries and recognize
large increases in the vulnerability of a given country over time. As such,
they have the potential to add value as a ‘‘first screen’’ of vulnerability and
as a supplementary tool to other types of analysis of crisis vulnerability. As
suggested in chapter 5, we think the indicators would have been useful
in anticipating the Asian currency and banking crises.
In this chapter, we summarize our key results. Furthermore, in thinking
about future evaluation of such leading indicators of crises, two obvious
questions arise: would publication of the indicators erode their usefulness
in an early warning system, and are there policy implications associated
with the better performing indicators? We discuss each of these questions
in turn.
Summary of Findings
Our main empirical findings can be summarized in 12 main points.

First, banking and currency crises in emerging markets do not typically
arrive without any warning. There are recurring patterns of behavior in
the period leading up to banking and currency crises. Reflecting this
95

Table 8.1 Currency and banking crises: best-performing indicators
High-frequency indicators
Real exchange rate Real exchange rate
Banking crisis Stock prices
Stock prices M2 multiplier
Exports Output
M2/reserves Exports
Output Real interest rate on bank deposits
Low-frequency indicators
Current account balance/GDP Short-term capital inflows/GDP
Current account balance/investment Current account balance/investment
the pack were a high ratio of short-term capital inflows to GDP and a
large current account deficit relative to investment (table 8.1).
Sixth, while there is a good deal of overlap between the best-perform-
ing leading indicators for banking and currency crises, there is enough
of a distinction to warrant treating the two separately. To highlight
two noteworthy differences, the two indicators that serve as proxies for
financial liberalization—namely, a rise in the real interest rate and an
increase in the money multiplier—turned out to be more important for
banking crises than for currency crises, whereas the opposite proved true
for the two indicators designed to capture currency/maturity mismatches
and excessively expansionary monetary policy—namely, a high ratio of
broad (M2) money balances to international reserves and excess M1 money
balances, respectively.
Seventh, while our data on sovereign credit ratings cover only a subsam-
ple of crises and relate to only two of the major rating firms (Moody’s
Investor Services and Institutional Investor), we find that changes in
sovereign credit ratings have performed considerably worse than the
better leading indicators of economic fundamentals in anticipating both
currency and banking crises in emerging economies. In addition, we
find no empirical support for the view that sovereign rating changes have
led financial crises in our sample countries rather than reacting to these
crises. In a similar vein, we have found that interest rate spreads (i.e.,
foreign-domestic real interest rate differentials) are not among the best-
performing group of leading indicators. More empirical work needs to
be done to determine whether these results are robust to other rating
agencies and other samples. Nevertheless, our findings suggest that those
who are looking to ‘‘market prices’’ for early warning of crises in emerging
economies would therefore be advised to focus on the behavior of real
exchange rates and stock prices—not on credit ratings and interest rate
spreads.
SUMMARY OF RESULTS AND CONCLUDING REMARKS 97

tendency, the better-performing leading indicators anticipated between
50 and 100 percent of the banking and currency crises that occurred
over the 26-year sample period. At the same time, even the best leading
indicators send a significant share of false alarms (on the order of one
false alarm for every two to five true signals).1
Second, using monthly data, banking crises in emerging economies
are more difficult to forecast accurately than are currency crises. Within
the sample, the average noise-to-signal ratio is higher for banking crises
than for currency crises, and the model likewise does considerably better
out-of-sample in predicting currency crises than banking crises. It is not
yet clear why this is so. It may reflect difficulties in accurately dating
banking crises—that is, in judging when banking sector distress turns
into a crisis and when banking crises end. For example, by our criteria,
banking distress in Indonesia and Mexico really began in 1992 (and not
in 1997 and 1994, respectively). The absence of high-frequency (monthly
or quarterly) data on the institutional characteristics of national banking
systems probably also is a factor.
Third, there is wide variation in performance across leading indica-
tors, with the best-performing indicators displaying noise-to-signal ratios
that are two to three times better than those for the worst-performing
ones.2 In addition, the group of indicators that show the best (in-sample)
explanatory power also seem, on average, to send the most persistent
and earliest signals. Warnings of a crisis usually appear 10 to 18
months ahead.
Fourth, for currency crises, the best of the monthly indicators were
appreciation of the real exchange rate (relative to trend), a banking
crisis, a decline in stock prices, a fall in exports, a high ratio of broad
money (M2) to international reserves, and a recession. Among the
annual indicators, the two best performers were both current account
indicators—namely, a large current account deficit relative to both GDP
and investment (table 8.1).
Fifth, turning to banking crises, the best (in descending order) of
the 15 monthly indicators were appreciation of the real exchange rate
(relative to trend), a decline in stock prices, a rise in the (M2) money
multiplier, a decline in real output, a fall in exports, and a rise in the
real interest rate. Among the eight annual indicators tested, the best of
1. The construction of the noise-to-signal ratio is described in chapter 2.

2. When an indicator has a noise-to-signal ratio above one, crises would be more likely
when the indicator was not sending a signal than when it was. Similarly, when an indicator
has a conditional probability of less than zero, it means that the probability of a crisis
occurring when the indicator is signaling is lower than the unconditional probability of a
crisis occurring—that is, merely estimating the probability of a crisis according to its historical
average. For example, if currency crises occur in a third of the months in the sample, the
unconditional probability of a crisis is one-third.

Eighth, in most banking and currency crises, a high proportion of the
monthly leading indicators— on the order of 50 to 75 percent— reach
their signaling thresholds. Indeed, both in and out of sample, we found
that fewer than one-sixth of crises occurred with only five or fewer of
the 15 monthly leading indicators flashing. In other words, when an
emerging economy is lurching toward a financial crisis, many of the
wheels come off simultaneously.
Ninth, although we have just scratched the surface on testing our
leading indicators out of sample, we are encouraged by the initial
results— at least for currency crises. We considered two out-of-sample
periods: an 18-month period running from the beginning of 1996 to the
end of June 1997 (just before the outbreak of the Asian financial crisis)
and a 24-month period running from January 1996 to the end of December
1997. Recall that because the indicators lead crises by anywhere from 10
to 18 months, part of the prediction period will lie outside the out-of-
sample observation period.
In each period, we concentrated on the ordinal ranking of countries
according to their crisis vulnerability.3 In chapter 5 we also illustrate for
a subset of the countries how one can calculate from this vulnerability
index the probability of a crisis for a given country over time. As regards
vulnerability to currency crises, the results for the two out-of-sample
periods were quite similar. The five most vulnerable countries (in descend-
ing order) for the 1996 to mid-1997 period were as follows: South Africa,
Czech Republic, Thailand, South Korea, and the Philippines (table 8.2).
For the somewhat longer 1996 to end of December 1997 period, the list
of the five most vulnerable countries is quite similar, although their ordinal
ranking is slightly different: Czech Republic, South Korea, Thailand, South
Africa, and Colombia. If the list were extended to the top seven, Malaysia
would have been included in both periods.
Perhaps the first question to ask is how many of the countries estimated
to be most vulnerable to currency crises in the out-of-sample periods
turned out to have undergone such crises? The answer, as shown in the
upper panel of table 8.2, is almost all of them. According to our index of
exchange market pressure, the Czech Republic, Thailand, South Korea,
and the Philippines all experienced currency crises in 1997 (that is, depreci-
ations or reserve losses that pushed the index of exchange market pressure
to three standard deviations or more above its mean). Colombia’s currency
3. Our preferred measure of vulnerability was an index equal to the weighted average of
‘‘good’’ indicators issuing signals in the out-of-sample period. By ‘‘good’’ indicators, we
mean those that had noise-to-signal ratios less than unity during the 1970-95 period. Taking
the monthly and annual indicators as a group, there were 18 ‘‘good’’ indicators. We used
the inverse of the noise-to-signal ratios as weights for the better indicators. We then ranked
each of the 25 countries in the sample according to the computed value of this index. The
index is meant to capture the probability of a crisis—not necessarily its severity.

Table 8.2 Country rankings of vulnerability to currency crises for
two periodsa
January 1996-June 1997 January 1996-December 1997
Experienced Experienced
Country Rank crisisb Country Rank crisisb
Most vulnerable
South Africa 1 Czech Republic 1 *
Czech Republic 2 * South Korea 2 *
Thailand 3 * Thailand 3 *
South Korea 4 * South Africa 4
Philippines 5 * Colombia 5 *
Least vulnerable
Chile 16 Chile 16
Venezuela 17 Peru 17
Uruguay 18 Venezuela 18
Mexico 19 Mexico 19
Peru 19 Uruguay 20
a. Weighted index is a sum of the weighted signals flashing at any time during the specified
period. Monthly and annual indicators are included. Weights are equal to the inverse noise-
to-signal ratios of the respective indicators.
b. An asterisk (*) indicates that the country experienced a crisis during the out-of-sample
period.
crisis arrives later, in the summer of 1998. Moreover, while South Africa
did not formally make the cut, it could reasonably be classified as a
near miss since it experienced a quasi-crisis in June 1998 (a 14 percent
devaluation cum a 13 percent decline in reserves that pushed the exchange
market pressure index 2.7 standard deviations above its mean). Malaysia,
which just makes it into the group of the seven most vulnerable, did have
a currency crisis in 1997.
Further information on the out-of-sample performance of the leading
indicators of currency crisis can be gleaned by looking for episodes in
which, to borrow from Sherlock Holmes, the ‘‘dogs were not barking’’—
that is, by looking to see how often crises occurred among those countries
estimated to have relatively low vulnerability. The lower panel of table
8.2 indicates the five countries that were estimated to have relatively low
vulnerability to currency crises in 1996-97. As with the high vulnerability
group, the ordinal rankings of countries are very similar across the two
out-of-sample periods, with Venezuela, Peru, and Uruguay slightly shift-
ing their relative positions in the least vulnerable list. Perhaps an explana-
tion as to why the index of vulnerability is relatively low for some of
these countries can be found in the fact that some of these countries were
still recovering from earlier crises (Mexico and Venezuela).
But what about Indonesia, which after all suffered the most severe
currency crisis (beginning in the summer of 1997) among the sample

countries during the out-of-sample period? Why did the model miss it
altogether?4 The explanation probably lies in two areas. First, most of the
best-performing (higher weight) leading indicators were not flashing in
Indonesia’s case. For example, in mid-1997 (just before the outbreak of
the Thai crisis), the real effective exchange rate of the Indonesian rupiah
was only 4 percent above its long-term average—far below its critical
threshold. In a similar vein, neither the decline in stock prices, nor the
decline in exports, nor the change in the ratio of M2 money balances to
international reserves had hit their threshold values.5 Second, at least
three of the factors important in the Indonesian crisis are not included in
our list of indicators: namely, currency/liquidity mismatches on the part
of the corporate sector, regional cross-country contagion effects, and politi-
cal instabilities (in this case associated with the Suharto regime). In this
connection, work reported in Kaminsky and Reinhart (2000) and extended
in chapter 6 suggests that the withdrawal of a common bank lender (in
this case, European and Japanese banks) had a lot to do with contagion
in emerging Asia—and Indonesia in particular—after the outbreak of the
Thai crisis.
The failure of our leading indicators to anticipate the Indonesian crisis
should not, however, obscure the fact that, of the five countries most
adversely affected by the Asian crisis (Thailand, South Korea, Indonesia,
Malaysia, and the Philippines), the indicators placed three of them (Thai-
land, South Korea, and the Philippines) in the top vulnerability group
and another (Malaysia) in the upper third of the country vulnerability
rankings. Given the well-documented failure of private credit ratings and
interest rate spreads to anticipate these Asian currency crises (with the
possible exception of Thailand), and given that these forecasts are based
solely on own-country fundamentals (that is, with no help from contagion
variables), this performance on relative-country vulnerabilities is notewor-
thy. By the same token, the relatively high estimated vulnerability of
several of the Asian emerging economies also challenges the oft-heard
view that the crisis was driven primarily by investor panic, with little
basis in weak country fundamentals.6
4. It should be recognized that none of the existing early warning models—including the
regression-based models—anticipated the Indonesian crisis.
5. Indonesia’s equity prices did suffer a severe decline, but it did not begin until August 1997.
6. Using a very similar approach but a slightly different set of indicators, Kaminsky (1998),
who presents a time series of calculated crisis probabilities for the Asian economies, finds
results that are in line with those shown in tables 1.6 and 1.7—namely, that estimated
currency-crisis vulnerability increased markedly before the 1997 event in Thailand and
moderately in Malaysia and the Philippines. Again, no such increase in estimated vulnerabil-
ity was present for Indonesia. South Korea was not in her sample. Radelet and Sachs (1998)
take the opposing view that the crisis in Asia was mainly attributable to investor panic. As
discussed in chapter 6, we only find that argument to be convincing in the case of Indonesia.

Table 8.3 Country rankings of vulnerability to banking crises for
two periodsa
January 1996-June 1997 January 1996-December 1997
Experienced Experienced
Country Rank crisisb Country Rank crisisb
Most vulnerable
Czech Republic 1 Czech Republic 1
South Korea 2 * South Korea 2 *
Greece 3 Thailand 3 *
South Africa 4 South Africa 4
Thailand 5 * Colombia 5 *
Least vulnerable
Venezuela 15 Chile 16
Chile 16 Argentina 17
Peru 17 Venezuela 18
Uruguay 18 Peru 19
Mexico 19 Uruguay 20
a. Weighted index is a sum of the weighted signals flashing at any time during the specified
period. Monthly and annual indicators are included. Weights are equal to the inverse noise-
to-signal ratios of the respective indicators.
b. An asterisk (*) indicates that the country experienced a crisis during the out-of-sample
period.
Turning to banking crises, the ordinal rankings of country vulnerability

again are quite similar across the two out-of-sample periods, although
the correspondence is slightly lower than was the case for currency crises:
four of the five countries estimated to be most vulnerable to banking
crises are the same across the two periods. Specifically, for the 1996 to
mid-1997 period, the five most vulnerable countries (again in descending
order) were Czech Republic, South Korea, Greece, South Africa, and Thai-
land (table 8.3). When the out-of-sample period is extended through the
end of 1997, Greece drops out of the top five and is replaced by Colombia.
As with the vulnerability rankings for currency crises, it is useful to
ask which of the countries estimated to be most vulnerable to banking
crises actually suffered that fate during the out-of-sample periods. As
suggested earlier, this is intrinsically a tougher question to answer for
banking crises than for currency crises because the identification and
dating of crises are subject to wider margins of error. Recall also that
because our 24-month early warning window for banking crises covers
both the 12-month period preceding the beginning of the crisis as well
as the 12-month period following the onset, successful predictions would
include some crises that began toward the end of 1995 and some that
started no later than early 1998 (as well as those that began in 1996 or 1997).
With these caveats in mind, the picture painted by table 8.3 can be
summarized as follows. Of the five countries estimated to be most vulnera-

ble during January 1996 to the end of June 1997, two experienced banking
crises that fall in our prediction window. Specifically, we consider South
Korea’s banking crisis to have begun in January 1997, with the loan losses
stemming from the bankruptcy of Hanbo Steel. In a similar vein, we date
Thailand’s banking crisis as starting in May 1996, when the Ministry of
Finance took control of Bangkok Bank of Finance (following a run on
deposits). A third member of the most vulnerable group, the Czech Repub-
lic, also experienced a banking crisis although the timing is not clear-cut.
The start of the Czech crisis could be dated in August 1996, reflecting the
closure of Kreditni Banka; alternatively, one could also defend a much
earlier starting date, namely September 1993, when Kreditni was initially
placed under supervision.7 Some researchers (e.g., Kaminsky and Reinhart
1999) also classify Malaysia and the Philippines as having registered
banking crises in 1997. The results for the longer out-of-sample period,
shown in the upper panel of table 8.3, are quite similar: the same three
countries (South Korea, Thailand, and Czech Republic) make up the list
of successful banking-crisis predictions. For the more recent sample,
Colombia is also added to the list of successful predictions. In April 1998
the bailout by Banco de la Republica of several finance companies facing
severe difficulties with mounting losses in their loan portfolios intensified
in earnest, and banking sector problems deepened throughout 1998-99.
What about the group of countries estimated to be least vulnerable to
banking crises? As seen in the lower panel of table 8.3, four of the five
countries in this category are common to both sample periods: Uruguay,
Venezuela, Peru, and Chile. Mexico appears only in the shorter period,
while Argentina makes the least vulnerable list only in the longer period.
For all five of the countries estimated to be least vulnerable, no banking
crisis appears to have taken place during the out-of-sample periods. As
was the case with the forecasting of currency crises, Indonesia (which is
ranked 11 or 12, depending on the sample chosen) emerges as a major
misclassification, although timing problems somewhat cloud the issue.
Many observers would regard the severity of Indonesia’s financial sector
problems in 1997 as constituting a ‘‘new’’ banking crisis; others might
argue that these difficulties constituted a continuation of the banking
problems that began in 1992 with the collapse of Bank Summa. In any case,
it is clear that the model was not picking up the increase in Indonesia’s
vulnerability in 1997. Mexico presents another timing problem. Mexico
remained in the throes of a banking crisis throughout the out-of-sample
period and thus could be classified as highly vulnerable. At the same
time, most studies (e.g., Demirgüç-Kunt and Detragiache 1998; IMF 1998b)
regard the Mexican banking crisis as having started at least as early as
7. The Czech banking crisis was not included in our in-sample test, and hence the model
is not calibrated to account for this crisis.

1994. Here, too, the model seems to have difficulty in identifying changes
in vulnerability when they occur in the context of continuing banking
problems.
Looking at both the high and low vulnerability groups, it is clear that
the early warning model is less successful out of sample in anticipating
banking crises than it is in anticipating currency crises. The problem is
not so much that the model misses many banking crises that do occur
but rather that it generates too many false positives or ‘‘noise,’’—that is,
it predicts more cases of banking crisis vulnerability than actually occur.
In this connection, it is worth noting that we classify only five or six
episodes as meeting our criteria for a banking crisis during the out-of-
sample period (that is, the period running roughly from late 1995 to early
1998). This list comprises South Korea, Thailand, the Czech Republic,
Indonesia, and Malaysia.8 Of these five crisis cases, three of the countries
concerned (South Korea, Thailand, and the Czech Republic) were mem-
bers of our ‘‘most vulnerable’’ group.9 This might be considered fair
performance. Difficulties in forecasting Asian banking crises in 1997 seem
to be common to the leading forecasting models—be they signals
approach models or regression-based models. For example, Demirgüç-
Kunt and Detragiache (1998), using a multivariate logit model, report that
the conditional probabilities for banking crises in the five most adversely
affected Asian economies were actually below the unconditional crisis
probabilities (Furman and Stiglitz 1998). Similarly, Kaminsky (1998) finds
that estimated crisis probabilities were rising sharply in the case of the
Thai banking crisis and moderately in the case of the Philippines, but not
for either Malaysia or Indonesia.
We conducted a number of experiments, which are described in chapter
5, to help gauge the robustness of our results on the ordinal ranking of
country vulnerability to currency and banking crises. In one exercise,
instead of basing the ordinal vulnerability rankings exclusively on weights
derived from the noise-to-signal ratios, we looked at both the proportion of
indicators signaling a crisis and the proportion of the top eight indicators
signaling a crisis. In another exercise, we looked at various indicators
signaling both banking and currency crises and calculated ‘‘average’’
vulnerability to banking and currency crises combined. And in yet another
set of exercises, we liberalized the optimal thresholds for each of the
indicators by 5 percent, thereby making it less likely that we would miss
crises that were unfolding, albeit at the cost of predicting crises that never
8. The Malaysian crisis would probably best be regarded as beginning in March 1998,
when the central bank announced losses at Sime Bank and elsewhere and when Malaysian
President Datuk Seri Mahathir bin Mohamad pledged state funds to prop up weak
institutions.
9. Malaysia was ranked fourteenth (out of 25 countries) in the shorter period and tenth in
the longer one.

occurred. While these robustness exercises not surprisingly generated
some changes in the ordinal rankings, perhaps the most important finding
was that the same ‘‘core’’ set of vulnerable countries—the Czech Republic,
South Korea, South Africa, Greece, Colombia, Thailand, the Philippines,
and Malaysia—consistently remained in the top tier of the vulnerability
list. It is also noteworthy that none of these sensitivity exercises anticipated
the Indonesian crisis.
All in all, we regard the out-of-sample performance of the signals
approach as encouraging— particularly as regards anticipating currency
crises in the Asian crisis countries.10 With the exception of Indonesia,
the model did well in identifying the countries with relatively high vulner-
ability. In addition, the model gave strong signals for Brazil, the Czech
Republic, South Africa, and Colombia, which also experienced crises (or
turbulence) outside the Asian region. The results for banking crises were
less impressive. While we would not place much confidence in the precise
estimated ordering of vulnerability across countries, we think the signals
approach looks promising for making distinctions between the vulnerabil-
ity of countries near the top of the list and those near the bottom—that
is, it may be useful as a ‘‘first screen,’’ which can then be followed by
more in-depth country analysis.
Some others are pessimistic about both the potential and actual out-of-
sample performance of signals-based leading-indicator models of cur-
rency crises, including their track record in anticipating the Asian financial
crisis.11 They argue that when such models do perform seemingly well,
it is often because they rely on ‘‘black box’’ simple contagion variables
(for example, the number of crises that have occurred in the previous
period), that the methodology embedded in the signals approach is biased
toward overpredicting crises in countries with good histories and this
explains its successes in predicting currency crises in Asia, and that both
in-sample and out-of-sample performance would be better if the good
indicator variables were entered linearly (rather than sending a signal
only when the indicator crossed its threshold) and if the weights on the
individual indicators were estimated by a regression (rather than selected
from an iterative noise-to-signal test one at a time). These critics also
argue that the correlation between the severity of observed currency crises
and crisis vulnerability predicted by the signals approach was low (at
least in 1996) and that (also in 1996) there did not seem to be a marked
distinction between the calculated currency crisis vulnerabilities of several
10. This is consistent with the results of a recent IMF study (Berg and Pattillo 1999), which
found that the signals model of Kaminsky, Lizondo, and Reinhart (1998) did a better job
of predicting the Asian crisis than the models of Frankel and Rose (1996) and of Sachs,
Tornell, and Velasco (1996).
11. See Berg and Pattillo (1999), The Economist (1998), Furman and Stiglitz (1998), IMF
(1998c), and Wyplosz (1997, 1998).

noncrisis countries (particularly Argentina and Mexico) and the Asian
crisis countries (Thailand and Indonesia). Further, they find that leading-
indicator models have poor in-sample performance in forecasting cur-
rency crises for developing countries, especially in emitting Type II errors
(false positives) and that publication of a vulnerability index could precipi-
tate a crisis. We do not find these criticisms to be persuasive.
In the two studies (Berg and Pattillo 1999; Furman and Stiglitz 1998)
that have explicitly run out-of-sample horse races between the Kaminsky-
Reinhart signals model and two other regression-based models of cur-
rency crises (Frankel and Rose 1996; Sachs, Tornell, and Velasco 1997),
both concluded that the signals approach does better.
Wyplosz (1998) bases his pessimistic conclusions on the in-sample per-
formance of leading-indicator models of currency crises in developing
countries using the Frankel and Rose (1996) model—not the Kaminsky-
Reinhart signals approach. Using an abbreviated search technique for the
optimal threshold for various indicators, Wyplosz finds that (using a 5
percent threshold) 62 of 86 currency crises are detected, while the model
signals wrongly—that is, emits false positives—in 43 percent of the crises.
Our results are more favorable than Wyplosz’s for developing-country
currency crises (in-sample). Out of sample, we find that the false positives
problem is more serious for the banking crises than for currency crises.
While we present some new results on cross-country contagion in chap-
ter 6, the out-of-sample results—both in earlier Kaminsky-Reinhart stud-
ies and in this book—do not rely at all on cross-country contagion; instead,
they reflect only own-country fundamentals.
There is (at least to our knowledge) no empirical evidence to support
the view that imposing a common absolute threshold for indicator variables
would produce better in-sample and out-of-sample performance than our
procedure of imposing a common percentile threshold and allowing the
absolute threshold to differ across countries. Nor, as we have argued earlier,
does it seem more reasonable on a priori grounds to impose the one-size-
fits-all restriction on countries with different histories—quite the contrary.
As for the alleged influence of our procedure in the context of forecasting
the Asian financial crisis, one would have thought that if this bias were
large, it might have led to a very successful prediction of crises in the Asian
countries, yet some of these same critics find that the signals approach does
very poorly in forecasting currency crises in these countries.
While more work is clearly needed to assess the robustness of the results
to different out-of-sample periods (since these differ and seem to generate
different outcomes across studies), we do not find that there was little
distinction in estimated currency-crisis vulnerabilities between most of the
Asian crisis countries, on the one hand, and some other (Latin American)
noncrisis countries, on the other. As indicated earlier, we found that
Thailand, the Philippines, and Malaysia had higher estimated currency-

crisis probabilities in 1996-97 than did Argentina and Mexico—not the
other way around. Thailand was near the top of our vulnerability list—
not near the bottom. Also, it is not obvious that out-of-sample comparisons
based on the severity of crises are more meaningful than those (as above)
that concentrate on the crisis/no crisis distinction.
In short, just as we emphasized that it is important not to oversell the
potential of early warning models to predict crises in emerging economies,
we think some of the critics are too quick to dismiss the usefulness of
these models because of a mixed out-of-sample performance based on
runs from a single period. We should also keep in mind the apparent
inability of non-model-based forecasts to foresee the Asian crisis. In our
view, much more empirical work will need to be done before we can
draw reliable conclusions on the out-of-sample performance of the sig-
nals approach.
Examining a somewhat more limited sample (20) of small developed
and emerging economies over the 1970-98 period, we looked for patterns
in the cross-country contagion of currency crises. Following Eichengreen,
Rose, and Wyplosz (1996), we define contagion as a case in which the
presence of a crisis elsewhere increases the probability of crisis at home,
even when the fundamentals have been taken into account. We considered
four channels through which shocks can be transmitted across borders:
two dealt with trade links (bilateral trade flows and trade competition
in third-country markets) and two channels addressed financial links
(correlation of asset returns in global portfolios and reliance on a common
bank lender). We also demonstrated how these four contagion channels
could be combined and weighted appropriately to form a ‘‘contagion
vulnerability index.’’
This exercise led us to our tenth main finding: that cross-country conta-
gion adds significantly to own-country fundamentals in furthering an
understanding of emerging market vulnerability to financial crises and
that (at least historically) contagion has operated more along regional
than global lines. According to our contagion vulnerability index, Brazil,
Argentina, and the Philippines had high vulnerability to the 1994 Mexican
peso crisis; Malaysia, South Korea, and Indonesia had high vulnerability
to the 1997 Thai crisis; and Argentina, Chile, and Uruguay had high
vulnerability to the 1999 Brazilian crisis. Although it is difficult to separate
financial contagion channels from trade channels (since countries linked
in trade also are linked in finance) we concluded that withdrawal of a
common bank lender (particularly Japanese banks) and high correlation
of asset returns were important in the contagion in Asia in 1997-98.
Eleventh, in addition to studying the antecedents of crises we also
drew on our data base for information on the aftermath of crises—with
particular attention to the speed with which emerging economies return
to ‘‘normal’’ after a currency or banking crises. We defined normal in

two alternative ways: first, as a period of ‘‘tranquility’’ that excludes not
only the crisis years but also the two- to three-year windows before
and after the crisis, and second, as the average of the two years just
preceding crises.12
One of our most robust findings was that the deleterious effects on
economic activity are more lingering for banking crises than for cur-
rency crises.13 For example, whereas it took about two years for economic
growth to return to the average of the two precrisis years after a currency
crisis, that recovery was not evident even three years after a banking
crisis. One possible explanation for this difference is that, whereas a
currency crisis sharply reduces external sources of funding, a banking
crisis curtails access to both external and domestic sources of finance for
households and firms—that is, the ‘‘credit crunch’’ is more severe in the
wake of banking crises. This more sluggish recovery pattern for banking
crises was also evident for exports, imports, and stock prices. For instance,
whereas exports recover relatively quickly (eight months) and ahead of
the rest of the economy following currency crises, they continue to sink
for two years following the onset of a banking crisis. Two other dimensions
of the protracted nature of banking crises are that it takes about three to
four years for a banking crisis to be resolved and it takes on the order of
a year and half between the onset of a banking crisis and its peak. All of
this highlights the challenge faced by the Asian crisis countries in sustain-
ing their recoveries: not only did the most affected countries in emerging
Asia suffer from currency crises that were accompanied by banking crises
(what Kaminsky and Reinhart 1999 dub ‘‘twin crises’’), but the banking
crises themselves are very severe.
Our analysis of the aftermath of crises does not lend support to the
notion that devaluations in emerging economies generate deflation.
Instead, we find that devaluations are inflationary, that the pass-through
to prices is incomplete (hence, devaluations lead to real depreciations),
and that it takes between two and three years after a devaluation for
inflation to return to the average of the two precrisis years.
Last but not least, we offer a number of suggestions for improving
early warning models of currency and banking crises. In our view, four
directions for future research merit priority.
As hinted above, more work needs to be done to determine the out-
of-sample forecasting properties of these models—be it signals approach
models or regression-based logit or probit models. In particular, it would
be useful to know how robust ‘‘who’s next’’ country rankings of vulnera-
12. More specifically, the ‘‘tranquil’’ period excludes the 24 months before and after currency
crises and the 24 months before and 36 months after banking crises.
13. In the cases where currency and banking crises coincide, the postcrisis performance
would show up in both the averages for currency and banking crises. See chapter 7 for details.

bility are in the face of changes in the forecasting period, different compos-
ite indicators, different definitions and transformations of the indicator
variables (e.g., alternative definitions of the effective real exchange rate
or of real exchange rate ‘‘misalignments,’’ and alternative ways of dating
banking crises or selecting the early warning ‘‘window’’), and the restric-
tions imposed in the different models (e.g., imposing thresholds versus
allowing indicators to enter linearly, imposing absolute thresholds versus
common percentile ones). It may turn out, as suggested by Berg and
Pattillo (1999), that combining certain features of the signals approach
and the regression-based models would improve forecasting (e.g., using
the signals approach to select the good indicators and then estimating
the weights and crisis probabilities using a regression-based format).
We think there is scope to bring other indicators into these horse races.
For example, Kaminsky (1998) has found that the share of short-term debt
in total foreign debt, as well as a proxy for capital flight (by residents of
emerging economies), do quite well in anticipating currency and banking
crises within the sample. Looking at the run-up to the Asian financial
crisis, Furman and Stiglitz (1998) likewise make a good case for including
the ratio of short-term external debt to international reserves as an indica-
tor in future early warning exercises. If monthly data could be obtained
both on real property prices and on the exposure of the banking system
to property, those too could prove very helpful.
A plausible extension would be to bring institutional characteristics of
weak banking systems into the forecasting of banking crises. There is a
strong presumption that the following all matter for vulnerability to
banking crises: weak accounting, provisioning, and legal frameworks;
policy-directed lending; the ownership structure of the banking system
(government ownership, foreign ownership, and so on); the incidence of
connected lending; the extent of diversification; the quality of banking
supervision; and the incentive-compatibility of the official safety net. Yet
it is only very recently that any of these factors have begun to enter the
empirical literature.14
The main constraint on making use of these institutional characteristics
is that one cannot get high-frequency measurements of them. Indeed, for
some of these characteristics (e.g., the share of government ownership),
it has proved difficult to get even annual data that is less than two or
three years old. This means that such variables have to be introduced as
zero-one dummy variables in a time-series context. There would be more
scope to take advantage of such factors in cross-section work—that is, in
explaining cross-country differences in the incidence of banking crises
over long periods.
14. See, for example, Demirgüç-Kunt and Detragiache (1998), who introduce law enforce-
ment and deposit-insurance variables into their banking crisis model.

Lastly, we think the ongoing work on modeling the nature of cross-
country contagion of crises should be extended. One of the lessons of the
last few major crises (that is, the Mexican crisis and the Asian/global
financial crisis) is that the channels of cross-country contagion are more
numerous and complicated than we thought earlier. Each of these factors
seem to play a part in contagion: trade links (bilateral and third party),
perceived similarities in macroeconomic and financial vulnerability, the
dynamics of competitive devaluations, induced effects on primary com-
modity prices, financial links operating via withdrawal of a common bank
or mutual fund lender, liquidity and margin-call effects operating via the
regulatory framework, and perceived changes in the rescheduling cum
capital-account convertibility regime (such as took place after the Russian
unilateral rescheduling/default in August 1998 and the Malaysian imposi-
tion of wide-ranging capital controls). We need to find ways to incorporate
more of these channels of contagion in our forecasting models.
Would the Publication of the Indicators Erode

Their Early Warning Role?
It is sometimes argued that if the indices of crisis vulnerability were made
publicly available on a timely basis, such publication could prompt a self-
fulfilling run on a country’s currency or its banks. Alternatively, it is
sometimes asserted that if countries really paid heed to the message of
the indicators and took preemptive action, then the indicators would
lose predictive power. This latter outcome would, of course, be highly
desirable. While neither of these arguments can be dismissed lightly, we
would regard both as exaggerated.
The conditions for generating this type of self-fulfilling runs are likely
to be relatively rare. As we have stressed throughout this book, the signals
approach is useful in identifying cases of high vulnerability to crises.
Explaining the precise timing of the crises remains an elusive goal. To
the extent that timing matters and that investment decisions are made
under uncertainty, there is little reason to expect that moderate increases in
the extent of vulnerability are likely to be sufficient to prompt a speculative
attack. As argued earlier, negative announcements of the readings in the
leading indicator index of business cycles—which have been published
for many years—do not cause a recession, although investors certainly
take these readings into account.
By the same token, we think it unlikely that publication would cause
the indicators to lose most of their predictive ability. This could certainly
occur if preemptive policy was an everytime, everywhere phenomenon
and if such preemptive policies were in fact successful in staving off
crises. These are strong assumptions. All too often, policymakers are
inclined to ignore distress signals on the grounds that, this time, the
situation is really different or that there are overriding political objectives

against corrective action. Furthermore, even if the signals are heeded and
corrective policy actions are taken, they may not be sufficient to prevent
the crisis. If the feedback from the indicators to corrective policy action
were strong and consistent, we would not have been able to identify
useful indicators in the first place. Of course, one could always speculate
that future policymakers will be wiser than their predecessors, but that
remains to be seen.
Do the Better Performing Indicators Carry

Policy Implications?
The empirical evidence presented in this book can be seen as supporting
the case for including leading indicators in the analytical tool kit for
diagnosing crisis vulnerability. But can one go farther and draw policy
implications from the performance of the better univariate indicators?
One should recall that the signals approach outlined in this book and
in earlier works by Kaminsky and Reinhart looks for empirical regularities
in the behavior of macroeconomic and financial variables in the run-up
to currency and banking crises. We thus cannot fully identify from this
exercise the channels by which policies affect economic outcomes.
For some of the indicators, the results have clear implications for macro-
economic and exchange rate policies; for others there is no obvious link.
For example, the performance of M2/reserves as an indicator of currency
crises is suggestive of the desirability of avoiding large discrepancies
between liquid liabilities and liquid assets. In this regard, the policy
implication would be to encourage emerging market countries to maintain
high liquidity ratios, prearranged lines of credit, and an ample stock of
reserves. Much of the behavior of other indicators, notably rising real
interest rates and money multipliers, are associated with financial liberal-
ization. Indeed, the reliability of these indicators in anticipating banking
crises may warn against hasty liberalizations. On the other hand, real
exchange rate overvaluations are an important indicator of both currency
and banking crises, but the burning policy question that remains unan-
swered is how emerging market countries can avoid these costly periodic
overvaluations. Real exchange rate targeting has been tried by countries
as diverse as Brazil, Chile, Indonesia, and Colombia, but the outcomes
were quite dissimilar—particularly as regards their consequences for
inflation.15 In the case of some other indicators, such as stock prices, the
policy implications are even less obvious.
In short, while we regard the empirical work on early warning indica-
tors as consistent with many stories of the origins of currency and banking
crises, one has to be careful not to overinterpret the results, as alternative
explanations of crises often yield observationally equivalent implications.
15. Calvo, Reinhart and Végh (1995) present empirical evidence on this issue.

Appendix
Data and Definitions
Currency Crisis Index

The index is a weighted average of exchange rate and reserve changes,
with weights such that the two components of the index have equal
conditional volatilities. Since changes in the exchange rate enter with a
positive weight and changes in reserves have a negative weight attached,
readings of this index that were three standard deviations or more above
the mean were cataloged as crises. For countries in the sample that had
hyperinflation, the construction of the index was modified. While a 100
percent devaluation may be traumatic for a country with low to moderate
inflation, a devaluation of that magnitude is commonplace during hyper-
inflations. A single index for the countries that had hyperinflation episodes
would miss sizable devaluations and reserve losses in the moderate infla-
tion periods, as the historic mean is distorted by the high-inflation episode.
To avoid this, we divided the sample according to whether inflation in
the previous six months was higher than 150 percent and then constructed
an index for each subsample. Our cataloging of crises for these countries
coincides fairly tightly with our chronology of currency market disrup-
tions. Eichengreen, Rose, and Wyplosz (1995) also include interest rates
in this index; however, our data on market-determined interest rates for
developing countries does not span the entire sample.
The Indicators
Sources include the IMF’s International Financial Statistics (IFS), the Interna-
tional Finance Corporation’s (IFC) Emerging Market Indicators, and the
111

World Bank’s World Development Indicators. When data were missing
from these sources, central bank bulletins and other country-specific
sources were used as supplements. Unless otherwise noted, we used 12-
month percentage changes.
M2 multiplier: the ratio of M2 to base money (IFS lines 34 plus 35)
divided by IFS line 14.
Domestic credit/nominal GDP: IFS line 52 divided by IFS line 99b
(interpolated). Monthly nominal GDP was interpolated from annual or
quarterly data.
Real interest rate on deposits: IFS line 601, monthly rates, deflated
using consumer prices (IFS line 64) expressed in percentage points.
Ratio of lending rate to deposit rate: IFS line 60p divided by IFS line
601 was used in lieu of differentials to ameliorate the distortions caused
by the large percentage point spreads observed during high inflation.
In levels.
Excess real M1 balance: M1 (IFS line 34) deflated by consumer prices
(IFS line 64) less an estimated demand for money. The demand for real
balances is determined by real GDP (interpolated IFS line 99b), domestic
consumer price inflation, and a time trend. Domestic inflation was used
in lieu of nominal interest rates, as market-determined interest rates were
not available during the entire sample for a number of countries; the time
trend (which can enter log-linearly, linearly, or exponentially) is motivated
by its role as a proxy for financial innovation and/or currency substitution.
Excess money supply (demand) during precrisis periods (mc) is reported
as a percentage relative to excess supply (demand) during tranquil times
(mt)—that is, 100 ⳯ (mc ⳮ mt)/mt.
M2 (in US dollars)/reserves (in US dollars): IFS lines 34 plus 35 con-
verted into dollars (using IFS line ae) divided by IFS line 1L.d.
Bank deposits: IFS line 24 plus 25.
Exports (in US dollars): IFS line 70.
Imports (in US dollars): IFS line 71.
Terms of trade: the unit value of exports (IFS line 74) over the unit
value of imports (IFS line 75). For those developing countries where
import unit values (or import price indices) were not available, an index
of prices of manufactured exports from industrial countries to developing
countries was used.
Real exchange rate: based on consumer price indices (IFS line 64) and
defined as the relative price of foreign goods (in domestic currency) to
the price of domestic goods. If the central bank of the home country pegs
the currency to the dollar (or deutsche mark), the relevant foreign price
index is that of the United States (or Germany). Hence for all the European
countries the foreign price index is that of Germany, while for all the
other countries consumer prices in the United States were used. The trend
was specified as, alternatively, log-linear, linear, and exponential; the best

fit among these was selected on a country-by-country basis. Deviations
from trend during crisis periods (dc) were compared with the deviations
during tranquil times (dt) as a percentage of the deviations in tranquil
times (i.e., 100 ⳯ (dc ⳮ dt)/dt).
Reserves: IFS line 1L.d.
Domestic-foreign interest rate differential on deposits: monthly rates
in percentage points (IFS line 601). Interest rates in the home country are
compared with interest rates in the United States (or Germany) if the
domestic central bank pegs the currency to the dollar (or deutsche mark).
The real interest rate is given by 100 ⳯ [(1 Ⳮ it )pt /ptⳭ1ⳮ1 ].
Output: for most countries, industrial production (IFS line 66). How-
ever, for some countries (the commodity exporters) an index of output
of primary commodities is used (IFS line 66aa).
Stock prices (in dollars): IFC global indices are used for all emerging
markets; for industrial countries the quotes from the main bourses are
used.
Overall budget balance/GDP: consolidated public-sector balance as
share of nominal GDP (World Bank Debt Tables).
Current account balance a share of GDP: (World Bank, World Develop-
ment Report database available in CD ROM).
Current account balance a share of investment: current account
divided by gross investment (World Bank, World Development Report
database available in CD ROM).
Short-term capital inflows: Short-term capital flows as a percent of
GDP, (World Bank, World Debt Tables, database available in CD ROM).
Foreign direct investment (FDI): FDI as a share of GDP (World Bank,
World Debt Tables, database available in CD ROM).
General government consumption/GDP: General government con-
sumption, national income accounts basis as a percent of GDP, annual
growth rate (World Bank, World Development Report database available
in CD ROM).
Central bank credit to the public sector/GDP: Annual growth rate
(World Bank, World Development Report database available in CD ROM).
Net credit to the public sector/GDP: Annual growth rate (World Bank,
World Development Report database available in CD ROM).
DATA AND DEFINITIONS 113

References
Ades, Alberto, Rumi Masih, and Daniel Tenegauzer. 1998. A New Framework for Predicting
Financial Crises in Emerging Markets. New York: Goldman Sachs. December.
Agenor, Pierre-Richard, Jagdeep S. Bhandari, and Robert P. Flood. 1992. Speculative Attacks
and Models of Balance of Payments Crises. IMF Staff Papers 39 (June): 357-94.
Bank for International Settlements (BIS). 1998. The Maturity, Sectoral, and Nationality Distribu-
tion of International Bank Lending. Basel: Bank of International Settlements.
Berg, Andrew, and Catherine Pattillo. 1999. Are Currency Crises Predictable? A Test. IMF
Staff Papers, no. 2 (June): 107-38.
Blanco, Herminio, and Peter M. Garber. 1986. Recurrent Devaluation and Speculative Attacks
on the Mexican Peso. Journal of Political Economy 94 (February): 148-66.
Borensztein, Eduardo, and José de Gregorio. 1998. Devaluations and Inflation after Currency
Crises. Photocopy. Washington: International Monetary Fund.
Calomiris, Charles. 1997. The Post-Modern Bank Safety Net: Lessons from Developed and Develop-
ing Countries. Washington: American Enterprise Institute.
Calvo, Guillermo A. 1998. Varieties of Capital-Market Crises. In Guillermo A. Calvo and
M. King, eds., The Debt Burden and Its Consequences for Monetary Policy. New York:
MacMillan Press.
Calvo, Guillermo A., and Morris Goldstein. 1996. What Role for the Official Sector? In
Guillermo A. Calvo, Morris Goldstein, and Eduard Hochreiter, eds., Private Capital
Flows to Emerging Markets after the Mexican Crisis. Washington: Institute for Interna-
tional Economics.
Calvo, Guillermo A., and Enrique Mendoza. 1996. Mexico’s Balance-of-Payments Crisis: A
Chronicle of a Death Foretold. Journal of International Economics 41, no. 3/4 (November):
235-64.
Calvo, Guillermo A., and Enrique Mendoza. 2000. Rational Contagion and the Globalization
of Securities Markets. Journal of International Economics, volume 51:79-113.
Calvo, Guillermo, Carmen M. Reinhart, and Carlos Vegh. 1995. Targeting the Real Exchange
Rates: Theory and Evidence. Journal of International Economics, volume 47:97-133.
Calvo, Guillermo A., and Carmen M. Reinhart. 2000. Fear of Floating. Photocopy. University
of Maryland.
115

Calvo, Sara, and Carmen M. Reinhart. 1996. Capital Flows to Latin America: Is There
Evidence of Contagion Effects? In Guillermo A. Calvo, Morris Goldstein, and Eduard
Hochreiter, eds., Private Capital Flows to Emerging Markets after the Mexican Crisis. Wash-
ington: Institute for International Economics.
Cantor, Richard, and Frank Packer. 1996a. Determinants and Impact of Sovereign Credit
Ratings. Federal Reserve Bank of New York Economic Policy Review (October): 1-15.
Cantor, Richard, and Frank Packer. 1996b. Sovereign Risk Assessment and Agency Credit
Ratings. European Financial Management 2: 247-56.
Caprio, Gerald Jr., and Daniela Klingebiel. 1996a. Bank Insolvency: Bad Luck, Bad Policy,
or Bad Banking? In Annual World Bank Conference on Development Economics. Washington:
World Bank.
Caprio, Gerald Jr., and Daniela Klingebiel. 1996b. Bank Insolvency: Cross-Country Experi-
ences. Washington: World Bank.
Claessens, Stijn, and Thomas Glaessner. 1997. Are Financial Sector Weaknesses Undermining
the Asian Miracle? Washington: World Bank.
Cooper, Richard. 1971. Currency Devaluation in Developing Countries. In G. Ranis, ed.,
Government and Economic Development. New Haven, CT: Yale University Press.
Corsetti, Giancarlo, Paolo Pesenti, and Nouriel Roubini. 1998. What Caused the Asian
Currency and Financial Crisis? Part II: The Policy Debate. National Bureau of Economic
Research Working Paper 6834. Cambridge, MA: National Bureau of Economic Research.
Corsetti, Giancarlo, Paolo Pesenti, Nouriel Roubini, and Cedric Tille. 1998. Structural Links
and Contagion Effects in the Asian Crisis. New Haven, CT: Yale University Press.
Council on Foreign Relations (CFR). 1999. Safeguarding Prosperity in a Global Financial System:
the Future International Financial Architecture. Washington: Institute for International Eco-
nomics.
Cumby, Robert, and Sweder van Wijnbergen. 1989. Financial Policy and Speculative Runs
with a Crawling Peg: Argentina 1979-81. Journal of International Economics 17 (August):
111-27.
Demirgüç-Kunt, Asli, and Enrica Detragiache. 1998. The Determinants of Banking Crises
in Developing and Developed Countries. IMF Staff Papers, 45 no. 1: 81-109.
Diamond, Douglas, and Phillip Dybvig. 1983. Bank Runs, Liquidity, and Deposit Insurance.
Journal of Political Economy 91 (June): 401-19.
Diebold, Francis, and Glen Rudebusch. 1989. Scoring the Leading Indicators. Journal of
Business 62, no. 3 (July): 369-91.
Dooley, Michael P. 1997. A Model of Crises in Emerging Markets. National Bureau of Economic
Research Working Paper 6300. Cambridge, MA: National Bureau of Economic Research.
Dornbusch, Rudiger, Ilan Goldfajn, and Rodrigo O. Valdés. 1995. Currency Crises and
Collapses. Brookings Papers on Economic Activity 2: 219-95.
Doukas, John. 1989. Contagion Effect on Sovereign Interest Rate Spreads. Economic Letters
29: 237-41.
The Economist. 1997. Rating Agencies: Risks Beyond Measure. (13 December).
Edwards, Sebastian. 1986. Are Devaluations Contractionary? Review of Economics and Statistics
68: 501-08.
Edwards, Sebastian. 1989. Real Exchange Rates, Devaluation, and Adjustment: Exchange Rate
Policy in Developing Countries. Cambridge, MA: MIT Press.
Eichengreen, Barry, and Andrew Rose. 1998. Staying Afloat When the Wind Shifts: External
Factors and Emerging-Market Banking Crises. National Bureau of Economic Research
Working Paper 6370 (January). Cambridge, MA: National Bureau of Economic Research.
Eichengreen, Barry, Andrew Rose, and Charles Wyplosz. 1995. Exchange Market Mayhem:
The Antecedents and Aftermath of Speculative Attacks. Economic Policy 21 (October):
249-312.
Eichengreen, Barry, Andrew Rose, and Charles Wyplosz. 1996. Contagious Currency Crises.
National Bureau of Economic Research Working Paper 5681 (July). Cambridge, MA:
National Bureau of Economic Research.

Eschweiler, Bernhard. 1997a. Emerging Asia: The Fallout After the FX Crisis. Asian Financial
Markets (17 October). Singapore: JP Morgan.
Eschweiler, Bernhard. 1997b. Did the Market See the Asian Crisis Coming? Paper presented
at World Bank Conference on the Asian Crisis, Washington (4 October).
Fischer, Stanley. 1993. The Role of Macroeconomic Factors on Growth. Journal of Monetary
Economics 32, no. 3: 395-415.
Flood, Robert, and Nancy P. Marion. 1999. Perspectives on the Recent Currency Crisis
Literature. International Journal of Finance and Economics 4, no. 1 (January): 1-26.
Forbes, Kristen, and Roberto Rigobon. 1998. Measuring Stock Market Contagion: Conceptual
Issues and Empirical Tests. Photocopy. Cambridge: MIT.
Frankel, Jeffrey A., and Andrew K. Rose. 1996. Exchange Rate Crises in Emerging Markets.
Journal of International Economics 41, no. 3/4 (November): 351-68.
Frankel, Jeffrey A., and Sergio L. Schmukler. 1996. Crisis, Contagion, and Country Funds:
Effects on East Asia and Latin America. In Reuven Glick, ed., Managing Capital Flows
and Exchange Rates: Perspectives from the Pacific Basin. Cambridge, UK: Cambridge Univer-
sity Press.
Furman, Jason, and Joseph E. Stiglitz. 1998. Economic Crises: Evidence and Insights from
East Asia. Brookings Papers on Economic Activity 2 (June): 1-114
Galbis, Vicente. 1993. High Real Interest Rates under Financial Liberalization: Is There a Problem?
International Monetary Fund Working Paper 7. Washington: International Monetary
Fund (January).
Gavin, Michael, and Ricardo Hausman. 1996. The Roots of Banking Crises: The Macroeco-
nomic Context. In Ricardo Hausman and Liliana Rojas-Suarez, eds., Banking Crises in
Latin America. Washington: Inter-American Development Bank.
Gerlach, Stefan, and Frank Smets. 1995. Contagious Speculative Attacks. European Journal
of Political Economy, vol. 11: 5-63.
Glick, Reuven, and Andrew Rose, 1998, Contagion and Trade: Why Are Currency Crises
Regional? National Bureau of Economic Research Working Paper 6806. Cambridge:
National Bureau of Economic Research.
Goldfajn, Ilan, and Rodrigo O. Valdés. 1995. Balance of Payments Crises and Capital Flows:
The Role of Liquidity. Photocopy. Cambridge: MIT.
Goldfajn, Ilan, and Rodrigo O.Valdés. 1998. Are Currency Crises Predictable?, European
Economic Review 42 (May): 887-895.
Goldstein, Morris. 1996. Presumptive Indicators/Early Warning Signals of Vulnerability to
Financial Crises in Emerging Market Economies. Photocopy. Washington: Institute for
International Economics.
Goldstein, Morris. 1997. The Case for an International Banking Standard. POLICY ANALYSES IN
INTERNATIONAL ECONOMICS 47. Washington: Institute for International Economics.
Goldstein, Morris. 1998a. Early Warning Indicators and the Asian Financial Crisis. Unpub-
lished paper. Washington: Institute for International Economics.
Goldstein, Morris. 1998b. The Asian Financial Crisis: Causes, Cures, and Systemic Implications.
POLICY ANALYSES IN INTERNATIONAL ECONOMICS 55. Washington: Institute for Interna-
tional Economics.
Goldstein, Morris. 1998c. Commentary: The Causes and Propagation of Financial Stability:
Lessons for Policymakers. Maintaining Financial Stability in a Global Economy. Kansas
City: Federal Reserve Bank of Kansas City.
Goldstein, Morris. 1998d. The Case for International Banking Standards. In Gerard Caprio,
William Hunter, George Kaufman, and Danny Leipziger, eds., Preventing Banking Crises.
Washington: World Bank.
Goldstein, Morris, and Philip Turner. 1996. Banking Crises in Emerging Economies: Origins
and Policy Options. Bank for International Settlements Economic Paper 46, (October).
Basel: Bank for International Settlements.
REFERENCES 117

Gonzales-Hermosillo, Brenda, Ceyla Pazarbasioglu, and Robert Billings. 1997. Determinants
of Banking Systems Fragility: A Case Study of Mexico. IMF Staff Papers (September):
295-314.
Honohan, Patrick. 1997. Banking System Failures in Developing and Transition Countries: Diagno-
sis and Prediction. Bank for International Settlements Working Paper 39 (January). Basel:
Bank for International Settlements.
International Monetary Fund (IMF). 1997. International Capital Markets. Washington:
International Monetary Fund.
International Monetary Fund (IMF). 1998a. World Economic Outlook. Washington: Inter-
national Monetary Fund.
International Monetary Fund (IMF). 1998b. International Capital Markets. Washington:
International Monetary Fund (IMF). 1998c. World Economic Outlook. Washington: Interna-
tional Monetary Fund.
International Monetary Fund (IMF). 1999. External Evaluation of IMF Surveillance, Report of
a Group of Independent Experts. Washington: International Monetary Fund.
International Monetary Fund. 2000. World Economic Outlook. Washington: International Mon-
etary Fund.
Kamin, Steven B. 1988. Devaluation, External Balance, and Macroeconomic Performance: A Look
at the Numbers. Studies in International Finance, no. 62 (August). Princeton, NJ: Princeton
University, Department of Economics, International Finance Section.
Kamin, Steven B., and John H. Rogers. 1997. Output and the Real Exchange Rate in Developing
Countries: An Application to Mexico. International Finance Discussion Papers 580 (May).
Washington: Board of Governors of the Federal Reserve.
Kaminsky, Graciela L. 1998. Currency and Banking Crises: A Composite Leading Indicator.
International Finance Discussion Paper Series, 629 (December). Washington: Board of
Governors of the Federal Reserve.
Kaminsky, Graciela L., Saul Lizondo, and Carmen M. Reinhart. 1998. Leading Indicators
of Currency Crises. IMF Staff Papers, 45 no. 1 (March): 1-48.
Kaminsky, Graciela L., and Carmen M. Reinhart. 1998. Financial Crises in Asia and Latin
America: Then and Now. American Economic Review 88, no. 2 (May): 444-48.
Kaminsky, Graciela L., and Carmen M. Reinhart. 1999. The Twin Crises: The Causes of
Banking and Balance-of-Payments Problems. American Economic Review 89, no. 3 (June):
473-500.
Kaminsky, Graciela L., and Carmen M. Reinhart. 2000. On Crises, Contagion, and Confusion.
In Journal of International Economics, 51 no. 1 (June): 145-68.
Kiguel, Miguel A., and Nita Ghei. 1992. Devaluation in Low-Inflation Economies. Policy Research
Working Paper 1224. Washington: World Bank.
Kodres, Laura E., and Matthew Pritsker. 1998. A Rational Expectations Model of Financial
Contagion. Mimeo. Washington: Board of Governors of the Federal Reserve.
Krueger, Anne O. 1978. Liberalization Attempts and Consequences. Cambridge, MA: Ballinger.
Krugman, Paul. 1979. A Model of Balance of Payments Crises. Journal of Money, Credit, and
Banking 11 (August): 311-25.
Krugman, Paul. 1998. What Happened to Asia?
Kumar, Manmohan S., and Willian Perraudin. 1998. Assessing Emerging Market Currency
Risk. In Robert Chote, ed., Financial Crises and Asia. London: Centre for Economic
Policy Research.
Kumar, Manmohan S., William Perraudin, and Vincenzo Zinni. 1998. Predicting Emerging
Market Currency Crises. Credit Suisse First Boston Discussion Paper. London: Credit
Suisse First Boston.
Larraı́n, Guillermo, Helmut Reisen, and Julia von Maltzan. 1997. Emerging Market Risk
and Sovereign Credit Ratings. OECD Development Centre Technical Papers 124. Paris:
Organization for Economic Cooperation and Development.

Lee, Suk Hun. 1993. Are the Credit Ratings Assigned by Bankers Based on the Willingness
of Borrowers to Repay? Journal of Development Economics 40 (April): 349-59.
Leiderman, Leonardo, and Alfredo Thorne. 1996. The 1994 Mexican Crisis and Its Aftermath:
What are the Main Lessons? In Guillermo A. Calvo, Morris Goldstein, and Eduard
Hochreiter, eds., Private Capital Flows to Emerging Markets after the Mexican Crisis. Wash-
ington: Institute for International Economics.
Lindgren, Carl-Johan, Gillian Garcia, and Matthew Saal. 1996. Bank Soundness and Macroeco-
nomic Policy. Washington: International Monetary Fund.
Lizondo, Saul, and Peter J. Montiel. 1989. Contractionary Devaluation in Developing Coun-
tries: An Analytical Overview. IMF Staff Papers 36 (March): 182-227.
Masson, Paul. 1998. Contagion: Monsoonal Effects, Spillovers, and Jumps Between Multiple
Equilibria. IMF Working Papers 142 (September). Washington: International Mone-
tary Fund.
Meese, Richard A., and Kenneth Rogoff. 1983. Empirical Exchange Rate Models of the
Seventies: Do They Fit Out of Sample? Journal of International Economics 14 (February):
3-24.
Mishkin, Frederic S. 1996. Understanding Financial Crises: A Developing Country Perspec-
tive. Annual World Bank Conference on Development Economics. Washington: World Bank.
Moody’s. 1998. Moody’s Rating Record in the East Asian Financial Crisis. White paper,
Moody’s Investors Service (May).
Moore, Geoffrey H. 1959. The 1957-58 Business Contractions: New Model or Old? American
Economic Review 49 (May): 292-308.
Morley, Samuel A. 1992. On the Effect of Devaluation during Stabilization Programs in
LDCs. Review of Economics and Statistics 74, no. 1 (February): 21-7.
Obstfeld, Maurice, and Kenneth Rogoff. 1995. The Mirage of Fixed Exchange Rates. Journal
of Economic Perspectives 9 (Fall): 73-96.
Radelet, Steven, and Jeffrey Sachs. 1998. The East Asian Financial Crisis: Diagnosis, Remedies,
Prospects. Brookings Papers on Economic Activity 1: 1-90.
Reinhart, Carmen M. 2000. The Mirage of Floating Exchange Rates. American Economic
Review. Forthcoming.
Rojas-Suarez, Liliana. 1998. Early Warning Indicators of Banking Crises: What Works for
Emerging Markets? With Applications to Latin America. Photocopy.
Rojas-Suarez, Liliana, and Steven R. Weisbrod. 1995. Financial Fragilities in Latin America:
The 1980s and 1990s. International Monetary Fund Occasional Paper 132. Washington:
Rojas-Suarez, Liliana, and Steven R. Weisbrod. 1996. The Do’s and Don’ts of Banking Crisis
Management. In Ricardo Hausman and Liliana Rojas-Suares, eds. Banking Crises in
Latin America. Washington: Inter-American Development Bank and Johns Hopkins
University Press.
Rose, Andrew K., and Lars E. O. Svensson. 1994. European Exchange Rate Credibility before
the Fall. European Economic Review 38 (May): 1185-1216.
Rosenberg, Michael R. 1998. Currency Crises in Emerging Markets: A Guide to Speculative Attack
Models and Early Warning Systems. New York: Merrill Lynch.
Sachs, Jeffrey, Aaron Tornell, and Andres Velasco. 1996. Financial Crises in Emerging Mar-
kets: The Lessons From 1995. Brookings Papers on Economic Activity 1: 147-215.
Sheng, Andrew. 1996. Bank Restructuring: Lessons from the 1980s. Washington: World Bank.
Stock, James H., and Mark W. Watson. 1989. New Indices of Coincident and Leading
Economic Indicators. National Bureau of Economic Research Macroeconomics Annual 1989.
Cambridge, MA: National Bureau of Economic Research.
Sundararajan, V., and Tomas Baliño. 1991. Banking Crises: Cases and Issues. Washington:
Valdés, Rodrigo O. 1997. Emerging Markets Contagion: Evidence and Theory. Working Paper
7 (March). Santiago: Banco Central de Chile.
REFERENCES 119

Wolf, Holger C. 1997. Regional Contagion Effects in Emerging Markets. Princeton University
Working Papers in International Finance, G-97-03.
World Bank. 2000. Global Economic Prospects and the Developing Countries 2000. Washington:
World Bank.
Wyplosz, Charles. 1997. International Capital Market Failures: Sources, Costs, and Solutions.
Unpublished paper, Geneva: Graduate Institute of International Studies.
Wyplosz, Charles. 1998. Speculative Attacks and Capital Mobility. Paper presented at Fourth
Dubrovnik Conference (24-26 June). Dubrovnik, Croatia.
Zhang, Alan. 1999. Testing for ‘Moral Hazard’ in Emerging Market Lending. Institute for Interna-
tional Finance Discussion Paper no. 99-1. Washington: Institute for International Finance.

Index
Ades, Alberto, 14n weighting signals for currency and

Africa, 18 banking crises, 62t
aftermath of crises, 85-93 ASEAN-4 economies
‘‘normal’’ recovery, 106-07 causes of financial crisis, 12
recovery process, 85-89 see also Indonesia; Malaysia;
representative time profile, caveats, Philippines; Thailand
89-93 Asia, 18
annual indicators banking crisis severity, 92, 92t
banking crises, 38t, 60t, 61t, 96-97, 97t currency crisis severity, 92-93, 92t
currency crises, 57t, 59t, 96, 97t exposure to property sector, 43
Argentina out-of-sample application, 67-71
bank cluster, 83 probability of currency crisis for four
banking crisis starting, 24t countries, 69f
borderline signals, 58 real GDP forecasts, 3t, 3
Brazil contagion, 81 short-term debt, 40t
characteristics in contagion episode, Asian financial crisis
82t causes, 12
composite contagion vulnerability characteristics of affected countries, 82t
index, 81t composite contagion vulnerability
contagion financial and trade clusters, index, 79, 81
80t credit rating behavior, 50
crisis vulnerability, alternative currency and maturity mismatches, 40
measures, 63t debt ratings before, 4-5t
currency crisis starting, 22t difficulties in forecasting, 103
probability of contagion in bank ‘‘fear of floating,’’ 13
cluster, 83t interest rate spreads, 6
short-term debt, 40t model predicting countries most
stock returns, 83 affected, 100
tequila effects, 73 onset and flashing signals, 68
121

probability of contagion in bank excluded indicators, 58
cluster, 83t export recovery, 87
rating agencies’ actions on eve and good/false signals, 27
aftermath, 51t identifying end of, 21
ratio of short-term external debt to lasting effects, 86, 107
international reserves, 108 least vulnerable countries, 102
recovery, 92 macroeconomic indicators, 42
spillover effects, 3 microeconomic indicators, 42t, 42
asymmetric-information approach, 14, 16 more difficulty in forecasting, 96
trade in financial assets, 76 output costs, 13n
Australia, 88 predictability, 37
protracted nature, 87
bailouts Quadratic Probability Score (QPS), 67t
beginning of banking crisis indicated, ranking of indicators compared, 36
20 ranking the monthly indicators, 34t
credible guarantees, 56 rankings of country vulnerability,
default versus crisis, 49 101-02, 101t
balance of payments crises, 89 recurring patterns before, 95-96
Baliño, Tomás, 13n, 14n severity compared by region and
bandwagon theory, 14
period, 92t, 92-93
bank deposit withdrawals, 20
signaling window, 27
bank deposits
signals approach, 15
defined, 112
starting dates, 24t, 25t
lead time, 41t
thresholds, 28, 29t
length of recovery, 86t
vulnerability, country rankings, 101t,
optimal thresholds, 29t
101-02
periodicity/transformation used, 26t
vulnerability, predicting, 36
ranking the indicators
weighting the signals in emerging
banking crises, 34t
markets, 62t
currency crises, 35t
banking panics, 20
Bank of International Settlements (BIS),
12n, 78n banking sector
bank lending-deposit interest rate banking-specific information, 42
spread, microeconomic indicators, common lender role, 76
42t economic recovery, 87
bank runs, 14, 37 health of microeconomic indicators,
Venezuela, 87 42-43
bank stocks, 20 high-frequency data absent, 96, 108
banking crises institutional characteristics, 108
annual indicators, 38t problems, 13
best annual indicators, 96-97, 97t beginning of banking crises, lead time,
best monthly indicators, 96, 97t 41t
composite indicator and conditional Berg, Andrew, 104n, 105, 108
probabilities, 66t Billings, Robert, 42
costs of, 1 Blanco, Herminio, 37
currency crash relationships, 2, 2n Bolivia
currency crises crises vis-à-vis, 13, 35t banking crisis starting, 24t
currency crisis following, 2n composite contagion vulnerability
dating onset, 20-21 index, 81t
disadvantages, 2 contagion financial and trade clusters,
downgrades, predicting, 50-52, 52t 80t
early warning indicators, 14 crisis vulnerability, alternative
episodes meeting criteria, 103 measures, 63t
events stressed, 20-21 currency crisis starting, 22t

weighting signals for currency and recovery, 92
banking crises, 62t short-term debt, 40t
borderline indicators, 58, 59t, 60t, 61t vulnerability to banking crises, 101t
Borensztein, Eduardo, 89, 90t vulnerability to currency crises, 99n
Brazil weighting signals for currency and
bailout package, 8 banking crises, 62t
bank cluster, 83 China, 36n
banking crisis starting, 24t Claessens, Stijn, 8n
characteristics in contagion episode, Colombia
82t banking crisis starting, 24t
composite contagion vulnerability banking-crisis predictions, 102
index, 81t, 106 composite contagion vulnerability
contagion, 81t index, 81t
financial and trade clusters, 80t contagion financial and trade clusters,
crisis vulnerability, alternative 80t
measures, 63t crisis vulnerability, alternative
currency crisis starting, 22t measures, 63t
short-term debt, 40t currency crisis starting, 22t
signals and timing of crisis, 56 short-term debt, 40t
weighting signals for currency and weighting signals for currency and
banking crises, 62t banking crises, 62t
common lenders, 76, 83, 100, 106
Calomiris, Charles, 8n
common shocks, 75, 83
Calvo, Guillermo A., 2, 3, 3n, 12n, 13n,
competitive devaluations, 75
36, 38, 40, 75, 76
composite contagion vulnerability index,
Calvo, Sara, 110n
77-84
Cantor, Richard, 47, 89
formula, 78
capital flight, 108
maximum value, 79
capital inflow composition, 39
Mexican peso crisis, 106
Caprio, Gerald, Jr., 1n, 13n, 14n, 25t
three recent crisis episodes, 79-84, 81t
causes of financial crises, patterns,
composite index, 58, 71
finding, 11
composite indicator, 64-71
central bank credit to the public sector/
crisis probability, 64
GDP
annual indicators empirical results, 66-67
banking crises, 38t forecasting accuracy, 67, 67t
currency crises, 39t formula, 62
defined, 113 southeast Asia application, 67-71
optimal thresholds, 29t conditional crisis probabilities
periodicity/transformation used, 26t annual indicators
change in banks’ equity prices, banking crises, 38t
microeconomic indicators, 42t currency crises, 39t
Chile banking crises, 34t
banking crisis starting, 24t compared to economic fundamentals
composite contagion vulnerability banking crises, 46t
index, 81t currency crises, 46t
contagion financial and trade clusters, composite indicator
80t banking crises, 66t
crisis vulnerability, alternative currency crises, 66t
measures, 63t currency crises, 34t
currency crisis starting, 22t estimated, 18
inflation rate, 92n financial and trade clusters, 77t
probability of contagion in bank Conference Board, 41
cluster, 83t confidence, crises in, 89
INDEX 123

contagion, 73-84 unwillingness to expose, 7n
Asian financial crisis, 71 vulnerability ratings, weighting the
composite contagion vulnerability signals, 62t
index, 77-84 critical value. see thresholds
countries sharing financial and trade cross-market hedging, 76, 83
clusters, 80t Cumby, Robert
cross-country, 73, 109 currency crashes, 2, 2n
extending research, 109 currency crises
currency crises, 3 actual and estimated, 98-99
defining, 74-75, 106 banking crises vis-à-vis, 13
empirical studies, 76-77 banking crisis as indicator, 2
Indonesia, 70, 73 best annual indicators, 96, 97t
Mexican peso crisis, 79 best monthly indicators, 96, 97t
regional focus, 106 composite indicator and conditional
shift-contagion, 74 probabilities, 66t
simple ‘‘black box’’ variables, 104 contagion factor, 3
testing of, 18 costs, 1, 3
theories and their implications, 75-76 defined, 19
trade and financial clusters, 77-79 downgrades, predicting, 50-52, 52t
Cooper, Richard, 90t economic growth, 88
Corsetti, Giancarlo, 7, 12n, 75n examples of country-specific
costs of banking crises, 1-2 thresholds, 31t
Council on Foreign Relations, 8n excluded indicators, 58
credit booms export recovery, 87
as causes of financial crises, 12-13 index of currency market turbulence,
defined, 12 19
credit crunch, 87, 107 as leading indictor of financial crises,
credit expansion, 68 13
credit rating (Institutional Investor), less difficulty in forecasting, 96
optimal thresholds, 29t less-lasting effects, 86, 107
credit ratings prediction, compared to banking
Asian financial crisis, 6-7 crises, 37
changes as early warning, 6 Quadratic Probability Score (QPS), 67t
downgrades, predicting, 50-52, 52t ranking the indicators, 35t
inflation, 89 ranking of indicators compared, 36
Institutional Investor index, 27 recurring patterns before, 95-96
creditworthiness severity compared by region and
data on, 7 period, 92t, 92-93
guarantors, 8 signals approach, 15, 57t
crisis periods, 85 signals approach performance
crisis vulnerability strengths, 71
borderline indicators, 58, 59t, 60t, 61t signals approach as promising, 104
composite indicator, 64-71, 98, 99t starting dates, 22t-23t
‘‘core’’ set of countries, 104 thresholds, 28, 29t
crises that showed few signals, 74t vulnerability, country rankings, 99t
cross-country analysis, 56-58, 57t vulnerability, predicting, 36
dynamics of process, 64 weighting the signals in emerging
emerging markets, alternative markets, 62t
measures, 63t, 63-64 currency crisis index, defined, 111
‘‘first screen,’’ 95, 104 currency forecasts, 6
five most vulnerable countries, 98 currency market turbulence, index, 19
‘‘good’’ indicators, 98 current account, balance as share of
ranking of countries, 98, 99t investment, 25

current account balance/GDP Diebold, Francis, 65, 65n
annual indicators domestic credit/GDP
banking crises, 38t defined, 112
currency crises, 39t lead time, 41t
defined, 113 length of recovery, 86t
optimal thresholds, 29t optimal thresholds, 29t
periodicity/transformation used, 26t periodicity/transformation used, 26t
current account balance/investment ranking the indicators
annual indicators banking crises, 34t
banking crises, 38t currency crises, 35t
currency crises, 39t domestic-foreign interest rate differential
optimal thresholds, 29t on deposits
periodicity/transformation used, 26t defined, 113
predicting banking crises, 39 lead time, 41t
Czech Republic Dooley, Michael P., 8n
banking crisis starting, 24t Doukas, John, 74
crisis vulnerability, alternative Dybvig, Phillip, 37
measures, 63t
currency crisis starting, 22t early warning indicators
as extreme case, 64 advantages of, 6
vulnerability to banking crises, 101t banking crises, 14
vulnerability to currency crises, 99n do the indicators flash early enough?,
weighting signals for currency and 40-42
banking crises, 62t Indonesian crisis, 100n
publication and erosion of role, 109-10
de Gregorio, Jose, 89, 90t selected, 26n
debt, developing country versus signals approach reliability, 106
industrial country, 89 subjective assessments versus, 11n
debt ratings suggestions for improving, 107-09
Moody’s, 4-5t themes in literature, 14-15
Standard & Poor’s, 4-5t timing of, 16
default episodes, 49 12-month measures, 26
deflation, 107 Edwards, Sebastian, 90t
Demirgüç-Kunt, Asli, 1n, 13n, 14n, 37, Egypt
102, 103, 108n banking crisis starting, 24t
Denmark crisis vulnerability, alternative
banking crisis starting, 24t measures, 63t
composite contagion vulnerability currency crisis starting, 22t
index, 81t weighting signals for currency and
contagion financial and trade clusters, banking crises, 62t
80t Eichengreen, Barry, 1n, 3, 13n, 14n, 19n,
currency crisis starting, 22t 37, 38, 74, 76, 106, 111
deregulation effects, 87 empirical results, signals approach, 33-43
Detragiache, Enrica, 1n, 13n, 14n, 37, 102, Eschweiler, Bernhard, 6
103, 108n Europe, 18
devaluations, 19-20, 64 excess real M1 balance
competitive, 75 defined, 112
deflation, 107 lead time, 41t
developing world, 89 length of recovery, 86t
earlier episodes, 89 optimal thresholds, 29t
literature review, 90t-91t periodicity/transformation used, 26t
developing countries, 49. see also ranking the indicators
transitional economies banking crises, 34t
Diamond, Douglas, 37 currency crises, 35t
INDEX 125

Exchange Rate Mechanism (ERM) crises, periodicity/transformation used, 26t
3, 6, 88 as poor indicator, 38
exchange rates Frankel, Jeffrey A., 77, 104n, 105
expectations, 53 Furman, Jason, 13n, 103, 104n, 105, 108
explicit pegged, 13
first- and second-generation models, G-7 countries, 36n
15n Galbis, Vicente, 36
fixed, and speculative attacks, 15 Garber, Peter M., 37
index of currency market turbulence, Garcia, Gillian, 1n, 2
19 Gavin, Michael, 13, 13n
managed floating regimes, 13 GDP, new indicators expressed as share
exogenous events, 17 of, 24-25
explicit pegged exchange rate, 13 general government consumption/GDP
exports (in US dollars) annual indicators
defined, 112 banking crises, 38t
lead time, 41t currency crises, 39t
length of recovery, 86t defined, 113
Malaysia, 78t optimal thresholds, 29t
optimal thresholds, 29t periodicity/transformation used, 26t
periodicity/transformation used, 26t Gerlach, Stefan, 75n
ranking the indicators Ghei, Nita, 90t
banking crises, 34t Glaessner, Thomas, 8n
currency crises, 35t Glick, Rueven, 76, 77
Thailand, 78t Goldfajn, Ian, 6, 53
exposure to property sector, by banks, 43 Goldstein, Morris, 2, 2n, 3n, 6, 7, 12n,
external debt composition, 37 13n, 14n, 25t, 38, 43, 51t
Gonazales-Hermosillo, Brenda, 42
‘‘fear of floating,’’ 13 government, explicit and implicit, 15
financial liberalization Granger causality tests, 45
Asian financial crisis, 68 Greece
banking crises vis-à-vis currency banking crisis starting, 24t
crises, 97 crisis vulnerability, alternative
money multiplier, 36 measures, 63t
policy implications, 110 currency crisis starting, 22t
real interest rates, 36 vulnerability to banking crises, 101t
reserves, 36 weighting signals for currency and
financial markets, 52-53 banking crises, 62t
financial and trade clusters, 77t, 77-79 guarantors for loans, 8
Finland
banking crisis starting, 24t Hausman, Ricardo, 13, 13n
composite contagion vulnerability hedging strategies, 76, 83
index, 81t herding behavior, 76, 81
contagion financial and trade clusters, Hong Kong
80t characteristics in contagion episode,
currency crisis starting, 22t 82t
Fischer, Stanley, 89 rating agencies in Asian financial
Flood, Robert, 15n crisis, 51t
Forbes, Kristen, 74 Honohan, Patrick, 2n, 13n, 14n
foreign direct investment (FDI) hyperinflation, 19-20
annual indicators
banking crises, 38t imports (in US dollars)
currency crises, 39t defined, 112
defined, 113 lead time, 41t
optimal thresholds, 29t length of recovery, 86t

optimal thresholds, 29t as early warning, 6
periodicity/transformation used, 26t market leading or following, 45-46
ranking the indicators optimal thresholds, 29t
banking crises, 34t performance as indicator, 97
currency crises, 35t as poor predictors, 49, 53
recovery patterns, 86 why financial crises not anticipated,
index of current market turbulence, 19 7-8
indicators. see leading indicators international financial organizations,
Indonesia crisis vulnerability, exposing, 7n
bailout package, 8 international interest rates, 37, 75, 92
banking crisis starting, 24t, 70n International Monetary Fund (IMF), 2n,
banking crisis vulnerability, 102 3t, 3, 12n, 13n, 14n, 49n, 102, 104n
characteristics in contagion episode, anticipating recent crises, 7
82t bailouts, 49
composite contagion vulnerability investor panic, 100, 100n
index, 79, 81t, 81 Israel
contagion, 70, 73 banking crisis starting, 24t
financial and trade clusters, 80t composite contagion vulnerability
crisis vulnerability, alternative index, 81t
measures, 63t contagion financial and trade clusters,
currency crisis starting, 22t 80t
debt underestimated, 7 crisis vulnerability, alternative
fundamentals, 70-71 measures, 63t
interest rate spreads, 6 currency crisis starting, 22t
model not accounting for, 99-100 weighting signals for currency and
banking crises, 62t
political factors, 73n
probability of contagion in bank
Japan, banking crisis, 20, 88
cluster, 83t
probability of currency crisis, 69f
Kamin, Steven B., 89n, 90t
quality of indicator, 64
Kaminsky, Graciela L., 1n 2, 3, 3n, 13n,
short-term debt, 40t
15n, 16n, 18, 19n, 21, 24, 25t, 27n,
silence of signals, 73
29n, 33, 33n, 34t, 35t, 35n, 39, 46n,
vulnerability to banking crises, 101t 53, 58, 66t, 68n, 69f, 71n, 76, 77t, 77,
weighting signals for currency and 78, 86, 87, 87t, 88t, 92t, 92, 100, 100n,
banking crises, 62t 102, 103, 104n, 105, 107, 108, 110
inflation Kiguel, Miguel A., 90t
Chile, 92n Klingebiel, Daniela, 1n, 13n, 14n, 25t
credit ratings, 89 Kodres, Laura E., 76
Latin America, 88 Krueger, Anne O., 91t
recession severity, 89 Krugman (1979) model, 38, 39
return to precrisis rate, 89 Krugman, Paul, 8n
inherent instability theory, 14 Kumar, Manmohan S., 14n
Institutional Investor
credit ratings Larraı́n, Guillermo, 7, 45, 48, 52
compared to fundamentals, 46t, 47 Latin America, 18
periodicity/transformation used, 26t banking crisis severity, 92, 92t
index described, 27, 27n contagion, 74
interbank debt growth, microeconomic currency crisis severity, 92-93, 92t
indicators, 42t default episodes, 49
interest rate on deposits, microeconomic inflation, 88
indicators, 42t probability of contagion in bank
interest rate spreads cluster, 83t
Asian financial crisis, 6 short-term debt, 40t
INDEX 127

lead time, of leading indicators, 40-42, banking crises, 34t
41t currency crises, 35t
leading indicators M2/reserves
composite indicator formula, 623 lead time, 41t
false alarms, 62, 96 length of recovery, 86t
flashing, 16 optimal thresholds, 29t
flashing early enough, 40-42 policy implications, 110
lead time, 40-42, 96 ranking the indicators
‘‘new,’’ 24-26 banking crises, 34t
original, 21, 24 currency crises, 35t
policy implications, 110 Malaysia
publication and early warning role, banking crisis starting, 24t, 103n
109-10 bilateral and third-party trade, 81
weighting scheme, 16, 16n characteristics in contagion episode,
why or how they affect probability, 17 82t
wide variation in performance, 96 composite contagion vulnerability
leading-indicator analysis index, 79, 81t
business cycles, 18 contagion financial and trade clusters,
potential, 18 80t
Lee, Suk Hun, 47 crisis vulnerability, alternative
Leiderman, Leonardo, 6, 12n measures, 63t
lemons problem, 75 currency crisis starting, 22t
lending-deposit interest rate ratio debt underestimated, 7
lead time, 41t exports (in US dollars), 78t
length of recovery, 86t as extreme case, 64
optimal thresholds, 29t fundamentals, 70
fundamentals, importance, 84
interest rate spreads, 6
banking crises, 34t
probability of contagion in bank
cluster, 83t
liberalization
perils of, 14-15
rating agencies in Asian financial
premature financial, 14-15
crisis, 51t
liberalization period, 87
Lindgren, Carl-Johan, 1n, 2
thresholds, 31t
liquidity, role of, 75
weighting signals for currency and
Lizondo, Saul, 13n, 15n, 19n, 33, 35n, 39,
banking crises, 62t
89, 104n
managed floating regimes, 13, 70n
low-income developing countries, 36n
Lucas critique Marion, Nancy P., 15n
described, 17-18 market expectations, 53
normal behavior, 27n Masih, Rumi, 14n
Masson, Paul, 75
M2 (in US dollars)/reserves (in US Meese, Richard A., 17
dollars) Mendoza, Enrique, 36, 38, 40, 76
defined, 112 methodology, 11-32
periodicity/transformation used, 26t generalized least squares, 50-51
M2 multiplier guidelines, 11-18
defined, 112 indicators, 16, 21-27
lead time, 41t limitations, 17-18
length of recovery, 86t monthly (versus annual) data, 14, 14n
optimal thresholds, 29t noise problems, 105
periodicity/transformation used, 26t out-of-sample tests, 17
ranking the indicators probability of crises, 16

Quadratic Probability Score (QPS), robustness check, 33-38, 103-04
65-66, 67t Montiel, Peter J., 89
ranking, 16 Moody’s
regression techniques, 15, 16 bank financial strength ratings, 49-50n
sample countries listed, 18 foreign currency debt ratings, 4-5t
signaling window, 27 report on rating record, 6n
signals, noise, and crises probabilities, sovereign credit ratings, 26t
31-32 Moody’s Investor services, monthly
thresholds, 28-31 changes, 27
Type I errors, 58 moral hazard play, 8, 8n
Type II errors, 55-56 Morley, Samuel A., 89, 91t
unbalanced panel, 46, 46n multiple equilibrium explanations, of
Mexican peso crisis, 6, 8 banking crises, 37
causes, 11-12 multivariate logit model, 15, 103
characteristics of affected countries, 82t
composite contagion vulnerability NAFTA (North American Free Trade
index, 79, 81t, 106 Agreement), 8
current account balance as indicator, net credit to the public sector/GDP
25 annual indicators
banking crises, 38t
Mexico
bailout package, 8
defined, 113
banking crisis staring, 25t
banking crisis vulnerability, 102
composite contagion vulnerability
net profits to income, microeconomic
index, 81t
indicators, 42t
contagion financial and trade clusters,
noise, assessed as problem, 105
80t
noise-to-signal ratio
crisis vulnerability, alternative
annual indicators
measures, 63t
banking crises, 38t
currency crisis, 3
currency crisis starting, 23t
banking crises, 33, 34t
currency and debt crisis, 2 compared to fundamentals
interest rate policy, 2 banking crises, 46t
market signals as warnings, 6 currency crises, 46t
probability of contagion in bank currency crises, 35t
cluster, 83t defined, 31-32
real exchange rate, 29, 30f, 31 eliminating variables, 58, 62
short-term debt, 40t financial and trade clusters, 77t
tequila effects, 73 flaw of model, 103
thresholds, 29, 31t inverse of, weighting by, 62
vulnerability to banking crises, 101t reading probabilities, 96n
vulnerability to currency crises, 99n ‘‘traditional’’ banking indicators, 42
weighting signals for currency and nonparametric ‘‘signals’’ approach. see
banking crises, 62t signals approach
Middle East, 18 North American Free Trade Agreement
Mishkin, Frederic S., 16n (NAFTA), 8
money multiplier, financial liberalization, Norway
36 banking crisis staring, 25t
monthly data, choice of, 14, 14n composite contagion vulnerability
indicator on a given month defined, 26 index, 81t
monthly indicators contagion financial and trade clusters,
banking crises, 60t, 61t, 96-97, 97t 80t
currency crises, 57t, 59t, 96, 97t currency crisis starting, 23t
INDEX 129

Obstfeld, Maurice, 6 characteristics in contagion episode,
operating costs to assets, microeconomic 82t
indicators, 42t composite contagion vulnerability
origin of financial crises, 11 index, 79, 81t
output contagion financial and trade clusters,
after banking crises, 2 80t
defined, 113 crisis vulnerability, alternative
lead time, 41t measures, 63t
length of recovery, 86t currency crisis starting, 23t
optimal thresholds, 29t as extreme case, 64
periodicity/transformation used, 26t interest rate spreads, 6
ranking the indicators managed float regime, 70n
banking crises, 34t probability of contagion in bank
currency crises, 35t cluster, 83t
recovery patterns, 86 probability of currency crisis, 69f
Thailand, 68 short-term debt, 40t
overall budget deficit/GDP trade exposure, 81
annual indicators vulnerability to currency crises, 99n
banking crises, 38t weighting signals for currency and
currency crises, 39t banking crises, 62t
defined, 113 political triggers, 17, 73n
optimal thresholds, 29t predictability, banking crises, 37
periodicity/transformation used, 26t Pritsker, Matthew, 76
probability of a crisis, conditional on
composite index of fundamentals
Packer, Frank, 47, 89 P(C/F), 83
Pattillo, Catherine, 104n, 105, 108 probability of a crisis, conditional on a
Pazarbasioglu, Ceyla, 42 signal P(C/S), 31
percentage of crises annual indicators
annual indicators banking crises, 38t
banking crises, 38t currency crises, 39t
currency crises, 39t banking crises, 34t
banking crises, 34t composite indicator, 64
compared to fundamentals currency crises, 34t
banking crises, 46t formula for estimating, 65
currency crises, 46t probability of crisis conditioned on
Perraudin, William, 14n fundamentals and a crisis elsewhere
Peru related to a common lender P(C/F,
banking crisis staring, 25t CE), 83-84
composite contagion vulnerability probit regression model, 15, 45
index, 81t banking crises, 47-48, 48t
contagion financial and trade clusters, currency crises, 48-49, 48t
80t ratings’ predictive ability versus, 47
crisis vulnerability, alternative property sector, exposure to, 43
measures, 63t proportion of crises accurately called,
currency crisis starting, 23t formula, 32
vulnerability to banking crises, 101t public sector, governments as
vulnerability to currency crises, 99n guarantors, 8
banking crises, 62t
Quadratic Probability Score (QPS), 65-66,
Pesenti, Paolo, 7, 12n, 75n
67t
Philippines
boom-bust cycle, 68, 70 Radelet, Steven, 6, 12n, 38, 68n, 100n

rate of growth on loans, microeconomic representative time profile, 89, 92
indicators, 42t rescue/bailout packages, expectation of
ratio of lending rate to deposit rate IMF, 8
defined, 112 reserve losses, 19
definition discussed, 21n reserves
ration of M2 (in dollars) to foreign defined, 113
exchange reserves, 36 financial liberalization, 36
real exchange rate lead time, 41t
defined, 21n, 112-13 optimal thresholds, 29t
Indonesia, 100 periodicity/transformation used, 26t
lead time, 41t ranking the indicators
length of recovery, 86t banking crises, 34t
Mexico, 29, 30f, 31 currency crises, 35t
optimal thresholds, 29t Rigobon, Roberto, 74
performance as indicator, 97 risk-weighted capital-to-asset ratio,
periodicity/transformation used, 26t microeconomic indicators, 42t
policy implication, 110 Rogers, John H., 89n
quadratic probability scores, 67t Rogoff, Kenneth, 6, 17
ranking the indicators Rojas-Suarez, Liliana, 13n, 42, 87
banking crises, 34t Rose, Andrew, 1n, 3, 13n, 14n, 19n, 36,
currency crises, 35t 38, 74, 76, 106, 111
real exchange rate index, defined, 21n Rose, Andrew K., 6, 76, 77, 104n, 105
real interest rate Rosenberg, 6
Roubini, Nouriel, 7, 12n, 75n
differential, periodicity/transformation
Rudebusch, Glen, 65, 65n
used, 26t
Russia, 8, 36n
differential, ranking the indicators,
Saal, Matthew, 1n, 2
financial liberalization, 36
Sachs, Jeffrey, 6, 12n, 38, 68n, 100n, 104n,
lead time, 41t
105
length of recovery, 86t Schmukler, Sergio L., 77
optimal thresholds, 29t self-fulfilling runs, 109
postcrisis behavior disparity, 86n shift-contagion, 74
ranking the indicators short-term capital flows
banking crises, 34t defined, 113
currency crises, 35t optimal thresholds, 29t
real interest rate on deposits, defined, thresholds, 28
112 as top indicator of banking crises, 38,
periodicity/transformation used, 26t 108
real interest rate differential short-term capital inflows/GDP
length of recovery, 86t annual indicators
ranking the indicators, banking crises, banking crises, 38t
recapitalization, bank, 2 periodicity/transformation used, 26t
recessions, 89 Thailand, 68
recognition-admission-resolution process, short-term debt
87-88 exposure, 38
recovery, from financial crises, 85-89, 86t ratio to international reserves, 40
Reinhart, Carmen M., 1n, 2, 3, 3n, 13n, selected countries, 40t
15n, 18, 19n, 21, 24, 25t, 27n, 29n, 33, signal
33n, 35t, 35n, 39, 46n, 53, 68n, 71n, borderline, 58, 59t, 60t, 61t
76, 77t, 77, 78, 86, 87, 87t, 88t, 92t, good/false, 27
92, 100, 102, 104n, 105, 107, 110, 110n indicator behavior and, 27
Reisen, Helmut, 7, 45, 48, 52 Mexico, 29, 30f
INDEX 131

signaling window, 27 weighting signals for currency and
silence of, 73 banking crises, 62t
weight individual indicators, 58, 62 sovereign credit ratings, 6-7, 25-26
signaling window, 27 economic fundamentals as
Mexico, 29, 30f determinants, 47
signals, crisis probability, 65 as leading indicators, challenged, 50
signals approach performance as indicator versus
basic premise, 18-19 fundamentals, 97
countries in sample, 18 as poor predictors, 49
critics of, 104-05 predictive ability evaluated, 45-49
disadvantages, 16n why financial crises not anticipated,
empirical results, 33-43 7-8
future questions, 108 Spain
methodological guidelines, 15-16 banking crisis staring, 25t
noise and crises probabilities, 31-32 composite contagion vulnerability
out-of-sample exercises, 55-71 index, 81t
out-of-sample usefulness, 36 contagion financial and trade clusters,
performance strengths, 71 80t
as promising model, 104 currency crisis starting, 23t
timing of early warnings, 16 speculative attacks, on interest rates
Smets, Frank, 75n currency crisis defined, 19
South Africa models of vulnerability, 15, 15n
banking crisis staring, 25t spillovers, 3. see also contagion
borderline signals, 58 Standard & Poor’s long-term debt
ratings, 4-5t
Stiglitz, Joseph E., 13n, 103, 104n, 105,
measures, 63t
108
currency crisis starting, 23t
stock prices, recovery patterns, 86
vulnerability to banking crises, 101t
stock prices (in dollars)
vulnerability to currency crises, 99t
defined, 113
lead time, 41t
banking crises, 62t
length of recovery, 86t
South Korea
bailout package, 8
performance as indicator, 97
borderline signals, 58
characteristics in contagion episode, banking crises, 34t
composite contagion vulnerability stock returns, 76
index, 81t Argentina, 83
contagion, 81 Sundararajan, V., 13n, 14n
financial and trade clusters, 80t Svensson, Lars E.O., 6
crisis vulnerability, alternative Sweden
measures, 63t banking crisis staring, 25t
currency crisis starting, 23t composite contagion vulnerability
debt underestimated, 7 index, 81t
as extreme case, 64 contagion financial and trade clusters,
official reserves, 7 80t
rating agencies in Asian financial currency crisis starting, 23t
crisis, 51t thresholds, 31t
trade exposure, 81 takeovers by public sector, banking, 20
vulnerability to banking crises, 101t Tenegauzer, Daniel, 14n
vulnerability to currency crises, 99n terms of trade

defined, 112 trade and financial clusters, 77t, 77-79
lead time, 41t trade issue, 76
length of recovery, 86t financial assets, 76
optimal thresholds, 29t trade issues, composite contagion
periodicity/transformation used, 26t vulnerability index, 78
predictive value, 47 tranquil periods, 85, 107n
ranking the indicators transitional economies
banking crises, 34t credibility problems, 89
currency crises, 35t credit ratings as predictors, 49-50, 50t
tesobonos, 6 excluding from sample, 36n
Thailand weighting signals for currency and
bailout package, 8 banking crises, 62t
banking crisis starting, 25t Turkey
banking exposure before crisis, 83 banking crisis staring, 25t
bilateral and third-party trade, 81 composite contagion vulnerability
composite contagion vulnerability index, 81t
index, 79, 81t contagion financial and trade clusters,
contagion 80t
financial and trade clusters, 80t crisis vulnerability, alternative
to Philippines, 84 measures, 63t
credit rating behavior, 50 currency crisis starting, 23t
crisis vulnerability, alternative weighting signals for currency and
measures, 63t banking crises, 62t
currency crisis starting, 23t Turner, Philip, 2, 13n, 14n
debt underestimated, 7 twin crises, 86, 107
emerging markets compared, 83, 83t
Ukraine, 8
exports, 78t
rescue/bailout, 8
fundamentals, 68
unconditional probability of crisis
IMF anticipation of, 7
banking crises, 34t
interest rate spreads, 6
official reserves, 7
formula, 31
United Kingdom, 88
quality of indicator, 64
United States
rating agencies in Asian financial Conference Board composite indices,
crisis, 51t 41
short-term capital inflows, 68 costs of savings and loan crisis, 2n
short-term debt, 40t leading-sector composite index, 41
spillover effects, 3 Mexican peso crisis, 8
vulnerability to currency crises, 99n Uruguay
weighting signals for currency and banking crisis staring, 25t
banking crises, 62t composite contagion vulnerability
Thorne, Alfredo, 6, 12n index, 81t
thresholds, 28-31 contagion financial and trade clusters,
for composite indicator, 64 80t
optimal, 28, 29t crisis vulnerability, alternative
proportion of monthly indicators, 98 measures, 63t
short-term capital flows, 28 currency crisis starting, 23t
signal-to-noise ratio, 28 vulnerability to banking crises, 101t
Tille, Cedric, 75n vulnerability to currency crises, 99n
time-series probabilities of crisis, 65, weighting signals for currency and
67-71, 100n banking crises, 62t
timing, of financial crises, 85-89
Tornell, Aaron, 104n, 105 Valdés, Rodrigo O., 6, 53, 74
INDEX 133

Végh, 110n vulnerability to currency crises, 99n
Velasco, Andres, 104n, 105 weighting signals for currency and
Venezuela banking crises, 62t
bank runs, 87 volatility patterns, 29, 29n
banking crisis staring, 25t von Maltzan, Julia, 7, 45, 48, 52
composite contagion vulnerability
Watson, Mark W.
index, 81t
Weisbrod, Steven R., 13n, 87
contagion financial and trade clusters,
Wolf, Holger C., 76
80t
Wyplosz, Charles, 3, 19n, 74, 76, 104n,
105, 106, 111
measures, 63t
currency crisis starting, 23t Zhang, Alan, 8n
vulnerability to banking crises, 101t Zinni, Vincent, 14n

Reinhart, Carmen M. - Goldstein, Morris Arthur - Forecasting Financial Crises - Early Warning Signals For Emerging Markets

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

Reinhart, Carmen M. - Goldstein, Morris Arthur - Forecasting Financial Crises - Early Warning Signals For Emerging Markets

Uploaded by

Copyright:

Available Formats

ASSESSING

INSTITUTE FOR INTERNATIONAL ECONOMICS

Institute for International Economics | http://www.iie.com

4 Rating the Rating Agencies 45

5 An Assessment of Vulnerability: Out-of-Sample Results 55

Institute for International Economics | http://www.iie.com

7 The Aftermath of Crises 85

8 Summary of Results and Concluding Remarks 95

Appendix A: Data and Definitions 111

Institute for International Economics | http://www.iie.com

Institute for International Economics | http://www.iie.com

Institute for International Economics | http://www.iie.com

Institute for International Economics | http://www.iie.com

Institute for International Economics | http://www.iie.com

C. Fred Bergsten, Director

Institute for International Economics | http://www.iie.com

Institute for International Economics | http://www.iie.com

Institute for International Economics | http://www.iie.com

2 ASSESSING FINANCIAL VULNERABILITY

Institute for International Economics | http://www.iie.com

Source: International Monetary Fund, World Economic Outlook.

Although the contagion of financial disturbances usually runs from

Institute for International Economics | http://www.iie.com

Mexico Ba2 Ba2 Ba2 Ba2

The Philippines Ba2 Ba2 Ba2 Ba2

Standard & Poor’s October 1997

Foreign currency debt BB positive BB positive BBⳭ positive BBⳭ stable

Source: Radelet and Sachs (1998).

6 ASSESSING FINANCIAL VULNERABILITY

Institute for International Economics | http://www.iie.com

We found that the Fund—in both bilateral and multilateral surveillance—largely

There are, of course, several reasons interest rate spreads or changes

Institute for International Economics | http://www.iie.com

8 ASSESSING FINANCIAL VULNERABILITY

Institute for International Economics | http://www.iie.com

Institute for International Economics | http://www.iie.com

Our approach to identifying early warning indicators of financial crises in

Institute for International Economics | http://www.iie.com

12 ASSESSING FINANCIAL VULNERABILITY

Institute for International Economics | http://www.iie.com

Institute for International Economics | http://www.iie.com

䡲 asymmetric information and ‘‘bank run’’ stories that stress liquidity/

14 ASSESSING FINANCIAL VULNERABILITY

Institute for International Economics | http://www.iie.com

In operational terms, this eclectic view of the origins of financial crises

Institute for International Economics | http://www.iie.com

16 ASSESSING FINANCIAL VULNERABILITY

Institute for International Economics | http://www.iie.com

Institute for International Economics | http://www.iie.com

Putting the Signals Approach to Work

䡲 Africa: South Africa

18 ASSESSING FINANCIAL VULNERABILITY

Institute for International Economics | http://www.iie.com

A currency crisis is defined as a situation in which an attack on the

I ⳱ (⌬e/e) ⳮ (␴e /␴R) * (⌬R/R) (2.1)

where ␴e is the standard deviation of the rate of change of the exchange

Institute for International Economics | http://www.iie.com

20 ASSESSING FINANCIAL VULNERABILITY

Institute for International Economics | http://www.iie.com

In addition to the 15 early warning indicators originally considered in

Institute for International Economics | http://www.iie.com

22 ASSESSING FINANCIAL VULNERABILITY

Institute for International Economics | http://www.iie.com