No. 51
Marek D¹browski (ed.)
Currency Crises
in Emerging-Market Economies:
Causes, Consequences
and Policy Lessons
with contribution from Ma³gorzata Antczak, Rafa³ Antczak,
Monika B³aszkiewicz, Georgy Ganev, Ma³gorzata Jakubiak, Ma³gorzata
Markiewicz, Wojciech Paczyñski, Artur Radziwi³³, £ukasz Rawdanowicz,
Marcin Sasin, Joanna Siwiñska, Mateusz Szczurek,
and Magdalena Tomczyñska
and editorial support of Wojciech Paczyñski
W a r s a w ,
2 0 0 2
Summary Report from the research project on "Analysis of
Causes and Course of Currency Crises in Asian, Latin
American and CEE Countries: Lessons for Poland and
Other Transition Countries", grant No. 0144/H02/99/17 of
the State Committee for Scientific Research (KBN).
The publication was financed by Rabobank Polska S.A.
Key words: currency crisis, financial crisis, contagion, emerging markets, transition economies, exchange rates, monetary policy, fiscal policy, balance of payments, debt, devaluation.
DTP: CeDeWu Sp. z o.o.
Graphic Design – Agnieszka Natalia Bury
Warsaw 2002
All rights reserved. No part of this publication may be
reproduced, stored in a retrieval system, or transmitted in any
form or by any means, without prior permission in writing
from the author and the European Commission.
ISSN 1506-1647 ISBN 83-7178-285-3
Publisher:
CASE – Center for Social and Economic Research
ul. Sienkiewicza 12, 00-944 Warsaw, Poland
e-mail: case@case.com.pl
http://www.case.com.pl
Currency Crises in Emerging-Market Economies ...
Contents
Abstract
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9
2. Definition of Currency Crisis and its Empirical Exemplification . . . . . . . . . . . . . . . . . . . . . . . . . . .12
3. Theoretical Models of Currency Crises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17
3.1. First-generation Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17
3.2. Modified First-generation Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18
3.3. Second-generation Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19
3.4. Third-generation Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20
4. Early Warning Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21
5. Causes of Currency Crises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25
5.1. Fiscal Imbalances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25
5.2. Current Account Deficit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .27
5.2.1. Evolution of the Theoretical Views . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .27
5.2.2. Results of Empirical Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .28
5.3. Currency Overvaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .29
5.3.1. How to Measure a Real Exchange Rate Overvaluation? . . . . . . . . . . . . . . . . . . . . . . . . . . .30
5.3.2. Results of Empirical Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .30
5.4. The Role of Exchange Rate Regimes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32
5.5. Structural Weaknesses of the Banking and Corporate Sectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . .33
5.5.1. Relationship Between Banking and Currency Crises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .33
5.5.2. The Most Frequent Weaknesses of the Banking Sector . . . . . . . . . . . . . . . . . . . . . . . . . . .34
5.5.3. Review of Empirical Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .34
5.6. Political Instability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35
6. Crisis Management – How to Defend an Exchange Rate if at All? . . . . . . . . . . . . . . . . . . . . . . . . .37
7. Contagion Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .39
7.1. Definition Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .39
7.2. The Channels of Crises Propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .39
7.3. Empirical Investigation of Contagion Effect in CIS Countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .40
8. Economic and Social Consequences of Currency Crises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .43
9. Crisis Prevention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .48
9.1. The Role of International Liquidity in Preventing Currency Crises . . . . . . . . . . . . . . . . . . . . . . . . . .48
9.2 The Role of the IMF in Preventing Currency Crises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .50
10. Research Conclusions and Policy Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .54
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .56
CASE Reports No. 51
3
M. D¹browski (ed.)
List of figures
Figure 2.1. Typology of financial crises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12
Figure 2.2. Severity of the currency crises measured by reserve losses and nominal depreciation
of the domestic currency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14
Figure 2.3. Real interest rate differentials around crisis dates, monthly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14
Figure 2.4. Nominal and real interest rates before and after the crisis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15
Figure 2.5. Broad money monetization in the crisis countries, 1992–1998 . . . . . . . . . . . . . . . . . . . . . . . . . . . .16
Figure 4.1. Early warnings indicator (statistically significant) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21
Figure 4.2. Behavior of selected warning indicators (stylized facts) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .22
Figure 4.3. Reserves over short-term debt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23
Figure 4.4. Ratio of the official international reserves to reserve money . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23
Figure 4.5. Real effective exchange rates in crisis countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .24
Figure 4.6. Real exchange rates in relation to US dollar, CPI based . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .24
Figure 5.1. Links between fiscal variables and currency crisis – review of empirical research . . . . . . . . . . . . . .25
Figure 5.2. The value of fiscal indicators in predicting a crisis in selected FSU economies . . . . . . . . . . . . . . . . .26
Figure 5.3. Summary of the empirical researches showing the role of a current account deficit in predicting
currency crises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .29
Figure 5.4. Summary of the empirical researches showing the role of real exchange rate overvaluation
in predicting currency crises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31
Figure 5.5. Performance of bank crises as a signal of currency crises and vice versa . . . . . . . . . . . . . . . . . . . . .35
Figure 6.1. Declared and actual changes in the exchange rate regime during the crisis . . . . . . . . . . . . . . . . . . .38
Figure 7.1. Contagion crisis probability in transition countries basing on trade links – results of the probit
model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .41
Figure 7.2. Trade matrix of the crisis-affected countries in 1997 (% of total exports) . . . . . . . . . . . . . . . . . . . .41
Figure 7.3. Contagion crisis probability in transition countries: comparison of results of the probit
and balance of payments models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .42
Figure 8.1. Real GDP growth rate before and after the crisis (median) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .44
Figure 8.2. Costs of crises in terms of lost output relative to trend . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .45
Figure 8.3. Net capital inflow before and after the crisis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .45
Figure 8.4. CPI, y-o-y percentage change before and after the crisis; monthly data (medians for subgroups) . .46
Figure 8.5. Unemployment rate before and after the crisis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .47
Figure 9.1. Evaluating optimal international liquidity – marginal cost and benefit . . . . . . . . . . . . . . . . . . . . . . . .49
Figure 9.2. Optimal liquidity holding versus cost of the crisis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .50
Figure 9.3. Crisis cost to the policy maker, as of end-99 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .50
Figure 9.4. Compliance with the IMF quantitative performance criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .51
Figure 9.5. Compliance with IMF structural conditionality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .52
4
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
Marek D¹browski
Professor of Economics
Chairman and one of the founders of the CASE – Center for Social and Economic Research in Warsaw
From 1991 involved in policy advising for governments and central banks of Russia, Ukraine, Kyrgyzstan, Kazakhstan,
Georgia, Uzbekistan, Mongolia, and Romania; 1989–1990 First Deputy Minister of Finance of Poland; 1991–1993 Member
of the Sejm (lower house of the Polish Parliament); 1991–1996 Chairman of the Council of Ownership Changes, the advisory body to the Prime Minister of Poland; 1994–1995 visiting consultant of the World Bank, Policy Research Department;
from 1998 Member of the Monetary Policy Council of the National Bank of Poland. Recently his area of research interest
is concentrated on macroeconomic policy problems and political economy of transition.
CASE Reports No. 51
5
Currency Crises in Emerging-Market Economies ...
Abstract
Currency crises have been recorded for a few hundreds
years but their frequency increased in the second half of the
20th century along with a rapid expansion of a number of
fiat currencies. Increased integration and sophistication of
financial markets brought new forms and more global character of the crises episodes.
Eichengreen, Rose and Wyplosz (1994) propose the
operational definition, which helps to select the episodes
most closely fitting the intuitive understanding of a currency
crisis (a sudden decline in confidence towards a specific currency). Among fundamental causes of currency crises one
can point to the excessive expansion and over-borrowing of
the public and private sectors, and inconsistent and nontransparent economic policies. Over-expansion and overborrowing manifest themselves in an excessive current
account deficit, currency overvaluation, increasing debt burden, insufficient international reserves, and deterioration of
other frequently analyzed indicators. Inconsistent policies
(including the so-called "intermediate" exchange rate
regimes) increase market uncertainty and can trigger speculative attack against the domestic currency. After a crisis has
already happened, the ability to manage economic policies
in a consistent and credible way becomes crucial for limiting
the crisis' scope, duration and negative consequences.
Among the dilemmas that the authorities face in such circumstances is the decision on readjustment of an exchange
rate regime, as the previous regime is usually the first institutional victim of any successful speculative attack.
The consequences of currency crises are usually severe
and typically involve output and employment losses, fall in
CASE Reports No. 51
real incomes of a population, deep contraction in investment and capital flight. Also the credibility of domestic economic policies is ruined. In some cases a crisis can serve as
the economic and political catharsis: devaluation helps to
temporarily restore competitiveness and improve a current
account position, the crisis shock brings the new, reformoriented government, and politicians may draw some
lessons for future.
The responsible macroeconomic policy can help to
diminish a risk of an occurrence of a currency crisis. It
involves balanced and transparent fiscal accounts, proper
monetary-fiscal policy mix, and low inflation, avoiding
indexation of nominal variables and intermediate monetary/exchange rate regimes. On the microeconomic level
key elements include privatization, demonopolization and
introduction of efficient competition policy, prudential
re-gulation of the financial sector, trade openness, and
simple, fair and transparent tax system. All the above
should help elimination of soft budget constraints, overborrowing on the side of both private and public sector
and moral hazard problems. All these measures need to
be strengthened by legal reforms, efficient and fair judiciary system, implementation of international accounting,
reporting and disclosure standards, transparent corporate and public governance rules, and many other elements. Reforms can be supported by the IMF and other
international organizations, which on their part should
depoliticize their actions and decision-making processes,
sticking to the professional criteria of country assessment
and their consequent execution.
7
Currency Crises in Emerging-Market Economies ...
1. Introduction
Financial crises cannot be considered new phenomena
of the last decade or even of the entire 20th century. Older
history registers many episodes of government defaults and
bank runs. However, currency crashes – one of the forms of
financial instability – were not so frequent until the middle
of the last century1. The reason for this was very simple.
The world monetary system was dominated by a number of
strong currencies based on the gold standard and their
satellites (currency boards in colonies) and consequently
only the extreme events such as the World War I and Great
Depression could temporarily damage this system. However, with abandoning the direct gold standard after the Great
Depression and the World War II stability of individual currencies became dependent on national economic policies,
and, therefore, more vulnerable to market speculation.
Although the Bretton Woods system introduced after the
World War II tried to return to an indirect gold standard
(parities of individual currencies were set in relation to the
US dollar and the latter was officially backed by the US gold
reserves) and currency stability (through fixed but
adjustable pegs) it contained the fundamental inconsistency
leading to its final crash in the beginning of 1970s. This
inconsistency originated from an attempt to follow the socalled impossible trinity (see Frankel, 1999), i.e. exchange
rate stability, monetary independence, and financial market
integration. As long as the importance of the last element
was limited (convertibility of most of currencies in the afterwar period was very limited) the Bretton Woods system
could sustain, avoiding major disturbances. However,
progress in free capital movements and very expansionary
fiscal and monetary policies in the US during the Vietnam
War brought the definite collapse of this system in 1971.
As a result, the last decades of the 20th century were
dominated by fiat money individually managed by each
country, that were not always following the price stability
1
goal. Additionally, the number of independent countries
(and monetary authorities) rapidly increased, first resulting
from decolonization processes in Africa, Asia, and Latin
America and later due to the collapse of the Soviet Union,
Yugoslavia and the whole Soviet block. An increasing number of countries opted to liberalize capital accounts and at
the same time progressing technological revolution made
cross-border capital transaction much easier than a few
decades earlier. All these events increased the global integration of financial markets on the one hand, but also a
possibility to speculate against exchange rates of individual currencies and, thus, the frequency of currency crises,
on the other.
The decade of 1990s brought a new experience in this
field. While earlier currency crises were caused mainly by
the evident domestic macroeconomic mismanagement
(that gave theoreticians an empirical ground for a construction of the so-called first generation models of currency
crises), during the last decade crises also hit economies
widely regarded as following solid policies and having a good
reputation. This new experience started with the 1992 ERM
crisis when the British pound and the Italian lira were forced
to be devalued. This was particularly surprising in the case
of the UK that successfully went through a series of very
ambitious economic reforms in the 1980s.
At the end of 1994, the serious currency crisis hit Mexico, and during the next few months it spread to other Latin
American countries, particularly to Argentina (the so-called
Tequila effect). Although Argentina managed to defend its
currency board, the sudden capital outflow and banking crisis caused a one-year recession. Currency crises were not
the new phenomena in the Western Hemisphere where
many Latin American countries served as the textbook
examples of populist policies and economic mismanagement through several decades. However, the two main vic-
Bordo and Schwartz (1998) provide an interesting overview of 19th and 20th century financial and currency crises.
CASE Reports No. 51
9
M. D¹browski (ed.)
tims of the "Tequila" crisis – Mexico and Argentina – had
been widely regarded as examples of successful reforms and
thus the experienced turbulence seemed to be unjustified,
at least at first sight.
Two years later an even more unexpected and surprising
series of financial crises hit South East Asia. The Asian Tigers
had for many years enjoyed a reputation of fast growing,
macroeconomically balanced and highly competitive
economies, which managed to make a great leap forward
from the category of low-income developing countries to
middle or even higher-middle income group virtually during
the life of just one generation. However, a more careful analysis could easily specify several systemic weaknesses, particularly related to the financial and corporate sectors. Additionally, as in the case of Mexico the crisis management in its early
stage was not especially successful and only provoked further
devaluation pressure and a financial market panic.
The external consequences of the Asian crisis became
much more serious than in the case of Mexico. While the
latter had a regional character only, the former affected the
whole global economy spreading to other continents. The
Asian crisis started in Thailand in July 1997 and its first round
of contagion hit Malaysia, Indonesia and the Philippines in
summer 1997. The next wave caused serious turbulence in
Hong Kong, the Republic of Korea, and again in Indonesia in
the fall of 1997 and beginning of 1998. Singapore and Taiwan
were affected to lesser extent. Development in Asia undermined investors’ confidence in other emerging markets,
particularly in Russia and Ukraine, both of which were characterized by chronic fiscal imbalances. Both countries, after
resisting several speculative attacks against their currencies
at the end of 1997 and the first half of 1998, finally entered
the full-scale financial crisis in August – September 1998.
Following Russia and Ukraine, also other post-Soviet
economies (Moldova, Georgia, Belarus, Kyrgyzstan, Uzbekistan, Kazakhstan, and Tajikistan) experienced forced devaluations and debt crisis. Finally, Russian developments triggered eruption of a currency crisis in Brazil in early 1999,
and some negative contagion effects for other Latin American economies, particularly for Argentina.
In the meantime, cumulated negative consequences of
the Asian and Russian crises damaged confidence not only in
relation to the so-called emerging markets but also affected
the financial markets of developed countries. In the last
quarter of 1998, the danger of the US and worldwide recession pushed the Federal Reserve Board to ease significantly
the US monetary policy. However, some symptoms of the
global slowdown such as a substantial drop in prices of oil
and other basic commodities could not be avoided.
The new crisis episodes stimulated both theoretical discussion and a large body of empirical analyzes trying to identify the causes of currency crises and their rapid propagation, their economic and social consequences, methods of
their preventing and effective management after a crisis had
happened. On the theoretical ground, the new experience
brought the so-called second- and third-generation models
of currency crises. Both theoretical and empirical discussions started to pay more attention on the role of market
expectations and multiple equilibria as well as global integration of financial markets and their functioning.
While most of recent empirical studies available in economic literature concentrated on Latin America and Asia,
the main objective of this paper2 is to expand this analysis to
the countries of Central and Eastern Europe and the former
Soviet Union. This additional empirical input allowed for
reexamination of the existing theoretical models and accumulated empirical observations and verification of policy
conclusions and recommendations proposed by other
authors.
This paper contains an overview of the main findings and
conclusions of the several studies related both to the concrete currency crisis episodes3 and selected aspects of these
crises from the comparative perspective4. In section 2 we
try to clarify the definition of a currency crisis (comparing to
other forms of financial crises) and make it operational. In
section 3 we present the evolution of theoretical models of
currency crises during the last two decades. Section 4 gives
an overview of the symptoms informing about danger of a
crisis occurrence, i.e. the so-called early warning signals
(indicators). Based on this, section 5 gives a more in-depth
2
This paper summarizes main findings of the research project on "Analysis of Causes and Course of Currency Crises in Asian, Latin American and
CEE Countries: Lessons for Poland and Other Transition Countries", grant No. 0144/H02/99/17 of the Scientific Research Committee (KBN) carried
out by the CASE under direction of Marek D¹browski, from October 1, 1999 to September 30, 2001.
3 Eleven country monographs covered two countries of Latin America (Mexico 1994–1995 and Argentina 1995), four Asian countries (Thailand
1997, Malaysia 1997–1998, Indonesia 1997–1998, and the Republic of Korea 1997–1998), and five emerging market economies in Europe (Bulgaria
1996–1997, Russia 1998, Ukraine 1998, Moldova 1998, and Turkey 2000).
4 Thirteen comparative studies dealt with definitions, theoretical models and causes of currency crises, crisis management and propagation (contagion effect), economic, social and policy consequences of the crises, and crisis prevention.
10
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
analysis of some fundamental causes of currency crises: fiscal imbalances, current account deficits, currency overvaluation, the performance of specific exchange rate regimes,
structural weaknesses in the banking and corporate sectors,
and political instability. In section 6 we discuss crisis management, concentrating on changes in the exchange rate
regime in a crisis period. Section 7 is devoted to crisis propagation, concentrating on the empirical example of the CIS
in 1998–1999. In section 8 we analyze the economic, social
CASE Reports No. 51
and policy consequences of currency crises. In particular,
we try to assess to what extent crisis shocks can contribute to an improvement of macroeconomic fundamentals. Section 9 deals with crisis prevention. In this context
we discuss two particular issues: the optimal level of central bank’s international reserves and the role of the IMF in
preventing crises, using the empirical case of five CIS countries. Section 10 contains general conclusion and policy
recommendations.
11
M. D¹browski (ed.)
2. Definition of Currency Crisis and its Empirical
Exemplification
The existing terminology related to currency crises is
not precise and can create certain confusion. This relates
not only to popular policy discussions but also to analytical
and theoretical works of an academic character.
Any clarification attempt should involve two stages of
discussion. First, we need to distinguish currency crisis from
a more general category of financial crisis, and some other
similar notions such as balance of payments crisis. The second stage involves building a clear operational definition of a
currency crisis, which can be used in empirical research.
The notion of a financial crisis seems to be the broadest,
involving all kinds of instability related to monetary and
financial systems. Generally speaking, this is a sudden
decline in confidence in relation to government/central bank
and banking sector ability to respect their liabilities (on the
committed terms).
Historically, financial crises were defined in two ways.
The narrow definition, associated with the monetarist
school linked financial crises with banking panics (Friedman
and Schwartz, 1963). However, this definition seems too
narrow from the point of view of the existing spectrum of
episodes of financial instability. What Friedman and
Schwartz have in mind could be better called as systemic
banking crises understood as an inability of commercial
banks to respect their liabilities (see below).
The second, very general definition of systemic financial
crisis outlined by Minsky (1972) and Kindleberger (1978)
involved broad categories of crises, including sharp declines
in asset prices, failures of financial and nonfinancial institutions, deflations or disinflations, disruptions in foreign
exchange markets, or some combinations of the above. In
this approach the root cause of financial instability lies in the
breakdown of information flows hindering the efficient functioning of financial markets (Antczak, 2000). Contrary to the
5
12
Friedman-Schwartz one, the Minsky-Kindleberger definition
seems too general, and, therefore, not very useful in practice. In fact, it covers the very broad set of various macroand microeconomic disturbances, including the regular
cyclical developments.
Being the broadest among the discussed categories (see
Figure 2.1), financial crises cover other narrower definitions (WEO, 1998; pp. 111–112; Antczak, 2000). As it was
mentioned earlier, banking crisis refers to the actual or
potential bank runs or failures that induce banks to suspend
the internal convertibility of their liabilities or which compels
the monetary authorities to intervene to prevent this by
extending assistance on a large scale. The public debt crisis is
a situation in which a government cannot service its foreign
and/or domestic obligations. The balance of payment crisis is
a structural misbalance between a deficit in a current
account (absorption) and capital and financial account
(sources of financing) that after exhausting international
reserves leads to a currency crisis. The balance of payment
crisis was a synonym of a currency crisis in the light of the
first-generation theoretical model (see below).
Figure 2.1. Typology of financial crises
Banking crisis
Financial Public debt crisis
crisis
Balance of payments crisis ⇒ Currency crisis
Source: Based on Antczak (2000).
Finally, a currency crisis can be understood as a sudden
decline in the confidence to an individual currency5 usually
leading to a speculative attack against it.
All the above-mentioned specific forms of a financial crisis are very often inter-related. A systemic banking crisis can
trigger a speculative attack against national currency as it happened, for example, in Bulgaria in 1996 (Ganev, 2001) and in
I am grateful to Witold M. Or³owski for suggesting this general definition of currency crisis.
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
Turkey in 2000–2001 (Sasin, 2001d). The opposite sequence
of events is also very often in place (Argentine 1995, Russia
1998). The same concerns interrelations between a public
debt crisis and a currency crisis. The former may provoke
massive capital outflow and consequently a balance of payments/currency crisis (Russia and CIS 1997–1998). On the
other hand, substantial devaluation/depreciation of the
domestic currency leads to an increase in the domestic value
of the foreign exchange denominated debt and can dramatically worsen the debt to GDP ratio. Thus, one specific form
of a crisis may (although does not need) lead to its development into a full-scale financial crisis involving the entire financial market. We will discuss these interlinks in more detail
in the subsequent sections of this paper (especially sections 5.1. and 5.5).
Having discussed general terminological issues we will
now try to make a definition of a currency crisis more precise and operational6. One possibility is to make a pure
expert assessment as to whether any particular country
experienced a currency crisis as suggested by Glick and
Rose (1999). However, such an approach bears risk of being
too arbitrary.
In a popular and intuitive understanding, a currency crisis should result in a substantial devaluation/depreciation of
the domestic currency. However, the question arises as to
how substantial should the change in a nominal exchange
rate be and over how long period should this be measured.
In particular, answers to the above questions may prove difficult in the case of floating (flexible) exchange rate regimes
when changes in nominal rate are expected, by definition,
to accommodate fluctuations in demand for domestic
money that is unstable at least in short term.
If a country follows a fixed peg and market pressure
forces government and monetary authorities to abandon or
change this peg, situation is clearer7. However, recently less
and less countries follows this kind of rigid exchange rate
regimes (apart from countries that introduced currency
boards or joined the monetary union). Consequently
researchers need to find some kind of quantitative criteria
allowing for identification of specific cases of currency crises.
Frankel and Rose (1996) propose one such criterion,
defining a "currency crash" as a nominal depreciation of a
currency of at least 25 percent in a year, provided that
this represents a 10 percentage points increase in the rate
of depreciation from the previous year.
However, speculative attacks against a currency are not
always successful, particularly in cases of currency board
arrangements (examples of Argentina in 1995 and Hong
Kong in 1997). If nothing changed in the behavior of the
nominal exchange rate one could eventually claim that these
countries avoided crisis. However, defending the exchange
rate can be very costly in terms of decrease in official
reserves (and domestic money stock) and temporarily higher interest rates with the resulting negative consequences
for output and employment. Hence, it would be unjustified
to eliminate such cases from the analysis of currency crises.
The above leads to a multi-factor definition of currency crisis that was formulated, among others, by Rodrik
and Velasco (1999), Bussiére and Mulder (1999), Tanner
(1999), Aziz, Caramazza and Salgado (2000), Goldfajn and
Valdés (1997) and van Rijckeghem and Weder (1999).
Eichengreen, Rose, and Wyplosz (1994) proposed the
index of exchange market pressure (EMP) being the weighted average of the changes in exchange rate, official reserves,
and interest rate measured relative to a foreign currency.
The crisis is defined by the EMP reaching an extreme value,
i.e. by a factor times the standard deviation above the sample mean. Weighting in the EMP index results from different
volatility of the components, and weights are different for
every country. Sample mean and benchmark standard deviation can also be differentiated among countries.
Trying to find more precise definitions, WEO (1998, p.
115) introduced the additional notion of currency crashes,
related to the situation when the exchange rate component of the EMP index accounted for more than 75 percent
of its overall value. We will not follow such a detailed specification in our analysis, using the operational definition of
currency crises according to Eichengreen, Rose and
Wyplosz (1994) proposal.
Following this concept, Jakubiak (2000) tested 14 historical episodes8 considered by experts as currency crises,
using the individual components of the EMP index or similar concepts of measuring a decline in confidence to a specific currency.
6
Szczurek (2001) provides review of some operational definitions.
This kind of situation seems to be considered by Bordo and Schwartz (1998, p. 1) defining currency crisis as "...a clash between fundamentals and
pegged exchange rates, whether fixed or crawling". The similar approach is used by Ötker and Pazarbasioglu (1997).
8 The sample covered Argentina (1995), Brazil (1999), Bulgaria (1997), the Czech Republic (1997), Georgia (1998), Indonesia (1997), Korea (1997),
the Kyrgyz Republic (1998), Malaysia (1997), Mexico (1994), Moldova (1998), Russian Federation (1998), Thailand (1997), and Ukraine (1998). Due to
data unavailability some countries had to be excluded from individual tests.
7
CASE Reports No. 51
13
M. D¹browski (ed.)
Figure 2.2. Severity of the currency crises measured by reserve losses and nominal depreciation of the domestic currency
Country
b
Crisis date
Loss of reserves
Nominal depreciation against USD
a
Reserves/M2 at
in percent
at a crisis date
3 months after the
a crisis date
crisis
Mexico
Dec 1994
8.2%
64.5%
54.6%
97.62%
Argentina
Mar 1995
18.3%
41%
0.0%
0.0%
Bulgaria
Feb 1997
25.8%
16.8%
100.98%
53.92%
Czech Republic
May1997
29.6%
23.0%
5.44%
9.00%
Thailand
Jul 1997
23.3%
23.0%
24.34%
54.01%
Malaysia
Jul 1997
23.2%
18.4%
4.19%
35.85%
Indonesia
Aug 1997
18.3%
5.2%
16.78%
40.36%
Korea
Dec 1997
17.0%
33.2%
45.64%
18.83%
Russian Fed.
Aug 1998
14.9%
40.6%
26.72%
186.63%
Ukraine
Sep 1998
25.2%
58.1%
51.11%
52.31%
Moldova
Nov 1998
111.2%
35.0%
55.41%
40.76%
Kyrgyz Rep.
Nov 1998
87.96%
18.7%
19.40%
25.35%
Georgia
Dec 1998
59.9%
24.5%
16.8%
43.51%
Brazil
Jan 1999
24.2%
53.5%
64.08%
37.40%
Note: a – Monetary authorities’ reserve loss is calculated from the month when the stock of these reserves peaked until the crisis date (following
Choueiri and Kaminsky, 1999);
b – Scope of depreciation is calculated from the month before the crisis until one month ("depreciation at the crisis date") or three months later;
end-period exchange rates are used.
Source: Jakubiak (2000).
Generally, in most of the examined cases Jakubiak found
the relevance of three basic criteria proposed in the Eichengreen, Rose, and Wyplosz (1994) EMP index, i.e. loss of
international reserves or serious nominal depreciation plus
high real interest rates.
According to Figure 2.2, central banks of all the countries covered by this study experienced heavy losses of their
official reserves. In four out of fourteen analyzed cases
(Malaysia, Indonesia, Bulgaria, and the Kyrgyz Republic) they
lost less than 20% of their international reserves at the time
of a crisis. Malaysia and Bulgaria kept floating exchange rate
regime when the decisive stage of respective currency
crises occurred. However, Malaysian reserves deteriorated
again during the next year after the first wave of the crisis
and started to be rebuilt only at the end of 1998. Bulgaria
experienced a banking crisis in 1996, during which its central bank has already severely depleted its foreign reserves.
In the case of Indonesia, after floating of the rupiah in August
1997, the official reserves continued to fall, and the lowest
level was recorded in February 1998, giving a cumulative
decline by 24% comparing to the pre-crisis period. At the
other end of the spectrum, Mexico, Argentina, Russia,
Ukraine and Brazil lost over 40% of their international
reserves.
Figure 2.3. Real interest rate differentials around crisis dates, monthly
Month before/
after the crisis
m-6
m-5
m-4
m-3
m-2
m-1
CRISIS
m+1
m+2
m+3
m+4
m+5
m+6
Argentina
1995
2.57
1.94
1.99
2.08
2.17
3.30
11.65
11.36
8.40
4.29
4.07
3.31
3.86
Brazil
1999
11.32
11.51
23.61
30.63
24.87
23.03
29.84
37.97
35.60
26.91
19.79
15.77
15.25
Mexico
1994
6.90
7.93
5.84
4.85
3.95
4.16
4.53
11.59
12.08
23.68
23.35
8.45
-1.44
Moldova
1998
14.90
17.89
17.20
17.25
17.98
15.97
23.22
26.49
26.73
25.20
24.61
21.18
24.53
Russia
1998
-1.30
-1.58
-1.11
1.51
3.68
5.66
3.96
-32.33
-35.25
-48.19
-62.23
-75.96
-83.80
Thailand
1997
2.99
3.06
2.45
2.05
1.47
1.50
3.21
1.47
1.15
0.72
-0.04
-0.22
-1.03
Ukraine
1998
6.32
6.76
7.76
10.21
11.93
11.98
12.90
9.12
5.13
-0.01
-0.69
-3.18
1.89
Note: The interest rates differentials are calculated as the real deposit rates in a crisis country minus the real deposit rate in the US.
Source: Jakubiak (2000).
14
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
Figure 2.4. Nominal and real interest rates before and after the crisis
Nominal Rates (Deposit or Similar)
25
20
15
10
5
0
c-24
c-18
c-12
c-6
crisis
month
c+6
c+12
c+18
c+24
Real Rates (Deposit or Similar)
6
4
2
0
-2
-4
-6
-8
c-24
c-18
c-12
c-6
crisis
month
c+6
c+12
c+18
c+24
Note: Monthly data; graphs plot medians for the sample. Real rates were calculated using CPI index.
Source: B³aszkiewicz and Paczyñski (2001).
In addition to a fall in reserves all countries but Argentina experienced devaluation/depreciation of their currencies, sometimes very substantial. Only in the Czech Republic the scale of depreciation of the koruna can be considered
as modest.
What concerns the third component of the Eichengreen, Rose and Wyplosz (1994) EMP index, high domestic
interest rates were present in each of the analyzed countries at times when their currencies came under pressure
(see Figure 2.3). There were also significant increases after
a crisis hit, or shortly before, indicating the attempts of central banks to defend the currency.
CASE Reports No. 51
However, the relative magnitude of interest rates varied
across the countries. The highest rates were observed in
Brazil around the 1999 crisis. Significant increases were also
recorded in Argentina, Russia, Thailand, and Ukraine.
The similar results were obtained in comparative analysis of nominal and real interest rates in 41 developing and
transition economies carried out by B³aszkiewicz and
Paczyñski (2001) – see Figure 2.4.
A sudden fall in demand for money could be another
indicator of the loss of confidence in the currency subject to
market pressure. However, in Jakubiak’s (2000) sample currency crises did not manifest themselves clearly by the fall in
15
M. D¹browski (ed.)
Figure 2.5. Broad money monetization in the crisis countries, 1992–1998
Country
1992
1993
Argentina
11.2%
16.3%
Brazil
20.9%
22.9%
Bulgaria
Czech Rep..
61.1%
Georgia
Indonesia
38.3%
38.7%
Korea
35.7%
37.3%
Kyrgyz Rep.
Malaysia
67.3%
73.7%
Mexico
22.7%
24.6%
Moldova
10.4%
Russia
Thailand
69.5%
71.8%
Ukraine
13.1%
Note: Crisis period is indicated by the grey color.
Source: Jakubiak (2000).
1994
19.4%
25.1%
61.9%
62.3%
1995
18.8%
26.3%
55.7%
68.7%
40.4%
36.8%
42.3%
36.6%
78.6%
25.7%
11.4%
77.8%
25.1%
14.5%
71.0%
12.1%
72.2%
9.6%
the demand for money measured as the change in the level
of monetization (ratio of broad money to an annualized
value of GDP in current prices – see Figure 2.5).
There was a visible fall in demand for money in Bulgaria
in 1996–1997, in Argentina in 1995 and in the Czech Republic in 1997. Monetization was also depressed during the two
years following crises in Bulgaria and in the Czech Republic.
The results are ambiguous in relation to East Asia in
1997–1998 and the CIS countries in 1998 where in some
cases a crisis year brought even an increase in the overall
monetization level.
This a bit surprising results concerning changes in
demand for money can originate from measurement problems. First, due to lack of data on domestic currency aggregates Jakubiak (2000) used broad money data, which included foreign currency deposits. A substantial devaluation normally leads to an increase in the domestic currency denominated value of the foreign exchange deposits, thus artificially improving the broad money to GDP ratio. Second, endof-year stocks of broad money to nominal GDP ratios were
used (data availability was again ground for such a presentation) representing cumulative flow for all the year. Consequently, if a crisis happened at the end of the year this ratio
16
1996
21.1%
25.7%
42.8%
70.9%
5.8%
46.1%
38.4%
12.8%
83.1%
23.7%
17.6%
14.6%
75.4%
9.4%
1997
24.0%
26.4%
23.7%
65.6%
6.6%
49.4%
42.1%
11.9%
88.5%
25.1%
18.9%
16.4%
85.9%
11.8%
1998
27.5%
29.9%
26.7%
63.9%
7.6%
53.4%
50.7%
13.8%
93.2%
24.6%
17.8%
99.0%
13.1%
could be distorted. Third, CIS countries generally represented a chronically low level of monetization (even before
the crisis), so we could not expect significant fluctuations of
this ratio.
Generally, the Jakubiak (2000) study demonstrates that
the best known recent historical episodes of currency crises
in emerging markets meet the definition criteria proposed
by Eichengreen, Rose and Wyplosz (1994). On the other
hand, testing them according to additional criteria such as
changes in broad money monetization gives less univocal
results. However, an attempt to do the opposite, i.e. to
identify the episodes of currency crises using the EMP index
can produce misleading results, i.e. selecting "tranquil" periods as crises.
Sasin (2001a) mentions a trade-off faced by researchers
choosing a selection algorithm: the more out of the "true"
crises it captures, the more of "tranquil" periods it wrongly
marks as crises. Hence, any quantified definition of a currency crisis cannot serve as an automatic selection tool. An
additional expert-type assessment (clearly, involving a certain dose of arbitrariness) is always necessary. This type of
mixed approach has been adopted in empirical studies summarized in this paper.
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
3. Theoretical Models of Currency Crises
One can distinguish three generations of theoretical
models trying to explain the phenomenon of currency
crises. Each of them developed as a result of empirical
experience of the following waves of crisis episodes9. The
first-generation models were constructed after balance-ofpayment crises in Mexico, Argentina, and Chile in the1970s
and the early 1980s. The ERM crisis in 1992 and the Mexican crisis of 1994–1995 acted as stimulus for working out
the second-generation models. Finally, first attempts to
build the third-generation models started after the Asian
crisis of 1997–1998.
3.1. First-generation Models
Looking back, Mundell (1960) balance of payments
model10 was among the first attempts at presenting the
interdependence between the ability to maintain a currency peg and the level of the international reserves of a central bank. However, as mentioned earlier, the classical firstgeneration model was elaborated later on the basis of
empirical experience of the series of balance of payments
crises in Latin America in the decade of 1970s and in the
early 1980s. Krugman (1979) provided the first version of
such a model and Flood and Garber (1984) later simplified
and extended it11.
The first-generation model relates to a small open economy whose residents have a perfect foresight and consumes
a single, tradable good of a fixed, exogenously determined
domestic supply. There are no private banks and money
supply is equal to the sum of domestic credit issued by the
central banks and the domestic-currency value of foreign
reserves maintained by the central bank, which earn no
interest.
Central bank accommodates any changes in domestic
money demand through the purchases or sales of international reserves. Therefore, if domestic credit expansion
(usually caused by the monetization of a fiscal deficit)
exceeds the fixed money demand, international reserves
will be declining at the rate of credit expansion leading to
their final depletion.
The higher the initial stock of reserves and/or the lower
the rate of domestic credit expansion, the longer it takes
before an exchange rate peg is attacked and collapses. In the
absence of speculation, collapse of the peg exchange rates
occurs after depletion of reserves. The larger the initial portion of domestic credit in the money stocks the sooner the
collapse. Finally, the higher interest rate elasticity of money
demand, the earlier the crisis occurs.
Central bank is able to defend an exchange rate peg until
it has reserves and then has no choice but to allow an
exchange rate to float freely. Thus, rational agents observing
expansionary monetary policy can expect that a decline in
reserves and resulting collapse of an exchange rate peg at
some point is inevitable, even without a speculation. However, active speculators working in a competitive environment
and trying to avoid losses or earn gains at a time of collapse
will force the crisis occurrence before central bank's reserves
are depleted. The collapse may happen when the "shadow
floating exchange rate" becomes equal to the exchange rate
peg. This is the equilibrium exchange rate prevailing after the
full depletion of foreign reserves and forced abandoning of
the peg. As long as the peg exchange rate is more depreciated than the implicit shadow rate, it is not attacked because
potential speculators face a danger of losses.
9
The content of this section draws heavily from Antczak (2000), Siwiñska (2000) and Szczurek (2001) were used as an additional input.
In standard economic textbooks it is referred to as the Mundell - Fleming model.
11 Agenor, Bhandari and Flood (1992) provide an excellent overview of the evolution of the first-generation models.
10
CASE Reports No. 51
17
M. D¹browski (ed.)
Summing up, the first-generation crisis occurs as a result
of an expansionary macroeconomic policy incompatible
with a peg exchange rate. The speculative attack against
currency is provoked by a threat of depletion of central
bank’s international reserves caused by an excessive fiscal
deficit, being monetized by monetary authorities. The international reserves in the Krugman (1979) model take the role
of exhaustible resource in the equivalent model of Salant
and Henderson (1978) elaborated in order to question the
idea of commodity prices stabilization board proposed by
Hotelling (1931).
3.2. Modified First-generation Models
The basic first-generation model has been subsequently
extended in three main directions. The first one concerns an
active governmental involvement in crisis management and
sterilization of reserve losses. The second group of modifications abandons assumption on the perfectly foreseen
speculative attacks and introduces uncertainty. Finally, the
third direction is related to the target zone models.
The first modification refers to an experience of some
currency crises in the 1990s (UK, Mexico; later also in Russia
and Ukraine) when the negative money-supply effect of
reserve loses has been sterilized allowing a smooth money
growth through the period of attack. Hence, in this presentation, money supply, exchange rate, foreign price level, and
interest rate level remain constant during the attack. The difference with the canonical model lies in the structure of
money supply: during capital outflow declining international
reserves are substituted with domestic credit. However, this
only accelerates time of depleting the central bank international reserves, collapse of exchange rate peg and speculative
attack against currency. The straightforward policy conclusion is that an exchange rate peg cannot survive if the authorities plan to sterilize reserve losses and speculators expect it
(see Flood, Garber, and Kramer, 1995). Under sterilization
policy and free capital mobility a currency peg proves unsustainable regardless of the size of international reserves.
The second modification of the canonical model removed
the assumption of perfectly foreseen speculative attacks. Market participants are never sure when an attack will take place
and how much the exchange rate will change as its result.
Therefore, uncertainty becomes a crucial element in speculators’ calculations (see Flood and Marion, 1996).
18
The model contains non-linearity in private behavior,
which reveals multiple solutions. If agents expect more currency variability in future, it affects the domestic interest rate
through the uncovered interest parity relation and feeds into
the demand for money, making the exchange rate more variable. The shift in expectations, therefore, alters the relevant
shadow rate for determining whether an attack is profitable
and changes the attack time. Crises can still be the outcome
of inconsistency in macroeconomic policies (canonical first
generation models), but they can also arise from self-fulfilling
prophecies about exchange-market risk related to some or
all fundamentals. The existence of non-linearity in private
behavior assures that an economy can jump suddenly from
no-attack equilibrium to attack equilibrium.
The third kind of modification has originated from the
empirical observation that in many peg exchange rate
regimes, an exchange rate can still fluctuate within a certain
band around the official parity. This rises a question of how an
exchange rate behaves within the band. Krugman (1988) initiated the work on what later became commonly called as the
target zone models.
These models assume that authorities are fully committed
to maintain the band and follow policy of fixing the money
stock at a level allowing to keep the exchange rate within the
band. However, there are random disturbances in the money
supply process. Let us suppose that such a random disturbance increases the money stock. In the flexible exchange rate
system the exchange rate would depreciate. In target zone
models with strong and credible commitment of the authorities, speculators realize that the future money stock is more
likely to decrease than to increase. As a result, the exchange
rate is also more likely to appreciate than to depreciate in the
future. Speculators, therefore, will be willing to sell foreign
exchange today (expecting a lower price in the future). The
exchange rate will appreciate.
The opposite mechanism is at work for shocks
decreasing the money stock and putting appreciation pressure on the exchange rate. Maintaining the exchange rate
within the band requires from authorities to increase
money stock in the future and the exchange rate is more
likely to depreciate than appreciate. In an anticipation of
this fact speculators will be willing to sell domestic currency (expecting a lower price in the future). The exchange
rate will depreciate. As result one can conclude that speculation will be stabilizing the exchange rate, and speculators substitute fully credible authorities in maintaining the
exchange rate within the band.
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
If the authorities are less credible, shock in the money
stock will make speculators uncertain about its interpretation. The shock may be a result of random disturbance that
will be corrected by authorities in the future or may be a
result of a change in policy. The latter implies an increase in
the money stock (if exchange rate faces depreciation pressure due to a shock) and raises doubts about commitment
of authorities towards the target zone regime. Speculators
may interpret the weakening of currency as a signal of
future problems with fundamentals. In such case they will
be buying foreign currency instead of selling it. Therefore,
speculation will be a destabilizing factor.
Consequently, in the lack of full credibility case, domestic interest rates must be positively related to changes in the
exchange rate within the band. On the contrary, if speculators stabilize exchange rate, interest parity condition would
require domestic interest rate inversely related to the
exchange rate.
Destabilizing effects of exchange rate movements within
the band often led the authorities to intervene in the foreign
exchange market long before the limit of the band is
reached, limiting the variability of the exchange rate and narrowing the de facto width of the band. Therefore, the existence of a band does not give, in practice, a bigger flexibility.
3.3. Second-generation Models
The second-generation models were developed after
speculative attacks against the ERM in Europe in 1992 and
Mexican peso in 1994. They addressed serious drawbacks of
the first generation models. In the latter the governments and
central banks acted like lemmings: once engaged in a policy
incompatible with peg exchange rates, they were heading for
the disaster of reserve depletion (Szczurek, 2001). In reality,
the governments have more options: for example, they can
change their policy when balance of payments gets worse, or
devalue without depleting the reserves first. The second-generation models allow the governments to optimize its decision. The loss function is usually dependent on the exchange
rate and variables referring to both actual depreciation and
the prior public expectations of depreciation.
The second-generation model first proposed by Obstfeld
(1994) and further developed by Obstfeld (1997), Velasco
(1996), Drazen (1999) and many other authors12 draws on
12
game theory, more precisely non-cooperative game with
three players: authorities possessing a finite stock of reserves
to defend currency regime and two private players. The size
of committed reserve stock defines the payoffs in the two
players’ game, played by private agents. The outcome of the
game depends on the size of international reserves possessed
by authorities. In the first case called the "High Reserve" game
an amount of the official reserves is higher than the combined
stock of domestic money hold by both traders. Even if both
players sell their resources to a central bank, its reserves
remain at the level high enough to maintain the exchange rate
peg and speculators incur losses. The exchange rate peg will
survive the attack.
In the second case called the "Low Reserve" game the
level of central bank’s reserves is so low that each player
can solely take out a currency peg. A trader who manages
to exchange his entire domestic currency assets into foreign
currency receives a capital gain (in domestic currency
terms; alternatively she avoids a capital loss in foreign currency terms) net of transaction costs. If both traders sell,
each of them will be able to obtain half of the official reserve
stock and share a capital gain. In this variant the collapse of
an exchange rate peg is unavoidable.
The most interesting is the third "Intermediate Reserve"
game where neither trader alone can deplete the official
reserves but both can do it if they happen to coordinate a
speculative attack. The payoff structure is such that either
player fails in an individual attack, bearing the loss, while the
second players has zero payoff (no gain, no loss). But if both
attack, each registers a gain. Therefore, there are two Nash
equilibria: if both players sell the currency the exchange rate
peg must collapse, but if neither player believes the other will
attack the currency peg survives. So the intermediate state of
fundamentals (illustrated by the stock of international reserves
of a central bank) makes the crisis possible, but not certain.
In the canonical model fundamentals are either consistent
with a currency regime or not. In Obstfeld (1994) the same is
true for extreme values of fundamentals, but there is also a
large room within which fundamentals are neither so strong as
to make a successful attack impossible, nor so weak as to
make it inevitable. In this case speculators may or may not
coordinate their actions in order to attack the peg.
The second-generation crisis models require three elements: a reason for authorities to abandon its exchange rate
peg, a reason to defend it, and increasing cost of defending the
current regime when its collapse is anticipated or self-fulfilled.
A survey of second-generation models is provided in Eichengreen, Rose and Wyplosz (1996).
CASE Reports No. 51
19
M. D¹browski (ed.)
In order to have an incentive to attack the exchange rate
regime, there needs be something awkwardly fixed in domestic economy. The arguments in favor or against maintaining the
current (peg) regime may be of various natures. Obstfeld
(1994) based his example on unemployment and authorities
willingness to relax monetary policy, which cannot be implemented as long as the commitment to a peg exchange rate.
So, the logic of second-generation models arises from the
fact that defending exchange rate parity can be expensive (via
higher interest rates) if the market believes that it will ultimately fail. As a result, speculative attack on currency can
develop either as a result of a predicted future deterioration
in fundamentals, or purely through self-fulfilling prophecy.
3.4. Third-generation Models
The outburst of Asian crises in 1997–1998 brought a new
challenge for the theory. Clearly, these crises differed from the
ones experienced in Latin America during 1980s (first generation crises) or the ERM crisis of 1992 (giving the pretext for
building the second-generation models). First, none of the fundamental problems such as high fiscal deficit, expansionary
monetary policy, or high inflation – typical for the first-generation models – was observed in the Asian countries. Second,
the authorities in these countries did not face any dramatic
trade-offs between political and economic goals, an issue on
which second-generation models were based. Third, all Asian
countries experienced a boom-bust cycle in their asset markets, preceding the currency crisis. Fourth, the currency crises
were only part of a widespread financial crisis, which included
also collapses of many banks and non-banking financial institutions as well as bankruptcies of large non-financial corporations (Krugman, 1998a).
Two major approaches dominated in the post-1997 theoretical literature. The first one, represented by McKinnon
and Phil (1996) as well as Krugman (1998a, 1999) modeled
the "over-borrowing syndrome", and emphasized the role of
moral-hazard-driven lending by unregulated banks and financial institutions (Corsetti, Pesenti, and Roubini 1998a, b).
According to this view, a rational agent can expect the
government rescue operation of any large bank or corporation with good political connections in case it had solvency
problem. This assumption has two kinds of implications.
Expectation of future bailing out means a sort of hidden subsidy to investment, thus stimulating a boom-bust cycle on the
asset market. On the other hand, part of a private sector
20
"over-borrowing" may be interpreted as an implicit government debt. The currency side of a financial crisis can therefore be understood as a consequence of the anticipated fiscal
costs of financial restructuring and its partial monetization.
Generally, this approach referred, in some way, to the
first-generation models, which stressed the key role of the
policy fundamentals. The difference was that while canonical and modified first-generation models concentrated on
fiscal and monetary factors leading to speculative attacks,
the above mentioned interpretation of the Asian crises
extended this analysis also to microeconomic flaws.
In the alternative view represented by Radelet and Sachs
(1998), a self-fulfilling pessimism of international lenders
caused financial fragility of the Asian countries. The authors
stressed that while there were significant underlying problems within Asian economies at both macroeconomic and
microeconomic level, the imbalances were not severe
enough to cause a financial crisis of such magnitude. Radelet
and Sachs (1998) blamed a combination of a set of factors
such as panic in the international investment community,
policy mistakes in crisis management, and poorly designed
international rescue programs, for triggering a fully fledged
financial panic resulting in currency crises, bank runs, massive bankruptcies, and political disorder. Chang and Velasco
(1998) have proposed the similar approach, i.e. explaining
Asian currency crisis as a product of a bank run.
Thus, this direction of theoretical effort referred to second-generation models putting the attention on multiply equilibria and self-fulfilling character of the speculative attacks.
Analysis of a herding behavior went in the similar direction.
The first theory of herding (Chari and Kehoe, 1996)
underlines that a bandwagon effect is driven by an assumption
that some investors have private information. Another explanations focus on the principal-agent problem, i.e. that these
are agents rather than principals who manage money investment processes. When money managers are compensated
based on comparison with other money managers, they have
more incentives to follow the others (herd behavior) even if
they are wrong, rather than to make independent decisions.
Building on this discussion, Krugman (1999) elaborated a
third-generation model concentrating on microeconomic
weaknesses (such as moral hazard and resulting over-borrowing), which may trigger, in the world of high capital mobility, speculative attacks against the existing exchange rate
regimes. This model attempts to consider both fundamental
factors and multiply equilibria in the market behavior.
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
4. Early Warning Signals
The decade of 1990s brought a large body of empirical
literature trying to design a system of the so-called early
warning signals (indicators) that could help to predict currency crises. These studies have been motivated mainly by
practical considerations of providing policymakers with an
analytical tool helping them to avoid crisis episodes and
their heavily devastating consequences (see section 8).
Focusing on symptoms of mounting imbalances and misalignments (countries’ vulnerability to crisis) rather than on
their fundamental roots they can serve as a good starting
point for a deeper analysis of crisis-generating factors.
A commonly used approach involves comparing the
behavior of a set of macroeconomic variables before a crisis
with that during tranquil times. One of the possible variations
of this methodology is to monitor the stylized facts in the
period preceding the currency crisis. The pre-crisis behavior
of a variable is compared to its behavior during non-crisis
periods for the same group of countries or for the group of
countries where no crisis occurred. The aim is to find variables that display anomalous performance before a crisis but
do not provide false signals predicting crisis, which will never
happen (see WEO, 1998, p. 126; Tomczyñska, 2000).
Figure 4.1. Early warnings indicator (statistically significant)
Sector
1.
Monetary Policy
2.
Fiscal Policy
3.
Real Sector
4.
External Sector
5.
Global Variables
6.
Institutional and
Structural
Variable
International reserves
M2/int. Reserves
real exchange rate
inflation
money
money multiplier
credit growth
central bank credit to banks
real interest rates
fiscal deficit
government consumption
credit to public sector
real GDP growth or level
employment/unemployment
trade balance
exports
terms of trade
Foreign interest rates
Domestic-foreign interest rate
differential
foreign real GDP growth
banking crisis
financial liberalization
openness
crisis elsewhere
Number of studies
considered
12
3
14
5
3
1
7
1
1
5
1
3
9
3
3
3
3
4
Statistically significant
results
11
3
12
5
2
1
5
1
1
3
1
3
5
2
2
2
2
2
2
2
1
2
1
1
1
1
1
1
1
1
Source: Tomczyñska (2000) following the analyzes of Kaminsky, Lizondo, Reinhart (1998).
CASE Reports No. 51
21
M. D¹browski (ed.)
Figure 4.2. Behavior of selected warning indicators (stylized facts)
Indicator
real exchange rate
Foreign exchange reserves
Exports
terms of trade
Before Crisis
Monetary policy
Strong increase and beginning of decline
just before the crisis
Strong increase
Fiscal policy
Increase
Real Sector
Decline
External Sector
Very strong appreciation
Strong decline
Strong deterioration
Deterioration
M2 multiplier
real interest rates
Domestic credit/GDP
Financial Liberalization
Increase
Fluctuation and decline
Strong increase
M2/reserves
Excess real M1 balances
fiscal deficit/GDP
Output
After Crisis
Decline
Decline
Improvement
Start to increase
Depreciation
Start to increase
Increase
Poor, with predisposition to improve
over longer time
Decline and remains on low level
Increase
Decline
Note: The pre- and post-crisis behavior is compared to the average behavior during normal periods. Observation of the 12-month percentage
changes of variables, except for interest rate, real exchange rate and excess real M1 balances which are in levels.
Source: Tomczyñska (2000) following Kaminsky and Reinhart (1999).
Kaminsky, Lizondo and Reinhart (1998) presented a very
detailed overview of such indicators. Any signal identified
within the 24-months window before the crisis was considered as a good one, while any signal outside that period was
regarded as a false alarm.
The leading indicators were grouped into the following
broad categories: domestic macroeconomic variables,
external sector variables, public finance, global variables,
and institutional and structural variables. After the additional selection based on other empirical studies authors came
to conclusions summarized in Figure 4.1. It shows the
number of studies, in which any particular indicator was
considered and results were statistically significant.
The above results were partly confirmed by the crosscountry empirical analyzes carried out by the members of the
CASE research team. Sasin (2001a) examined the panel of 46
developed and developing countries for the period of 1990s.
The special attention was devoted to distinction between variables emphasized by the fist-generation and second-generation
models, including multiple equilibria and contagion effect. Considerable amount of predictability was found in respect to such
"classical" indicators as overvaluation of a real exchange rate and
the level of central bank’s international reserves. Multiple equilibria did not get much support from the investigated data while
contagion, through various channels, was clearly present.
B³aszkiewicz (2000) conducted an econometric probit
analysis aiming to establish the most important determinants
of the currency crises in East Asia in 1997–1998. Although
22
her results occurred to be mixed (the probit modeling
turned out to be very sensitive to changes in the sample size
and introduction of new variables and brought up an important issue of causality) they stress importance of fundamental problems such as current account deficits, total debt,
short term debt, slowing GDP growth rate, exchange rate
appreciation or excessive capital inflow rather than just a
market panic as suggested by Radelet and Sachs (1998).
The next step in investigating "early warning" indicators
(after completing their list) is looking into their individual
behavior. Example of such an analysis is provided, among
others, by Kaminsky and Reinhart (1998) (see Figure 4.2).
Jakubiak (2000) carried out the similar exercise examining
a set of macroeconomic variables in 14 emerging economies.
She found that the ratio of central bank’s international reserves
to short-term external debt gradually decreased during the
months preceding crises (Figure 4.3). Mexico, Korea, Indonesia, Russia, Bulgaria, and Thailand represented cases where
international reserves were well below countries’ short-term
external liabilities. Argentina also recorded low level of the
examined ratio but no earlier than at the crisis date. Although
this ratio was on a decline in Malaysia and Brazil, as they were
approaching the crisis date, its value stood above 1.5 in both
cases, thus not indicating dramatic external liquidity problems.
The same is true for Ukraine although this country recorded
very rapid deterioration of the reserves-to-debt ratio.
Jakubiak (2000) study also confirms that the ratio of the
official international reserves to reserve money is of partic-
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
Figure 4.3. Reserves over short-term debt
3.5
3.0
Ukraine
Malaysia
Brazil
Thailand
Argentina
Indonesia
Russian Federation
Bulgaria
Korea Republic
Mexico
2.5
2.0
1.5
1.0
0.5
0.0
month t-3
month t-2
month t-1
crisis
Notes: The stock of reserves in taken from the months prior to the crisis dates, and at the crisis dates. Short-term external debt is the end-year
value reported in Global Development Finance.
Source: Jakubiak (2000).
Figure 4.4. Ratio of the official international reserves to reserve money
2.5
2.0
Argentina
Brazil
Bulgaria
Czech Republic
Georgia
1.5
1.0
0.5
0
m-12
m-9
m-6
m-3
crisis
m+3
m+6
m+9
m+12
4
3. 5
Korea
3
Kyrgyz Republic
Malaysia
Mexico
2. 5
2
Moldova
1. 5
Russia
Thailand
1
Ukraine
0. 5
0
m-12
m-9
m-6
m-3
crisis
m+3
m+6
m+9
m+12
Source: Jakubiak (2000).
CASE Reports No. 51
23
M. D¹browski (ed.)
Figure 4.5. Real effective exchange rates in crisis countries
150
140
130
Czech Republic
120
Malaysia
Moldova
110
100
Russia
90
Ukraine
80
Bulgaria
70
60
m-24
m-18
m-12
m-6
crisis
m+6 m+12
m+18
m+24
Source: Jakubiak (2000) using IFS IMF data.
ular importance in the case of peg/fixed exchange rate
regimes. A fall of this indicator below one indicates potential
difficulties with the commitment to a declared nominal
anchor. As seen from Figure 4.4 problems with insufficient
backing of the monetary base by central bank’s reserves
could be observed in Argentina, Brazil, Bulgaria, Russia,
Ukraine, Georgia, Kyrgyz Republic, and, to some extent, in
Mexico during pre-crisis months. East Asian economies
recorded relatively safe values of this measure in the months
prior to the crisis. Nevertheless, they were substantially
lower during the last months preceding the crisis than at
other times. This indicator did not have any importance in
relation to Malaysia, which kept floating exchange rate
regime around the crisis time.
Similarly, the real appreciation of a currency proved an
important factor. As we see from Figure 4.5 real effective
exchange rates have been indeed appreciating through several months prior to the crises in Russia, Malaysia, Czech
Republic, and, to some extent, also in Moldova. The same
has been true for Bulgaria but only when we consider the
banking crisis in 1996, which preceded the drastic currency
crash at the beginning of 1997. Real effective exchange rate
had not been appreciating in Ukraine during at least six
months prior to the crisis.
For the rest of "crisis" countries where real effective
exchange rate index was unavailable Jakubiak (2000) calculated the CPI based real exchange rate index versus US dollar, which gave less clear results (see Figure 4.6) because
countries under consideration traded also in other currencies. However, beginning of a currency crisis was marked by
the large real exchange rate depreciation in every economy.
Later on, real exchange rate usually recovered partially.
Figure 4.6. Real exchange rates in relation to US dollar, CPI based
120
100
Brazil
Indonesia
80
Korea
Mexico
Thailand
60
40
20
m-24
m-18
m-12
m-6
crisis
m+6 m+12
m+18
m+24
Source: Jakubiak (2000).
24
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
5. Causes of Currency Crises
This chapter contains the survey of results of empirical
research related to the selected underlying causes of currency crises. We start from fiscal imbalances and later move
to a current account deficit, currency overvaluation, the role
of specific exchange rate regimes, structural weaknesses in
the banking and corporate sectors, and political instability.
5.1. Fiscal Imbalances
The role of fiscal imbalances in generating currency
crises seems to be obvious at least from the time the firstgeneration models appeared in theoretical literature (see
section 3). Fiscal deficit, if monetized by a central bank at
the pace exceeding increase in demand for money must
sooner or later lead to an exhaustion of official reserves and
a currency crash. This link is a more indirect and delayed in
the case of non-monetary financing of the fiscal deficit, i.e.
when government borrows on the market. In this case
excessive borrowing, increasing yields on government
bonds and perspective of the possible debt trap influence
investors’ expectations and push them from the optimistic
equilibrium to the pessimistic one. Thus, investors’ behavior
and crisis spiral can be well explained in terms of the second-generation models (see Siwiñska, 2000 for a detailed
analysis of this mechanism).
When we turn to empirical research, the picture
becomes less clear. Results depend on the sample selection
and adopted analytical methods. A short review of the selected previous research efforts is presented in Figure 5.1.
Still, this kind of causality, i.e. fiscal imbalances leading to
currency crises, was established in less formalized case
studies related to transition economies (Markiewicz, 1998
for Hungary, Bulgaria, Russia, Ukraine and Kyrgyzstan;
Antczak R., 2001 for Russia; Markiewicz, 2001 for Ukraine;
Ganev, 2001 for Bulgaria; Radziwi³³, 2001 for Moldova).
Siwiñska (2000) conducted two kinds of comparative
analysis related to this issue. In the first one she examined
the behavior of a number of fiscal variables before and dur-
Figure 5.1. Links between fiscal variables and currency crisis – review of empirical research
Authors.
Eichengreen,
Rose, Wyplosz
(1994)
Sample
1967–1992,
23 countries,
mostly of ERM
Univariate Analysis
Econometric Analysis
Links found, but only in non-ERM sample.
Effect of budget deficit and
In the non-ERM sample the hypothesis of the equal distribution of public debt on the
the fiscal deficits in crisis and non-crisis periods is rejected; the
probability of a crisis is
same hypothesis is accepted in the ERM sample.
insignificant.
Frankel, Rose 1971–1994, 105 Links not found.
(1996)
countries
Fiscal deficit tends to be small and shrinking in countries
experiencing a crisis.
Kaminsky,
1970–1995, 20
Links found.
Reinhart (1999) countries
Fiscal deficit found to be higher in the two years prior to currency
crisis, as compared to tranquil period. Fiscal deficit as an indicator
of crisis has accurately called 27% of currency crisis (the lowest
share out of the considered indicators).
Aziz,
1975–1997, 20
Weak links found.
Caramazza,
industrial and 30 Fiscal deficit found to be on average larger for two years before
Salgado (2000) developing
the crisis across the whole sample, but the result was not
countries
significant at 95-percent level and not robust across different sub
sample.
Source: Siwiñska (2000) on the basis of: Aziz, Caramazza and Salgado (2000), Eichengreen, Rose and Wyplosz (1994), Frankel and Rose (1996),
Kaminsky and Reinhart (1999).
CASE Reports No. 51
25
M. D¹browski (ed.)
ing the crisis, compared to tranquil periods. The sample
consisting of 30 developing countries for the period of
1980–1999 and 20 transition countries for the period
1992–1999 constituted the base for this comparison.
Both in developing countries (only for the period of
1990–1999 – earlier the "tranquil" periods were also characterized by the large deficits) and transition economies, the
crisis was preceded by larger, than in normal times, fiscal
deficits. In CIS countries the difference has been up to 80%.
The external public debt in developing countries failed
to display any distinct pattern, but sovereign foreign debt in
transition economies was, on average, higher before and
during the year of a crisis. The domestic debt in developing
countries was higher than tranquil average for up to two
years before the crisis. This stands in line with Obstfeld
(1994) model. The lack of comparative data for CIS countries did not allow the author to analyze the domestic public debt for this sub-sample.
The second analysis carried out by Siwiñska (2000) compares the fiscal developments in a number of transition
economies in the second half of 1990s in a more descriptive
way. The focus was put on the coincidence between fiscal
imbalances and occurrence of currency crises. Figure 5.2
presents summary of this analysis related to countries of the
former Soviet Union. Most of them suffered from the series
of 1998–1999 currency crises.
The ratio of general government fiscal deficit to GDP
proved the most reliable indicator in predicting currency
crises. This tendency was confirmed by the larger sample
covering also Central European countries. This indicator
issued a right signal in 7 out of 8 cases when a currency crisis did occur within next 24 months and produced only two
wrong signals out of 10 cases when there was no crisis.
The amount of external public liabilities owed to commercial creditors also performed quite well but only when
the arbitrary measures (marked by an asterisk in the exhibit) were introduced. Still, this indicator had a tendency to
overstate the danger of a crisis. The indicators of the
indebtedness level occurred even less reliable.
Generally, Siwiñska (2000) study confirms an important
role of fiscal imbalances in triggering currency crises in
developing and transition countries. In particular, large budget deficits and rapidly growing short and medium-term liabilities were among the key determinants of the series of
1998–1999 crises in the CIS countries. When the investors
became increasingly concerned about the ability of governments to honor their obligations they began to withdraw
their capital, central banks were faced with significant
Figure 5.2. The value of fiscal indicators in predicting a crisis in selected FSU economies
Deficit of
general
government
External debt
Short – term
external debt
Debt service to
reserves
External public
liabilities owed
to private
creditors
A
No. Of
countries,
where
indicator was
larger and
crisis did
emerge
within 24
months
3 (+);
6(+*)
B
No. Of
countries,
where
indicator was
larger and
crisis did not
emerge
within 24
months
1 (+);
2(+*)
2
1 (+*)
C
No. of
countries,
where
indicator was
not larger
and crisis did
emerge
within 24
months
4(+)
1 (+*)
D
No. of
countries,
where
indicator was
not larger
and crisis did
not emerge
within 24
months
4
2
6 (+);
5 (+*)
5
2
0
2 (+);
4 (+*)
1 (*+);
0 (+)
1 (+*)
True
indications
(A+D)
Wrong
indications
(B+C)
Tendency
to issue a
signal
before a
crisis
A/(A+C)
Tendency
not to issue
a signal
when there
is no crisis
D/(D+B)
7 (+);
10 (+*)
5 (+)
2 (+*)
3/7 (+);
6/7 (+*)
4/5 (*+)
3
5 (+);
6 (+*)
5
8 (+*);
9 (+)
7
5/14(+);
8/14 (+*)
2/7
3/5
5
5
7
5
2/7
1
5 (+);
3 (+*)
5 (+);
4 (+*)
7 (+);
8 (*+)
5 (+);
4 (*+)
2/7 (+);
4/7(+*)
1 (+);
4/5 (+*)
4 (+*)
4/5
Notes: (+) denotes that an indicator was larger than its average plus one standard deviation across all FSU; (*) indicates that the indicator was relatively large, but did not exceed the statistical threshold. In case of budget deficit "relatively large" was over 5% of GDP, during at least two years; in
case of public debt owed to private creditors – over 4% of GDP, and external debt – over 40 of GDP; (+*) denotes, that both (+) and (*) were considered.
Source: Siwiñska (2000).
26
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
reserves and exchange rate pressures and governments
with a liquidity crisis.
However, one must remember that fiscal imbalances in
CIS economies reflected deeper structural shortcomings
and the lack of consistent reforms: soft budget constraints
across the economy, weak governments, inefficient tax systems, Soviet-type budget expenditures.
5.2. Current Account Deficit
5.2.1. Evolution of the Theoretical Views
Over the past few decades there have been important
changes in the way the economist perceive the current
account – from "current account deficit matters" through
"current account deficit is irrelevant as long as the public
sector is balanced" and, again, "deficit matters", finally to
"deficit may matter" (see Edwards, 2001)13.
Until mid-1970s, the discussion on external imbalances
was dominated by the "elasticity approach" focusing on the
necessity to have trade flows balanced. Because most of
developing countries run large and persistent current
account deficits at that time recurrent devaluations were
the standard remedy to cope with this problem.
However, effects of these devaluations depended on
export and import price (exchange rate) elasticities.
Researchers were divided in their assessment concerning
these results. Optimists saw devaluations as effective tools
for restoring trade balances. For example, Cooper (1971)
examined 21 major devaluation cases during the period of
1958–1969 and found them successful in bringing the current account back to balance. On the other hand, representatives of the "elasticity pessimism" argued that elasticity
occurred to be small and countries had to arrange a large
exchange rate adjustment in order to improve their external position that in turn caused a huge income and output
contraction. According to "pessimists" view since developing
countries exported mainly commodities and there was no
prospect for surge in demand for them in the world market
the devaluation was ineffective and the answer was not to
devalue but to encourage industrialization through import
substitution policies.
After the oil shock and dramatic deterioration of the current
account balance of oil importing countries in the 1970s the trade-
13
flow/elasticity approach was replaced by the inter-temporal
approach. This approach based on two arguments: (i) the current
account is equal to national savings minus investment; (ii) both savings and investment decisions are based on inter-temporal factors
such as permanent income, expected return on investment projects,
etc. As far as current account deficit reflected new investment perspectives but not falling saving rates there was no reason to be concerned about it. The deficit meant only that economic agents
expecting future prosperity brought by new investment opportunities were smoothing their consumption paths–moving it from the
future to the present. The influential paper of Jeffrey Sachs (1981)
strongly insisted on this view.
In addition, the new approach made a distinction
between the deficits that result from fiscal imbalances and
those reflecting private sector decisions. The public sector
was thought to act rather on political than on economic
ground, so the current account deficit induced by the fiscal
deficit was regarded as "bad", while private sector's decisions were assumed rational and the current account deficit
responding to them was perceived "good". It was assumed
that in the future private sector would be able to make necessary corrective actions (i.e. repay debt) while public sector most probably would not. This view was most frequently associated with the former UK Chancellor of the Exchequer, Nigel Lawson.
However, the debt crisis of 1982 exposed the obvious
inadequacy of the Lawson Doctrine. In fact, in some of the
crisis affected countries (example of Chile – see Edwards and
Edwards (1991)) large current account deficits were accompanied by the balanced fiscal accounts. Emphasis was put
again on the current account deficit and the real exchange
rate overvaluation (see e.g. Fischer, 1988; Cline, 1988).
This view was even stronger reiterated after the
1994–1995 Mexican crisis. This led Lawrence Summers, the
then Deputy US Treasury Secretary to warning that any current account deficit in excess of 5% of GDP should be subject of attention (Summers, 1996; Edwards, 2001). This
gave the beginning of the new Doctrine of 5% adopted in
practice of both the IMF and private investors in the late
1990s (some other analysts used the threshold of 4%). Still,
this view is not shared by everybody.
For instance, Dornbusch (2001) distinguishes between
old- and new-style currency crises. The first type – current
account crisis – involves a cycle of overspending and real
appreciation that worsens a current account, decreases foreign reserves and finally leads to devaluation. The new-style
This section follows closely Sasin (2001b) review, which in turn draws from Edwards (2001).
CASE Reports No. 51
27
M. D¹browski (ed.)
crises, which can be called "capital account crises", are connected with the potential insolvency of the balance sheets of
a significant part of the economy (for example, large portfolio
of non-performing loans) or by maturity (and currency) mismatches and manifest themselves by a sudden capital flight.
Other authors try to determine "current account sustainability" (see e.g. Milesi-Ferretti and Razin, 1996;
Edwards, 2001). Because of the lasting improvement in capital market access, persistent terms of trade improvement
and productivity growth emerging market economies can,
as it is predicted by the inter-temporal models, finance moderate current accounts on an ongoing basis.
The weakest notion of sustainability implies that the present value of the (future) current account deficits (plus debt)
must equal the present value of the (future) surpluses, or in
other words that a country will (in infinity) repay its debt.
This criterion is certainly not satisfactory. The debt repayment prospects may be too distant and say nothing about
the appropriateness of a present deficit: virtually any present deficit can be (somehow) undone by sufficiently large
surplus in the (unspecified) future.
According to the stronger notion, the deficit is sustainable if it can be reverted into sufficient surplus in the foreseeable future and debt repaid on an ongoing basis (in a
sense of non-increasing debt/GDP ratio) without drastic
policy changes and/or a crisis. This definition is a starting
point for a calculation of a sustainable current account. If the
actual deficit lasts longer above sustainable level and a country does not undertake corrective measures (restrain of
domestic demand or devaluation) it can perhaps expect an
externally forced adjustment.
Summing up the discussion on whether the current
account "matters" as a potential cause of currency crises
Edwards (2001, p.37) gives a partly positive answer. He
states that "if this question is interpreted very narrowly, in the
sense that countries with an (arbitrarily defined) large current
account deficit, almost inevitably face a crisis, then the answer
is "no." If, however, it is interpreted more broadly, as suggesting
that there are costs involved in running "very large" deficits (...)
the answer is a qualified "yes". Also Dornbusch (2001) concludes that "...it is safe to say that a rapid real appreciation –
say over 2 or 3 years – amounting to 25 percent or more, and
an increase in the current account deficit that exceeds 4 percent of GDP, without the prospect of a correction, takes a country into the red zone."
5.2.2. Results of Empirical Research
Sasin (2001b) provides an extensive review of the existing empirical research, examining links between current
account deficits and currency crises. In most cases the studies do not show a strong and significant correlation between
(high) current account deficit and currency crises in a proper econometric treatment.
Figure 5.3 contains the summary of various attempts to
predict currency crises, which usually include current
account deficit as one of explaining variables. In the second
column the t-statistics for the null hypothesis that the current account deficit is irrelevant in crisis prediction is presented.14 Whenever author tests more than one specification additional t-statistics are presented.
As can be seen from Figure 5.3, the reviewed papers
present mixed results. One possible explanation can be
that much depends on the sample selection and the specification of the variables included. This possibility was formally confirmed by Sasin (2001a) who found that depending on specification one could obtain values for t-statistics
from -5 to 8.
As a proper econometric methodology is unable to
provide strong proofs in favor of the impact of external
deficits on crises, the narrative case studies can shed some
light on the issue. For example, Milesi-Ferretti and Razin
(1996) attempt to answer when and why current account
problems may end in a currency crisis. They discuss the
cases of Australia (persistent current account deficit, no
drastic policy actions, no crisis), South Korea (early
1980s), Israel, Ireland, (high deficit, preventive policy
reversal, no crisis), Chile, Mexico (deficit, crisis). Basing
on the improved portfolio approach (with financing constraints) the main determinants of current account deficit
sustainability ex ante included: the size of the export sector, level of international competitiveness, level of domestic savings, composition of external liabilities, strength of
the financial system, degree of political stability and fiscal
balances.
Their analysis shows that among countries with a current account deficit, the group that did experience currency
crises differs from those that did not in the following
respects: overvaluation of the real exchange rate (with rapid
growth of imports), relatively small export sector, high
external debt burden, low and declining saving ratio (espe-
14
Values over 1.9 indicate that, with 95% confidence, the current account has an impact on the emergence of currency crises. For around 1.6 the
confidence level is 10%.
28
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
Figure 5.3. Summary of the empirical researches showing the role of a current account deficit in predicting currency crises
Study
Edwards (2001)
Milesi-Ferretti and Razin
(1998)
Abhuvalia (2000)
Caramazza et.al. (2000)
Bussiere and Mulder
(1999)
Eichengreen, Rose and
Wyplosz (1996)
Frankel and Rose (1996)
Berg and Pattillo (1999)
Kaminsky, Lizondo and
Reinhart (1998)
Sasin (2001a)
Results (t-statistics)
1) 1.64 2) 1.64
3) 1.44 4) 0.31
1) 1.25
2) 2
3) 1.65
4) 0.36
5) 1.3
6) 2.25
7) 0.6
8) 0.77
9) 1.71 10) 2.05
1) 1.94
2)1.44
3) 1.52
4)1.43
1) 2.17 2) 2.5
3) 1.84 4) 0.51
1) 2.03
1) 0.8
2) 1.9
3) 1.2
1) 1.03 2) 0.22
1) 5.6
2) 9.5
3) 5.5
-not t-statistics, but
"noise to signal ratio" –
export change: 0.42 (the
third best result)
1) 1.1
2) 3.1
3) -0.1(wrong sign)
4) 1.2
5) -0.4 (wrong sign)
6) 2.8
Notice
four definitions of crisis and different specifications
different samples
1-4) during current account reversals
5-10) prediction of overall crash
two samples
two different set of contagion controls
different specification of crisis index
Early Warning System with 5 regressors
different specification of variables
1) default and 2) predictive power
1) "indicator model"
2) linear model
3) "piecewise linear model"
univariate "signal" analysis, "noise to signal ratio"; 0-perfect prediction,
0.5-no information, >0.5 worse than unconditional guess
- 1,3,5) fixed effect linear model; 2,4,6) probit; 1,2) full sample; 3,4)
emerging markets; 5,6) developed economies
- Actually Sasin checks around 10,000 specifications and concludes that
an average significance for current account is 0.5 (2 for developed and
0.1 for emerging economies).
Source: Sasin (2001b).
cially Chile and Mexico15), political instability (elections in
Mexico) and the composition of capital inflows16. All countries arranged devaluation but some were preventive and
successful (Australia, Israel) while some countries were
forced to do it by the crises (Chile, Mexico).
On the other hand, B³aszkiewicz and Paczyñski (2001)
investigating a sample of 41 crisis-affected developing and
transition countries found that almost all of them suffered
from persistent current account imbalances and that deficit
oscillating around 5% of GDP should be considered as a
"red flag" by policymakers. However, it does not mean the
opposite – that each country with such a deficit must suffer
a currency crisis.
The above observations seem to be confirmed by the
current account developments in transition economies.
While Central European and Baltic countries have been able
to run substantial current account deficits (sometimes well
exceeding 5% of GDP) on a sustainable basis, some CIS
countries (for example, Russia and Ukraine) have been
pushed, in fact, into a current account surplus. The first
group is advanced in the EU accession negotiation, its financial markets and banking systems are consider as being relatively stable and these countries managed to create a
favorable business climate for FDI and portfolio investments. On the other hand, the investment climate in the
CIS is much worse than that of the EU candidates, and the
1998–1999 series of financial crises effectively closed access
of these countries to the international financial market.
5.3. Currency Overvaluation
Analysis of the real effective exchange rate seems to be
very closely related, at least at the first glance, to the issue
of the current account deficit discussed in the previous subsection. However, empirical results differ substantially (see
below). While a current account position does not deter-
15
Ireland and Israel also experienced a decline in saving rates but it was an effect of public sector imbalances that were later much improved. Australia’s saving rate is low as well but it is seemingly counterbalanced by the efficiency of its banking system.
16 Australia relied more on equity investments. Significant part of its debt was denominated in domestic currency which meant that risk was shared
by foreign investors. Ireland, in turn, had a large and sustainable inflow of FDI’s.
CASE Reports No. 51
29
M. D¹browski (ed.)
mine univocally probability of a currency crisis occurrence,
this relation seems to be more explicit in the case of
exchange rate overvaluation17. One can try to explain this
difference by the specific level of sustainable current
account balance for each country and, perhaps, by problems
with measuring a real exchange rate and its overvaluation.
5.3.1. How to Measure a Real Exchange Rate
Overvaluation?
While the idea that an overvalued currency violating the
law of one price can lead to trade and balance of payments
disequilibrium and cause devaluation sounds reasonable, the
operative measuring of this overvaluation is not an easy task.
This is strictly connected with the methodological difficulties
to find a concrete numerical level of the equilibrium
exchange rate in an individual economy. In addition, in an
open economy such an equilibrium level can be subject to
frequent changes, responding to various nominal and real
shocks. As a result, the overvaluation can arise as a consequence of (see Sasin, 2001b):
• changes in an external environment, for example, in
terms of trade or as a result of real depreciation of the
major trade partner’s currencies;
• domestic supply-side shocks;
• policy related causes, for example, limited credibility of
an exchange-rate-based disinflation program;
• large foreign capital inflow, which put an upward pressure on the exchange rate.
In practice estimation of the over-/under-valuation of the
exchange rate can be done in two ways (Sasin, 2001b). The
first approach is based on the assumption that equilibrium
RER implies balanced current account or some current
account position considered as sustainable. As the initial
step, the long-run level of domestic saving and investment is
estimated, then the normal capital flows (assuming some
long-term equilibrium interest rate differentials, growth rate
of the economy, etc.) are determined. If the two sides are
different it means that the real exchange rate is not in balance. Afterwards, the equilibrium value being able to equate
the two sides is assessed basing on estimated exchange rate
elasticity of various macroeconomic variables. Subtracting
prevailing exchange rate from the equilibrium one gives the
measure of RER overvaluation.
17
30
The second approach uses the econometric and statistical tools and rather abstracts from a detailed countryspecific knowledge. The easiest way is to explicitly use the
notion of purchasing power parity. One selects the price
index (CPI, WPI, PPI) and a period of time, and then
decides that the average of index-based RER over the chosen period constitutes an equilibrium rate. Subtracting that
average from the current index-based RER gives the RER
overvaluation.
Other popular method, slightly more demanding,
requires empirical estimation of the real exchange rate
determinants what is usually done through one-equation
regression. Basing on the common knowledge and various
models, relevant theoretical fundamentals are selected
(these usually include terms of trade, degree of openness,
government expenditures, etc.) and included into the
regression. Afterwards the fundamentals are decomposed
into permanent and temporary components. The permanent ones are included in the estimated real exchange rate
equation and equilibrium (fitted) rate is inferred. Even more
sophisticated method requires an underlying model of the
exchange rate – it is usually a monetary model.
Theoretically, the overvaluation can be undone by the
exactly opposite processes to the one that led to its occurrence. However, because of downward price and wage rigidity it is more difficult to arrange a smooth real depreciation
(restore the equilibrium) than to allow the real appreciation.
Goldfajn and Valdes (1996) assumed that, after controlling
for other macroeconomic fundamentals, the real exchange
rate overvaluation could be undone in two ways: by cumulative inflation differentials and by nominal depreciation (including a currency crisis). Afterwards, they calculated the probability that overvaluation would end smoothly without a sharp
nominal depreciation (crisis) for a large set of countries over
the period of 1960–1994. Probability of reverting the prolonged (over six-month) overvaluation successfully was 32%
for an appreciation not exceeding 15%, 24% for 20% misalignment, 10% for 25% threshold, 3% for 30% overvaluation and zero for overvaluation exceeding 35%.
5.3.2. Results of Empirical Research
Sasin (2001b) presents an ample empirical evidence that
overvaluation can explain currency crises. Figure 5.4 con-
This section draws on Sasin (2001b).
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
Figure 5.4. Summary of the empirical researches showing the role of real exchange rate overvaluation in predicting currency crises
Study
Edwards (2001)
Milesi-Ferretti and Razin
(1998)
Ahluvalia (2000)
Caramazza et.al. (2000)
Bussiere and Mulder (1999)
Frankel and Rose (1996)
Berg and Pattillo (1999)
Goldfajn and Valdes (1996)
Kaminsky, Lizondo, and
Reinhard (1998)
Sasin (2001a)
Results (t-statistics)
1) 0.03
2) 1.05
3) 0.59
4) 0.12
1) 4.75
2) 4.9
3) 3.8
4) 3.04
5) 3.25
6) 3.75
7) 6
8) 5.6
9) 6
10) 2.8
1) 2.7
2) 3
3) 1.46
4) 2.48
1) 2.17
2) 1.70
3) 0.62
4) 1.07
1) 1.9
1) 1.51
2) 2.53
1) 15.9
2) 13.5
3) 3.35
1) 1.69
2) 1.53
3) 2.63
4) 1.51
Not t-statistics, but"noise
to signal ratio"
1) 0.19 (the best result)
1) 4.7
2) 5.4
3) 1.6
4) 2.8
5) 4.1
6) 2.6
Notice
four definitions of crisis
overvaluation as deviations from PPP
different samples
1-4) during current account reversals
5-10) prediction of overall crash
two samples
two different set of contagion controls
different specifications of crisis index
Early Warning System with 5 regressors
1) default
2) predictive power
1) "indicator model"
2)linear model
3)"piecewise linear model"
different models and nominal vs. real devaluation
univariate "signal" analysis, "noise to signal ratio"; 0-perfect
prediction, 0.5-no information, >0.5 worse than unconditional
guess
1,3,5 -fixed effect linear model; 2,4,6 -probit; 1,2 -full sample; 3,4 emerging markets; 5,6 -developed economies
Actually Sasin checks around 10,000 specifications and concludes
that an average significance for RER is 4 with standard deviation of
about 2
Source: Sasin (2001b).
structed similarly to Figure 5.3 contains the summary of
empirical research testing such a hypothesis18.
The results presented in Figure 5.4 indicate strong support for the hypothesis that real overvaluation is linked to
currency crises. The advocates of the opposite view have
only few arguments. The most often raised is the BalassaSamuelson effect, according to which real overvaluation (as
revealed by price index) should not impair competitiveness
because of higher increase of productivity in tradable goods
sector. Sasin (2001b) confirms this view to some extent, discovering that the significance of the real exchange rate to
currency crisis prediction is much lower for the emergingmarket economies. On the other hand, as noted by Dornbusch (2001), the Balassa-Samuelson model is often used to
justify overvaluation in the presence of large current
account deficits, while, according to this model, real
exchange rate appreciation, other things being equal, should
not deteriorate trade equilibrium.
Some researchers, skeptical about the econometric
methodology prefer to use "before-after analysis", which
usually depicts the stylized facts associated with currency
crises. Aziz, Caramazza and Salgado (2000) provide a recent
example. They categorize crises into subgroups: crises in
industrial countries, in emerging economies, crises characterized by currency crashes19, by reserve losses, "severe"
crises, "mild" crises, crises accompanied by banking sector
problems, crises with fast and slow recoveries. Afterward,
they analyze how the given variable (real exchange rate)
behaves on average in the time window before and after a
crisis. Their results, once again, support the hypothesis of a
significant role of currency overvaluation in causing currency crises.
The similar conclusion is formulated by Jakubiak
(2001a) who analyzed probability of ending peg in 14
emerging-markets economies. Real exchange rate appreciation increases this probability. B³aszkiewicz and Paczyñski
(2001) who estimated changes in real exchange rates for
the group of crisis-affected developing and transition countries also found a gradual real appreciation through 48
months preceding a crisis.
18
In the second column the t-statistics for the null hypothesis that the overvaluation is irrelevant in crisis prediction is presented. Values over 1.9 indicate that, with 95% confidence, the real overvaluation has an impact on the emergence of currency crises. For around 1.6 the confidence level is 10%.
19 Crises in which currency depreciation accounts of more than 75% of a crisis index.
CASE Reports No. 51
31
M. D¹browski (ed.)
5.4. The Role of Exchange Rate Regimes
The first reaction of many economists after a recent
series of financial crises both in the developed countries (the
ERM crisis in 1992–1993) and in emerging markets in
1995–1999 (Mexico, Thailand, Malaysia, Korea, Philippines,
Indonesia, Russia, Ukraine, and Brazil) was the strong critique of the peg exchange rate regimes. These regimes
became generally blamed for generating currency crises
(see e.g. Obstfeld and Rogoff, 1995; IIE, 1999; Sachs, 1998;
Mishkin, 2000; Mishkin, 2001) and policy recommendations,
including those of the IMF, started to go towards advertising
more flexible exchange rate arrangements.
Doubts concerning a pegged exchange rate usually concentrate on four arguments:
1. Exchange rate peg itself becomes an easy target of
speculative attack.
2. Exchange rate stability (predictability in the case of
crawling peg) stimulates excessive capital inflow and overborrowing in foreign currencies, particularly if financial and
corporate sectors experience serious regulatory weaknesses.
3. Stable exchange rate helps currency overvaluation
(what does not need to be necessarily true as a real
exchange rate is rather unmanageable under free capital
movement at least using monetary policy instruments).
4. Unsustainable peg ends up with sharp devaluation,
which triggers sudden capital outflow and heavily damages
balance sheets of financial and non-financial corporations.
The above critique seems to be a bit superficial, not necessarily well balanced, and sometimes excessively emotional. As Jakubiak (2001a, p. 43) correctly points out "...it is
clear that the eruptions of the currency crises were not caused
by the fixed regime choices themselves, but rather by the inconsistent macroeconomic policies within this financial framework".
Empirical case studies conducted by Jakubiak (2001a)
demonstrate a variety of the actual exchange rate regimes in
the fourteen crisis-affected countries – from an incomplete
currency board (Argentina) through various kinds of relatively
stable exchange rate arrangements (Brazil, Georgia, Indonesia,
Korea, Malaysia, Mexico, Moldova, Russia, Thailand, and
Ukraine) to mostly floating rates (Bulgaria, Czech Republic,
Kyrgyz Republic)20. On the other hand, logit estimates show
the increasing probability of ending a peg when real exchange
rate appreciates and with more months spent with a peg.
20
32
A floating exchange rate regime does not provide a guarantee that a country avoids a currency crisis and its painful
consequences that could, for example, include massive
insolvency of banks and enterprises with foreign exchange
denominated liabilities. On the other hand, it has certain
advantages comparing to a peg. First, the real free float
excludes possibility of a market speculative test of the sustainability of the declared exchange rate trajectory. Second,
as Mishkin (2001, p. 35) notes "...movements in the exchange
rate are much less nonlinear in a pegged exchange rate regime".
Third, the floating exchange rate makes all borrowers (both
private and public) more aware of the exchange rate risk.
Fourth, significant depreciation or appreciation of the
exchange rate may serve as an early warning signal for policymakers that economic policy needs some adjustment.
Floating exchange rate may be, however, a difficult and
costly regime for developing or transition countries suffering
from chronic high inflation, high currency substitution, and
lack of credibility of domestic monetary policy. This is the
reason why a fully floating exchange rate is not a very popular variant in emerging-market economies even among
those that recently went through currency crises and were
forced to abandon a formal peg. Calvo and Reinhart (2000)
call this phenomenon the "fear of floating". In this context
even some of the above mentioned critics of the peg
exchange rate regimes (see Mishkin, 2000 and 2001) accept
the possibility of introducing the hard peg in the form of a
currency board or dollarization/ euroization.
This brings us to the central point of the discussion concerning monetary/exchange rate regimes and their responsibility for generating currency crises. In fact, this is not the
exchange rate regime itself (e.g. fixed rate) which creates a
danger of a speculative attack but an attempt to manage
simultaneously both the exchange rate and the domestic
money supply. The intermediary regimes such as the
adjustable peg, crawling peg, target band, crawling band or
managed float all violate, in fact, the principle of the "impossible trinity" (Frankel, 1999). According to this principle any
country must give up one of the three policy goals: exchange
rate stability, monetary independence, and financial market
integration. Having all three simultaneously proves impossible. Assuming that the free (or relatively free) capital
mobility is irreversible, the future choices of a monetary/exchange rate regime will need to be restrained to
basically two options – either monetary independence or
the exchange rate fixing.
The de facto implemented regimes often differed from the declared ones – see Jakubiak (2001a).
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
Why the intermediate regimes are difficult to operate
and can provoke speculative attacks? The first and most fundamental reason is that compromised solutions are unlikely
to provide the advantages of any of extreme regimes, i.e.
neither an exchange rate anchor nor a sufficient discretion
in managing domestic liquidity. On the contrary, they may
bring both a substantial exchange rate variability (actual or
expected when a peg is not perceived credible) and make
money supply exogenous (i.e. being out of control of monetary authorities). Second, compromised regimes can prove
technically very difficult to manage due to the fluctuating
demand for money and changing market expectations.
Moreover, the pressure of current economic and political
conditions may bring temptation to go beyond the compromise. Third, transparency, and, therefore, credibility of
intermediate regimes is lower than that of the extreme
solutions.
5.5. Structural Weaknesses of the
Banking and Corporate Sectors
As we look at the history of currency crises, it is apparent that these were often accompanied by banking crises
(see e.g. Glick and Hutchison, 2000). The series of Asian
crises in 1997–1998 and CIS crises in 1998–1999 brought
the researchers’ attention back to the structural and institutional weaknesses of commercial banks and their role in
provoking currency troubles. In addition, Asian crises illustrate the importance of the proper corporate governance,
particularly if the large financial-industrial conglomerates
dominate an economy. The similar conclusion might be
drawn from the 1998 Russian crisis.
This recent experience gave impulse to development of
a new (third) generation of currency crisis theoretical models discussed in details in section 3 of this paper.
5.5.1. Relationship Between Banking
and Currency Crises
In analyzing the impact of a banking crisis on probability
of provoking a currency crisis one can find several potential
transmission mechanisms21. First, economic agents have
good reasons to expect that authorities will prevent the
21
financial system from collapse and bail it out by monetary
expansion. In practice, this means preference towards higher inflation relative to giving up the exchange rate stability.
Second, domestic agents can run on domestic banks, withdraw deposits and then convert this money into the hard
currency (this kind of behavior is very likely to occur in both
developing and transition economies). This puts a pressure
on the exchange rate and usually leads to depreciation or
devaluation. Third, a banking crisis impairs credit relations,
worsens the position of domestic corporations, and brings
about an economic slowdown. As the availability of profitable investment opportunities is decreasing, foreign
investors might choose to withdraw their assets. As they
would likely be reluctant to leave their money in hands of
presumably insolvent banks and would expect other
investors to think along similar lines, they might consequently expect a massive capital outflow and currency collapse. Thus, in order to avoid potential losses, they would
probably choose to withdraw as quickly as possible triggering a sharp devaluation. This kind of scenario was recently
observed in Turkey in 2000–2001 (Sasin, 2001d).
However, the opposite causality – from a currency crisis
to a banking crisis – is also often in place. The main mechanism at work is connected with the significant role of foreign exchange liabilities in balance sheets of commercial
banks and large corporations of many developing and transition economies. Even if commercial banks follow strict
rules of balancing foreign exchange assets and liabilities (that
was clearly not the case in many crisis-effected countries)
non-financial corporations (banks’ major clients) often run
huge currency mismatches in their balance sheets. Sharp
depreciation (devaluation) of a national currency rapidly
deteriorates the financial situation of corporations with
large unhedged foreign exchange liabilities thus negatively
influencing both liquidity and solvency position of the commercial banks. If banks run open foreign exchange positions
on their own this only accelerates eruption of the systemic
banking crisis. The second and probably less important
channel is the impact of devaluation on the higher expected
inflation and, therefore, on short-term interest rates.
Because the banking system is funded on short-term money
(maturity mismatch) the increase in the respective interest
rates poses an increased burden on their balance sheets.
The balance sheets of shortly indebted corporations also
deteriorate and negatively affect the banking system
through an increase in non-performing loans.
The entire subsection 5.5 bases on Sasin (2001c).
CASE Reports No. 51
33
M. D¹browski (ed.)
There are also common factors to both crises and they
manifest themselves through financial market reactions. If the
economic situation is unfavorable, prospects for future profitable investment are weak, politicians or economic agents act
irresponsibly, and foreign creditors are not sure whether
exchange reserves are sufficient to cover potential short-term
obligations there can be a shift in the market sentiment and a
sudden capital outflow. Investors can stop rolling over the debt
demanding immediate repayment. There are only two alternatives at the authorities’ disposal to counteract the situation,
i.e. an increase in interest rates or devaluation. Both actions
have a very harmful impact on the unhedged and imbalanced
financial and corporate sectors. Often both take place, so a
currency and banking crises – the twin crises – coincide.
5.5.2. The Most Frequent Weaknesses of the
Banking Sector
Banks act as intermediaries whose primary function is
maturity transformation: their liabilities consist mainly of
short-term deposits while assets of medium- and long-term
loans. By their nature, they must be highly leveraged institutions. In addition, they face several kinds of risks connected
with borrower creditworthiness, unexpected changes in a
yield curve and exchange rate, and with non-credit operation on the equity or real estate markets. A small change in
the balance sheet performance translates into very large
changes in banks’ capital. When the value of their assets less
liabilities falls below some point banks are believed to be
undercapitalized; when it turns to be negative, banks are
considered insolvent (negative net worth).
Banking problems can be classified according to various
criteria. The most popular approach distinguishes between
the liabilities-side distress (bank run) and the asset-side
problems (boom-bust cycle).
Bank run spiral can be well described in terms of multiple equilibria and game theory (see Sasin 2001c). It is worth
noticing, however, that recent banking crises in industrialized countries such as the Nordic (early 1990s) or Japanese
(1990s) ones do not have a liability-side character. Such a
character seems, in turn, typical in emerging markets where
large creditors (very often non-residents) choose to leave
following the disclosure of some bad news concerning the
asset side of the financial system. This was a case in recent
crises in Bulgaria in 1996, Indonesia (and Asia in general) in
1997–98, Russia in 1998 and Turkey in 2000–2001. However, all these were not really examples of runs against otherwise solvent financial institutions (as in the classical bank-run
models) but rather against presumably insolvent banks suffering from the asset deterioration.
The asset-side crisis usually takes a form of an endogenous boom-bust cycle with over-lending. Investors’ overoptimism, herding behavior and "disaster myopia" typical for
the boom phase of the cycle22 may be additionally stimulated by both loose monetary and fiscal policies and the
numerous regulatory and microeconomic weaknesses
encouraging moral hazard behavior of both banks and their
clients (see Krugman, 1998b). In particular, one can mention
expectations of the government bailing out borrowers in
trouble, political influence on banks lending decisions, badly
designed deposit insurance (guarantee) schemes, improper
sequencing of financial market and capital account liberalization, weak banking supervision, weak prudential and
accounting standards, weak property rights protection and
contract enforcement mechanisms, and many others. Many
of the above characterized Asian and CIS economies.
5.5.3. Review of Empirical Research
Analyzing the panel of 21 industrial, 37 developing and
32 emerging-market countries over the period of
1975–199723 Glick and Hutchison (2000) find that out of
these 90 countries 72 had banking crises and 79 experienced at least one currency crisis. There were 90 banking
crisis episodes and 202 currency crisis episodes. Out of 90
banking crises 37, i.e. 41% of the total have been twincrises.24 The frequency of banking crises increased four
times between the decade of 1970s and 1990s while the frequency of currency crises remained more or less the same.
The occurrence of twin crises rose as well. All types of
crises, in particular the twin crises, were more frequent in
the financially liberalized emerging markets.
The same authors also provide empirical evidence on
the interrelation between currency and banking crises (see
22
Sometimes, these are activities that would be classified as illegal or criminal in developed countries. We have in mind, for example, massive insider/connected lending, stripping bank/corporation profits/assets, looting, and corruption.
23 "Emerging markets" were defined as countries with relatively open capital markets while "developing country" sample included other developing and transition economies.
24 "Twin" crises were defined as banking crises accompanied by a currency crisis in the previous, same, or following year.
34
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
Figure 5.5. Performance of bank crises as a signal of currency crises and vice versa
Freq of accompanying
Freq of accompanying
Number
Number
currency crises (%)
banking crises (%)
CumulaCumulaof
of
Currency crises as bank crisis
Bank crises as currency crisis
tive
tive
freq.
currency
banking
indicator (index)
indicator (index)
freq.
crises
crises
t-1
t
t+1
t-1
t
t+1
7
7
5
11
16
15
18
41
202
All
90
0.98
1.44
1.42
1.38
1.40
0.98
7
8
5
10
18
15
19
42
160
Developing
71
0.82
1.66
1.35
1.32
1.59
0.82
11
14
6
9
24
20
29
50
78
Emerging
46
0.77
2.46
1.96
1.87
2.30
0.87
Note: Predictability index: the higher the value of the index, the better predictability, the value of 1 indicates ambiguous informative content.
Source: Glick and Hutchison (2000); Sasin (2001c).
Figure 5.5). The frequency of banking crises accompanied
by currency crises is higher than the frequency of currency
crises accompanied by banking crises. Currency crises tend
to cluster one year after a banking crisis while banking crises
accompanied a currency crisis usually in the previous year.
Both findings support the view that banking crises provoke
currency crises rather than the opposite.
Also the comparison of the predictability index developed by Glick and Hutchison (2000) reveals that currency
crises in period t and t+1 can be well predicted by the
occurrence of a banking crisis in period t. The predictability
is stronger for emerging markets (values of 2.46 and 1.96
respectively). The occurrence of a currency crisis in period
t does not contain any information regarding the next period probability of a banking crisis (the value of 0.98).
The referred analysis was univariate. When exploiting a
cross-correlation among variables in the multivariate model
the results change somehow. Currency crises as predictors
of banking crisis remain insignificant (except for contemporaneous events). The usefulness of banking crises as predictors of currency crises decreases substantially. They issue a
proper (and very strong) signal of an approaching currency
crisis for the emerging markets only. The contemporaneous correlation is significant for both developing and
emerging markets.
Kaminsky (1998) develops a so-called "signal approach"
to assess what are the determinants of both currency and
banking crises. The estimated coefficients are noise-to-signal ratios (ntsr): the coefficient equal to zero indicates perfect predictability power of the variable, the coefficient
equal to one indicates the power of a simple unconditional
guess (i.e. the variable in question is neutral) while values
greater than one disqualify the variable as a predictor. She
notices that it is a little harder to predict banking crises
(compounded ntsr=0.8) than currency crises (ntsr=0.7).
The best variable to predict both crises is the real exchange
CASE Reports No. 51
rate overvaluation (ntsr=0.2 for currency crises and 0.3 for
banking crises). Consistently with other studies she finds
that a banking crisis is a very good indicator of a currency
crisis (ntsr=0.3) while the opposite is not true (ntsr=1.2).
5.6. Political Instability
The role of a political factor in provoking currency crises
seems to be obvious but certainly needs in a more concrete
specification. The notion of political instability/fragility may
involve a broad range of situations. On the one extreme,
these can be such dramatic events as external or domestic
military conflicts. Some countries suffer chronic constitutional problems like conflict between executive and legislative branches of governments (very frequent in CIS and
Latin American countries) or between federal (central) and
regional governments (examples of Russia, Brazil and
Argentina). Many developing and transition countries do not
have a well-established configuration of political parties
what creates difficulties in forming stable and predictable
governments. Finally, even in the most stable democracies
there are episodes of minority governments, unstable government coalitions, irresponsible behavior of individual
politicians or political parties, and uncertainty related to
forthcoming election results. On the other hand, even the
very stable authoritarian regime can be challenged by
democratic aspirations of societies.
Political instability creates uncertainty among money
holders what, in turn, increases the risk margin associated
with a specific currency or debt instrument. This may trigger the sudden capital outflow, stop capital inflow or push
money holders towards currency substitution.
Political instability decreases chances of correcting
unsustainable policies inconsistent with a currency peg and,
35
M. D¹browski (ed.)
therefore, accelerates the moment of a speculative attack
(under the first- and third-generation crisis models). It also
increases probability of inconsistent policy goals what may
trigger a speculative attack according to the second-generation crisis models.
Finally, political instability complicates crisis management
(see section 6), deepening its scale and negative consequences. In some cases political instability can be endogenous vis-a-vis a crisis. When crisis starts it usually undermines legitimacy of a government that often results in political destabilization, which in turn makes any bold anti-crisis
steps very difficult.
36
Looking at the recent crisis episodes, one can find several cases of significant role of political fragility in causing or
at least triggering a crisis. This relates, for example, to Mexico (Paczyñski, 2001), Argentina both in 1995 and even
more in 2001–2002 (Jakubiak, 2001b), Brazil, Russia
(Antczak R., 2001), Ukraine (Markiewicz, 2001), Moldova
(Radziwi³³, 2001) and Turkey (Sasin, 2001d). The outburst of
crises in many of these countries contributed to serious
political destabilization, initiating a negative "crisis spiral".
This happened in Bulgaria (Ganev, 2001), Russia (Antczak R.,
2001), Indonesia (Sasin, 2001f) and recently in Argentina
(2001–2002).
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
6. Crisis Management – How to Defend
an Exchange Rate if at All?
After a currency crisis’ outburst its further dynamics
depends very much on government ability to take the
required corrective measures, introduce a consistent policy
package, act consequently and quickly enough, influence financial markets developments and calm market panics. Political
factors mentioned above (see subsection 5.6) play an important role here. Also support of the international financial institutions (notably the IMF) may help a government in regaining
influence on financial markets and help to convince market
participant to resist temptation to withdraw their capital.
As analyzing all the potential scenarios of crisis management would go far beyond the thematic boundaries of this
paper we will concentrate on one important dimension: how
to manage an exchange rate if market pressure is already
there? Because a currency crisis is, by its nature, connected
with speculative attack against the exchange rate this must
be a central point in any discussion on crisis management.
Hypothetically, when the speculative attack starts (or
can be expected soon) the first dilemma, which policymakers face is their capacity to defend an exchange rate. This
concerns the level of central bank’s liquid foreign exchange
reserves, room of maneuver in increasing interest rates and
fiscal adjustment, access to external aid, opportunity to persuade financial markets high probability of future positive
scenario. If such capacity does not exist it is better to give
up an exchange rate earlier in an orderly way than later in
the atmosphere of the market panic. This is a very popular
topic of ex post discussions in countries, which were eventually forced to devalue, for example in Russia. Of course, it
is usually not easy to assess ex ante the government and
central bank capacity to defend an exchange rate but sometimes the situation is quite clear from the very beginning.
Assuming that authorities have capacities to defend an
exchange rate the next question concerns the rationale of
doing it. In this respect every government must balance (ex
ante) costs and benefits of each policy option. Defending
exchange rate is likely to involve loss of central bank’s foreign reserves and monetary contraction (if an intervention
is non-sterilized; if it is sterilized it only increases the effectiveness of a speculative attack – see subsection 3.2), higher interest rates, lower external competitiveness, output
contraction, and higher unemployment. On the other hand,
currency depreciation, although helpful for trade and current account balance (at least in the short term), can be disastrous for inflation and inflationary expectation, future
credibility of monetary and exchange rate policy, and is
bound to increase domestic currency value of the foreign
exchange denominated debt (see Jakubiak, 2001a)25. This
last factor serves sometimes as a crucial argument in favor
of authorities’ attempt to avoid sharp devaluation as they
are afraid about potential default on their own foreign
exchange denominated debt and massive bankruptcies of
banks and enterprises. In addition, if country successfully
resists the attack against its currency it may strengthen its
credibility for future, incurred costs notwithstanding. This is
particularly important in the case of hard peg solutions such
as a currency board.
If speculative attack is successful and authorities give up
the exchange rate level they must at the same time decide
on the future strategy. The first possible option is to allow
an exchange rate to float expecting that market will find a
new equilibrium level. This option minimizes a risk of missing a new exchange rate target (and associated further credibility loses) but involves a danger of serious overshooting
(towards depreciation) at least for some period that, however, can be long enough to cause a massive insolvency of
commercial banks and enterprises. The alternative choice
involves re-pegging of an exchange rate on the new level
25
On the contrary, if government, banks or enterprises run large domestic currency denominated liabilities at fixed interest rates a surprised devaluation can provide a certain relief to debtors. The key question is, of course, how it will influence borrowing conditions in future. In addition, it is rather uncommon situation in developing and transition economies, which generally suffer huge currency and maturity mismatches between their assets and liabilities.
CASE Reports No. 51
37
M. D¹browski (ed.)
Figure 6.1. Declared and actual changes in the exchange rate regime during the crisis
Country
Brazil
Thailand
Mexico
Russia
Georgia
Ukraine
Korea
Indonesia
Malaysia
Moldova
Kyrgyz Rep.
Czech Repub.
Bulgaria
Declared exchange rate
regime prior to the crisis
Adjustable band with dual
exchange rate
Basket peg
Crawling band
Exchange rate band
Conventional peg
Horizontal band
Exchange rate band
Crawling band
Managed float
Managed float
Managed float
Horizontal band
Free float
Exchange regime into
which the country
switched after the crisis
independently floating
independently floating
independently floating
managed float
independently floating
horizontal band
independently floating
independently floating
managed float***
independently floating
managed float
managed float
currency board****
FLT index over 12
months prior to the
crisis
0.080
FLT index over 12
months after the crisis
0.081
0.081
0.124
0.185
0.202
0.256**
0.271
0.285
0.358
0.683
0.943
1.048
0.437
0.360
1.093
0.566
0.558
0.530
0.495
1.221
0.672
0.414
0.796
0.128
0.505*
Notes: FLT index is defined as…
* over 11 months;
** counted back form the beginning of the Thai crisis. It is commonly agreed that the crisis in Korea erupted in December 1997, but won was
under serious pressure form the summer of 1997. For this reason period from July to November 1997 was excluded from the pre-crisis calculations.
*** Malaysia switched to the conventional peg arrangement in September 1998, a year after the currency crisis started.
**** Bulgaria adopted a currency board arrangement in July 1997 that is 4 months after its currency collapsed.
Source: Jakubiak (2001a).
and/or according to a new formula. The key problem is connected with the ability of authorities to find a proper new
level and formula (this is not an easy task in the environment
of the market turmoil) and convince financial markets that
the new peg will be credible and defendable. This kind of
approach has more chance for success if it has a preemptive
character (prior to crisis) and when the new peg is hard
enough in institutional terms (for example, a currency
board) rather than when abandoning the previous regime is
forced by a market panic.
In both cases the success of the adopted strategy (in terms
of minimizing costs of a crisis) depends on the comprehensiveness and consistency of the entire adjustment package
(including fiscal policy, banks rehabilitation, structural measures, etc.), determination and credibility of government
actions, and external support (for example, from the IMF).
Jakubiak (2001a) tried to determine empirically what kind
of exchange rate regime the 13 crisis-affected countries actually chose after the previous one collapsed. In order to identify the real regimes (which sometimes differ significantly
from the declared ones) she used the so-called "effective
exchange rate flexibility index" (FLT) following Poirson
(2001). The index is calculated as a ratio of monthly nominal
exchange rate depreciation to the monthly relative change in
reserves. It is based on the idea that if a country follows a
26
38
floating regime, the volatility of its exchange rate should be
relatively large, while the monetary interventions – small.
Conversely, if a country follows a currency peg, the movements in the exchange rate are expected to be close to zero.
The indices were calculated using monthly data and then
averaged for the 12 months prior to the crisis, in order to
get a better picture of what was the pre-crisis situation. The
index could take the values between zero (no exchange rate
movements) and infinity (no interventions). The small values
of the index meant that the authorities were intervening
heavily on the foreign exchange to offset market forces and
a country followed a relatively hard peg.
Results presented in Figure 6.1 suggest two kinds of conclusions. First, there was certain discrepancy between the
declared and actual regimes both before and after the crisis.
Second, most of countries, which had de iure or de facto currency pegs before the crisis, decided to allow for a greater
exchange rate flexibility in the period directly following the
crisis (apart from Argentina and Hong Kong, which managed
to defend their currency boards)26. On the contrary, countries, which had more flexible exchange rates before, narrowed their fluctuations after the crisis. Bulgaria was the most
extreme case as it adopted the currency board regime.
Some of them, for example, Russia, Ukraine and Georgia later stabilized their exchange rates again.
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
7. Contagion Effect
The new feature brought by currency crises in the
decade of 1990s is their fast propagation from one country
to another. This happened after the collapse of British
pound in November 1992, after the Mexican crisis in
December 1994 (the so-called Tequila effect), after the Thai
crisis in July 1997 (sometimes called the "Asian flue") and
again after the Russian crisis in August 1998. A large body of
both theoretical and empirical literature started to analyze
what was referred to as a "contagion" effect.
7.1. Definition Problems
Rawdanowicz (2001, p. 80)27 rightly points out that
although the term contagion seems to be intuitively understandable, it is surrounded by many misconceptions. In some
papers authors have not explicitly defined contagion at all
(e.g., Glick and Rose, 1999) or have provided a specific definition for particular purposes of their research, like in Eichengreen, Rose and Wyplosz (1996, p. 19): "The contagion effect
with which we are concentrated can be thought of as an increase
in the probability of a speculative attack on the domestic currency which stems not from "domestic fundamentals" such as
money and output but from the existence of a (not necessarily
successful) speculative attack elsewhere in the world".
In other papers more formal and clear-cut distinctions
have been put forward. For instance, Masson (1998) distinguishes three types of phenomena related to the spread of
crises: monsoonal effects – arising from a common shock
such as an interest rate hike in the US; spillovers – a crisis in
one country worsens the macroeconomic fundamentals in
other countries, for instance, via trade linkages; contagion –
crises spread unrelated to fundamentals, being an effect, for
example, of a shift in market sentiment or in the perception
27
of market conditions. Though Masson’s definitions have conceptual appeal, some problems with differentiating between
spillovers and contagion can arise on the empirical side.
Forbes and Rigobon (1999; 2001) proposed a different
definition. They coined the term shift-contagion, i.e. "a significant increase in cross-market linkages after a shock to an
individual country (or group of countries)". Cross-market linkages can be measured by various statistics, like the correlation in returns on assets, the probability of a speculative
attack or the transmission of shocks or volatility. This definition is very neat for testing purposes, however, it focuses
mechanically on one specific channel and therefore ignores
developments in other segments of the economy as well as
overall economic background.
7.2. The Channels of Crises Propagation
Describing the channels of crisis propagation Rawdanowicz (2001) follows Forbes and Rigobon (2001) and
the WEO (1999b, chapter III) concept.
The first channel connected with multiple equilibria was
described by Masson (1998). According to this author, a crisis in one country affects investors’ expectations, which
then leads to a shift from a good to a bad equilibrium in
other countries. Expectations are thus the mechanism by
which crises spread. Clearly, this channel would not work
during tranquil periods. Masson argues that multiple equilibria are associated with pure contagion.
The second channel, i.e. endogenous liquidity shock,
occurs when a crisis in one country causes liquidity strains
in other countries. Investors, in order to survive, maintain
margin calls, etc. and have to alter their portfolios. Very
often this entails selling assets in other countries, which in
The entire section 7 draws heavily from Rawdanowicz (2001).
CASE Reports No. 51
39
M. D¹browski (ed.)
turn sparks crises. The drop in liquidity can also induce
tighter credit rationing in other countries, which can lead to
the selling of assets.
The third channel, called political contagion, can be activated when there is a political conflict concerning exchange
rate target, on the one hand, and policy objectives such as
competitiveness, dynamics of output or fight with unemployment, on the other. A country may then decide to sacrifice its peg exchange rate to meet these other objectives.
Such a possibility is revealed by a successful speculative
attack, which, in turn, impacts on other countries in a similar position, which may also experience pressure on their
currencies. This channel has been analyzed thoroughly by
Drazen (1999) and applied to the 1992 ERM crisis. In fact,
this is a modification of a second-generation crisis model.
The fourth channel relates to trade. Demand for foreign goods in a crisis-hit country is slashed due to (i) devaluation/depreciation of the domestic currency and (ii)
lower economic activity dampened by higher interest
rates. In practice, the exchange rate channel is the most
important as it has a direct and immediate impact as
opposed to the aggregate demand channel. Lower demand
in the crisis-hit country induces strains on countries that
export significant amounts to this country. A drop in their
exports may lead to current account problems and trigger
a crisis too. The same effect can be expected if two
exporting countries compete in third markets and one of
them devalues its currency.
While considering the trade channel, the issue of timing
and expectations should be brought to attention. In the
spillovers paradigm (see Masson, 1998) depreciation in the
crisis-hit country worsens export prospect in other countries. This process is rather long lasting – depending on the
structure of trade contracts – but probably not shorter than
3 months. Faster reaction involves the expectation mechanism. Trade linkages may enter the reaction function of
financial market actors.
The fifth channel deals with common aggregate shocks
such as a change in world interest rates, a slowdown in
world output growth or changes in bilateral exchange rates
among major world currencies.
In practice, the above channels may occur simultaneously and the separation of their effects is statistically difficult if
not impossible.
7.3. Empirical Investigation of Contagion
Effect in CIS Countries
Rawdanowicz (2001) analyzed the spread of the Russian
crisis among selected CIS and CEE countries without formal
differentiation between contagion and spillovers. In the first
approach author employed a simple probit model testing
trade linkages between crisis-affected countries and Russia.
The crisis was defined with a binary variable (1 – crisis
occurred, 0 – no crisis occurred). Russia represented the
ground-zero country. Out of the 24 transition countries in
the sample, 6 were identified as crisis-hit countries, on the
basis of expert assessment (Belarus, Georgia, Moldova,
Kazakhstan, the Kyrgyz Republic, and Ukraine). Given the
numerous problems with data availability and reliability the
crisis variable was explained only with the use of trade
shares and a single macroeconomic variable.
The trade variable was constructed as the cumulative
share of exports to the ground-zero country and other countries that were previously hit by the crisis in total exports of
a country in 1997 (the last year before the Russian crisis). For
countries that did not experience the crisis it was the cumulative share of exports to all crisis-hit countries. Due to data
constraints the macroeconomic variable covered only the
ratio of total reserves minus gold (at the end of the third
quarter of 1998) to exports (for 1998 as a whole, fob).
Figure 7.1 specifies probabilities of a crisis in each analyzed country basing on estimation results provided by the
above-mentioned probit model. Given a cut-off point of
50%, the model failed to predict one crisis out of six, which
really did happen (the Kyrgyz Republic), and selected two
potential crisis episodes, which did not happen (Lithuania
and Tajikistan). What concerns the Kyrgyz Republic, the official trade statistics probably underestimated the actual share
of its export to Russia (including unregistered "shuttle"
export) and other crisis-hit countries28. Adopting a higher
cut-off point of 75% the model failed to predict two crises,
which really did happen but avoided issuing a false signal in
relation to crises, which did not happen.
Figure 7.2 shows a trade matrix between the crisiseffected countries what gives an additional, illustrative confirmation of an importance of the trade channel of crisis
transmission.
28
Uzbekistan, being an important trade partner of the Kyrgyz Republic, also experienced macroeconomic problems similar to currency crisis.
However, due to lack of credible statistics the country was dropped from the examined sample. In addition, Uzbekistan and Kazakhstan adopted
numerous trade restrictions in relation to Kyrgyz export. Kyrgyz export also suffered from a decline of gold prices in 1998.
40
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
Figure 7.1. Contagion crisis probability in transition countries basing on trade links – results of the probit model
Country
Ukraine
Moldova
Kyrgyz Rep.
Belarus
Georgia
Kazakhstan
Armenia
Azerbaijan
Bulgaria
Czech Rep.
Estonia
Hungary
Actual
crisis*
1
1
1
1
1
1
0
0
0
0
0
0
Fitted
Cut-off = Cut-off =
probability
0.50
0.75
0.9154
1
1
0.9474
1
1
0.1411
0
0
1.0000
1
1
0.5739
1
0
0.8924
1
1
0.0000
0
0
0.0011
0
0
0.0000
0
0
0.0002
0
0
0.3009
0
0
0.0018
0
0
Country
Latvia
Lithuania
Poland
Romania
Slovakia
Slovenia
Tajikistan
Turkey
Turkmenistan
Albania
Croatia
Macedonia
Actual
Fitted
Cut-off = Cut-off =
crisis* probability
0.50
0.75
0
0.1640
0
0
0
0.5277
1
0
0
0.0000
0
0
0
0.0109
0
0
0
0.0424
0
0
0
0.0013
0
0
0
0.5309
1
0
0
0.0000
0
0
0
0.0000
0
0
0
0.0000
0
0
0
0.0000
0
0
0
0.0656
0
0
Note: * binary variable: 1– presence of a crisis; 0 – no crisis.
Source: Rawdanowicz (2001).
Figure 7.2. Trade matrix of the crisis-affected countries in 1997 (% of total exports)
A\B
Russia
Ukraine
Moldova
Kyrgyz Rep.
Belarus
Georgia
Kazakhstan
Russia
8.5
0.4
0.2
5.4
0.2
2.9
Ukraine
26.2
2.1
0.0
5.8
0.3
0.7
Moldova
58.2
5.6
0.0
4.0
0.5
0.2
Kyrgyz Rep.
16.3
0.8
0.0
1.5
0.2
14.3
Belarus
64.5
5.9
1.3
0.1
0.0
0.7
Georgia
30.0
3.5
0.0
0.0
0.4
Kazakhstan
33.9
4.8
0.0
1.0
0.7
0.0
1.7
Note: % of country B's exports to country A in terms of country B's total exports.
Source: Rawdanowicz (2001).
In order to gain more insights into how the Russian crisis spread Rawdanowicz (2001) conducted additional backof-the-envelope calculations within the framework of a balance of payments model proposed by Masson (1999). This
model is capable of demonstrating how a large enough
shock to the current account can trigger a crisis if foreign
debt servicing exceeds a certain level. Borrowing costs
reflect expectations of crisis.
Results of these estimations indicated that most of the
countries under investigation (18 out of 24) had fundamentals
that were conducive to the outbreak of crisis at the end of
1997. Only two fell in the zone of multiple equilibrium (Bulgaria and Poland) and 5 featured "healthy" fundamentals (the
Czech Republic, Hungary, Russia, Slovenia, Turkmenistan).
Among the countries with a low probability of crisis, Russia
and Turkmenistan deserve closer examination. Russia, despite
its very low reserves (even by the standards of the sample
countries) and significant external debt, recorded a substantial
trade surplus of 6.2% of GDP. It should be noted, however,
that the high trade surplus in Russia was overestimated, as the
CASE Reports No. 51
deficit in shuttle trade was not included. Due to a decline in oil
and natural gas prices at the end of 1997 and 1998, Russian
trade and current account surplus rapidly disappeared in the
course of 1998. In addition, through all the decade of the 1990s
Russia suffered a negative capital account balance (massive capital flight). In the case of Turkmenistan, very high international
reserves (over 47% of GDP) made this country, according to
the model, resistant to balance of payments shocks.
The high probability of a crisis occurrence in many examined countries was caused by their high debt and trade deficit
ratios. If we assume the different level of sustainable trade
(current account) deficit in individual countries (see subsection
5.2) these results can be misleading. In fact, assessment of fundamentals quality should include a broader set of macroeconomic and institutional variables such as public debt, fiscal balance, current account balance, inflation, unemployment, the
structure of foreign capital flows, exchange rate regime.
Taking into account the above reservations Rawdanowicz (2001) made a synthesis of the results obtained in the
two above reported econometric analyzes (Figure 7.3).
41
M. D¹browski (ed.)
Figure 7.3. Contagion crisis probability in transition countries: comparison of results of the probit and balance of payments models
Crisis
Good fundamentals
Significant
trade linkages
Insignificant
trade linkages
Bad fundamentals
Belarus, Georgia,
Kazakhstan, Moldova,
Ukraine
Kyrgyz Republic
No crisis
Good fundamentals
Bad fundamentals
Lithuania, Tajikistan
Bulgaria*,
Czech Republic,
Hungary,
Poland*,
Slovenia
Albania, Armenia,
Azerbaijan, Croatia,
Estonia, Latvia,
Macedonia, Slovakia
Turkey, Turkmenistan
Notes: * – multiple equilibria.
Source: Rawdanowicz (2001).
Significance of trade linkages was determined on the basis of
the probit model whereas quality of fundamentals was
assessed using Masson (1999) balance-of payments model.
Most countries that did not experience a crisis (Albania,
Armenia, Azerbaijan, Croatia, Estonia, Latvia, Macedonia,
Slovakia, Turkey, Turkmenistan) had bad fundamentals
although there was insufficient crisis propagation to trigger
crises there (at least in trade linkages, as no formal inferences
about financial linkages could be drawn). However, as was
noted earlier, the criteria for "bad fundamentals" were too
"sensitive" and selected cases, which did not really belong to
this category. Thus, the group might incorporate countries
that either had bad fundamentals and no crisis propagation or
had relatively good fundamentals and crisis propagation (if
any) was not strong enough to trigger financial turmoil.
Five countries (Bulgaria, the Czech Republic, Hungary,
Poland, and Slovenia) proved to be in "healthy" condition
(Bulgaria and Poland had tendencies to multiple equilibria)
and had insignificant trade linkages. As these countries did
not experience a crisis, they probably did not experience
enough crisis propagation via financial channels.
Lithuania and Tajikistan turned out to be interesting
cases. These countries were not defined as crisis-hit though
both the probit model and balance of payments model indicate that they should have had crises. Although Tajikistan
was not chosen as a crisis-hit country it experienced depreciation of its ruble in November 1998. What concerns
Lithuania, 45% of its exports in 1997 went to crisis-hit
42
countries (mainly Russia and Belarus), its external debt
stood at 33.8 per cent of GDP (28.2 per cent excluding debt
to IFIs). The factor, which helped this country to retain good
reputation in financial investors’ eyes and resist speculative
pressure, was the currency board regime.
Peculiarities of the actual trade structure and trade problems of the Kyrgyz Republic have been explained earlier in
this section. In addition, financial channels should be also
taken into consideration. At the end of 1997, 18.7% of the
country’s external public debt was owed to CIS creditors on
a non-concessionaire basis and this was roughly equal to the
level of its official international reserves excluding gold.
Thus, the withdrawal of Russian and Kazakh creditors (see
IMF, 1999) could have impacted the som, especially in the
face of a very shallow exchange rate market. Moreover, the
high dollarization and low banking deposits indicated a lack
of domestic confidence to the Kyrgyz currency.
Summarizing the above results, one can conclude that all
the crisis-hit countries might be classified as having conditions conducive to crisis. They experienced not only the balance-of-payments fragility but also chronic fiscal problems
and numerous structural flaws (D¹browski, 1999). Thus,
there were not "innocent victims" in the sample, i.e. countries with good fundamentals still suffering crises from a
pure contagion effect. The shock propagation through both
trade and financial channels determined only the actual timing of the individual crises.
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
8. Economic and Social Consequences of Currency Crises
In the public discussion about currency crises and other
kinds of financial turbulence the majority of opinions point
at crises’ severe costs (see e.g. Stiglitz, 1998). According to
this dominant view, crises are unfavorable incidents and
should be avoided using all possible means. On the other
hand, one can expect that crises, punishing evident cases of
economic mismanagement, could have disciplining effect on
governments (and indirectly on their electorate), push necessary reforms (Rodrik, 1996), and may automatically correct imbalances created by politicians29. The comparison
with a mechanism of bankruptcy on the micro-level seems
to be a good parallel here.
Among the countries analyzed, Bulgaria seems to be the
best case of such a positive self-correcting mechanism,
involving a change of government, comprehensive package
of economic reforms and introducing a currency board just
after the 1996–1997 crisis (Ganev, 2001). Other such positive examples involve Mexico (Paczyñski, 2001) and Thailand (Antczak M., 2001).
However, it is not easy to find a more comprehensive
and balanced picture of potential and actual crises consequences in the existing crisis literature. First, most of theoretical models and empirical researches concentrate on factors causing currency crises rather than on their effects.
Among the latter the attention is put on the immediate negative consequences such as output decline, income contraction, higher unemployment, fiscal costs of restructuring of
the financial sector (e.g. WEO, 1998; Milesi-Ferretti and
Razin, 1998; D¹browski, 1999) rather than on medium and
longer term policy implications.
Generally, one can distinguish the following categories of
crisis costs (B³aszkiewicz and Paczyñski, 2001): (i) fiscal and
quasi-fiscal costs resulting from increased burden of public
debt service (due to devaluation and higher interest rates)
and necessity to restructure financial institutions and some
29
big corporations; (ii) costs related to lost economic growth;
(iii) social costs connected with unemployment, decline in
real incomes, worsened health and education situation, etc.,
leading to bigger poverty; and (iv) political costs. Most of
these issues are difficult to measure, especially in international comparisons.
However, B³aszkiewicz and Paczyñski (2001) attempted
a series of comparative analyzes of various crisis effects with
special attention given to transition economies of Eastern
Europe and the former USSR. They tried to get a more
comprehensive picture of the crisis consequences, including
policy changes. The whole investigated sample (ALL) covered 41 countries, and some regional sub-samples were
also identified. Transition economies (TR) group contained
23 countries, of which 6 countries (FSU 98 – Belarus, Georgia, Kazakhstan, Kyrgyz Republic, Russia and Ukraine) were
considered as those affected by the Russian 1998 crisis and
its contagion effects. Southeast Asian countries, which
underwent 1997 series of currency crises (Indonesia, Korea,
Malaysia, Philippines and Thailand) were nicknamed
ASIA97. Finally Latin American sub-sample (LAM) included
Argentina, Bolivia, Brazil, Chile, Mexico, and Venezuela.
In the first approach, B³aszkiewicz and Paczyñski (2001)
analyzed the behavior of several macroeconomic variables
before and after crises and described both their positive and
negative consequences.
Looking at median GDP growth rate (see Figure 8.1)
for the entire investigated sample (ALL), the economic stagnation around a crisis year is clear. The signs of the slowdown could be observed two years before a crisis with the
1.5% recession in the crisis year. However, from there on
economies tend to grow faster than before.
Transition economies already suffered from recession
three years before a crisis but it was connected, at least
partly, with the early transition output decline caused main-
Additionally, crises brought the experience to the international financial organizations (Kohler, 2001).
CASE Reports No. 51
43
M. D¹browski (ed.)
Figure 8.1. Real GDP growth rate before and after the crisis (median)
10
8
6
4
2
0
-2
-4
-6
-8
-10
c-3
c-2
c-1
crisis
c+1
c+2
c+3
year
ALL
ASIA97
TR
LAM
FSU98
Source: B³aszkiewicz and Paczyñski (2001).
ly by structural reasons. Then the situation improved a bit
but a crisis worsened it again and the negative growth rate
dropped to -6.5%. This rate returned to its pre-crisis level
a year after the crisis and economic growth resumed
thereafter.
The same pattern can be identified for the FSU98 subgroup. The 1998–1999 crises occurred exactly when these
economies started to recover after several years of early
transition output decline. Consequently, the impact of a
transition itself was perhaps more pronounced than the
impact of the crises. However, the crisis brought a major
setback to a weak recovery that started to be observed in
FSU countries around 1997. Three years after the crises
growth rates remained positive staying in the moderate
range of 3–4%30.
In South-East Asia, the 1997 crisis brought the median
rate of output growth down from more than 8% in 1994 to
an estimated 4.8% in 2001. In terms of the scale of growth
contraction, the ASIA97 group stood out in the entire examined sample (with the recession of 7.4% a year after the
shock). However, this group was also exceptional in
terms of its pre-crisis results with growth rates of around
8% annually.
The economic growth in Latin America started slowing
down earlier with the sharpest drop a year before the crisis.
However, two years after crises the GDP growth rate surpassed its pre-crisis high.
30
44
While disaggregating GDP median growth rates by
demand components B³aszkiewicz and Paczyñski (2001)
found that investments and imports contracted the most
sharply. The former suffered more seriously and for longer
periods with ASIA97 recording the highest and longest
investment contraction and LAM group – the most limited
and shortest one.
Estimation of the potential output loss caused by crises
was the next step done by B³aszkiewicz and Paczyñski
(2001). There are, however, methodological problems connected with such a simulation. It is not clear whether the
growth path before the outburst of a crisis was sustainable.
This problem can be handled by adding some subjective
corrections to the average growth rates before crisis
episodes (used as an approximation of a trend).
The obtained results (Figure 8.2) do not differ significantly from those obtained for a larger sample of countries
in WEO (1998). There were, however, large differences
between regional sub-samples. On one side, LAM countries
turn out to be relatively weakly affected in terms of lost
GDP growth. On the end, ASIA97 group suffered massive
GDP losses what could be explained by high growth rates
before 1997 (and, even after some downward corrections,
high growth trend). Transition economies (TR) and in particular the FSU98 group recovered relatively quickly and
with only limited output losses.
The last year in the crisis window for the FSU98 group was the IMF forecast from May 2001.
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
Figure 8.2. Costs of crises in terms of lost output relative to trend
Average recovery time
WEO (1998) sample
ALL sample (Blaszkiewicz & Paczynski, 2001)
TR sub-sample
FSU98 sub-sample
ASIA97 sub-sample
LAM sub-sample
1.6 E(1.5)
1.4
1.1
1.1
2.4
0.8
Cumulative loss of
output (% of GDP)
4.3 E(4.8)
6.4
4.8
2.9
16.9
1.8
Crises with output
losses (%)
61 E(64)
71
69
71
100
50
Notes: In the WEO (1998) sample the numbers in parentheses refer to emerging economies group. Average recovery time is calculated as the
average time of returning to the trend growth path. Cumulative lost of output is calculated as a sum of differences between observed and trend growth
figures, until an economy returns to trend growth path. Last column shows the percentage of countries that experienced output loss after a financial
crisis.
Source: B³aszkiewicz and Paczyñski (2001).
Figure 8.3. Net capital inflow before and after the crisis
350
300
250
200
150
100
50
0
-50
c-4
c-3
c-2
c-1
median for crisis window
crisis
quarter
c+1
c+2
c+3
c+4
median for Q194-Q100; all countries
c+5
c+6
2Q MA
Source: B³aszkiewicz and Paczyñski (2001).
All the investigated countries suffered from persistent
current account imbalances before the crisis. This related
particularly to transition economies where the median current account deficit was equal to 7.4% of GDP. However,
the impact of currency crises on trade balance occurred to
be limited. Even though it started to improve after devaluations, it turned positive only in Asia and Latin America. In
the third year after the crises, the trade balance for the
whole sample deteriorated again. In Latin America imports
outperformed exports already in the second year after the
crisis. Trade balance in transition economies improved only
for one year after devaluation. The only "textbook case"
scenario in which currency devaluation improved trade
competitiveness was observed in the Asia97 group.
The pattern for CA recovery was similar – it is only in
Asia where crises brought the CA into a surplus position. In
CASE Reports No. 51
other groups, current account deficit shrank only during the
first year after a crisis, but widened thereafter.
The above results confirm the well-known observation
that devaluation brings usually the short-term windfall gains
only, while causing many other adverse effects such as higher inflation (see below).
Crisis-generated changes in the current account balances must be reflected on the capital account (and, in fact,
these are often developments on capital account side,
which trigger dramatic shifts in a current account). Up to
one year before a crisis, countries in the examined sample
were experiencing a median net capital inflow of more than
US$300 million per quarter. This figure decreased significantly in the quarters preceding the crisis, turning negative
in the first quarter after the crisis episode (see Figure 8.3).
Then, net capital inflows remained repressed for at least
45
M. D¹browski (ed.)
Figure 8.4. CPI, y-o-y percentage change before and after the crisis; monthly data (medians for subgroups)
100
90
80
70
60
50
40
30
20
10
0
c-24
c-20
c-16
c-12
All
c-8
c-4
Trans
crisis
c+4
FSU98
c+8
c+12
ASIA97
c+16
c+20
c+24
LAM
Source: B³aszkiewicz and Paczyñski (2001).
next two years, staying below the pre-crises levels and
below the median value for the whole sample in the
1Q94–1Q00. The same held true for portfolio and foreign
direct investments. Not surprisingly the drop was sharper
for short-term flows. Three years after the crisis FDI and
portfolio investments still remained substantially lower than
before.
In the period before a crisis a clear disinflation trend
could be observed and inflation was generally low, staying at
single digit levels. A crisis represented a dramatic change in
the trend with CPI rising on average (median of the sample)
by more than 40% in 12 months after the crisis. Figure 8.4
suggests that on average it took at least another year until
12-month rate of CPI growth returned to a single digit level.
In transition economies, currency crises had the
strongest inflationary impact what was probably connected
with their fresh inflationary history, low monetization level
and a relatively high level of currency substitution. On the
other hend, Asian countries experienced the very moderate
inflationary consequences, with median CPI rising from
around 4% before the crisis to some 10% a year later. Aziz,
Caramazza, and Salgado (2000) obtained similar results.
A drop of real wages was visible only in the first year
after a crisis. In the second year wages tended to rebound
strongly. However, the downward impact on real incomes
was possibly stronger, due to increased unemployment (see
Figure 8.5). With the exception of Latin American countries, unemployment started to increase already a year
before a crisis and continued a strong upward trend after-
46
wards. This could suggest that devaluation did not help in
protecting employment.
Finally, B³aszkiewicz and Paczyñski (2001) tried to investigate to what extent currency crises helped to correct previous economic policies and whether these corrections
were sustainable. Due to limited cross-country data availability they could test only one proxy indicator – the behavior of the current account before and after the crisis. The
current account balance relative to GDP (in percent) was
studied in three years preceding and three years following a
crisis year. A "problem" with current account imbalance was
defined as deficit larger than 4% of GDP in any two of the
three preceding years or deficit of more than 6% of GDP in
a year before a crisis. A current account "problem" after a
crisis was defined in a similar way: deficit larger than 4% of
GDP in any two of the three following years or deficit larger than 6% of GDP in the third year after a crisis.
Such a threshold allowed for identifying 13 crises where
current account was potentially a problem beforehand. Of
these, in 9 cases the situation improved and current account
deficits were reduced (or turned into surpluses) during
three years following crises, while in 4 cases there was no
much improvement. Additionally, in 5 cases, the current
account deficits widened significantly after financial turbulence, while no problems were indicated before a crisis. In
addition, the data were not always available for all three
years after crises, so the statistics might actually look even
worse (i.e. crises tend to bring even less improvement to
CA balances). The general conclusion from that exercise
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
Figure 8.5. Unemployment rate before and after the crisis
10
8
6
4
2
0
c-3
c-2
All
c-1
crisis
FSU98
c+1
ASIA97
c+2
c+3
LAM
Note: Presented figures are sample medians.
Source: B³aszkiewicz and Paczyñski (2001).
was that there was no common pattern in the sample – in
some countries previous imbalances were removed by a
crisis, while in the others they remained in place or even
newly emerged.
Generally, B³aszkiewicz and Paczyñski (2001) study
does not provide us with the clear answers related to currency crisis consequences. Obviously, the economic and
social costs of the crises estimated in terms of lost output,
higher inflation, higher unemployment, lower real wages,
and negative fiscal implications of higher debt burden and
CASE Reports No. 51
financial sector restructuring are substantial and painful.
The question, as to whether crises can bring catharsis to
economies in terms of better economic policy and removing previous imbalances cannot be answered in a univocal
way. There are examples of positive changes but they
relate only to a part of crisis episodes. Perhaps the fear of
crisis can be considered as the most important but indirect
(and rather immeasurable) effect disciplining governments
and central banks.
47
M. D¹browski (ed.)
9. Crisis Prevention
Any attempt to give a comprehensive answer how to
avoid currency crises would have involve discussing a broad
spectrum of issues related to both macroeconomic policies
and structural/institutional measures, going well beyond this
paper’s agenda. To give a good example, Mishkin (2001, p. 13)
paper on the prevention of financial crises in emerging market countries analyzes 12 rather broad areas of policy
reforms: (1) prudential supervision; (2) accounting and disclosure requirements; (3) legal and judicial systems; (4) market based discipline; (5) entry of foreign banks; (6) capital
controls; (7) reduction of the role of state-owned financial
institutions; (8) restrictions on foreign denominated debt;
(9) elimination of the "too-big-to-fail" principle in the corporate sector; (10) sequencing financial liberalization; (11)
monetary policy and price stability; and (12) exchange rate
regimes and foreign exchange reserves. This list is clearly
not complete as Mishkin mainly focused on the financial
market turbulence. One can add here the fiscal policy and
fiscal management, privatization of non-financial enterprises,
trade policy or broad range of political/institutional factors.
In order to limit our analysis to reasonable size and avoid
repeating observations and conclusions formulated in the
diagnostic part of this paper (sections 4-7), we will concentrate below on two specific questions: (i) the role foreign
exchange reserves (international liquidity); and (ii) the role
of the IMF programs in preventing currency crises.
9.1. The Role of International Liquidity in
Preventing Currency Crises
Many economists (e.g. Feldstein, 1999; Mishkin, 1999;
Radelet and Sachs, 1998) call for higher foreign exchange
reserves holdings as a measure helping to prevent currency
31
48
crises31. At first glance, this proposal seems to make sense:
even with less-than perfect economic policy, country could
survive any speculative attack, provided that its central bank
has enough foreign exchange reserves. What is more, if the
reserves are high enough, the attack (bound to fail) will
never happen, according to first- and second-generation crisis models (see section 3). In addition, numerous empirical
studies suggest that insufficient international liquidity was a
good predictor of the recent crises (see e.g. Radelet and
Sachs, 1998, Tornell, 1999, Bussière and Mulder, 1999).
This proposal requires, however, a serious theoretical
and empirical assessment, taking into consideration both
benefits and costs of maintaining high liquidity. First, one
must answer the question what is the sufficient level of the
international reserves, which gives a full guarantee that any
speculative attack will be resisted. Theoretically, the minimum "safe" level of central bank’s international reserves
under a fixed exchange regime is equal to its high-powered
money (and this is a standard norm under the currency
board regime). However, if a central bank wants to act as
the "lender of last resort" the level of foreign exchange
reserves’ backing should be higher, somewhere between
the equivalent of monetary base and broad money (probably at least the equivalent of M1, to be able to stop a bank
run). If a central bank wanted to sterilize its foreign
exchange interventions to avoid the liquidity squeeze in the
economy, any level of international reserves would not be
sufficient enough to provide a full anti-crisis assurance.
Second, keeping international reserves is costly for a
central bank, and consequently for a state budget, which
gets most of central bank profit. These costs come from the
interest rate differences, which can have various forms in
practice. For example, if central bank (or government) must
borrow abroad, in order to increase its official reserves, this
is the difference between the borrowing costs and foreign
The entire subsection 9.1 heavily draws from Szczurek (2001).
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
Figure 9.1. Evaluating optimal international liquidity – marginal cost and benefit
US$m
400
350
300
250
200
Cost
150
100
50
Benefit
1
2
3
4
5
Reserves/GDP
Source: Szczurek (2001).
interest rate income, which central bank can gain from prudent location of its international reserves (these are usually
US Treasury bonds and other similar instruments). If accumulation of international reserves pushes the central bank
into the structural surplus position, this is the difference
between interest rate on domestic currency denominated
money market instruments and the above mentioned interest rate on reserves investment.
Szczurek (2001) attempted to estimate the optimal level
of foreign exchange reserves using a simple policy optimization model. As we see from Figure 9.1 there are two possible equilibria. One is at international liquidity equal to
zero. Increasing liquidity costs more than it brings because
fundamentals are too bad for a slight improvement of liquidity to change the probability of a crisis much. The second
equilibrium is in the point where downward sloping marginal benefit and marginal cost curves cross each other. It is
clearly the optimal level of liquidity: the total gain (which is
the surface below the marginal benefit curve) exceeds the
total cost.
Szczurek (2001) used the above model for two applications. First, he tried to estimate what is the optimal holding
of international liquidity. The second application involved
finding out the perceived cost of the currency crisis to policy-makers.
Obviously, a crisis results in certain costs both to the
economy and the policy makers. What concerns the economy it can be observed empirically as, for instance, the
deviation of the post-crisis GDP growth from its long-term
CASE Reports No. 51
trend (see section 8). The second cost element relates to
reputation losses of policy makers themselves, and probably
is the function of the degree of the rigidity of the foreign
exchange regime, and the length for which the regime was
maintained, past inflation experience, but also personality of
the central banker, etc.
By assuming that the analyzed countries hold optimal
(from the point of view of the policy maker) international
liquidity, it was possible to estimate "how much the crisis is
feared", or the total cost of the currency crisis (including the
reputation cost), as viewed by the policymaker.
The sample covered 33 developing and transition
economies participating in the global financial market. The
overall results proved that international liquidity did matter
very much in averting currency crises. The results for the
average (from the pool) country are shown in Figure 9.2.
Vertical scale represents optimal liquidity (as a
reserves/short-term debt ratio), corresponding to the
assumed cost of the currency crisis (as a percentage of GDP).
The results indicate that the IMF’s recommendation of
keeping foreign exchange reserves’ stock equal to the
short-term foreign debt seems to be insufficient. Assuming the cost of the crisis at just 1% of GDP, optimal international liquidity is 1.6 times the short-term foreign debt.
If we believe the crises are more costly than 1% of GDP,
the reserves held will be a multiple of the short-term foreign debt.
How much the policymakers fear the crisis? Figure 9.3
shows quite a wide disparity of implicit currency crisis cost
49
M. D¹browski (ed.)
Figure 9.2. Optimal liquidity holding versus cost of the crisis
LLBIS
5
4
somewhat surprising (as the peg exchange rate regimes
should involve higher reputation costs of currency depreciation): apart from Malaysia, the first nine countries on the
list had a floating or managed floating exchange rate.
9.2 The Role of the IMF in Preventing
Currency Crises
3
2
1
2
4
6
8
10
% GDP
Source: Szczurek (2001).
Figure 9.3. Crisis cost to the policy maker, as of end-99
Country
MALAYSIA
CZECH REPUBLIC
POLAND
CHILE
SLOVAK REPUBLIC
KOREA
PHILIPPINES
SOUTH AFRICA
RUSSIA
ARGENTINA
COSTA RICA
ECUADOR
CHINA,P.R.: MAINLAND
COLOMBIA
BRAZIL
ROMANIA
PAKISTAN
DOMINICAN REPUBLIC
% of GDP
12.1%
5.0%
3.3%
2.7%
2.1%
2.0%
1.8%
1.5%
1.4%
1.3%
1.2%
0.9%
0.9%
0.8%
0.8%
0.4%
0.4%
0.3%
Source: Szczurek (2001).
for the policy makers: from 12.1% of GDP (in Malaysia) to
0.3% (in the Dominican Republic). The results seem to have
little to do with the exchange rate arrangements, which is
As the basic statutory mission of the International Monetary Fund (IMF) is to deal with balance-of-payments disproportions it is reasonable to ask what has been the role of
this organization in preventing currency crises of the last
decade (or resolving them when already happened). This
question is particularly important as many well-known economists question effectiveness of the IMF in the recent crisis
episodes. Meltzer (1998) critique of the IMF and US Treasury role before and during the Mexican crisis, and Radelet
and Sachs (1998) critique on the IMF role in Thailand and
Indonesia are just two examples of a large body of such
opinions.
Antczak, Markiewicz and Radziwi³³ (2001) investigated
the role of the IMF in preventing the 1998–1999 series of
currency crises in five CIS countries – Russia, Ukraine,
Moldova, Georgia and Kyrgyzstan (RUMGK). One can distinguish two stages of the IMF cooperation with the CIS countries. In the first stage covering the period of 1992–1995 the
IMF offered mainly the short term programs – the Structural Transformation Facility (STF) and the Stand-by arrangements (SBA). The STF financial line was designed especially
for transition economies at the beginning of 1993, and
involved rather limited and weak conditionality (D¹browski,
1998). It aimed to provide less advanced transition
economies (mainly in the FSU) with the limited financial support and prepare them to receive a standard Fund aid in
future. The second stage of the IMF assistance (1995–1998)
was dominated by the medium-term Extended Fund Facility
(EFF) and Enhanced Structural Adjustment Facilities (ESAF)32
containing a much more comprehensive conditionality. Generally, the size of net IMF financing was substantial and picked
up before the crisis (1997–1998) but was also volatile at least
in the case of Russia, Ukraine and Moldova.
32
EFF is a standard medium term program while ESAF represents a special type of the medium term financial aid provided for low and lowermedium income countries, on concessionaire terms. In the investigated sample Kyrgyzstan and Georgia were eligible to receive this type of assistance.
After the crisis Moldova became qualified together with Georgia and Kyrgyzstan to receive the Poverty Reduction and Growth Facility (PRGF), the successor of ESAF.
50
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
As deep fiscal imbalances constituted the most important cause of the CIS currency crises in 1998–1999 (see section 5.1 and Siwiñska (2000), Antczak R. (2001), Markiewicz
(2001), Radziwi³³ (2001)), the IMF attitude to this problem
seems to be a crucial indicator of the relevance and effectiveness of its adjustment programs. Unfortunately, the IMF
record in this sphere was far from being satisfactory. The
actual fiscal balances occurred to be always worse than
those estimated by the IMF as sustainable even under very
optimistic assumptions.
There were several reasons of this failure both on the
IMF and on client-countries’ side. First, forecasts of GDP
growth rates were systematically overestimated, particularly in the case of Russia, Ukraine and Moldova (Antczak,
Markiewicz and Radziwi³³, 2000). The excessive optimism
related to perspectives of overcoming the transformation
output decline and positive growth effects of the proposed
reform measures might serve as an explanation in the beginning of transition only. Later it was rather the conscious mistake repeated systematically both by governments of the
analyzed countries and the IMF. The governments had the
strong political reason to do it as presenting an optimistic
growth prospects could soften the resistance against
unpopular adjustment measures and increase a room of
maneuver in fiscal planning. The IMF permanently accused
by the populist-minded politicians and economists of being
too tough and imposing unnecessary austerity on poor
countries, increasingly wanted to present its actions as
growth oriented and focusing on poverty alleviation33. Also
very formally, medium term programs such as the EFF or
ESAF had to contain economic growth objectives in order
to be accepted by the IMF Executive Board. Hence, wishful
thinking was often taken as serious forecast.
The false growth assumptions had an important impact
on the sustainability of the programs. Prospects of high
growth rates limited the pressure on fiscal adjustments as
projected debt accumulation did not significantly exceed a
forecast real GDP growth and was, therefore, not seen as
an important problem.
In addition, fiscal performance criteria were calculated
initially in the IMF programs exclusively on a cash basis while
arrears became the important and persistent source of
financing budget deficit (calculated on the accrual basis)34.
The failure of governments to collect projected revenues
and execute planned expenditures also pushed them
towards dubious non-cash and other quasi-fiscal operations
that both decreased the efficiency of fiscal management and
seriously distorted economic life. The IMF recognized these
distortive practices as a serious problem relatively late. As
result, stopping arrears and non-cash settlements was
added to the conditionality lists.
Figure 9.4. Compliance with the IMF quantitative performance
criteria
Country
Program
Compliance
(full = 100 )
Russia
Russia
Ukraine
Ukraine
Ukraine
Moldova
Moldova
Moldova
Georgia
Georgia
Kyrgyzstan
Kyrgyzstan
SBA (1995)
EFF (1996)
SBA (1995)
SBA (1996)
SBA (1997)
SBA (1993)
SBA (1995)
EFF (1996)
SBA (1995)
ESAF (1996)
SBA (1993)
ESAF (1994)
100
77
(off track) 76
99
(off track) 62
86
100
82
100
89
(off track) 70
86
Note: The index is calculated as the simple average of compliance on
each performance criteria at each test date. Compliance on a given condition at a given date is evaluated using the following scale: met=100,
waived=50, met after modification=50, waived after modification=30,
not met=0.
Source: Antczak, Markiewicz and Radziwi³³ (2001) on the basis of
Mercer-Blackman and Unigovskaya (2000).
Finally, the IMF presented a substantial degree of tolerance in relation to countries’ expansionary fiscal policies and
their failures to meet the fiscal targets agreed under the
Fund supported programs. Formally, the IMF did express its
concern about the fiscal imbalances in all their programs and
Article IV consultations. Additionally, the large and increasing share of the overall number of structural conditionality
benchmarks was related to the reform of the fiscal sector.
However, this effort was relatively poorly reflected in
the performance criteria, which were relatively lenient and
exhibited poor conditionality. Their implementation went
even worse. Nevertheless, the IMF continued to provide
support to countries with very high fiscal imbalances, often
praising them for the progress towards the market economy, stabilization, and long-term growth. Even if fiscal slip-
33
The earlier mentioned renaming ESAF into PRGF is one of the examples of such a PR-type window dressing.
Accordingly, Buiter (1997) stated that the cash deficit indicators used widely by the IMF in its programs were "myopic" and "more than useless"
in the evaluation and design of macroeconomic policy packages.
34
CASE Reports No. 51
51
M. D¹browski (ed.)
Figure 9.5. Compliance with IMF structural conditionality
Compliance
Number of structural benchmarks
(full = 100 )
Public
Fiscal
Finan- Privati- Other Total
Trade/ Pricing
Enterzation
sector
cial
Exchaprise
nge
sector
System
Russia
SBA
3
1
2
6
50
Russia
EFF
2
1
1
18
7
6
2
37
73
Ukraine
SBA
2
1
1
5
3
12
68
Ukraine
SBA
1
2
2
1
2
1
2
11
83
Moldova
SBA
4
1
3
2
3
13
81
Moldova
SBA
2
1
2
5
75
Moldova
EFF
6
1
1
2
1
4
1
16
90
Georgia
SBA
1
3
8
2
3
17
77
Georgia
ESAF
2
1
5
5
4
5
22
79
Kyrgyzstan
SBA
4
1
1
6
0
Kyrgyzstan
ESAF
1
7
10
8
4
5
35
79
Note: The index of compliance (last column) is calculated as the simple average of compliance on each structural benchmark at each test date.
Compliance on a given benchmark at a given date is evaluated using the following scale: met=100, met to certain extent or with insignificant delay=50,
insufficient information about outcome=50, not met=0.
Source: Antczak, Markiewicz and Radziwi³³ (2001) on the basis of Mercer-Blackman and Unigovskaya (2000).
Country
Program
pages led to a program’s going off-track, new programs
were granted almost immediately.
Figure 9.4 presents the level of compliance with the fiscal and monetary performance criteria during implementation of the SBA, EFF and ESAF programs in RUMGK in the
period of 1994–1998 using methodology of Mercer-Blackman and Unigovskaya (2000). The presented picture is not
very optimistic.
The level of compliance with structural and institutional
benchmarks seems to be higher than that of quantitative criteria (Figure 9.5) although one must take into account
methodological difficulties with measuring the qualitative
conditionality and its execution. Some of these benchmarks
could be met only formally, for example, through preparing
draft law or even its approval without real implementation.
On the other hand, there is a lot of evidence that a chronic
fiscal crisis and financial sector fragility in the CIS countries
have very strong structural roots. It can imply that either the
methodology proposed by Mercer and Unigovskaya (2000)
provides too optimistic results or the IMF conditionality in
this sphere has been generally weak.
The official evaluation of the IMF programs in RUMGK
done by the Policy Development and Review Department of
the IMF (2001a; 2001b) indicates rather high compliance
with the adopted conditionality. However, Antczak,
Markiewicz and Radziwi³³ (2001) question this opinion, particularly in relation to Russia. According to their opinion,
from the very beginning of a transition process the IMF was
not insistent enough on its conditionality, especially in the
area of fiscal adjustment. They also claim that the actual con-
52
ditionality was effectively much weaker than it was suggested by the relatively high scores on compliance presented by
IMF sources. This inconsistency stemmed, among others,
from several techniques of circumventing the imprecise and
nontransparent conditionality invented by the client countries and tolerated by the Fund.
Summing up, Antczak, Markiewicz and Radziwi³³ (2001)
express their opinion that the IMF conditionality in relation
to RUMGK exhibited excessive leniency, allowing them to
avoid fiscal and other kinds of necessary adjustment. While
flexibility was sometimes necessary, it was clearly abused in
the analyzed case. The last conclusion is also relevant to
many other countries and regions. For example, Brazil had
eight separate stand-by programs between 1965 and 1972,
Peru had 17 different arrangements between 1971 and 1977
(McQuillan, 1998) and Turkey – 16 unsuccessful programs
between 1961 and 1999 (Sasin, 2001d).
In addition to weak and inconsistently executed conditionality of its programs, the IMF was softening governments’ budget constraints in RUMGK by providing nonmonetary sources of deficit financing. As a result, the political support for fiscal tightening was even more difficult to
generate than it would have been the case of the absence of
IMF programs (Antczak, Markiewicz and Radziwi³³, 2001).
The support of the IMF for the development of nonmonetary deficit financing in RUMGK was twofold. First, the
IMF indirectly financed CIS governments, accepting the central bank credit to government backed by the increase in
gross international reserves coming from the IMF loans.
Second, agreements with the IMF also allowed for negotia-
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
tions with other creditors on rescheduling of debt payments. Russia, as the legal successor of the Soviet debt,
received the largest rescheduling of debt payments from
the Paris Club.
The unjustified "seal of good housekeeping" further softened budget constraints, opening the door to cooperation
with other multilateral and bilateral donors. IMF programs
constituted a stamp of approval, helping to upgrade countries’ ratings and new capital inflows. This catalytic role is
explicitly recognized as one of the main functions of the IMF
in support of adjustment. However, when the IMF underwrites unsustainable policies, the effect is disastrous for the
borrowing country: external debt accumulates and incentives for adjustments are diminished. Wrong signals sent to
private creditors and governments had a detrimental impact
on these countries. Policies based on increasing debt-toGDP ratios were unsustainable, but could be maintained as
long as financial markets stayed not fully aware of this.
Worse, a general sense of implicit guarantees ("too big
to fail"), especially in the case of Russia was built up over the
course of years. While the role of IMF in the modern world
should be the prevention of crises through the surveillance
of national policies, transparent information and reduction
of moral hazard, the practice was exactly opposite. Meltzer
(1999) concludes: "Moral hazard lending to Russia, encouraged by the bail out of foreign lenders to Mexico, permitted
Russia and other countries to finance large unbalanced budgets
by borrowing externally. The result is a much larger financial
problem for international lenders and for the economies of
other countries".
One can ask what were the sources of such serious
flaws of the IMF programs in CIS countries and (probably)
in many other regions, particularly if one takes into account
the enormous professional expertise and experience of this
organization?
If we try to answer this question looking from the perspective of the client country, the issue of the so-called
reform ownership plays a crucial role. The most fundamental problems can be solved only if national authorities with
broad a political support assume the ownership and full
responsibility for reforms and necessary policy corrections
CASE Reports No. 51
as it happened in most of Central European and Baltic countries. Unfortunately, in countries of weak reform ownership
(most of the CIS), policies were assumed (and reluctantly
followed) just to please the IMF and receive disbursements,
rather than to solve the problems of the country.
But even in such cases, the programs could induce better policies, through enhancing the credibility of reforms
and helping reformist governments to overcome political
opposition to the program. But this is possible only if the
IMF program provides a binding commitment. With soft
conditionality and soft financing this goal cannot be reached.
So why the IMF is soft and, instead of promoting good policies in a decisive manner, it "pays due regard to the domestic
social and political objectives"(IMF, 2001a), even if these
objectives are detrimental to long-term growth and stabilization?
The answer has most likely of a political nature. The IMF
is the international public institution dependent on their
shareholders, i.e. member countries. This means that it cannot be too tough in relation to its shareholders, particularly
towards big and politically influential countries like Russia,
Brazil, Mexico, Indonesia or Korea, to give just a few examples from the last decade. On the other hand, the biggest
shareholders, G-7 countries use the IMF (and other development organization like the World Bank or regional development banks) as the flexible instruments of their foreign
policies and foreign aid. Facing their own domestic fiscal
constraints it is much easier to use such the specific "extrabudgetary fund" like the IMF than ask parliament to approve
the explicit budget expenditure lines related to foreign aid
(see Meltzer, 1999, D¹browski, 1998).
Another problem is related to the system of incentives
facing the IMF bureaucracy (see Vaubel, 1991). The IMF as
an institution has a stake in the "success" of the program. It
is, therefore, difficult for its officials and staff to declare the
program failure, even if it cannot impose its real implementation. Finally, the IMF wants to stay in the country, as it considers some influence on policy ("to have the seat at the
table") to be better than no influence at all. This is, however, not a good solution if it leads the Fund to underwrite bad
policies.
53
M. D¹browski (ed.)
10. Research Conclusions and Policy Recommendations
Currency crises and other forms of financial crises have
been present in the economic history for many centuries.
However, their frequency and specific forms changed over
time as a result of economic development and evolution of
currency regimes. In the second half of the 20th century,
rapid expansion of a number of the nationally managed independent currencies led to a higher frequency of crisis
episodes. Progressing integration and increasing sophistication of the product and financial markets brought new forms
and more global character of the crises events (contagion
effect) in the last decade.
Currency crises became a very popular topic of academic and political debate with hundreds of conferences and
seminars and thousands of publications in recent years.
Unfortunately, not in all cases quantity was transformed into
quality. Many analyzes have a very fragmental and superficial
character and offers unclear or even misleading conclusions.
Confusion starts with terminology. Notion of a crisis is
often used in the highly imprecise and too extensive way.
Hence, the first research task was to bring some order to
the terminology and propose the maximally precise and
empirically operational definition of a currency crisis. The
conclusion is that definition of Eichengreen, Rose and
Wyplosz (1994) helps to select the empirical episodes most
closely fitting with intuition understanding of what a currency crisis is (a sudden decline in confidence to a specific currency). However, a certain dose of arbitrary expert assessment seems to be still necessary as the additional selection
tool. Otherwise, any formalized definition may always select
the cases, which can hardly be considered as the real currency crises and omit the evident crisis episodes.
The similar problems concern diagnosis of what leads to
currency crises. The most frequent cases of misunderstanding are connected with taking the symptoms of already
existing imbalances and distortions (usually expressed in the
form of the so-called early warning indicators) as original
causes of the problem. The latter can be summarized as
54
excessive expansion and over-borrowing of the public and
private sectors on the one hand, and inconsistent and nontransparent economic policies on the other. Over-expansion
and over-borrowing manifest themselves in the excessive
(unsustainable) current account deficit, currency overvaluation, increasing debt burden, insufficient international
reserves, and deterioration of many other frequently analyzed indicators. Inconsistent policies increase market
uncertainty and country risk premium, contribute to shortening of the lending horizon, and can trigger speculative
attacks against currencies. The frequently discussed role of
pegged (fixed) exchange rate regimes should be seen precisely in this context. An attempt to simultaneously control
exchange rate and domestic liquidity in the world of free
capital movement (and currency substitution) violates the
principle of the "impossible trinity" and can be consider by
the market players as a signal of major policy inconsistency.
After a crisis already hit a country, the ability to run a
consistent and credible economic policy is crucial for limiting its size, length and negative consequences. One of the
key questions that authorities face is how to readjust an
exchange rate regime, that is usually the first institutional
victim of the successful speculative attack.
One could try to go further by asking what are the
causes of over-expansion/over-borrowing and policy
inconsistencies? The first attempt at an answer could simply point at bad economic policies. This, however, seems
to be too easy and too superficial. To get a deeper diagnosis one must analyze the broad set of political and institutional variables such as the electoral and government systems, federalism, constitutional protection of public
finance stability, central bank’s independence, transparency of public finances and government policies, external
constraints coming from international treaties, and many
others. The experience of last decade shows that all these
parameters are extremely important and badly need further empirical investigation.
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
The above mentioned "internationalization" of crisis
episodes also contributed to research and analytical confusion expressed in a certain diagnostic fatalism and lack of
conviction that international financial markets can behave in
rational and predictable way. This triggered a lot of antiglobalization thinking, proposals to reintroduce capital controls, or subordinate international capital markets to strong
regulatory control (what would probably be helpful but, at
least at present, remains politically unrealistic). True, international financial markets presented a lot of panic reactions
in the aftermath of the Asian and Russian crises. However,
later on, the evident learning process started and the recent
reactions in the case of individual crises (for instance, in
Turkey or Argentine) are much more selective and countryspecific. In addition, the size of capital inflow to emerging
markets moderated after 1997, being now probably more
in line with their absorption capacities.
However, even during the most dramatic period of
1997–1998 it was very difficult to find any single crisisaffected country, which did not exhibit enough domestic
vulnerability to be prone to speculative attack against its
currency. The contagion effect could only accelerate what
was anyway unavoidable and it was very unlikely that it
could hit a completely "innocent" country.
The consequences of currency crises in developing and
transition economies are usually severe with output and
employment loses, depreciating real income of population,
deeply contracting investment and capital inflows, ruined
country credibility, etc. In some cases, a crisis can serve as
the economic catharsis: devaluation helps to restore competitiveness and improve current account position. Even if
this happens, the results are not necessarily sustainable. The
same concerns political changes: the crisis shock can bring
the new, more responsible and reform-oriented government but it may also initiate the long lasting political stalemate and destabilization.
The most powerful albeit not immediate positive effects
of currency crises are connected with their educative
power. In many countries politicians are so afraid of being
responsible for causing a potential crisis that they are ready
to draw some lessons from those who failed. And this kind
of practical experience may be even more convincing than
any well-balance advice of the IMF or other international
organizations.
The latter must also seriously rethink their role and their
potential co-responsibility for the recent crisis episodes. In
fact, such a self-evaluation process already started and both
the IMF and the World Bank dramatically increased their
CASE Reports No. 51
openness, dissemination of various official documents and
analytical reports, pushed member countries towards
greater transparency of their statistics, regulations, and policies, worked out new international standards and codes, etc.
All these measures should increase transparency of the international financial transactions and help international financial
markets to panic less and react in a more selective way.
However, the most challenging need for the IMF and
other IFIs is to depoliticize their actions and decisions and
stick to the professional criteria of country assessment and
their consequent execution. Otherwise, their professional
influence will further shrink as it happened in the last few
years. In fact, this challenge concerns the IFIs major shareholders even more than their staff and management. G-7
governments must resist temptation to use these organizations as easy off-budget sources of providing politically
motivated foreign aid.
On the national level, there is a large agenda of institutional, structural, and macroeconomic measures, which
would help countries to avoid building dangerous imbalances and systemic pathologies, diminishing in this way a
danger of currency crises. In the macroeconomic sphere,
they involve the balanced and transparent fiscal accounts,
proper monetary-fiscal policy mix, and consequent antiinflationary policies, avoiding various kinds of indexation of
nominal variables and intermediate monetary/exchange rate
regimes. On the microeconomic level, privatization, demonopolization, trade openness, competition policy and simple, fair and transparent tax system should help to avoid soft
budget constraints, over-borrowing and moral hazard problems. Financial sector requires a great deal of transparency,
tough and well-executed prudential regulations, and private
ownership involving the first-class multinational financial
institutions. In most of the emerging-market economies the
mentioned microeconomic measures must be strengthened
by legal reforms, efficient and fair judiciary system, international accounting, reporting and disclosure standards, transparent corporate and public governance rules, and many
others. This clearly implies an extensive agenda of political
and institutional reforms aiming at making governments and
public governance more accountable, efficient and businessfriendly and concentrated on supply of the basic public
goods rather than trying to substitute market mechanism in
the spheres where the latter work better.
The above list is far from being complete.
55
M. D¹browski (ed.)
References
Ahluwalia, Pavan (2000): Discriminating Contagion – An
Alternative Explanation of Contagious Currency Crises in
Emerging Markets. IMF Working Paper, WP/00/14.
Agénor, Pierre-Richard, Jagdeep S. Bhandari, and Robert
P. Flood (1991): Speculative Attacks and Models of Balance of
Payments Crises. NBER Working Paper, No. 3919, November.
Antczak, Ma³gorzata (2001): The 1997 Currency Crisis in
Thailand, [in:]Marek D¹browski (ed.): The Episodes of Currency Crises in Latin American and Asian Economies, CASE
Reports, No. 39.
Antczak, Rafal (2000): Theoretical Aspects of Currency
Crises. CASE – Center for Social and Economic Research,
Warsaw, Studies and Analyses, No. 211.
Antczak, Rafal (2001): The Russian Crisis of 1998, [in:]
Marek D¹browski (ed.): The Episodes of the Currency Crises
in the European Transition Economies, CASE Reports, No. 40.
Antczak, Rafal, Ma³gorzata Markiewicz, and Artur Radziwi³³
(2001): Financial Crises in FSU Countries: The Role of the IMF.
[in:] Marek D¹browski (ed.): Currency Crises in Emerging Markets – Selected Comparative Studies, CASE Reports, No. 41.
Aziz Jahangir, Francesco Caramazza, and Ranil Salgado
(2000): Currency Crises: In Search of Common Elements. IMF
Working Paper, WP/00/67.
Berg Andrew, and Catherine Pattillo (1999): Predicting
currency crises: the indicator approach and an alternative. Journal of International Money and Finance, vol. 18, pp. 561-586.
B³aszkiewicz, Monika (2000): What Factors Led to the
Asian Financial Crisis: Where or Were not Asian Economics
Sound? CASE – Center for Social and Economic Research,
Warsaw, Studies and Analyses, No. 209.
B³aszkiewicz, Monika (2001): The South Korean Currency
Crisis, 1997–1998. [in:] Marek D¹browski (ed.): The
Episodes of Currency Crises in Latin American and Asian
Economies, CASE Reports, No. 39.
B³aszkiewicz, Monika, and Wojciech Paczyñski (2001):
The Economic and Social Consequences of Financial Crises, [in:]
Marek D¹browski (ed.): Currency Crises in Emerging Markets
– Selected Comparative Studies, CASE Reports, No. 41.
Bordo, Michael, and Anna J. Schwartz (1998): Under
What Circumstances, Past and Present, Have International
Rescues of Countries in Financial Distress Been Successful?,
NBER Working Paper, No. 6824, December.
Buiter Willem H. (1997): Aspects of Fiscal Performance in
some Transitions Economies Under Fund-Supported Programs.
IMF Working Paper, No. WP/97/31, April.
56
Bussiere Matthieu, and Christian Mulder (1999): External
Vulnerability in Emerging Market Economies – How High Liquidity Can Offset Weak Fundamentals and the Effects of Contagion. IMF Working Paper, WP/99/88.
Calvo, Guillermo A., and Carmen M. Reinhart (2000):
Fear of Floating. NBER Working Paper, No. 7993, November.
Caramazza Francesco, Luca Antonio Ricci, and Ranil Salgado (2000): Trade and Financial Contagion in Currency Crises.
IMF Working Paper, WP/00/55.
Chang, Roberto and Andres Velasco (1998): Financial
Crises in Emerging Markets. NBER Working Paper, No.
6606, June.
Chari, V.V., and Patric Kehoe (1997), Hot Money. NBER
Working Paper, No. 6007, April.
Chinn, Menzie D. (1998): Before the Fall: Were East
Asian Currencies Overvalued? NBER Working Paper, No.
6491, April.
Choueiri, Nada, and Graciella Kaminsky (1999): Has the
Nature of Crisis Changed? A Quarter of Currency Crises in
Argentina. IMF Working Paper, WP/99/152, November.
Cline, William R. (1988): International Debt: Progress and
Strategy. Finance and Development, vol. 25 No.2.
Corsetti, Giancarlo, Paolo Pesenti, and Noel Roubini
(1998a): Paper Tigers? A Model of Asian Crisis. NBER Working
Paper, No. 6783, November.
Corsetti, Giancarlo, Paolo Pesenti, and Noel Roubini
(1998b): What Caused the Asian Currency and Financial Crisis? NBER Working Paper, No. 6833 and 6834, December.
Cooper Richard N. (1971): Currency devaluation in developing countries. Essays in International Finance, vol. 86,
Princeton University
D¹browski, Marek (1998): Western Aid Conditionality and
the Post-Communist Transition 1990–1994. Journal of Policy
Reform, Vol. 2, pp. 169-193.
D¹browski, Marek (1999): Последствия российского финансового кризиса для соседних стран
(Consequences of the Russian Financial Crisis for the neighboring countries). CASE – Center for Social and Economic
Research, Warsaw, Studies and Analyses, No. 198.
Dornbusch, Rudiger (2001): A primer on emerging market
crises. MIT, http://www.mit.edu/people/rudi/papers.html
Drazen, Allan (1999): Political Contagion in Currency
Crises. NBER Working Paper, No. 7211, July.
Edwards, Sebastian (2001): Does the current account
matter? NBER Working Paper, No. 8275, May.
Edwards, Sebastian, and Alejandra Cox Edwards (1991):
Monetarism and Liberalization. The Chilean Experiment. The
University of Chicago Press.
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
Eichengreen, Barry, Andrew K. Rose, and Charles
Wyplosz (1994): Speculative Attacks on Pegged Exchange
Rates: An Empirical Exploration with Special Reference to the
European Monetary System. NBER Working Paper, No.
4898, October.
Eichengreen, Barry, Andrew K. Rose, and Charles
Wyplosz (1996): Contagious Currency Crises. NBER Working
Paper, No. 5681, July.
Feldstein, Martin (1999): Self-Protection for Emerging
Market Economies. NBER Working Paper, No. 6907, January.
Fischer, Stanley (1988): Real Balances, the Exchange Rate,
and Indexation: Real Variables in Disinflation. Quarterly Journal of Economics, vol. 103, No. 1.
Flood, Robert P., and Peter M. Garber (1984): Collapsing
Exchange Rate Regimes: Some Linear Examples. Journal of
International Economics, Vol. 17, pp. 1-13.
Flood, Robert P., and Nancy Marion (1998): Perspectives
on Recent Currency Crisis Literature. NBER Working Paper,
No. 6380, January.
Flood, Robert P., Peter M. Garber, and Charles Kramer
(1995): Collapsing Exchange Rate Regimes: Another Linear
Example. NBER Working Paper, No. 5318, October.
Forbes, Kristin J., and Roberto Rigobon (1999): No Contagion, Only Interdependence: Measuring Stock Market Comovements. NBER Working Paper, No. 7267, July.
Forbes, Kristin J., and Roberto Rigobon (2001): Measuring Contagion: Conceptual and Empirical Issues, [in:] Stijn
Claessens and Kristin J. Forbes (eds.): International Financial
Contagion, Kluwer Academic Publishers.
Frankel, Jeffrey A. (1999): No Single Currency Regime is
Right for All Countries or at All Times. NBER Working Paper,
No. 7338, September.
Frankel, Jeffrey A., and Andrew K. Rose (1996): Currency Crashes in Emerging Markets: Empirical Indicators. NBER
Working Paper, No. 5437, January.
Friedman, Milton and Anna Schwartz (1963): A Monetary
History of the United States, 1867–1960. Princeton University Press.
Ganev, Georgy (2001): The Bulgarian Currency Crisis of
1996–1997. [in:] Marek D¹browski (ed.): The Episodes of
the Currency Crises in the European Transition Economies,
CASE Reports, No. 40.
Glick, Reuven, and Michael Hutchison (2000): Banking
and Currency Crises: How Common are the Twins? Hong Kong
Institute for Monetary Research.
Glick, Reuven, and Andrew K. Rose (1999): Contagion
and trade: Why are currency crises regional? Journal of International Money and Finance, vol. 18, pp. 603-617.
CASE Reports No. 51
Goldfajn Ilan, and Rodrigo O. Valdes (1996): The Aftermath of Appreciation. NBER Working Paper, No. 5650, July.
Goldfajn Ilan, and Rodrigo O. Valdes (1997): Are Currency Crises Predictable? IMF Working Paper, WP/97/159.
Hotelling, Harold (1931): The Economics of Exhaustible
Resources, Journal of Political Economy, vol. 39, pp. 137-25.
IIE (1999): Safeguarding Prosperity in a Global Financial
System. The Future International Financial Architecture, A
Council of Foreign Relations-Sponsored Report, Institute
for International Economics, Washington, D.C
IMF (1999): Kyrgyz Republic: Recent Economic Developments. IMF Staff Country Report, No. 99/31, April
IMF (2001a): Conditionality in Fund-Supported Programs –
Policy Issues. Policy Development and Review Department,
February 16, 2001 February 16.
IMF (2001b): Structural Conditionality in Fund-Supported
Programs. Policy Development and Review Department,
February 16, 2001 February 16.
Jakubiak, Ma³gorzata (2000): Indicators of Currency Crisis:
Empirical Analysis of Some Emerging and Transition Economies.
CASE – Center for Social and Economic Research, Warsaw,
Studies and Analyses, No. 218.
Jakubiak, Ma³gorzata (2001a): Choice of the Exchange
Rate Regime and Currency Crashes – Evidence of Some Emerging Economies, [in:] Marek D¹browski (ed.): Currency Crises
in Emerging Markets – Selected Comparative Studies, CASE
Reports, No. 41.
Jakubiak, Ma³gorzata (2001b): The 1995 Currency Crisis
in Argentina, [in:] Marek D¹browski (ed.): The Episodes of
Currency Crises in Latin American and Asian Economies, CASE
Reports, No. 39.
Kaminsky, Graciella (1998): Currency and Banking Crises
– The Early Warning of Distress. IMF Working Paper,
WP/99/178.
Kaminsky, Graciela, Lizondo Saul, and Carmen Reinhart
(1998): Leading Indicators of Currency Crises. IMF Staff
Papers, Vol. 45, No. 1.
Kaminsky, Graciela, and Carmen Reinhart (1999): The
Twin Crises: the Causes of Banking and Balance-of-Payments
Problems, The American Economic Review, Vol. 89, No. 3.
Kindleberger, Charles P. (1978): Maniacs, Panics, and
Crashes: A History of Financial Crises. Basic Books, New
York.
Köhler, Horst (2001): Home Free? Turning Crisis into
Opportunity. Asian Wall Street Journal, January 8.
Krugman, Paul (1979): A model of balance of payments
crises. Journal of Money, Credit, and Banking, vol. 11, pp.
311-325.
57
M. D¹browski (ed.)
Krugman, Paul (1988): Target Zones and Exchange Rate
Dynamics. NBER Working Paper, No. 2481, January.
Krugman, Paul (1998a): Asia: What Went Wrong. Fortune, March 2.
Krugman, Paul (1998b): What Happened to Asia? MIT,
mimeo, January.
Krugman, Paul (1999): Balance Sheets, the Transfer Problem, and Financial Crises, preliminary draft, www.mit.edu/people/krugman, January.
Markiewicz, Ma³gorzata (1998): Fiscal policy and disinflation in transition economies. CASE – Center for Social
and Economic Research, Warsaw, Studies and Analyses,
No. 127.
Markiewicz, Ma³gorzata (2001): The Ukrainian Crisis of
1998, [in:] Marek D¹browski (ed.): The Episodes of the Currency Crises in the European Transition Economies. CASE
Reports, No. 40.
Masson, Paul R. (1998): Contagion – Monsoonal Effects,
Spillovers, and Jumps Between Multiple Equilibria. IMF Working Paper, WP/98/142, September.
Masson, Paul R. (1999): Contagion: macroeconomic models with multiple equilibria. Journal of International Money
and Finance, Vol. 18 (1999), p. 587-602.
McQuillan, Lawrence J. (1999): The Case against the
International Monetary Fund. Hoover Institution, Essays in
Public Policy, November.
Meltzer, Allan H. (1998): Statement before the Joint Economic Committee of the United States Congress: Asian Problems and the IMF, February 24
Meltzer, Allan H. (1999): What's Wrong with the IMF?
What Would Be Better? [in:] William Hunter, George Kaufman, and Thomas Krueger (eds.): The Asian Financial Crisis,
Boston: Kluwer Academic Publishers.
Mercer-Blackman, Valerie, and Anna Unigovskaya (2000):
Compliance with IMF Program Indicators and Growth in Transition Economies. IMF Working Paper, No. WP/00/47, March.
Milesi-Ferretti, Gian Maria, and Assaf Razin (1996): Sustainability of Persistent Current Account Deficits. NBER Working Paper, No. 5467.
Milesi-Ferretti, Gian Maria, and Assaf Razin (1998): Current account reversals and currency crises: empirical regularities. IMF Working Paper, WP/98/89.
Minsky, H.P. (1972): Financial Stability Revisited: The Economics of Disaster, [in:] Reappraisal of the Federal Reserve Discount Mechanism. Vol. 3, Board of Governors of the Federal Reserve System, Washington, D.C., pp. 95-136.
Mishkin, Frederic S. (1999): Lessons from the Asian Crisis,
Journal of International Money and Finance, vol. 18, pp. 709-723.
58
Mishkin, Frederic S. (2000): International Experience with
Different Monetary Policy Regimes. NBER Working Paper, No.
7044, September.
Mishkin, Frederic S. (2001): Financial Policies and the Prevention of Financial Crises in Emerging Market Countries.
NBER Working Paper, No. 8087, January.
McKinnon, Ronald, and Huw Phil (1996): Credible Liberalization and International Capital Flows: the "Overborrowing
Syndrome", [in:] Takatoshi Ito and Anne O. Krueger (eds):
Financial Deregulation and Integration in East Asia. The University of Chicago Press.
Mundell, Robert A. (1960): The Monetary Dynamics of
International Adjustment under Fixed and Flexible Exchange
Rates. Quarterly Journal of Economics, v. LXXIV, no. 2, pp.
227-257.
Obstfeld, Maurice (1994): The Logic of Currency Crisis.
NBER Working Paper, No. 4640, September.
Obstfeld, Maurice (1997): Models of Currency Crises with
Self-fulfilling Features. NBER Working Paper, No. 5285, February.
Obstfeld, Maurice, and Kenneth Rogoff (1995): The
Mirage of Fixed Exchange Rates. NBER Working Paper, No.
5191, July.
Otker, Inci, and Ceyla Pazarbasioglu (1997): Speculative Attacks and Macroeconomic Fundamentals: Evidence
from Some European Currencies. European Economic
Review, vol. 41.
Paczyñski, Wojciech (2001): The Mexican Peso Crisis
1994–1995, [in:] Marek D¹browski (ed.): The Episodes of
Currency Crises in Latin American and Asian Economies. CASE
Reports, No. 39.
Poirson, Helene (2001): How Do Countries Choose Their
Exchange Rate Regime? IMF Working Paper, No. WP/01/46,
April.
Radelet, Steven, and Jeffrey Sachs (1998): The Onset of
the East Asian Financial Crisis. NBER Working Paper, No.
6680, August.
Radziwi³³, Artur (2001): The Moldovan Currency Crisis,
1998, [in:] Marek D¹browski (ed.): The Episodes of the Currency Crises in the European Transition Economies. CASE
Reports, No. 40.
Rawdanowicz, £ukasz (2001): Propagation of Currency
Crises – The Case of the Russian Crisis, [in:] Marek D¹browski (ed.): Currency Crises in Emerging Markets – Selected Comparative Studies, CASE Reports, No. 41.
Rodrik, Dani (1996): Understanding Economic Policy
Reform. Journal of Economic Literature, Vol. 34, pp. 9-41,
March
CASE Reports No. 51
Currency Crises in Emerging-Market Economies ...
Rodrik, Dani, and Andres Velasco (1999): Short Term
Capital Flows. NBER Working Paper, No. 7364, September.
Sachs, Jeffrey (1981): The Current Account and the
Macroeconomic Adjustment Process. NBER Working Paper,
No. 0796, November.
Sachs, Jeffrey D. (1998): Proposals for Reform of the Global Financial Architecture. Paper prepared for the UNDP
Meeting on Reform of the Global Financial Architecture,
New York, December 8.
Salant, Stephen W., and Dale W. Henderson (1978):
Market Anticipation of Government Policy and the Price of
Gold. Journal of Political Economy, Vol. 86, pp. 627-48.
Sasin, Marcin (2001a): Predicting Currency Crises, the Ultimate Significance of Macroeconomic Fundamentals in Linear
Specifications with Nonlinear Extensions. CASE – Center for
Social and Economic Research, Warsaw, Studies and Analyses, No. 224.
Sasin, Marcin (2001b): The Importance of the Real
Exchange Rate Overvaluation and the Current Account Deficit
in the Emergence of Finacial Crises, [in:] Marek D¹browski
(ed.): Currency Crises in Emerging Markets – Selected Comparative Studies, CASE Reports, No. 41.
Sasin, Marcin (2001c): Financial Systems, Financial Crises,
Currency Crises, [in:] Marek D¹browski (ed.): Currency Crises
in Emerging Markets – Selected Comparative Studies, CASE
Reports, No. 41.
Sasin, Marcin (2001d): The Turkish 2000 Financial Market
Crisis of Confidence. [in:] Marek D¹browski (ed.): The
Episodes of the Currency Crises in the European Transition
Economies, CASE Reports, No. 40.
Sasin, Marcin (2001e): The Malaysian Currency Crisis,
1997–1998, [in:] Marek D¹browski (ed.): The Episodes of
Currency Crises in Latin American and Asian Economies, CASE
Reports, No. 39.
Sasin, Marcin (2001f): The Indonesian Currency Crisis,
1997–1998, [in:] Marek D¹browski (ed.): The Episodes of
Currency Crises in Latin American and Asian Economies, CASE
Reports, No. 39.
CASE Reports No. 51
Siwiñska, Joanna (2000): Currency Crises and Fiscal Imbalances – the Transition Countries Perspective. CASE – Center for
Social and Economic Research, Warsaw, Studies and Analyses,
No. 219.
Stiglitz, Joseph (1998): Must Financial Crises Be This Frequent and This Painful? McKay Lecture, Pittsburgh, Pennsylvania, September 23.
www.worldbank.org/knowledge/chiefecon/stiglitz.htm.
Summers, Lawrence H. (1996): Commentary, [in:]
Ricardo Hausmann and Liliana Rojas-Suarez (eds.): Volatile
Capital Flows, Inter-American Development Bank.
Szczurek, Mateusz (2001): International Liquidity, and the
Cost of Currency Crises, [in:] Marek D¹browski (ed.): Currency Crises in Emerging Markets – Selected Comparative Studies,
CASE Reports, No. 41.
Tanner, Evan (1999): Exchange Market Pressure and Monetary Policy – Asia and Latin America in the 1990s, IMF Working Paper, WP/99/114.
Tomczyñska, Magdalena (2000): Early Indicators of Currency Crises. Review of Some Literature. CASE – Center for
Social and Economic Research, Warsaw, Studies and Analyses, No. 208.
Tornell, Aaron (1999): Common Fundamentals in Tequila
and Asian Crises, NBER Working Paper, No. 7139, May.
Van Rijckeghem, Caroline, and Beatrice S. Weder
(1999): Sources of Contagion: Finance or Trade? IMF Working
Paper, WP/99/146.
Vaubel, Roland (1991): The Political Economy of the
International Monetary Fund: A Public Choice Approach, [in:]
Roland Vaubel, and Thomas D. Willett (eds.): The Political
Economy of International Organizations, Boulder, Westview:
205-245.
Velasco, Andres (1996): Fixed exchange rates: Credibility,
flexibility and multiplicity. European Economic Review, Vol.
40, pp. 1023-1035.
WEO (1998): World Economic Outlook, IMF, May 1998.
WEO (1999): World Economic Outlook, IMF, May 1999.
59
32
Praca zbiorowa pod redakcj¹ A. Wojtyny: Alternatywne strategie dezinflacji
33
Praca zbiorowa pod redakcj¹ A. Wojtyny: Wspieranie wzrostu gospodarczego poprzez konsolidacjê reform
34
J. Pankow, L. Dimitrov, P. Kozarzewski: Effects of Privatization of Industrial Enterprises in Bulgaria. Report on
Empirical Research
35
Praca zbiorowa pod redakcj¹ S. Golinowskiej: Edukacja i rynek pracy
36
S. Golinowska, P. Kurowski (eds.): Rational Pension Supevision, First Experiences of Central and Eastern European
States in Comparison with other Countries
37
J. Pañków (ed.): Fiscal Effects from Privatization: Case of Bulgaria and Poland
38
G. Ganev, M. Jarociñski, R. Lubenova, P. WoŸniak: Credibility of the Exchange Rate Policy in Transition Countries
39
M. D¹browski (ed.): The Episodes of Currency Crises in Latin American and Asian Economies
40
M. D¹browski (ed.): The Episodes of Currency Crises in the European Transition Economies
41
M. D¹browski (ed.): Currency Crises in Emerging Markets – Selected Comparative Studies
42
Praca zbiorowa pod redakcj¹ J. Cukrowskiego: Renta emisyjna jako Ÿród³o finansowania bud¿etu pañstwa
43
P. Bujak, M. Jarmu¿ek, S. Kordel, W. Nawrot, J. Œmietaniak: Œredniookresowa projekcja dzia³alnoœci inwestycyjnej
Otwartych Funduszy Emerytalnych na regulowanym rynku gie³dowym akcji
44
E. Balcerowicz, A. Bratkowski: Restructuring and Development of the Banking Sector in Poland. Lessons to be
Learnt by Less Advanced Transition Countries
45
E. Kocenda, J. Valachy: Secondary Privatization in the Czech Republic: Changes in Ownership and Enterprise
Performance in Voucher-Privatized Firms
46
M. Simoneti, A. Böhm, M. Rems, M. Rojec, J.P. Damijan, B. Majcen: Secondary Privatization in Slovenia: Evolution of
Ownership Structure and Company Performance Following Mass Privatization
47
P. Kozarzewski, R. Woodward: Secondary Privatization in Poland (Part I): Evolution of Ownership Structure and
Company Performance in Firms Privatized by Employee Buyouts
48
B. B³aszczyk, M. Górzyñski, T. Kamiñski, B Paczóski: Secondary Privatization in Poland (Part II): Evolution of Ownership Structure and Performance in National Investment Funds and their Portfolio Companies
49
I. Grosfeld, I Hashi: The Evolution of Ownership Structure in Firms Privatized through Wholesale Schemes in the
Czech Republic and Poland
50
B. B³aszczyk, R. Woodward: Secondary Privatisation: The Evolution of Ownership Structures of Privatised
Enterprises