Evolution and Performance of Exchange Rate Regimes

229O C C A S I O N A L PA P E R

Evolution and Performance of
Exchange Rate Regimes

Kenneth S. Rogoff, Aasim M. Husain, Ashoka Mody,
Robin Brooks, and Nienke Oomes

INTERNATIONAL MONETARY FUND
Washington DC
2004

229O C C A S I O N A L PA P E R

Evolution and Performance of
Exchange Rate Regimes

Kenneth S. Rogoff, Aasim M. Husain, Ashoka Mody,
Robin Brooks, and Nienke Oomes

INTERNATIONAL MONETARY FUND
Washington DC
2004

© 2004 International Monetary Fund
Production: IMF Multimedia Services Division

Figures: Theodore F. Peters, Jr.
Typesetting: Alicia Etchebarne-Bourdin

Cataloging-in-Publication Data
Evolution and performance of exchange rate regimes/Kenneth S.

Rogoff . . . [et al.]—Washington, D.C.: International Monetary Fund [2004].
p. cm.—(Occasional paper; 229)
Includes bibliographical references
ISBN 1-58906-327-9

1. Foreign exchange rates. 2. Foreign exchange rates—Developing coun-
tries. 3. Inflation (Finance). 4. Business cycles. I. Rogoff, Kenneth S.
II. Occasional paper (International Monetary Fund); 229.
HG3811.E96 2004

Price: US$25.00
(US$22.00 to full-time faculty members and

students at universities and colleges)

Please send orders to:
International Monetary Fund, Publication Services
700 19th Street, N.W., Washington, D.C. 20431, U.S.A.
Tel.: (202) 623-7430 Telefax: (202) 623-7201

E-mail: [email protected]
Internet: http://www.imf.org

recycled paper

Contents

Preface vii

I. Overview 1

II. The Evolution of Exchange Rate Regimes: A Fresh Look 4
5
New Regime Classifications 6
Divergence Between Stated and Actual Policies 12
Regime Transitions 14
Implications for Assessing Regime Performance 14
Appendix I. The Natural Classification 16
Appendix II. Determinants of Exchange Rate Regime Choice
21
III. Regime Performance: Inflation and Business Cycles 22
23
Summary of Empirical Analysis of Exchange Rate Regimes
Analytical Considerations 24
Macroeconomic Performance and Crisis Probabilities: Summary 28
35
Statistics
Regime Performance and Levels of Development 37
Achieving Credibility in Developing and Emerging Economies
Appendix. Data and Regression Results for Economic 49
50
Performance Analysis
10
IV. Summary
13
References 24

Box 25

2.1. Anchor Currency Choice 26

Tables 27

2.1. Annual Transition Probabilities 27
3.1. Economic Performance Across Exchange Rate Regimes
3.2. Average Annual Inflation and Real Per Capita GDP Growth: 28

Comparison of Dual (or Multiple) and Unified Exchange
Rate Systems, 1970–99
3.3. Average Annual Inflation Rates Across Exchange Rate
Regimes, 1970–99
3.4. Average Annual Real Per Capita GDP Growth Across Exchange
Rate Regimes, 1970–99
3.5. Average Annual Growth Volatility Across Exchange Rate
Regimes, 1970–99
3.6. Real Exchange Rate Volatility Across Exchange Rate Regimes,
1970–2002

iii

CONTENTS

3.7. Probability of Crises During Specific Regimes Using the Natural 29
Exchange Rate Regime Classification
15
Appendix Tables 16

A2.1. Main Features of Various De Facto Classifications 18
A2.2. Natural Classification Categories 38
A2.3. Studies on Determinants of Exchange Rate Regimes (Likelihood 40
41
to Float) 42
A3.1. Variable Description 43
A3.2. List of Countries
A3.3 Comparing IMF De Jure and Natural Classifications 44
A3.4. Inflation Performance Across Country Groups 45
A3.5. Growth Performance Across Country Groups
A3.6. Volatility of Real GDP Growth Performance Across Country 46
47
Groups
A3.7. Emerging Markets in the 1990s 7
A3.8. Inflation Performance in Developing Countries: Announcement 7
8
and Duration Effects 8
A3.9. Robustness Tests 9
11
Figures 11
12
2.1. Natural Classification Regimes by De Jure Category, 1973–99 30
2.2 Countries with Dual/Parallel Foreign Exchange Markets
2.3 Natural Classification Regime Distribution 30
2.4 De Jure Regime Distribution
2.5. Natural Classification Regime Distribution by Country Group 31
2.6. IMF De Facto Regime Distribution
2.7. Levy-Yeyati–Sturzenegger Regime Distribution 31
2.8. Natural Classification Regime Transitions 32
3.1. Inflation Performance Across Regimes: Confidence Effect
3.2. Inflation Performance Across Regimes: Confidence and 32
33
Discipline Effects
3.3. Inflation Performance in Advanced Countries, Emerging Markets, 34

and Developing Countries: Confidence Effect 34
3.4. Inflation Performance in Advanced Countries, Emerging Markets,
34
and Developing Countries: Confidence and Discipline
Effects 35
3.5. Growth Performance Across Regimes
3.6. Growth Performance in Advanced Countries, Emerging Markets, 36
and Developing Countries
3.7. Volatility of Real GDP Growth Performance Across Regimes
3.8. Volatility of Real GDP Growth Performance in Advanced
Countries, Emerging Markets, and Developing Countries
3.9. Volatility of Real GDP Growth and Contamination Across
Regimes: Evidence from Emerging Markets
3.10. Volatility of Real GDP Growth Across Regimes: Emerging
Markets for the 1990s Only
3.11. The Inflation Benefit Associated with Announcement in
Developing Countries
3.12. The Inflation Benefit Associated with Consistent Macroeconomic
Policies in Developing Countries

iv

Contents

The following symbols have been used throughout this paper:
. . . to indicate that data are not available;
— to indicate that the figure is zero or less than half the final digit shown, or that the item

does not exist;
– between years or months (e.g., 2003–04 or January–June) to indicate the years or

months covered, including the beginning and ending years or months;
/ between years (e.g., 2003/04) to indicate a fiscal (financial) year.
“n.a.” means not applicable.
“Billion” means a thousand million.
Minor discrepancies between constituent figures and totals are due to rounding.
The term “country,” as used in this paper, does not in all cases refer to a territorial entity that
is a state as understood by international law and practice; the term also covers some territorial
entities that are not states, but for which statistical data are maintained and provided interna-
tionally on a separate and independent basis.

v



Preface

The issue of the appropriate exchange rate regime for a country has been perenni-
ally lively, with a new set of considerations coming to the fore in the 1990s. The role
played by international capital flows and domestic financial systems in determining
the performance of exchange rate regimes has gained prominence in the policy debate
over the appropriate exchange rate arrangement.

Using recent advances in the classification of exchange rate regimes to draw new
lessons about the performance of alternative regimes, this paper’s key message is that,
as economies and their institutions mature, the value of exchange rate flexibility in-
creases. In reaching this overarching conclusion, the paper takes an empirical per-
spective to form a judgment of actual experience. The study takes as a given that mul-
tiple currencies will continue to exist in the foreseeable future. Hence, the policy
analysis is intended for countries that choose to retain their own currencies.

The study was prepared by a staff team under the direction of Kenneth Rogoff,
when he was Economic Counsellor and Director of the IMF’s Research Department.
The team was led by Aasim Husain and Ashoka Mody, and included Robin Brooks
and Nienke Oomes. The authors would like to acknowledge contributions from
Andrew Berg, who worked on the subsection “Learning to Float” in Section III; Paolo
Mauro and Grace Juhn, who provided the subsection “Empirical Findings on Factors
Affecting Regime Choice” in Appendix II; and Antonio Spilimbergo, who did the esti-
mations for the subsection “Regimes and Crisis Probabilities” in Section III. The
paper also benefited from useful comments and sugg.estions from Tamim Bayoumi,
Michael Bordo, Agustín Carstens, Barry Eichengreen, Inci Ötker-Robe, Jacques Polak,
Carmen Reinhart, David J. Robinson, Miguel Savastano, Federico Sturzenegger, Lars
Svensson, and numerous other colleagues throughout the Fund. Young Kim and
Eisuke Okada provided valuable research assistance. Atish Ghosh generously shared
his data set. Sheila Kinsella managed the Research Department end in the preparation
of this paper. Gail Berre of the External Relations Department edited the paper and
coordinated production of the publication.

An earlier draft of the paper was presented at an informal seminar of the IMF’s
Executive Board, and the current version has benefited from the comments made on
that occasion. The views expressed are those of the authors, however, and do not nec-
essarily reflect those of national authorities or IMF Executive Directors.

vii



I Overview

T his study assesses the historical durability and who were skeptical of the regimes’ sustainability.
performance of alternative exchange rate Many such episodes were associated with costly fi-
regimes, with special focus on developing and nancial crises, especially in emerging markets. One
emerging market countries. It suggests that the pop- influential view predicted that exchange rate
ular bipolar view of exchange rates is neither an ac- regimes would move in a bipolar manner to the ex-
curate description of the past nor a likely scenario tremes of hard pegs, which would be relatively im-
for the next decade. While the study confirms that mune to speculative pressures or free floats
emerging market countries need to consider adopt- (Eichengreen, 1994; and Fischer, 2001). An increas-
ing more flexible exchange rate regimes as they de- ing number of countries did announce their intent to
velop economically and institutionally, it also finds allow greater exchange rate flexibility. Among de-
that fixed or relatively rigid exchange rate regimes veloping and emerging market economies, however,
have not performed badly for poorer countries. For the de jure announcement to float did not typically
countries that have relatively limited financial mar- translate into de facto fully floating exchange rates.
ket development and relatively closed capital mar- Countries, it appeared, had a fear of floating (Calvo
kets, fixed exchange rate regimes appear to offer and Reinhart, 2002).
some measure of credibility without compromising
growth objectives—with the important proviso that These observed trends and policy ambivalence re-
monetary policy must be consistent in avoiding a flected a variety of opposing considerations in the
large and volatile parallel market premium. As adoption and performance of exchange rate regimes.
countries develop economically and institutionally, In their discussions of papers on exchange rate
there are considerable benefits to adopting a more regimes in September and November 1999, IMF Ex-
flexible exchange rate system—although, of course, ecutive Directors concluded that there were no sim-
the following analysis provides only a general guide ple prescriptions for the choice of a country’s ex-
and should not be interpreted as a one-size-fits-all change rate regime.1 Instead, they emphasized the
prescription. For developed countries that are not in importance of macroeconomic fundamentals and the
a currency union (or headed toward one), relatively consistency of the exchange rate regime with under-
flexible exchange rate regimes offer higher growth lying macroeconomic policies. Several also thought
without any cost in anti-inflation credibility—pro- that a range of alternatives between the polar ex-
vided they are anchored by some other means, such tremes of rigidity and flexibility were viable. More
as an independent central bank with a clear anti- recently, however, the IMF has been urged—from
inflation mandate. One perhaps surprising finding outside as well as within—to take a more prescrip-
of the quantitative analysis is the remarkable dura- tive role in its surveillance of members’ exchange
bility of exchange rate regimes outside of emerging rate policies and regime choice, underscoring the
market countries, with only 7 percent of all coun- importance of an improved understanding of the per-
tries changing regimes in an average year over the formance of alternate regimes (Calomiris, 1998; In-
1940–2001 period. ternational Financial Institution Advisory Commis-
sion, 2000; Mussa, 2002; and IMF, Independent
Debates on the appropriate exchange rate regime Evaluation Office, 2003c).
for a country are perennially lively. In the 1990s, a
new set of considerations came to the fore, particu- While recognizing the central importance of
larly the role played by international capital flows macroeconomic fundamentals, this study uses re-
and domestic financial systems in determining the cent advances in the classification of exchange rate
performance of exchange rate regimes. Just when regimes to draw new lessons from the performance
pegged regimes were gaining respectability as pro-
viding nominal anchors, several pegs (and crawling 1See the summings up of IMF Board discussions in Mussa and
pegs) faced speculative pressures from investors others (2000).

1

I OVERVIEW

of alternative regimes. The findings indicate that, as ior of exchange rates that is supplemented by data on
economies and their institutions mature, the value the movement of foreign exchange reserves and in-
of exchange rate flexibility increases. This conclu- terest rates, as well as judgments on the true intent of
sion reflects distinctions among advanced, emerg- policymakers. Based on such an effort, the IMF now
ing, and other developing economies. Emerging compiles the de facto exchange rate regimes of its
markets have stronger links to international capital member countries, dating back to 1990 (IMF, 1999
markets than do other developing economies. Un- and 2003b). The de facto regime classification prin-
like advanced economies, however, emerging mar- cipally used in this study is the “Natural” classifica-
kets face a variety of institutional weaknesses that tion proposed by Reinhart and Rogoff (2004) which
manifest themselves in higher inflation, problems of is available from the 1940s for virtually all IMF
debt sustainability, fragile banking systems, and member countries. Among its distinguishing features
other sources of macroeconomic volatility, all of is the use of parallel market exchange rates to deter-
which potentially undermine the credibility of poli- mine the actual operation of an exchange rate regime
cymakers. Thus, while the non–emerging market and the identification of a separate category of freely
developing countries (hereinafter referred to as de- falling regimes that are characterized by high infla-
veloping economies) may gain credibility through tion, and thus, implicitly, by weak macroeconomic
pegging their exchange rates, emerging market management.
economies find it harder to do so and could benefit
from investing in learning to float. More advanced This study has two additional main sections. Sec-
economies, with their stronger institutions, are best tion II first discusses several alternative exchange
positioned to enjoy the benefits of flexibility with- rate regime classification systems and reviews per-
out the risk of losing policy credibility. spectives they offer. It describes trends in the distri-
bution of regimes, noting the difference between de
To be clear, this study takes as a given the current jure trends, which show a move to flexibility, and de
conjuncture of a multiplicity of currencies. As such, facto trends, which show that intermediate ex-
the conclusions apply to those countries that have change rate arrangements are still pervasive. The
their own currencies. It is possible, however, that the section also examines the transitions between
current context may evolve, and a sufficiently large regimes and finds that de facto regimes tend to be
number of countries may, in the next decade and be- long-lived. The bulk of the de facto regime transi-
yond, elect to join currency unions, leading to fewer tions in the past half century have occurred in the
currencies in circulation. This would change the be- wake of exceptional events, such as the breakdown
havior of governments and international business, of the Bretton Woods system, the creation of the Eu-
and, hence, change the economic performance of al- ropean Economic and Monetary Union, and the col-
ternative regimes in ways that the following does not lapse of the Soviet Union. In the absence of such
attempt to predict.2 events, the present global distribution of regimes is
not likely to change substantially. Over the longer
Because analytical arguments on the economic term, however, political economy considerations
influence of exchange rate regimes often lead to may guide regime choice in some countries, possi-
opposing conclusions, this study bases its perspec- bly resulting in their election to form or join a cur-
tive on actual experience. Empirical observations rency union. Such transitions, of course, are beyond
are used to form judgments on how offsetting fac- the scope of this analysis.
tors play out in different country groups. The sim-
ple groupings do not allow for complexities at the Section III studies the performance of exchange
level of individual countries, however, by reflect- rate regimes in terms of inflation and business cy-
ing, for example, their economic size and internal cles. It finds that the advantages of exchange rate
heterogeneity. flexibility increase as a country becomes more inte-
grated into global capital markets and develops a
Empirical analysis of exchange rate regime per- sound financial system. Free floats have, on aver-
formance depends, of course, on the classification of age, registered faster growth than other regimes in
regimes. The conclusions of this study rely on the advanced countries, without incurring higher infla-
distinction between de jure and de facto regimes. tion. Conversely, in developing countries with lim-
Owing to the importance of this distinction, attempts ited access to private external capital, pegs and
have been made in recent years to characterize de other limited-flexibility arrangements have also
facto regimes using information on the actual behav- been associated with lower inflation, without an ap-
parent cost in terms of lower growth or higher
2IMF (2003a) concludes, however, that while Group of Three growth volatility. In emerging market economies
(G-3) exchange rate volatility has real effects, especially on some with higher exposure to international capital flows,
countries with high debt ratios and mismatches in trade and finan- however, the more rigid regimes have had a higher
cial flows, the overall effects are small. incidence of crises. The analysis also indicates that

2

Overview

macroeconomic performance under all types of de with results obtained using other classifications to as-
facto regimes was weaker in countries with dual or sess the robustness of the conclusion or to explain
multiple exchange rates that deviated substantially why the differences arise. Fourth, the need for cau-
from official rates, suggesting important gains from tion arises from the fact that, although a country’s
exchange rate unification. regime is conventionally classified as fixed, if its cur-
rency is fixed with respect to a single other currency,
The analysis and results in this study are subject to then performance is a function of multiple relation-
a number of qualifications. First, empirical findings ships with all partner currencies. The combining of
may reflect in part the influence of economic perfor- multiple relationships into one has both descriptive
mance on the choice of regime, rather than the other and prescriptive consequences. For example, in clas-
way around. Second, an inherent difficulty arises in sifying Argentina as a hard peg case, one loses sight
classifying regimes in a fully specified manner. A of the fact that, in relation to the great majority of its
country’s true exchange rate regime is, properly trading partners, the peg to the dollar made it a
speaking, a super regime consisting of a sequence of floater. Finally, further analysis is needed to jointly
regimes and not just the one that prevails at a particu- classify exchange rate regimes and capital account
lar point in time. Thus, the harmful effects of a openness. For all these reasons, while the conclu-
regime may be observed only when it collapses, lead- sions and policy implications drawn in this study
ing to a misattribution of the poor performance to the offer new cross-country perspectives on exchange
successor regime. Third, some of the conclusions de- rate regimes, the results should be interpreted with
pend on the choice of the Natural classification. To suitable caution, especially for individual cases.
the extent possible, such conclusions are compared

3

II The Evolution of Exchange Rate
Regimes: A Fresh Look

I s there an observed tendency for exchange rate oping countries. The so-called “middle” along
regimes to drift to the polar extremes of hard pegs the flexibility dimension continues to constitute
and free floats, with a hollowing of the middle be- half of all regimes, as it has throughout the past
tween the two? Have regime changes become signif- three decades. Freely floating regimes remain
icantly more frequent in the post–Bretton Woods rare. The moderate increase in the number of
era? And have certain regimes historically proven pegs in the 1990s was mainly in the euro area
more difficult to sustain, particularly in countries and the transition economies.
more open to capital flows? Policy debates centered
around these questions have forced a growing recog- • The frequency of regime transitions today is sim-
nition that the exchange rate regime a country actu- ilar to what it was 50 years ago. Since 1940,
ally operates (its de facto regime) often differs around 7 percent of all countries have changed
meaningfully from its announced (or de jure) their regime in a given year, with emerging mar-
regime. This divergence affects potentially the kets tending to switch regimes more frequently
analysis of historical trends in exchange rate than other countries. Apart from transitions re-
regimes, their macroeconomic performance, and the lated to major global or regional events in
answers to salient policy questions. economies experiencing severe macroeconomic
stress, changes in de facto regimes in the post–
In recognition of the divergence between actual Bretton Woods period have been about as fre-
and operational regimes, a number of efforts have quent as during the period of fixed parities.
been undertaken to develop a classification of de
facto rather than de jure regimes. The IMF now pub- This section also provides a brief discussion of the
lishes regime classifications that take into account different approaches to exchange rate regime classi-
the actual functioning of regimes; these are available fication and documents the evolution of regimes
from 1990, and findings based on this classification across the world from 1940. It considers transitions
are reported in IMF (2003b). The Natural classifica- across regimes, and concludes with some observa-
tion, developed by Reinhart and Rogoff (2004), ex- tions of how the choice of a classification system
tends back to the 1940s, and overlaps significantly might affect the assessment of the performance of al-
with the IMF de facto classification in the 1990s. ternate regimes. Throughout the section, differences
The Natural classification also draws analytically across economies that are at different stages of de-
useful distinctions that facilitate the interpretation of velopment and integration into global capital mar-
countries’ economic behavior and performance. kets are highlighted by dividing countries into three
groups—advanced, emerging market, and other de-
This section describes the evolution of exchange veloping economies.3
rate regimes across the world using primarily the
Natural classification, but provides also comparisons 3Emerging market economies are those that are included in the
with other (including the de jure) classifications. Morgan Stanley Capital International (MSCI) index, and comprises
Below are the main findings. Argentina, Brazil, Chile, China, Colombia, Czech Republic, Egypt,
Hungary, India, Indonesia, Israel, Jordan, Korea, Malaysia, Mex-
• Historically, the actual operation of exchange ico, Morocco, Pakistan, Peru, the Philippines, Poland, Russia,
rate regimes seems to have differed from the South Africa, Thailand, Turkey, and República Bolivarian de
announced framework about 50 percent of the Venezuela. With the exception of Israel, advanced economies are
time. Many countries have exhibited a fear of those that are classified as upper-income economies by the World
floating; as a result, the actual flexibility of Bank. All other economies constitute the other developing coun-
their exchange rate was substantially less than tries group. Small variations in the composition of the emerging
announced. markets group do not alter the thrust of the findings reported below
on the evolution of regimes and regime transitions. Recognizing
• Intermediate regimes remain prevalent, espe- the significant variation in financial integration across countries
cially among emerging markets and other devel-

4

New Regime Classifications

New Regime Classifications termine the de facto flexibility of exchange rate
regimes.7 In addition, the IMF itself moved to a de
Until the late 1990s, most empirical studies of ex- facto classification system in 1999. The IMF de facto
change rate regimes relied on the de jure regime classification combines available information on the
classification reported in the IMF’s Annual Report exchange rate and monetary policy framework and
on Exchange Arrangements and Exchange Restric- authorities’ formal or informal policy intentions with
tions (AREAER), which was then based on coun- data on actual exchange rate and reserves movements
tries’ official notifications to the IMF. The de jure to reach a judgment about the actual exchange rate
classification distinguished between three broad cat- regime.8
egories—pegged regimes, regimes with limited flex-
ibility (usually within a band or cooperative arrange- Despite these advances, analysis sometimes re-
ment), and more flexible arrangements (those with quires a more nuanced characterization of regimes.
managed or free floats)—which were then divided Countries experiencing episodes of macroeconomic
into 15 subcategories.4 instability often have very high inflation rates, which
may be reflected in high and frequent exchange rate
Although comprehensive in terms of country and depreciation. Classification of such regimes as float-
historical coverage, the de jure classification system ing, intermediate, or pegged is problematic because
had a serious drawback: in practice, exchange rate the macroeconomic disturbances could be incor-
regimes often differed from what they were offi- rectly attributed to the exchange rate regime. In addi-
cially announced to be. For example, some pegged tion, in countries with significant parallel foreign ex-
regimes devalued frequently, while many floats typi- change markets, where rates differ substantially
cally moved within a tight band. Consequently, the from official ones, movements in parallel rates rather
de jure classification characterized inaccurately the than in official rates provide a more realistic barom-
distribution of operative currency regimes across the eter of underlying monetary policy. In particular,
world and over time. Moreover, empirical analyses countries with a fixed official rate but with high in-
employing this classification to test theories of flation and a rapidly depreciating parallel rate cannot
regime choice or to assess the relationship between be considered as having a monetary stance that is
regime choice and economic performance risked consistent with a pegged regime. Moreover, to assess
reaching incorrect conclusions and drawing mislead- the relationship between regimes and longer-term
ing policy implications.5 economic performance, it is helpful to identify
longer-term regimes rather than shorter-term spells
Recognizing the merits of classifying regimes within a regime, such as the widening of a horizontal
more realistically, a number of new de facto classifi- band or a onetime devaluation followed by a re-peg.
cation systems have been proposed. Ghosh and oth- By employing a relatively short horizon over which
ers (1997) classify regimes on a de facto basis using the de facto regime is assessed, classification algo-
information on actual exchange rate movements. rithms, such as the one employed by Levy-Yeyati
Subsequently, the evidence on macroeconomic per- and Sturzenegger, can record potentially a large
formance under alternative de jure regimes was reex- number of regime changes that are related to short
amined by Ghosh, Gulde, and Wolf (2003) by check- periods of disturbances—possibly transient eco-
ing the robustness of these results against a hybrid de nomic or political shocks—and that do not involve a
jure/de facto classification.6 Another classification change in the regime itself.
system, devised by Levy-Yeyati and Sturzenegger
(2002 and 2003), discards the de jure classification Reinhart and Rogoff’s (2004) Natural classifica-
altogether and instead employs purely statistical tion addresses these shortcomings by separating
techniques to exchange rate and reserves data to de- episodes of severe macroeconomic stress and incor-

and over time within the emerging markets group, this study also 7The Levy-Yeyati–Sturzenegger data set, which goes back to
reports results for the 1990–2001 period where relevant. 1974, attempts to classify, on an annual basis, about 180 countries
in terms of actual flexibility. About one-third of the observations
4See Ghosh, Gulde, and Wolf (2003) for a description of the de in their sample cannot be classified by their algorithm, however,
jure classification system, as well as historical data on countries’ because of missing data or because the regime was a peg to an
classification under this system. undisclosed basket.

5For an early recognition of this concern, see Edwards and 8See IMF (1999), Section IV, for details. The IMF de facto
Savastano (1999). classification is, in effect, a hybrid classification system that com-
bines data on actual flexibility with information on the policy
6The hybrid classification—referred to as the “consensus” framework. Using historical data and information on countries’
classification by Ghosh, Gulde, and Wolf (2003)—discards ob- exchange arrangements, Bubula and Ötker-Robe (2002) put to-
servations in which the de jure classification does not match a de gether a database containing IMF de facto classifications going
facto one, based on actual exchange rate movements. Effectively, back to 1990.
this procedure narrows the sample by 35 percent over the
1970–99 period.

5

II THE EVOLUTION OF EXCHANGE RATE REGIMES: A FRESH LOOK

porating information on dual/parallel market ex- cations may result in a high frequency of recorded
change rates.9 Their classification distinguishes regime transitions because of changes in the pattern
regimes that are freely falling as a separate category of actual exchange rate movements. The Natural
and, in cases where the dual/parallel exchange rate classification addresses this issue by employing a
differs substantially from the official rate, uses move- five-year horizon to gauge actual exchange rate
ments in the former rate to classify the regime. Also, flexibility. While this helps to distinguish regimes
a five-year horizon is used to gauge the true flexibil- from spells, it limits the Natural classification’s
ity of the longer-term exchange rate regime. The Nat- ability to detect short-term currency market pres-
ural classification divides de facto regimes into five sures, such as those that culminated in the CFA
coarse categories—fixed, limited flexibility, man- franc devaluation in early 1994, that could have
aged floating, freely floating, and freely falling—and longer-term macroeconomic effects. Hence, the
into 14 fine subcategories. The Reinhart-Rogoff data Natural classification is not necessarily appropriate
set is comprehensive, covering virtually all IMF for analyzing issues, such as the near-term impact
members, in most cases, back to 1946. Hence, it fa- of changes in a country’s exchange rate spell. From
cilitates richer historical analysis of regime distribu- a global perspective, however, the Natural classifi-
tions, transitions, and performance than other de cation, with its special features and rich historical
facto classifications.10 coverage, has the potential to yield important new
insights into the history of regimes and their effect
Some qualifications should be noted, however, on macroeconomic performance.
with respect to de facto classifications, including
the Natural classification. The absence of exchange Divergence Between Stated and
rate variability that is used to classify regimes may Actual Policies
reflect the absence of real shocks to the economy
rather than a fixed exchange rate regime. Reinhart, Comparison of the de jure and Natural classifica-
Rogoff, and Spilimbergo (2003) find, however, that tions highlights the divergence between stated and
countries that have had relatively stable exchange actual policies, particularly at the polar extreme of
rates have not been subjected to fewer or smaller flexibility. Focusing on the broad classification cate-
terms-of-trade shocks.11 Also, de facto classifica- gories over the period 1973–99 (for which there are
tions are based on past movements of exchange overlapping data), Figure 2.1 shows that only about
rates as well as other variables. Hence, they are half of the observations—where each observation
backward looking and do not incorporate informa- corresponds to a given country’s regime in a particu-
tion on policy intentions, which may in turn affect lar year—were classified in the same broad category
economic performance.12 This argument cuts both under both the de jure and the Natural classifica-
ways, however. Stated, and even informal, exchange tions. The divergence was particularly striking
rate policy intentions may be forward looking but among so-called floating regimes, where only 20
may also be misleading.13 Finally, de facto classifi- percent were de facto free floats while 60 percent
were either intermediate or pegged regimes and an-
9The Natural classification relies on a broad set of descriptive other 20 percent had freely falling currencies.14 Al-
statistics and detailed country chronologies of exchange rate though almost all de jure hard pegs were in fact op-
arrangements to group regimes. As noted by Reinhart and Rogoff, erated as hard pegs, fewer than 40 percent of de jure
this is analogous to natural taxonomic schemes in biology, where soft pegs were de facto pegs, either hard or soft.
species are grouped according to their characteristics. About 60 percent of de jure intermediate regimes ac-
tually operated as intermediate regimes.15
10Technical aspects of the fine and coarse versions of the Nat-
ural classification system are described in Appendix I, which also In the 1970s and 1980s, the differences between
contains a summary comparison of the various regime classifica- actual and stated policies reflected to a large extent
tion systems. the prevalence of dual/parallel foreign exchange
markets. In the early 1970s, almost one-half of all
11In principle, of course, countries with relatively stable ex- countries and one-third of advanced economies had
change rates may have been subject to fewer or smaller other real active dual/parallel markets with exchange rates that
shocks, including policy shocks, or to shocks that happened to
offset the terms-of-trade shocks they experienced. 14Unless otherwise noted, all subsequent references to de
facto regimes and regimes’ actual operations are to the Natural
12For example, de facto classifications (other than that of the classification.
IMF) do not distinguish unsuccessful pegs—those regimes where
the authorities try to peg the exchange rate but are unable to do 15Surprisingly, during the run-up to the European Monetary
so. The IMF de facto classification, by contrast, incorporates Union, all the euro area countries were listed as intermediate
information on policy intentions and, in principle, retains a regimes in the de jure classification until 1999.
forward-looking element.

13This does not mean, of course, that formal announcement of a
de facto regime does not affect macroeconomic performance. In-
deed, as the results described in Section III indicate, the effect of
announcing the true de facto regime has been significant for cer-
tain regimes.

6

Divergence Between Stated and Actual Policies

Figure 2.1. Natural Classification Regimes Figure 2.2. Countries with Dual/Parallel
by De Jure Category, 1973–99 Foreign Exchange Markets

(In percent of annual observations) (In percent of annual observations)

De facto hard peg Managed floating 60
De facto other peg Freely floating 50
Limited flexibility Freely falling 40
30
De jure 20
hard peg 10
0
De jure
other peg 1973 76 79 82 85 88 91 94 97 2000
Sources: Reinhart and Rogoff (2004); and IMF staff estimates.
De jure
intermediate

De jure of mismatches between countries’ classifications in
float the de jure and Natural classifications did not. This
was due mainly to the increase in freely falling
0 20 40 60 80 100 regimes in the 1990s, which included the transition
economies of central and eastern Europe and the for-
Sources: Reinhart and Rogoff (2004); Ghosh, Gulde, and Wolf mer Soviet Union, and the de jure classification
(2003); and IMF staff estimates. of euro area currency regimes as intermediate until
1999.
deviated substantially from official rates (Figure
2.2). Foreign exchange markets have since been uni- The frequency of freely falling regimes is also on
fied in most countries. In emerging markets and a declining trend, despite a brief resurgence follow-
other developing countries, the unification occurred ing the breakup of the Soviet Union. Rogoff (2003)
mainly in the 1990s as capital flows to emerging notes that this in turn reflects the decline in infla-
market economies accelerated and efforts were in- tion across the world in recent years. Hence, ac-
tensified by the international community, including counting for dual/parallel markets and free falling
the IMF, to encourage countries to accept Article regimes, while critical in drawing lessons from the
VIII of the IMF’s Articles of Agreement. Although history of regimes, is less likely to be as relevant in
the number of countries with dual/parallel exchange the future.
rates that deviated substantially from official rates
declined to 9 in 2001 from 30 in 1995,16 the number Differences Across Country Groups

16These data are based on Reinhart and Rogoff (2004), and are As noted, compared to the Natural classification
not identical to the IMF’s classification of unified versus the de jure classification significantly overstates the
dual/multiple rates. By multiple exchange rates, Reinhart and Ro- number of true floats and pegs, suggesting that
goff refer to cases where one or more rates is market determined, fewer countries are at the polar extremes than im-
as opposed to cases where multiple official rates are all fixed and plied by their announcements. Figures 2.3 and 2.4
simply act as a differential tax on a variety of transactions. show that few countries, especially emerging mar-
kets and other developing countries, actually allow

Another important difference is that dual/multiple markets are
typically legal, whereas parallel markets may or may not be legal.

7

II THE EVOLUTION OF EXCHANGE RATE REGIMES: A FRESH LOOK

Figure 2.3. Natural Classification Regime their exchange rates to float freely. Among emerg-
Distribution ing markets, the proportion of de facto free floaters
has remained relatively small at 4–7 percent since
(In percent of annual observations) the mid-1980s (Figure 2.5).17 Even among ad-
vanced economies, only about 20 percent allow
Hard peg Managed floating their currencies to float freely, although close to 40
Other peg Freely floating percent state that they have floating regimes. These
Limited flexibility Freely falling figures also show that fewer countries actually peg
their exchange rates than announcements would
100 suggest. De facto pegs accounted for about one-
third of all de facto regimes in recent years, while
80 de jure pegs comprised about one-half of all de jure
regimes. The number of hard pegs was significantly
60 higher, however, under the Natural classification
than under the de jure.18 While the proportion of de
40 facto pegs has increased slightly since the early
1990s, this mainly reflects the monetary union in Eu-
20 rope and the adoption of pegs by some of the coun-
0 tries that were previously experiencing freely falling
currency values. Interestingly, hard pegs accounted
1940 50 60 70 80 90 2000 for most of the recent increase in pegs in other devel-
Sources: Reinhart and Rogoff (2004); and IMF staff estimates. oping countries, while soft pegs accounted for much
of the increase in emerging markets.
Figure 2.4. De Jure Regime Distribution
Intermediate regimes have been, and continue to
(In percent of annual observations) be, considerably more prevalent than suggested by
the de jure classification. While de jure intermediate
Hard peg Intermediate regimes rose to about a quarter of all exchange rate
Floating regimes in the late 1990s from around 10 percent in
Soft peg the mid-1970s, the proportion of de facto regimes
with an intermediate degree of flexibility has re-
100 mained at about one-half since the mid-1970s.19
Within intermediate regimes, however, managed
80 floats have become more prevalent in emerging mar-

60 17For other developing countries, the increase in de jure floats
in the late 1980s and early 1990s was in reality a rise in freely
40 falling regimes, and part of the decline in free floats since the
mid-1990s reflected a reduction in freely falling currency values
20 as macroeconomic stabilization progressed in many of these
0 countries (e.g., Azerbaijan, Bulgaria, Iran, the Kyrgyz Republic,
and Ukraine).
1970 75 80 85 90 95
Sources: Ghosh, Gulde, and Wolf (2003); and IMF staff estimates. 18The definition of hard pegs differs slightly across classifica-
tions. In the de jure classification, such pegs constitute monetary
unions and currency boards. The Natural classification also in-
cludes preannounced pegs. Of the 43 countries listed as hard pegs
by the Natural classification in 2001, only five had preannounced
pegs, of which only one (Malaysia) was in the emerging markets
group. Excluding preannounced pegs from the hard peg category
does not affect the finding that hard pegs are more prevalent
under the Natural classification than under the de jure. The find-
ing of a general absence of a bipolar tendency among emerging
markets in the 1990s (discussed below) is actually accentuated by
such an adjustment, however.

19Among advanced economies, the proportion of intermediate
de facto regimes expanded sharply around the time of the collapse
of the Bretton Woods system but shrunk steadily in the 1980s and
1990s as the euro area countries moved toward monetary union.
Among emerging market economies and other developing coun-
tries, the proportion of intermediate regimes rose markedly in the
1970s, but has remained relatively flat since then.

8

Divergence Between Stated and Actual Policies

kets over the past decade, while other developing Figure 2.5. Natural Classification Regime
countries have tended to move in the opposite direc- Distribution by Country Group
tion toward more limited flexibility.
(In percent of annual observations for each group)
A historical retrospective using the Natural classifi-
cation also suggests that the breakup of the Bretton Hard peg Managed floating
Woods system was much less of a watershed event for Other peg Freely floating
emerging markets and other developing countries than Limited flexibility Freely falling
for advanced economies. De facto pegs in advanced
economies declined sharply as the Bretton Woods Advanced Countries
system collapsed, while among emerging markets and 100
other developing countries the decline in pegs was
more gradual and continued through the 1980s.20 80

Even when compared with other de facto classifi- 60
cations, the Natural classification records fewer
regimes near the polar extremes of full flexibility 40
and rigid pegs. At a broad level, the IMF de facto
classification yields similar results to the Natural 20
classification—two-thirds or more of Natural classi-
fication free floats, pegs, and intermediate regimes 0
are classified the same way by the IMF de facto clas-
sification. The IMF classification, however, picks up 1940 50 60 70 80 90 2000
many more free floats than the Natural classifica-
tion, especially among emerging markets, where as Emerging Markets
many as one-third were listed as freely floating 100
regimes in 2001 (Figure 2.6).21 Similarly, the preva-
lence of pegs is higher than in the Natural classifica- 80
tion, especially for other developing countries, of
which about half were listed as pegged regimes in 60
2001.22 The Levy-Yeyati–Sturzenegger de facto
classification also records many more free floats and 40
pegs and, consequently, many fewer intermediate
regimes than the Natural classification (Figure 2.7). 20
Surprisingly, over half of emerging markets are clas-
sified as floats in the Levy-Yeyati–Sturzenegger 0 50 60 70 80 90 2000
classification in the late 1990s, both before and after Developing Countries
the Asian crises, and free floats are more prevalent 1940
than in the de jure classification, drawing into ques-
tion the degree to which the former presents a more 100
accurate picture of actual regimes than the latter.
80
20As the prevalence of de facto pegged regimes has evolved,
the choice of anchor currency among peggers has undergone sig- 60
nificant change, with virtually all peggers now anchoring to either
the dollar or the euro (Box 2.1). 40

21Of all the observations classified as free floats by the IMF de 20
facto regime that were also classified by the Natural classification,
only about 27 percent were classified by the latter as free floats, 0
while 18 percent were freely falling regimes, 33 percent were
managed floats, 18 percent were limited flexibility regimes, and 3 1940 50 60 70 80 90 2000
percent were pegs. About 30 percent of the IMF de facto free
floats were not classified by the Natural classification, usually be- Sources: Reinhart and Rogoff (2004); and IMF staff estimates.
cause qualitative evidence suggested the presence of a significant
parallel market, but parallel exchange rate data were not available.
That said, Bubula and Ötker-Robe (2002) also find, using the IMF
de facto classification, that intermediate regimes have been more
prevalent than suggested by the de jure classification.

22Among advanced countries, however, euro area economies
are listed as limited flexibility regimes rather than pegs in the
IMF de facto classification, as they were listed in the de jure clas-
sification, until 1999 (until 2001 for Greece).

9

II THE EVOLUTION OF EXCHANGE RATE REGIMES: A FRESH LOOK

Box 2.1. Anchor Currency Choice

While there is a large empirical literature on the con- 1990–2001
ditions under which countries adopt fixed or floating
regimes (discussed in Appendix II), less has been writ- The overall distribution of anchor currencies did not
ten on the determinants of anchor currency choice. The change much in the 1990s, apart from the introduction
question of interest is: Once countries choose to peg of the euro in 1999. The behavior of transition
their exchange rates to an anchor currency—including economies during this period, however, is illustrative of
by means of crawling pegs or bands—what determines the dynamics of anchor currency choice. Following the
the choice of this anchor? breakup of the Soviet Union in the early 1990s, most
transition economies fell initially in the freely falling
The theory of optimal currency areas suggests that category for several years, and then increasingly started
countries benefit from adopting the same anchor as a tying their currencies to the deutsche mark or the U.S.
trade partner, because this reduces their bilateral ex- dollar. Interestingly, the choice of anchor was almost
change rate variability. Meissner and Oomes (2004) perfectly divided among regional lines: while Central
provide empirical evidence of these network externali- and Eastern European countries chose to anchor to the
ties. The authors find that, after controlling for other deutsche mark, and later to the euro, most former So-
factors—such as country size, openness, and colonial viet Union republics chose the U.S. dollar as their an-
history—the probability of choosing a particular anchor chor—with the exception of Estonia, which adopted a
currency increases with the amount of trade with other currency board arrangement with the deutsche mark,
countries that use this same anchor. These externalities and Latvia, which chose the SDR. As Meissner and
may explain why virtually all countries that have cho- Oomes (2004) show, this divide between the euro and
sen to peg their exchange rates in some way to another the dollar cannot be explained solely on the basis of
currency have converged over the last 50 years to using trade flows with Europe or the United States but is par-
either the U.S. dollar or the euro as their anchor cur- tially the result of network externalities arising from
rency (see figure below). trade partners’ anchor currency choices.

1940–72 Anchor Currency Choices

Between 1940 and 1972, the U.S. dollar was the most (In percent of annual observations)
popular anchor currency chosen by advanced countries,
followed by the British pound and the German deutsch U.S. dollar Euro
mark. For developing countries, the predominant anchor
currencies were the U.S. dollar, the British pound, and Pound sterling Deutsche mark
the French franc, with the latter two choices being deter-
mined largely by colonial history. French franc Other

1973–89 100

Following the collapse of the Bretton Woods sys- 80
tem, the British pound disappeared entirely from the
menu of anchor choices. Pegs to the U.S. dollar de- 60
clined in popularity among advanced countries as an
increased number of free and managed floaters 40
emerged, and the majority of advanced countries that
retained pegs ended up tying their currencies in some 20
form to the deutsche mark, and later to the euro. De-
veloping countries largely switched to using the U.S. 0
dollar as anchor, except the group of former French
colonies that continued to peg to the French franc. 1940 50 60 70 80 90 2000

Source: Meissner and Oomes (2004).

Bipolar Hypothesis and Fear of Floating regimes or move to hard pegs.23 That speculative at-
tacks against hard pegs were rare and could appar-
The Natural classification raises questions about ently be warded off seemed to lend support to the
the general validity of the bipolar hypothesis. Start-
ing in the mid-1990s, some observers had predicted 23For example, Eichengreen (1994, pp. 4–5) argues that coun-
that emerging market countries would, over time, tries “will be forced to choose between floating exchange rates on
move to the polar extremes of exchange rate flexibil- the one hand and monetary unification on the other.” Obstfeld and
ity; that is, they would either adopt freely floating Rogoff (1995, p. 74) claim that for countries with an open capital

10

Divergence Between Stated and Actual Policies

hypothesis.24 The increase in free floats and hard Figure 2.6. IMF De Facto Regime
pegs since 1990 in the de jure—and to a smaller ex- Distribution
tent in the IMF de facto classifications, as illustrated
in Figures 2.4 and 2.6, respectively—appeared to (In percent of annual observations)
support the bipolar view. As noted above, however,
the Natural classification indicates that there has Hard peg Managed floating
been no “hollowing out of the middle.” While a few Other peg Freely floating
emerging markets indeed moved in the 1990s to de Limited flexibility
facto hard pegs (Argentina and Malaysia) or
free floats (Indonesia, Korea, and South Africa), just 100
as many transitioned from freely falling to interme-
diate regimes (Brazil, Peru, Poland, Russia, and 80
República Bolivariana de Venezuela).25 As a result,
the middle remained as large as it was a decade ago. 60
Moreover, transitions since 1990 to de facto pegs
among emerging markets have been more in the soft 40
category (China, Egypt, Jordan, and Peru) rather
than the hard category.26 20

The tendency of countries to allow less exchange 0 1990 92 94 96 98 2000
rate flexibility in practice than in policy statements is
consistent with the fear of floating. As Calvo and Sources: Bubula and Ötker-Robe (2002); and IMF staff estimates.
Reinhart (2002) argue, fear of floating—a reluc-
tance to allow exchange rates to fluctuate freely— Figure 2.7. Levy-Yeyati–Sturzenegger
could arise for various reasons, including policy Regime Distribution
credibility concerns; fear of Dutch disease in case of
large appreciations; and fear of inflation, currency (In percent of annual observations)
mismatches, and/or balance sheet effects (on ac-
count of high liability dollarization) in case of large Peg Managed floating
Freely floating
account, “there is little, if any, comfortable middle ground be- Limited flexibility
tween floating rates and the adoption of a common currency.”
More recently, Summers (2000, p. 8) argued that, for economies 100
with access to international capital markets, “the choice of appro-
priate exchange rate regime . . . increasingly means a move away 80
from the middle ground of pegged but adjustable fixed exchange
rates towards the two corner regimes.” Fischer (2001, p. 22) con- 60
cluded on the basis of the IMF de facto classification that “In the
last decade, there has been a hollowing out of the middle of the 40
distribution of exchange rate regimes in a bipolar direction, with
the share of both hard pegs and floating gaining at the expense of 20
soft pegs.”
0
24According to the Natural classification, Brazil, Korea, and
Malaysia had limited flexibility regimes prior to their recent capi- 1974 77 80 83 86 89 92 95 98
tal account crises, while Mexico, the Philippines, and Thailand
had de facto pegs (but not hard pegs) before their respective Sources: Levy-Yeyati and Sturzenegger (2003); and IMF staff
crises. Russia was not classified in 1997–98, while Argentina was estimates.
classified as a hard peg through 2001. Of all the major recent cri-
sis countries, only Turkey had a managed floating regime prior to
its crisis.

25Hernandez and Montiel (2001) argue that, while several
Asian countries have increased the flexibility of their exchange
rates in the postcrisis period, generally they have not adopted
truly freely floating regimes.

26Peru was classified as a de facto soft peg during 1999–2001
by the Natural classification on the basis of a two-year rather than
a five-year window to allow for a possible structural break in the
variability of the exchange rate toward the end of the sample pe-
riod. Peru would fall just short of the criteria for a de facto peg in
1999–2001 if a five-year window, which would also span the pe-
riod prior to 1999, were used.

11

II THE EVOLUTION OF EXCHANGE RATE REGIMES: A FRESH LOOK

depreciations.27 As Figure 2.1 indicates, the vast ma- Figure 2.8. Natural Classification Regime
jority of countries that say they float actually do not. Transitions
Moreover, many countries that say they have inter-
mediate regimes in fact have de facto pegs. (Number of transitions)

Regime Transitions To/from freely To more flexible To less flexible
regime
Major global and regional events have influenced falling regime
exchange rate regime transitions. The collapse of 90 2000
the Bretton Woods system in 1973 was, of course, 16
the outcome of pressures built up in a relatively Advanced Countries 90 2000
rigid system of exchange rate regimes and was fol-
lowed by a sharp increase in flexible arrangements 14 90 2000
(Figure 2.8). The debt crisis of the 1980s and the
transformation of the economies of Central and 12
Eastern Europe and the former Soviet Union in the
early 1990s were also accompanied by a relatively 10
high frequency of regime transitions, especially
into and subsequently out of the freely falling 8
category. In the latter half of the 1990s, as several
large emerging markets faced external financing 6
crises, the frequency of exchange rate regime transi-
tions among this group rose once again. Then in 4
1999, a major transition occurred among advanced
economies with the adoption of a monetary union in 2
the euro area. 0

Once the transitions into and out of the freely 1940 50 60 70 80
falling category—as well as those that occurred as a
result of global events—are distinguished, it turns 7 80
out that the frequency of changes in exchange rate Emerging Markets
regimes today is remarkably similar to that of 50
years ago. As Figure 2.8 illustrates, the average 6
number of countries transitioning to a different 5
regime (excluding transitions into and out of the 4
freely falling category) in any given year since the 3
collapse of the Bretton Woods system was about the 2
same as during the Bretton Woods period. 1
0
Thus, the interesting finding is that countries have
changed their de facto exchange rate regime rela- 1940 50 60 70
tively infrequently. On the basis of data going back
to the 1940s, about 7 percent of all countries transi- 25
tioned to a different regime in an average year, and Developing Countries
the typical exchange rate regime had a duration of
about 14 years (Table 2.1). If the 1970–75 period is 20
excluded and Eastern and Central European and for-
15
27See also Reinhart (2000). Hausmann, Panizza, and Stein
(2001) find that exchange rate volatility declines with the de- 10
crease of amounts countries can borrow internationally in their
own currency, which the authors consider an indicator of a coun- 5
try’s ability to avoid currency mismatches. The extent of ex- 0
change rate pass-through turns out to be less significant. Alesina
and Wagner (2003) identify conditions under which countries de- 1940 50 60 70 80
clare a de jure float but, because of fear of floating, restrict ex-
change rate flexibility. Sources: Reinhart and Rogoff (2004); and IMF staff estimates.

mer Soviet Union countries along with the euro area
countries are removed from the sample, transitions
were even less frequent. In the adjusted sample, the
average regime duration rises to just over 16 years,
while the proportion of countries changing regime in
any given year declines to about 6!/4 percent.

De facto pegged regimes have tended to change
less frequently and last longer than other regimes.
For all de facto pegs since 1940, the probability of
exiting to a different regime in any given year was

12

Regime Transitions

Table 2.1. Annual Transition Probabilities

(Historical rate of regime transitions, in percent)

Natural Classification 7.0
All countries, 1940–2001 3.5
Pegs only
All countries, adjusted sample, 1940–2001 6.2
(excluding 1970–75)1 2.5
Pegs only 7.0
Advanced economies, 1940–2001 5.1
Pegs only 9.7
Emerging markets, 1940–2001 6.7
Pegs only 14.4
Emerging markets, 1989–2001 9.8
Pegs only 6.1
Developing countries, 1940–2001 2.4
Pegs only
6.8
De jure Classification2 4.9
All countries, 1973–2001
Pegs only

Source: Reinhart and Rogoff (2004); Ghosh, Gulde, and Wolf (2003); and IMF staff estimates.
1Excludes euro area and former command economies in Europe and the former Soviet Union.
2Natural classification transition rates for all regimes and pegs over the same period were 9.3 percent and 5.1
percent, respectively.

about 3!/2 percent.28 Since the Natural classification the Bretton Woods system accounted for a sizable
classifies as pegs only those that are successful, portion of such exits during this period.30
countries that attempt to peg but are able to sustain
the peg only briefly tend not to be classified as pegs. Emerging markets, however, have tended to switch
This, together with the fact that the Natural classifi- regimes more frequently, and have gone into the
cation does not treat onetime devaluations followed freely falling category more often, than other coun-
by a re-peg as a change in the longer-term regime,29 tries. Since 1940, the annual regime transition rate
reduces the observed exit rate from de facto pegs. It among emerging markets has averaged about 10 per-
is also worth noting that regime transitions are less cent, compared with 7 percent for advanced countries
frequent in the de jure classification than in the Nat- and about 2 percent for other developing countries.
ural classification, suggesting that countries tend to On average, about 3 percent of emerging markets, ex-
change their stated exchange rate policy objectives cluding those already in the freely falling category,
even less frequently than their de facto exchange rate have transitioned to a freely falling regime every year.
policies. The average annual exit rate from de facto By contrast, only 0.5 percent of all advanced coun-
and de jure pegs during 1973–99 has been about the tries and less than 2 percent of other developing coun-
same; however, this is partly because the collapse of tries have switched to a freely falling regime in any
given year. The transition rate out of pegged regimes
among emerging markets has also been higher (about

28These conclusions contrast with the results obtained by Klein 30Masson (2001) obtains very similar results for regime transi-
and Marion (1997), Eichengreen and others (1998), and Dut- tion rates and regime duration using the Ghosh and others (1997)
tagupta and Ötker-Robe (2003) among others, who find the classification, but finds that transitions using the Levy-Yeyati–
longevity of pegs to be much shorter. This is mainly because the Sturzenegger classification are considerably more frequent. Mas-
Natural classification attempts to identify longer-term regimes son suggests that the difference in historical transition rates may
rather than short-term “spells,” which are analyzed in the other arise from sampling problems—a fair number of observations are
studies. inconclusive in the Levy-Yeyati–Sturzenegger data and thereby
omitted—and methodological differences that tend to accentuate
29For example, the Natural classification does not treat the de facto flexibility (and hence transition rates) in the Levy-
1994 CFA franc devaluation as a change in regime. By contrast, Yeyati–Sturzenegger algorithm in periods of heightened ex-
the Levy-Yeyati–Sturzenegger classification, which uses a one- change market pressures. Using the IMF de facto classification,
year horizon to measure the variability of the official exchange Bubula and Ötker-Robe (2002) conclude that intermediate
rate, picks up significantly more transitions: for example, a regimes are unlikely to disappear in the future.
switch from peg to “dirty float” for each of the CFA franc zone
countries in 1994 with a switch back to peg in 1995.

13

II THE EVOLUTION OF EXCHANGE RATE REGIMES: A FRESH LOOK

7 percent) than in advanced and other developing The persistent popularity of intermediate regimes—
countries (5 percent and 2!/2 percent, respectively).31 especially among emerging markets and other devel-
oping countries—as identified by the Natural classifi-
If historical transition rates continue and no further cation, suggests that such regimes may provide
major global shocks occur, intermediate regimes will important advantages. Indeed, the absence of a gen-
remain prevalent in the future, and the overall distrib- eral bipolar tendency may indicate that intermediate
ution of de facto regimes will be similar to that at regimes are able to capture some of the benefits of
present. Given that pegs have had a somewhat longer both extremes while avoiding many of the costs.
average duration than other regimes in the past, the
historical transition rates imply that the proportion of Finally, the relatively long average duration of
pegs could increase slightly over time. Similarly, Natural classification regimes may suggest that
since relatively few countries, especially developing regime transitions involve significant costs. The
countries, have had true free floats in the past, the higher transition rates for emerging markets indi-
historical likelihood of transitioning into a free float cate, however, that either these costs decline as
has been low, implying that the share of free floats countries experience higher capital flows or, more
among all regimes is likely to remain modest in the likely, that higher capital flows in the absence of ad-
future. As other developing countries become in- equate financial infrastructure and safeguards make
creasingly integrated into global financial markets, it harder to sustain regimes, particularly pegged
however, their regime transitions may well resemble regimes. Again, evidence in support of this channel
those seen among emerging markets during the may be obtained potentially by assessing the (histor-
1990s. In that case, the proportion of pegged regimes ical) likelihood of crises under alternative exchange
among developing countries will tend to decline rate regimes across different types of economies.
gradually in the future, while managed floats and free
floats will gradually increase. Over the longer term, Appendix I.
of course, political economy considerations may The Natural Classification
guide regime choice decisions in some countries. For
example, some may choose to join currency unions in This appendix summarizes the data and algorithm
the not-so-distant future. Prospects for transitions of used to construct the Natural classification and pro-
that nature cannot be assessed on the basis of histori- vides a brief summary of the main features of vari-
cal transition rates, however, and are clearly beyond ous de facto classifications (see Table A2.1).
the scope of this analysis.
The Natural classification, which classifies ex-
Implications for Assessing change rate regimes into fine and coarse categories (as
Regime Performance summarized in Table A2.2), employs monthly data on
official and market-determined exchange rates for the
Empirical analysis seeking to uncover the link be- period 1940–2001.33 The data on market-determined
tween countries’ exchange rate regimes and their exchange rates are drawn from various issues of
macroeconomic performance depends critically on Pick’s Currency Yearbook, Pick’s Black Market Year-
how regimes are classified. The wide variation be- book, and Pick’s World Currency Report, while the
tween countries’ stated exchange rate regimes and official rate data are from the same sources as well as
their actual practice suggests that results obtained by the IMF’s International Financial Statistics. The
employing the de jure classification could be off the quotes are end-of-month exchange rates. Annual clas-
mark and that use of a classification that more accu- sifications are simply the modal monthly classifica-
rately captures true regime flexibility can lead to dif- tions for each country in each year.
ferent conclusions. The Natural classification, with
its special features and historical coverage, is a The procedure employed by the Natural classifica-
promising candidate for such analysis.32 tion to classify regimes is as follows:

31These calculations do not treat switches within the pegged As discussed in Appendix II, however, it is difficult to find empir-
category (e.g., from hard to other pegs) as a transition. The aver- ical regularities between a large set of potential determinants of
age duration of pegs in other developing countries is strongly af- regime choice—including standard measures of the broader pol-
fected by the CFA franc zone countries, many of which have re- icy context—and between countries’ actual regimes.
tained de facto pegs throughout the sample period.
33While data on market-determined exchange rates are avail-
32The issue of causation affects potentially the analysis of able only for the period 1946–98, Reinhart and Rogoff (2004)
regime performance: better macroeconomic performance may be were able to classify most countries for the years 1940–45 and
associated with certain regimes because countries with strong 1999–2001 on the basis of official exchange rate data only be-
performance may choose systematically to adopt those regimes. cause few countries had active parallel markets in those years.
Observations where the parallel market was known to be substan-
tial but where parallel rate data were not available are marked
“unclassified” by the Natural classification.

14

Appendix I

Table A2.1. Main Features of Various De Facto Classifications

Ghosh, Gulde, IMF (1999, 2003b); Bubula Levy-Yeyati and Reinhart and
and Wolf (2003) and Ötker-Robe (2002) Sturzenegger (2003) Rogoff (2004)

Period 1973–99 1990–present 1974–2000 1940–2001

Frequency Annual Annual and monthly Annual Annual and monthly

Number of countries 165 190 156 153

Number of regime types 25 fine, 9 coarse 15 fine, 8 coarse 4 14 fine, 5 coarse

Advantages Uses quantitative and Uses quantitative and Uses information on Uses dual/parallel exchange
volatility of foreign rate information
qualitative information qualitative information exchange reserves
Separates freely falling
(survey of IMF desk (survey of IMF desk Systematic approach; episodes
no judgment needed
economists) economists; discussions Long time series; monthly
exchange rate movements to
with authorities; news identify regime

Fine taxonomy articles; press reports) Systematic approach; no
judgment needed
All IMF member
countries classified;
classification
continuously updated

Disadvantages Relies to large extent Requires subjective Exchange rate stability Exchange rate stability may
on stated policy judgment, which may
intentions, which may differ across countries or reserve changes occur for reasons other than
deviate substantially and over time
from actual practice may occur for policy intervention

reasons other than

policy intervention A few countries are not

classified for all years

Requires subjective Reserves data may
judgment, which may
differ across countries not cover derivatives
and over time
Many observations
Not all countries are not classified—only
classified for all time 15 years per country
periods classified on average

Other countries
affect classification
(due to cluster
analysis)

First, a separation is made between countries with is verified according to rules analogous to those de-
either official dual or multiple rates or active parallel scribed in step 3 below, it is then classified accord-
(black) markets. ing to the announcement.35

If there is no dual or black market, a check is If there is no preannounced exchange rate path, if
done to see if there is an official preannounced the announced regime cannot be verified by the data
arrangement, such as peg, crawling peg, or band. If (which is often the case), and if the 12-month rate of
there is, the announced regime is verified by exam- inflation is below 40 percent, the regime is classi-
ining the mean absolute monthly change over the fied on the basis of actual exchange rate behavior as
period following the announcement.34 If the regime follows:

34The advantage of using mean absolute deviations, rather than 35When the announced regime is a peg to an undisclosed basket
variances or standard deviations, is that this minimizes the impact of currencies, tests are done to see if the basket peg is really a de
of outliers. For example, when the exchange rate is fixed but sub- facto peg to a single dominant currency (or to the SDR). If no
ject to periodic large devaluations, the variance or standard devia- dominant currency can be identified, the episode is not labeled as
tion would overstate the extent of exchange rate flexibility in the a peg. While this suggests that the Natural classification could
period around the devaluation. miss some de facto basket pegs, Reinhart and Rogoff (2004)
argue that this is almost certainly not a major issue.

15

II THE EVOLUTION OF EXCHANGE RATE REGIMES: A FRESH LOOK

Table A2.2. Natural Classification Categories

Fine Coarse Description

11 No separate legal tender
21 Preannounced peg or currency board arrangement
31 Preannounced horizontal band that is narrower than or equal to ±2 percent
41 De facto peg
52 Preannounced crawling peg
62 Preannounced crawling band that is narrower than or equal to ±2 percent
72 De facto crawling peg
82 De facto crawling band that is narrower than or equal to ±2 percent
93 Preannounced crawling band that is wider than or equal to ±2 percent
10 3 De facto crawling band that is narrower than or equal to ±5 percent
11 3 Moving band that is narrower than or equal to ±2 percent (i.e., allows for both
appreciation and depreciation over time)
12 3 Managed floating
13 4 Freely floating
14 5 Freely falling
15 6 Dual market in which parallel market data are missing

• If the absolute monthly percent change in the ex- If the 12-month rate of inflation exceeds 40 per-
change rate is equal to zero for four consecutive cent, the episode is classified as freely falling.37
months or more, that episode is classified (for
however long it lasts) as a de facto peg, if there The remaining regimes—those that have not al-
are no dual or multiple exchange rates in place.36 ready been classified by steps one through four—be-
come candidates for managed or freely floating. To
• If the probability is 80 percent or higher that the distinguish between the two, the degree of exchange
monthly exchange rate change remains within a rate flexibility is measured by a composite statistic.
plus/minus 1 percent band over a rolling five-
year period, then the regime is classified as a de Appendix II.
facto peg or crawling peg over the entire five- Determinants of Exchange Rate
year period. If the exchange rate has no drift, it is Regime Choice
classified as a fixed parity; if a positive drift is
present, it is labeled a crawling peg; and, if the The Natural classification data show some links
exchange rate also goes through periods of both between de facto regime flexibility and certain
appreciation and depreciation, it is a moving macroeconomic and financial variables, such as
peg. trade openness and dollarization. A review of the lit-
erature suggests, however, that it is difficult to find
• The approach regarding de facto bands, as well as empirical regularities between potential exchange
preannounced bands, follows a parallel two-step rate regime determinants and actual regimes that
process. Thus, if there is more than an 80 percent hold consistently across all countries, time periods,
probability that the monthly exchange rate change and regime classifications. Systematic robustness
remains within a plus/minus 2 percent band over a checks of the determinants of regime choice employ-
rolling five-year period, then the regime is classi- ing the Natural classification support this result.
fied as either a de facto narrow band, a narrow
crawling band, or a moving band throughout the Macroeconomic and Financial
entire period during which it remains continu- Characteristics of Regimes
ously above the 80 percent threshold.
Optimum currency area (OCA) theory holds that
36This allows for the identification of relatively short-lived de variables, such as large size and low openness to
facto pegs as well as those with a longer duration. For instance,
this exercise allowed for identification of the Philippines’ de facto 37In the rare cases where inflation is over 40 percent but the
peg to the U.S. dollar during 1995–97 in the run-up to the Asian market rate nevertheless follows a confirmed, preannounced band
crisis, as well as the numerous European de facto pegs to the or crawl, the preannounced regime takes precedence.
deutsche mark prior to 1999.

16

Appendix II

trade, are likely to be associated with floating ex- Empirical Findings on Factors
change rates. One reason for this may be that trade
openness raises the transactions benefits from com- Affecting Regime Choice
mon currencies, and should be expected to lead,
therefore, to a decline in the number of independent Systematic prediction of exchange rate regime
currencies. The data appear to support the OCA the- choice is elusive. A review of a reasonably broad col-
ory prediction that countries that trade a lot will tend lection of previous studies shows that different em-
to have less flexible exchange rate regimes. Ad- pirical studies using the de jure and other de facto
vanced economies that have a high trade openness regime classifications have often obtained different
ratio have tended to have pegged regimes, while the results, suggesting that it is very difficult to draw
prevalence of free floats has been notably higher in general conclusions about how countries choose their
advanced countries with low external trade ratios, exchange rate regimes. Although certain characteris-
such as Australia, Japan, and the United States. A tics have been shown to be important in determining
similar pattern holds among other developing coun- exchange rate regime choice in some groups of coun-
tries, where the prevalence of managed floats has tries, and certain characteristics may distinguish
been markedly higher and pegs significantly lower countries in some regimes from those in different
in the countries that rely less on external trade. The regimes, no result appears fully robust to changes in
pattern among emerging markets has been less clear, country coverage, sample period, estimation method,
although relatively closed economies in this group and exchange rate regime classification.
have had a much higher likelihood of being in the
freely falling category. Several empirical studies have analyzed the deter-
minants of exchange rate regime choice in a cross
Higher dollarization appears to be associated with section of countries. Among the first studies of this
less flexible exchange rate regimes among emerging kind are Heller (1978), which analyzes the determi-
markets, consistent with fear of floating. Fear of nants of exchange rate regimes with data from the
floating appears to be stronger in highly dollarized mid-1970s, soon after the generalized floating that
emerging markets, where pegged regimes are more followed the breakup of the Bretton Woods system,
prevalent, than in less-dollarized countries in the as well as Dreyer (1978); Holden, Holden, and Suss
group. Conversely, emerging markets with low and (1979); Melvin (1985); Bosco (1987); Savvides
medium degrees of dollarization are more likely to (1990); Cuddington and Otoo (1990 and 1991);
have managed or freely floating regimes. Fear of Rizzo (1998); and Poirson (2001). Some studies,
floating does not explain, however, why other devel- such as those by Collins (1996), Edwards (1996 and
oping countries with high dollarization ratios appear 1999), and, more recently, Frieden, Ghezzi, and
to prefer regimes with limited flexibility to pegs. A Stein (2001), have used random effects panel data to
possible explanation for this could be that many of analyze also the determinants of changes in ex-
these countries became highly dollarized following a change rate regime. As such, they can be seen as
freely falling episode and lacked the credibility nec- somewhat related to the recent literature on predict-
essary to defend a peg. A regime with limited flexi- ing exchange rate crises. Nevertheless, these studies
bility allowed them to obtain the benefits of a rela- are included in this review because they report find-
tively stable currency, while at the same time ings on the role of country characteristics that are
maintaining some ability to adjust to shocks. relatively stable over time (such as openness) in de-
termining exchange rate regime choice. Another re-
There is little systematic relation, however, in the cent study, by Berger, Sturm, and de Haan (2000),
degree of capital account openness across de facto uses panel data in an attempt to identify the long-run
regimes. Emerging markets and other developing determinants of exchange rate regime choice. Addi-
countries tend to have more capital controls and tional studies addressing changes in exchange rate
lower capital flows in relation to GDP than advanced regimes include Masson (2001), Klein and Marion
economies. Nevertheless, the variation in capital ac- (1997), and Duttagupta and Ötker-Robe (2003).
count openness does not appear to be related to the
flexibility of countries’ currency regimes. Among The vast majority of previous studies have at-
advanced economies, the volume of capital flows in tempted to explain countries’ de jure exchange rate
countries with de facto pegged regimes tends to be regime choice. A few studies have constructed and
higher than in those with intermediate regimes and used measures of the degree of de facto flexibility
significantly higher than in those with freely floating on the basis of the actual observed volatility of ex-
regimes. The relationship is more mixed, however, change rates and reserves, including Holden,
for emerging markets and other developing coun- Holden, and Suss (1979) and, more recently, Poir-
tries, possibly because capital controls are often in- son (2001). Table A2.3 summarizes the approaches
effective, so the expected inverse relation between and findings of these studies with regard to the im-
controls and observed capital flows may not hold. pact of several variables on observed exchange rate
regime choice. Most studies considered some of the

17

II THE EVOLUTION OF EXCHANGE RATE REGIMES: A FRESH LOOK

Table A2.3. Studies on Determinants of Exchange Rate Regimes (Likelihood to Float)

Author Cuddington Honkapohja

Heller Dreyer Holden, Holden, Melvin Savvides and Otoo and Pikkarainen
(1978) (1978) and Suss (1979) (1985) (1990)
(1990, 1991) (1994)

Sample 86 countries 88 developing 76 countries 64 countries 39 developing 66 countries 125 countries

countries countries

Time frame 1976 1976 1974–75 1976–78 1976–84 1980, 83, 86 1991

Methodology Discriminant Probit OLS on a Multinomial Two-stage Ordered/ Logit and
analysis probit non-ordered probit
continuous logit
Mult./bin.
measure logit

Explanatory variables

(OCA factors)1 – – ^^ – ^^ • – ±^ –^
Openness + + ^^ + –^
Economic development + + ^^ +^ +
Size of economy – ^^
Inflation differential – + + ^^ + – ±
Capital mobility + +^
Geographical trade + ^^ + +

concentration –
International financial
– ^^ – –
integration

(Other macro/external/ + ^^
structural factors)1 – ^^ – ^^

Growth – ^^
Negative growth + ^^
Inflation
Moderate-to-high inflation
Reserves
Capital control
Terms-of-trade volatility
Variability in export

growth
External variability/

openness
Real exchange rate

volatility
Product diversification
Current account
External debt
Growth of domestic

credit
Money shocks
Foreign price shocks

(Political/historical
factors)1

Political instability
Central bank independence
Party in office has majority
Number of parties in

coalition
Coalition government

18

Appendix II

Table A2.3 (concluded)

Author Frieden, Berger,

Collins Edwards Edwards Rizzo Ghezzi and Sturm and de Poirson
(1996) (1996) (1999) (1998) (2001)
Stein (2000) Haan (2000)

Sample 24 Latin 63 countries 49 developing 123 countries 26 Latin 65 developing 93 countries
American countries
American and middle countries

and Caribbean income

countries

Time frame 1978–92 1980–92 1980–92 1977–95 1960–94 1980–94 1990–98

Methodology Probit (panel) Probit (panel) Probit (panel) Probit Ordered logit Probit (panel) Ordered probit
(panel)

Explanatory variables

(OCA factors)1 +^ +^ – ^^ + ^^ –
Openness • + ^ – ^^ ± ^ ±
Economic development + ^^ – ^^ + +^
Size of economy + ^^ + ^^ +^ +
Inflation differential +^
Capital mobility – ^^ – ^^ + ^^ – ^^
Geographical trade +^ +^
+ ^^ + ^^ – ^^ +^ –^
concentration + ^^ + ^^ + ^^ + –^
International financial +^
– ^^ – ^ ± ^^
integration ±^ +^

(Other macro/external/ + ^^ + ^^
+ ^^ +
structural factors)1 – ^^ – ^^
+ ^ + ^^
Growth
± ^^
Negative growth –^ +^
+ ^^ + ^
Inflation
+ ^^ + ^^
Moderate-to-high –^ –^
++
inflation + ^^ ––

Reserves

Capital controls

Terms-of-trade volatility

Variability in export

growth

External variability/

openness

Real exchange rate

volatility

Product diversification

Current account – ^^

External debt

Growth of domestic

credit

Money shocks

Foreign price shocks

(Political/historical
factors)1

Political instability
Central bank independence
Party in office has majority
Number of parties in

coalition
Coalition government

1+ indicates that the coefficient of explanatory variable is positive and – that it is negative; ± indicates the coefficient is either positive or negative de-
pending on the specification or method used; ^^ indicates the coefficient is statistically significant in most cases; ^ indicates the coefficient is statistically
significant in some specifications; and • indicates not significant but sign not reported by the author.

19

II THE EVOLUTION OF EXCHANGE RATE REGIMES: A FRESH LOOK

optimum currency area variables, such as trade two studies, and not significantly associated with
openness (typically measured as imports plus ex- any particular exchange rate regime by another three
ports, divided by GDP), the size of the economy studies.
(gross domestic product in common currency), the
degree of economic development (GDP per capita), There are a few possible exceptions, notably size
and geographical concentration of trade (the share of the economy and inflation. Size of the economy
of trade with the country’s main partner). Among turns out to be positively associated with floating in
macroeconomic variables, several studies included almost all studies, though not always significantly.
inflation (whether the country’s own inflation or in- Inflation is almost always positively and signifi-
flation in excess of partner countries) and foreign cantly associated with floating. In the case of infla-
exchange reserves. Many studies included an indi- tion, however, there are serious questions regarding
cator of either capital controls, which were typically the functional form of the relationship. In a number
also drawn or constructed from the IMF’s Annual of studies, the authors use the inflation rate or the in-
Report on Exchange Arrangements and Exchange flation differential rather than their logarithms or
Restrictions, or de facto capital openness (e.g., the similar transformations, leaving open the possibility
ratio of foreign assets of the banking system to the that the results might be driven by a few influential
money supply). Some studies included measures of observations. Moreover, Collins (1996) finds that
volatility of domestic output, exports, domestic high inflation affects exchange rate regime choice in
credit, or the real exchange rate, although no two the opposite direction than that of low/moderate in-
studies seem to have looked at the same measure of flation, and significantly so.
volatility. A few studies considered variables related
to political economy or institutional strength. Most New empirical tests using the Natural classifica-
studies analyzed some variables that were not in- tion confirm that it is difficult to explain how coun-
cluded in any preceding (or subsequent) studies. tries choose their exchange rate regimes on the basis
Collectively, the studies considered more than 30 of simple empirical regularities. These results are
potential determinants of exchange rate regime consistent with previous work based on other ex-
choice. (Only the variables considered by more than change rate regime classifications (Juhn and Mauro,
one study are included in Table A2.3.) 2002). For a number of potential determinants of
regime choice—including economic size, trade
No result appears to be reasonably robust to openness, and capital controls—the variation across
changes in country coverage, sample period, estima- regimes is statistically significant. With the possible
tion method, and exchange rate regime classifica- exception of economic size and trade openness,
tion. For example, openness—the most frequently however, none of the variables is consistently signif-
analyzed variable—is found to be significantly asso- icant across varying specifications in probit and
ciated with floating regimes by three studies, signifi- multinomial logit regression analysis. This suggests
cantly associated with fixed exchange rates by three that the macroeconomic, structural, and institutional
studies, and not significantly associated with any variables postulated in various theories are not ro-
particular exchange rate regimes by another five bust predictors of exchange rate regime choice. Of
studies. Per capita GDP is found to be significantly course, this does not preclude the potential impor-
associated with floating regimes by three studies, tance of certain variables for specific groups of
significantly associated with fixed exchange rates by countries, in certain time periods, or across some of
the regime categories.

20

III Regime Performance: Inflation and
Business Cycles

H ow does economic performance differ across could have adverse consequences. Case studies
exchange rate regimes? Since theoretical pre- illustrate a variety of approaches to achieving
dictions are varied and often conflicting, this section greater flexibility.
explores the question empirically for the period
1970 to 1999 using the Natural classification of de • In advanced countries, free floats registered
facto exchange rate regimes. This section offers an faster growth than other regimes without incur-
overarching conclusion while recognizing the limita- ring higher inflation. This benefit may reflect the
tions of such analyses—in particular, the possibility typically more pronounced nominal rigidities in
that economic performance influences the choice of mature economies, giving flexible exchange
regimes as much as regimes influence performance rates an important role in reallocating resources
and that characterizing regimes is inherently difficult following real shocks. Moreover, with financial
because a country’s unique history of regimes may maturity, widespread availability of debt denom-
be more relevant for economic outcomes than inated in domestic currency and in hedging in-
merely the ongoing regime. The findings suggest struments reduces the adverse consequences
that exchange rate flexibility becomes more valuable from currency mismatches that give rise to the
as countries mature in terms of their access to inter- fear of floating.
national capital markets and as they develop sound
financial systems. Below are three country group- Though on average the value of exchange rate
ings, each of which responds differently to exchange flexibility was found to increase with financial matu-
rate flexibility. rity, the results also suggest that the performance of
any regime can be enhanced by consistent macro-
• In developing countries, with their low expo- economic management. In particular, unified ex-
sure to international capital movements, rela- change rate systems have been associated with supe-
tively rigid regimes, such as pegs and intermedi- rior performance, and the declining trend of regimes
ate flexibility arrangements, appear to have with dual exchange rates that depart substantially
enhanced policy credibility and thus have helped from official rates (noted in Section II) is, therefore,
achieve lower inflation at little apparent cost in a welcome one. Similarly, freely falling regimes,
terms of lost growth, higher growth volatility, or characterized by dysfunctional macroeconomic poli-
more frequent crises. The superior performance cies, have also been poor performers. The good
of pegged regimes required commitment, as news, once again, is that the incidence of freely
shown through public announcement of that falling regimes has declined steadily over the past
goal, and was further improved through consis- decade. Furthermore, in developing and emerging
tent macroeconomic policies that allowed for economies the intermediate regimes—those lying
longer regime duration. between the two poles, or the two corners of pegs
and free floats—have not fared systematically worse
• In contrast, for emerging markets, with their than the polar regimes, which is consistent with their
higher exposure to international capital flows, longevity (also described in Section II).
the rigidity of regimes, particularly in the 1990s,
was associated with more frequent banking The section is organized as follows. First, results of
crises and especially costly “twin” crises that in- earlier empirical analysis are summarized, followed
cluded both financial sector and balance of pay- by a brief summary of the analytical issues to help in-
ments turbulence. Moreover, rigid systems were terpret the results. Next, the study takes a first look at
not associated with obvious gains in terms of inflation, growth, growth volatility, and the incidence
lower inflation or higher growth. At the same of crises across different exchange rate regimes, but
time, the move to full flexibility was inhibited by does so without controlling for other factors that af-
the concern that large swings in exchange rates fect economic performance. It then controls for other
determinants of economic performance and thus at-

21

III REGIME PERFORMANCE: INFLATION AND BUSINESS CYCLES

tempts to isolate the conditional relationship between from greater variability of real exchange rates under
exchange rate regimes and economic performance. flexible systems, there is little difference in the be-
Following is an assessment of whether the credibility havior of key macroeconomic aggregates across dif-
underlying different exchange rate regimes can be en- ferent exchange rate arrangements. Mussa (1986) had
hanced through announcement of the regime and reached similar conclusions earlier. Indeed, in their
through policies that allow for longer regime duration, review of the literature up to that point, Edison and
and includes case studies on how emerging markets Melvin (1990) despair that the empirical effort to
can enhance their ability to effectively float their cur- contrast economic performance across exchange rate
rencies. A recap concludes the section. An appendix regimes would ultimately prove inconclusive.
summarizes the data and the econometric results dis-
cussed in this section. A recent generation of papers offers a more nu-
anced assessment. Using data for the post–Bretton
Summary of Empirical Analysis of Woods era for over 100 countries, the analysis initi-
ated by Ghosh and others (1997) culminated in the
Exchange Rate Regimes comprehensive contribution of Ghosh, Gulde, and
Wolf (2003). These works deal with several empiri-
In guiding exchange rate regime choice, economic cally difficult issues. While they rely primarily on
theory has proved to be an insufficient guide for pol- the de jure regime classification, they do make some
icymakers. Empirical clarification is, thus, crucial. effort to distinguish between the regime announced
In part, theoretical ambiguity arises because the ef- by national authorities and the one actually prac-
fects of particular regimes operate with varying ticed. They also consider the perennially hard ques-
strength in different economies. In empirical analy- tion of the direction of causality: do exchange rate
sis, therefore, country types need to be distin- regimes lead to particular macro outcomes or does
guished, and this study contrasts the performance of performance determine the choice of regimes? An-
regimes in developing economies, emerging mar- other important contribution is that by Levy-Yeyati
kets, and advanced economies. Emerging markets, and Sturzenegger (2002), who develop a different
the subject of much of the recent policy discussion, measure of de facto regimes (as discussed in Section
differ from developing economies in terms of their II) and also attempts to deal with the causality issue.
higher exposure to international capital flows, but
they continue nevertheless to exhibit important insti- The results of these studies, however, continue to
tutional and financial sector weakness. As a conse- conflict with each other, reflecting the differences in
quence, emerging markets face higher inflation, their methods of classifying regimes. Ghosh, Gulde,
greater risk of debt unsustainability, more fragile fi- and Wolf (2003) find that inflation under fixed ex-
nancial systems, and higher propensity to macroeco- change rate regimes is significantly lower than under
nomic volatility. Therefore, the emerging markets intermediate or freely floating arrangements, due to
are characterized by more serious problems of credi- greater confidence in the currency (a credibility ef-
bility in the formulation of economic policy (see, for fect) and lower money growth (a discipline effect),
example, Fraga, Goldfajn, and Minella, 2003). and that the benefit of pegged exchange rate regimes
in terms of inflation performance is fairly robust to
A more fundamental ambiguity arises in evaluat- the endogeneity of regime choice. The study does not
ing exchange rate regimes, where theoretical predic- find evidence of a strong link between exchange rate
tions lead to opposing conclusions. For example, regimes and economic growth, however, especially
while pegged regimes are generally thought to lower after controlling for country-specific effects and pos-
inflation, they may only postpone its manifestation. sible simultaneity bias. This result contrasts with
The growth effects of regimes depend on what is as- Levy-Yeyati and Sturzenegger (2002), who use a de
sumed about the shock-absorbing capacity of differ- facto classification of regimes and finds for a similar
ent regimes and how important these shock ab- sample that flexible exchange rates are associated
sorbers are in raising investment and productivity. with higher growth in developing countries—which
Also, flexible exchange rates may dampen the includes the groups of countries referred to in this
volatility resulting from real external shocks, but this study as emerging markets. No similar association
very flexibility may add to the volatility, with ad- exists among industrial countries. Both Ghosh,
verse economic consequences that lead to a fear of Gulde, and Wolf, and Levy-Yeyati and Sturzenegger
floating (Calvo and Reinhart, 2002). find, however, that fixed exchange rate regimes are
associated with somewhat higher output volatility.38
Empirical analysis, however, has not delivered
clear results either. In a well-known contribution, 38Several missing and inconclusive observations in the Levy-
Baxter and Stockman (1989) compare the time-series Yeyati–Sturzenegger classification raise concerns about their con-
behavior of key economic aggregates during and clusions (see Section II).
after the Bretton Woods system and finds that, aside

22

Analytical Considerations

Against that background, this section reexamines peg. Hence, emerging markets are less likely to be
the link between exchange rate regimes and eco- able to import credibility than are other developing
nomic performance along four dimensions: inflation, countries whose interaction with international capi-
output growth, growth volatility, and the incidence of tal markets is more limited. Tornell and Velasco
crises. The assessment is based on the recently con- (2000) raise the possibility that the inflationary gains
structed Natural classification, which identifies the from fixed regimes are illusory. No exchange rate
prevailing de facto exchange rate regime as noted in system, the authors argue, can ultimately act as a
Section II. The relative longevity of regimes under substitute for sound macroeconomic policies. Far
the Natural classification renders the reverse causal- from exerting discipline, fixed exchange rate
ity problem less serious than, for example, under the regimes may create an incentive for governments
Levy-Yeyati–Sturzenegger classification, where with short time horizons to cheat, delivering tem-
regime classifications change as often as every year. porarily higher growth through larger deficits, with
Nevertheless, the section undertakes supplementary the full inflationary cost of such policies borne out
analysis to assess if the findings are robust to the re- following the eventual collapse of the peg.
verse causality concern. Also, to allow for the possi-
bility that the pressures under a particular regime are The theoretical implications of exchange rate
manifested after its collapse in a new regime, the sec- regimes for economic growth and volatility are simi-
tion examines the lagged influence of regimes so larly murky, with various opposing claims.39 In favor
that, in effect, the performance in the first year of a of pegs, Dornbusch (2001) argues that lower infla-
new regime continues to be attributed to the previous tion associated with rigid exchange rate regimes
regime. This turns out to be important in the analysis would reduce interest rates and uncertainty, spurring
of volatility in emerging markets. In addition, regime investment and growth.40 Also, when a country ties
announcement and duration are considered as factors its currency tightly to that of another through a cur-
that may influence regime performance. The analysis rency board arrangement, transaction costs may be
covers up to 158 countries from 1970 to 1999, lowered, thereby increasing trade between the two
throughout which developing, emerging, and ad- countries. Frankel and Rose (2002) find that such
vanced economies are distinguished. Where appro- expansion of trade is not offset by diversion away
priate, the 1990s, which was a period of rapidly ris- from other trade partners and that by increasing the
ing capital flows, is distinguished from earlier years. openness of the economy this form of exchange rate
rigidity also raises output growth. An argument in
Analytical Considerations favor of exchange rate flexibility is the possibility of
rapid resource reallocation following real shocks
An important prediction from economic theory is when short-run price rigidity is significant (Levy-
that exchange rate pegs act as a disciplining device, Yeyati and Sturzenegger, 2003). Broda (2001) finds
allowing policymakers in countries with a propen- evidence that terms-of-trade shocks are amplified in
sity for high inflation to import credibility and, countries that have more rigid exchange rate
hence, lower inflation from abroad (Giavazzi and regimes. Edwards and Levy-Yeyati (2003) take that
Giovannini, 1989; and Dornbusch, 2001). As a pol- empirical analysis one step further and concludes
icy prescription, nominal exchange rate rigidity—or that the inability of rigid regimes to absorb such
an exchange rate anchor—came back into favor in shocks translates, in practice, into lower growth.
the late 1980s and early 1990s, especially in Latin Similarly, Calvo (1999) argues that the need to de-
America, where exchange rate–based stabilizations fend a peg following a negative external shock may
were viewed as particularly helpful following a his- result in high real interest rates and also stifle
tory of high inflation (Edwards, 2001). In this line of growth.
reasoning, the harder the peg, the more effective it is
in enhancing credibility (Edwards and Magendzo, While flexible exchange rate regimes may, in prin-
2003a). ciple, dampen real shocks to the economy, could the
very flexibility of the exchange rate introduce a new
The proposition that pegs provide an inflation ad- element of volatility? As noted above, a robust find-
vantage is far from universally held, however. For ing is that nominal exchange rate volatility is associ-
advanced economies, pegged exchange rate regimes
should not be necessary for achieving credibility. 39For a useful summary, see Bailliu, Lafrance, and Perrault
Even where such regimes could play a role, achiev- (2002).
ing and maintaining hard pegs is not a straightfor-
ward process. In particular, as exposure to interna- 40Such a beneficial outcome may have prevailed in the postcon-
tional capital flows increases, a larger fraction of the vertibility Bretton Woods period from 1959–71, when inflation
monetary aggregates must be backed to maintain the and exchange rate volatility were low and growth was relatively
strong (Bordo, 2003). It is not clear, however, whether this was
the consequence of the rigidity in exchange rate regimes or the
consequence of a generally favorable economic environment.

23

III REGIME PERFORMANCE: INFLATION AND BUSINESS CYCLES

Table 3.1. Economic Performance Across Exchange Rate Regimes

Inflation Growth Volatility Crisis

Fixed May enhance monetary May reduce transaction May increase volatility in High risk of speculative
Flexible policy credibility and lower costs, and raise trade the presence of real attacks against currency,
inflation; emerging markets and growth; may also shocks and nominal especially when exposed
less likely to be able to reduce interest rates and rigidities to volatile capital flows;
import credibility; inflation, uncertainty, also raising susceptibility to banking
moreover, may be “bottled investment and growth sector distress
up” under weak macro-
economic management Low risk of currency
crises and banking crises
Declining importance of Higher growth due to Possible spilling over of
“imported” credibility with shock absorbers and real exchange rate
stronger institutions and fewer distortions following volatility into real activity
financial sectors real shocks

ated with high real exchange rate volatility. Rogoff Under pegs, the exchange rate may become progres-
(1999) argues that such variability does not, in prac- sively overvalued, weakening the financial system;
tice, have significant effects on output and consump- without, or with only limited, lender-of-last-resort
tion in advanced economies but may be harmful in capabilities, authorities may be unable to deal with
developing countries. Even if the higher volatility domestic financial distress.
has harmful effects, however, pegged regimes may
not be the appropriate policy response because the Table 3.1 summarizes these predictions for eco-
volatility may only appear to be contained and may nomic performance across regimes.
have real (adverse) effects on private investment due
to the greater uncertainty over regime sustainability. Macroeconomic Performance and
Crisis Probabilities: Summary
Indeed, just as the inflation-reducing benefits of Statistics
exchange rate rigidity were being emphasized in the
early 1990s, a fundamental reevaluation of the pol- Drawing on both the de facto and de jure classifi-
icy prescription was under way following the early cations, the following portion describes the associa-
crises associated with rigid regimes. For early recog- tion between exchange rate regimes and various di-
nition of this concern, see, for example, Eichengreen mensions of economic performance. No attempt is
(1994), and Obstfeld and Rogoff (1995). The latter made in this preliminary description to control for
study notes that in 1995, following the collapse of other factors that may also influence economic out-
the British pound in September 1992 and of the comes. First, a summary of average macroeconomic
Mexican peso in December 1994, that “Many recent performance under alternate regimes is presented.
efforts to peg exchange rates within narrow ranges Second, because the occurrence of crises has been
have ended in spectacular debacles.” The authors particularly highlighted in recent policy discussions,
went on to conclude “These events are not unprece- the relationship between regimes and the frequency
dented but their ferocity and scope have called into of banking and currency crises is documented.
question the viability of fixed rates among sovereign
nations in today’s world of highly developed global Regimes and Performance: Summary
capital markets.” The subsequent fall of tightly man- Measures for Inflation, Growth, and Volatility
aged regimes in East Asia (1997), Russia (1998),
Brazil (1999), and Argentina (2002) have served as a Conflicting policy objectives and large macroeco-
continuing warning against pegged regimes, espe- nomic imbalances will lead to poor economic per-
cially in emerging markets subject to volatile capital formance irrespective of the exchange rate regime.
flows. Pegged exchange rates—or those with limited For the purposes of this discussion, there are at least
flexibility—invite speculative activity against the ex- two sets of conditions under which the exchange rate
change rate and lead to abandonment of the peg, cur- regimes may have no independent influence on
rency overshooting, and large output costs (Larraín macroeconomic outcomes through the prevailing
and Velasco, 2001). Pegged regimes may also be
subject to a higher incidence of banking crises.

24

Macroeconomic Performance and Crisis Probabilities: Summary Statistics

Table 3.2. Average Annual Inflation and Real Per Capita GDP Growth:
Comparison of Dual (or Multiple) and Unified Exchange Rate Systems,
1970–991

(In percent)

Unified exchange rate Average Annual Average Per Capita
Dual (or multiple) exchange rates Inflation Rate GDP Growth

Source: Author’s calculations. 22.0 1.8
1Figures in parentheses are medians. (7.7) (2.1)
175.6 0.6
(15.1) (1.4)

severity of economic distortions. First, the preva- under other systems as freely floating, their identifi-
lence of dual (or multiple) rates—and, hence, a po- cation as a separate category in the Natural classifi-
tentially large differential in official and parallel cation can make a significant difference to the rela-
market exchange rates—is a consideration in deter- tive rankings of regimes. For example, according to
mining the operative regime as well as a factor influ- the de jure classification (the last column in Table
encing economic outcomes through the prevailing 3.3), pegs have a much lower inflation rate than
severity of economic distortions, as observed in floating regimes. Under the Natural classification,
Reinhart and Rogoff (2004). Second, the authors iso- however, freely floating regimes (the bottom row of
late countries with annual inflation rates above 40 Table 3.3) have, on average, lower inflation than ex-
percent into a separate freely falling category, with change rate pegs. This reversal occurs because, as
the implication that the macroeconomic imbalances noted, many freely falling episodes are in the float-
in such conditions overwhelm the possible effects of ing regime category according to the de jure classifi-
the exchange rate regime. cation. As noted below, when other influences on in-
flation are taken into account, the advantage of
The evidence suggests that dual exchange rates pegged and intermediate regimes over the floating
are associated with significantly worse economic regime reappears even in the Natural classification;
performance. Over the period 1970–99, the average however, not distinguishing the freely falling cate-
per capita income growth rate in countries with dual gory renders that advantage much larger.
exchange rates was about 0.6 percent per year; in
contrast, countries with unified rates grew at three The performance of intermediate regimes is not
times the pace, at about 1.8 percent per year (Table especially different from that of other regimes (Ta-
3.2). Similarly, annual inflation in countries with bles 3.3, 3.4, and 3.5). This is consistent with the
dual exchange rates was about 175 percent, while longevity of these regimes, as documented in Sec-
under unified rates it was about 22 percent. These tion II. If this comparison had revealed consistently
performance differences primarily reflect instances poorer performance under intermediate regimes,
of large departures from official rates—the differ- there would have been a greater basis for expecting a
ences in median performance are less egregious. shift to the polar extremes of pegs and free floats.
With increasingly integrated capital markets, large
gaps between official and parallel rates have become Finally, as documented by Mussa (1986), Baxter
untenable, and the move to unified exchange rates and Stockman (1989), and Flood and Rose (1995),
has been almost universal (see Section II). real exchange rates are more variable the greater the
flexibility of the regime (Table 3.6). Exchange rate
By their construction, freely falling regimes per- volatility is considerably higher under managed float-
form significantly worse than other regimes on all ing and freely floating regimes than under pegged and
counts: they have higher inflation, lower growth limited flexibility regimes. This reflects the fact that
rates, and higher volatility (Tables 3.3, 3.4, and 3.5). real rates tend, at least in the short run, to move closely
With the worldwide decline in inflation, the inci- with nominal rates. Notably, more flexibility under the
dence of freely falling regimes is on the decline (Ro- de jure classification is not associated with greater
goff, 2003). For retrospective analyses, however, be- variability of the real exchange rate because regimes
cause freely falling episodes are typically classified that are declared flexible are often tightly managed.

25

III REGIME PERFORMANCE: INFLATION AND BUSINESS CYCLES

Table 3.3. Average Annual Inflation Rates Across Exchange Rate Regimes, 1970–991

(In percent)

Peg Limited Managed Freely Freely Unknown Total
Flexibility Floating Floating Falling
Pegged 17.9 12.4 33.9
Intermediate (6.8) 9.6 14.2 24.5 391.7 (6.5) (7.9)
Floating (7.9) (10.4) (23.2) (39.9)
Total 11.2 25.7 36.0
(3.5) 13.0 16.7 9.2 147.6 (15.9) (10.8)
(9.1) (15.1) (3.8) (66.1)
20.3 445.6 138.5
(11.5) 10.1 11.3 8.1 408.9 (22.2) (10.8)
(7.5) (8.4) (4.5) (68.6)
17.1 55.5 49.7
(6.5) 11.1 14.2 9.9 305.3 (7.6) (8.7)
(8.3) (10.8) (4.8) (57.0)

Source: Author’s calculations.
1Figures in parentheses are medians.

Regimes and Crisis Probabilities Consider, first, the frequency of banking crises.42
More rigid regimes had a higher likelihood of bank-
In the 1990s, several economies with rigid ex- ing crises, especially in the 1990s. For all countries,
change rate regimes were victims of severe eco- during the period 1980–97, the probability of a
nomic crises. A concern thus arose not just for the banking crisis in a given year varied between about 3
prospects of the economies directly subject to the and 4.5 percent, with no clear variation across ex-
crises but also for the possible contagion of crises change rate regimes (Table 3.7).43 The highest prob-
across countries with similar economic features fol- abilities of a banking crisis occurred in the emerging
lowing a general loss of investor confidence. The oc- market economies, however, where the evidence also
currence of crises has, therefore, acquired greater suggests that the probability of a crisis increased
prominence in the policy discussions on the choice with the rigidity of the exchange rate regime. More-
of exchange rate regimes. Despite the policy interest, over, the association between rigidity and probabil-
few systematic studies have examined the links be- ity of banking crises in emerging markets became
tween crises and exchange rate regimes. stronger in the 1990s.

The evidence presented in this section suggests The finding that banking crises are more likely to
that popular perception in this regard has some sta- occur under rigid regimes is in contrast to that of
tistical basis. While the evidence on currency crises Ghosh, Gulde, and Wolf (2003), who conclude that,
is mixed, the frequency of banking and twin crises, if anything, floating regimes are the most likely to
where banking and currency turbulence come to- experience banking crises. The difference in findings
gether, has been higher under more rigid regimes but is the consequence of the latter’s use of the de jure
mainly for emerging markets, and particularly so in
the 1990s. As noted in the introduction, emerging 42Crisis probabilities were obtained as the ratio of crises
markets are more exposed to international capital episodes under a particular regime divided by the number of
flows than are other developing economies, but com- regime years. Each crisis was treated as a single episode even if it
pared to advanced industrialized economies emerg- lasted for multiple years. The estimates presented drop the year of
ing markets have fragile financial sectors.41 the crisis itself as well as the years immediately preceding and
following the regime change to minimize the influence of the
41Emerging markets are defined here as those countries classi- regime transition on the occurrence of crises.
fied as such by Morgan Stanley Capital International on the basis
of several factors but including also their access to international 43The data for banking crises are obtained from Demirgüç-
capital markets. As discussed in the appendix, any definition of Kunt and Detragiache (1998), who declare a banking crisis to
emerging markets is likely to include or exclude countries on the have occurred when any one of the following four conditions
margin in ways that are more or less appropriate. Extensive ro- held: nonperforming loans exceeded 10 percent of banking sys-
bustness tests were undertaken, and only the most robust results tem assets; a bailout cost 2 percent or more of GDP; large-scale
are highlighted in the text. nationalization occurred; or other emergency measures, such as
bank holidays, deposit freezes, and special guarantees had to be
undertaken.

26

Macroeconomic Performance and Crisis Probabilities: Summary Statistics

Table 3.4. Average Annual Real Per Capita GDP Growth Across Exchange Rate Regimes, 1970–991

(In percent)

Peg Limited Managed Freely Freely Unknown Total
Flexibility Floating Floating Falling
Pegged 2.0 1.0 1.6
Intermediate (2.0) 2.6 1.6 –3.2 –1.1 (0.6) (1.6)
Floating (2.6) (1.6) (0.5) (–0.7)
Total 2.8 2.7 2.1
(2.4) 2.6 1.9 2.7 0.0 (2.7) (2.3)
(2.9) (2.1) (2.2) (0.4)
3.6 –1.6 0.6
(2.9) 1.7 1.6 2.2 –3.1 (–0.3) (1.7)
(1.8) (2.2) (2.3) (–1.2)
2.1 0.8 1.5
(2.2) 2.4 1.7 1.8 –1.3 (0.6) (1.8)
(2.6) (2.0) (2.0) (–0.6)

Source: Author’s calculations.
1Figures in parentheses are medians.

Table 3.5.Average Annual Growth Volatility Across Exchange Rate Regimes, 1970–991

(In percent)

Peg Limited Managed Freely Freely Unknown Total
Flexibility Floating Floating Falling
Pegged 4.0 4.3 4.0
Intermediate (2.7) 3.8 3.6 5.7 4.3 (2.9) (2.7)
Floating (2.3) (2.6) (3.3) (3.4)
Total 1.6 6.1 2.6
(1.2) 2.0 2.6 3.3 3.8 (2.5) (1.8)
(1.6) (1.8) (1.7) (3.4)
3.1 4.9 3.8
(1.8) 2.4 4.1 1.9 6.4 (2.9) (1.9)
(1.5) (1.9) (1.1) (4.6)
3.7 4.5 3.7
(2.4) 2.8 3.5 2.7 4.7 (2.9) (2.4)
(1.8) (2.3) (1.3) (3.7)

Source: Author’s calculations.
1Figures in parentheses are medians.

classification, which has many more countries clas- based on a measure of currency crises employed by
sified as floating than does the Natural classification. Berg, Borensztein, and Pattillo (2004).44 The evi-
As noted in Section II, many of these de jure floaters dence for the 1990s is less clear cut and suggests that,
are classified under the Natural classification as among emerging markets, pegged regimes had more
freely falling; other floaters did not actually float frequent currency crises. An alternative measure of
and so were de facto under more rigid regime cate- currency crises, using different thresholds for ex-
gories. As a consequence, using the de jure classifi- change rate depreciation and loss in reserves (Bordo
cation leads to an overstatement of the likelihood of
banking crises under floating regimes and an under- 44Berg, Borensztein, and Pattillo (2004) recognize a crisis as
statement of crisis probabilities under more rigid having occurred when the weighted average of one-month
regimes. changes in the exchange rate and reserves is more than three
(country-specific) standard deviations above the country
Currency crises over the years 1970 to 2000 tended average.
to occur more frequently in intermediate regimes,

27

III REGIME PERFORMANCE: INFLATION AND BUSINESS CYCLES

Table 3.6. Real Exchange Rate Volatility Across Exchange Rate Regimes, 1970–20021

Limited Managed Freely Freely

Peg Flexibility Floating Floating Falling Unknown Total

Pegged 6.3 8.9 25.1 7.0 53.6 6.6 12.7
Intermediate 3.2
Floating 10.5 4.8 10.5 30.6 42.3 28.4 12.1

Total 5.6 5.2 11.6 8.4 17.3 14.8 10.4

6.1 17.9 13.7 37.0 9.2 12.0

Source: Author’s calculations.
1Volatility is measured as the three-year centered standard deviation of the annual real effective exchange rate (IFS, line RECZF). Nicaragua is excluded
from this table because its exchange rate has been extremely volatile, and its inclusion unduly influences the averages.

and others, 2001), shows that pegs and limited flexi- Regime Performance and
bility had a significantly higher risk of currency cri- Levels of Development
sis than managed or freely floating regimes45 for
emerging markets. While the previous section reported correlations,
this section takes the more demanding step of at-
Finally, twin crises have been almost uniquely an tempting to isolate, over the period 1970–99, the
emerging market phenomenon: they have never oc- association between exchange rate regimes and the
curred in the group of countries classified as devel- performance measures of interest, after controlling
oping and rarely in advanced economies. Moreover, for other variables that may also influence perfor-
the incidence of twin crises in emerging markets is mance.46 But even after such controls are included,
highest under pegged regimes and falls as flexibility reverse causality, or endogeneity, remains a con-
in regimes increases. Kaminsky and Reinhart (1999) cern in such analyses: in other words, the observed
have noted that twin crises have particularly high relationships may reflect the influence of the per-
costs. Such crises typically start with domestic fi- formance variable on the choice of the regime
nancial distress, which accelerates when a currency rather than the other way around. This problem
crisis also sets in, leading to a vicious cycle. Costs cannot be fully resolved, but it is mitigated by the
are high in terms of bailout of the financial sector as relatively long duration of the typical regime under
well as in terms of reserves lost. Larraín and Velasco the Natural classification, implying that temporary
(2001) provide a theoretical discussion of why cur- changes in performance do not influence the choice
rency boards may be particularly prone to twin of regime. The problem is also mitigated by using
crises. Rigid regimes may promote excessive risk as an explanatory variable the regime prevailing in
taking during periods of booms in capital inflows, the previous one or two years where the results pre-
when the expectation of an exchange rate guarantee sented are unchanged when that is done, except, as
reduces the incentive to hedge foreign currency ex-
posure. The sudden withdrawal of flows leaves the 46See the appendix for a detailed discussion of the methodol-
domestic financial sector susceptible to severe dis- ogy. In addition to variables that are used conventionally to ex-
tress. At the same time, the commitment to an ex- plain the different dimensions of performance (discussed below),
change rate target limits lender-of-last-resort opera- two further sets of controls are used throughout. First, common
tions. If depositors withdraw domestic currency shocks across countries, such as spikes in oil prices or changes in
from domestic banks to buy the foreign reserve cur- the volatility of G-3 currencies, influence all economies beyond
rency at the central bank under a fixed exchange the effect channeled through observed variables. These are con-
rate, the panic withdrawal can lead to a self-fulfilling trolled for through the use of time dummies. Second, while an in-
crisis as foreign currency reserves are depleted. Ar- creasing number of country control variables can be added, cer-
gentina’s massive collapse is a cautionary tale of tain unobserved or difficult-to-measure country characteristics
how some of these forces can contribute to the un- may reflect important dimensions of institutions and policy credi-
raveling of even a hard peg. bility. These, in turn, are likely to be correlated with exchange
rate regimes; to control for these unobserved characteristics,
45Bubula and Ötker-Robe (2003) continue to find vulnerability country dummies are included. The implication of this approach
in the intermediate regimes in the 1990s but does not distinguish is that regime performance is judged by changes that occur within
emerging markets. a country rather than across countries.

28

Regime Performance and Levels of Development

Table 3.7. Probability of Crises During Specific Regimes Using the Natural Exchange Rate
Regime Classification1

(In percent)

___________B_an_k__C_r_i_si_s_(_1_9_8_0_–_9_7_)__________ ___________B_an_k__C_r_i_si_s_(_1_9_9_0_–_9_7_)__________

Limited Managed Freely Limited Managed Freely
Peg Flexibility Floating Floating
Peg Flexibility Floating Floating

All 3.4 4.7 4.5 3.9 3.1 7.1 3.0 3.8
Advanced 0.0 6.5 0.0 4.2
Emerging 0.0 2.7 2.3 4.1 15.4 8.0 3.8 0.0
Developing 2.6 7.1 4.5 —
11.4 7.5 7.0 0.0

2.8 7.0 3.6 —

___B_a_l_a_n_ce__o_f_P_a_y_m_e_n_t_s_C__ri_s_is_(_1_9_7_0_–_2_0_0_0_)___ ___B_a_l_a_n_ce__o_f_P_a_y_m_e_n_t_s_C__ri_s_is_(_1_9_9_0_–_2_0_0_0_)___

Limited Managed Freely Limited Managed Freely
Peg Flexibility Floating Floating
Peg Flexibility Floating Floating

All 4.1 4.1 9.2 4.6 4.7 5.2 9.2 4.3
Advanced 3.6 5.8 8.6 4.9
Emerging 3.3 3.9 7.1 4.9 8.8 6.1 6.9 0.0
Developing 0.0 2.8 15.4 —
4.6 5.6 10.0 0.0

5.2 2.0 9.7 —

__________T_w__in__C_r_is_e_s_(_1_9_8_0_–_9_7_)__________ __________T_w__in__C_r_is_e_s_(_1_9_9_0_–_9_7_)__________

Limited Managed Freely Limited Managed Freely
Peg Flexibility Floating Floating
Peg Flexibility Floating Floating

All 1.6 1.4 0.8 0.0 3.2 2.6 0.0 0.0
Advanced 0.0 2.2 0.0 0.0
Emerging 0.0 0.7 0.0 0.0 15.4 4.0 0.0 0.0
Developing 0.0 0.0 0.0 —
7.7 3.0 1.8 0.0

0.0 0.0 0.0 —

Source: Author’s calculations.
1Probabilities are calculated by dividing the number of occurrences of a crisis under a particular regime by the total number of regime years. Each cri-
sis is counted only once, and hence, if it persists over multiple years, the subsequent years are not taken into account for this calculation. Additionally, the
years an exchange rate regime transition takes place (i.e., the year preceding, the year during, and the year following the transition) are excluded from
this computation. A dash (—) indicates that no crisis data were available for developing countries under freely floating exchange rate regimes.

discussed below, in the analysis of volatility in 2003b)—support this conclusion. It turns out, how-
emerging markets.47 ever, to be important to distinguish between different
country types. On average, pegged and intermediate
Inflation Performance Across regimes have been associated with significantly
Exchange Rate Regimes lower inflation rates than floating regimes, but this
reflects an inflation benefit that accrues primarily to
A wide range of empirical studies have found that developing economies and not to emerging markets
fixed exchange rate regimes deliver lower inflation. or advanced economies.48
IMF (1997) found that the median inflation rates for
fixed regimes have been lower than those for floating Before examining the differentiation across coun-
regimes, though the difference declined over time. try groups, it is useful to note that the findings with
More rigorous studies that control for other determi-
nants of inflation—for example, Ghosh, Gulde, and 48This finding is consistent with, but goes beyond, that ob-
Wolf (2003) and Edwards and Magendzo (2003a and tained by Ghosh, Gulde, and Wolf (2003), who, using the de jure
classification, find little difference in inflation performance
47Moreover, as summarized in Appendix II, it is difficult to across regimes in upper-income (advanced) countries but do find
identify country characteristics that consistently predict exchange greater rigidity to be associated with lower inflation in
rate regimes. Since regimes are strongly persistent, they are likely middle- and lower-income countries. Distinguishing further be-
to be the best predictors of expected regimes. tween emerging markets and developing economies shows the
latter to be the primary beneficiaries of exchange rate rigidity.

29

III REGIME PERFORMANCE: INFLATION AND BUSINESS CYCLES

Figure 3.1. Inflation Performance Across respect to inflation performance across all countries
Regimes: Confidence Effect1 are similar whether the de jure or the Natural classifi-
cation is used. Figure 3.1, which pools all countries,
(In percent) contrasts the results using the de jure classification
with those for the Natural classification. The bars in
12 the figure represent the difference in inflation in inter-
mediate and floating rate regimes relative to pegged
De jure classification regimes, after controlling for a variety of factors
thought to influence inflation in all regimes.49 Be-
Natural classification cause the regressions control for money growth, they
can be thought of as capturing the value of credibility
10 rather than greater discipline, whereby pegs generate
lower inflation through control on the growth of
8 6.22 money supply, for example. Using the de jure classifi-
(5.13) cation, floating regimes are associated with signifi-
6.18 cantly higher inflation than pegged regimes, on aver-
(6.58) 4.48 age 6.2 percent.
(2.91)
6 For the Natural classification, pegged regimes
continue to exhibit significantly lower inflation than
4 freely floating regimes, though the margin by which
they do so (4.5 percent) is smaller. In addition, inter-
1.63 mediate regimes now perform significantly better
(by 2.9 percent) than floating regimes in terms of in-
2 (2.11) flation. Separating out the freely falling category re-
duces the average inflation rate for freely floating
0 Floating regimes, thereby reducing the inflation advantage of
pegged regimes and giving some inflation advantage
Intermediate to intermediate regimes.

Source: Authors’ calculations. The effect on inflation through potentially greater
1Figures in parentheses are t-statistics.The bars depict differ- monetary discipline under restrictive regimes is sig-
ences in performance relative to pegged exchange rate regimes, nificantly smaller than that due to enhanced credibil-
conditioning on a range of other variables. See appendix in ity. Figure 3.2 captures the discipline effect by at-
Section III for details. tributing differences in rates of money supply
growth to the regimes themselves and thus imputing
Figure 3.2. Inflation Performance Across additional discipline effects based on these differ-
Regimes: Confidence and Discipline Effects1 ences. Using the de jure classification, the discipline
effect adds to the inflation advantage of pegged
(In percent) regimes somewhat. The same holds for the Natural

12 7.64 49The results presented in graphical form are the coefficients on
(3.55) dummy, or categorical, variables representing the exchange rate
De jure classification regime. These coefficients should be interpreted as a regime’s per-
Natural classification 5.17 formance (relative to the excluded pegged regime), conditional
(3.08) upon the other included variables in the regression. The two inter-
10 mediate regimes in the coarse Natural classification have been ag-
gregated to one regime for consistency with the de jure classifica-
7.71 tion. Throughout, when using the Natural classification, the freely
falling countries are also identified as an additional category. In
8 (3.72) line with the discussion on performance and crisis probabilities
above, freely falling regimes do not perform as well as other
6 regimes in most instances. Because their performance does not
have a direct implication for policy discussion on exchange rate
4 regime performance, however, the results are not presented in the
main text but are included in the tables in the appendix, which dis-
2.01 cusses these issues in further detail. All the inflation regressions
(2.32) control for real GDP and money growth, in addition to trade open-
ness, the degree of central bank independence, terms-of-trade
2 shocks, and the fiscal balance. In addition, as noted above, each re-
gression has country-specific fixed effects and year dummies.
0 Floating

Intermediate

Source: Authors’ calculations.
1Figures in parentheses are t-statistics.The bars depict differ-
ences in performance relative to pegged exchange rate regimes,
conditioning on a range of other variables. See appendix in
Section III for details.

30

Regime Performance and Levels of Development

Figure 3.3. Inflation Performance in Figure 3.4. Inflation Performance in
Advanced Countries, Emerging Markets, Advanced Countries, Emerging Markets,
and Developing Countries: Confidence and Developing Countries: Confidence and
Effect1 Discipline Effects1

(In percent) (In percent)

15 Limited flexibility 10.43 15 10.01
(3.83) (3.65)
Managed floating Limited flexibility
Managed floating
10 Freely floating
10 Freely floating
6.21
5.97 (0.25)
(0.21)
5 2.54 2.20 2.42 2.75
5 2.61 (1.49) (1.53)
(1.57) (1.27)
(2.11)

0 0

–0.56 –1.09 –1.40 –1.23 –0.76
(–1.41) (–1.75) (–2.11) –2.69 –2.66 (–0.36) (–1.18)

–5 (–1.45) (–1.40) –2.96 –5.10
(–0.63)
–10 Emerging Developing –5 (–3.13) Developing
markets countries Emerging countries
Advanced –10 markets
countries
Advanced
countries

Source: Authors’ calculations. Source: Authors’ calculations.
1Figures in parentheses are t-statistics.The bars depict differ- 1Figures in parentheses are t-statistics.The bars depict differ-
ences in performance relative to pegged exchange rate regimes, ences in performance relative to pegged exchange rate regimes,
conditioning on a range of other variables. See appendix in conditioning on a range of other variables. See appendix in
Section III for details. Section III for details.

classification, where the inflation advantage attrib- high rate of regime transitions in this group of coun-
uted to exchange rate pegs and to intermediate tries), there is some evidence that, as in developing
regimes relative to floating rises modestly. But over- economies, inflation rises with flexibility. This may
all these effects are small. explain the fear of floating50 because the reduction in
pass-through to domestic prices may take time. For
A different story emerges when inflation perfor- advanced countries, there is evidence that inflation
mance is distinguished across the advanced coun- actually declines with increased exchange rate flexi-
tries group, emerging markets, and developing bility. While the direction of the results typically fa-
countries (Figures 3.3 and 3.4). In developing coun- vors floating over pegged, the results in alternate
tries, inflation performance deteriorates with ex- specifications are not always so clear, and the appro-
change rate flexibility. The results indicate that priate conclusion appears to be that floating regimes
pegged regimes have the lowest inflation, about 2.5 do no worse than pegged regimes in terms of infla-
percent per annum lower than countries with inter- tion performance in advanced economies. These
mediate flexibility; floating regimes experience in- differences across advanced countries, emerging
flation that is about 8 percent a year more than markets, and developing countries are similarly ap-
regimes with intermediate flexibility. Note that, parent in Figure 3.4, which incorporates the above-
though the difference between pegs and intermedi- mentioned discipline effects via monetary policy.
ate regimes is not statistically significant, the differ-
ence between pegs and floating regimes is highly Overall, these results suggest that there may be
significant. This result holds up consistently in a va- some merit to pegged and intermediate regimes in
riety of empirical specifications (see, for example, developing countries perhaps reflecting the fact that,
Table A3.4 in the appendix). in the absence of sound institutions and a strong

In emerging markets, inflation performance shows 50See Tables A3.4 and A3.7 in the appendix. Also, hard pegs
no significant relationship with greater exchange rate may have better inflation performance in emerging markets rela-
flexibility. When, however, the regime prevailing in tive to other regimes; however, the effect is only marginally sig-
the prior one or two years is used as the explanatory nificant statistically.
variable (to minimize the influence of the relatively

31

III REGIME PERFORMANCE: INFLATION AND BUSINESS CYCLES

Figure 3.5. Growth Performance Across track record, these regimes can enhance policy cred-
Regimes1 ibility and discipline monetary policy. This does not,
of course, imply a blanket recommendation of
(In percent) pegged regimes because many country-specific fea-
tures would need to be taken into account in making
4 De jure classification 0.26 0.36 that decision, including the appropriate level at
3 Natural classification (0.57) (0.47) which to peg the exchange rate. As countries gain
access to international capital markets, there appears
2 Floating to be no evidence of inflation reduction through the
adoption of rigid regimes.
1 0.00
Per Capita Income Growth Across
(0.00) Exchange Rate Regimes

0 Does the inflation advantage of pegged and inter-
mediate over floating regimes in developing coun-
–0.02 tries help growth (through reduced interest rates and
(–0.05) lower uncertainty, as Dornbusch, 2001, suggests);
does it come at the expense of growth; or does the
–1 exchange rate regime make no difference to growth
(as Eichengreen, 2001, concludes)?
–2
For the full sample of countries (Figure 3.5),
–3 both the de jure and de facto classifications show
virtually no relationship between exchange rate
–4 Intermediate flexibility and growth.51 For developing economies
(Figure 3.6), growth appears to decline with in-
Source: Authors’ calculations. creased flexibility, though the effect is not statisti-
1Figures in parentheses are t-statistics.The bars depict differ- cally significant. Thus, the association observed
ences in performance relative to pegged exchange rate regimes, above of lower inflation with greater rigidity appar-
conditioning on a range of other variables. See appendix in ently does not come at the expense of growth; but
Section III for details. neither does lower inflation have a measurable fa-
vorable effect through, for example, lower interest
Figure 3.6. Growth Performance in rates and reduced uncertainty. For emerging mar-
Advanced Countries, Emerging Markets, kets, the relationship between growth and regimes
and Developing Countries1 is noisy, as with inflation.

(In percent) In contrast, for advanced countries free floats do
significantly better than other regimes in terms of
4 2.73 Limited flexibility growth performance. Indeed, the results suggest that
for advanced economies exchange rate rigidity is
3 (1.97) Managed floating monotonically associated with slower growth, which
is even more apparent when regimes are lagged, as
2.02 Freely floating reported in the appendix. Because in the advanced
(1.52) countries no inflation benefit is associated with
2 greater rigidity—indeed, if anything, inflation per-

1 0.46 0.26 51The results are based on separate panel regressions of real per
0.11 (0.56) (0.34) capita GDP growth on the relevant set of regime dummies, with
the pegged regime as the omitted category. Again, the two inter-
(0.14) mediate categories in the Natural classification have been aggre-
gated for better comparison with the de jure classification. Each
0 regression follows Ghosh, Gulde, and Wolf (2003) in controlling
for factor accumulation (investment ratio, education, and popula-
–1 –0.48 tion level and growth); trade openness; terms-of-trade shocks; im-
(–0.71) –0.77 portance of the government sector (tax ratio and central govern-
–2 ment balance); and conditional convergence. Finally, each
(–1.28) regression includes country-specific fixed effects and year dum-
mies, as above.
–1.75
(–1.03)

–3 –2.47
(–0.88)

–4 Emerging Developing
markets countries
Advanced
countries

Source: Authors’ calculations.
1Figures in parentheses are t-statistics.The bars depict differ-
ences in performance relative to pegged exchange rate regimes,
conditioning on a range of other variables. See appendix in
Section III for details.

32

Regime Performance and Levels of Development

formance worsens with more regime rigidity—there Figure 3.7.Volatility of Real GDP Growth
appears to be an overall benefit from floating.52 Performance Across Regimes1

The beneficial influence of flexible regimes as (In percent)
countries become more advanced is consistent with
the view that floating permits more rapid adjustment 4
following shocks, and that with stronger institu-
tions—in particular, deep financial sectors—ad- De jure classification
vanced economies are not subject to the offsetting
risks of floating. Bordo and Flandreau (2001), in line 3 Natural classification
with Calvo and Reinhart (2002), note that where do-
mestic financial markets are underdeveloped, bor- 2
rowing in foreign currency creates significant risks of
sharp changes in an enterprise’s net worth when ex- 1 0.46 0.82 0.82
change rates are flexible. As borrowing in domestic (1.19) (3.11) (1.24)
currencies becomes a viable option, the costs of flex-
ibility fall.53 Bordo (2003) makes the further argu- 0
ment that advanced economies have always been
more successful in managing the trade-off between –0.06
achieving credibility and retaining flexibility. The au- (–0.31)
thor suggests that even during the period of the clas-
sical gold standard when exchange rates were fixed, –1
the margin permitted by gold points allowed tempo-
rary changes in exchange rates. Of importance is the –2 Floating
observation that these exchange rate changes were
expected to be temporary—to deal with shocks—and Intermediate
hence a reversion to the parity was expected. In con-
trast, where credibility is low, deviations can generate Source: Authors’ calculations.
the expectation of further deviations. 1Figures in parentheses are t-statistics.The bars depict differ-
ences in performance relative to pegged exchange rate regimes,
Growth Volatility Across Exchange conditioning on a range of other variables. See appendix in
Rate Regimes Section III for details.

Finally, consider the relationship between ex- 3.7).55 However, while there is essentially no rela-
change rate regimes and output growth volatility.54 tionship for developing countries, volatility appears
When using the Natural classification, growth to increase with flexibility in the other two groups
volatility does not appear to vary systematically of countries (Figure 3.8). The increase in volatility
across regimes and across all countries (Figure with flexibility in advanced economies comes at
apparently little or no cost, as Rogoff (1999) sug-
52Again, this result parallels that obtained by Ghosh, Gulde, gests, and as implied by the earlier findings that
and Wolf, who find that free floats register significantly higher flexibility is associated with higher growth and
growth than pegged regimes in upper– and upper–middle income lower inflation.
countries; the authors also find little evidence among lower– and
lower–middle income countries of a link between regimes and For emerging markets, the story is more complex
growth performance. (Figure 3.8). Here there appears at first to be higher
volatility associated with more flexibility. Two con-
53The idea that better institutions are a driving force behind the siderations, however, caution against that conclu-
difference in the results across advanced and developing coun- sion. First, the volatility associated with the col-
tries is further supported by the robustness tests reported in the lapse of rigid regimes is likely to register during
appendix. subsequent regimes—an important consideration

54Figure 3.7 is based on separate panel regressions of the three- 55This is in contrast with the outcome under the de jure classifi-
year centered standard deviation of real per capita GDP growth cation, where, relative to free floats, the standard deviation of real
on the relevant set of regime dummies, with pegged regimes the per capita GDP growth is significantly lower for pegged regimes
omitted category. Again, the two intermediate categories in the (0.82 percent) and intermediate regimes (0.88 percent). The re-
Natural classification have been aggregated for better comparison gressions for the de jure classification are very sensitive to the in-
with the de jure classification. Other controls are similar to the clusion of country dummies. Without such fixed effects, growth
growth regressions above. They include factor accumulation (in- volatility is greater under pegged than floating regimes, though
vestment ratio, education and population growth); trade open- not significantly so. With country dummies, the opposite is true.
ness; terms-of-trade growth; and the size of government (tax ratio This explains the differences in results reported here from those
and central government balance). Each regression also includes of Ghosh, Gulde, and Wolf (2003) and, presumably, also of Levy-
country-specific fixed effects and year dummies. Yeyati and Sturzenegger (2003), since neither study includes
country-specific fixed effects. For the Natural classification, there
is little discernible pattern or significance with or without country
dummies.

33

III REGIME PERFORMANCE: INFLATION AND BUSINESS CYCLES

Figure 3.8.Volatility of Real GDP Growth Figure 3.9.Volatility of Real GDP Growth
Performance in Advanced Countries, and Contamination Across Regimes:
Emerging Markets, and Developing Evidence from Emerging Markets1
Countries1
(In percent)
(In percent)

4

4 3.59 2 0.60
(1.59) (0.79)
Limited flexibility
–0.92
3 Managed floating 2.70 0 (–1.04)

Freely floating (2.24)

2 1.34 –2 –0.84 –0.44
(–1.27) –1.84 (–0.25)

(–2.14) –3.08
(–1.48)
1.06 (1.86) –4
Freely
0.60 (1.96) floating
(1.08)
1 0.39 –6
(0.56)
0.14
0 (0.14) –8

–1 –0.15 One-year lagged regime
–0.48 (–0.40)
(–1.13) –10 Two-year lagged regime

–2 Emerging Developing –12 Managed
markets countries floating
Advanced Limited
countries flexibility

Source: Authors’ calculations. Source: Authors’ calculations.
1Figures in parentheses are t-statistics.The bars depict differ- 1Figures in parentheses are t-statistics.The bars depict differ-
ences in performance relative to pegged exchange rate regimes, ences in performance relative to pegged exchange rate regimes,
conditioning on a range of other variables. See appendix in conditioning on a range of other variables. See appendix in
Section III for details. Section III for details.

for emerging markets with their relatively high rate Figure 3.10.Volatility of Real GDP Growth
of transitions. Figure 3.9 investigates whether this Across Regimes: Emerging Markets for the
phenomenon is quantitatively important. It com- 1990s Only1
pares the estimated volatility effects of regimes that
prevailed in the previous one and two years (on the (In percent)
assumption that the spillover effects will be mani-
fested mainly in the first two years of transition to a 4
new regime). Now, the volatility in pegged regimes
is actually higher relative to limited flexibility and 2 0.58
freely floating, with little difference relative to
managed floating. These results are not all statisti- (0.66)
cally significant, but they point to significant
spillover effects when transitions from pegged 0
regimes occur. The implication is that the apparent
relationship in emerging markets between flexibil- –2 –2.97
ity and higher volatility is due largely to the volatil- (–1.33)
ity following the collapse of rigid regimes being at- –1.76
tributed to subsequent more flexible regimes. (–1.45)
Second, the transmission of volatility from rigid to
flexible regimes appears more so the case for the –4 –3.54
1990s, when the countries identified here as emerg-
ing markets began to tap international capital in a (–1.87)
significant manner (Figure 3.10). Together with
their higher likelihood of twin crises, as reported –6
above, this appears to further strengthen the case
against rigid regimes for emerging markets. –8 –6.59
(–1.30)
34 One-year lagged regime
–8.99
–10 Two-year lagged regime (–1.69)

–12 Managed Freely
floating floating
Limited
flexibility

Source: Authors’ calculations.
1Figures in parentheses are t-statistics.The bars depict differ-
ences in performance relative to pegged exchange rate regimes,
conditioning on a range of other variables. See appendix in
Section III for details.

Achieving Credibility in Developing and Emerging Economies

Achieving Credibility in Developing consistency of the macroeconomic stance with re-
and Emerging Economies spect to the exchange rate regime, its duration (the
number of years that a particular regime has been in
The results given above suggest that for develop- force) was treated as an additional dimension of its
ing economies a benefit in the form of lower infla- characteristics.
tion has been associated with pegged and interme-
diate regimes and that such a benefit has not come The results imply that in developing countries
at the expense of lower growth or higher volatility. pegged regimes delivered an inflation benefit even
Moreover, the inflation benefit of these relatively with no track record (i.e., with zero duration). In ad-
rigid regimes was found to accrue primarily dition, the duration dimension for pegged regimes
through a credibility effect rather than through was highly significant and negative (–0.2 percent per
greater monetary discipline. Following is an inves- year). Additionally, the calculations show that coun-
tigation as to how developing countries can further tries that maintained pegged regimes over a period
enhance the credibility of their exchange rate of 10 years, for example, may have earned an addi-
regimes to improve macroeconomic performance. tional inflation benefit of more than 80 percent over
With respect to emerging markets, which do not the initial inflation gain (see Figure 3.12).
seem to derive appreciable benefits from rigid
regimes but also fear to float, the performance of Taken together, the lesson appears to be that de-
floating regimes and whether they can be improved veloping countries that announce their peg and are
is examined below. able to maintain them over longer durations derive
greater benefits from the rigidity in exchange rate
Announcement Effects regimes. While this finding is prima facie encourag-
ing, it may be an insufficient policy guide in the
Is there an incremental inflation benefit associated context of the growing importance of international
with officially announcing an operative pegged capital flows. Obstfeld and Rogoff (1995) note that
regime? The presumption is that announcement im- most countries with long-lasting pegs adopted them
plies a stronger commitment to maintaining the peg in times when global capital markets were relatively
and hence to policies that are supportive of that shallow. Having achieved credibility during that
regime. To consider the announcement effect, the
overlap between de facto and de jure regimes was Figure 3.11.The Inflation Benefit
identified. The statistical task was to determine if the Associated with Announcement in
overlap added value to the regime. Developing Countries1

Among developing countries, this announcement (In percent)
effect is large and significant for pegged regimes.
As Figure 3.11 shows, once a separate announce- 20
ment effect is allowed for, the small number of de-
veloping countries that pursued exchange rate pegs Announcement effect
without explicitly announcing that policy exhibited
average inflation that, if anything, was somewhat 15 Pure regime effect
higher than that in other (especially intermediate)
regimes. In other words, the inflation benefit of 10 4.90 2.28
pegged regimes identified above did not derive (3.21) (0.54)
merely from operating a tightly managed exchange 3.87 1.31
rate. The big gain came only when the peg was offi- –6.68 (0.23)
cial. In Figure 3.11 it is also interesting to note that 5 (2.85) (–1.49)
the announcement of other regimes had the opposite Freely
effect of raising inflation. Thus the announcement 0 Managed floating
benefit differentiated pegs from other regimes in an floating
important way. –5 –6.47
(–1.40)
Regime Duration –10 –9.18

Regimes that last longer presumably do so be- (–2.05)
cause macroeconomic policies are maintained in a
consistent manner over time. If so, longer duration –15
should add to a regime’s credibility and be associ-
ated with superior performance. To proxy for the –20 Limited
flexibility
Pegged

Source: Authors’ calculations.
1Figures in parentheses are t-statistics.The bars depict differ-
ences in performance relative to pegged exchange rate regimes,
conditioning on a range of other variables. See appendix in
Section III for details.

35

III REGIME PERFORMANCE: INFLATION AND BUSINESS CYCLES

Figure 3.12.The Inflation Benefit Associated Opposing views exist on the feasibility of learn-
with Consistent Macroeconomic Policies in ing to float. One fairly pessimistic view (Eichen-
Developing Countries1,2 green, Hausmann, and Panizza, 2004) hold that the
risk of a sharp depreciation under floating rates will
(In percent) depress investment activity in most emerging mar-
kets because (unhedged) foreign currency borrow-
20 Cumulative 10-year duration effect 15.06 ing will always be significant. This handicap, the
15 Pure regime effect (3.82) authors argue, cannot be overcome without coordi-
10 nated international action to facilitate countries’
5 –2.16 2.01 1.04 borrowing in their own currency. As a result, float-
0 (–2.13) (0.91) (0.49) ing exchange rates, in this view, will remain mostly
–5 a mirage.
–10 –1.61 –0.13
–15 (–1.25) (–0.17) Another perspective starts by noting that floating
–20 is relatively new, and the experience with it thus far
–18.31 has been fairly positive (Edwards and Savastano,
(–3.00) 1999; and Larraín and Velasco, 2001). While in prac-
tice emerging market floaters are far from pure, in
Pegged Limited Managed Freely that intervention is common and the authorities gen-
flexibility floating floating erally take the exchange rate directly into account in
setting monetary policy, there has been meaningful
Source: Authors’ calculations. and in some ways effective floating. Inflation objec-
1Figures in parentheses are t-statistics.The bars depict differ- tives have been met and countercyclical policy has
ences in performance relative to pegged exchange rate regimes, been possible. Goldstein (2002) summarizes the
conditioning on a range of other variables. See appendix in available evidence as suggesting that emerging
Section III for details. markets can conduct floating in combination with
2Measured as the additional inflation benefit from maintaining inflation-targeting monetary policy and measures to
a given regime for 10 years. discourage currency mismatch in a way that credibly
achieves low inflation; buffers external shocks, such
less-demanding period, they were often able to as to the terms-of-trade; and provides some indepen-
maintain the pegs even with greater exposure to in- dence of monetary policy. Ho and McCauley (2003)
ternational capital. Those countries seeking to es- argue, in their review of recent experience that
tablish credibility in the current context, however, where exchange rate considerations have been op-
are likely to find that a more challenging task. posed to inflation targets, inflation has typically
been the primary objective of policy.56
Learning to Float
There is reason to believe emerging market
What does the future hold for emerging markets, economies can improve the flexibility and effective-
particularly middle-income, open–capital account ness of their floats over time. Such learning to float
countries? These rigid regimes run the risk of trig- could take place through two main channels. The first
gering crises, and concerns arising from large is through the acquisition of confidence and experi-
movements or excessive volatility of the exchange ence on the part of the authorities. The authorities
rate limit the extent of flexibility that policymakers themselves need to learn how to conduct monetary
are willing to allow. These countries currently man- policy appropriate to a flexible exchange rate. It may
age their exchange rates to varying degrees while take time, for example, for the central bank to refine
pursuing domestic monetary policies—increasingly the new internal procedures and communication
some variant of inflation targeting—to anchor in- strategies involved in inflation targeting. Moreover,
flationary expectations. The finding reported ear- the authorities may need time and experience to build
lier that floating becomes a superior alternative as trust in their own framework and to become comfort-
institutional capabilities become stronger raises the able with allowing substantial exchange rate flexibil-
possibility that the more developed emerging mar- ity. The second is through modified behavior on the
ket economies may wish to anticipate a further part of private agents, who may adjust their behavior
move to floating and hence begin to invest in learn- as they observe flexible exchange rates in action and
ing to float. come to appreciate the risks involved in unhedged for-

56See also Goldstein and Turner (2004); Berg, Borensztein, and
Mauro (2002); Goldfajn, and Olivares (2001), and references
therein for useful surveys. See also Schmidt-Hebbel and Werner
(2002) on the experiences of Chile, Mexico, and Brazil.

36

Appendix

eign exchange positions. This adjustment in behavior exposure of Mexican firms to devaluation risk has
would, in turn, reduce banking system dollarization lessened with the flexible exchange rate regime in
as lenders and borrowers appreciate and price the place since the 1994/1995 crisis.
risks involved in currency mismatch (Ize and Levy-
Yeyati, 2003). Similarly, expectations that the central Finally, Brazil is a third country that has been float-
bank will in fact allow exchange rate flexibility may ing its exchange rate while at the same time building a
diminish incentives to accept excessive foreign cur- track record of low and stable inflationary expecta-
rency–denominated capital inflows (Caballero and tions through inflation targeting. Fraga, Goldfajn, and
Krishnamurthy, 2002). Finally, as private agents ob- Minella (2003) describe the challenges associated
serve that the authorities can keep inflation low in the with building credibility in an environment that is
context of a floating exchange rate regime, their infla- characterized by significant volatility. It notes that in
tion expectations may respond less to movements in an emerging market environment exchange rate pol-
the exchange rate, thus reducing the pass-through icy and inflation targeting cannot be easily dissociated
from exchange rates to inflation. The dynamics asso- because a history of monetary instability tends to
ciated with learning to float would allow a sort of vir- make the exchange rate a focal point for inflationary
tuous circle, at least for those countries that can expectations, and foreign currency borrowing subjects
demonstrate some initial effectiveness in floating.57 domestic firms and financial institutions to significant
risks. The authors recommend a gradual learning
The experience of at least three emerging market process that includes high levels of communications
floaters may be consistent with this dynamic. Con- and transparency on the part of the central bank.
sider the case of Chile, which in the late 1990s tran-
sited from a framework with both an inflation target Appendix.
and an explicit exchange rate band to a more pure Data and Regression Results for
form of floating. Chile went through two episodes of Economic Performance Analysis
exchange rate pressure, in late 1998 and late 2000.
In the first episode, associated with the Russian and This appendix describes the data used in Section III
long-term capital management crises, interest rates and reports the detailed regression results that lie be-
increased sharply as the authorities defended the ex- hind the key findings discussed with respect to eco-
change rate within the band. Thus, the weakening nomic performance across exchange rate regimes.
exchange rate was accompanied by a sharp interest
rate increase, as well as a sharp recession. In the sec- Much of the data are taken from Ghosh, Gulde,
ond episode of exchange market pressure, in late and Wolf (2003), including the de jure classification
2000, the authorities allowed the currency to float, in of exchange rate regimes, the three measures of eco-
line with the new exchange rate arrangement intro- nomic performance (inflation, growth, growth
duced in August 1999, according to which the ex- volatility), and the control (or explanatory) variables
change band was discontinued and intervention was used in the regression analysis. Each variable is cov-
limited to extreme circumstances. ered at an annual frequency from 1970 to 1999 for
up to 158 countries. The control variables are drawn
Mexico is another country that has seen inflation from the literature and are thought to provide a suit-
come down in the context of a regime that has also be- able explanation of the variations in the performance
come gradually more flexible. Once the immediate measures. Table A3.1 provides a detailed description
postcrisis period was over in 1995, the authorities of the data. It lists each variable, provides a brief de-
paid substantial attention to the exchange rate in the scription, and notes which of the subsequent regres-
conduct of their monetary policy. Over time, they sions feature these variables. Using this data has the
adopted more formal inflation targeting and allowed advantage of allowing the evaluation of performance
substantial movements in the real exchange rate. In- under the Natural classification to be directly com-
flation and both nominal and real interest rates have pared to a well-respected baseline that assesses per-
come down fairly steadily. Inflation persistence has formance across the de jure regimes.
declined over time, suggesting perhaps an increasing
credibility of the monetary authorities.58 At the same Three groups of variables are not covered in the
time, Martínez and Werner (2002) conclude that the Ghosh, Gulde, and Wolf (2003) data. The first group
is the Natural regime classification, available at an an-
57Many authors, including Mussa and others (2000), Goldstein nual frequency from http://www.puaf.umd.edu/fac-
(2002), and Jeanne (2003), have noted that implementing a float- ulty/papers/reinhart/papers.htm.59 The second group
ing exchange rate regime with a credible monetary policy may in is the crisis variables. The banking crisis variable is
turn increase the effectiveness of floating. Also, as countries are taken from Demirgüç-Kunt and Detragiache (1998).
more successful in floating their currencies, there will be less rea-
son to keep exchange controls that are not necessary. 59The data are also available at monthly frequency.

58See Carstens and Werner (2000), and Edwards and Savastano
(1998).

37

III REGIME PERFORMANCE: INFLATION AND BUSINESS CYCLESTable A3.1. Variable Description
38
Variable Description Regression Used

IMF de jure regime Pegged Pegged regimes 13.1, 13.3, 13.5
dummies Intermediate Intermediate regimes 13.1, 13.3, 13.5
Floating Floating regimes 13.1, 13.3, 13.5

Natural regime Pegged Pegged regimes (fine classification 1~4) 13.2, 13.4, 13.6, 14.1~18.2
dummies Limited flexibility Limited flexibility regimes (fine classification 5~9) 14.1~18.2
Managed floating Managed floating regimes (fine classification 10~12) 14.1~18.2
Intermediate Limited flexibility or managed floating regimes (fine classification 5~12) 13.2, 13.4, 13.6
Freely floating Freely floating regimes (fine classification 13) 13.2, 13.4, 13.6, 14.1~18.2
Freely falling Freely falling regimes (fine classification 14) 13.2, 13.4, 13.6, 14.1~18.2

Dependent Variable Control Variables 13.1, 13.2, 14.1~14.9, 17.1, 17.2, 18.1, 18.2
13.1, 13.2, 14.1~14.9, 17.1, 17.2, 18.1, 18.2
Inflation (scaled Money growth Broad money growth 13.1, 13.2, 14.1~14.9, 17.1, 17.2, 18.1, 18.2
consumer price Real GDP growth Real GDP growth 13.1, 13.2, 14.1~14.9, 17.1, 17.2, 18.1, 18.2
inflation; p/(1+p)) Trade openness Exports plus imports of goods and services (percent of GDP) 13.1, 13.2, 14.1~14.9, 17.1, 17.2, 18.1, 18.2
Central bank turnover rate Central bank governor turnover rate (per five years) 13.1, 13.2, 14.1~14.9, 17.1, 17.2, 18.1, 18.2
Terms-of-trade growth Terms-of-trade growth 18.1
Government balance Central government balance (percent of GDP) 18.1
Common pegged Overlap of natural classification pegged with IMF de jure pegged 18.1
Common limited flexibility Overlap of natural classification limited flexibility with IMF de jure intermediate 18.1
Common managed floating Overlap of natural classification managed floating with IMF de jure intermediate 18.2
Common freely floating Overlap of natural classification freely floating with IMF de jure floating 18.2
Pegged duration Linear time trend for pegged (natural) that measures regime duration 18.2
Limited flexibility duration Linear time trend for limited flexibility (natural) that measures regime duration 18.2
Managed floating duration Linear time trend for managed floating (natural) that measures regime duration 18.2
Freely floating duration Linear time trend for freely floating (natural) that measures regime duration
Freely falling duration Linear time trend for freely falling (natural) that measures regime duration 13.3, 13.4, 15.1~15.9, 17.3, 17.4
13.3, 13.4, 15.1~15.9, 17.3, 17.4
Per capita real Investment ratio Gross fixed investment (percent of GDP) 13.3, 13.4, 15.1~15.9, 17.3, 17.4
GDP growth Trade openness Exports plus imports of goods and services (percent of GDP) 13.3, 13.4, 15.1~15.9, 17.3, 17.4
Terms-of-trade growth Terms-of-trade growth, three-year backward average
Average years of schooling Average number of years of schooling of total population age 25 and older 13.3, 13.4, 15.1~15.9, 17.3, 17.4
(per five-years) 13.3, 13.4, 15.1~15.9, 17.3, 17.4
Tax ratio General government revenue (percent of GDP), three-year backward average 13.3, 13.4, 15.1~15.9, 17.3, 17.4
Government balance Central government balance (percent of GDP), three-year backward average 13.3, 13.4, 15.1~15.9, 17.3, 17.4
Initial income/U.S. income Ratio of per capita GDP to U.S. per capita GDP in 1970 (international prices) 13.3, 13.4, 15.1~15.9, 17.3, 17.4
Population growth Population growth, three-year backward average
Population size Population (logarithm)

Volatility of real GDP Investment ratio Gross fixed investment (percent of GDP), three-year centered standard deviation 13.5, 13.6, 16.1~16.9, 17.5, 17.6
Exports plus imports of goods and services (percent of GDP) 13.5, 13.6, 16.1~16.9, 17.5, 17.6
growth (three- Trade openness Terms-of-trade growth, three-year centered standard deviation 13.5, 13.6, 16.1~16.9, 17.5, 17.6
Average number of years of schooling of total population age 25 and older 13.5, 13.6, 16.1~16.9, 17.5, 17.6
year centered Terms-of-trade growth (per five years)
General government revenue (percent of GDP), three-year backward average 13.5, 13.6, 16.1~16.9, 17.5, 17.6
standard Average years of schooling Central government balance (percent of GDP), three-year backward average 13.5, 13.6, 16.1~16.9, 17.5, 17.6
Ratio of per capita GDP to U.S. per capita GDP in 1970 (international prices) 13.5, 13.6, 16.1~16.9, 17.5, 17.6
deviation) Population growth, three-year backward average 13.5, 13.6, 16.1~16.9, 17.5, 17.6

Tax ratio Crisis Variables none
none
Government balance From Demirgüç-Kunt and Detragiache (1998)
From Berg, Borensztein and Pattillo (2004)
Initial income/U.S. income

Population growth

Bank crisis
Balance of payments crisis

Appendix
39

III REGIME PERFORMANCE: INFLATION AND BUSINESS CYCLES

Table A3.2. List of Countries

Advanced _E_m__e_r_g_in_g_M__a_r_k_et_s_1_ ______________________D__e_v_e_lo_p_i_n_g_C_o__u_n_tr_ie_s_____________________________
C__o_u_n_t_ri_e_s___

Australia Argentina Albania Georgia Panama
Austria Brazil Algeria Ghana Papua New Guinea
Belgium Chile Antigua and Barbuda Grenada Paraguay
Canada China Armenia Guatemala Romania
Cyprus Colombia Azerbaijan Guinea Rwanda
Denmark Czech Republic Bahrain, Kingdom of Guinea-Bissau St. Lucia
Finland Egypt Bangladesh Guyana St. Vincent and the
France Hungary Barbados Haiti
Germany India Belarus Honduras Grenadines
Greece Indonesia Belize Iran, Islamic Republic of Samoa
Hong Kong SAR Israel Benin Iraq Senegal
Iceland Jordan Bolivia Jamaica Seychelles
Ireland Korea, Republic of Bosnia and Herzegovina Kazakhstan Sierra Leone
Italy Malaysia Botswana Kenya Slovak Republic
Japan Mexico Bulgaria Kyrgyz Republic Sri Lanka
Kuwait Morocco Burkina Faso Lao People’s Democratic Sudan
Luxembourg Pakistan Burundi Suriname
Netherlands Peru Cameroon Republic Swaziland
New Zealand Philippines Cape Verde Latvia Syrian Arab Republic
Norway Poland Central African Republic Lebanon Tajikistan
Portugal Rep. Bolivariana de Chad Lesotho Tanzania
Qatar Comoros Liberia Togo
Singapore Venezuela Congo, Dem. Rep. of Libyan Arab Jamahiriya Tonga
Slovenia Russian Federation Congo, Rep. of Lithuania Trinidad and Tobago
Spain South Africa Costa Rica Madagascar Tunisia
Sweden Thailand Côte d’Ivoire Malawi Turkmenistan
Switzerland Turkey Djibouti Maldives Uganda
United Arab Emirates Dominica Mali Ukraine
United Kingdom Dominican Republic Malta Uruguay
United States Ecuador Mauritania Vietnam
El Salvador Mauritius Zambia
Equatorial Guinea Moldova Zimbabwe
Estonia Mozambique
Ethiopia Myanmar
Fiji Nepal
Gabon Nicaragua
Gambia,The Niger
Nigeria

1Emerging market economies are those that are included in the Morgan Stanley Capital International (MSCI) index. With the exception of Israel, which
is in the MSCI index, advanced economies are those that are classified as upper-income economies by the World Bank. All other economies constitute
the developing countries group.

The authors declare a banking crisis to have occurred emerging market, or a developing country. Advanced
when any one of the following four conditions held: countries are defined using the World Bank defini-
nonperforming loans exceeded 10 percent of banking tion for upper-income countries, following Ghosh,
system assets; a bailout cost 2 percent or more of Gulde, and Wolf (2003). In dividing the rest of the
GDP; large-scale nationalization occurred; or other world into two further groups, the analytical distinc-
emergency measures, such as bank holidays, deposit tion of relevance was their degree of exposure to in-
freezes, and special guarantees, had to be undertaken. ternational capital markets. Those considered to
The currency or balance of payments crisis variable is have high exposure were classified as emerging mar-
taken from Berg, Borensztein, and Pattillo (2004), kets, and the rest were designated developing.60
which declares a crisis as having occurred when the Table A3.2 lists the country composition of the
weighted average of one-month changes in exchange
rate and reserves is more than three (country-specific) 60Strictly, since all nonadvanced (non–upper income) countries
standard deviations above the country average. are commonly referred to as developing, the two categories used
in this paper could have been referred to as emerging and other
The final group of variables defines whether a developing.
country is classified as an advanced economy, an

40

Appendix

Table A3.3. Comparing IMF De Jure and Natural Classifications1

Dependent Variable _______I_n_fl_a_ti_o_n_______ Per Capita Volatility of
Classification ___R_e_a_l_G__D_P__G_r_o_w_t_h___ ___R_e_a_l_G__D_P__G_r_o_w_t_h___

De jure Natural De jure Natural De jure Natural

Regression number 13.1 13.2 13.3 13.4 13.5 13.6
Intermediate (de jure)
Floating (de jure) 0.062 0.000 –0.001
Intermediate (natural) (6.58)*** (–0.05) (–0.31)
Freely floating (natural) 0.062
Freely falling (natural) (5.13)*** 0.003 0.008
Money growth (0.57) (3.11)***
Real GDP growth 0.113
Investment ratio (0.83) 0.016 0.000 –0.019 0.005
Trade openness –0.958 (2.11)** (0.00) (–0.27) (1.19)
Central bank turnover rate (–3.40)*** 0.045 0.008
Terms-of-trade growth (2.91)*** 0.004 0.007 (1.24)
Average years of schooling 0.022 0.218 (0.47) (1.06) 0.012
Tax ratio (1.29) (13.51)*** (2.69)***
Government balance 0.048 0.132 –0.025 0.019
Initial income/U.S. income (3.34)*** (1.05) (–3.67)*** (2.02)** –0.012
Population growth –0.005 –0.555 0.004 (–0.17)
Population size (–0.25) (–2.59)*** –0.083 –0.090 (0.93)
Constant (–2.62)*** (–2.83)*** –0.010 0.008
Observations –0.361 0.027 (–0.76) (1.22)
R-squared (–4.99)*** (1.90)* 0.025 0.029 –0.008
0.026 (2.47)** (2.94)*** (–0.40) 0.019
0.366 (2.05)** –0.006 (2.02)**
(7.72)*** 0.007 0.030 0.026 (–0.40) 0.004
1,946 (0.39) (2.09)** (1.83)* 0.123 (0.82)
(1.02) –0.005
–0.239 0.002 0.002 (–0.36)
(–4.07)*** (–0.83) (–0.60) 0.032 –0.003
(3.72)*** (–0.16)
0.090 0.004 –0.006 1,878 –0.008
(1.51) (0.14) (–0.25) (0.55)
1,946 0.111
0.008 0.004 (0.91)
(0.25) (–0.11)
0.031
–0.063 –0.064 (2.53)**
(–2.51)** (–2.39)** 1,878

–0.275 –0.321
(–1.68)* (–1.90)*

0.013 0.015
(1.01) (1.07)

0.091 0.093
(2.14)** (2.07)**

1,762 1,762

0.58 0.70 0.29 0.31 0.26 0.26

Source: Author’s calculations.
1Figures in parentheses are t-statistics; *significant at 10 percent; **significant at 5 percent; ***significant at 1 percent.

advanced, emerging market, and developing country (the actual exposure a country faces). In the spirit of
groups. this paper, a de facto definition was appropriate, an
approach also followed by Prasad and others (2003).
To distinguish between emerging and developing Because there are no well-defined or generally
economies, exposure to international capital can be accepted thresholds of exposure to international capi-
determined either in a de jure sense (the extent of for- tal, the cutoff between high and low exposure can be
mal capital controls in place) or in a de facto sense

41