A Retrospective on the Bretton Woods System: Lessons for International Monetary Reform 9780226066905

Citation preview

A Retrospective on the Bretton Woods System

A National Bureau of Economic Research Project Report

A Retrospective on the Bretton Woods System Lessons for International Monetary Reform

Edited by

Michael D. Bordo and Barry Eichengreen

The University of Chicago Press Chicago and London

MICHAELD. BORDOis professor of economics at Rutgers University. BARRYEICHENGREEN is professor of economics at the University of California, Berkeley. Both are research associates of the National Bureau of Economic Research.

The University of Chicago Press, Chicago 60637 The University of Chicago Press, Ltd., London

0 1993 by the National Bureau of Economic Research All rights reserved. Published 1993 Printed in the United States of America 02010099989796959493 ISBN: 0-226-06587-1

12345

(cloth)

Library of Congress Cataloging-in-Publication Data A Retrospective on the Bretton Woods system : lessons for international monetary reform / edited by Michael D. Bordo and Barry Eichengreen. p. cm.-(A National Bureau of Economic Research project report) Includes bibliographical references and index. 1. International finance-History. 2. Monetary policy. I. Bordo, Michael D. 11. Eichengreen, Barry J. 111. Series. HG3881.R425 1993 92-30913 332.4'556-dc20 CIP

Q The paper used in this publication meets the minimum requirements of the American National Standard for Information Sciences-Permanence of Paper for Printed Library Materials, ANSI 239.48-1984.

National Bureau of Economic Research Officers George T. Conklin, Jr., chairman Paul W. McCracken, vice chairman Martin Feldstein, president and chief executive oficer

Geoffrey Carliner, executive director Charles A. Walworth, treasurer Sam Parker, director of finance and administration

Directors at Large John H. Biggs Andrew Brimmer Carl F. Christ George T. Conklin, Jr. Don R. Conlan Kathleen B . Cooper Jean A. Crockett George C. Eads

Martin Feldstein George Hatsopoulos Lawrence R. Klein Franklin A. Lindsay Paul W. McCracken Leo Melamed Robert T. Parry

Peter G. Peterson Douglas D. h r v i s Robert V. Roosa Richard N. Rosett Bert Seidman Eli Shapiro Donald S . Wasserman

Directors by University Appointment Jagdish Bhagwati, Columbia William C. Brainard, Yale Glen G . Cain, Wisconsin Franklin Fisher, Massachusetts Institute of Technology Saul H. Hymans, Michigan Marjorie B. McElroy, Duke

James L. Pierce, California, Berkeley Andrew Postlewaite, Pennsylvania Nathan Rosenberg, Stanford Harold T. Shapiro, Princeton Craig Swan, Minnesota Michael Yoshino, Haward Arnold Zellner, Chicago

Directors by Appointment of Other Organizations Marcel Boyer, Canadian Economics Association Rueben C. Buse, American Agricultural Economics Association Richard A. Easterlin, Economic History Association Gail Fosler, The Conference Board A. Ronald Gallant, American Statistical Association Robert S. Hamada, American Finance Association

Charles Lave, American Economic Association Rudolph A. Oswald, American Federation of Labor and Congress of Industrial Organizations Dean P. Phypers, Committee for Economic Development James F. Smith, National Association of Business Economists Charles A. Walworth, American Institute of Certijied Public Accountants

Directors Emeriti Moses Abramovitz Emilio G . Collado Thomas D. Flynn

Gottfried Haberler Geoffrey H. Moore James J. O’Leary

George B . Roberts William S. Vickrey

Relation of the Directors to the Work and Publications of the National Bureau of Economic Research 1. The object of the National Bureau of Economic Research is to ascertain and to present to the public important economic facts and their interpretation in a scientific and impartial manner. The Board of Directors is charged with the responsibility of ensuring that the work of the National Bureau is carried on in strict conformity with this object. 2. The President of the National Bureau shall submit to the Board of Directors, or to its Executive Committee, for their formal adoption all specific proposals for research to be instituted. 3. No research report shall be published by the National Bureau until the President has sent each member of the Board a notice that a manuscript is recommended for publication and that in the President’s opinion it is suitable for publication in accordance with the principles of the National Bureau. Such notification will include an abstract or summary of the manuscript’s content and a response form for use by those Directors who desire a copy of the manuscript for review. Each manuscript shall contain a summary drawing attention to the nature and treatment of the problem studied, the character of the data and their utilization in the report, and the main conclusions reached. 4.For each manuscript so submitted, a special committee of the Directors (including Directors Emeriti) shall be appointed by majority agreement of the President and Vice Presidents (or by the Executive Committee in case of inability to decide on the part of the President and Vice Presidents), consisting of the three Directors selected as nearly as may be one from each general division of the Board. The names of the special manuscript committee shall be stated to each Director when notice of the proposed publication is submitted to him. It shall be the duty of each member of the special manuscript committee to read the manuscript. If each member of the manuscript committee signifies his approval within thirty days of the transmittal of the manuscript, the report may be published. If at the end of that period any member of the manuscript committee withholds his approval, the President shall then notify each member of the Board, requesting approval or disapproval of publication, and thirty days additional shall be granted for this purpose. The manuscript shall then not be published unless at least a majority of the entire Board who shall have voted on the proposal within the time fixed for the receipt of votes shall have approved. 5 . No manuscript may be published, though approved by each member of the special manuscript committee, until forty-five days have elapsed from the transmittal of the report in manuscript form. The interval is allowed for the receipt of any memorandum of dissent or reservation, together with a brief statement of his reasons, that any member may wish to express; and such memorandum of dissent or reservation shall be published with the manuscript if he so desires. Publication does not, however, imply that each member of the Board has read the manuscript, or that either members of the Board in general or the special committee have passed on its validity in every detail. 6. Publications of the National Bureau issued for informational purposes concerning the work of the Bureau and its staff, or issued to inform the public of activities of Bureau staff, and volumes issued as a result of various conferences involving the National Bureau shall contain a specific disclaimer noting that such publication has not passed through the normal review procedures required in this resolution. The Executive Committee of the Board is charged with review of all such publications from time to time to ensure that they do not take on the character of formal research reports of the National Bureau, requiring formal Board approval. 7. Unless otherwise determined by the Board or exempted by the terms of paragraph 6, a copy of this resolution shall be printed in each National Bureau publication.

(Resolution adopted October 25, 1926, us revised through September 30, 1974)

Contents

Preface Michael D. Bordo and Barry Eichengreen

I.

xi

OVERVIEWS AND ORIGINS 1. The Bretton Woods International Monetary System: A Historical Overview Michael D. Bordo Comment: Rudiger Dornbusch Comment: Richard N. Cooper General Discussion

3

2. Bretton Woods and Its Precursors: Rules versus Discretion in the History of International Monetary Regimes 109 Albert0 Giovannini Comment: Anna J. Schwartz Comment: Charles Wyplosz General Discussion 3. The Political Origins of Bretton Woods G. John Ikenberry Comment: John S. Ode11 Comment: L. S. Pressnell General Discussion 11.

BRETTON WOODS IN OPERATION 4. The Adjustment Mechanism Maurice Obstfeld Comment: Robert Z. Aliber

vii

155

20 1

viii

Contents Comment: Vittorio U . Grilli General Discussion

5. The Provision of Liquidity in the Bretton Woods System Hans Genberg and Alexander K. Swoboda Comment: Stanley W. Black Comment: John Williamson General Discussion

6. International lkansmission under Bretton Woods Alan C. Stockman Comment: Toru Iwami Comment: Bennett T. McCallum General Discussion

269

317

7. The Role of International Organizations in the 357 Bretton Woods System Kathryn M. Dominguez Comment: Albert0 Alesina Comment: William H. Branson General Discussion

8. Devaluation Controversies in the Developing Countries: Lessons from the Bretton Woods Era Sebastian Edwards and Julio A. Santaella Comment: Stanley Fischer Comment: Albert Fishlow General Discussion

9. The Collapse of the Bretton Woods Fixed Exchange Rate System Peter M. Garber Comment: Willem H . Buiter General Discussion

10. Panel Session I: Retrospectives Michael Mussa, chair Edward M. Bemstein, W. Max Corden, and Robert Solomon

405

46 1

495

ix

Contents

111.

THELEGACY OF BRETTON WOODS 11. Interest Differentials under Bretton Woods and the Post-Bretton Woods Float: The Effects of Capital Controls and Exchange Risk 515 Richard C. Marston Comment: Paul Krugman Comment: Allan H. Meltzer General Discussion 12. Attitudes toward Inflation and the

Viability of Fixed Exchange Rates: Evidence from the EMS Susan M. Collins and Francesco Giavazzi Comment: Michele Fratianni Comment: Niels Thygesen General Discussion

547

13. Panel Session 11: Implications for International 587 Monetary Reform Barry Eichengreen, chair C. Fred Bergsten, Stanley Fischer, Ronald I. McKinnon, Robert Mundell, and Martin Feldstein IV.

CONCLUSION 14. Epilogue: Three Perspectives on the Bretton

Woods System Barry Eichengreen

62 1

Contributors

659

Author Index

663

Subject Index

669

This Page Intentionally Left Blank

Preface

In 1944, delegates from forty-four countries assembled at the Mt. Washington Hotel in Bretton Woods, New Hampshire, to negotiate a “new world order” governing exchange rates, foreign lending, and international trade. Among their most notable achievements was the Bretton Woods System of pegged but adjustable exchange rates and a new institution, the International Monetary Fund, to oversee the operation of exchange rates. The quarter century that followed, what came to be known as the “Bretton Woods years,” appears in retrospect as a golden age of exchange rate stability and rapid economic growth. The exchange rates of the major industrial countries remained fixed for extended periods. Inflation was moderate by subsequent standards. World trade expanded buoyantly. National income in the G7 countries rose more rapidly than in any comparable period before or since. It is tempting to assume, as many have done, that the key to this admirable performance lay in the international monetary agreement concluded in 1944. The beginning of the end of the Bretton Woods era came on 15 August 1971, when U.S. President Richard Nixon shut the gold window, suspending the official convertibility of the dollar into gold at $35 an ounce and cutting the exchange rate system loose from its moorings. Parities were realigned but stabilized only temporarily. By 1973, the transition to a new regime of floating exchange rates was complete. The real and financial turbulence of the succeeding period far outstripped anything witnessed during the Bretton Woods years. The twentieth anniversary of the suspension of gold convertibility and of the collapse of the pegged-rate system seemed an appropriate time to revisit the Mt. Washington Hotel in order to reassess the lessons of Bretton Woods. In the autumn of 1991, more than sixty academics and policy makers therefore returned to the scene for the conference whose proceedings are published here. This was not the first conference, of course, reassessing Bretton Woods. xi

xii

Preface

But, for a number of reasons, the organizers felt that it might be possible to view Bretton Woods in a new and revealing light. Theoretical advances in economics-for example, work on time consistency and credibility-had begun to build a framework within which the concept of a “Bretton Woods regime” might be systematically analyzed. Ongoing policy problems had added to the nostalgia with which the pegged exchange rates of Bretton Woods were viewed. The exchange rate gyrations of the 1980s had sharpened dissatisfaction with floating exchange rates and heightened interest in the pegged-rate systems of previous years. The European Monetary System, established in 1979, had lent increasing stability to the exchange rates of the participating countries and provoked discussion of exchange rate stabilization over a wider area. In the fall of 1991, the members of the European Community were moving rapidly to unify their currencies-that is, to fix their exchange rates once and for all through the establishment of a common currency and a European central bank. Clearly, arguments to stabilize exchange rates were back in vogue. Simultaneously, economic and political upheavals in Eastern Europe and in the successor states of the Soviet Union, the stagnation of portfolio capital flows to Latin America, and the troubled progress of the Uruguay Round of GATT negotiations combined to produce calls for a new Bretton Woods Agreement to reconstruct international institutions in ways that would address these pressing problems. It was with these issues in mind that the NBER conference convened to consider what made possible the successful conclusion of the original Bretton Woods Agreement and to analyze its effects. A noteworthy and enjoyable aspect of the meeting was that it assembled in one place three distinct generations of international monetary economists. Among the participants were a number of academics and officials who had been “present at the creation.” Along with a subsequent generation of economists and officials currently at the center of academic and policy debates, they participated in panels and roundtables in which historical issues and policy implications were discussed. Formal papers were presented by a still younger generation of scholars whose professional careers by and large postdate the Bretton Woods era. The papers, as will be evident from the table of contents to this volume, fall into a number of groups. The first set provides an overview of Bretton Woods from three different vantage points. Chapter 1, by Michael D. Bordo, surveys the literature on Bretton Woods and analyzes its performance in comparative perspective. A companion paper by Albert0 Giovannini offers a detailed comparison of the behavior of exchange rates under Bretton Woods and its predecessors. G. John Ikenberry’s chapter then provides a political scientist’s analysis and explanation for the conclusion of the Bretton Woods Agreement. The next set of chapters addresses first the successful operation of Bretton Woods and then its collapse. Maurice Obstfeld, Hans Genberg and Alexander K. Swoboda, and Alan C . Stockman analyze three central facets of the system’s operation: balance-of-payments adjustment, the international transmis-

xiii

Preface

sion of business cycles, and the provision of international liquidity. Kathryn M. Dominguez assesses the role of the IMF in Bretton Woods's operation, while Sebastian Edwards and Julio A. Santaella view this and related questions from the vantage point of the developing countries. Finally, Peter M. Garber reassesses the long-standing debate over the causes of Bretton Woods's collapse. Much as historical experience can shed light on subsequent economic developments, subsequent developments provide an additional lens through which history can be viewed. Chapters by Richard C. Marston and by Susan M. Collins and Francesco Giavazzi therefore compare aspects of the Bretton Woods experience with the post-1973 float and the European Monetary System, respectively. The volume also contains a pair of panel discussions that reflect on the historical experience with Bretton Woods and look forward to the future of the international monetary system. The comments in these panels reflect the personal views of the participants, which may contain policy prescriptions. The volume concludes with an epilogue by Barry Eichengreen summarizing the main findings to emerge from the conference and their implications for international monetary reform. As with all NBER conferences, as organizers of this one we benefited greatly from the intellectual and logistical support provided by the Bureau. Martin Feldstein and Geoffrey Carliner helped with the scholarly menu, while Kirsten Foss Davis and Ilana Hardesty dealt with the gastronomic one and, with characteristic good humor, managed to ferry the conference participants to northern New Hampshire at the height of the leaf-turning season. Jane Konkel in Cambridge and Julie McCarthy in Chicago shepherded the manuscript into print. The Ford Foundation provided financial support for the research program of which this publication is a part. We are grateful to all of them. Michael D . Bordo and Barry Eichengreen

This Page Intentionally Left Blank

1

The Bretton Woods International Monetary System: A Historical Overview Michael D. Bordo

After twenty years of floating exchange rates, there is now considerable interest, among those concerned over its perceived shortcomings, in an eventual return by the world to a fixed exchange rate regime. This interest has been enhanced by the apparent success of the European Monetary System (EMS) and the prospects for European monetary unification. The Bretton Woods system was the world’s most recent experiment with a fixed exchange rate regime. Although it was originally designed as an adjustable peg, it evolved in its heyday into a de fact0 fixed exchange rate regime. That regime ended with the closing by President Richard Nixon of the gold window on 15 August 1971. Twenty years after that momentous decision, a retrospective look at the performance of the Bretton Woods system is timely. This paper presents an overview of the Bretton Woods experience. I analyze the system’s performance relative to earlier international monetary regimesas well as to the subsequent one-and also its origins, operation, problems, and demise. In the survey, I discuss issues deemed important during the life of Bretton Woods and some that speak to the concerns of the present. The survey is limited to the industrial countries-the G10 and especially the G7. I do not examine the role of the International Monetary Fund (IMF), the fundamental organization of Bretton Woods, in the economies and international economic relations of the developing nations. Section 1.1 compares the macro performance of Bretton Woods with the preceding and subsequent monetary regimes. The descriptive statistics on Michael D. Bordo is professor of economics at Rutgers University and a research associate of the National Bureau of Economic Research. For helpful comments and suggestions the author would like to thank Forrest Capie, Max Corden, Barry Eichengreen, Lars Jonung, Charles Kindleberger, Adam Klug, Allan Meltzer, Donald Moggridge, Hugh Rockoff, Anna Schwartz, Leland Yeager, and the NBER conference participants. His thanks for providing data on Japan go to James Lothian and Robert Rasche. Valuable research assistance has been provided by Bernhard Eschweiller and Johan Koenes.

3

4

Michael D. Bordo

nine key macro variables point to one incontrovertible conclusion. Both nominal and real variables exhibited the most stable behavior in the past century under the Bretton Woods system, in its full convertibility phase, 1959-71. While Bretton Woods was relatively stable, it was also very short lived. From the declaration of par values by thirty-two countries on 18 December 1946 to the closing of the gold window on 15 August 1971, it lasted twenty-five years.’ However, most analysts would agree that, until the Western European industrial countries made their currencies convertible on 27 December 1958, the system did not operate as intended. On this calculation, the regime lasted only twelve years. Alternatively, if we date its termination at the end of the Gold Pool and the start of the two-tier system on 15 March 1968, it was in full operation only nine years. This raises questions about why Bretton Woods was statistically so stable and why it was so short lived. (1) Was Bretton Woods successful in producing economic stability because it operated during a period of economic stability, or did the existence of the adjustable peg regime produce economic stability? Alternatively, was its statistical stability an illusion-belied by the presence of continual turmoil in the foreign exchange markets? (2) Why did the system crumble after 1968 and end (so far) irrevocably in August 1971? It is the hope of the conference organizers that answers to these questions and many others can be provided by this and other papers to be presented here. Section 1.2 surveys the origins of Bretton Woods: the perceived problems of the interwar period; the plans for a new international monetary order; and the steps leading to the adoption of the Articles of Agreement. Section 1.3 examines the preconvertibility period from 1946 to 1958, the problems in getting started exemplified by the dollar shortage and the weakness of the IMF, and the transition of the system to convertibility and the gold dollar standard. Section 1.4 analyzes the heyday of Bretton Woods from 1959 to 1967 in the context of the gold dollar standard and its famous three problems: adjustment, liquidity, and confidence. I review both the problems and the many proposals for monetary reform. Section 1.5 considers the emergence of a “de facto” dollar standard in 1968 and its collapse in the face of U.S.-induced inflation. Finally, section 1.6 summarizes the main points of the paper, discusses some lessons learned from the Bretton Woods experience for the design of a fixed exchange rate regime, and raises questions answered by the other papers in the conference volume. 1. The par value system was preserved by the Smithsonian Agreement, 18 December 1971, until its final abandonment on 1 March 1973.

5

The Bretton Woods International Monetary System

1.1 The Performance of Bretton Woods in Comparison to Alternative Monetary Regimes The architects of the Bretton Woods system wanted a set of monetary arrangements that would combine the advantage of the classical gold standard (i.e., exchange rate stability) with the advantage of floating rates (i.e., independence to pursue national full employment policies). They sought to avoid the defects of floating rates (destabilizing speculation and competitive beggarthy-neighbor devaluations) and the defects of the fixed exchange rate gold standard (subordination of national monetary policies to the dictates of external balance and subjection of the economy to the international transmission of the business cycle). As a consequence, they set up an adjustable peg system of fixed parities that could be changed only in the event of a fundamental disequilibrium. The architects derived their views of an ideal international monetary arrangement from their perception of the performance of the pre-World War I classical gold standard and of the sequence of floating rates and gold exchange standard that characterized the interwar period. As background to the historical survey of Bretton Woods, I compare descriptive evidence on the macro performance of the international monetary regime of Bretton Woods with that on the performance of preceding and subsequent regimes. The comparison for the seven largest (non-Communist) industrialized countries (the United States, the United Kingdom, Germany, France, Japan, Canada, and Italy) is based on annual data for Bretton Woods (1946-70), the present regime of floating rates (1974-89), and the two regimes preceding Bretton Woods: the interwar period (1919-39) and the classical gold standard (1881-1913). The Bretton Woods period (1946-70) is divided into two subperiods: the preconvertible phase (1946-58) and the convertible phase (1959-70).* The comparison also relates to the theoretical issues raised by the perennial debate over fixed versus flexible exchange rates. According to the traditional view, adherence to a (commodity-based) fixed exchange rate regime, such as the gold standard, ensured long-run price stability for the world as a whole because the fixed price of gold provided a nominal anchor to the world money supply. By pegging their currencies to gold, individual nations fixed their price levels to that of the world. The disadvantage of fixed rates is that individual nations were exposed to both monetary and real shocks transmitted from the rest of the world via the balance of payments and other channels of transmission (Bordo and Schwartz 1989). Also, the common world price level under the gold standard exhibited secular periods of deflation and inflation reflecting shocks to the demand for and supply of gold (Bordo 1981; Rockoff, 1984). However, a 2. I also examined the period 1946-73, which includes the three years of transition from the Bretton Woods adjustable peg to the present floating regime. The evidence is similar to that of the period 1946-70, so it is not presented here.

6

Michael D. Bordo

well-designed monetary rule could avoid the long-run swings that characterized the price level under the gold standard (Cagan 1984). The advantage of floating exchange rates is to provide insulation from foreign shocks. The disadvantage is the absence of the discipline of the fixed exchange rate rulemonetary authorities could follow inflationary policies. Theoretical developments in recent years have complicated the simple distinction between fixed and floating rates. In the presence of capital mobility, currency substitution, policy reactions, and policy interdependence, floating rates no longer necessarily provide insulation from either real or monetary shocks (Bordo and Schwartz 1989). Moreover, according to recent real business cycle approaches, there may be no relation between the international monetary regime and the transmission of real shocks (Baxter and Stockman 1989). Neveriheless, the comparison between regimes may shed light on these issues, One important caveat is that the historical regimes presented here do not represent clear examples of fixed and floating rate regimes. The interwar period is composed of three regimes: general floating from 1919 to 1925, the gold exchange standard from 1926 to 193 1, and a managed float to 1939.3The Bretton Woods regime cannot be characterized as a fixed exchange rate regime throughout its history: the preconvertibility period was close to the adjustable peg envisioned by its architects; the convertible period was close to a de facto fixed dollar ~ t a n d a r dFinally, .~ although the period since 1973 has been characterized as a floating exchange rate regime, at various times it has experienced varying degrees of management. Table 1 . 1 presents descriptive statistics on nine macro variables for each country, the data for each variable converted to a continuous annual series from 1880 to 1989. The nine variables are the rate of inflation, real per capita growth, money growth, short- and long-term nominal interest rates, shortand long-term real interest rates, and the absolute rates of change of nominal and real exchange rates. The definition of the variable used (e.g., M1 vs. M2) was dictated by the availability of data over the entire period. For each variable and each country, I present two summary statistics: the mean and the standard deviation. For inflation, I also show (in parentheses) the standard deviation of the forecast error based on a univariate regression. For all the countries taken as a group, I show two summary statistics: the grand mean and a simple measure of convergence measured as the mean of the absolute differences between each country’s summary statistic and the grand means of the group of c o u n t r i e ~I. ~comment on the statistical results for each variable. 3. To be more exact, the United States stayed on the gold standard until 1933 and France until 1936. For a detailed comparison of the performances of these three regimes in the interwar period, see Eichengreen (1989a). 4. Within the sample of seven countries, Canada floated from 1950 to 1961. 5. This is a very crude measure of convergence or divergence between the different countries’ summary statistics. Because it is based on the average for the whole period, it suppresses unusual movements within particular subperiods. These will be. discussed in the text.

Table 1.1

Descriptive Statistics of Selected Open Economy Macro Variables, the 6 7 Countries, 1881-1989

Bretton Woods: Total, 1946-70

Bretton Woods: Preconvertible, 19 4 6 5 8

Bretton Woods: Convertible, 1959-70

Floating Exchange, 1974-89

M

SD

M

SD

M

SD

M

SD

A. Inflation PGNPa United States

2.4

2.8

Germany

2.7

1.5 (.7) 1.5 (1.2) 1.8

5.6

3.7

3.5 (2.6) 2.5 (2.8) 6.2 (3.9) 5.1 (5.0) 5.7 (5.0) 3.1 (2.4) 16.0 (8.1)

2.6

United Kingdom

2.6 (3.0) 2.2 (2.3) 4.0 (4.1) 4.1 (4.2) 4.6 (4.6) 3.0 (2.7) 11.5 (7.9)

2.4 (1.7) 6.1 (4.0) 1.3 (.7) 3.2 (2.2) 2.4 (1.6) 3.0 (3.0) 4.6 (2.6)

4.6 2.1 5.6

3.4 3.2 5.5

(1.7)

3.6 (3.5) 1.0 (3) 1.1 (.9) 2.1 (2.6)

9.4 3.3 8.8

France

5.6

Japan

4.5

Canada

2.7

Italy

3.8

Mean Convergence

3.6 .9

4.6 2.0

3.9 1.3

6.0 2.9

3.9 .9

1.8 .6

7.2 2.9

3.3 1.2

B . Real per capita growth’ United States United Kingdom Germany

2.0 2.1 5.0

2.8 1.8 3.3

1.8 2.1 7.3

3.4 2.2 3.9

2.9 2.3 3.6

1.9 1.4 2.6

2.1 1.5 2.2

2.7 4.2 1.9

(continued)

4.2 2.1 5.9

5.5 3.5 3.8

2.6 7.9 12.9

Interwar, 1919-38 M - 1.8 - 1.5

-2.1 2.2 - 1.7

- 1.9 - 1.1

SD 7.6 (8.2) 7.8 (8.2) 4.7 (4.8) 9.1 (9.4) 7.3 (8.5) 6.0 (6.3) 11.7 (10.7)

Gold Standard, 1881-191 3

M .3 .3 .6 -

.o

4.6 .4 .6

SD 3.1 (3.1) 3.1 (3.0) 2.6 (2.6) 5.0 (4.5) 5.5 (5.6) 1.4 (1.3) 3.2 (3.3)

1.o

7.7 1.5

1 .o 1 .o

3.4

.2 1.2 2.6

8.1 4.5 8.5

1.8

5.0 2.4 2.9

-1.1

1.1

1.7

1 .O

Table 1.1

(continued)

Bretton Woods: Total, 1946-70 M

SD

Bretton Woods: Reconvertible, 1946-58 M

SD

Bretton Woods: Convertible, 1959-70 M

SD

France Japan Canada Italy

3.9 8.1 2.5 5.6

2.2 2.7 2.6 3.3

4.6 7.3 1.9 5.2

2.7 2.8 3.2 4.4

3.9 8.9 3.8 5.8

1.3 2.4

Mean Convergence

4.2 1.8

2.7 .4

4.3 2.1

3.2 .6

4.5 1.7

1.8

C . Money growth" United States United Kingdom Germany France Japan Canada Italy

6.3 3.2 12.8 11.5 17.3 6.0 13.3

5.8 3.2 6.0 7.5 15.9 4.0 7.8

6.4 1.7 17.6 14.7 18.2 5.0 15.9

8.3 2.9 5.6 7.2 18.5 3.9 10.5

Mean Convergence

10.I 4.2

7.2 2.8

11.4 6.0

3.4 4.0 4.0 4.2

1.9 2.5 1.5 1.9

2.0 2.3 4.1 3.2

D . Short-term interest rate United States United Kingdom Germany France

Floating Exchange, 197489 M

1.7 3.5 1.6 2.5

.5

7.0 5.5 10.9 8.6 14.6 9.4 12.4

8.1 3.7 .9 1.8 1.1

1.5

SD

Interwar, 1919-38 M

SD

Gold Standard, 1881-1913 M

SD

2.6 2.2

1.3 2.0 .2 .9

7.2 6.1 8.8 4.7

I .5 1.4 2.3 1 .0

4.6 3.8 2.8 4.1

2.2 .5

2.3 .7

1.2 .7

6.8 1.5

1.5 .3

3.7 .8

1.5 2.9 4.7 6.6 2.5 4.3 2.0

8.6 13.5 5.7 8.8 5.7 11.0 13.4

2.4 5.6 4.5 3.4 6.2

.6 .8 1.3 6.4 .5 1.1 3.6

8.6 4.1 10.1 8.5 9.7 4.7 6.2

6.1 2.1 5.7 2.2 5.8 7.4 3.2

5.9 1.7 4.1 3.5 10.8 5.3 3.1

9.8 2.5

3.5 1.4

9.5 2.7

4.6 1.1

2.0 1.7

7.5 2.0

4.6 1.8

5.0 2.0

4.8 5.8 4.0 5.1

1.6 1.6 1.7 1.9

8.9 11.2 5.9 10.3

2.6 2.1 2.4 2.6

3.5 3.0 4.8 3.1

2.0 1.8 1.6 1.4

4.8 2.8 3.2 2.5

.9 .8 .9 .6

1.1

1.9

1.5 1.1

5.5 4.9

Japan Canada Italy

N.A.

Mean Convergence

4.2 .8

1.8 .3

3.5 1.3

1.2 .2

5.1 .5

1.4 .2

8.5 1.9

2.5 .3

2.9 .9

E . Long-term inrerest rate United States United Kingdom Germany France Japan Canada Italy

3.9 5.2 6.3 5.7 7 .O 4.5 6.0

1.3 1.8 .7

.8

3.0 3.9 5.9 5.8

.4 .8 .5 .5

1.1 1.3 .7 1.0

.1

N.A. 3.8 6.3

.8 .4

.1 1.0 .7

10.4 12.1 7.8 10.9 7.1 10.3 13.7

2.1 2.8 1.5 2.4 1.8 2.3 3.3

4.2 4.1 6.9 4.6

1.5 .7

5.0 6.6 6.7 5.7 7.0 5.9 5.7

Mean Convergence

5.5 .9

1.o .5

4.8 I .2

.6 .I

6.1 .6

.9 .3

10.3 1.7

0.3 -.l 2.2 - .9 1.9 - .3

3.9 3.4 2.6 5.2 2.5 4.2

-1.2 -2.4 3.0 -3.3 2.7 .1

4.7 3.3 3.6 6.9 2.8 3.4

2.4 2.3 1.6 1.2

.4 1.1 1.5 1.4 1.1 .7

2.5 1.3 2.5 2.1 1.4 2.5

F . Real short-term interest rateb United States United Kingdom Germany France Japan Canada Italy

6.5 2.9

.8 2.0

6.8 2.2

.8 1.3

N.A.

N.A.

5.9 4.8

.4 1.3

N.A.

N.A.

.5

2.0

5.2 9.2

2.0 3.4

N.A.

N.A.

2.0 .9

2.4

.5

1.3 .4

3.2 .7

.7 .2

.6 .7 1.8 .8

3.8 2.9 3.7 3.2

.3 .2 .2 .3

.5 .4

N.A.

N.A.

N.A. N.A.

N.A.

4.7 5.9

.8 .6

3.5 4.2

.4 .5

2.3 .5

5.1 .9

.9 .3

3.6 .3

.3 .1

2.8 5.1 1.9 2.8 3.5 3.2

3.8 4.2 5.1 1.2 1.4 -.8

6.7 7.1 5.2 14.7 8.8 1.3

4.8 2.9 2.4 2.8 -1.5

2.0 2.3 2.3 6.4 5.5

N.A.

N.A.

N.A. N.A.

Mean Convergence

1 .o

.5

3.6 .6

- .2 2.1

4.1 .9

1.7 .6

1.0 .3

2.0 .5

3.2 .7

2.5 1.9

7.3 2.7

2.3 1.5

3.7 1.8

G. Real Long-term interest rateb United States United Kingdom Germany

.8 1.1 4.3

3.6 2.8 2.8

- .7 - .8 4.3

4.4 2.6 4.4

2.5 3.2 4.3

.7

1.0 1.0

3.9 2.2 4.4

3.8 3.7 .9

4.6 5.4 6.9

6.8 7.1 6.0

3.7 3.0 2.9

2.2 2.5 2.4

(continued)

Table 1.1

(continued)

Bretton Woods: Total, 1946-70

France Japan Canada Italy Mean Convergence

M

SD

.4 1.7 1.3 - .4

4.4 1.3 3.8 12.1

-1.2 N.A. 3.8 6.3

6.2

1.3 1.o

4.4 2.2

.7 2.4 1.8 2.5 15.9 1.6 7.4

6.3 3.8 7.7 37.2 1.9 20.6

H.Nominal exchange rate' United States United Kingdom Germany France Japan Canada Italy

Bretton Woods: Preconvertible, 1946-58 M

SD

Bretton Woods: Convertible, 1959-70

Floating Exchange, 1974-89

SD

M

M

Gold Standard, Interwar, 1919-38

1881-1913

SD

SD

M

SD

M 3.5 N.A. 3.5 4.2

6.5

1.0 1.3 .6 2.2

2.7 2.0 3.6 .5

3.1 4.2 2.8 5.3

1.o N.A. 4.7 5.9

15.1

.4

1.8 1.7 3.0 2.2

2.0 2.9

3.1 1.9

2.7 .7

1.1 .4

2.7 1.1

3.4 1.0

4.8 1.3

6.1 3.6

3.5 .4

2.4

3.6 2.4 4.4 22.0 2.2 14.1

8.3 5.3 11.3 42.6 2.0 27.4

.7 1.4 1.3 1.1 .2

.8 3.9 2.1 3.3 .2 1.9 .2

10.1 9.3 10.7 8.8 3.7 10.9

4.7 8.2 7.8 9.5 2.4 9.0

6.8 3.9 17.8 6.7 2.6 13.6

7.9 9.5 16.9 8.9 3.4 20.1

.2 .2 .3 2.9

.2 .1 .2 4.5

1.4

1.5

.8

.8

.8 .2

'

.8 .6

.o

.4 .5 1.4

.o

4.6 4.0

11.2 10.1

8.1 6.6

16. I 12.6

.8 .4

1.8 1.2

8.9 1.8

6.9 2.3

8.6 4.8

11.1 4.9

.8 .9

1.1 1.3

United Statesd United Kingdom Germany France Japan Canada Italy

1.7 3.5 2.8 4.1 3.0 2.4 8.0

1.o 5.5 5.1 5.6 1.5 2.3 18.7

4.7 3.8 6.2 4.4 2.4 13.1

7.1 7.3 7.7 4.3 2.3 25.2

1.7 2.5 1.9 2.5 2.1 1.2 2.4

1.0 3.5 1.8 2.9 1.2 1.7 1.6

9.4 8.8 9.2 9.6 3.8 8.6

4.3 8.2 7.7 8.9 2.0 7.8

6.5 5.8 8.9 7.8 3.2 13.3

6.9 9.2 6.9 7.2 2.8 16.9

1.7 2.4 4.3 6.6 2.6 2.1

1.5 1.2 5.0 5.6 2.2 1.7

Mean Convergence

3.6 1.4

5.7 3.7

5.8 2.6

9.0 5.4

2.0 .4

2.0 .7

8.2 1.4

6.5 1.9

7.6 2.4

8.3 3.2

3.3 .9

2.9 1.6

Mean Convergence 1. Real exchange rateEC

Sources: See the appendix. Note: For inflation, the standard deviation of the forecast error based on a univariate regression is shown in parentheses. The forecast error is calculated as the standard error of estimate of the fitted equation ln(P,) = a b In (P,- ,), where P, is the price index in year t. 'Mean growth rate calculated as the time coefficient from a regression of the natural logarithm of the variable on a constant and a time trend. Talculated as the nominal interest rate minus the annual rate of change of the CPI. cAhsolute rates of change. Trade-weighted nominal and real exchange rate starting in 1960. Calculated as the nominal exchange rate divided by the ratio of foreign to the U.S. CPI.

+

12

Michael D. Bordo

Injution. The classical gold standard had the lowest rate of inflation, and the interwar period displayed mild deflation. The rate of inflation during the Bretton Woods period was on average, and for every country except Japan, lower than during the subsequent floating exchange rate period. The average rate of inflation in the two Bretton Woods subperiods was virtually the same. However, this comparison conceals the importance of two periods of rapid inflation in the 1940s and 1950s and in the late 1960s (see fig. 1.1).6Thus, the evidence based on country and period averages of very low inflation in the gold standard period and of a lower inflation rate during Bretton Woods than the subsequent floating period is consistent with the traditional view of price behavior under fixed (commodity-based) and flexible exchange rates. In addition, the inflation rates show the highest degree of convergence between countries during the classical gold standard and to a lesser extent during the Bretton Woods convertible subperiod compared to the floating rate period and the mixed interwar regime. This evidence also is consistent with the traditional view of the operation of the classical price specie flow mechanism and commodity arbitrage under fixed rates and insulation and greater monetary independence under floating rates.’ The Bretton Woods convertible subperiod had the most stable inflation rate of any regime judged by both the standard deviation and the forecast error.* By contrast, the preconvertible Bretton Woods period exhibited greater inflation variability than either the gold standard or the recent float. However, most of this difference can be accounted for by the high variability of inflation in Italy during the 1940s and 1950s. The evidence of a high degree of price stability in the convertible phase of Bretton Woods is also consistent with the traditional view that fixed rate (commodity-based) regimes provide a stable nominal anchor; however, the remarkable price stability during this period may also reflect the absence of major shocks. Real per Capita GNI? Generally, the Bretton Woods period exhibited the most rapid growth of any monetary regime, especially the convertible period, and, not surprisingly, the interwar period the lowest (see fig. 1.2). Output variability was also lowest in the convertible subperiod of Bretton Woods, but, because of higher variability in the preconvertible period, the Bretton Woods system as a whole was more variable than the floating period. Both pre-World War I1 regimes exhibit higher variability than their post-World War I1 counterparts. The Bretton Woods regime also exhibited the lowest divergence of output variability between countries of any regime, with the interwar regime the highest. The lower variability of real output during Bretton Woods than during 6. The data sources for fig. 1.1 and all subsequent figures are listed in the data appendix. 7. For similar evidence, see Bordo (1981), Darby, Lothian, et al. (1983), and Darby and Loth-

ian ( 1989). 8. For similar results using the Kalman filter, see Meltzer and Robinson (1989).

The Bretton Woods International Monetary System

13

%

30

-20

I

1 1850

19'1 0

19 0

19 0

1930

1

1940

1950

1

19'10

I

1950

1960

19 0

90

80 70 60

50 40

30

I

I

20 10

0 -10

-20

I

1850

19 0

1 9'1 0

-

France

19 0

1950

---- Italy

1 9LO

-..-.

1950

Japan

Fig. 1.1 Inflation rates, G7 countries, 1880-1989

1960

19 0

14

-20

Michael D. Bordo

'

18$0

19 0

1 9'1 0

19 0

19'30

I

1950

1940

40

I

1960

1990

I E

30

It I'

,I I'

20

10

0

-10 1 ,

( I

@ I ( I

-20

I,

-30

-France ---- Italy

-..-.

Japan

Fig. 1.2 Per capita income growth rates, G7 countries, 1880-1989

19 0

15

The Bretton Woods International Monetary System

other periods may reflect a lower incidence of real shocks, it may reflect a lower incidence of monetary surprises, or it may be the result of countercyclical monetary and fiscal p o l i c i e ~In . ~ turn, the greater convergence of output variability under Bretton Woods may reflect the operation of the fixed exchange rate regime, which created conformity between countries’ business fluctuations (Bordo and Schwartz 1989; Darby and Lothian 1989). Money Growth (M,). Money growth was considerably more rapid across all countries after World War I1 than before. There is not much difference between Bretton Woods and the subsequent floating regime. Within the Bretton Woods regime, money growth was more rapid in the preconvertibility period than in the convertibility period. Money growth rates showed the least divergence between countries during the fixed exchange rate gold standard and the convertible Bretton Woods regime, with the greatest divergence in the preconvertible Bretton Woods period and the interwar period. Of key importance for the viability of the Bretton Woods system, however, is the fact that money growth in the United States, the center of the system, was considerably lower than the average of the G7 countries in both subperiods but increased both absolutely and relatively between the two subperiods (see fig. 1.3 and also fig. 1.29 below). Like inflation and real output variability, money growth variability was lowest in the convertible Bretton Woods period. This, however, was not the case for the preconvertible period, which was the most variable of any regime. Money growth also exhibited the greatest divergence in variability between countries. To the extent that one of the properties of adherence to a fixed exchange rate regime is conformity of monetary growth rates between countries, these results are sympathetic to the view that the Bretton Woods system really began in 1959. Short- and Long-Term Interest Rates. The underlying data can be seen in figures 1.4 and 1.5. As in other nominal series, the degree of convergence of mean short-term interest rates is highest in the convertible Bretton Woods period. Long-term rates are most closely related in the classical gold standard, with the convertible Bretton Woods period not far behind. These findings are similar to these of McKinnon (1988), who views them as evidence of capital market integration under fixed exchange rates. The lack of convergence in the preconvertibility Bretton Woods period reflects the presence of pervasive capital controls. Convergence of nominal interest rates would not be expected under floating exchange rates. Convergence of standard deviations is also highest in the gold standard period, followed by Bretton Woods. Long-term 9 . For evidence that, by using a different detrending filter than the logarithm first-difference used here, real output variability is not greater in the floating period than in the fixed period, see Baxter and Stockman (1989). See also Sheffrin (1988), Bergman and Jonung (1990), and Backus and Kehoe (1992).

Michael D. Bordo

16

30

20

10

0

-10

-20

1850

,z

100

19'1 0

13L0

U.S.

1320

- - - - U.K.

19'30

-..-.

I

1940

Germany

I

1950

1 ,Lo

.I..... Canada

b)

80

60

40

20

0

-20

-40

-

France

----

Italy

-..-.

Japan

Fig. 1.3 Money growth rates, G7 countries, 1880-1989

13'70

1 ,Lo

The Bretton Woods International Monetary System

17

-

18 16

14 12

-

10

-

8 -

6 4 -

-

-

2 0 '

18'90

19 0

19'1 0

19 0

1950

19Lo

19'50

I

1960

1970

19 0

9 5 -

13

11

7 -

-

3 1

1

1850

19 0

19'1 0

-

19 0 France

1950

1 ,Lo

1950

1 ,Lo

---- Japan

Fig. 1.4 Short-term interest rates, G7 countries, 1880-1989

1970

19 0

Michael D. Bordo

18

20 18 16

14 12

10

-

-

-

8 -

6 -

-

4 -

21

1850

19 0

1 9'1 0

19 0

1930

-France ----

Italy

19LO

-..-.

1950

110

Japan

Fig. 1.5 Long-term interest rates, G7 countries, 1880-1989

1990

19 0

19

The Bretton Woods International Monetary System

rates were most stable and least divergent under the classical gold standard, followed by the two Bretton Woods subperiods, with floating exchange rates the least stable. The evidence that nominal interest rates are more stable and convergent between countries under fixed exchange rate (commodity-based) regimes is consistent with the traditional view. Real Short- and Long-Term Interest Rates. The real interest rates are ex post rates calculated using the rate of change of a consumer price index.I0 (For the underlying data, see figs. 1.6 and 1.7.) Unlike the nominal series, the degree of convergence in means between real short-term interest rates is lowest in the floating exchange rate period, next lowest in the Bretton Woods convertible period, and highest in the preconvertible period. For long-term real rates, as in the case of nominal rates, convergence is highest under the gold standard, followed by the Bretton Woods convertible regime. It is lowest under preconvertible Bretton Woods. The real short-term interest rate is most stable across countries during the Bretton Woods convertible period, when it also shows the least amount of divergence in standard deviations. The same holds for real long-term interest rates. The behavior of real interest rates across regimes is consistent with McKinnon’s (1988) explanation. He argued that fixed exchange rates encourage capital market integration by eliminating devaluation risk. This reduces variability in short-term real interest rates. Similarly, real long-term interest rates are stabilized by pooling across markets, which reduces capital market risk. Nominal and Real Exchange Rates.” The lowest mean rates of change of the nominal exchange rate and the least divergence between rates of change occurred during the Bretton Woods convertible and the gold standard periods, with the former exhibiting the lowest degree of divergence. Exchange rates during the preconvertibility Bretton Woods regime changed almost as much as during the floating period. This mainly reflected the major devaluations of 1949 (see fig. 1.8 and table 1.2). Nominal exchange rates were the least variable in the gold standard and convertible Bretton Woods periods and the most variable and most divergent in the Bretton Woods preconvertible period. As with the nominal exchange rate, the lowest mean rate of change in the real exchange rate across countries and the least divergence between countries was in the Bretton Woods convertible period, with the gold standard period next in size of these measures (see fig. 1.9). The highest rate of change was in the floating period. Similarly, the lowest standard deviation across countries and the least divergence between standard deviations was the Bretton Woods convertible period,with the gold standard again next in these rankings. The other regimes were characterized by much greater variability and divergence. 10. Define the real interest rate as r, = i, - A log P,,where i, is the nominal interest rate. and A log P, = log P, - log P,-,,is the percentage change in the consumer price index. 11. For use of this measure, see Grilli and Kaminsky (1991).

Michael D. Bordo

20

'

30

25

20 15

10

5 0

-5 -10

-15 -20

I

b)

1860

19 0

19'1 0

19 0

- - - - U,K

-us.

1 9'30

-..-.

19kO

Germany

1950

.......

1960

1970

19 0

Canada

30

20

10

0

-10

-20

-30

-40

1850

19 0

19'1 0

-

19LO France

19'30

I

1940

1950

1980

1990

---- Japan

Fig. 1.6 Real short-term interest rates, G7 countries, 1880-1989

19 0

21

The Bretton Woods International Monetary System

25

I

20 15 10

5

0 -5 -10

-15

-20

-60

'

1.3'90

1900

19'1 0

I

1920

1950

19Lo

19'50

1 ,Lo

1990

-France ---- Italy - - Japan Fig. 1.7 Real long-term interest rates, G7 countries, 1880-1989

19 0

Michael D. Bordo

22

16

12

8 4

0

140

-

120

-

100 90

---- Germany

.......

Canada

b)

130

110

U.K.

,.. ..:::.: ..

I

-

I

-

::

I

8

I

8 I, I1

-

I,

I1 I,

"I

..

:: :: : : : :

i ;

: :

"I

III

llp

I,,,

:

: :

*:

I.

50 40 30

20 10 0

-

France

- - - - 'ta'y

' ' ' . ' ' _Japan

Fig. 1.8 Absolute change in nominal exchange rates, G7 countries, 1880-1989

Table 1.2

Exchange Rate and Gold Arrangements, the G10 Countries, 1946-1971 ~~

1946'

1948

1949

Belgium

BF 43.83

9122, Devaluation to 50 BF/$

Canada

$1.00

9119, Devaluation to $1.10

1950

FR 119.1

Italy




4125, Official rate set at 360 U/$ by monetary authorities

1958

Float 8/11, Rate set at 420 Fr/$

1/30, Official par value set at 4.20 DM/$

Sept., Rate set at 625 L/$ by monetary authorities

Japan

1957

9/30, Par value suspended

9120, 350 Frl$

1/26, Par valueb suspended

6/20/ DM 9/19, Devaluation introduced at to 4.20 DM/$ 3.33 DM/$

Germany

1953

12127, BF becomes convertible de facto

< France

1951

12/29, Official par value set at 493 .lFrl$ (12127, Convertible de facto) 12127,

Convertible de facto 12/21, Convertible de facto

5/1 1, Official par value set at 360Y/$

Table 1.2

(continued)

1946'

1948

1949

Netherlands

G 2.65

9/2 1, Devaluation to 3.80 G/$

Sweden

7/13, Official rate 3.60 SKr/$ set by monetary authorities

9/20, Devaluation to 5.17 SKr/$

United Kingdom

€0.248

9/18, Devaluation to 0.357 W $

United States

1953

1957

1958 12/27, Convertible de facto

1115, Official par value set at 5.17 SKI-/$

12/27, Convertible de facto 12/27,

Convertible de facto

1961

1962

I964

1967

1968

1969

I970

1971

2/15, BF becomes convertible under Article VIII

Belgium

Canada

Float f

Germany

1951

$3510~.gold I960

France

1950

>

New Fr ( = 100 2/15, Fr becomes old Fr) convertible under Article VIII

316, Revaluation to 4.00 DM/$ 2/15, Convertible under Article VIII

5/2, Official Par value set at $1.08

5/31. Float

8/10,

Devaluation to 5.55 Fr/$

10126, Revaluation to 3.66 DM/

v

519, Float

Italy

3/30, Official par value set at 625 V $

2/15, Convertible

under Article VIII 4/1, Y becomes

Japan

convertible under Article VIII Netherlands

3/7, Revaluation to 3.62 G/$

5/9, Float

2/15, Convertible under Article VIII Sweden

2/ 15, Convertible under Article VIII

United Kingdom

2/15 Convertible

under Article VIII United States

Nov., London Gold Pool established

11/18,

Devaluation to 0.417 f/$ 3/15, Gold Pool

suspended. 3/17, two-tier gold market instituted

8/15, Suspension

of gold convertibility

Source: Various IMF publications. Nore: BF = Belgian franc; Fr = French franc; DM = deutsche mark; L = lira; Y = yen; G = guilders; SKr = Swedish krona. 'Initial parity: price of U.S. $1 .OO. bMultiple exchange rates were in effect from January 1948 to September 1949. From 26 January to 16 October, a rate of 214.4 Fr/U.S.$ applied to all foreign exchange actions in nonconvertible currencies and to selected imports paid in convertible currencies. For all other transactions, the effective rate was the average of the 214.4 rate and the free rate. On 16 October 1948, the average rate was made applicable to all transactions except nontrade transactions in convertible currencies. I,

y”* =

Y[TJ7FY

+

7F:>1,

because this satisfies equality of marginal utilities, which implies IT.& = A?* = x and the analogous condition for good I: This implies that a positive taste shock for one of the goods in the home country (a full in ax,e.g., which raises home consumption of good X ) is transmitted negatively to foreign consumption of that good. If utility were nonseparable, foreign consumption of the good that is not subject to the shock rises or falls depending on the sign of the cross-derivative in utility, which could reinforce or mitigate the negative international transmission in consumption. n*xd* such that xd

6.4

+

Production

The international transmission of technology disturbances in a model with production has received less attention except within one-sector models or at the microeconomics level for particular industries. l k o recent papers, Backus, Kehoe, and Kydland (in press) and Baxter and Crucini (in press), address transmission issues (among other topics) in two-country, one-sector, real-business-cycle models. In these models, nation-specific shocks to technology have some persistence, so a positive technology shock in the home country (alone) raises both current output and the expected marginal product of capital. If factors are sufficiently mobile internationally, factors move from the foreign to the home country, reducing foreign output and creating negative international transmission in production. If people have pooled risk in asset markets to a sufficient degree (as in the models cited), the shock is transmitted positively to foreign consumption: domestic and foreign consumption move together. The negative transmission to foreign production and positive transmission to foreign consumption make it difficult for these models to explain the observation that the international correlation of consumption is positive but smaller than that of production. In fact, the cross-country correlation of industrial production was higher in the Bretton Woods period than after 1973 (see Baxter and Stockman 1989). One explanation is that there were (and are) seriously incomplete international financial markets or that, because of information costs, people did (and do) not make use of available financial markets to diversify internationally. Other explanations include nontraded goods

326

Alan C. Stockman

(Stockman and Dellas 1989)’ and preference shocks that changed consumption independently across countries (Stockman and Tesar 1991). Productivity shocks are transmitted internationally even if factors are internationally immobile. There are three main channels of transmission. First, if foreign and domestic goods are perfect substitutes in consumption as in the one-sector models cited above, then increased output in the home country reduces the real interest rate. The fall in the real interest rate can reduce foreign output temporarily through intertemporal substitution in labor supply and in the utilization rate of capital. This interest-rate channel creates negative transmission of a home productivity shock to foreign output. This channel of transmission also operates, although with less strength, when home and foreign goods are imperfect substitutes. A second channel of transmission also occurs through demand. Suppose that leisure and consumption are substitutes in utility in the sense of a negative cross-derivative in the utility function. Then a home productivity disturbance in a risk-pooled world raises foreign consumption. Since this reduces the foreign marginal utility of leisure, it raises foreign employment and output. This tends to create positive international transmission to foreign output. A third channel of transmission occurs through demand for capital inputs. Here temporary and permanent productivity shocks can be transmitted differently to foreign countries. A temporary productivity shock in the home country creates a trade surplus in the home country as domestic consumption and foreign consumption rise. With repeated shocks, consumption tends to be more highly correlated across countries than is output. On the other hand, a permanent or highly persistent productivity shock in the home country can cause a trade deficit because the shock raises the expected marginal product of investment and imports of (or to be used as) capital. If equilibrium investment in the home country rises more than output, the home country can have a trade deficit in the short run. If the productivity shock is persistent but not permanent, then it dies away as time passes, so eventually the increased home output (from higher productivity and a higher capital stock) exceeds the increase in investment. Then the trade deficit can turn to a trade surplus. When productivity shocks are positively correlated across countries, home and foreign output can move together even without any international transmission of disturbances. This creates the difficult empirical issue of distinguishing the transmission of shocks from correlated disturbances in different countries. Evidence from Costello (1991) and Waldmann (1990) indicates that the cross-country correlation of Solow residuals, intended to measure technology shocks, is lower than the cross-country correlation of outputs. If this were the 7. Nonseparability in utility between consumption of nontraded and traded goods can create less than perfect correlation in consumption of traded goods even with complete markets because a change in production and consumption of the home nontraded good alters equilibrium home consumption of the traded good, tending to move home and foreign consumption of traded goods in opposite directions.

327

International Transmission under Bretton Woods

only source of exogenous disturbances, we might attribute the excess correlation of output (over productivity shocks) to transmission.*

6.5 International ’kansmission in the Presence of Nontraded Goods Nontraded goods can play an important role in international transmis~ion.~ Consider a productivity shock in the home country nontraded-goods industry. This is likely to raise home consumption of nontraded goods, and, because traded and nontraded goods are complements in consumption (see above), this raises home demand for traded goods. In equilibrium, home consumption of traded goods rises-partly through increased foreign output of traded goods and partly through decreased foreign consumption of traded goods. (Home output of traded goods might increase because of the increase in demand, or it might decrease because of the increase in factor productivity in the nontraded-goods sector, leading factors to migrate to that sector.) So the productivity disturbance in the home nontraded sector can be transmitted positively to foreign production and negatively to foreign consumption (see Stockman and Tesar 1991). Similar results follow from preference shocks. Consider a preference shock in the home country that temporarily raises the marginal utility of consumption of all goods. This causes a home-country trade deficit, raises tradedgoods output in both countries, and raises nontraded-goods output in the home country (because of complementarity between traded and nontraded goods in consumption). Output of nontraded goods in the foreign country falls both because foreign factors move from the nontraded- to the traded-goods industry in response to the higher relative price of traded goods (due to the taste shock) and because foreign consumption of traded goods falls and (through complementarity)reduces foreign demand for nontraded goods. So the shock is transmitted positively to foreign output of traded goods, negatively to foreign output of nontraded goods, and negatively to foreign consumption. The taste shock also raises the relative price of the domestic consumption bundle, which includes home nontraded goods, so it causes real appreciation in the home country that appears under a pegged exchange-rate system as a rise in the home price level relative to the foreign price level. The importance of nontraded goods for international transmission is enhanced when governments “create” nontraded goods out of traded goods through restrictions on international trade; overall, trade restrictions were somewhat higher under Bretton Woods (particularly before the Kennedy Round tariff reductions) than they are today. 8. This runs into problems associated with differences across countries in the timing of the effects of productivity shocks on output. The response to a common shock may occur more rapidly in one country than another despite the absence of “transmission” across countries. 9. For an entirely different channel of international transmission of monetary disturbances in the presence of nontraded goods than that summarized below, see Miller and Todd (1991).

328

Alan C. Stockman

6.6 International 'lkansmission with Sticky Prices Sticky prices are widely regarded as an important feature of macroeconomic data. Svensson (1986) added short-run price stickiness to a model similar to the one discussed above, using it to investigate the international transmission of fiscal policy in Svensson (1987) and the international transmission of monetary disturbances under floating exchange rates in Svensson and van Wijnbergen (1989). To discuss international transmission under Bretton Woods, I will examine the effects of monetary disturbances with fixed exchange rates in Svensson's model. The setup of the model remains mostly the same: the representative household in the home country maximizes (1) subject to the sequence of constraints (2), (4),and (3, and there is an analogous problem for the foreign representative household. Now, however, firms set nominal prices one period in advance as in Svensson (1986) and Svensson and van Wijnbergen (1989). Output in each country per capita is determined by demand subject to the capacity constraints

The capacity constraints imply that, if demand exceeds capacity, there is rationing:

4 Ix,

and yf

Iy,.

The foreign money supply adjusts to keep the exchange rate pegged at e, = e V t . Define the exogenous part of the state vector by s, = ( X I , Y,,

0,).

Assume that the vectors is random and independently drawn over time. An equilibrium is a set of functions of the state vector s and the predetermined variables-nominal prices and the levels of the beginning-of-period money stocks of each country-used to describe output and each country's consumption of each good, nominal goods prices set for the following period, the foreign money-growth rate, asset holdings, and asset prices, such that each representative consumer maximizes utility subject to its budget and cashin-advance constraints, firms maximize market value by choice of nominal prices, and markets clear with a fixed exchange rate. Assume that firms operate in the monopolistically competitive environment described in Svensson (1986), which builds on Dixit and Stiglitz (1977), and assume that a stationary rational expectations equilibrium exists. Although many of the characteristics of the equilibrium of this model are similar to those discussed by Svensson in his model with flexible exchange

329

International Transmission under Bretton Woods

rates, international transmission under the pegged-rate system considered here differs from the discussion by Svensson and van Wijnbergen for a flexible exchange-rate system. Notice that the price-setting problem facing firms is a stationary problem. Svensson’s proof that firms set nominal prices proportionally to the current money supply applies to this model, so at time t - 1 firms set period-t prices by the formulas

where k and k* are constants. The possible stationary equilibria of the model divide into three parameter regions for each country, depending on whether the cash-in-advance constraint binds, the output-capacity (full-employment) constraint binds, or neither Constraint binds (so there are nine regions in all because there are two countries). Obviously, small changes in monetary policy have no effects on output if the full-capacity constraint is binding (except at the border of that region). But monetary policy has real effects in the other regions. First suppose that the cash-in-advance constraint for spending on the domestic good but nor the domestic full-capacity constraint binds. Then home output is 2x =

M

= wfk. p, Similarly, if the cash-in-advance constraint for spending on the foreign good but not the foreign full-capacity constraint binds, then foreign output is

‘

Now consider the equilibrium region in which neither the cash-in-advance constraint nor the full-capacity constraint binds for either good. In the interior of this region, home output and foreign output per capita solve the equations U,(x,, yJ

=

hf - A

-

P‘

- -

Of

and

U,(x,, y,) =

X*

P,*

A*

= 0;

’

where A and A* are positive constants that come from the first-order conditions for choices of end-of-period money holdings. A and A* are constants because of the assumption that shocks are i.i.d. (so that expected future variables are independent of the current state). Now consider the international transmission of monetary policy in this fixed-exchange-rate model with sticky prices. First suppose that, for each good, the cash-in-advance constraint binds but that the full-capacity constraint does not. In this region, a permanent rise in the home money supply (a positive i.i.d. shock to the home country’s money-growth rate) raises home output. It has no effect on foreign output unless it affects the foreign money sup-

330

Alan C. Stockman

ply. Since the exchange rate in period-t asset markets (after the close of period-t product markets is e,

=

MtA*k 7 NtAk*

an increase in the home money supply leads to an equal proportional increase in the foreign money supply to keep the exchange rate pegged: - M

A*k

o,*= o,e, L Nt- ,Ak* '

So in this region-where for each good the cash-in-advance constraint binds but the full-capacity constraint does not-an increase in the home money supply raises real output in both countries: there is positive international transmission to foreign output. (Under a flexible-exchange-rate system, in contrast, there is no transmission to foreign output in this region if the home monetary expansion leaves the foreign money supply unaffected.) Now consider the region in which neither the cash-in-advance constraint nor the liquidity constraint binds for either good. Svensson and van Wijnbergen demonstrate that, under a floating-exchange-rate system, a home monetary expansion has two effects, with opposite signs, on foreign output. A home monetary expansion raises output and consumption of the home good, while, given the foreign rate of money growth, the effect on foreign output depends on the sign of the cross-derivative in the utility function, U , 2 .Foreign output rises, remains the same, or falls depending on whether U , , is positive, zero, or negative.lO This cross-derivative is positive (or negative) if and only if the intertemporal elasticity of substitution in aggregate consumption is greater (or less) than the intratemporal elasticity of substitution between consumption of home and foreign goods. So under flexible exchange rates there is positive international transmission of output in response to a permanent monetary disturbance if the elasticity of intertemporal substitution exceeds the elasticity of intratemporal substitution between home and foreign goods. Otherwise, a permanent increase in the home money supply is transmitted negatively to foreign output. Under pegged exchange rates, there is a third effect. The foreign money supply rises proportionally to the increase in the home money supply, implying equal percentage changes in the marginal utility of consumption of home 10. This follows directly from the expressions for home and foreign output in this region. Suppose, e.g., that output and consumption of the home good increases. If U,,> 0, this raises the marginal utility of consumption of the foreign good, so we require an increase in output and consumption of the foreign good to bring its marginal utility back to the unchanged equilibrium level (for given foreign money growth).

331

International Transmission under Bretton Woods

output and the marginal utility of consumption of foreign output. This implies that foreign output rises regardless of the relative magnitudes of the intertemporal and intratemporal elasticities of substitution in consumption:

and

While this model predicts short-run real effects of monetary disturbances, it does not predict differences in inflation across countries under pegged exchange rates. Figures 6.1-6.3 below show that, despite a period of stable nominal exchange rates from 1962 until the end of the decade, there were sizable differences in inflation across countries (measured using consumer price indexes). Even during the period before the devaluation of sterling in 1967, the average inflation rate was higher in Japan, Italy, and the Netherlands than in Canada, Germany, France, the United Kingdom, and Belgium and lower in the United States. From 1962 to 1967, inflation in the United States averaged 1.9 percent per year, while inflation in Japan, Italy, and the Netherlands was above 4 percent per year. In the model discussed here, nominal prices are predetermined each period, so there is no short-run effect of the home monetary expansion on nominal prices in this model. But prices change in the periods following a monetary disturbance. Since nominal prices for date t 1 are set at date t in proportion to the period-t money supply, an increase in the home money supply at date t could in principle affect the relative price of home and foreign goods starting at date t 1. The assumption of i.i.d. shocks, however, prevents this from occurring: the foreign money supply and the foreign nominal price of the foreign good rise in proportion to the change in the home money supply, to keep the exchange rate pegged. So this model produces real effects of monetary disturbances through sticky prices, but it cannot (in this form) explain why short-run inflation rates may differ across countries with pegged exchange rates. To do that probably requires a model with real costs of short-run arbitrage, as in Dumas (19871, where arbitrage occurs with a delay because of the time required to ship goods from one country to another. For some time following an unexpected disturbance (but not permanently), nominal prices in the home country can differ from nominal prices in the foreign country. Although Dumas’s model is real, it is reasonable to conjecture that a monetary version of his model would imply that monetary shocks in a country can have short-term effects on domestic inflation even under a pegged exchange-rate system. Further theoretical study of this issue would be worthwhile because the next section presents evidence that suggests that monetary policy was able

+

+

332

Alan C. Stockman

to affect inflation in the short run under the Bretton Woods system despite pegged exchange rates. 6.7 Evidence of the International Pansmission of Inflation The Bretton Woods system consisted of an initial phase in which most currencies were nonconvertible, international trade in goods and financial assets was severely restricted, and international trade that did occur was mainly bilateral barter. Because of the peculiar difficulties arising from inconvertibility and extensive controls as well as an absence of reliable data, the analysis below concentrates on the second period of Bretton Woods, after 1958, in which currencies were convertible and for which more reliable data become available. For much of the analysis, the sample period is restricted further. Figure 6.1 shows nominal exchange rates against the U.S. dollar for a set of eight countries under this second phase of the Bretton Woods system. There is no long period of fixed exchange rates: the longest such period is from 1962:3, when Canada pegged to the U.S. dollar (or 1961:2, after Germany’s revaluation), through 1967:3, before the devaluation of the pound sterling. Figure 6.2 shows inflation under Bretton Woods in the same nine countries (measured with consumer price indexes). Even for subperiods during which nominal exchange rates were pegged, inflation differentials across countries were large. Average inflation during 1963-67, shown in figure 6.3, ranged from 1.9 percent per year in the United States to rates above 4 percent in the Netherlands, Italy, and Japan. Figure 6.4 shows real exchange rates, defined as the relative price of the U.S. CPI bundle of goods in terms of the home country’s CPI bundle (i.e., as e*P[United States]/P[n] for nation n). The most striking aspect of the figure is that these relative prices change dramatically when nominal exchange rates change with currency devaluations or revaluations. This fact is well known and often cited as evidence that nominal prices are sticky, so that devaluations cause changes in relative prices. A further striking observation is that there appears to be little tendency for the relative price changes accompanying devaluations to reverse themselves in subsequent months or years. l 1 Because the Bretton Woods experience was brief after the restoration of currency convertibility in 1958, the evidence on international transmission available from the Bretton Woods period is quite limited. To draw inferences with as much evidence as possible from these limited data, this paper makes use of the combined cross-sectionavtime-series feature of the data by testing and imposing constraints on coefficients across equations for various coun11. These revelations and devaluations are, in one sense, the biggest economic shocks to occur under the Bretton Woods system. They had major effects on real exchange rates. By far the largest changes in real exchange rates during this period occurred with these changes in official pegs. Yet they appear to have had little effect on international trade, output, employment, and other variables (see Baxter and Stockman 1989).

333

International Transmission under Bretton Woods

1.15

U.K.

1.10

1.05

France 1.oo

Canada 0.95

Italy Belgium

0.90

Netherlands 0.85

Japan Germany

0.80

1

1

1958

1

.

1960

1

1

1962

1

1

1964

1

1

1966

1

1

1968

1

1

1970

1

1972

Fig. 6.1 Exchange rates under Bretton Woods, 1958-72

0.090

-

0.060

-

0.030

-

0.000

-

z.

1958

1960

1962

1964

1966

1968

1970

1972

Fig. 6.2 Inflation under Bretton Woods, 1958-72 (monthly series on twelvemonth inflation rates)

Alan C. Stockman

334 %

m

US.A.

Can

Ger

Fra

U.K.

Be1

Ita

Neth

Jap

Fig. 6.3 Average inflation, 1962-67

1.6 1.5

-

1,4

..._

: ; $..,

Japan I,, :,*.q ...... >' :-. .,$

i.,?.:

Netherlands

0 . 8 ' .

1

1958

1

1

1960

'\, . ..

I

1

1962

1

I

1964

I

I

1966

I

I

1968

I

I

1970

I

1972

Fig. 6.4 Real exchange rates under Bretton Woods, 1958-72 (US.CPI divided by national CPI, adjusted by exchange rate)

335

International Transmission under Bretton Woods

tries. Furthermore, the discussion below focuses on international differences in inflation across countries rather than on the level of inflation in each country. By examining differences between pairs of countries, the analysis can abstract from issues concerning worldwide price changes that affected countries equally and focus on the limits to international transmission of inflation during this period. At first glance, it appears that there is no relation between differences in inflation across countries and differences in money-supply growth rates across countries (measured by any monetary aggregates). Simple correlations are all close to zero. The average correlation between contemporaneous differences of quarterly growth rates of M1 and CPIs-across all pairs of countries in a sample including Canada, France, Germany, Italy, the United Kingdom, and the United States for periods during which the exchange rate was fixed-is negative. Correlations close to zero also appear at most leads and lags, except when one of the countries in the pair is the United States. Similarly, bivariate Granger-causality tests show no relation between international differences in inflation and money-growth rates. For the sample of countries including Canada, France, Germany, Italy, the United Kingdom, and the United States, almost every pairing shows no Granger causality in either direction. These results are consistent with the monetary approach to the balance of payments if differences in inflation across countries reflect measurement error, in which case they are unrelated to changes in nominal money demand and, as a result, nominal money supply. They are not clearly supportive of the monetary approach if international differences in inflation result from changes in the relative prices of nontraded goods, in which case higher-inflation countries might be expected to have greater growth rates of nominal money demand and so nominal money supplies. But the implications of the data change dramatically when other relevant variables are included in the analysis: there is evidence that changes in the money supply in one country led to subsequent changes in that country's inflation rate (given foreign inflation) in the short run. That is, there was only partial international transmission of foreign inflation to the domestic country in the short run. The short sample of data is not able to provide strong evidence about the long run; the evidence reported here is not inconsistent with the idea that the effects of domestic money growth on domestic inflation vanished in the long run so that the international transmission of foreign inflation to each country was complete in the long run. On the other hand, the data for the Bretton Woods period do not provide strong evidence for or against any hypothesis about the long run. Because the time series are short but there are data for several nations, the results reported below are estimated jointly across country pairs, allowing disturbances to be contemporaneously correlated (although uncorrelated at all leads and lags), and impose restrictions that certain coefficients are identical

336

Alan C. Stockman

across equations. (The coefficient restrictions were also tested as described below.) Specifically, let S(U, b, X) =

X(U)

- x(b)

be an operator that takes the difference in the variable x between countries a and b. For example, G[France, Germany, d log(p)] = d log(p)France- d log(p)ammy. Denote the price level measured by the CPI (other measures give results similar to those reported below) by p , the narrow money supply by M 1, the three-month nominal interest rate by i , and real GDP by y. Consider joint estimation of the equations

1

1

6 nation, US, d log@) (6)

=

1

1

a(L)6 nation, US, d log(p)

I

+ p(L)8 nation, US, d log(M1) + T(L, nation)6(nation, US, i)

[

+ q(L,nation) 6

[

1+

nation, US, d log(y)

&(nation)

for nation = Canada, France, Germany, Italy, and the United Kingdom. All variables are quarterly and expressed as deviations from means. They are seasonally adjusted at the source, although the equation also included seasonal dummies. Begin by imposing the restriction that all coefficients are identical across nations. A test of this restriction for each set of coefficients (one set at a time) showed that some of these restrictions were rejected. Consider, for example, the hypothesis that T(L, nation) and q(L,nation) are the same for all nations except Canada; that is, T(L, nation) = T(L) and q(L, nation) = q(L)for all nations except Canada (where T[L] and q[L]do not depend on the nation). Wald tests of these hypotheses generate chi-square statistics with degrees of freedom equal to the number of restrictions. Almost all these tests led to rejections at p-values smaller than .OW1 . I 2 As a result, these coefficients are left unconstrained in the results reported below. (Those estimates show clearly that T[L]and q[L] differ across nations.) On the other hand, one cannot reject (at any reasonable significance level) the hypothesis that the coefficients a(L)and p ( L ) are independent of the nation, with one exception: a(4) for Canada is significantly different than a(4) for other countries. So, with this exception, the estimates reported below impose the restrictions that the vectors a and p are independent of the nation. They also assume that V(E) = Z is a symmetric positive definite matrix and that V(u,,u,-J = 0 for all t and s. While a(0) = 0, the parameters p(O), r(O), and q(0) were estimated along with the other parameters. Because standard models imply that the differentials in money-growth rates are endogenous and 12. The Canadian-French case was an exception: thep-value was .18. This might suggest that the coefficients for Canada and France could be constrained to be identical, although the estimates reported here do not impose this constraint.

337

International Transmission under Bretton Woods

could be correlated with the disturbances, the estimates reported below were obtained using instrumental variables. The instruments were-in addition to all lagged variables in the equation-current and lagged values of the U.S. federal funds rate and the growth rate of the U.S. monetary base. The lag length for each variable is four quarters (experiments indicated that longer lags do not affect the results). The first results, for the period 196O:l-1971:2, appear in table 6.1. The actual period differs across countries because of data availability (see the data appendix). The estimates show strong effects of lagged differentials in money growth on the inflation differential.I3 The sum of the coefficients on the money-growth differential is .28, indicating (subject to the usual caveats on this interpretation) that a sustained 1 percent rise in domestic money growth relative to foreign money growth leads to a .28 percent rise in the inflation differential under pegged exchange rates. The effect of the contemporaneous money-growth differential is small; most of the effect is associated with lagged money-growth differentials.l4 Estimated coefficients on interest-rate differentials and income-growth differentials vary significantly from country to country. Wald tests of the null hypothesis that the (lagged or current and lagged) money-differential coefficients are jointly zero are rejected at more than the .0001 level. There is little evidence of autocorrelation of residuals, as the table shows, and tests that the disturbances are normally distributed fail to reject that hypothesis at the usual significance levels. The results are nearly identical when the equations are estimated for a shorter sample period that excludes Canadian data before 1962:3, when the Canadian dollar floated against the U.S. dollar, and after 1970:1, when it floated again; German data starting in 1969:4, when the deutsche mark was revalued; and U.K. data starting in 1967:4, when the pound was devalued. Again, there is a strong connection between inflation differentials and lagged money-growth differentials. The natural interpretation of these results is that the governments of these countries had some control over money growth despite pegging their exchange rates: central banks were able to and sometimes did “sterilize” changes in international reserves to affect domestic money growth. l5 Moreover, these differential rates of growth of money appear to have affected inflation differentials. One cannot, however, rule out other (reverse-causality) interpretations 13. When the equation is reestimated allowing for eight lags of money-growth differentials, the chi-square statistic for the hypothesis that the coefficients on lags 5-8 are jointly zero is only 2.08 (p-value = .72). And all four of these coefficients were very close to zero individually. So only the four lags were included in subsequent estimates. 14. The equation also includes nominal-interest differentials and GDP-growth differentials. Full results were reported in the first draft of this paper and are available on request from the author. 15. Iwami (1991) discusses monetary policies under Bretton Woods and compares that system to the gold standard; he argues that U.S. policy under Bretton Woods did not follow the analogous “rules of the game” for a gold standard. Iwami argues that U.S. neglect of the rules of the game allowed other countries to pursue expansionary policies and that this system was viable for the short run but not the long run, so it eventually broke down after its credibility had been strained.

338 Table 6.1

Alan C. Stockman Effects of Money-GrowthDifferentialson Inflation Differentials SE

?-Stat.

,081

,019 ,016 .017 ,018 ,020 ,073 ,068 ,066 .071

,610

,144

1.03 6.15 3.24 3.36 1.90 1.35 - .42 1.81 1.14 4.22

Coeff.

p-Value

~

.020 ,098 ,056 ,062 ,039 ,098 ,029 .120

Canada

France

Germany

.64

.58

.01

.oo .08 .20 .67 .09 .27

.oo

Italy

~

R2

.32

.oo

UK ~

.79

.74

.68

Autocorrelations of Residuals Canada

France

Germany

Italy

UK ~

1st 2d 3d 4th SE p-value'

.I0 .16 .04 .03 .I6 .82

- .03 - .31 .15 .04 .19 .52

.I1 .28 .06 - .06 .I1 .55

- .23 - .01 .18 - .07 .17 .38

.I3 -.18 - .28 -.I1 .19 SO

Nore: These are joint instrumental variables estimates of eqq. (6): the dependent variable is &[nation, US, d log(p)]. The equation also includes current and lagged values of interest-rate differentials and differentials of GDP growth rates; the coefficients of these variables were allowed to differ across countries. When money-growth differentials are excluded from the equations, the R2 vector is .35 for Canada, .54 for France, .76 for Germany, .72 for Italy, and .54 for the United Kingdom. 'p-value refers to the Box-Pierce Q-statistic for the first four autocorrelations.

of the results. For example, it is possible that differentials in inflation were caused by changes in relative prices of nontraded goods (which show up in retail prices of traded goods because of nontraded components of value added such as retailing and transportation). To explain the correlation between inflation differentials and lagged money-growth differentials, the real disturbances would have to raise the monetary aggregates through their effects on demands for money and credit before raising the prices of nontraded goods (raising the international inflation differential). There is very little difference between the estimates reported in table 4.1 and ordinary least squares (seemingly unrelated regression) estimates of the equations. Tests of exogeneity of contemporaneous money-growth differentials, interest differentials, and income-growth differentials (using Hausman's specification test) fail to reject exogeneity at any reasonable significance level.

339

International Transmission under Bretton Woods

This provides some evidence that the relation between money-growth differentials and inflation differentials reflects some degree of monetary-policy autonomy under the Bretton Woods system rather than feedback from inflation differentials (induced by real shocks) to money growth differentials as predicted by the monetary approach to the balance of payments. Figure 6.5 plots the residuals from these equations, and figure 6.6 plots the actual and fitted values. Although the period includes dates of revaluations and devaluations, there is no visual indication of unusual behavior of residuals around-before or after-those dates. This result parallels the finding in Baxter and Stockman (1989) that changes in the exchange-rate system, and in the variability of real exchange rates, seem to be unconnected with macroeconomic or intemational-trade quantity variables. In this case, the major changes in real exchange rates that occur at the time of finite revaluations seem to have little effect on inflation differentials or the relation between inflation differentials and money-growth differentials. Experimentation with the data revealed that the connection between moneygrowth and inflation differentials varies with the overall level of nominal interest rates. Table 6.2 presents results from an equation that adds the contemporaneous level of the U.S. short-term interest rate (three-month Treasury-bill rate) to the equation, with a coefficient that differs across countries. The set of instrumental variables remains unchanged from previous tables: the only contemporaneous variables included among instruments were the growth rate of the U.S. monetary base and the U.S. Federal funds rate. The contemporaneous level of interest rates has a statistically significant effect on the interest differential between the United States and France and between the United States and Germany (although in opposite directions in the two cases). A Wald test that the coefficients of the U.S. nominal interest rate are jointly zero leads to rejection of that hypothesis at the ,0001 level. Holding the level of the nominal interest rate fixed strengthens the effects of lagged money-growth differentials on inflation differentials. The sum of coefficients on moneygrowth differentials rises from .28 to .48,and the sum on lagged differentials rises from .26 to .44. Table 6.3 presents estimates of a similar set of equations for the inflation differential between Germany and other nations in the sample. France is excluded from this sample because chi-square tests indicated that the French coefficientsdiffered from the coefficientsof other countries. The U.S. nominal interest rate is included as in table 6.2. The results continue to show a positive effect of lagged money-growth differentials on the inflation differential and virtually no contemporaneous effect. The sum of the money coefficient is .26. Similar results appear when other countries are used for comparison. As before, there is little evidence of misbehavior of the residuals of the equations in the form of autocorrelation or departures from normality. Nor do the residuals behave any differently around periods of devaluations. These results suggest that, when central banks engaged in sterilization pol-

Alan C. Stockman

340

Conada minus U.5.A

France minus U.S.A 10

0.8

I

-I 0 60

61

62

63

64

65

66

67

68

69

70

71

60

Germany minus U S.A

,I , 61

62

, 63

, , , , , , , 64

65

66

67

68

69

70

I

71

Italy minus U.S.A I

I

08 0 6

04 02 0 -0 2

-0 6 -0 8

, ,

-0.54 60

61

62

,

63

, 64

, ,

65

66

I

67

1, 68

, ,

69

70

, 71

-10

U K minus U S A 08.

-104 60

, 61

, , , , , , , , , ,

62

Fig. 6.5

63

64

65

66

67

68

69

70

71

Plots of residuals from table 6.1

60

, ~,

61

62

63

, , , , , ,

64

65

66

67

68

69

I

,

70

71

International Transmission under Bretton Woods

341

France minus U.S.A

Canada minus U.S.A. 2.5

’ 21 j

6

h

1

2c

actual

1.5

actual

fitted

10

0.5

0

-0 5

-1

6 60

, , , , , , , , , , ,

61

62

63

64

65

66

67

68

69

70

71

-1

0 61

62

63

65

66

67

60

69

70

71

65

66

67

68

69

70

71

64

Italy minus U S.A

Germany minus U.S A.

1 actual

-1 2

, , , , , , , , , , ,

-1 6

60

61

62

63

64

65

66

65

66

67

68

69

70

71

61

62

63

64

U.K. minus U S.A

61

62

63

64

I

I

67

I

68

I

1

69

70

I

71

Fig. 6.6 Inflation differentials: actual and fitted values (actual inflation differentials and fitted values of equations from table 6.1)

342

Alan C. Stockman Inflation Differentials and Money-Growth Differentials with the Level of the U.S. Interest Rate Included

Table 6.2

6d log (Ml), 6d log (MI),-, 6d log M log 1% (M1),-4 & f l o g( P L , 6d log (P),+ ad log (P),-~ , 6d log ( J J ) , + ~ others 6d log ( P ) , + ~Canada ,

R2

Coeff.

SE

r-Stat.

p-Value

,040

,021 ,019

,019 ,021 ,021 ,070 ,070 ,065 ,071

1.89 6.12 3.88 4.03 2.70 .83 - .65 1.45 .24

.07

,117 ,076 ,088 ,057

,151

4.54

,059

- ,046 ,094 ,017 ,688

.oo

.oo .oo

.01 .41 .52 .I6 .81

.oo

Canada

France

Germany

Italy

UK

.67

.60

.88

.72

.68

Nore: These are joint instrumental variables estimates of eqq. (6) as in table 6.1, except the U.S.

three-month Treasury-bill rate is included in each equation with a coefficient that can differ across equations.

icies under the Bretton Woods system, they were able to affect not only the domestic money supply in the short run but also the domestic inflation rate in the short run. This sterilization limited the extent to which changes in foreign inflation were transmitted in the short run to the country. (Because the sample is short, the estimates cannot provide much evidence about long-run effects of sterilization.) The question then arises as to whether these short-run limits on the international transmission of inflation spilled over to real variables. First, notice that the estimates given above imply that there were predictable differences in inflation across countries under Bretton Woods, related to prior differences in money-growth rates. But these predictable changes in inflation differentials appear not to have been fully reflected in changes in nominal-interest differentials. Instead, they occurred mainly as changes in ex post real interest-rate differentials. Because these interest differentials were predictable, it may be reasonable to interpret them as ex ante real interest-rate differentials. (Such ex ante real interest differentials are not necessarily ruled out by arbitrage; they are consistent with expected changes in relative prices of the CPI bundles of goods in different countries.) Table 6.4 shows estimates of a set of nominal-interest-differential equations specified in accordance with the earlier equations for inflation differentials. There is some evidence of a positive effect of lagged money-growth differentials on nominal-interest differentials. (The hypothesis that the coefficients on the money-growth differentials are jointly zero is rejected at the .OW1level.) But the positive effect of lagged money differentials is small. Money growth rates are quarterly growth rates, while the nominal interest differential is in

343

International Transmission under Bretton Woods Germany as the Comparison Country

Table 6.3

Coeff.

- ,007 ,125 ,029 .075 ,035 .loo - ,031 .067 - .272 ,059

SE

t-Stat.

p-Value

,023 ,024 ,021 ,034 ,026 .089

- .30 5.21

.78 .01 .35

,085

.082 .085

,141

1.06

2.19 1.31 1.12 .36 .82 -3.17 .42

.09 .26 .32 .73 .46 .03 .68 ~

Canada

Italy

UK

us

.75

.37

.64

.89

R2

Note: These are joint instrumental variables estimates of eqq. (6) as in table 6.2, except the base country against which differentials are computed is Germany rather than the U.S.

Money and Nominal-InterestDifferentials

Table 6.4

SE

t-stat.

- ,027 .007 .040 ,044 .013 - .069 -.185 - ,030 - .429

,023 ,018 .019 .019 .028 ,076 .069 .073 ,071 ,221

-1.16 .31 2.04 2.32 2.32 .16 - 1.00 -2.51 - .41 - 1.93

Canada

France

Germany

Italy

UK

.68

.81

.82

.94

.51

,065

R2

Two-Tail p-Value

Coeff.

.26 .71 .06

.03 .03 .87 .33 .02 .68 .07

Nore: These are joint instrumental variables estimates of an equation like (6), except the dependent variable is the difference betweea the country’s short-termnominal-interest rate and the U.S. three-month Treasury-bill rate, G(nation, US,i).

percentage points at annual rates. So a coefficient of about four on the moneygrowth differential would indicate that a 1 percentage point rise in the moneygrowth differential raises the nominal-interest differential by 1 percentage point. In fact, the sum of the first four lagged coefficients is only about 1.5, and the coefficient on contemporaneous money is negative. So, besides the

344

Alan C. Stockman

positive effect of lagged money differentials, there is weak evidence of a liquidity effect on nominal interest rates in the first quarter.16 The fact that an increase in the money-growth differential-which is associated with a lagged rise in the inflation differential-is accompanied by a fall in the expected real-interest differential (and perhaps a shorter-run fall in the nominal-interest differential) is some evidence in favor of the hypothesis that countries had (and used) some degree of monetary autonomy under the Bretton Woods system. Under the alternative (reserve-causation) hypothesis, which says that the rise in the nominal supply is an endogenous response to a real shock that ( a )raises the demand for money and (b)raises the relative price of domestic in terms of foreign goods with a lag, one might expect a rise in the international real-interest differential. Additional evidence on this point comes from attempts to model the changes in money-supply differentials based on past changes in inflation differentials and other variables. There appears to have been little relation between changes in the money-supply growth differential and lagged changes in income, inflation, or nominal-interest differentials. Estimates of equations for money-growth differentials show no strong relation between lagged inflation differentials and the current moneygrowth differential (see table 6.5). In fact, the point estimates of the coefficients of lagged inflation differentials on the money-growth differential are mainly negative. When lagged differentials of the balance of payments or balance of trade are added to this equation, they have no additional predictive content for the differentials in money-supply growth rates. So this evidence supports the hypotheses that changes in money-growth differentials were responsible for subsequent changes in inflation differentials, that variables that normally affect money-demand-growth differentials had little effect on shortrun differentials in money-supply growth, and that short-run international differentials in money-supply growth were not strongly related to inflation or money growth in other countries. In other words, there was little international transmission of inflation in the short run under Bretton Woods.

6.8 Conclusions International transmission of real and nominal disturbances occurs through many channels. Some are independent of the exchange-rate system, while others depend on that system. Some create positive international transmission, while others create negative transmission across countries. The data available for the Bretton Woods system after the establishment of convertible currencies are sufficiently short that little evidence can be obtained about most of the main issues regarding these channels of international transmission. Distinguishing common shocks to real variables from their international transmis16. Leeper and Gordon (1991) find no liquidity effects in the U.S. data for this period when changes in the money stock are treated as endogenous as in these estimates.

345 'Lgble 6.5

International Transmission under Bretton Woods Money-Growth Differentials Coeff.

SE

- ,027 ,007

.023 ,018 .019 .019 ,028 .076 ,069 .073 ,071 .221

,040 .044 ,065 ,013 - ,069 - ,185 - ,030 - ,429

R2

?-Stat. - 1.16

.37 2.04 2.32 2.32 .16 - 1.00 -2.51 - .41

- 1.93

Rvo-Tail p-Value .26 .71 .06 .03 .03 .87 .33 .02 .68 .07

Canada

Germany

Italy

UK

.29

.50

.94

.36

Nore: These are joint instrumental variables estimates of an equation like (6). except the dependent variable is the difference between the country's nominal money-supply growth rate and that of the United States, G(nation, US, Ml)).

sion and isolating the various channels through which transmission occurs require longer time series than the Bretton Woods system made available to us. But it is possible to obtain some evidence about the short-run international transmission of inflation under the Bretton Woods system. This paper has presented evidence that there was little short-run international transmission of inflation under Bretton Woods and that countries exercised some control over their own short-run inflation rates through monetary policy, despite pegged exchange rates. The evidence indicates that a country with a money-growth rate 1 percent above another country's subsequently experienced higher inflation of between one-fourth and one-half percent. This paper has argued that it is reasonable to interpret this as a causal relation from money-growth differentials to inflation differentials, rather than reverse causation. The reversecausation hypothesis, that real shocks affecting the demand for money also affected international inflation differentials with a lag, raises the question of precisely what these disturbances were. In addition, the evidence indicates that increases in the money-growth differential and subsequent increases in the inflation differential were associated with decreases in ex ante real interestrate differentials, which might be expected from liquidity effects if the causality ran from money growth to inflation. The reverse-causation hypothesis might suggest the opposite reaction for real interest rates if increases in the money supply occur in response to higher demands for credit. And the reverse-causation hypothesis cannot explain why money-growth differentials do not appear to be positively related to lagged inflation differentials. So the natural interpretation of the evidence is that countries had some scope for

346

Alan C. Stockman

monetary-policy independence under Bretton Woods and exercised it in ways that limited international transmission. This conclusion raises new puzzles with both theoretical and empirical components. Obviously, the international transmission of monetary disturbances and inflation under the Bretton Woods system was more complicated than simple models suggest. How were countries able to conduct independent monetary policies under the Bretton Woods system of pegged exchange rates? The simplest version of the monetary approach to the balance of payments implies that independent monetary policy is impossible in a country that pegs its exchange rate because the one instrument of monetary policy, the supply of the monetary base, must be used to peg the exchange rate. Other models suggest a possible role for independent monetary policy. Some (portfolio balance) models explicitly introduce two instruments: the monetary base and the currency denomination of the assets that the central bank buys and sells when it conducts open-market operations. By buying assets in one currency and selling them in another, the central bank can in principle affect relative rates of return on the assets, and the exchange rate, without altering the money supply. This permits, in principle, separation of the pegged exchange-rate policy from money-supply policy. Was this the main operative channel that limited short-run international transmission of inflation under Bretton Woods? Alternatively, there may be other channels through which central banks can conduct independent monetary policies under pegged exchange rates, perhaps involving distribution effects and the choices of markets and instruments used for open-market operations. Were these the main channels? If the United States was a reserve-currency country, did it control the long-run inflation rates of Bretton Woods nations, as Darby et al. (1983) argue? Did barriers to international trade in goods limit arbitrage and allow each country some shortrun control over its own nominal variables? If so, would these barriers have allowed long-run control, as long as price levels did not diverge too much? If this is the key channel that limited international transmission of inflation under Bretton Woods, what are its implications for flexible exchange rates? Does it imply that countries could use monetary policies to affect international relative prices under floating exchange rates? To what extent does this alter the international transmission of real disturbances under either exchange-rate system? These and other related questions call for future research.

Data Appendix Data are from OECD Main Economic Indicators, Citibase (all U.S. series), and Darby et al. (1983) (French real income). Canada. Data cover the entire period from 1960:l through 1971:2. France. Inflation, money-growth, and interest-rate data cover the entire pe-

347

International Transmission under Bretton Woods

nod from 1960:l through 1971:2. French real income data are from Darby et al., and I excluded the quarter of the general strike in May 1968. Because the data analysis employs lags, this reduces the sample period in France to 1961:1-1968: 1 and 1969:4-1971:2. Germany. Data cover the entire period from 1960: 1 through 1971:2. Italy. Inflation, money-growth, and real income data cover the entire period from 1960:l through 1971:2. The short-term interest rate is the auction rate on six-month Treasury bills, from OECD Main Economic Indicators: Historical Statistics, and is available from 1960:1 through 1969:2. United Kingdom. Data cover the entire sample, except M1 begins in 1963:1. United States. Data from Citibase.

References Backus, David K., Patrick J. Kehoe, and Finn E. Kydland. In press. International Real Business Cycles. Journal of Political Economy. Baxter, Marianne, and Mario Crucini. In press. Explaining Savings-Investment Correlations. American Economic Review. Baxter, Marianne, and Alan C. Stockman. 1989. Business Cycles and the Exchange Rate Regime: Some International Evidence. Journal of Monetary Economics 23(May):377-400. Cardia, Emanuela. 1991. The Dynamics of a Small Open Economy in Response to Monetary, Fiscal, and Productivity Shocks. Journal of Monetary Economics 28(December):411-34. Cole, Harold L., and Maurice Obstfeld. 1991. Commodity Trade and International Risk Sharing: How Much Do Financial Markets Matter? Journal of Monetary Economics 28(August):3-24. Costello, Donna. 1991. A Cross-Country, Cross-Industry Comparison of the Behavior of Solow Residuals. University of Florida. Typescript. Darby, Michael R., James R. Lothian, Arthur E. Gandolfi, Anna J. Schwartz, and Alan C. Stockman. 1983. The International Transmission of Znjlation. Chicago: University of Chicago Press. Dellas, Harris. 1986. A Real Model of the World Business Cycle. Journal of Znternational Money and Finance 5(September):381-94. Dixit, A. K., and J. E. Stiglitz. 1977. Monopolistic Competition and Optimal Product Diversity. American Economic Review 67(June):297-308. Dumas, Barnard. 1987. Time to Ship and the Dynamic Behavior of Deviations from the Law of One Price. University of Pennsylvania. Typescript. Engel, Charles. 1991. Is Real Exchange Rate Variability Caused by Relative Price Changes? An Empirical Investigation. Working paper. University of Washington. Frenkel, Jacob A., and Assaf Razin. 1987. Fiscal Policies in the World Economy: An Intertemporal Approach. Cambridge, Mass.: MIT Press. Iwami, Toru. 1991. The Bretton Woods System as a Gold Exchange Standard. Discussion Paper no. 91-F-11. University of Tokyo, Research Institute for the Japanese Economy. Leeper, Eric, and David Gordon. 1991. In Search of the Liquidity Effect. International Finance Discussion Paper no. 403. Washington, D.C.: Federal Reserve Board.

348

Alan C. Stockman

Lucas, Robert E., Jr. 1982. Interest Rates and Currency Prices in a Two-Country World. Journal of Monetary Economics 10(November):335-60. Lundvik, Petter. 1990. Business Cycles in a Small Open Economy: Sweden, 18711987. Working paper. Stockholm: Institute for International Economic Studies. Mendoza, Enrique. 1991. Real Business Cycles in a Small Open Economy. American Economic Review 81(September):797-818. Miller, Preston, and Richard M. Todd. 1991. Monetary Policy Transmission When There Are Nontraded Goods. Working Paper no. 481. Federal Reserve Bank of Minneapolis. Mussa, Michael. 1986. Nominal Exchange Rate Regimes and the Behavior of Real Exchange Rates. Carnegie-Rochester Conference Series on Public Policy 25: 117214. Stockman, Alan C. 1980. A Theory of Exchange Rate Determination. Journal of Political Economy 88( 1980):673-98. . 1983. Real Exchange Rates under Alternative Nominal Exchange Rate Systems. Journal of International Money and Finance 2(August): 147-66. . 1988a. Real Exchange Rate Variability under Pegged and Floating Nominal Exchange Rate Systems: An Equilibrium Theory. Carnegie-Rochester Conference Series on Public Policy 29(Autumn):259-94. Stockman, Alan C., and Harris Dellas. 1989. International Portfolio Nondiversification and Exchange Rate Variability. Journal of International Economics 26(May):27 1-90. Stockman, Alan C., and Linda Tesar. 1991. Tastes and Technologies in a Two-Country Model of the Business Cycle: Explaining International Comovements. NBER Working Paper no. 3566. Svensson, Lars E. 0. 1985. Currency Prices, Terms of Trade, and Interest Rates: A General Equilibrium Asset-Pricing, Cash-in-Advance Approach. Journal of International Economics 18(February): 17-41. . 1986. Sticky Goods Prices, Flexible Asset Prices, Monopolistic Competition, and Monetary Policy. Review of Economic Studies 3(July):385-405. . 1987. International Transmission of Fiscal Policy. Scandinavian Journal of Economics 89(July):305-34. Svensson, Lars E. O., and Sweder van Wijnbergen. 1989. Excess Capacity, Monopolistic Competition, and the International Transmission of Monetary Disturbances. Economic Journal 99(September):785-805. Waldmann, Robert James. 1990. Assessing the Relative Sizes of Industry- and NationSpecific Shocks to Output. Working paper. Harvard University.

Comment

TOW Iwami

Alan C. Stockman’s paper consists of two parts. The former (secs. 6.1-6.6) considers various forms of disturbances (shocks) and possible ways in which they are transmitted internationally. The latter (sec. 6.7) is devoted to testing the possibility that independent monetary policies cause inflation differentials, at least in the short run, contrary to the widely accepted view of the pegged exchange-rate system. Tom Iwami is professor of economics at the University of Tokyo.

349

International Transmission under Bretton Woods

The logical relation between the former and the latter parts is difficult to understand. In the former, Stockman mainly discusses the effects of real disturbances, while his argument in the latter is not about the transmission of inflation but about differential inflation rates and about how these differentials are caused by monetary rather than real factors. In considering theoretical models, Stockman mentions actual settings of the period and compares the logical consequences of each model with the results of empirical research to date. Undoubtedly, this is the correct procedure. However, I am skeptical that the various models in the former part are relevant to the realities of the Bretton Woods era. Let us take examples of “disturbances to preference,” “technology shock,” and the “presence of nontraded goods.” The extent of “preference” effects is dubious, while technology must have been strongly related to the high-growth performance of the period and hence to economic fluctuations as well. But the relevance of Stockman’s discussion is limited because a productivity shock is assumed to take place in the nontradable goods industry. I will discuss below the productivity shock mainly to tradables. The reader would also like to know what kind of fluctuations in real variables the author deduces from the previous research. Stockman refers to the findings of, for example, Baxter and Stockman that changes in real macroeconomic quantities were not so much affected by the type of an exchange-rate system, on the one hand, and that the cross-country correlation of industrial production was higher under the Bretton Woods system than in the later period, on the other.’ Then I wonder how these two statements are consistent with each other. While Stockman concludes that inflation differentials resulted from autonomy in monetary policy, differences in productivity growth and in prices of nontradables among countries would also lead to inflation differentials. The best example of this is the fact that, of the countries represented in his figure 6.3, Japan has the highest average inflation rate; such a high inflation rate is caused by the higher inflation rate of nontradables, whereas import price is determined internationally. The greater price increase of nontradables reflects the ever growing demand for labor force and real estate, a demand associated with rapid economic growth. In order to assess relative inflation, an econometric model based on the CPI is misleading. Table 6C.1 shows that the EPI (export price index) inflation rate of Japan is the lowest among the seven developed countries in the period 1960-70, despite the highest average inflation rate for the CPI. If the inflation rates in Japan had been quite low for every price index, the major cause would have been anti-inflationary management of the money supply. The higher inflation rate of the GDP deflator rejects such a possibility. 1. See Marianne Baxter and Alan C. Stockman, “Business Cycles and the Exchange Rate Regime: Some International Evidence,”Journal ofMonetary Economics 23, no. 3 (1989): 377-400.

350

Alan C. Stockman

Table 6C.1

International Divergence of Inflation Rates (yearly average, %) ~~

~

~

1950-60

1960-70

1950-70

1970-80

CPI: United States Japan Germany France Britain Italy Canada

2.09 4.01 1.88 5.58 3.33 3.15 2.20

2.75 5.74 2.59 4.04 4.05 3.64 2.72

2.42 4.87 2.23 4.81 3.69 3.39 2.46

7.82 8.97 5.08 9.63 13.09 13.97 8.04

GDP deflator: United States Japan Germany France Britain Italy Canada

2.61 3.67’ 2.84 6.03 4.08 3.19 3.43

3.10 4.30 3.71 4.35 4.23 4.50 3.01

2.86 4.09b 3.27 5.18 4.16 3.84 3.22

7.39 7.62 5.31 9.49 13.95 16.40 8.76

1.50 2.21 2.03 5.00 2.87 .54

...

1.52 1.28 1.32 2.86 3.08 2.49 1.77

1.51 1.75 1.68 3.92 2.98 1.51

9.31 7.53 5.10 8.07 13.57 15.42 9.68

1.26 .29 3.89 4.80 2.60 - .55 1.28

1.52 .28 .76 2.49 3.11 .55 2.16

1.39 .28 2.31 3.64 2.85

14.55 3.69 5.15 9.14 14.51 16.14 11.29

WPI: United States Japan Germany France Britain Italy Canada EPI: United States Japan Germany France Britain Italy Canada

...

.oo

I .72

Source; Toru Iwami, “Japan’s Experiences under the Bretton Woods System” (Discussion Paper no. 92-F-1, University of Tokyo, Faculty of Economics, 1992). Original data from International Financial Statistics: Supplement (1987). Note: The figure marked a stands for 1955-60 only and for 1955-70 only. Inflation rates are p) = (log Pr - log Po)/r, where p is the average rate of inflation, Pt is calculated as log (1 the price index of the fth year, and Po is the price index of the benchmark year.

+

Thc lower WPI inflation rate is the result of higher-productivity growth specific to large-scale firms, while the cause of the still lower inflation rate of the EPI is twofold: first, the composition of exports shifted to goods of higherproductivity growth; second, exporters may have cut export prices further, to get a larger market share abroad at the lower margins. The real exchange rate

351

International Transmission under Bretton Woods

of the CPI is rising, while that of the EPI is falling (see fig. 6C.1).2 It would be more fruitful to interpret the inflation differentials in terms of the dynamic aspect of postwar economic development. Unfortunately, Stockman’s argument is confined to the short-run effects because of the limited time-series data. Another important question is how far his argument goes and what the implications are of his econometric models. The independence of monetary policy, if any, does not imply that the balance of payments does not constitute a restraint on macroeconomic policy. The developed countries seem to have obeyed the “rules of the game” in general under the Bretton Woods system, with the exception of the United state^.^ This asymmetry enabled the developed countries to expand their money supply despite the pegged exchange rates. The fact that the key-currency country, the United States, did not respond to the balance of payments deficits by contracting the money supply is a main characteristic that distinguishes the Bretton Woods system from the classical gold ~ t a n d a r dBoth . ~ the success in solving the international liquidity shortage and the consequent breakdown of the whole system resulted from U. S . macroeconomic policy preferring internal rather than external balances. The remarkable differences in inflation rates between the two periods before and after the closing of the gold window suggest that the pegged exchange rate was the factor that imposed discipline on excessive monetary expansion, even in the United States. One might argue that the inflation of the 1970s was due mainly to the oil crisis. However, without the dollar devaluation, OPEC countries would not have raised oil prices so dramatically. The oil crisis was one of the side effects generated by the dollar devaluation. It was no coincidence that the so-called golden age of capitalism (i.e., greater economic growth and a lower inflation rate) and Bretton Woods co-occurred.5 Admittedly, the era of fixed exchange rates with full convertibility of the major currencies is too short to provide us with enough data to analyze empirically the long-term effects of real as well as monetary shocks. Partly because of such limitations, Stockman must content himself on the whole with raising questions rather than answering them. 2. The preceding discussion is based on Toru Iwami, “Japan’s Experiences under the Bretton Woods System” (Discussion Paper no. 92-F-1, University of Tokyo, Faculty of Economics, 1992). The classic statement of productivity effects on the real exchange rate is Bela Balassa, “The Purchasing-Power Doctrine: A Reappraisal,” Journal of Political Economy 72 (1964): 584-96. 3. Ronald I. McKinnon stresses that the asymmetry between the United States and the rest of the world continued to exist regardless of the exchange-rate system (see his comments in chap. 13 in this volume). 4. See Toru Iwami, “The Bretton Woods System as a Gold-Exchange Standard” (Discussion Paper no. 91-F-11, University of Tokyo, Faculty of Economics, 1991). 5. On the golden age of capitalism, see Stephen A. Marglin and Juliet B . Schor, eds., The Golden Age of Capitalism: Reinterpreting the Postwar Experience (Oxford: Oxford University Press, 1990).

352

Alan C. Stockman

140

:..,. .............. ..........

130 120110-

..............

...................

*..

...................

..

..

70 60 -

-

-

.......

I

.

.......

501 1 1 1 1 1 1 i I I 7 1950 1952 1954 1956 1958 1960 1962 1964 1966 1968 1970 8

-WP I

.....

1

EPI

...

1

CP I

Fig. 6C.1 Yeddollar real exchange rate based on WPI, CPI, and EPI (195071) (1950 = 100)

Source: Tom Iwami, “Japan’s Experiences under the Bretton Woods System’’ (Discussion Paper no. 92-F-1, University of Tokyo, Faculty of Economics, 1992). Original data from International Financial Statistics: Supplement (1987). Note: Exchange rate is expressed in terms of yeddollar.

Comment

Bennett T. McCallum

Alan Stockman’s useful paper consists of two distinct parts. The first of these is the exposition in sections 6.1-6.6 of a theoretical framework for the analysis of the transmission, from one nation to another, of the effects of macroeconomic shocks. Although a form of nominal price stickiness is introduced in section 6.6, for the most part the shocks considered are real (as opposed to monetary) in nature-shocks to technology and preferences. Stockman’s development of this theoretical framework is expertly crafted, just as one would anticipate in light of his role as a major contributor to the subject.’ The second part of the paper, by contrast, consists of an empirical exploration (presented in sec. 6.7) of the proposition-implied by some popular Bennett T. McCallum is H. J. Heinz Professor of Economics at Camegie Mellon University and a research associate of the National Bureau of Economic Research. 1. Among those contributions are Alan C. Stockman, “Real Exchange Rates under Alternative Nominal Exchange Rate Systems,” Journal of International Money and Finance 2 (1983): 14766, “The Equilibrium Approach to Exchange Rates,’’ Federal Reserve Bank of Richmond Economic Review 73, no. 2 (1987): 12-31, and “Real Exchange Rate Variability under Pegged and Floating Nominal Exchange-RateSystems: An Equilibrium Theory,” Carnegie-Rochester Conference Series on Public Policy 29 (1988): 259-94.

353

International Transmission under Bretton Woods

models-that nations had no scope for monetary autonomy under the Bretton Woods arrangements. This exploration is not tightly linked to the framework developed in the first part of the paper but concerns the transmission of inflation. The following comments will, like Stockman’s presentation at the conference, focus on the second of these two distinct subjects. I enjoyed studying Stockman’s empirical work and would certainly agree with his main conclusion, namely, that the evidence from the period 1958-71 is inconsistent with the proposition mentioned above. This empirical work is rather ingenious in its design; by focusing on differentials across countries and imposing equality restrictions on distributed-lag coefficients where permitted by the data, Stockman was able to get some estimates that are statistically significant and come close to being economically intelligible. In my opinion, the results fall somewhat short of full intelligibility, however, since interestrate differentials are not first-differenced as the other variables are. Except for that, Stockman’s equation (6) would be a distributed-lag money demand equation of conventional specification (solved for the price level and firstdifferenced). It might also be noted that Stockman has been careful to report exactly what he is doing in his econometric work. There is one interpretive quibble that I will mention, nevertheless, before going on to more important matters. It concerns the statement that “a sustained 1 percent rise in domestic money growth relative to foreign money growth leads to a .28 percent rise in the inflation differential,” .28 being the sum of the coefficients on the d log(Ml), variables in table 6.1. But since lagged values of the dependent variable are included as regressors, we need to divide by 1.0 minus the sum of their coefficients (i.e., by 1.0 - .27 = .73) to ge the steady-state effect. That would give a value of .38, for countries other than Canada, which is still rather low in relation to 1.0. (For Canada, the number is .28/[1 - .799] = 1.39.) Let us now consider Stockman’s main conclusion. It is that, in his words, “there was little short-run international transmission of inflation under Bretton Woods and that countries exercised some control over their own short-run inflation rates through monetary policy, despite pegged exchange rates.” With regard to this conclusion, it is important that the two “short-run” qualifications are included. If they were excluded, this conclusion would suggest that it is wrong to tell one’s students that the adoption of fixed exchange rates precludes a nation from choosing its own inflation rate. But, when I tell my students that, what I am referring to is a nation’s average inflation rate over a number of years-perhaps even a decade. Now Stockman’s results indicate that, over short spans of time, nations can chose their money growth rates and that these influence their inflation rates. But these results do not deny that, if a nation maintains a 20% inflation rate for ten years, then it will be unable to keep its exchange rates fixed if, for example, most other nations are inflating at only 2% per year. Consequently, Stockman is right to state his conclusion

354

Alan C. Stockman

in terms of there being some short-term or temporary scope for monetary autonomy under Bretton Woods. Having said that, I must go on to admit that, at first glance, Stockman’s data seem rather inconsistent even with the longer-run version of the noautonomy proposition. Specifically, the series plotted in his figure 6.1 and 6.3 appears inconsistent with the usual no-autonomy notion because Japan, Italy, and the Netherlands are three nations in his sample with high average inflation rates (according to fig. 6.3), yet figure 6.1 indicates that all three of these had currencies that appreciated relative to the dollar over the years of convertibility under Bretton Woods. Reflection suggests that it would be more appropriate to base average inflation rates for the comparison on a somewhat longer span of time than 196267. Not having access to Bordo’s paper (chap. 1 in this volume) at the time, I calculated my own averages and decided to base them on the period 1957-70, using annual CPI data from the IMF’s International Financial Statistics Yearbook. The chosen span drops off 1971 because by then the system was breaking down, and it includes 1957 and 1958 because the inflation experience of those years would seem to be relevant for the 1959-70 period of convertibility. These choices may appear rather ad hoc but nevertheless seem appropriate. Be that as it may, table 6C.2 shows that the average inflation rates over 1957-70 all fall within 1.25 points of the U.S. rate (2.58%) except for France and Japan-which are about 4.5%, just over two points greater than the U.S. value. And these two exceptions are not troublesome from the perspective of the usual notion because France was in fact forced to devalue in 1958 and again in 1969 whereas Japan’s inflation rate was offset over these years by its spectacular rate of productivity growth-something that a country cannot achieve by macroeconomic policy measures. So the data are actually consistent with the longer-term version of the no-autonomy proposition. In conclusion, I would like to comment on the recently popular practice of testing the validity of the purchasing power parity (PPP) doctrine as a longrun proposition. This topic is not explicitly on Stockman’s agenda but is in fact closely related because long-run validity of PPP is much the same as longrun incompatibility of fixed exchange rates with inflation rate discrepancies. Specifically, the recent practice I am concerned with is that of testing for the presence or absence of cointegration of nominal exchange rates with relative price levels. If these two series are not cointegrated, but instead the implied real exchange rates are nonstationary (in the sense that their ARMA representations have unit roots in the AR polynomial), then the PPP is said to fail even as a long-run proposition.2 2. Three references that come to hand are Mark P. Taylor, “An Empirical Examination of LongRun Purchasing Power Parity Using Cointegration Techniques,” Applied Economics 20 (1988): 1369-81; Cletus C. Coughlin and Kees Koedijk, “What Do We Know about the Long-Run Real Exchange Rate?” Federal Reserve Bank of St. Louis Review 7 2 , no. 1 (1990):36-48; and Robert McNown and Myles S. Wallace, “National Price Levels, Purchasing Power Parity, and Cointegration: A Test of Four High Inflation Economies,” Journal of International Money and Finance 8 (1989): 533-45. But there are others that could be cited.

355

International Transmission under Bretton Woods

Table 6C.2

CPI Inflation Rates, 1957-70 CPI, 1975 Base

Country Belgium Canada France Germany Italy Japan Netherlands United Kingdom United States

1957 48.3 51.0 34.7 55.0

37.9 31.8 42.2 35.2 52.3

1970

Ratio, X

66.9 70.2 65.4 74.2 58.4 58.0 66.0 54.2 72.1

1.3851 1.3764 1.8847 1.3491 I S409 1.8239 1.5640 1.5398 1.3786

X

Average ,07143 - 1 .0254 ,0249 ,0499 ,0233 ,0338 .0473 ,0350 ,0338 ,0250

~

Source: IMF, International Financial Statistics Yearbook (1980).

But I would argue that to draw such a conclusion is unwarranted. Suppose the relevant real exchange rate is a random walk with zero drift and small variance and that its evolution is independent of monetary policy actions in the relevant countries. Then it remains true that major differences in money growth rates over a decade would have major effects on the (nominal) exchange rate and tend to dominate its behavior. Indeed, the practical messages of the PPP doctrine would retain their validity, provided that the doctrine is interpreted as a long-run matter. That would remain true, furthermore, even if real exchange rates were dependent at high frequencies on money growth behavior, provided that there is independence at low frequencies. Note also that the logic of the cointegration test would imply that one should in principle reject PPP if any of the system’s shock processes that affect the real exchange rate have even a small permanent component. As Stockman has emphasized in previous writing^,^ one would surely expect this to be the case. But, even if the process is entirely permanent, as in my example, one should not jump to the usual conclusion. Thus, in my opinion, the barrage of cointegration tests of PPP that we have seen in the literature serves primarily as another (unnecessary) indication of the profession’s enthusiasm for the application of new and undigested statistical techniques.

General Discussion Sebastian Edwards followed up on Bennett McCallum’s comment that much of what Stockman’s regressions are picking up is not nominal disturbances but structural disturbances that permanently altered the real exchange rate. 3. Especially in Stockman, “The Equilibrium Approach to Exchange Rates,” and “Real Exchange Rate Variability under Pegged and Floating Nominal Exchange-Rate Systems.”

356

Alan C. Stockman

The disturbances that are important include productivity differentials via the Balassa effect, where one would expect countries with more rapid productivity growth to exhibit equilibrium real appreciation; terms of trade shocks; and changes in tariffs and commercial policy. Alexander Swobodu questioned Stockman’s claim that there was no world inflation rate. He suggested that the approach taken in Bordo’s table 1.1 (see chap. 1 in this volume) comparing the degree of convergence toward an average across regimes, would reveal a world inflation rate for the Bretton Woods period as well. He also suggested testing for the effects of aggregate money growth (across countries) on average inflation. He also did not agree with Stockman’s interpretation of his findings of little influence of changes in international reserves on money growth as suggesting sterilization. The same results could follow if countries set money growth targets, which in turn induced changes in international reserves, yet without any causality running from reserves to money. Huns Genberg suggested that the author should test for feedback effects through interest rates and the price level. Richard Murston thought it was possible that, instead of the regressions identifying a money demand equation, as McCallum argued in his Comment, they were identifying an endogenous money supply.

7

The Role of International Organizations in the Bretton Woods System Kathryn M. Dominguez

With the world at war, participants from each of the Allied countries convened on 1 July 1944 in Bretton Woods, New Hampshire, to create a new international monetary system. The breakdown of the interwar gold standard, and the mutually destructive economic policies that followed, convinced leaders that a new set of cooperative monetary and trade arrangements was a prerequisite for world peace and prosperity. The outcome of the conference, known as the Bretton Woods Agreement, included the creation of an adjustable peg exchange rate system and the establishment of two international organizations that would maintain economic cooperation among the participating countries. To this end, the conference participants drafted charters for the International Monetary Fund (IMF) and the International Bank for Reconstruction and Development (the World Bank). At the Havana conference in 1947, participants formulated a charter for the International Trade Organization (ITO).’ Member Kathryn M. Dominguez is associate professor of public policy at the Kennedy School of Government, Harvard University, and a faculty research fellow of the National Bureau of Economic Research. The author is grateful to Alfred0 Cuevas for outstanding research assistance and Albert0 Alesina, Bill Branson, Michael Bordo, Barry Eichengreen, Peter Kenen, Maurice Obstfeld, Leslie Pressnell, and Richard Zeckhauser for helpful comments and suggestions. The International Finance Section at Princeton University and the NBER provided financial support. 1. Discussions and disagreements between the United States and Britain on trade policy began as early as 1940 and continued throughout the war in the context of Article VII of the Mutual Aid Agreement (Lend Lease). In September 1943, the two countries reached a short-lived “understanding” on trade, the Washington Principles. But by 1944 the United States backed away from this broad, across-the-board approach to trade liberalization, and the United Kingdom followed suit. As a consequence, the British delegation to the Bretton Woods conference was under a strict cabinet directive not to discuss trade policy with their US.opposite numbers. The United Kingdom reluctantly agreed to support a U.S. proposal for an international conference on trade and employment (the Havana conference) during the Anglo-American Financial Agreement negotiations.

357

358

Kathryn M. Dominguez

countries subsequently ratified the charters for the IMF and the World Bank,z while the General Agreement on Tariffs and Trade (GATT) eventually subsumed some of the original goals of the IT0 . This paper examines the roles played by these organizations in maintaining the Bretton Woods system. Theory indicates that, even if countries understand that cooperation will lead them to a Pareto superior outcome, they need not cooperate unless they are convinced that other countries are also committed to doing so. In this context, international organizations can facilitate cooperation by serving as commitment mechanisms. Cooperation in the Bretton Woods system involved the maintenance of stable exchange rates and unrestricted trade among member countries. The commitment mechanisms that the Bretton Woods institutions provided member countries included rules of cooperation, financial resources to enable them to play by the rules, and a centralized source of information on each others’ commitment to the rules. In practice, the two Bretton Woods institutions and the GATT had limited success convincing their members to maintain cooperative arrangements. The evidence suggests that both the carrot and the stick that the institutions employed were weak commitment mechanisms. First, the main institutional carrot, financial assistance, was not always available to countries that played by the rules of the game. Countries were ineligible for assistance once their accumulated debt exceeded their capacity to repay. Second, the historical record shows that the institutions rarely wielded the stick, in that they did not consistently enforce the rules of the system. Third, the GATT’s relatively relaxed rules of the game effectively provided countries a trade controls escape route from the limits that the IMF required be observed for exchange controls. While all three organizations were ultimately unable to convince countries to maintain the cooperative behavior envisioned by their architects in the 1940s, the organizations survived the collapse of the Bretton Woods system by evolving with the changing economic environment. The IMF, in particular, broadened its role as a centralized source of information on member country economic performance. Postwar history suggests that information monitoring and sharing has been a relatively effective commitment mechanism for all three international organizations. The paper is organized as follows. Section 7.1 describes the international cooperation problem in theory. A stylized exchange rate game is presented to highlight individual country incentives to cooperate. Practical difficulties in achieving decentralized cooperation are then described along with three potential solutions to the problem. Section 7.2 examines the goals of the architects of the Bretton Woods system and the institutions that were created to facilitate the achievement of those goals. Section 7.3 presents two examples that illustrate the conditions under which institutions can provide countries 2. With the exceptions of the Soviet Union, Liberia, and New Zealand, all the nations that participated in the Bretton Woods conference ratified the charters for the IMF and the World Bank.

359

The Role of International Organizationsin the Bretton Woods System

effective incentives to maintain cooperation, and section 7.4 presents empirical evidence on the actual performance of the postwar institutions. Section 7.5 discusses the more recent evolution of the IMF’s role in maintaining international cooperation. Section 7.6 presents conclusions and lessons for future cooperative arrangements.

7.1 The International Cooperation Problem in Theory Many of the participating countries at the Bretton Woods conference contributed both to the establishment and to the breakdown of cooperative arrangements in the interwar period. They were thus well aware of the incentives that led countries to defect from the gold standard. Game theory allows us to examine these incentives formally. When two countries interact in a game in which each can do better individually by taking a particular action, a unique Nash equilibrium exists in which both take the action even though they are jointly worse off. In such games, it is easy to show that an outcome in which neither country takes the action is Pareto superior to the uncooperative solution to the game. Even when both countries understand this, it is typically difficultto arrive at this better equilibrium without the help of a commitment mechanism. Neither country will cooperate unless each can be convinced that the other is also committed to doing so. Section 7.1.1 provides a stylized example of the sort of exchange rate game played by countries in the interwar period. In the game, countries have an incentive to devalue but can be made better off if they commit not to do so. Section 7.1.2 discusses why the Pareto-superior cooperative solution is difficult to achieve, and section 7.1.3 presents three possible solutions to this problem. 7.1.1 The Devaluation Game Consider a model in which two countries, home (H) and foreign (F),face an established fixed exchange rate system. Assume that both countries initially declare par values against gold. Each country in this model then has two policy options, to defect from the system by devaluing the domestic currency against gold or to maintain the par value of the domestic currency. If one country devalues and the other maintains its par value, the country that devalues gains a trade surplus (or).3 However, a country that devalues also incurs a cost (C) for defecting from the fixed exchange rate system. The existence of this devaluation cost is a common assumption in the literature. Eichengreen describes it as “a transactions cost associated with the existence of more than one currency (analogous to extra costs of interstate trade in the United States if there existed 50 state monies, all floating against one another)” (Eichen3. This abstracts from complications arising from J-curve effects and the failure of the MarshallLerner condition to hold.

360

Kathryn M. Dominguez

green 1987, 7). More generally, countries would presumably not agree to be members of a fixed exchange rate system unless they believe to some degree that exchange rate instability is costly. Assume, however, that the benefits from unilateral devaluation outweigh the costs (a> C). The payoffs for each country in this game are described in figure 7.1. This game is an example of the classic prisoner's dilemma and has an equilibrium in dominant strategies in which both countries devalue. And, as long as devaluation is costly (C > 0), this is the only Nash equilibrium for the game.4 Moreover, when C > 0, the Nash equilibrium is Pareto inferior to the cooperative solution where neither country devalues. This is the type of cooperation problem that participants at the Bretton Woods conference in 1944 had in mind when they set about creating a new international monetary system. During the 1930s, many countries defected from the gold standard system that was established after World War I. Countries engaged in competitive devaluations, hoping both to conserve gold and to shift world demand toward domestic output. The devaluations largely offset each other, but countries combined beggar-thy-neighbor exchange rate policies with trade and capital restrictions that left all countries worse off.* Much of the discussion that led up to the Bretton Woods conference centered around the creation of international organizations that would insure against a repeat of the collapse in cooperation that occurred in the 1930s. Before examining the possible roles of these organizations in the maintenance of the system, it is useful to determine whether there exist conditions under which countries have individual incentives to achieve the cooperative equilibrium. The Folk Theorem suggests that, if we place our model in a repeated game setting, a cooperative outcome may emerge without the help of external institutions.6 In the one-period game, each country maximizes its payoff taking the actions of the other as given. If, instead, the game is played repeatedly, countries can condition their actions in each period on what has occurred in the past. Let Py (Pr) be the payoff to the home (foreign) country in period t and IT^) its total stream of payoffs from the game, appropriately discounted, m

Ti"

6'Py,

= 1=0

where 6 is the common discount factor (0 < 6 < 1). 4. If devaluation is costless (C = 0 ) , the only Nash equilibrium will be one in which both countries devalue, but it no longer involves the use of (strongly) dominant strategies. 5 . Bordo (chap. 1 in this volume) suggests that the perception, during and after World War 11, that countries' policies in the interwar period were destructive was incorrect. However, for purposes of explaining the goals of the participants at the Bretton Woods conference, it is the perception, and not the reality, that is important. 6 . For a general discussion of equilibrium concepts in repeated games, see Kreps (1990, chap. 14).

361

The Role of International Organizations in the Bretton Woods System FOREIGN NOT DEVALUE

DEVALUE

NOT DEVALUE DEVALUE Fig. 7.1 Payoff matrix for devaluation game

Consider the following strategy: 1. If no country in the past history of the game has devalued, do not devalue. 2. If a country in the past history of the game has devalued, devalue. This tit-for-tat strategy can be shown to lead to a subgame perfect equilibrium. In other words, if both countries follow this strategy, then neither benefits from deviating from either phase 1 (cooperation) or phase 2 (punishment). During phase 2, both countries play their one-shot equilibrium strategies, so punishment is sustainable. A comparison of the gains from deviating in any one period, less the costs incurred during the punishment phase, against the gains from following the proposed strategy shows that the cooperative phase is feasible. If the home country follows the cooperative strategy, its expected payoff is zero (IT"* = 0). If the home country deviates in any period, it stands to gain (a - C) immediately, but will incur costs (C) forever after.7 The expected payoff is therefore (2) In order to sustain cooperation, the payoffs to cooperation must exceed the payoffs from devaluing (nH5 IT"*). This implies (3)

Thus, as long as the cost of devaluation is positive (C > 0) and the discount rate (6) is high enough, the cooperative solution is one possible subgame perfect equilibrium. If devaluation is costless (C = 0), then the threat of reversion to the devaluation equilibrium has no bite, and the cooperative solution is not feasible.8 This result, therefore, suggests that countries can maintain 7. This assumes that, once a country devalues, it is never forgiven; its reputation as a defector is irreversible. An alternative assumption is that, after the defecting country has been punished with a retaliatory devaluation, both countries return to the cooperative equilibrium. In this case, costs are incurred in only two periods. Axelrod's (1984) version of tit for tat is of this latter form; it is a one-round punishment strategy. The qualitative results of the game do not depend on which of these assumptions holds. 8. Of course, if devaluation is costless (C = 0), then the payoffs in the cooperative and competitive devaluation equilibriums are identical, so there is no incentive to cooperate.

362

Kathryn M. Dominguez

cooperative behavior (fixed par values) as long as the game is repeated and there exist positive costs to devaluation.

7.1.2 Practical Difficulties with Decentralized International Cooperation While the devaluation game illustrates that a cooperative outcome is both feasible and sustainable under fairly unrestrictive conditions,g the experience of countries in the interwar period indicates that there must be more to the problem than the game suggests. The game abstracts from at least three important and potentially interrelated problems: the existence of more than two countries, incomplete information, and asymmetries among countries. The number of potential equilibrium outcomes increases when more than two countries are introduced to the model. As the number of countries increases, it becomes less likely that countries will achieve the cooperative outcome. The problem is a classic one, collective action, As the number of countries increases, so also does the number of potential defectors. In the two-country game, both countries were assumed to know with certainty both the benefits and the costs of devaluation. Moreover, in order to focus on generic incentives, benefits and costs were assumed to be identical across countries. In practice, neither class of assumptions is likely to hold. The benefits of a unilateral devaluation are likely to be a complicated function of the magnitude of the devaluation, the elasticities of import and export demands, and the prospective actions of the other country. It is unlikely that each country can predict the effect of its own devaluation, let alone the costs that such an action might impose on other countries. Each country’s incentive to maintain cooperation necessarily depends on its knowledge of the costs and benefits of defection; if these are uncertain or unknown, then the likelihood of a cooperative outcome may be reduced. The benefits and costs of devaluation in practice may also vary across countries. The effect of a devaluation is likely to be greatest for smaller countries with the most open economies. Country heterogeneity could affect the equilibrium outcome of the game in various ways depending on the relative sizes of a and C for each country. For example, if both a and C are large for the smaller country and small for the larger country, a cooperative solution may still be feasible. However, if countries share the same C but one receives a larger a,that country will have a greater incentive to defect and a better prospect that the other country, recognizing the asymmetry, will not defect in response.

7.1.3 Three Possible Solutions to the Cooperation Problem There are several different approaches that one can take to solve the cooperation problem. Economists often focus on rules-based solutions, while po9. The cooperative outcome also depends on the assumption that both countries follow the titfor-tat strategy.

363

The Role of International Organizations in the Bretton Woods System

litical scientists are more likely to study the role of institutions and negotiations in the achievement of cooperative outcomes. Rules-based solutions are typically designed to be simple but rigid. Rules are formulated to be simple, so that they can easily be followed, but rigid, so that countries cannot maneuver around them. Institutions and negotiations tend, in contrast, to be flexible but to allow for complexities in arrangements. The advantage of a rules-based solution is that, once formulated, it is easy to implement: “If appropriate rules can be found, rule-based regimes have the advantage over non-cooperative regimes of leading to superior outcomes, while at the same time preserving the reality of national autonomy and decentralization in economic decision-making” (Cooper 1985, 1227). A major problem with the rules-based approach is formulating rules that are acceptable to all countries. Rules typically predetermine the distribution of the gains from cooperation, and countries that perceive that they have bargaining power may be reluctant to play by the rules. Alternatively, hegemonic theories suggest that countries with more bargaining power are more likely to agree to rules because they can set the rules: “Hegemonic structures of power, dominated by a single country, are most conducive to the development of strong international regimes whose rules are relatively precise and well obeyed” (Keohane 1980, 132). Apart from the problems for countries of setting and agreeing to abide by rules, this solution often collapses in the face of change. Rules are unlikely to cover all contingencies, and, as soon as conditions change, countries are unlikely to stick to the rules. A second solution to the cooperation problem, international negotiations, is the starkest alternative to cooperation by rules. Negotiations provide a process of communication between Countries. This process can potentially lead to the formulation of rules or the creation of an organization, the third solution, but it need not. International negotiations often take the place of the rules solution when cooperation is in danger of collapsing. lo More generally, negotiations are typically ongoing and not necessarily cumulative; thus, they represent one of the most flexible means for countries to achieve a cooperative solution. International negotiations may also serve to promote cooperation at the domestic level. In countries in which there is no policy consensus at the domestic level, governments may welcome the pressure to comply with policies agreed on in international negotiations. In this manner, “international negotiations sometimes enable government leaders to do what they privately wish to do, but are powerless to do domestically” (Putnam 1988, 457). The third potential solution to the cooperation problem is the creation of international organizations. Organizations often serve to promote cooperation 10. Indeed, the G7 summit negotiation process arose in the wake of the Bretton Woods system’s collapse (Putnam and Bayne 1987).

364

Kathryn M. Dominguez

by enforcing rules-based solutions and providing a forum for international negotiations. In addition, organizations can serve to promote cooperation by providing a centralized source of information to their members. Milgrom, North, and Weingast (1990) show that, when information problems are substantial, repeat play and tit-for-tat strategies are insufficient to sustain cooperation. They study the medieval “law merchant” system in this context and find that, by efficiently gathering and disseminating information to traders, the organization played a pivotal role in history: “The history of long-distance trade in medieval and early modem Europe is the story of sequentially more complex organization that eventually led to the ‘Rise of the Western World’. In order to capture the gains associated with geographic specialization, a system had to be established that lowered information costs and provided for the enforcement of agreements across space and time” (Milgrom, North, and Weingast 1990,4). The participants at the Bretton Woods conference incorporated all three solutions to the cooperation problem in the creation of the new international monetary system. The par value system was rules based, and institutions were created to facilitate compliance with the rules, provide a forum for further negotiations, and establish a centralized information system.

7.2 Postwar Goals and the Bretton Woods Agreement With the breakdown of gold standard arrangements in the early 1930s, countries engaged in numerous policy actions that led to exchange rate instability. Countries resorted to competitive depreciation, exchange controls, and tariff warfarel’ in unilateral efforts to emerge from the world depression at the expense of neighboring countries. In a classic study, the League of Nations (1944) warned of the self-defeating nature of these beggar-thy-neighbor policies. It was against this background that leaders in various countries recognized the need for a new approach to international cooperation. 7.2.1

Goals

Seven hundred thirty participants from forty-five countries met on 1 July 1944 in Bretton Woods, New Hampshire, to create a new international monetary system.I2 While the United States was the official host, the conference was the culmination of the efforts of two men: John Maynard Keynes and Harry Dexter White.13 Keynes and White both began circulating proposals for a new international monetary system domestically as early as 1941. Both men 11. For example, the United States passed the Smoot-Hawley Tariff in 1930. 12. F’rior to the main conference, technicians from seventeen of these countries met for a preliminary drafting conference in Atlantic City, N.J., in June 1944. 13. An excellent detailed account of the prehistory of the Bretton Woods conference and the negotiations that took place at the conference is contained in Horsefield (1969).

365

The Role of International Organizations in the Bretton Woods System

believed that the economic stresses that countries faced in the interwar years contributed to the start of World War 11. While significantly different in detail, both proposals included a return to a modified gold standard exchange rate system and the creation of international organizations that would facilitate cooperation among member countries. One of the main concerns of the United States at the time centered around the growth of preferential trading systems from which its exports were excluded. The most important of these arrangements was the imperial preference system. The trading blocs served to divert trade away from countries outside the bloc using a combination of differential tariffs and exchange controls. It was this latter policy that White’s plan was most bent on eliminating. The White plan centered on the creation of two organizations, an international stabilization fund and a bank for reconstruction and development. The Fund’s roles included promoting currency stability, encouraging capital flows, and facilitating international settlements. The stabilization Fund was to be contributory, with total resources of $5 billion (the U.S. contribution was to be $2 billion). While the U.S. plan gave the Fund limited resources, it granted it substantial decision-making power. Most important, the Fund was to have veto power over a country’s decision to change its exchange rate. Just as the American plan focused on the United States’s main economic concerns, the British plan focused on the United Kingdom’s main economic concerns: unemployment and the convertibility of sterling. In 1944, the U.K. economy was in disarray, and sterling balances were large relative to Britain’s gold reserve. The key component of the Keynes plan was the International Clearing Union (ICU), a bank for central banks with its own international currency called bancor. Keynes’s ICU resembled the British overdraft system. Debit balances in this system took the form of overdrafts rather than loans. In this way, the burden of balance of payments adjustments would rest with creditor countries, who were required to accept bancor as payment for net exports. Keynes’s plan was both more tolerant of exchange controls and potentially more expensive for creditor countries. The U.S. burden, for example, could in principle exceed $20 billion dollars, the total drawing rights of the other member countries. The final Bretton Woods Agreement was a compromise between the two plans, with a more limited financial commitment from the creditor countries than the Keynes plan and a more tolerant view toward exchange rate management than the White plan. The compromise plan established two international organizations, the IMF and the World Bank. Both organizations required contributions from members based on their relative economic resources. Although exchange rates were to be fixed in value against gold, member countries were permitted to adjust the values of their currencies under certain conditions. Also, exchange controls were allowed on capital transactions but not normally on current transactions.

366

Kathryn M. Dominguez

7.2.2 The Exchange Rate Arrangement One of the principal duties given the IMF in the Bretton Woods Agreement was the promotion of exchange rate stability. The conference participants set out the rules for establishing and maintaining the new par value system principally in Articles IV and XX of the IMF charter. IMF members that were not occupied by the enemy during World War I1 were obliged to establish par values, expressed in terms of either gold or the U. S . dollar, within thirty days of the official commencement of the Bretton Woods system. All current account exchange transactions were to be made within 1% bands of the established par values. The rules did not permit members to change par values (other than a one-time change of lo%), except to correct a fundamental balance of payments disequilibrium and only after consultation with the IMF. Moreover, if a member changed the par value of its currency over the objections of the IMF, then that member would be ineligible to use IMF resources. One of the more heated debates among the architects of the Bretton Woods system was over the scarce currency clause in Article VII of the IMF charter. The British and other European countries were concerned about the possibility of a postwar depression. They argued that this could lead to a circumstance in which the total amount that countries were in deficit to the Fund exceeded the amount of available credit. As a consequence, the IMF might not have the resources to provide adequate financial assistance, even though countries were following the rules of the game. The scarce currency clause effectively allows the Fund in this circumstance to put pressure on surplus countries. Once the Fund declares a surplus country’s currency to be scarce, debtor countries have the right to discriminate against transactions in the scarce currency. Under Article VIII of the IMF charter, countries that have declared par Values are required to make their currencies convertible for current account transactions. Article XIV, however, provided countries a convertibility escape clause. This article allowed countries to maintain existing exchange controls for an initial three-year transition period after the establishment of the Fund and thereafter with the provision that they justify their position to the Fund. 7.2.3 Creation of International Organizations Of the two organizations created at Bretton Woods, the IMF was the most important in terms of member country day-to-day operations. The World Bank was designed chiefly to supply the capital needed for postwar reconstruction and long-term development projects. Although the two institutions are explicitly separate in terms of charter, funding, and staff, membership in the IMF is a prerequisite for membership in the World Bank. The main features of the IMF are similar to those in White’s original plan. Each member of the IMF has a quota equal to its subscription to the Fund. The original quotas totaled $8.8 billion with a U.S. contribution of $2.75 billion. A member’s quota determines its financial contribution, its voting

367

The Role of International Organizations in the Bretton Woods System

power in the IMF (based on one vote for each $lOO,OOO of quota), and its access to the financial resources of the IMF. l4 Subscriptions called for by quotas represent the principal source of assets for the IMF. Members were required to pay 25% of their quota in gold and the rest in their domestic currency. The IMF also derives income from charges on member drawings and has the authority to borrow to augment quota resources when necessary. As described earlier, the establishment of an initial par value is, by Article XX, Section 4(c), a prerequisite to the use of the IMF’s resources:I5“The rules governing access to the use of Fund’s resources apply uniformly to all members. However, Article V, Section 12(f) (ii) and (iii), allows the Fund to make balance of payments assistance available on special terms to ‘developing members’ in difficult circumstances” (Chandavarkar 1984, 3 1). Balance of payments assistance takes the legal form of a purchase or drawing (not a loan) of a strong currency (or SDRs)16for the members’ own currency. Member drawings fall into four categories. Drawings up to the first 25% of a country’s quota are in the gold tranche. The next three categories are called credit tranches. Transactions in the first credit tranche bring the IMF’s holdings of a member’s currency above 100% but not above 125% of its quota. Drawings in the second, third, and fourth credit tranches require substantial justification and typically involve conditionality, terms and conditions to guarantee that the country is able to repurchase its currency in a timely fashion. Any drawing or standby arrangement exceeding 25% of the member’s quota within any twelve-month period (unless the IMF holds less of the member’s currency than 75% of the quota) and any cumulative drawing that exceeds 200% of a member’s quota require a waiver.” Standby arrangements were first introduced in 1952. A standby arrangement involves the IMF granting financial assistance to members in advance of difficulties: “Indeed, a stand-by arrangement presents the contradiction that the drawing country does not have to establish need at the time the arrangement is entered into and the Fund in effect waives any power to judge need at the time the drawing is made” (Dam 1982, 122). While standby arrangements do not involve justification by the member country, they typically involve commitments to performance criteria (see Gold 1970). The second organization created at the Bretton Woods conference was the International Bank for Reconstruction and Development. It was designed to help finance investment projects for reconstruction and development, particularly in underdeveloped regions. Along with providing long-term loans and 14. For a detailed description of the quota system, see Altman (1956). 15. Exceptions to this were possible for members whose metropolitan territories had been occupied by the enemy (Article XX, Section 4[d]). 16. Special drawing rights (SDRs) were introduced in 1969 to supplement reserves. 17. In the 1960s, the IMF established a number of compensatory financing facilities that granted automatic waivers of the 200%rule.

368

Kathryn M. Dominguez

technical assistance, the World Bank was to promote private foreign investment by guaranteeing and participating in loans by private investors. Member countries’ subscriptions in the World Bank take the form of shares of capital stock. Twenty percent of subscribed capital is paid in; of this, 2% is in the form of gold or U.S. dollars, and 18% is paid to the Bank in each member’s domestic currency (and cannot be used for loans without the consent of the member whose currency is to be lent). The remaining 80% of the World Bank’s subscribed capital is subject to call by the Bank only when required to meet its own obligations on its borrowings or guarantees. The drafters of the Bank’s Articles of Agreement were intent on avoiding the perceived capital market failures of the interwar period. To that end, the Bank’s charter contains a number of protective provisions governing loans to be made or guaranteed by the Bank. World Bank loans “must be for productive purposes and, except in special circumstances, must be to finance the foreign exchange requirements of specific projects of reconstruction or development” (IBRD 1954, 7). The borrower need not be a member government, but the loan must be guaranteed by the member government in the country where the project is located. The borrower must be in a position to repay the loan, and the Bank is required to make arrangements to ensure that the loan is used for its original purpose. Finally, “the Bank must be satisfied, before making or guaranteeing any loan, that in the prevailing market condition the borrower would be unable to obtain the loan from private sources under reasonable conditions” (IBRD 1954, 7). The division of labor between the IMF and the World Bank has always been somewhat blurred. In the first complete draft of his plan, White stated, “The objectives of the Bank, it will be noted, are similar in some respects to those of the Fund, but a careful examination will reveal that in their most important aspects they are different” (Oliver 1975, 297; cited in Feinberg 1988, 546). The Fund was to provide short-term balance of payments assistance, while the Bank was to provide longer-term project assistance. Keynes originally advocated close financial collaboration between the two organizations, but later changed his tune. In an often-quoted passage, Keynes stated, “I should like to see the Board of the Fund composed of cautious bankers, and the Board of the Bank of imaginative expansionists” (Moggridge 1980, 194; cited in Feinberg 1988,547). Resolution VII at the Bretton Woods conference recommended the creation of a third organization whose purpose was to promote cooperation in international trading arrangements. Preparatory discussion in 1946 and 1947 led to the Havana conference, which produced a charter for the International Trade Organization (ITO). The United States had pushed for a powerful IT0 that would work to abolish tariffs, quotas, and preferential trading arrangements. Most of the other countries were concerned about safeguarding their weaker national economies against U.S. export competition. After extensive negotia-

369

The Role of International Organizationsin the Bretton Woods System

tions, the original objectives of the I T 0 proposals were only nominally maintained. The combination of an equal vote for every country and escape clauses basically left every country to do as it liked. In this form, the charter was unacceptable to the United States, and the Senate failed to ratify it.18 During the deliberations on the ITO, a multilateral trade agreement known as the GATT was drafted as a stop-gap measure. Twenty-three countries signed the GATT in 1947, but it was not until the mid-1950s that the GATT officially became a permanent international organization. The GATT’s mission is to set and regulate a code of conduct for international trade. The GATT is founded on three principles: nondiscrimination among trading partners (the most-favored-nation clause), no export subsidies or quantitative restrictions, and offsetting reductions in old tariffs to compensate for any introduction of new tariffs. It was hoped that, if countries complied with these three principles, then at the very least trade restrictions between countries would not increase. The GATT provides a rules-based cooperative solution for trade disputes among countries. In the context of GATT trade rounds, countries agree to provide tariff concessions as long as all other countries also do so. If a country raises its tariff above the agreed level or imposes a trade restriction on the product, then the other countries retaliate with a “compensatory suspension of concessions.” The GATT was not equipped with either carrots or sticks to compel its members to honor negotiated trade agreements: “A violation of the General Agreement-for example, the nullification of a concession-leads not to the imposition of punitive measures, but rather to the creation of a mere right on the part of the injured contracting parties to withdraw concessions (or other GATT obligations) running to the offending contracting party” (Dam 1970, 21). While the IMF and the World Bank can refuse members access to resources if they deviate from the rules, all that the GATT can effectively do is allow its members to play a tit-for-tat strategy. The architects of the postwar international order recognized that, if they did not coordinate rules on trade and exchange restrictions, countries could easily use one as an escape valve from rules against restrictions in the other. In order to avoid this possibility, Article XV:4 of the GATT provides that contracting parties shall not, by exchange action, frustrate the intent of the provisions of the GATT or, by trade action, frustrate the intent of the provisions of the Articles of Agreement of the IMF. The IMF for its part agreed to provide the GATT with both relevant statistical information on member country exchange restrictions and analysis of the causes and effects of import restrictions maintained for balance of payments reasons. l9 18. Of the fifty-six nations represented in Havana, fifty-three signed the IT0 charter, but only one nation subsequently ratified it. 19. For further discussion of the relation between the IMF and the GATT, see de Vries and Horsefield (1969, chap. 16) and Dam (1970, chap. ).

370

Kathryn M. Dominguez

7.3 Maintaining Cooperation-the Role of Financial Assistance Whereas the IMF and the World Bank both provide member countries financial incentives to maintain cooperation, the GATT does not. Do carrots provide an effective commitment mechanism for organizations that hope to promote international cooperation? In this section, I analyze two games that illustrate under what conditions financial assistance can play an effective role. 7.3.1 The Role of Financial Assistance in International Negotiation

The most general mandated function of each of the postwar institutions was to promote cooperation by providing a forum in which members can consult and negotiate with each other on international monetary and trade matters. To investigate the role of financial assistance in international negotiations, an example is useful. Consider first an example of domestic tax policy negotiations between opposing groups within the home government. Assume that both groups agree that taxes should be raised to eliminate an existing government budget deficit but that the groups disagree on the allocation of the total tax burden to be borne by each group. Assume further that, if the two groups cannot negotiate a compromise, the government will be forced to rely on an inflation tax to finance the deficit and that this will, in turn, force the country to devalue its currency. Alesina and Drazen (1991) characterize the process leading to the compromise as a “war of attrition.” Each group believes that, the longer it refuses to compromise with the other, the more likely it is that the other group will concede. Whichever group concedes first ends up bearing a disproportionate share of the tax increase. Even though the country as a whole benefits from any compromise, Alesina and Drazen suggest that compromises will often be delayed as groups attempt to shift the allocation of burdens in their favor. One possible solution to this problem is for the executive branch of the government to take a more active role in the negotiating process. For example, the executive might declare that it will impose a penalty on whichever group refuses to support a government-sponsored compromise. Assuming that the executive is in office for a finite term, the problem with this solution is that, as long as the executive’s objective is to maximize the country’s welfare (defined as the unweighted sum of the welfares of both groups), it will never have an incentive to follow through with its threat. This is the classic problem of time consistency, that, once a compromise is reached, it is not in the executive’s interest to punish either group. If the groups are rational, they will foresee this and disregard the executive’s threat. Alternatively, an outside party or organization, one that is not subject to the time inconsistency problem that the domestic government faces, can play an important role. Assume both that the IMF has a longer time horizon than the

371

The Role of International Organizations in the Bretton Woods System

domestic governmentz0and that the IMF’s objective is to maximize global welfare. During the period that the Bretton Woods system was in place, any set of domestic policies that would lead a country to devalue would reduce global welfare from the IMF’s perspective. It is in the IMF’s interest in this circumstance to provide the executive in the home government with an incentive to follow through with its threat. The IMF can offer the government financial assistance if it achieves a compromise and staves off a currency devaluation. As long as the value of the financial assistance exceeds the value of the penalty imposed on the group that refuses to compromise, the executive’s threat will be credible. Governments face similar credibility problems in the context of trade policy. Although a country can be made better off with less restrictive trade policies, certain import-competing industries have incentives to pressure members of the legislature not to agree to trade concessions. Any threats to penalize the protectionist groups will not be credible unless they are time consistent. But, without the ability to provide financial assistance, the GATT can do little to help governments forge a trade policy compromise. 7.3.2 The Role of Financial Assistance in the Par Value System A second function primarily of the IMF, but also to some extent of the World Bank, is to provide members with financial assistance to facilitate their compliance with the par value system. A modified version of the devaluation game described in section 7.1 can clarify the potential role of financial assistance in an adjustable peg exchange rate system. The two countries, home (H) and foreign (F),now move sequentially, the home country making the first move (in period 1) and moving in all subsequent odd-numbered periods. Likewise, the foreign country moves in all even-numbered periods. This assumption allows us to examine countries’ immediate reactions to each other’s moves.21 Let eH (eF)be the price of one ounce of gold in units of the home (foreign) country’s currency. To place the model in the Bretton Woods system context, the countries now have three policy options during a move: (i) a large devaluation against gold (Ae = M); (ii) a small devaluation against gold (Ae = m); and (iii) no devaluation (Ae = 0). Assume that a large unilateral devaluation produces a large trade surplus for the devaluing country (a = 1). A small devaluation produces a smaller trade surplus (a < 1). Assume initially that countries cannot borrow money. The payoffs for each country in any given period (regardless of whose turn it is to move) will be P, - C if either devalues and P, if neither does. Both 20. In other words, assume that the government has a higher rate of discount than does the IMF. 21. For an analogous alternating-move infinite horizon model setup, see Maskin and Tirole (1988).

372

Kathryn M. Dominguez

countries incur a devaluation cost in period t if a country devalues in that period. Here international monetary stability is considered a public good so that the cost of its disintegration is borne by both countries (see Eichengreen 1987). Let = P’; - 1,C be the total payoff to country j ( j = H, F) in period t, in which Z,is an indicator function equal to one if a country devalues in period t and zero otherwise, and let 4 be the price of one ounce of gold in units of country j ’ s currency in period t. The home country will receive one of five possible payoffs in each period, depending on its own actions and those of the foreign country:

(4)

- 1 if - a if 0 if

1 if

- 1. ln(1 r) 1

(9)

+

-

+

Define v* to be the smallest integer value of v that satisfies the inequality. This tells us that, after time t = max[2v* - 1, 01, the IMF will not provide full 25. Here we assume away issues of moral hazard. The IMF is assumed to be able to distinguish between countries that are truly experiencing balance of payments difficulties and those that claim that they are, when they are not, in order to receive IMF permission to devalue.

375

The Role of International Organizations in the Bretton Woods System

financing of F's deficits. The discounted payoff to the home country for following the deviation strategy (D) in its first n* turns is =HD

=

-6(1 - 82"')C I 0 = 1 - 6

+'.

More generally, at any given odd period t = 2v - 1 < 2v* - 1, an additional deviation by the home country, followed by subsequent adherence to F, yields a negative gain of -6C. Therefore, any finite number of deviations, such that 0 < n* Iv*, does not pay. However, once the home country devalues the n* 2 v* 1 time, H will force the foreign country beyond its borrowing limit. If the home country deviates for v* 1 periods, its discounted payoff is

+

+

This indicates that, if the cost of devaluations is small enough, v* + 1 deviations will pay. That is, from period 2v* + 1 onward, the game degenerates into either the CD or the s strategy game, although the foreign country is formally following the IMF strategy, F. Next consider the foreign country's incentives to deviate from F. Assume that a number n (0 < n < v*) deviations have occurred in the past and that it is the foreign country's turn to move. Under F, the foreign country must devalue in such a way as to obtain a surplus B < 1, yielding payoffs - 6C. If the foreign country deviates from F-by, for example, engineering a surplus equal to one, then it loses its right to borrow from the IMF in the future. Such a devaluation would also induce the home country to devalue in the following period. This indicates that, if the foreign country deviates from F, the game collapses into the CD game previously described. These results are surprising in that they indicate that the promise of IMF financial assistance provides countries no more incentive to maintain fixed exchange rates than does the s strategy, which they can achieve without IMF participation. Once a country hits its borrowing limit, the necessary condition for F to lead to a subgame perfect equilibrium is 1 - B 2 C 2 (1 - 6)/ (1 6 ) = r/(2 r), which is identical to what we found earlier for It is only under the strict condition that countries never hit their borrowing limit that the F strategy provides a subgame perfect equilibrium when s does Much of the debate at the Bretton Woods conference between Keynes and White was over the level of financial resources that would be made available

+

+

26. For both the CD and the s strategies, a is an arbitrary constant, such that a 5 1 . Under the F strategy, 6 is the maximum surplus a country can obtain when trying to restore balance. That is, 6 is the surplus that the foreign country needs to restore long-term balance after the home country has deviated n* times. 27. In this case, the necessary condition for F to be a subgame perfect equilibrium is that the countries' discount rates are equal to the IMF's interest rate: 6-I = 1 + r, r > 0.

376

Kathryn M. Dominguez

to member countries in need of assistance. White wanted to limit country access to resources because he felt that the United States would ultimately bear the bulk of the financial burden of a more generous system. According to Edward Bernstein, the chief technical adviser and spokesman for the U.S. delegation at Bretton Woods, under the White plan “the Fund would give each country the currency it needed to meet its deficit and the country would give its currency to the Fund. But then it would be obligated to repay the money it drew and it had to begin to correct its balance of payments. In the Keynes Plan there was no obligation to repay unless it developed a balance of payments surplus” (Black 1991, 37-38). In the context of the model presented in this section, under the Keynes plan countries would be less likely to hit their borrowing limit and, as a consequence, would have a greater incentive to stick to the rules of the game.

7.4 Empirical Evidence from 1944-71 The effectiveness of institutions for deterring breach of contract might best be judged like that for peacetime armies-by how little they are used. (Greif, Milgrom, and Weingast 1991, 1)

Each of the postwar institutions has come in for criticism, not so much because they have been little used, but because international monetary and trade relations did not achieve the level of cooperation that was first promised. An empirical assessment of this proposition, however, is difficult. First, it is difficult to formulate a testable hypothesis. What one might like to know is how countries would have behaved had the IMF, the World Bank, and the GATT not existed. If economic conditions were comparable before and after World War 11, one could examine the different levels of cooperation achieved between countries in the two periods. Alternatively, had a significant number of countries not become members of the Bretton Woods institutions, one might compare economic growth and cooperative arrangements across the two groups of countries. But most of the developed countries (with the notable exceptions of Switzerland and the former Soviet Union) and the majority of developing countries are members, precluding any such comparison. A second problem that arises with any quantitative study of the effectiveness of these institutions is that much of the requisite data remains confidential. This is particularly a problem for an assessment of the effectiveness of international negotiations within each of the organizations. For example, country requests for par value changes or drawings that were not approved by the IMF are not necessarily part of the public record. In light of these problems, this section presents available empirical evidence on the main accomplishments of each of the three postwar institutions relative to their original missions. I begin with brief summaries of the activities of the World Bank and the GAlT. There is less to say about these two

377

The Role of International Organizations in the Bretton Woods System

institutions because the goals of both are intrinsically open ended relative to those of the IMF. The empirical assessment of the IMF is provided in two parts. The IMF’s role in the par value system is presented in section 7.4.1, and section 7.4.2 presents the IMF’s record for balance of payments assistance. The World Bank’s original mission was to provide capital for European reconstruction. The World Bank’s resources were from the beginning limited, but the intention was that the Bank would encourage private investment by providing loan guarantees. The president of the World Bank in 1947, John McCloy, “thought of the Bank as a temporary institution which would go out of business if it were successful, for it would no longer be needed as an intermediary between productive borrowers and private lenders” (Oliver 1975, 259). It was soon realized that the task of reconstruction was beyond the Bank’s scope, and the Marshall Plan, implemented in 1948, largely took over the job.**The Bank then shifted its resources and focus to financing development projects in underdeveloped regions. Rather than serving as a guarantor of private investment, as the Bretton Woods participants had envisioned, the Bank took on an intermediary role, borrowing funds from private investors and lending them to developing countries. The average yearly World Bank loan level for the period 1947-57 was only $283 million. Average project lending in the next ten years increased substantially to $764 million. But it was not until the late 1960s, after Robert McNamara became president, that lending steadily began to increase. In 1978 alone, World Bank loans exceeded $6 billion.29 The Bank’s slow start was in part due to its passive approach toward development lending. In 1950, the president of the Bank, Eugene Black, explained that the reason that the Bank had not made many loans to developing countries “has not been lack of money but the lack of well-prepared and well-planned projects ready for immediate execution” (Mikesell 1972, 72). Applications for development loans reportedly consisted of lists of projects that the member governments had under review with no indication of priorities or feasibility. It was against this background that the Bank organized its first economic survey mission to Colombia in 1949. Although these missions initially received mixed receptions, the World Bank increasingly took the view that it needed to assist countries in formulating long-term development programs. Robert Garner, vice president of the Bank in 1947, reflected later that “advice was more important than money” (Oliver 1975, 255). It is difficult to give an overall assessment of the Bank’s record. After a shaky start, the Bank greatly expanded its loan portfolio and actively involved itself in analyzing and centralizing information on member country develop28. In 1947, the Bank did make four reconstruction loans: to France ($250 million), the Netherlands ($195 million), Denmark ($40 million), and Luxembourg ($12 million). 29. World Bank loan data are from various issues of World Bank Annual Reports.

378

Kathryn M. Dominguez

ment programs. However, even given its impressive and improving recent record, the Bank’s resources are meager compared to the financing needs of developing countries. The ITO’s mission, taken up by the GATT, was to facilitate an open, liberal, and competitive international trading system. To the GATT’s credit, postwar trade restrictions have declined substantially, but cooperation in trade has not been uniform across countries or industries and has progressed slowly. The GATT’s principles are subject to numerous exceptions. Customs unions, free-trade areas, and certain preferential trade arrangements are all excluded from the principle of nondiscrimination. Agricultural subsidies are excluded, as are import quotas for developing countries. The GAIT has been most successful in its role as facilitator of multilateral trade negotiations among developed countries. As a result of six major rounds of negotiations in the first twenty years of the GATT, tariff rates of the industrial countries fell from an average of 40% in 1947 to 13% by the late 1960s. There has been less success with liberalizing developing country trade policies. Most recently, the GATT has come in for criticism over the relevance of its rules of the game. While the GATT continues to focus on reducing tariffs worldwide, countries have found an effective escape valve by creating socalled nontariff barriers. Countries have learned to replace tariffs with other forms of trade restriction that are not subject to GATT rules. The IMF’s two original responsibilities were to administer the par value system and to provide members financial assistance to enable them to maintain their par values in the face of short-term balance of payments shortfalls. The next two subsections present an empirical assessment of the IMF’s record in these two areas. 7.4.1

The Par Value System

The IMF elected to treat an exchange rate change as unauthorized on only one occasion, France in 1948. “Indeed, the Fund on some occasions did not even require a member to assert that it was in ‘fundamental disequilibrium’ when passing upon a proposed change in par values. As time went on and it became apparent that a key problem under Bretton Woods was not the instability that White had feared but rather an unwillingness of members to make par value changes promptly enough, considerable effort was expended on making it clear that the ‘fundamental disequilibrium’ requirement was not really a limitation on prompt and small exchange rate changes” (Dam 1982,92). Initially, the IMF was determined that new members declare par values within the thirty-day period specified in Article XX 4(a). But, over time, the IMF became less resolute on this as well as some of the other exchange rate rules set out in the Articles. In a series of decisions over the course of 19475 1, the Executive Board declared that any change in a member’s exchange

379

The Role of International Organizations in the Bretton Woods System

rates (whether or not it involved a change in the member’s par value) was subject to review by the IMF. But on numerous occasions countries changed both exchange rates and par values without prior approval of the IME30 There were forty-four proposed exchange rate changes between 1948 and 1949. In 1948, both France and Mexico suspended their par values without IMF approval and allowed their currencies to float. In 1949, both Belgium and Peru were granted permission by the IMF to allow their currencies to float temporarily. While Belgium declared a new par value two days later, Peru had yet to declare a new par value in 1965. Even without a declared par value, however, Peru was permitted to make IMF drawings. During 1949, the Bretton Woods system experienced its first major round of devaluations. During the summer of 1949, gold and dollar reserves of the sterling area fell by over 30%. In September, the U.K. government finally asked for approval for a 30% devaluation; this was immediately approved by the IMF. The devaluation of the pound sterling was followed by major devaluations by the other sterling bloc countries as well as by all the Western European c~untries.~’ Table 7.1 shows the extent of these devaluations; in all, nineteen countries devalued in 1949. While there was some grumbling over the perception that the IMF had really just rubber-stamped the U.K. request to devalue, there was more serious discussion within the IMF over the approval of the subsequent devaluation^:^^ “The Latin American Directors, exercised lest the devaluations of the outer sterling area in line with that of the pound sterling should harm the export prospects of their countries, thought that there was a need for a ‘definitional examination’ of competitive depreciation” (de Vries and Horsefield 1969, 100). The countries that opposed the IMF’s decision to approve the devaluations had reason to feel that this was an important issue. The countries that devalued in 1949 accounted for almost half the world’s exports and about 60% of the exports of industrial countries. In the years to follow, the IMF became increasingly tolerant of member countries’ refusals to play by the rules of the par value system. In 1950, Canada informed the IMF of its decision to allow its currency to float because of a heavy capital inflow (mainly from the United States during the Korean War). After debating the issue at length, the IMF made no official pronouncement. 30. For detailed descriptions of member country exchange rate policies during 1945-65, see de Vries and Horsefield ( 1969). 31. Obstfeld (chap. 4 in this volume) refers to the 1949 devaluations as competitive-the very policies that the Bretton Woods system was set up to avoid. An alternative interpretation is given by Edward Bernstein: “In the environment of 1949, when European recovery had been only partially achieved, it was impossible to make fine distinctions between the appropriate change in the parity of the Netherlands guilder, for example, and the parity of sterling. That is why the European devaluations in 1949 were nearly the same” (Bernstein 1972, 53). 32. In Bernstein’s recollections, he states, “Early in 1949, the U.S. Executive Director began to press for discussions on the devaluation of the European currencies” (Black 1991, 66). This may explain why the U.K. devaluation proposal was accepted by the Fund so quickly. Bernstein goes on to say that the U.S. view was not made public because of concern that this would lead to a speculative attack against the European currencies.

380

Kathryn M. Dominguez

Table 7.1

Devaluations by Countries in the Bretton Woods System, 1949-50

Country Greece Denmark Egypt Finland Netherlands Norway Sterling area except Pakistana

Devaluation Relative to $ (%)

Country

Devaluation Relative to $ (%)

33 30 30 30 30 30 30

Sweden France Germany Belgium Portugal Canada Italy

30 22 20 13 13 9 8

Canada’s exchange rate floated for twelve years, yet the country was not denied access to IMF resources.33 Ten years after the establishment of the par value system, only nine countries had accepted the obligations of Article VIII and had fully convertible currencies.” With the exceptions of the United States and Canada, all the developed member countries took advantage of the convertibility escape clause provided in Article XIV. Not only did they avail themselves of the automatic three-year transition period, but all the European currencies remained inconvertible for twelve years. Although most currencies were de facto convertible by 1959, the European countries did not officially assume Article VIII status until February 1961. Although a number of developing countries accepted Article VIII status before the developed members, most of them experienced chronic balance of payments problems throughout the Bretton Woods era. As a result of these difficulties, many developing countries became increasingly dependent on a wide range of exchange and trade restrictions. By the mid-l960s, “while the industrial countries were able to maintain their external economic relations with few limitations on the acquisition or use of foreign exchange, many of the developing countries continued to rely on restrictions, sometimes in combination with multiple exchange rates” (de Vries and Horsefield 1969, 294). By the end of 1966, the IMF had 104 members, twenty-three of whom had not established par values. After twenty years, only 58% of the member countries maintained 1% bands around fixed par values. However, table 7.2 shows that, if we include countries that maintained fixed unitary exchange rates, although without a par value, and countries with fluctuating rates that re33. However, in practice, Canada did not draw on Fund resources until 1962, the year it reestablished a par value. 34. The nine countries with Article VIII status in 1956 included Canada (1952), the Dominican Republic (1953), El Salvador (1946), Guatemala (1947). Haiti (1953), Honduras (1950), Mexico (1946), Panama (1946), and the United States (1946).

The Role of International Organizations in the Bretton Woods System

381 lsble 7.2

Adherence to Par Values, 1946-66 ~

Without Par Values

With Par Values

% Effectively

Year

All Members

1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966

40 45 48 48 49 50 54 55 56 58 60 64 68 68 68 74 82 102 102 103 104

8 8 8 8 9 7 10 6 7 9 11 13 15 12 9 13

21 23 23 24 24 25 20 22 20 22 24 27 31 33 37 40 47 54 54 57 64

70 67 63 67 61 62 57 62 59 59 57 63 63 68 72 80 84 85 84 86 87

17

30 30 28 23

Stable‘

Source: de Vries (1966, 506-7). ‘Includes countries with minor multiple currency practices in addition to par values and countries without par values but with fixed or stable unitary rates.

mained stable for at least three years, then the percentage of countries with effectively stable rates in 1966 rises to 87%. The par value system came under a new set of strains in the late 1960s and finally collapsed in the early 1970s: “This was due to rapid changes in competitive positions among the major industrial countries, reflecting divergent rates of productivity, growth in favor of continental Europe and Japan, to widening disparities in rates of inflation, and to the reluctance of many countries to make timely and adequate exchange rate changes” (Hooke 1982, 5). With the exceptions of devaluations by France in 1958 and 1969, revaluations by Germany in 1961 and 1969 and the Netherlands in 1961, and another sterling bloc devaluation in 1967, members of the developed countries made little use of exchange rate policy between 1950 and 1970.35The reverse was true for members from developing countries. Eighty-two percent of developing country members devalued by more than 30% between 1949 and 1966, and 25% 35. For detailed descriptions of member country exchange rate policies during 1966-71, see de Vries (1976).

382

Kathryn M. Dominguez

of these devalued by more than 75% (see de Vries 1968). In 1970, Canada was the first country to allow its currency to float, but it was soon followed by Germany in 1971 and the United Kingdom in 1972. In March 1973, the par value system officially collapsed following the U.S. announcement that it would devalue the dollar by 10%. The empirical evidence on the par value system indicates that the rules of the system were rarely enforced and that the goal of the architects of the system, stable exchange rates, was achieved only by a small number of countries for a short time period. During the so-called heyday of the Bretton Woods era, 1959-67, most developed countries did maintain stable and convertible exchange rates. However, few developing countries were able to maintain stable rates without the help of exchange and trade restrictions.

7.4.2 Member Drawings The participants at Bretton Woods had originally envisioned use of the IMF’s resources as a privilege granted to members that were otherwise in compliance with the IMF’s rules and in need of short-term balance of payments a ~ s i s t a n c e But, . ~ ~ as was the case with the par value system, the IMF took on an increasingly broad definition of member eligibility for Fund resources. Uruguay was the only original member that did not have the right to purchase exchange from the IMF because it had not agreed to a par value. Once Uruguay established a par value in 1960, however, the IMF decided to permit member countries to draw on Fund resources even if they had not established a par value. France was the first country to be declared ineligible to use IMF resources because of noncompliance with the par value system. In 1948, France introduced multiple currency practices that were not approved by the IMF. The only other country to be denied access to IMF resources in response to a par value violation was Czechoslovakia in 1953. Czechoslovakia eventually left the IMF in 1955, partly over this issue. Bolivia and Cuba were each denied access to IMF resources, in 1958 and 1964, respectively, for noncompliance with the conditions of earlier drawings. Bolivia was unable to carry out the conditions of its standby arrangement, and Cuba failed to repurchase its 1958 drawing within five years. Cuba, however, resigned from the IMF before the ban on future drawings was put in place. Like the World Bank, the IMF got off to a slow start with its lending practices: “There was a sharp division in the Board between those who believed that members had automatic rights to draw on the Fund’s resources and others . . . [who] took the line that access to the Fund’s holdings of dollars should be made subject to fairly strict conditions. This atmosphere resulted in many 36. Keynes and White had a difference of opinion on this point. Keynes felt that members should, at the very least, have an indisputable right to draw from the gold tranche.

383

The Role of International Organizations in the Bretton Woods System

decisions by the Board which in effect tied the Fund’s hands in its initial years: there were to be few transactions, no participation in European payments arrangements, and little action against restrictions” (de Vries and Horsefield 1969, 32). Indeed, in 1949, only $102 million was drawn from the Fund. Nothing was drawn in 1950. In the years 1951-55, borrowings averaged less than $100 million per year. Likewise, only five countries have ever been officially denied access to IMF r e ~ o u r c e s . ~ ~ In 1952, the Board decided to grant countries the unconditional right to draw up to 25% of their quota (the gold tranche) as a means of encouraging members to use IMF resources. However, the maximum amount that countries were eligible to draw from the IMF in any one-year period was 25% and not more than 200% of their quota in total. In 1953, in yet another effort to encourage more countries to make drawings, the IMF began routinely issuing waivers of the 25% rule.38 IMF drawings beyond the gold tranche are subject to conditionality. Members are required to pursue specific economic policies in order to receive Fund resources. These policies vary from case to case but typically include ceilings on domestic credit and public-sector expenditures, elimination (or reduction) of restrictions on trade and payments, and elimination of price controls. Conditionality is intended to help countries attain viable balance of payments without resort to trade or exchange restrictions. However, at least publicly, countries are rarely pleased to relinquish their policy discretion to the IMF: “vpically some-and sometimes many-of the requirements embodied in the Fund’s proposals for conditionality are difficult for the member to accept. . . . If performance clauses are not met, further drawings on the Fund automatically cease” (Polak 1991, 52). The stringency of IMF conditionality increases with the size of a member’s drawing. This may have discouraged members from making large drawings. Table 7.3 shows the amounts of member drawings over the period 1947-65. Before 1961, “no country had outstanding drawings or a stand-by arrangement with the Fund for amounts that in the aggregate exceeded 100% of its quota. The first approval for a larger amount was given in connection with a combined drawing and stand-by arrangement requested by Chile” (Mookerjee 1966, 432). The first sizable drawings in the fourth credit tranche did not occur until 1965.39 In the IMF’s first two decades of operation, drawings by industrial countries accounted for over half of total Fund credit. The share of developing country 37. For detailed descriptions of member country drawings over the period 1948-78, see de Vries and Horsefield (1969) and de Vries (1976, 1985). 38. Turkey was the first country allowed to draw over 25% of its quota over a twelve-month period starting in August 1953. Over the period 1956-65, waivers were granted in 144 of 155 standby arrangements and forty-three of seventy-fivedirect drawings. 39. This changed in the late 1970s when a number of developing countries were granted permission to draw amounts two and three times their quota. For example, in 1980, a standby arrangement for ’hrkey involved permission to draw up to 625% of its quota.

384

Kathryn M. Dominguez Amounts of Fund ’hansactions by ’hanches, 1947-65 (in millions of U.S. dollars)

Table 7.3

Year 1947 1948 1949 1950 195 1 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965

From Net Creditor Position

... ...

...

..

... ... ..

... 37.5 24.3

...

157.5

Gold Tranche

First Credit Tranche

398.8 89.9 56.7

68.9 118.1 44.8

6.6 13.9 158.5 49.9 12.5 294.6 421.2 54.6 42.6 61.7 722.5 167.6 8.4 1,140.7 462.7

28.0 43.2 62.5 12.5 12.5 369.0 452.1 55.7 64.0 106.9 869.0 214.1 15.9 514.2 69.7

...

...

Second Credit Tranche

Third Credit Tranche

Fourth Credit Tranche

... ...

... ...

... ... ... ...

28.0 8.5 2.5 20.8 103.2 211.7 69.2 104.5 785.0 93.4 112.3 60.2 777.3

... 8.1 .6 16.0 3.9 6.6 64.3 64.0 138.6 18.7 528.1

... ... .

I

.

20.2 44.8 57.5 491 .O

Source: Mookerjee (1966)

drawings did not exceed that for industrial countries until the late 1970s. Table 7.4 shows the relative magnitudes of drawings by developed and developing countries over the period 1947-78. The first large drawings on the Fund were made by the United Kingdom and France in 1948 before the Marshall Plan disbursements began. The Marshall Plan was in effect from 1948 through 1952, and, under its auspices, the United States provided $11.6 billion in grants and $1.8 billion in loans to European countries and $950 million in grants and $275 million in loans to Japan. Countries that were eligible for Marshall Plan funds were to make Fund drawings only in exceptional or unforeseen cases. But, even in this four-year period, when developed countries were essentially ineligible for Fund credit, total drawings by developing countries did not exceed $250 million. An increasing fraction of drawings after 1952 were in the form of standby arrangements. Although these arrangements were originally “considered as something in the nature of a confirmed line of credit that gave a member an absolute right to make purchases . . . [they have] become the main instrument for conditionality” (de Vries and Horsefield 1969, 533). This may explain why many of these credit arrangements were never actually used. Table 7.5 provides data on the amounts drawn under standby arrangements from 1947 to 1965. Only 23% of these arrangements were fully drawn on. This trend

Drawings on the IMF, 1947-71 (U.S. $million), 1971-78 (SDRs)

Table 7.4 Year

Developed Countries'

Other Countries

No. of Countries

1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978

43 1 132 29 0 0 34 129 0 0 56 1 540 190 50 19 1,775b 300 30 1,762c 1,885d 892 425 2,864' 2,476 1,513 1,473' 1,4169 599 533 3,22gb 4,062' 2,874 1,912k

37 76 73 0 35 51 100 62 27 131 437 148 130 260 703 284 304 188 443 556 410 689 395 326 427 613 577 525 1,874 2,530 2,036 59 1

8 11 6 0 2 6 6 3 2 11 20 14 12 14 22 18 15 22 23 34 29 37 36 41 35 30 26 24 50 57 55

31

Source: IMF, International Financial Statistics. 'Includes Industrial countries (Belgium, Canada, France, Germany, Italy, Japan, the Netherlands, Sweden, the United Kingdom, and the United States and other developed areas (Australia, Austria, Denmark, Finland, Iceland, Ireland, Israel, New Zealand, Norway, Portugal, South Africa, Spain, and Yugoslavia). b$l,500 million of which was to the United Kingdom. c$l,OOO million of which was to the United Kingdom and $525 million of which was to the United States. d$1,400 million of which was to the United Kingdom and $435 million of which was to the United States. '$1,400 million of which was to the United Kingdom and $745 million of which was to France. '$1,362 million of which was to the United States. ~$1,312million of which was to the United States. h$l,942 million of which was to Italy. '$1,700 million of which was to the United Kingdom. J$1,700 million of which was to the United Kingdom. k$l,250 million of which was to the United Kingdom.

386 Table 7.5

Kathryn M. Dominguez Amounts Drawn under Standby Arrangements, 1947-65 Amount Drawn as % of Amount Available

% of Total No. of Standby Arrangements

0

30

1-25

6 16 13 12 23

26-50 5 1-75 76-99 100

Source: Mookerjee (1966).

continued in the late 1960s and 1970s. Over time, standby agreements served less as a financial resource and more as a signal to both private banks and investors that the country had the IMF’s stamp of approval. The evidence on drawings and standby agreements confirms that the original rules of the game set at the Bretton Woods conference did not remain hard and fast. There were few instances in which countries were declared ineligible for IMF resources, even when they did not follow the rules of the par value system. There were even fewer instances of countries getting close to their credit limits. One explanation for this pattern is that the IMF used conditionality as its stick. Countries were de fact0 ineligible for Fund resources if they did not agree to pursue IMF-dictated adjustment policies. Over the Fund’s first three decades, the industrial countries were the major recipients of Fund resources and conditionality. However, since 1978, most drawings on the IMF have involved developing countries. With conditionality as the Fund’s major means of enforcing its rules, this change in the IMF’s clientele has effectively led to a two-tier membership system. Those developing countries that rely on IMF credit are subject to the rules of the game, while the developed countries are not.

7.5 The Evolution of the IMF’s Role in Maintaining Cooperation By the end of the 1960s, the IMF’s role in promoting international cooperation had fundamentally changed. The participants at the Bretton Woods conference assigned the IMF two main tasks: to enforce the rules of the par value system and to provide members with financial assistance when necessary to enable them to observe the rules. But the historical record indicates that the par value system was never widely adhered to, and it was soon to be abandoned altogether. Likewise, as the world’s private capital markets developed, developed countries no longer needed to rely on IMF resources for their financing needs. The IMF responded to the changing economic environment by turning its focus from the developed world to developing countries. But, rather than

387

The Role of International Organizations in the Bretton Woods System

competing with private capital markets for business, the IMF took on a new role as a monitor of developing country stabilization programs. Although the commercial banks might have preferred that the IMF directly provide them information about the economic prospects of countries, the confidential nature of member country information precluded such an arrangement: “Commercial bankers were not in a position to monitor the borrowing government’s implementation of economic policy measures, and they gradually came to view that the best approach was to ask the country to enter into an upper credit tranche stand-by arrangement with the Fund” (Mentre 1984,31). A simple two-period model of a country that wishes to borrow can provide insight into the potential monitoring role for an international organization. Section 7.5.1 presents the model, and section 7.5.2 presents empirical evidence.

7.5.1 A Monitoring Role for the IMF Consider a model in which a country’s population can be represented by a single agent, the government, with utility function U(CJ 6U(C,), in which U is a standard concave function, S is a discount factor (0 < S < l), and Ciis ith-period consumption. The country enters period 1 with endowment Y to be allocated between consumption in period 1 and investment in period 1. The country can borrow an amount b in period 1. In period 2, the capital invested during period 1 produces according to a concave production function, f l k ) , modified with an efficiency parameter 8. Debts must be repaid with interest accruing at rate r. If it is assumed that debt contracts are perfectly enforceable, then consumption is C, = Y + b - k in period 1 , and C, = 8flk) (1 r)b in period 2. If the government chooses k and b to maximize utility, it will follow the golden rule: ef(k*> = 1 + r. If contracts are not perfectly enforceable, then banks will need a means to punish debtors that default on their loans in period 2. It is common in the debt literature (see Bulow and Rogoff 1989, app.) to assume that banks cannot recoup anything from a sovereign country that defaults and cannot charge discriminatory interest rates. Adopting such a framework, the only recourse that banks have when a country defaults is to refuse to grant trade credit. This, in turn, causes a loss of a fraction p of the defaulting country’s output. Given such a penalty, a country that invested k and borrowed b in period 1 will repay its debt when 8fTk) - (1 r)b 2 (1 - p)eflk). Banks can foresee this choice and will not lend more than a country will repay.* This establishes a credit ceiling bc = p8fTk)/(l + r). As long as 8 is fully observable by both the country and the bank and the credit ceiling is binding,41 the country will choose k to maximize utility such that

+

+

+

40.Eaton and Gersovitz (1981) present the classic analysis of lender and borrower incentives under these conditions. 41. If the credit ceiling is not binding, the maximization problem again yields the golden rule.

388

Kathryn M. Dominguez

which says that capital is accumulated until its marginal productivity equals the effective interest rate, a weighted combination of the market rate and an implicit rate. The weights are proportional to 8, indicating that low-8 countries are likely to face higher effective interest rates than high-8 countries. If 8 is unobservable, the problem for the banks and the countries is more ~ o m p l i ca t edAssume .~~ that there are two types of countries, a proportion y of low-efficiency types (8,) and a proportion (1 - y) of high-efficiency types (Oh), with 8, > 8,. If banks can observe 8, they will offer low-8 types loans up to bf and high-8 types loans up to 4. The low-8 types will borrow all they can, and the high-8 types will borrow their optimal amount, b,* < 4. On average, a representative country therefore receives ybf (1 - y)b,* in loans. But if banks cannot observe 8, they can be sure to be repaid only on those loans that do not exceed the credit ceiling that applies to 8, types, bf.43 In this case, low4 countries are unaffected, while both the banks and high-8 countries are hurt. Banks earn fewer profits because potentially profitable loans to high-8 countries are not made. This inefficiency is difficult to avoid because l o w 4 countries have an incentive to imitate the behavior of high-8 types in order to borrow more. Define U(8,, 8,) to be the level of utility of a correctly identified low-8 type, and let U(€J,,8,) be the utility of a low-8 type that imitates a high-8 type, such that

+

(13)

4

I

+ 6; - 4) + 6 e,f(kf) - (1 + r)bf , v(e,,e,) = U(Y + b,* - k,*) + 6U{max[O,f(k,*) - (1 + r)b,*;(1 - p)B,flk,*)]).

U(e,, 8,)

=

u(Y

Visually comparing these levels of utility, we see that a low-8 type that imitates a high4 type can consume more in the first period than if it were correctly identified. This is because the consumption smoothing properties of a concave utility function assure that high-8 types consume more in both penods than low-8 types do. 42. Here the assumption is that 0 is given but unobservable and that countries have no control over efficiency. This assumption leads to a classic adverse selection problem. Countries that know that they have low 0s will try to exploit the bank’s inability to distinguish them from high4 countries. An alternative assumption made by Gertler and Rogoff (19W) is that investment (k)is unobservable. Lenders observe the total amount borrowed, realized output, and the country’s production function, but they do not observe what the borrower does with the funds. This assumption leads to a moral hazard problem. The IMF can serve an important role in the context of either problem. But if the problem is moral hazard, then the IMF rule would involve conditionality rather than monitoring. 43. Stiglitz and Weiss (1981) show that, even if banks can charge discriminatory interest rates, credit rationing will arise if borrowers are not distinguishable because the interest rate will itself affect the riskiness of loans.

389

The Role of International Organizations in the Bretton Woods System

One solution to this problem, from the standpoint of banks and high-8 countries, is for high-8 countries to signal their true type." The problem for the high-8 country is to maximize its utility subject both to the credit constraint and to the constraint that an imitating low-8 country's utility is not higher than it would be if correctly identified, U(8,, 8,) 5 U(8,, 8,). The firstorder conditions from this maximization indicate that a high-8 country is likely to borrow less when it needs to signal:

Borrowing adds to both the high-8 country's utility in period 1 and the imitator's utility because the latter depends on the actions of the country being imitated. The usual benefits from borrowing in period 1 , 88f(k)U(c,), are modified by the last term in the first-order condition. Under these circumstances, the costs of borrowing for the high-8 country will be larger the larger is €I/€),,and the lower is the loss of output to the imitator when it does not repay its loan (p). More generally, the high-8 types incur the full cost of repayment (1 r)Uf(C2,,)but do not reap the full benefits of investing because some of the benefits leak to the imitators. This model suggests that countries can benefit from a credible monitoring technology that would help banks distinguish between low-8 and high-8 types. Assume that the cost of monitoring i countries is C(i) = C + wi, where C(i) exhibits decreasing average costs or economies of scale in monitoring. This assumption reflects the fact that an organization, once set up, can monitor additional countries at small cost. Further, assume that the fixed cost C is high enough that it does not pay individual banks to incur the costs of monitoring. (Nor can high-8 types afford to pay for the costs of individual bank monitoring.) If there arej debtor countries and the IMF monitored all of them, its average monitoring cost would be Clj w. As long as C and j are large enough, high-8 countries will request the IMF to monitor them and can even offer to pay for the service. Further, as long as monitoring is credible, low-6 types (that seek to imitate high-8 types) have no incentive to go to the IMF to be monitored because they will be found out. Therefore, the very action of going to the IMF conveys all the information to the banks that is needed to distinguish correctly between the true high-8 types and imitating low-8 types.

+

+

7.5.2 Empirical Evidence The two oil crises in the 1970s were particularly damaging to the terms of trade of nonoil developing countries. While the IMF responded to these developments by introducing two temporary oil facilities in 1974 and 1975, it was the commercial banking sector that responded with substantial financial assistance. The proportion of external debt of developing countries owed to 44.For an analysis of a similar signaling problem, see Milgrom and Roberts (1982). 45. For details, see Appendix B .

390

Kathryn M. Dominguez

private banks rose dramatically over the period 1973-82. Table 7.6 presents these data, along with the relatively smaller magnitudes of developing country debt owed to governments and international institutions over the same period. Although the model described above suggests that the IMF could have helped private banks distinguish between types of developing country borrowers in this period, there is little evidence that any such certification took place. Commercial bank loans continued to be readily available for many developing countries throughout the 1970s, but banks gradually began to restrict capital flows toward certain regions. It was at this time that the IMF’s information and monitoring role began to take shape: “Both in Eastern Europe and later in Latin America, certain countries found their access to capital markets restricted, partly because the debt problems in neighboring countries had changed bankers’ assessment of their creditworthiness. In some cases, the Fund, at the request of the.debtor authorities, has been the conduit of information between the countries and their creditors, in an effort to help ensure that market sentiments be guided by more comprehensive and reliable economic information” (Brau and Williams 1983, 14). By 1978, certain heavily indebted countries began to have difficulty servicing their loans and approached both official and private creditors for debt restructurings. It was at this time that the IMF’s monitoring role became established. Both official creditors and bank creditors began to require that countries experiencing payments difficulties negotiate upper credit tranche arrangements with the IMF prior to the conclusion of their restructuring negotiations. In thirty-nine of forty-seven restructuring negotiations with commercial banks over the period 1978-83, the new terms were made conditional on an IMF arrangement (see Brau and Williams 1983, table 11, pp. 30-34). Table 7.6

External Debt of Nonoil Developing Countries, 1973-82 (in billions of US. $)

Year

Total Debt Outstanding

Government

International Institutions

Private Banks

1973 1974 1975 1976 1977 1978 1979 1980 1981 1982

130.1 160.8 190.8 228.0 278.5 336.3 391.1 467.6 550.8 614.2

37.3 43.4 50.3 57.9 67.6 79.1 89.1 101.7 113.4 125.7

13.7 16.6 20.3 24.8 31.0 38.4 45.6 53.2 62.7 71.0

60.8 77.9 95.1 114.8 137.3 169.1 199.7 229.5 275.5 300.8

Source: IMF, Annual Report (various issues). Nore: Nonoil developing countries include all Fund members except industrial countries and countries where oil exports account for at least two-thirds of total exports.

391

The Role of International Organizations in the Bretton Woods System

Upper credit tranche standby arrangements serve this monitoring role well because they typically involve substantial conditionality.46Further, the IMF disburses portions of its financial assistance over time, usually over the course of one or two years. This permits the IMF to monitor the adjustment program and potentially cancel financial support for a member that does not comply with the conditions of the arrangement.

7.6 Lessons for Future International Cooperative Arrangements While the institutions created by the Bretton Woods Agreement (and subsequently) fell short of achieving the goals that were originally set for them, they all survived the collapse of the Agreement. The IMF was not able to maintain the par value system, the World Bank was not able to satisfy the financing needs of postwar reconstruction and development, and the GATT was not able to eliminate trade restrictions between countries. But, to their credit, each of these organizations had the flexibility to evolve with economic circumstances and take on new roles in the maintenance of international cooperation. The models presented in this paper indicate that international organizations can facilitate cooperation by serving as commitment mechanisms. Even when countries understand that cooperation will lead to a Pareto-superior equilibrium, they will be reluctant to cooperate unless they are convinced that other countries are also committed to doing so. The postwar institutions all provided member countries with commitment mechanisms, but evidence suggests that some of these were not credible. The IMF and the GATT both provided member countries with a set of rules of cooperation. However, the record indicates that these rules were not consistently enforced. Likewise, the IMF and the World Bank provided members financial resources to enable them to play by the rules. But these resources were so restricted as to tie a country’s incentive to cooperate to the level of its accumulated debt. All three institutions provided members a centralized source of information on each other’s commitment to the rules. Of the three forms of commitment mechanism, evidence suggests that this was the most effective, in that it remains an important function for each institution. If the IMF, the World Bank, and the GATT had not been established after the war, would it have been necessary to create them subsequently? It is difficult to find evidence that they were indispensable. History suggests that more recent architects would have less ambitious goals than the ones formulated at Bretton Woods. Further, the evolution of commitment mechanisms used by 46. De Long and Eichengreen (1991, 2) argue that it was the conditionality that went with the Marshall Plan financial aid that deserves the credit for that program’s success: “Conditionality pushed governments toward political and economic orders that used the market to allocate resources and the government to redistribute wealth, and that turned out to be highly successful at inducing rapid economic growth.”

392

Kathryn M. Dominguez

the postwar organizations indicates that more recent organizations would have relied less on rules and more on the provision of centralized information to promote international cooperation. Our postwar experience with international organizations provides us with three broad lessons. First, commitment mechanisms are effective only if they are credible. The IMF’s original rules of the game were too strict and thus not credible. The Fund effectively turned to a more flexible commitment mechanism, conditionality, to influence member country behavior. Second, an international organization can convince domestic parties to undertake policies that improve global welfare by providing the country’s government with financial incentives that override its time inconsistency problem. Governments can draw on IMF and World Bank resources to credibly forge compromises among conflicting domestic interest groups. Third, evidence suggests that international organizations can effectively promote cooperation by providing their members with a credible monitoring technology. Both the IMF and the World Bank are able to certify their members’ commitment to cooperative behavior by exploiting access to confidential information on members’ economic performance.

Appendix A Proposition I . The CD strategy induces a subgame equilibrium in the devaluation game when C 1, A, can take any one of the values { - M , - m, 0, rn, M}.In all these cases except the last one, Af = M , the CD strategy requires H to devalue at t , yielding profits IT@” = 6(1 - C + nFD). If A, = M ,CD calls for no devaluation and yields IT@ + 6C. Consider deviations that - ef = {m,0 , - rn, consist of setting such that (recall e;F = e;F- ,) AH = -M},assuming that, starting at t 1 , F will stick to the CD strategy. Equations ( A l t ( A 8 ) show that the payoffs to H from following any one of the deviations AH are smaller than nSDand thus also smaller than nSD+ 6C:

+

(Al) If AH = rn and Af # rn, then nHm = 6(a - C nCD)< IT@^.

+

The Role of International Organizations in the Bretton Woods System

393

(A2) If AH = m and Ar = m, then nHm," = 6(a nFD)and nHm," 5 nHCD if a 5 1 - C as assumed. (A3) If AH = 0 and Ar # 0, then nd = 6( - C + r*') < nSD. (A4) If AH = 0 and AI = 0, then n H O , O = 6nCD< rHCD. (A5)IfAH = -mandAz # - m , then nH-"= 6( - a - C nFD)< IT*'. (A6) If AH = - m and Ar = - m , then nH-'"-" = 6( - a nP) < nHCD. (A7) If AH = - M and AI # - M , then nH-M = 6( - 1 - C nFcD) < IT@" - a2C 2 because C 5 - as assumed.

+

+

+

+

62

(A8) If AH = -A4 and Ar = -M, then n H - M . - M = 6( - 1 nFD)< nHCD - 62C

because C Q.E.D.

+

2

+

5 -as

1

62

assumed.

Proposition 2 . Strategy ( S ) is a subgame perfect equilibrium if ( 1 - a) > c > (1 - 6)/(1

+ 6).

Proof. Assume that no devaluation has occurred in the past; then any devaluation triggers competitive behavior. If it is the home country's turn to move and the foreign country has just devalued, then the home country has four options, he" = { - M , - m , m, M}. Clearly, nH-M< nH-"< n H m < +fM so that the only sensible option consists of setting A& = - e f = M, yielding the payoff #'. But it was shown previously that IF"< 0 when C > ( 1 - a)/( 1 + 6 ) . If a devaluation already occurred, then the proof of proposition 1 assures that the punishment phase of strategy s is also one-step unimprovable. Q .E.D .

Proposition 3. Assume that 6-I = 1 + r, r > 0 . Given the assumption of proposition 2 , 1 - a 2 C L ( 1 - 6)/(1 6 ) = r/(2 + r ) , and the necessary condition that a = &, the F strategy induces stability as a subgame perfection equilibrium.

+

Proof. The arguments presented in the text indicate that, if F follows F, H will deviate either an infinite number of times or not at all. This is because any number of deviations n 5 v* is worse than not deviating at all. An infinite number of deviations is better than any finite number n > v*. Thus, a first deviation by H implies an infinite number of deviations to follow. Under these conditions, the maximum discounted value of F's surpluses and deficits is

394

Kathryn M. Dominguez

+

- 6/( 1 6), an amount that F does not have the capacity to repay. This implies that the game with the IMF collapses to the one described in proposition 2 during the very first period and that its outcome is the same as in that game, although F is formally following strategy F. If E reacting to H’s D strategy, deviates from F by devaluing by more than is necessary to maintain external balance, it gains IT*” minus (1 r)6 because it must still repay its accumulated debts to the IMF. But - 6C > - 6 (1 r) > 7FSD - (1 r)6. If F postpones adjustment two periods, F’s profits will be - a3C - a2 - 6 = -6(S2C 6 + 1) net of payments to the IMF because F will not be eligible for loans in the intermediate periods. But -6C + 6’C + a2 + 6 = &[(az - l)C 6 + 11. This is nonnegative if 1 6 2 C(S2 - l ) , which is true for any C > 0. Alternatively, putting off the payment triggers the CD outcome, as it makes any further IMF intervention impossible. Profits in this case are -6(2 + r ) - S2 62,rrH’”,and this is smaller than -6(1 + r) < - 6C. Q.E.D.

+

+

+

+ +

+

+

Appendix B The borrowing country’s maximization problem in the case where debt contracts are perfectly enforceable is maxb,kU(Y

1

+ b - k ) + 6U[Of(k) - (1 + r)b ,

f.0.c.:

U’(C?) = 6(1

+ r)U‘(C,*),

u ~ c :=) esur(c,*y(k*).

Combining the first-order conditions, we obtain the golden rule for investment: Of’@*) = 1 + r, which determines k*. Substituting k* in either f.0.c. yields b*. Assume that, while all countries have the same 6, U ( ), andf( ), k* and b* differ among countries depending on the country’s level of efficiency (8). In order to show that k*(8) and b*(O)are monotonically increasing in 8, first differentiate the golden rule with respect to 8:

f [k*(e)lde + e f ”[k*(e)ldk*(e)= 0 ,

dk*(W - -f’[k*(@l > o. rearranging: -d8 ef“[k*(e)i Next differentiate the first f.0.c. taking dk*(O)/d8into account:

U”(C,)[db* - ““do] d8

395

The Role of International Organizations in the Bretton Woods System = 6(l

+ r)U”(C,) - ( 1 + r)db* + f(k*)de + rearranging:

[uyc,)+ - _ -

iqi

db*(B) d8 ~

+ r)ej-’(k*)urr(c,)]-dk*(e) + ~ ( +1 r)U”(c,Hk*) d8 U”(C,)

+ 6(1 + r)’U”(C,)

> 0.

If debt contracts with sovereign countries are not perfectly enforceable and the only recourse for banks when countries default is to refuse to grant trade credit, the country’s maximization problem includes an additional constraint. Banks will not lend more than they expect will be repaid. Borrowing countries will therefore be faced with a credit ceiling, such that b I8g(k). (Where 8g(k) = pef(k)/[l r] and p is the fraction of output, countries will lose if banks refuse to grant trade credit.) If this borrowing constraint is binding, the country’s problem becomes

+

max, U[Y - k + Oh(k)l + SU[ef(k) - ( 1 + r)Og(k)], f.0.c. u f ( c f ) [ e g f ( k c )- 11 + su‘(q)[efr(kc)i - 6(1 + r)U’(cg)8g’(kc)= 0, rearranging terms:

If 0 is unobservable but there are two types of countries, high efficiency (8,) and low efficiency (O,), high-8 countries will have an incentive to signal to banks their true type. The high-8 country’s problem is

The first-order conditions when the credit ceiling is not binding are

U’(c,,)(l -X)=6(1 +r)U’(c,,), combining:

396

Kathryn M. Dominguez

References Alesina, A., and A. Drazen. 1991. Why Are Stabilizations Delayed? American Economic Review 81(December):l170-88. Altman, 0 . L. 1956. Quotas in the International Monetary Fund. IMF Staf Papers ~ ( A u ~ u 129-50. s~): Axelrod, R. 1984. The Evolution of Cooperation. New York: Basic. Bemstein, E. 1972. The Evolution of the International Monetary Fund. In Bretton Woods Revisited, ed. A. L. K. Acheson, J. F. Chant, and M. F. J. Prachowny. Toronto: University of Toronto Press. Black, S. 1991. A Levite among the Priests: Edward M. Bernstein and the Origins of the Bretton Woods System. Boulder, Colo.: Westview. Brau, E., and R. C. Williams. 1983. Recent Multilateral Debt Restructurings with Official and Bank Creditors. IMF Occasional Paper no. 25. Washington, D.C., December. Bulow, J., and K. Rogoff. 1989. A Constant Recontracting Model of Sovereign Debt. Journal of Political Economy 97( 1):155-78. Chandavarkar, A. 1984. The International Monetary Fund: Its Financial Organization and Activities. IMF Pamphlet Series, no. 42. Washington, D.C. Cooper, R. 1985. Economic Interdependence and Coordination of Economic Policies. In Handbook of International Economics, ed. R. Jones and P. Kenen. Amsterdam: North-Holland. Dam, K. W. 1970. The GATT: Law and International Economic Organization. Chicago: University of Chicago Press. . 1982. The Rules of the Game: Reform and Evolution in the International Monetary System. Chicago: University of Chicago Press. De Long, B., and B. Eichengreen. 1991. The Marshall Plan as a Structural Adjustment Program. Harvard Institute of Economic Research Discussion Paper no. 1576. November. de Vries, M. G. 1966. Fund Members’ Adherence to the Par Value Regime: Empirical Evidence. IMF StafPapers 13(November):504-31. , 1968. Exchange Depreciation in Developing Countries. IMF Staf Papers 15(November):560-78. . 1976. The International Monetary Fund, 1966-1971. Vol. 1, Narrative. Washington, D.C.: IMF. . 1985. The International Monetary Fund, 1972-1978. Vol. 2, Narrative and Analysis, Washington, D.C.: IMF. de Vries, M. G., and J. K. Horsefield. 1969. The International Monetary Fund, 1945-1965. Vol. 2, Analysis, Washington, D.C.: IMF. Eaton, J., and M. Gersovitz. 1981. Debt with Repudiation: Theoretical and Empirical Analysis. Review of Economic Studies 48:289-309. Eichengreen, B. 1987. Hegemonic Stability Theories of the International Monetary System. Harvard Institute of Economic Research Discussion Paper no. 1305. March. Feinberg, R. 1988. The Changing Relationship between the World Bank and the International Monetary Fund. International Organization 42(Summer):545-60. Gertler, M., and K. Rogoff. 1990. North-South Lending and Endogenous Domestic Capital Market Inefficiencies. Journal of Monetary Economics 26:245-66. Gold, J. 1970. The Stand-by Arrangements of the International Monetary Fund. Washington, D.C.: IMF. Greif, A., P. Milgrom, and B. Weingast. 1991. The Merchant Guild as a Nexus of Contracts. Stanford University, Department of Economics, February. Typescript.

397

The Role of International Organizations in the Bretton Woods System

Hooke, A. W. 1982. The International Monetary Fund: Its Evolution, Organization, anddctivities. 2d ed. IMF Pamphlet Series no. 37. Washington, D.C. Horsefield, J. K., ed. 1969. The International Monetary Fund, 1945-1965. Vol. 1, Chronicle. Washington, D.C.: IMF. International Bank for Reconstruction and Development (IBRD). 1954. The International Bank for Reconstruction and Development, 1946-1 953. Baltimore: Johns Hopkins University Press. Keohane, R. 1980. The Theory of Hegemonic Stability and Changes in International Regimes, 1967-1977. In Change in the International System, ed. 0 . Holsti, R. Siverson, and A. George. Boulder, Colo.: Westview. Kreps, D. 1990. A Course in Microeconomic Theory. Princeton, N.J.: Princeton University Press. League of Nations. 1944. International Currency Experience: Lessons of the InterWar Period. Geneva. Maskin, E., and J. Tirole. 1988. A Theory of Dynamic Oligopoly I: Overview and Quantity Competition with Large Fixed Costs. Econometrica 56(May):549-69. Mentre, P. 1984. The Fund, Commercial Banks, and Member Countries. IMF Occasional Paper no. 26. Washington, D.C., April. Mikesell, R. 1972. The Emergence of the World Bank as a Development Institution. In Bretton Woods Revisited, ed. A. L. K. Acheson, J. F. Chant, and M. F. J. Prachowny. Toronto: University of Toronto Press. Milgrom, P., D. North, and B. Weingast. 1990. The Role of Institutions in the Revival of Trade: The Law Merchant, Private Judges, and the Champagne Fairs. Economics and Politics 1(March):1-23. Milgrom, P., and J. Roberts. 1982. Limit Pricing and Entry under Incomplete Information: An Equilibrium Analysis. Econometrica 50(March):443-59. Moggridge, D. 1980. The Collected Writings of John Maynard Keynes. Voi. 26, Activities 1941-1946. London: Macmillan. Mookerjee, S. 1966. Policies on the Use of Fund Resources. IMF Staff Papers 13(November):421-41. Oliver, R. W. 1975. International Economic Co-operation and the World Bank. London: Macmillan. Polak, J. 1991. The Changing Nature of IMF Conditionality. Princeton Essays in International Finance, no. 184 (September). Princeton, N.J.: Princeton University. International Finance Section. Putnam, R. D. 1988. Diplomacy and Domestic Politics: The Logic of Two-Level Games. International Organization 42(Summer):427-60. Putnam, R. D., and N. Bayne. 1987. Hanging Together: Cooperation and Conflict in the Seven-Power Summits. London: Sage. Stiglitz, J., and A. Weiss. 1981. Credit Rationing in Markets with Imperfect Information. American Economic Review 71(June):393-410.

COIlllllent

Alberto Alesina

The message of the paper by Kathryn Dominguez can be summarized by the following quote: “The IMF was not able to maintain the par value system, the Alberto Alesina is the Paul Sack Associate Professor of Political Economy at Harvard University, a faculty research fellow of the National Bureau of Economic Research, and a research fellow of the Centre for Economic Policy Research.

398

Kathryn M. Dominguez

World Bank was not able to satisfy the financing needs of postwar reconstruction and development, and the GATT was not able to eliminate trade restrictions between countries. But, to their credit, each of these organizations had the flexibility to evolve with economic circumstances and take on new roles in the maintenance of international cooperation.” It should be noted that Dominguez’s paper is almost exclusively concerned with the IMF; the role of the other two organizations is analyzed very briefly. I find the first part of the paragraph too harsh, and based on an overestimation of what these organizations could achieve and an underestimation of what they did achieve. In some sense, the statement is correct: the par value system did not last, and several devaluations occurred; the world is not a free trade area, and the World Bank did not solve all the financing need for reconstruction and development. I guess the question is what we can expect from international organizations like these. Can they really enforce “good behavior and cooperation” if key member countries do not want to cooperate? After all, the first twentyfive years after the Second World War were an economic dream, at least for the industrial countries, compared to the twenty-five years after the First World War. The second part of the statement does not necessarily save the reputation of these organizations; the fact that they adapted to changing circumstances may just be an example of a successful struggle of inefficient international bureaucracies to survive. The fact that they survived is not enough credit, in particular given the rather negative judgment cast on their performance in the first thirty years of their lives in the first part of the paragraph. I shall first discuss the theory of cooperation presented in the paper; then I shall turn to the empirical evidence. Dominguez argues, quite correctly, that one can think of three methods of cooperation: (1) rule-based cooperation, which is the strongest but least flexible type of cooperation; (2) international negotiations with no rules, which is the weakest but most flexible type of cooperation; and (3) cooperation enforced with a combination of rules with escape clauses and international negotiations, in the context provided by international organizations. The type of cooperation chosen in the Bretton Woods Agreement is clearly the third one. Dominguez suggests that it failed or, at least, was largely unsuccessful. Section 7.1 of the paper considers a standard repeated prisoner’s dilemma in which the “dilemma” is that noncooperative behavior leads to competitive devaluations. Dominguez argues that sovereign countries by themselves (i.e., simply by force of reputation and fear of retaliation) would have trouble enforcing cooperative behavior for several reasons, including information and monitoring costs. An international organization helps support cooperation by favoring diffusion of information, facilitating negotiation, and generally promoting mutual understanding. I agree with all these arguments. In fact, there is a direct connection between the role of international organizations and the

399

The Role of International Organizations in the Bretton Woods System

cost of noncooperative behavior. By joining, say, the IMF, a country set itself up in a situation in which, in order to devalue in a noncooperative fashion, it would have had to engage in some kind of discussion with the IMF, perhaps leave this organization, basically “make a fuss.” Thus, joining an international organization raised the costs of noncooperative behavior; in the language of repeated games, the costs of “cheating” are increased by joining the organization. Reputation and institutions complement each other; they are not substitutes. Even though institutional arrangements may increase the costs of noncooperation in the repeated prisoner’s dilemma games, in the end it is always the fear of losing reputation and being retaliated against that prevents a country from behaving noncooperatively. This simple consideration implies that one cannot expect miracles from international organizations of independent and sovereign nations. Up to this point, the paper considers each country as a homogeneous player. In section 7.3, the paper briefly touches on a very interesting issue: the interconnection between international cooperation and domestic political conflict. Recent developments in international relations theory emphasize how one cannot separate the dynamic of domestic political conflicts from the resolution of international policy coordination problems. I Putnam convincingly argues that politicians and bureaucrats engaged in international policy coordination play a “two-level game”: one “level” within the context of domestic politics, one “level” in the international arena. This interplay is viewed by Dominguez as one in which the IMF is a sort of “international social planner” that maximizes world welfare. On the contrary, the governments of each individual country are short lived and short sighted, suffer from time-inconsistency problems, and cannot resolve domestic distributional struggles, I have a lot of sympathy for this approach, particularly insofar as it captures domestic politics. However, the treatment of the IMF as essentially a “world social planner” maximizing “global welfare” is oversimplified. Is the IMF maximizing global welfare, or are IMF policies the result of some resolution of the conflicts of interests of different country members, which may or may not coincide with “global welfare”? More generally, how are the “weights” of the “global welfare function” derived? I would have liked to see much more in this paper on the politics of IMF interventions, from the point of view of both domestic and international politics. For instance, it would have been useful to investigate more closely what the IMF can do to resolve domestic distributional “wars of attrition,” which are discussed in the paper. De Long and Eichengreen argue that the difference between the successful adjustments in Western Europe after the Second World 1. See Robert D. Putnam, “Diplomacy and Domestic Politics: The Logic of Two Level Games,” International Organization 43, no. 3 (Summer 1988):427-60; and Robert Keohane and Joe Nye, “Transgovernmental Relations and International Organizations,”World Politics 27 (1974):39-62.

400

Kathryn M. Dominguez

War and the much less successful ones after the First World War is precisely due to the fact that domestic distributional conflicts were resolved more quickly and more cooperatively after the Second World War.2 This, in their view, was the result of the skillfully designed Marshall Plan. I wonder whether the more cooperative international climate that the three international organizations promoted had something to do with this success story. I found these “political” arguments very interesting and very innovative; I wish that they had played a more substantial role in Dominguez’s paper. The empirical part of the paper documents the view that one should have expected a lot from these three organizations, and in particular from the IMF, but that they did not deliver. According to the author, the IMF never enforced the par value system and let countries devalue their currencies or fluctuate without even checking whether a “fundamental disequilibrium” really justified such devaluations. The discussant’s role is to disagree, even though I have a lot of sympathy for the view pushed in the paper. I will make two points. First, the fact that the IMF never (except in a couple of cases) opposed a devaluation does not necessarily mean that its role was irrelevant. Other devaluations may not have been carried over in anticipation of difficulties with the IMF. Second, it is not completely clear what benchmark is to be used to decide whether there were too many (or too few!) devaluations and/or whether credit facilities were used too much or not enough. For instance, table 7.2 reports the adherence to par values in 1946-66. Is this rate of “effective stability” (see the last column of this table) high or low? Dominguez explicitly acknowledges the difficulty created by this lack of an obvious benchmark. Nevertheless, she feels that, in the end, a negative judgment should be cast on the IMF. Such a judgment has, unavoidably, a certain amount of arbitrariness. Nevertheless, I applaud her strong stand; much better to provoke the reader to “think” than writing too much of a “two-handed” conclusion! In the final part of the paper, Dominguez argues that, after the collapse of the Bretton Woods system, the IMF changed its role and became a monitoring agency geared toward reducing problems of asymmetric information between commercial banks and borrowing countries. A crucial, difficult question that is not truly addressed by the paper is, If the IMF did not exist, should it have been created in the mid-1970s after the collapse of the Bretton Woods system? Or did the IMF survive just because it was there? Do the new functions of the IMF justify its creation ex novo or only its survival? Did the IMF survive because of “sunk costs” arguments? In summary, this paper reads as if, with the help of international organizations, it should have been relatively easy to make the world a harmonious 2. Bradford De Long and Barry Eichengreen, “The Marshall Plan as a Structural Adjustment Program” (Cambridge, Mass.: Harvard Institute of Economic Research Discussion Paper Series no. 1576, November 1991).

401

The Role of International Organizations in the Bretton Woods System

place in which trading partners cooperate on economic matters. Since we are still far from this ideal situation, international organizations must have failed. I take a much more cynical view. Incentives not to cooperate are strong. From the creation of modern nation-states five hundred years ago, countries have been at war with each other very often, including in this present century, the first half of which was one of the bloodiest periods of modem history. After the Second World War, industrial countries have managed to survive more cooperatively than ever before. We did not have another Great Depression as in the 1930s; trade restrictions were reduced; except in the mid-1970s growth was relatively high and stable. We did not have a Third World War; even the Cold War is over. It is difficult to tell whether we would have had the same economic outcomes without the three organizations. However, while reading this well-executed paper, I had to remind myself every once in a while how much better the post-Second World War adjustment has been than the postFirst World War adjustment.

Comment

William H. Branson

This paper provides an interesting new analysis of the role of the Bretton Woods institutions in the world economy, focusing mainly on the IMF. In a series of game-theoretic models, the paper shows the IMF playing several roles. It can provide a commitment mechanism in cases within countries where there may be prisoner’s dilemmas between policymakers. Essentially, it can provide resources to support a ban on bad strategies. It can also provide information to private lenders via monitoring, giving countries incentives to be good borrowers. While illustrating the operation of the IMF, the paper also evaluates how the institutions, especially the IMF, have evolved with the changing world economic environment. I will come back to this evaluation at the end of this Comment. The first series of models in section 7.1 begins with a two-country policy game in which the noncooperative static equilibrium is a prisoner’s dilemma. The second model is a repeated version of the same game where tit-for-tat strategies yield cooperati0n.A many-country version of this model would face collective action and free rider problems. These models are presented to motivate a discussion of alternative solutions to the cooperation problem. These solutions are rules, which economists like but which are hard to write; international negotiations, which political scientists like but which yield indeterminate results; and international institutions, which can police the rules, William H. Branson is professor of economics and international affairs at Princeton University and a research associate of the National Bureau of Economic Research.

402

Kathryn M. Dominguez

organize negotiations, and provide information and monitoring. I suppose international bureaucrats favor international organizations. The point of all this is to see in section 7.2 that the participants at Bretton Woods set up a system that provides some of all three. This is still true, even after the par value system broke down in the early 1970s. In this sense, the Bretton Woods system still exists and may be working pretty well! A second model is introduced in section 7.3 to illustrate the link between cooperation within a country between two groups of decision makers and the IMF’s provision of resources. Here we have two groups struggling over who pays the taxes to support the beneficial supply of government services. The executive would like to threaten to penalize any group that does not support the government compromise. But the threat is time inconsistent because the penalty will reduce national welfare. The role of the IMF is to provide resources that make the threat credible. It seems to me that this story requires that the penalty somehow be “wasted.” Suppose that the penalty were transferred from the noncooperative to the cooperative group. Then, since national welfare is the sum of the two groups’ welfares, there could be a zero loss to national welfare from the penalty. In the context of this model of the role of IMF or World Bank provision of resources, I would rather assume that the losing groups are being bought off by the external agent. The second part of section 7.3 discusses the role of the IMF in providing resources to induce countries to maintain the par value system. The model is a dynamic version of the first devaluation game, in which the countries take turns deciding whether to change their gold price. The IMF provides financing for temporary current account deficits as long as the countries stay with the system. The result is an equilibrium in which the countries do cooperate as long as they have not reached their credit limits. This model is used to score a point for Keynes, who wanted no repayment for deficit countries, and against White, who wanted stricter credit limits. The period 1944-71 is interpreted in the paper as the rules phase of the IMF’s history, in which the rule was the par value system and the IMF was the enforcer. The experience of this period is evaluated in section 7.4. The verdict, with which I agree, is that the rules were too rigid. There were many devaluations early by the Europeans and later among the developing countries. By 1971, there were few fully convertible currencies, and one-quarter of the members had yet to declare par values. I think of the period after 1950 as one in which the center currencies were largely fixed to each other and the periphery was variable. This is in contrast to the period after 1971, in which the center currencies, now the dollar, yen, and ECU, float and the periphery generally pegs to some average. In the environment since the mid- 1970s, no country can maintain a fixed exchange rate since the center is variable. The length of time it took for the Western European countries to establish full convertibility makes current suggestions that the Central and Eastern Europeans move immediately to full convertibility at least questionable.

403

The Role of International Organizations in the Bretton Woods System

After 1971, the IMF turned to its monitoring role, especially with respect to developing countries. In section 7.5, the paper lays out nicely the nowstandard debt overhang model. Here, the banks are uncertain about which countries are good or bad borrowers, the bad borrowers have incentives to try to look like good ones, and monitoring is costly. In this case, the IMF can provide the monitoring via standby facilities, good borrowers come in for certification, and the private banking system lends to them. This is a nice model of IMF (and also World Bank) interaction with the borrowers. The paper ends with an evaluation of the three institutions, the IMF, the World Bank, and the GATT. All these institutions have had to cope with major changes in the world economic environment. These changes continue today with the relative decline of the United States, the emergence of a three-bloc trade world, and the potential transformation in Central and Eastern Europe. The IMF and the Bank seem to have adjusted relatively well. The IMF has supported the change in the exchange rate system; the “Bretton Woods” system lives on with major currencies more or less floating. The World Bank has adopted policy-based “structural adjustment” lending, improving its accessibility to the developing countries. The GATT has fared less well. After several rounds of multilateral tariff cuts, the multilateral system seems to be collapsing, and the institution is not reacting as flexibly as the IMF or the Bank.

General Discussion Sebastian Edwards argued that the IMF lost credibility for its exchange rate policy, not because there were too many changes in parities, but because the escape clause aspect of the adjustable peg made it difficult for the Fund to object to a proposed change. This problem was exacerbated by the clause preventing the Fund from objecting to a proposed change because of the domestic, social, or political policies of the member countries. He stressed the importance of conditionality, which developed in the 1950s and was legalized in the First Amendment to the Articles in 1968. One reason why the Fund’s resources were not fully used was conditionality: at some point, many countries found that it was difficult to meet the conditions. Edwards pointed out that the seal-of-approval role of the IMF evolved slowly over time from a precedent set by the League of Nations. John Williamson agreed with Edwards that the problem was not that there were too many changes in par values. Rather, there were two problem in the way the Fund dealt with its responsibility in exchange rate supervision-it allowed overshooting, especially in the case of devaluation in 1949, and no pressure was applied for changes in par values that were inappropriate. Dale Henderson pointed out that there is a relation between the amount of resources the Fund has and the extent of its monitoring role.

404

Kathryn M. Dominguez

Willem Buiter viewed the paper as documenting the progressive modernization of the IMF. The Fund originally had the joint roles of providing systemic order and individual country surveillance. The former role disappeared with the end of the official exchange rate component of Bretton Woods, when advanced countries realized that they no longer needed access to the Fund’s resources. The latter role became important for countries that still needed the Fund’s resources, especially less developed countries. Edward Bernstein discussed the original mandate of Bretton Woods. It was to create the IMF and the World Bank. The International Trade Organization was not considered at Bretton Woods. This point was expanded on by Leslie Pressnell. Bernstein then discussed the nature of the fundamental difference between the positions of the United States and the United Kingdom at the conference. The British delegation wanted the responsibilities of Fund members to be effective as long as the Fund provided them with financing. The American delegation believed that, once members had made a commitment to the institution, they should be allowed to follow whatever methods were applicable to the given situation. Bernstein concluded that the most important contribution of the IMF-that has survived Bretton Woods-is its role as an institution where monetary problems can be discussed. Man Corden expanded on the useful roles of the IMF. These include a monitoring role, a general information role, the important role played by the IMF’s country missions, the IMF’s role in creating a formidable climate of opinion, and a role as a propagator of economic orthodoxies. Maurice Obstfeld felt that more emphasis should be placed on the constructive role of the GATT in promoting the growth of trade and income under Bretton Woods. He suggested modeling the GATT in terms of changing payoffs in a game between governments and constituencies that favor free trade or protection within a country. Instead of the model used in the paper, that of the monitoring role of the IMF, one based on adverse selection, Obstfeld would prefer one based on moral hazard. Lars Jonung remarked on how little discussion there had been on the role of markets and spontaneous order as opposed to government interrelations in creating the cooperation of the Bretton Woods era.

8

Devaluation Controversies in the Developing Countries: Lessons from the Bretton Woods Era Sebastian Edwards and Julio A. Santaella

In 1973, the international monetary system forged in Bretton Woods experienced a final collapse, as the industrial nations abandoned all efforts to sustain a fixed exchange rate regime and decided to adopt freely floating exchange rates. In spite of this significant change in the international financial system, throughout the 1970s most of the developing countries continued to rely heavily on fixed exchange rates, mainly pegging to specific countries within the spirit of an optimum currency area. For example, the December 1979 issue of International Financial Statistics (IFS) reports that 85% of the developing countries had some sort of fixed exchange rate system at that time. During the 1980s and early 1990s, however, an increasing number of developing countries moved away from fixed exchange rates and adopted more flexible regimes. According to the December 1990 issue of IFS, the proportion of less developed countries (LDCs) that had some type of fixed exchange rate had declined to 67%. This movement toward greater exchange rate flexibility was, to a considerable extent, associated with the debt crisis unleashed in 1982. Those countries that had to cope with sudden cuts in external financing had very limited policy options. In an effort to engineer gigantic resource transfers to their creditors, most of these countries adopted adjustment packages that included, as an important component, the abandonment of fixed rate practices. It is in this context that, in the mid-l980s, we saw the end of long experiences with fixed exchange rates in countries such as Venezuela, ParaSebastian Edwards is the Henry Ford I1 Professor of International Business Economics at the Anderson Graduate School of Management, University of California, Los Angeles, and a research associate of the National Bureau of Economic Research. Julio A. Santaella is an economist at the International Monetary Fund. The authors are grateful to their discussants, Albert Fishlow and Stan Fischer, for helpful comments and to the editors of this volume for encouragement. Sebastian Edwards acknowledges support from the National Science Foundation and the University of California Pacific Rim Program. The authors are grateful to Roberto Schatan and Pablo Sanguinetti for their help.

405

406

Sebastian Edwards and Julio A. Santaella

guay, and Guatemala. Many countries rapidly adapted to their new circumstances. The exchange rate ceased to be a sacrosanct variable linked to the nationalistic destinies of countries; during the late 1980s, a large number of economies had become increasingly comfortable with managed exchange rate regimes. Recently, however, a number of observers and experts-including prominent members of the IMF Executive Board-have argued that the enthusiasm for devaluation and an active exchange rate policy has gone too far. It has been pointed out that, by relying too heavily on exchange rate adjustments, and by allowing developing countries to adopt administered systems characterized by frequent small devaluations, Fund programs have become excessively inflationary. According to this view, exchange rate policy in the developing countries should move toward greater rigidity-and even complete fixity-as a way to induce financial discipline and reduce inflation. This position, which is steadily gaining new supporters, has largely been influenced by current macroeconomic views that emphasize the role of expectations, credibility, and institutional constraints (see, e.g., Aghevli, Khan, and Montiel 1991; Agenor and Montiel, 1991; and Burton and Gilman, 1991). It would be illusory, however, to think that a return to greater exchange rate fixity will completely eliminate situations of “fundamental disequilibrium.” In fact, most supporters of nominal exchange rate anchors concede that, under conditions of severe exchange rate misalignment, it is generally advisable to implement adjustment packages that combine fiscal and credit restraint with a discrete nominal devaluation (see Burton and Gilman 1991). What is perhaps paradoxical is that precisely this type of pegged arrangement, where the currency may be occasionally devalued by a large amount, was extremely controversial during the Bretton Woods period. In fact, the “devaluation issue” was often at the forefront of conflict between national authorities and the staff of the International Monetary Fund. Even under conditions of obvious “fundamental disequilibrium,” the economic authorities in the developing countries tended to resist devaluing their currencies. Instead, they often imposed trade and exchange controls in an effort to avoid a balance of payments crisis.’ This historical resistance to devaluations had its roots in a deep skepticism about the effectiveness of exchange rate adjustment. In fact, it has been commonplace in the developing world to argue that large and discrete devaluationsand especially devaluations implemented within the context of IMF programs-have no effect on the external sector, result in output contractions, and worsen income distribution. (see Denoon 1986; Buira 1983; and SELA 1986). An important question in the current debate regarding the desirability of a 1. There has traditionally been a sense among some observers that LDCs have been forced by third parties-and in particular by the IMF-to devalue their currencies (see, e.g., Denoon 1986).

407

Devaluation Controversies in the Developing Countries

return to fixed rates revolves around the actual effectiveness of discrete and substantial devaluations in the context of a fixed rate regime. The purpose of this paper is to deal with this issue from a historical perspective. We analyze the conditions surrounding 48 major devaluation episodes in the developing countries that took place during the Bretton Woods period (1954-71). By focusing on the Bretton Woods era, we can examine the fundamental empirical features of large nominal devaluations in a historical environment with generalized fixed exchange rates. This paper differs from previous work in three fundamental respects. First, a formal distinction is made between devaluations undertaken within the context of an IMF program and unilateral devaluations implemented without a formal IMF-sponsored program.2 This distinction is particularly interesting because it allows a critical assessment of the role of the Fund; it provides a very natural benchmark for evaluating the results associated with IMF programs. In that sense, the traditional difficulty of finding appropriate “counterfactuals” to IMF programs is somewhat r e d ~ c e dIn . ~this analysis, we ask why some countries sought IMF supports while others undertook adjustment-cumdevaluation programs on their own. We also inquire whether, on average, IMF devaluers tended to fare better than non-IMF devaluers. Second, in addition to analyzing the economic aspects of these devaluations, we investigate some important political developments surrounding these episodes. We inquire, in particular, whether countries that received IMF assistance were characterized by a different political environment than those that did not approach the Fund. We also analyze the extent to which the political structure affects the degree of success of an adjustment-cum-devaluationprogram. Third, we compare the main features of these Bretton Woods devaluations with a number of more recent devaluations. The empirical approach followed here is based on Edwards (1988, 1989c) and combines nonparametric tests with cross-country regression analyses in an effort to understand the circumstances surrounding these forty-eight devaluations. A salient feature of our approach is that we analyze in detail the evolution of a number of key variables during the three years preceding and three years following the forty-eight devaluation episodes. In doing this, an effort is made to detect regularities across countries that will allow us to infer some general rules relating to the causes and effects of devaluations. At the same time, we point out peculiarities that help better understand the exchange rate history of a particular country. In addition to the groups of IMF and nonIMF devaluation countries, we defined a control group of twenty-four devel2. For lack of a better name, we call these unilateral devaluations. Notice, however, that, according to Bretton Woods rules, the Fund had to approve all nominal devaluations exceeding 10%. 3. Most studies evaluating the effectiveness of Fund programs have compared these programs with countries that have not undertaken an adjustment program. This has even been the case with recent efforts based on regression analyses.

408

Sebastian Edwards and Julio A. Santaella

oping nations that maintained a fixed nominal exchange rate for at least ten years; their behavior is compared to that of the devaluing countries. In these comparisons, a series of parametric and nonparametric tests were

8.1 The Simple Economics of Devaluation, Adjustment, and Credibility In this section, we briefly provide an analytic framework for the empirical analysis that follows. We first analyze the conditions leading to a situation of “fundamental disequilibrium” and external sector crisis and discuss the conditions under which devaluations are effecti~e.~ We then analyze the role of an external multilateral agency, such as the IMF, in a stabilization program. 8.1.1 Fundamental Disequilibrium and the Theory of Devaluations Fixed exchange rates introduce restrictions to macroeconomic policies: if a (small) country wants to maintain its parity, its inflation rate cannot exceed (for a significant period of time) the rate of world inflation. Historically, however, policymakers have tried to ignore the constraints imposed by fixed exchange rates by implementing rapid fiscal expansions. If, as is often the case, this increase in fiscal expenditures is mostly financed through domestic credit creation, we will have a number of macroeconomic effects. First, there will be an increase in the demand for tradable goods, a worsening of the current account, and, with other things given, a loss of international reserves. Second, there will be a higher demand for home goods, a higher domestic rate of inflation, a real exchange rate (RER)overvaluation, and a continuous erosion in the country’s degree of international competitiveness. As international reserves draw lower, the government will usually try to tackle the situation by imposing exchange, capital, and trade controls. The parallel premium for foreign exchange will increase, and the black market will grow in scope. Naturally, these controls will not solve the crisis; they will merely slow down the loss of reserves and postpone the required adjustment. At some point the authorities will realize-or will be forced to recognizethat the country is following an unsustainable path and that adjustment is required. This stylized story suggests that the conditions faced by countries facing “fundamental disequilibrium,” and thus the need to devalue, can be summarized as follows: (1) fiscal and credit policies become “inconsistent” 4.This episodic strategy has modem precedents in Cooper’s (1971a, 1971b, 1971c) classic studies of devaluation and, more recently, in Harberger and Edwards’s (1982) study of balance of payments crises. However, Cooper did not deal with the period preceding the devaluation, and, contrary to this study and to Harberger and Edwards’s, he did not use a control group for comparison. Other studies that make use of the episodic approach employed in this paper are Kamin (1988) and Eichengreen (1990). 5. For a more technical representation of the economics of devaluation, see, e.g., Edwards (1989~)and Khan and Lizondo (1987).

409

Devaluation Controversies in the Developing Countries

with the objective of maintaining a fixed exchange rate; (2) there is a rapid rate of domestic inflation; (3) there is a large current account deficit; (4) international reserves become very low and continue to decline; (5) the parallel market premium increases; and (6) the RER becomes rapidly overvalued. Of course, a situation of “fundamental disequilibrium” can also be caused by a real shock (like a change in the terms of trade) that creates a macroeconomic gap that has to be closed. The first fundamental step in an adjustment program is to tackle the sources of the initial disequilibrium: the fiscal imbalance has to be reduced and financial discipline reestablished. Another crucial element in the adjustment program is the correction of the situation of RER overvaluation. The relative price of tradables to nontradables faced by domestic agents has to increase to a level compatible with external equilibrium. There are two basic ways of achieving this required RER adjustment or RER devaluation. The first is to follow a disinjutionury policy, where the reduction in aggregate demand atrained through the fiscal adjustment forces a reduction in nominal prices of nontradable goods. This option, however, has two important drawbacks: under most circumstances it is too slow, and, if nominal prices (or wages) are inflexible downward, the transition will be characterized by unemployment and reduced production. The second basic alternative for reestablishing RER equilibrium is by engineering an increase in the domestic price of tradable goods through a nominal devaluation. In this case, of course, all the nominal devaluation is attempting to do is speed-up the adjustment. Even when realignment of relative prices is accomplished, the nominal devaluation is not the ultimate cause of the observed real exchange rate change; it is merely the vehicle through which the adjustment is attained.6 Naturally, for the nominal devaluation to be effective, in the sense of truly helping reestablish macroeconomic equilibrium in a smoother fashion, two main conditions have to be met. First, the devaluation has to be taken from a starting disequilibrium situation of RER overvaluation; second, the devaluation has to be accompanied by consistent macroeconomic and, especially, fiscal policies.’ If these conditions are not met, the devaluation will not be successful.8 In sections 8.2 and 8.3 below, we use data on forty-eight discrete 6. Eichengreen (1990) documents how, during the 1930s. some countries decided to follow a deflation while others chose to devalue their currency. 7. These may include the need to deindex labor and other contracts. 8. The above discussion clearly suggests that, in order for devaluations to “work,” there is no need, as is often suggested, for economic agents to have money illusion. Indeed, within this scenario, devaluations will facilitate the adjustment even when there are ultra-rational forwardlooking economic agents. In fact, in a fonvard-looking world, devaluations undertaken within the context described here will tend to be particularly effective. The reason for this is that these highly informed rational individuals will clearly understand that the devaluation is facilitating relative price changes and, thus, is inducing the required expenditure switching away from tradable goods. Consequently, economic agents will not react to this exchange rate change in a perverse way. However, if the devaluation is not accompanied by consistent macroeconomic policies, the in-

410

Sebastian Edwards and Julio A. Santaella

devaluation episodes to analyze whether the experience in the LDCs during the Bretton Woods period conforms to the view on devaluation cases presented here. 8.1.2 Credibility, Commitment, and Adjustment Recent work on stabilization and adjustment has emphasized the role of institutions and credibility. A number of authors have argued that, in order to put an end to macroeconomic and external sector disequilibrium, a “credible” change of the policy regime is needed (Sargent 1986; see also Edwards and Tabellini 1991a, 1991b). To the extent that a stabilization is not crediblethat is, to the extent that the public does not expect that the program will achieve its intended results-the cost of adjustment escalates, and the probability of success becomes smaller (Dornbusch 1991). This view leads governments naturally to look for ways of modifying and influencing expectations during a stabilization program. “Policy announcements” have been considered a possible vehicle for affecting inflationary expectations. However, it has been argued that in order for these announcements to be “credible”-and, thus, actually to affect expectations-it is necessary for the government to be able to precommit itself to a given course of action. This, of course, turns out to be difficult since societies many times lack the institutional setup required for the government to precommit itself credibly. Under certain circumstances, however, reputation can act as a substitute for precommitment. According to this view, the desire of a government to preserve its reputation-or, even possibly, improve it-provides it with a constrained set of policy options (see Persson and Tabellini 1990). Some authors have recently suggested that expectations can also be coordinated and that credibility can be established if it is supported by an external institution, such as the League of Nations in the 1920s and the International Monetary Fund after 1950 (see Sachs 1989; Edwards 1989b; and Santaella 1991). The reason is that, by granting its “seal of approval” to a stabilization plan, the external institution enhances the confidence in the program. In principle, this “seal of approval” will be independent of the financing that the external institution can p r ~ v i d eIn . ~fact, the presence of external involvement can endow the stabilizing government with a “commitment technology” that gives a greater assurance that the announced program will indeed be fully carried out. This framework has two important empirical implications: first, we would expect that countries that have more difficulty establishing independent credibility would

formed public will anticipate a devaluation-inflation spiral, making the situation even more critical than before. Some cross-country evidence on the effectiveness of devaluations can be found in Cooper (1971a, 1971b, 1971c), Edwards (1989~).Kamin (1988), and Gylfason and Radetzki (1985). 9. Accounts of the support given by external credits and loans to stabilizing countries are in League of Nations (1946) and Dornbusch and Fischer (1986).

411

Devaluation Controversies in the Developing Countries

be more likely to approach the IMF; second, to the extent that IMF-sponsored programs provide additional credibility, it would be expected that, with other things given, countries that undertake an IMF stabilization-cum-devaluation adjustment program would have an advantage over those nations that unilaterally implement adjustment programs. These empirical propositions are confronted with data from our forty-eight devaluation episodes in the sections that follow.

8.2 Macroeconomic Policy, Fundamental Disequilibrium, and IMF Programs In this and the following section, we analyze in detail forty-eight episodes of balance of payments and devaluation crises in the LDCs between 1954 and 1971. This investigation focuses on five important issues; (a) the role of “inconsistent” macroeconomic policies in generating “fundamental disequilibria” and in precipitating devaluation crises; (b) the differences, if any, between IMF and non-IMF devaluations; (c) the effectiveness of nominal devaluations as a means of restoring equilibrium and competitiveness; (d) the role of political forces in devaluation and IMF involvement; and (e) the determinants of successful stabilization-cum-devaluationpackages. 8.2.1 The Data Set Our data set consists of forty-eight major stepwise devaluations implemented by independent developing countries in the period 1948-7 1. In order for a devaluation episode to be included in our sample, it had to have the following properties: (1) the adjustment of the official rate had to exceed 14%; (2) the devaluation must have occurred after a period of at least two years where the country had a fixed exchange rate; and (3) the country in question must have had a population of at least one million people the year of the devaluation. Using the Znternational Financial Statistics (ZFS) tape and other sources, we identified sixty-nine devaluation episodes that met the three requirements set up above. Once those cases with no (or very little) data on the most important variables were eliminated, we were left with the forty-eight countries considered here. In that sense, then, an effort was made to identify, and then include, every one of the large step wise devaluation episodes that took place in the developing world during 1954-7 1. The final inclusion criteria were based exclusively on data availability. The exact dates of our forty-eight devaluations, as well as the inception and expiration dates of IMF standby programs, are shown in table 8.1. Twentytwo of the forty-eight devaluers implemented a unilateral (i.e., non-IMF) devaluation, while twenty-six had IMF programs. All the IMF programs considered here were standby arrangements, which were envisaged in 1952 by the Fund to control drawings in the credit trenches. Standby programs soon became the main instrument through which the IMF imposed conditionality. The

412

Sebastian Edwards and Julio A. Santaella

Table 8.1

Devaluation Episodes and IMF Programs in Selected Developing Countries; 1954-71

IMF Program ~ _ _ _ _ _ _ _

Country

Devaluation Date

I . Argentina 2. Argentina 3. Argentina 4. Argentina 5. Brazil 6. Chile 7. Colombia 8. Colombia 9. Colombia 10. Colombia 1 1. Costa Rica 12. Ecuador 13. Ecuador 14. Egypt 15. Ghana 16. Ghana 17. India 18. Indonesia 19. Israel 20. Israel 21. Israel 22. Jamaica 23. Korea 24. Korea 25. Malawi 26. Mexico 27. Nicaragua 28. Pakistan 29. Peru 30. Peru 3 1. Philippines 32. Philippines 33. Sierrahone 34. Spain 35. Spain 36. Sri Lanka 37. Trinidad-Tobago 38. Tunisia 39. Turkey 40. Turkey 41. Uruguay 42. Uruguay 43. Uruguay 44.Venezuela 45. Yugoslavia 46. Yugoslavia

28 Oct. 1955 2 Jan. 1959 19 Mar. 1962 18 June 1970 13 Feb. 1967 15 Oct. 1962 18June 1957 20Nov. 1962 2Sept. 1965 22 Mar. 1967 2 Sept. 1961 14 July 1961 17 Aug. 1970 7 May 1962 8 July 1967 27 Dec. 1971 6 June 1966 17 Apr. 1970 9Feb. 1962 19Nov. 1967 21 Aug. 1971 21 Nov. 1967 23 Feb. 1960 3 May 1964 20Nov. 1967 19 Apr. 1954 1 July 1955 1 Aug. 1955 22 Jan. 1958 31 Aug. 1967 22 Jan. 1962 21 Feb. 1970 22 Nov. 1967 18 July 1959 20 Nov. 1967 22 Nov. 1967 23 Nov. 1967 28 Sept. 1964 4Aug. 1958 3 Aug. 1970 15 Dec. 1959 9 May 1963 26 Apr. 1971 18 Jan. 1964 1 Jan. 1961 25 July 1965

Inception Date

Expiration Date

19 Dec. 1958 12Dec. 1961

18 Dec. 1959a 11 Dec. 1962b

13 Feb. 1967

12 Feb. 1968

19June 1957 1 Jan. 1962

18 June 1958 31 Dec. 1962

15 Apr. 1967 4Oct. 1961 8 June 1961 14 Sept. 1970 7 May 1962 25 May 1967

14Apr. 1968 3 Oct. 1962 7 June 1962 13 Sept. 1971 6 May 1963 24 May 1968

17 Apr. 1970

16Apr. 1971

...

...

... ...

... ...

...

16 Apr. 1954

...

...

... ... 15 Apr. 1955

... 18 Feb. 1957 18 Aug. 1967 12 Apr. 1962 20Feb. 1970

17 Feb. 1958' 17 Aug. 1968 11 Apr. 1963 19 Feb. 1971

17 Aug. 1959

16 Aug. 1960

...

...

...

... 1 Oct. 1964

30 Sept. 1965

17Aug. 1970

16 Aug. 1971

4 Oct. 1962 28 May 1970

3 Oct. 1963 27 May 1971

1 Jan. 1961

31 Dec. 1961 25 July 1966

... ... ...

26 July 1965

...

413

Devaluation Controversies in the Developing Countries

Table 8.1

(continued) IMF Program

Country

Devaluation Date

Inception Date

Expiration Date

47. Yugoslavia 48. Zaire

23 Jan. 1971 24 June 1967

22 Feb. 1971 6 July 1967

21 Feb. 1972 July 5, 1968

Sources: Pick’s Currency Yearbook and IMF Annual Reports and Reports on Exchange Resrricrions. Gmceled on 2 December 1959, when a new standby arrangement commenced. bCanceledon 16 May 1962. ‘Canceled on 9 February 1958, when a new standby arrangement commenced. dCanceled on 29 July 1971, when a new standby arrangementcommenced.

specific contents of these programs, however, were not homogeneous across our sample of twenty-six IMF programs; the concept of phasing (i.e., drawings in installments) was not introduced until 1956, and we had to wait until 1958 to observe whether drawings were made conditional on performance criteria.I0 Table 8.2 contains the percentage change in the (official) exchange rate the year of the crisis and in the three subsequent years. As pointed out, all the countries in our sample devalued their currencies by at least 14% after having maintained a fixed (official) exchange rate with respect to the U.S. dollar (or a stable managed float, like in the cases of Peru in 1958, Argentina in 1962, and Brazil in 1967) for two or more years. The non-IMF episodes are shown in panel A; the average depreciation of the exchange rate was 61%, while the median was 39%. The IMF programs are shown in panel B; on average, they devalued by 52% (43% was the median). As can be seen, while most of these countries returned to a fixed (or almost fixed) exchange rate, a few decided to follow a crawling peg system after the crisis. Also, through the years, some of the countries in our sample suffered recurrent crises and devaluations.” The nature of these devaluations and the specific circumstances that surrounded the episodes were very diverse. Some of the devaluations occurred in a unified exchange system and consisted in de jure modifications of the par value or the gold content of the currency agreed with the IMF. More frequently, however, there were de facto devaluations in which the parity of the official exchange rate was maintained but the depreciation was effected by the introduction of a regime based on multiple exchange rates. In other episodes, devaluations were implemented through the unification of multiple nominal exchange rates 10. For discussions of the evolution of IMF policies, see Dell (1981), Guitiin (1981). de Vries (1976, 1987), and Horsefield (1969). 1 1 . From today’s perspective, it is paradoxical to see Spain among the developing countries. However, during the Bretton Woods period, Spain’s per capita income was similar to that of many developing nations.

414

Sebastian Edwards and Julio A. Santaella

Table 8.2

Devaluation Crises and the IMF in the Bretton Woods Period Rate of Devaluation (exchange rate in local currency units per dollar)

Country

Year

Devaluation Year

1 Yr. After

2 Yr. After

3 Yr. After

1955 1970 1962 1965 1971 1966 1962 1967 1971 1967 1960 1964 1967 1955 1955 1967 1967 1967 1967 1958 1959 1964

158.2 14.3 130.5 50.0 78.2 58.7 66.7 16.7 20.0 16.0 30.0 96.7 16.7 40.0 43.2 15.9 16.2 24.1 16.0 221 .o 175.0 38.2

3.7 25.0 25.6

-1.2

8.1

7.2 16.7 - 10.2 1.1

29.4 7.1

1959 1962 1967 1957 1962 1967 1961 1961 1970 1962 1967 1970 1954 1958 1967 1962 1970 1959 1964 1970 1963

108.1 61.5 22.3 116.1 34.3 16.7 17.7 20.0 38.9 23.9 42.9 16.0 44.5 28.9 44.4 94.1 63.7 42.3 23.8 65.0 45.5

A . Non-IMF devaluers

Argentina Argentina Chile Colombia Ghana India Israel Israel Israel Jamaica Korea Korea Malawi Nicaragua Pakistan Sierra Leone Spain Sri Lanka Trinidad-Tobago Turkey Uruguay Venezuela B. IMF devaluers Argentina Argentina Brazil Colombia Colombia Colombia Costa Rica Ecuador Ecuador Egypt Ghana Indonesia Mexico Peru Peru Philippines Philippines Spain Tunisia Turkey Uruguay

.o

-29.6 - .4

.o .o .o

.9 100.0 6.3

.o .o .5

.9 .2

.o

.9

.o .o .o - .7

- 1.2 41.1 18.0

.o

7.1

.o .o .o .o .o

9.8

.o

14.3

.o

- .3

.o .o .o

-5.2 18.7

.o

.o .o .o

.o .o

- .9

.o .o

- .7

42.9 .3

- .2

1.3

- .6 - .7

.3

.3

- .5

.5 - .7

.3

.o .o .o

.o .o .o .I

.o

.o .o .o

.o .o .o

.4 13.9 26.6

61.5 24.9 2.1 4.7 50.0 6.9

.o

.o 5.7

.o .o .o .o .o .o .o

- 3.6 .o

.o

.o .o .o .o .o .o .o .o .o .o

5.4 - .2

- .8

215.8

26.7

.o .o

- .o

.o .o

415

Devaluation Controversies in the Developing Countries

Table 8.2

(continued) Rate of Devaluation (exchange rate in local currency units per dollar) ~~~

Country

Year

Devaluation Year

Uruguay Yugoslavia Yugoslavia Yugoslavia Zaire

1971 1961 1965 1971 1967

48.0 18.7 66.7 36.0 203.0

1 Yr. After

2 Yr. After

3 Yr. After

97.8

28.0

76.6

.o .o .o .o

.o .o

-8.2

.o

.o .o

9.3

.o

Source: International Financial Statistics and Pick‘s Currency Yearbook.

or through the (temporary) withdrawal of the central bank’s intervention in the exchange market, allowing the exchange rate to float momentarily to a higher parity at which a new peg was to be maintained. In the appendix, we present a brief description of each of the forty-eight devaluation episodes, and section 8.2.3 discusses in detail the experience with parallel and multiple exchange rate practices. 8.2.2 Macroeconomic Policies and Fundamental Disequilibria Under fixed nominal exchange rates, macroeconomic policies determine whether the exchange rate chosen by the authorities can be sustained in the longer run. Under most circumstances, if macroeconomic policies become “inconsistent,” international reserves will be eroded, the real exchange rate will experience an appreciation (i.e., overvaluation), and an exchange rate crisis-that is, a devaluation-will eventually occur. From an empirical point of view, it is not trivial to determine whether, for a particular country at a particular moment in time, macroeconomic policies have indeed become inconsistent with the fixed peg. In this section, we tackle this issue by comparing the evolution of macroeconomic policy in the devaluing countries with that of a control group of twenty-four fixed rate countries.’* Table 8 . 3 summarizes the behavior of five indicators of domestic credit and fiscal policies for our two groups of devaluing countries (IMF and non-IMF countries) and for the control group: (1) the rate of growth of domestic credit (panel A); (2) the rate of growth of domestic credit to the public sector (panel 12. This approach, of course, assumes that the control group followed sustainable policies. The countries, and years in the control group are the CBte d’Ivoire, the Dominican Republic, Ecuador, Egypt, El Salvador, Ethiopia, Greece, Guatemala, Honduras, Iran, Iraq, Jordan, Malaysia, Mexico, Nicaragua, Nigeria, Panama, Paraguay, Singapore, Sudan, Thailand, Thnisia, Venezuela, and Zambia. For more details, and for some of the most important caveats in using the control group approach, see Edwards (1989~).In this paper, years included as observations in the control group have been restricted to the Bretton Woods period.

416

Sebastian Edwards and Julio A. Santaella

Table 8.3

Indicators of MacroeconomicPolicy in Forty-eight Devaluation Episodes and a Control Group of Fixers ~~

3 Yrs. Prior NonIMF

2 Yrs. Prior

~

IMF

NonIMF

IMF

NonIMF

IMF

Control Group

12.6 19.6 29.5

11.0 17.8 24.9

15.6 18.7 28.5

6.9 15.4 22.7

12.3 17.1 21.5

8.6 14.7 22.5

B . Annual growth of domestic credit to the public sector (%) lstquartile 9.4 -?om

Fig. 12.1 Annual consumer price inflation rate, relative to Germany

Source: OECD, main economic indicators.

the 1984 referendum on the “scala mobile.” With the gradual convergence of inflation, realignments became less necessary and less frequent, until they were de facto abandoned in 1987.4 In the meantime, two more countries (Spain and the United Kingdom) joined the system, while Austria and Sweden-which are prevented from formal membership because they are not part of the European Community-follow an exchange rate policy consistent with full membership. By 1991, European monetary authorities-even the same central bank that in 1979 seemed to justify an inflation rate that was rising above 20 percent-are unanimously ready to sign on to a European system of central banks whose statutes declare price stability as the main objective of monetary policy. 4.Fordiscussions of the EMS experience prior to 1987, see Collins (1988) and Rogoff (1985a).

550

Susan M. Collins and Francesco Giavazzi

The aim of this paper is to provide new evidence about the hypothesis that the recent popularity of fixed exchange rates in Europe results from a convergence in attitudes toward inflation. So far, research on the European disinflation of the 1980s has concentrated on the extent to which membership in the EMS affected expectations and thus the unemployment cost of stabilizing prices (see, e.g., Giavazzi and Giovannini 1989; and Weber 1991). There has been less discussion of the reasons that led first to the decision to join the EMS and later (in most countries) to a change in policies that made the new exchange rate regime viable. Most existing work assumes that the turnaround was induced by “enlightened” policymakers whose unyielding commitment to price and exchange rate stability eventually produced a shift in private-sector expectations. An alternative view, explored in this paper, is that consumers’ perception of the trade-off between price and output stabilization changed first. Note that the second view raises an additional issue. Why did some countries pursue disinflation independently while others, by joining the EMS, seem to have attempted to replace their domestic central bankers with the Bundesbank? This issue is also discussed briefly. The paper is organized as follows. Section 12.1 compares the Bretton Woods and the EMS experiences with inflation differentials and exchange rate adjustments. Section 12.2 discusses both the empirical methodology and the survey data of consumer expectations about the future economic performance on which the analysis is based. Section 12.3 presents empirical evidence for the hypothesis that attitudes toward inflation and unemployment have shifted within Europe. Section 12.4 develops a theoretical framework that illustrates how changes in private-sector attitudes across countries might lead to a convergence in inflation rates. The model, which follows recent work by Alesina and Grilli (1991), provides a useful context for thinking about our empirical findings. Finally, section 12.5 contains our concluding discussion. 12.1 Bretton Woods in Light of the EMS

As documented in figure 12.lb, there have been two distinct phases in the EMS in terms of the inflation experience of its members. For the first few years, membership in the exchange rate mechanism did not seem to have any effect on the high-inflation countries. The year after the system was inaugurated, the inflation differential relative to Germany increased in France, Ireland, and Italy. Convergence began only after 1982. This visual evidence is confirmed by empirical research on the effects of EMS membership. One common finding is that, to the extent that membership did affect expectations, the shift in expectations occurred with a lag: around 1983 in France, 1982 in Ireland and Denmark, and late 1984 in Italy (see Giavazzi and Giovannini 1989, chap. 5; and Webber 1991). During this first stage of four to five years, the EMS had to accommodate countries with apparently very different attitudes toward inflation. In Italy, for

551

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

example, the inflation differential with Germany remained above 10 percent until 1984. Thus, the success of the system in its early years must be attributed in large part to its flexibility: that is, to the smooth working of realignments. Revisions of central parities happened frequently and were never delayed long enough to allow the buildup of large misalignments. However, they also required the agreement of all parties in the system, thus avoiding the risk of competitive devaluations.5 Moreover, capital controls allowed devaluations to occur without financial disruptions and allowed central banks to choose the timing of realignments, instead of being forced to realign by speculative attacks on reserves. In the latter half of the 1980s, when European inflation rates and-more important, as we shall argue in this paper-European attitudes toward inflation converged, frequent realignments became unnecessary. Eventually, intraEMS exchange rates became fixed. (There has been no change in central parities since January 1987.) As financial markets came to understand that realignments were no longer necessary, the need for capital controls also vanished. It therefore became possible to lift all administrative controls on intraEuropean financial transactions. Despite some similarities, the EMS experience contrasts sharply with the Bretton Woods experience. As for Germany-the “center” in the EMS-U.S. inflation was lower than inflation in Europe, at least until the mid-1960s. On average, between 1960 and 1966, inflation (GDP deflators) was 2 percent per year lower in the United States than in France, 1.6 percent per year lower than in the United Kingdom, and 3 percent per year lower than in Italy. Even in Germany, inflation was higher than in the United States during this period, notwithstanding the German attempt to put downward pressure on domestic prices by revaluing the deutsche mark in 1961. Except for the deutsche mark episode, however, exchange rates remained fixed. When realignments happened, they were dramatic events, forced by the unsustainable external position of a member country. The U.K. experience provides a clear example of the resultant difficulties. By 1966, the United Kingdom had accumulated a loss of competitiveness vis-2-vis the United States of almost 20 percent (measured using unit labor costs). Sterling was devalued in November 1967, forced by a speculative attack large enough to burst the dam provided by British exchange controls (see Bordo, chap. 1 in this volume). Table 12.1 compares the inflation performance and the role of exchange rates changes during Bretton Woods and the EMS. The first column of the 5. A clear example is France in March 1983. Jacques Delors, then finance minister, went to the realignment meeting asking for a devaluation of the franc that was viewed as “excessive” by his colleagues. The meeting was suspended, Delors returned to Paris,and the French government, facing exclusion from the exchange rate mechanism of the EMS, had to withdraw its request and change domestic policy accordingly. In the end, the devaluation of the franc was much smaller than the French had originally requested and was accompanied by devaluations of the Italian lira, the Belgian franc, and the Danish krone.

552

Susan M. Collins and Francesco Giavazzi

’Igble 12.1

Inflation Differentials and Realignments:Bretton Woods vs EMS Bretton Woods, Cumulative Position Relative to the U.S.

Denmark: Unit labor cost Exchange rate Italy: Unit labor cost Exchange rate Ireland: Unit labor cost Exchange rate France: Unit labor cost Exchange rate Belgium: Unit labor cost Exchange rate Spain: Unit labor cost Exchange rate Germany: Unit labor cost Exchange rate United Kingdom: Unit labor cost Exchange rate Sweden: Unit labor cost Exchange rate Japan: Unit labor cost Exchange rate

EMS, Cumulative Position Relative to Germany

1960-66

1967-7 1

1979-87

1988-91

35.7 .2

2.1 6.0

39.5 22.7

7.5 1.5

8.1 .6

- 12.0 - .9

81.2 53.2

10.4 - .9

26.7 .6

- 6.2 12.1

49.2 39.5

1.1 .8

-3.4 .2

- 10.6 12.0

33.5 43.7

2.5

10.0 -.l

- 14.3 - 2.2

12.3 29.1

2.7 .3

13.4 - .1

- 14.2 13.6

44.1

19.2 -4.2

12.4 -4.1

- 10.2 - 12.7

19.8 .6

-8.5 12.1

19.2 .4

-11.0 - 1.0

.5

- 23.4 - 3.6

.7

74.0

.o

Source: OECD, national income accounts. Unit labor costs are constructed using the index of labor cost per employee and the index of productivity. Both refer to the whole economy. Note: The data for “unit labor cost” show the total change in the index of relative labor cost per unit of output between each country and either the United States or Germany, over the period indicated. “Exchange rate” is the total change in each country’s exchange rate relative to the dollar or the deutsche mark over the period indicated-a positive sign indicates a cumulated depreciation.

table confirms that, until the mid-l960s, Bretton Woods central parities were rarely changed: our sample of countries reports only the 1961 deutsche mark revaluation. In the late 1960s, realignments became more frequent. However, there was no convergence of inflation rates. Instead, as U.S. inflation accelerated, the relative trend of unit labor costs reversed.

553

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

The third and fourth columns of 12.1 offer a comparison with the EMS. During the early 1980s, when inflation differentials relative to Germany were large, realignments avoided the buildup of large misalignments. Denmark provides a clear example of the difference. Between 1961 and 1966, the krone appreciated by 35.7 percent relative to the dollar, in real terms. Eventually, this was corrected in part when Denmark joined the sterling devaluation of 1967. A similar inflation divergence relative to Germany between 1979 and 1987, however, was accommodated by frequent devaluations, thus avoiding a significant real appreciation. Finally, the last column shows that, in the late 1980s, the move toward more fixed exchange rates was accompanied by a sharp convergence in inflation. Thus, as seen from the perspective of the EMS experience, Bretton Woods failed on two accounts. First, the system lacked the necessary flexibility to accommodate countries with different inflation rates. Second, although this is not a fault that can be ascribed to the design of the system, the Bretton Woods years did not see the convergence of attitudes toward inflation that characterized Europe in the late 1980s and that we believe is a necessary condition for the survival of a fixed rate regime. In the remainder of this paper, we first examine some new evidence that supports our view about the convergence of attitudes toward inflation. We then offer a theoretical framework that highlights the links between attitudes and the choice of an exchange rate regime. 12.2 Empirical Analysis

The empirical analysis examines two issues. First, is there any evidence of shifts in attitudes toward inflation versus unemployment within EMS member countries? Second, if so, when did any such shifts occur? In particular, we ask if there is any evidence of increased concern about inflation among countries that gave up some monetary sovereignty in following Germany’s leadership of the multiple peg system. (Note that simply joining the EMS in 1979 need not have entailed any such change.) We would also like to know when any such shifts occurred. For example, if attitudes did change, did this occur before or after the EMS was instituted? Did it occur before or after changes in actual inflationary policy and performance that some countries experienced? We assume that residents in each country have an expected loss function that depends on expectations about future inflation and unemployment. We use survey data from European households to provide information about how inflation and unemployment affect their assessments of general economic conditions, where the latter is interpreted as a measure of their “expected loss.” We then use regression analysis to infer the implied weights on inflation and unemployment in this loss function and to look for changes in these weights over time. Following a description of the data and methodology in this section, the empirical results are presented and discussed in section 12.3.

554

Susan M. Collins and Francesco Giavazzi

12.2.1 The Data To examine these issues, we use survey data on expectations about future economic performance from the European Community’s survey of consumer opinion, reported in European Economy, supplements B and C . 6 Surveys of households are taken three times a year (January, May, and October) in each of eight countries. The complete sample is available since 1974.’ These data can be interpreted as information about the average household in each country. Three of the survey questions are relevant for our purposes. The first asks respondents their perceptions of prospects for the general economic situation in their country over the next twelve months, relative to the current situation in their own country. The second asks their expectations about the changes in the trend of the price level (inflation) over the next twelve months, and the third asks their expectations about changes in the unemployment rate over the next twelve months.* Responses for these two questions are also relative to the current situation in the respondent’s country. The published indicators are weighted sums of these responses. Each series ranges between - 100 and 100, but the scale differs across questions. In particular, if respondents, on average, expected the inflation rate, the unemployment rate, and general economic conditions all to be the same over the next twelve months as they had been recently, then the indicators would be 50, 0, and 0 respectively (see n. 8 above). Table 12.2 shows the means and standard deviations of each variable for the eight members of the European Community. Looking first at general economic prospects (EP), the table shows that, on average, respondents were pessimistic. The average response is less than zero for all eight countries, implying that, on average, economic conditions were expected to deteriorate. France, Belgium, and Ireland appear to be the most pessimistic, with Ger-

+

6. We look at the nine countries that were members of the European Community during the 1970s, with the exception of Luxembourg, for which these data are not available. In some cases, the reported figures average across months. We obtained the actual figure for the relevant months from the European Commission. However, these data were in each case identical to the published figures. 7. The survey is given to a random sample of twenty-five hundred adults, most of whom are household heads, in each country in January and May and to five thousand adults in October. (Note that we do not treat the October observations differently in the empirical analysis.) For additional discussion of the survey, see Papadia and Basano (1981). 8. The three questions and possible responses are as follows: (1) “General economic situation in your country, prospects over the next twelve months?” The possible responses are “a lot better” (coded as I), “a little better” ( + %), “the same” (0). “a little worse” ( - Y2),“a lot worse” ( - l), or “don’t know.” (2) “Price trends over the next twelve months?” The possible answers are “more rapid increase” ( l), “same increase” (+ %), “slower increase” (0), “stability” ( - %), “fall slightly” ( - I), or “don’t know.” (3) “Unemployment level in your country in the next twelve months?” The possible responses are “increase sharply” ( l), “increase slightly” ( %), “remain the same” (0), “fall slightly” ( - %), “fall sharply” ( - l), or “don’t know.” For each variable, the “don’t know” responses are redistributed between the other answer categories according to the latter’s percentage distribution.

+

+

+

+

555

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

Table 12.2

Means and Standard Deviations: Household Expectations

Germany France Italy Belgium Denmark Netherlands United Kingdom Ireland”

General Economic Prospects

Expected Price Trends

Expected Unemployment Trends

- 9.08 ( 12.45) -23.78 (11.59) - 16.10 (13.07) - 25.14 (16.64) - 14.90 (15.10) - 15.82 (10.14) -11.45 (14.18) - 25.09 (16.59)

32.06 (13.17) 30.39 (14.07) 48.37 (8.27) 33.10 (8.86) 19.39 (14.82) 34.73 (1 8.97) 37.14 (12.56) 40.19 (15.50)

16.90 (16.52) 35.63 (14.73) 45.98 (10.10) 40.47 (18.70) 19.78 (17.56) 35.02 (28.39) 28.63 ( 18.54) 35.68 (19.10)

Source: European Economy, Supplements B and C, various issues, May 1974-May 1990. Note: Standard deviations are given in parentheses. ‘January 1975-May 1990.

many and the United Kingdom the least pessimistic. But recall that respondents in each country are asked about their expectations relative to recent performance in their own country. Therefore, responses for different countries are not on comparable scales. Cross-country comparisons must be interpreted with caution. Turning next to expected price changes ( P T ) , table 12.2 shows that Italians on average expected inflation rates to be about the same as in the past. (The mean response is close to 50.) Respondents in other countries expected inflation to slow somewhat (responses between 0 and 50), with the greatest slowdown in inflation expected in Denmark. Italy and Belgium-not Germanyare the countries with the least variation in expectations about price trends. Finally, table 12.2 shows that all countries, on average, expected unemployment to rise. Italians expected the largest increase in unemployment, on average, and Germany the smallest. 12.2.2 Empirical Methodology We interpret expected economic prospects as a measure of (minus) the expected loss function for residents in each country. We also assume that expected general economic conditions are a function of expected inflation (price trends) and expected unemployment: the greater the expected price increases

556

Susan M. Collins and Francesco Giavazzi

and the expected unemployment, the worse anticipated general economic condition~:~ (1)

-3 = E P

=

F(PT’, UP),

where E P is expected general economic situation in country i, P P is expected price trends in country i, and U P is expected unemployment in country i . Our next step is to take a linear approximation to the loss function F ( . ) . This equation is used in the estimation. (Future work will consider other specifications, such as a quadratic function.) The weights on PT and UT in determining EP provide indicators about attitudes toward inflation and unemployment. As discussed above, these survey responses are not directly comparable across countries. However, a decrease in a given country’s tolerance for unemployment relative to inflation should imply a fall in the weight (a smaller negative weight) on UT relative to PT in determining general economic prospects. Thus, changes in attitudes should imply structural shifts in equation ( l).l0 Note that we will estimate the actual weights that respondents placed on expected unemployment and expected price trends, not just the relative weight they placed on unemployment in the loss function (b in the model presented in sec. 12.4 below). This relative weight can be constructed from a ratio of parameters. Suppose that we wished to examine whether a shift in attitudes occurred at a given date s. We could simply construct a dummy variable that was zero before s and unity afterward, enter it interactively with PT and UT, and test whether the coefficients on these interacted variables differ from zero, in the appropriate directions. However, we do not wish simply to test for a prespecified breakpoint; we wish to look at when any shifts in attitudes occurred. These breakpoints need not occur in all countries and need not occur at the same time for each country that experienced shifts. To look for the timing of any structural shifts, we estimated a series of equations for each country, allowing the breakpoint to range from January 1976 to May 1989.” Thus, we estimated the following equation for each country:

9. Ideally, we would analyze survey resonses to questions about whether inflation or unemployment is considered a more serious problem. Such data are available for the United States and have been studied by Fischer and Huizinga (1982). However, these data are not available, in a time series, for European countries. 10. It is important to point out that there are alternative interpretations for changes in the coefficients in (2) below. One other possibility is that “general economic prospects” is really an indicator of expectations about future economic growth instead of an indicator of expected welfare. If so, changes in respondents’ perceptions about the structural relations among inflation, unemployment, and growth would cause the coefficients to change. In general, it is very difficult to distinguish between these two interpretations. 11. This formulation assumes that any possible change in attitudes occurred all at once. It would also be interesting to look for gradual changes in attitudes, e.g., using a specification that allows for time-varying parameters.

557

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

where D = 1 after the breakpoint, 0 otherwise. In most cases, there was evidence of serial correlation. Our estimations correct for first-order autoregressive error terms. We compare the values of the likelihood function across possible dates for a breakpoint to see which specification best fits the data. A convenient indicator for these comparisons is the posterior odds ratio, PR(S), for each possible breakpoint s. This ratio can be interpreted as the likelihood that the breakpoint occurred at time s relative to the likelihood that it occurred at s*, where s* is the breakpoint at which the value of the likelihood function is maximized.” PR(S) is equal to one for s = s* and is bounded between zero and one for all other dates, s. The ratio is useful for two reasons. First, it provides some information about how well the data can distinguish between alternative dates for the breakpoint. (For example, values close to one for many dates would suggest that it is difficult to pinpoint when the break occurred. Values close to zero at most dates, with a spike at one date, would suggest a clearly identifiable breakpoint. Multiple spikes would suggest more than one shift in attitudes within the sample period.) Second, since the ratio always ranges from zero to one, it facilitates cross-country comparison of the degree of confidence about the timing of shifts in attitudes about inflation and unemployment. For each country, we show a plot of the values of the posterior odds ratio across breakpoints and report the maximum likelihood parameter estimates and diagnostic statistics at breakpoint(s) corresponding to peaks in the value of the likelihood function. Estimates for C,, C, and C, provide information about shifts in attitudes. A negative (positive) estimate for C, can be interpreted as increased pessimism about overall economic prospects, given expectations about inflation and unemployment. Negative estimates for C, and C, can be interpreted as increases in the weights that respondents placed on expected unemployment and on expected inflation, respectively. And (C, + C,)/(C, + C,) > C,/C, can be interpreted as a decrease in the relative weight placed on unemployment expectations or as an indication that respondents have become willing to tolerate more unemployment in return for lower inflations.

12. The strong serial correlation suggests that there may be omitted variables in our equations to explain expectations about next year’s economic conditions. We tried including actual levels of inflation and unemployment for each country and an index of real oil prices. However, this did not significantly reduce the serial correlation problem. Note that including these variables did not qualitatively change the results discussed in the text. 13. The posterior odds ratio for a breakpoint at date s is defined as PR(S) = exp [LF(S) - LF(s*)], where LF(S) is the value of the log likelihood function, given a breakpoint at date s, and LF(S*) is the value of the log likelihood function at the breakpoint s*, where it is maximized. Thus, PR(S*) = 1. Our procedure is similar to the one followed by Mankiw, Miron, and Weil(l987). 14. While we discuss changes in the estimated 6-weight on unemployment relative to inflation-the estimated change in this ratio was statistically significant only in the cases of Belgium and Ireland.

558

Susan M. Collins and Francesco Giavazzi

12.3 Estimation Results Is there any evidence of a shift in attitudes about inflation among EMS member countries? If so, when did it occur, and in which direction was the change? This section discusses the results from our empirical analysis for each country. A summary and interpretation of these results is given at the end of the section. 12.3.1 France Figure 1 2 . 2 ~shows the values of the posterior odds ratios for alternative breakpoints for France. It shows that the likelihood function reaches a maximum when the breakpoint is October 1979. This suggests that there was a change in attitudes during the first year the EMS was in operation-not before it was instituted. There are also smaller peaks in the early 1980s. (The value of PR[S = May 19831 = .43, which can be interpreted as the likelihood that a shift occurred in May 1983 relative to the likelihood that it occurred in October 1979 is 43 percent. Similarly, PR[S = October 19861 = S 3. ) This suggests that there were additional shifts in attitudes during the early years of the exchange rate system. This result is quite interesting, in light of the fact that it was not until after 1983 that French monetary policy and inflation performance began to converge to policy and performance in Germany. The first two columns of table 12.3 show parameter estimates for France assuming an October 1979 breakpoint. The equations fit quite well. They show that, on average in the 1970s, French respondents put about three times the weight on expected unemployment that they placed on expected inflation in forming their assessments of general economic prospects. (C,/C, is about three.) The first column shows that, after October 1979, there is weak evidence that respondents became more pessimistic about general economic prospects (C, < 0) and that the weight they placed on price expectations increased (C, < 0). However, neither of these estimates is statistically significant. It is possible that there are not enough data to distinguish between a change in attitudes toward inflation and an increase in pessimism more generally. In the second column, the latter change is ruled out ( C , is constrained to equal zero). These estimates do show a statistically significant increase in the weight on inflation. They suggest that the weight on inflation relative to the weight on unemployment rose somewhat, from 0.28 before October 1979 to 0.39 afterward. Of course, this result raises the question of why French voters elected FranGois Mitterand, and his very expansionary platform, in 1981. The third column of table 12.3 provides estimates assuming a breakpoint in May 1983. Here, C, is significantly negative, suggesting that French respondents did anticipate less positive economic conditions in general as French macroeconomic policies under Mitterand became more restrictive. These results suggest that, before May 1983, the relative weight on unemployment

559

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

was 2.2 and provide weak evidence of a decrease after May 1983 to 1.4. These findings are consistent with the view that French households’ concerns about inflation increased gradually during 1982-83. 12.3.2 Italy Figure 12.26 shows the values of the posterior odds ratios for Italy across alternative breakpoints. Here, there is a clear, single spike of the likelihood function in October 1978. Interestingly, this is a few months before the EMS began operation. There is little evidence of additional shifts later on. The last column of table 12.3 shows parameter estimates assuming an October 1978 breakpoint. Again, the overall fit of the equation is quite good. In contrast to respondents in France, Italian respondents became more optimistic on average about general economic performance during the EMS period. Like French respondents, Italians appear to have become less willing to tolerate inflation after October 1978. The weight on price trends rises, and the shift is statistically significant-even allowing for a change in the constant term. The weight on unemployment relative to inflation declined from 0.7 to 0.5 after the breakpoint. 12.3.3 Germany Figure 12.2.c plots the posterior odds ratio across alternative breakpoints for Germany. It shows that the likelihood function peaks for breakpoints during October 19884ctober 1989. It also gives some evidence that attitudes were shifting in 1984-86. However, there are only six observations from October 1988 to the end of our sample, providing little information about such a recent shift in attitudes. We also wished to allow for a possible shift in attitudes following the collapse of the Berlin Wall in late 1989. Thus, in addition to looking at parameter estimates for breakpoints in October 1988 and January 1986, we also allow the coefficient on price trends to shift in October 1989, together with a general shift in attitudes in January 1986. Table 12.4 presents the three sets of estimates. Those for an October 1988 breakpoint are in column 1. They point to results that we did not expect. German respondents’ tolerance for inflation relative to unemployment appears to have increased sharply. Respondents also appear to have become more pessimistic about general economic prospects at the end of the 1980s. However, these results are somewhat strange in two respects. First, they suggest that Germany did not care at all about price trends after October 1988 (C, + C, is close to zero). Second, they suggest that, as German respondents expected unemployment to increase, they became more sanguine about general economic prospects (C, + C, > 0). Both may be an artifact of assuming that only one shift in attitudes occurred during the sample. The second and third columns of table 12.4 assume an earlier breakpointJanuary 1986. In the middle column, we allow for only one shift, while, in the last column, we allow for an additional shift-attitudes toward inflation

560

Susan M. Collins and Francesco Giavazzi

12 1 08

-

06

-

04

-

9

1 08

-

06

-

04

-

02

I

~

0 1

A,

A,

J

12 1 -

08

-

06 04

-

02

-

9

;iI

08

0.0

Jan 76

!

May84

Jan86

On87

A May 89

Fig. 12.2 Posterior odds ratio. a, France; b, Italy; c, Germany; d, Belgium; e, Netherlands; f,Ireland; g, Denmark; h, United Kingdom.

561

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

12-

1 -

08

-

06

-

1 -

562

Susan M. Collins and Francesco Giavazzi

Table 12.3

EP = C,

+ C;

Estimation Results: France and Italy UT + C ; P T + D { C , + C; UT + C;F’T} France

Breakpoint

Oct. 1979

Oct. 1979

May 1983

Italy, Oct. 1978

co

6.334 (1.376) - ,563 (7.256) - ,196 (-2.015) -4.213 (-,781) .041 ( .404) -.118 ( - 1.040) .511 (4.015)

3.255 (1.388) - ,532 (-7.759) - ,148 ( - 1.969)

8.406 (2.132) - ,600 ( - 8.738) - ,270 (-3.324) -9.680 ( - 1.795) ,146 (1.379) - ,049 ( - ,421) ,623 (5.403)

25.678 (3.345) - ,434 (-4.731) - ,656 (-4.333) 35.208 (3.79 1) - ,083 ( - ,621) - ,364 ( - 1.922) ,449 (3.325)

c, c 2

c3 c 4

c 5

Rho In L

- 129.90

.814

R2a

No. of obs. C,‘C2 (C, + C,)/(C,

49

+ C,)

2.9 1.9

...

- .010

( - ,124)

- ,172 (-1.955) .485 (3.765) -

130.23 ,800 49 3.6 2.6

- 130.736

- 129.027

,765 49 2.2 1.4

,856 49 .7 .5

Nore: The table shows maximum likelihood estimates, correcting for the first-order autocorrelation. “Rho” is the first-order autocorrelation coefficient. The sample is from May 1974 to May 1990. &statistics are given in parentheses. D is a dummy equal to zero before the breakpoint and one from then on. ’Based on transformed data, using autocorrelation coefficient, rho.

are allowed to shift again in October 1989 in response to developments in the former East Germany. We focus on the latter, which suggests three conclusions. First, German respondents do appear to have become more pessimistic, on average, about their economy in the mid-1980s. Second, the weight that Germans placed on expected price trends declined significantly in the mid1980s. In fact, C, + C, is close to zero. (There was no change in their weight on unemployment.) Increased German tolerance of inflation may help resolve the puzzle of why Germany has been willing to stay in an EMS that leads to higher German inflation. Finally, there appears to have been a further reduction in the weight that German respondents placed on price movements after October 1989. This is consistent with West Germans expecting German unification to increase inflation but, at the same time, feeling more positive about the likely performance of their economy. Unfortunately, there are too few observations after October 1989 to distinguish between a general shift in optimism and a shift in attitudes toward inflation.

563

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

Table 12.4

Estimation Results: Germany

EP=C,,+C,-UT+C,*PT+D(C,+C,*UT+C,.PT}+D,.C,*PT ~

Breakpoint

Oct. 1988

Jan. 1986

7.049 (3.153) - .576

- 12.058 ( - 1.374) ,750 (2.632) ,237 (.851)

12.440 (3.121) - ,553 ( - 9.505) - .378 ( - 3.268) - 10.612 ( - 2.155) .082 (.517) ,491 (2.636)

.381 (2.741)

(3.211)

- 133.375 .834 49

- 134.933 ,811 49

(-11.760)

- .229 ( - 3.267)

Rho

2.52 21.75

,444

1.46 -4.17

Jan. 1986, Oct. 1989 (PT only) 12.528 (3.921) - ,583 ( - 12.551) - ,367 (-4.023) -9.448 ( - 2.360) ,052 (.375) ,334 (2.046) ,309 (3.061) ,239 (1.604)

- 130.848 ,872 49 1.59 16.1

Nore: The table shows maximum likelihood estimates, correcting for first-order autocorrelation. “Rho” is the first-order autocorrelation coefficient. The sample is from May 1974 to May 1990. r-statistics are given in parentheses. D is a dummy equal to zero before the breakpoint and one from then on. D,is a dummy equal to zero before October 1989 and one from then on. ‘Based on transformed data, using autocorrelationcoefficient, rho.

12.3.4 Belgium and the Netherlands These two countries did not pull out of the Snake in the 1970s (see n. 3 above) but chose to follow German leadership. They also resemble Germany here in that parameter estimates in each imply a rise in relative tolerance for inflation. Looking first at Belgium, the plot of the posterior odds ratio in figure 12. U shows an initial peak in January 1981, with some evidence that attitudes had been shifting during the previous two years, and then a much larger peak in October 1985. The parameter estimates presented in table 12.5 show a significant increase in the weight on expected unemployment after January 198 1. After October 1985, there is evidence that the weight on inflation declined (the point estimate is close to zero). Turning next to the Netherlands, figure 12.2e shows that there appear to be a number of candidates for a breakpoint. In fact, all show attitudes shifting in

564

Susan M. Collins and Francesco Giavazzi

Table 12.5

Estimation Results: Belgium, Netherlands, Ireland EP = C, + C, . LIT + C, . PT D{C,+ C, . UT + C, . PT}

+

Be1giu m Breakpoint

Rho

The Netherlands

Jan.

Oct.

Oct.

May

1981

1985

1977

1986

39.284 (1.943) - .378 (-2.606) - ,762 ( - 1.663) -29.265 (1.426) - ,145 ( - .974) ,482 (1.022) ,687 (6.526)

21.381 (3.855) - .536 (-8.181) - ,417 (-3.105) - 16.931 ( - 2.767) - ,071 ( - ,460) ,274 (1.224) ,639 (5.554)

152.006 ,733 49

- 152.352

4.531 (1.182) - ,489 ( - 8.462) - ,391 ( - 3.733) 13.612 (2.398) - ,349 (-4.212) .138 ( 3 1 1) ,527 (4.169)

6.441 (1.992) - ,569 ( - 10.976) - ,349 (-3.870) ,016

- 127.991

- 126.936

,893

,922

49 1.25 3.31

- ,058 (.415) ,315 (1.427) ,360 (2.574)

49 1.63 15.03

-

SO 1.87

Ireland, Jan. 1986 21.613 (6.694) - ,608 (-15.235) - .616 (10.564) ,656 (.112) - ,275 ( - ,074) ,296 (1.680) ,109 ( ,684)

- 130.707

.745 49 1.29 3.44

.926 49 .99 2.76

Nore: The table shows maximum likelihood estimates, correcting for first-order autocorrelation. “Rho” is the first-order autocorrelation coefficient. The sample is from May 1974 to May 1990. [-statistics are given in parentheses, D is a dummy equal to zero before the breakpoint and one from then on. ”Based on transformed data, using autocorrelation coefficient, rho.

the same direction: increasing tolerance for inflation and/or decreasing tolerance for unemployment, Table 12.5 shows estimation results assuming breakpoints in October 1977 and in May 1986. For the earlier breakpoint, none of the changes in coefficients are statistically significant. (Constraining the constant term to be the same across subperiods did not alter this result.) However, the estimates suggest that the weight that Dutch respondents placed on unemployment relative to inflation rose from 0.5 to 2.0. The estimates for the later breakpoint show a further shift, to a relative weight on unemployment of 5.0 after May 1986. 12.3.5 Ireland Ireland also seems to have experienced a decrease in tolerance for unemployment. The likelihood function exhibits a series of peaks from late 1982 through 1986 (see fig. 12.2f). The last column of table 12.5 gives the parameter estimates assuming that the breakpoint was January 1986. These show

565

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

that the weight on price trends declined in the second subperiod while the weight on unemployment rose. Both coefficient estimates differ significantly from their estimated values in the earlier subperiod. The weight that Irish respondents placed on unemployment relative to inflation rose from slightly less than 1.0 before January 1986 to 2.8 afterward. 12.3.6 Denmark As shown in figure 12.2g, there appear to be two breakpoints for Denmark-an early one in 1976 and one a decade later. Table 12.6 presents parameter estimates for each. The early breakpoint suggests that, in the mid1970s, Danish respondents placed little weight on inflation in their general economic assessments-the relative weight on unemployment was 1 1 . 1 . This changed dramatically, as the relative weight on unemployment fell to just 0.5 after May 1976. (However, we have only a few observations in the earlier subperiod.) The second set of estimates, with a January 1986 breakpoint, is quite different. They point to a rise in Danish tolerance for inflation and a decline in tolerance for unemployment. (This shift resembles the ones discussed above for Germany, Belgium, the Netherlands, and Ireland.) Perhaps the relatively low explanatory power of the regressions for Denmark is due to the assumption of a single shift in attitudes. Without additional information about the early 1970s, it is difficult to test for multiple breakpoints. Instead, we allow the coefficient on PT to change only in January 1976, assuming a general breakpoint in January 1986. These results are reported in the third column of table 12.6. They imply that, prior to 1976, Danish respondents placed about the same weight on inflation and unemployment. During 1976-86, their concern about inflation increased significantly (C, < 0).After 1986, they became significantly more pessimistic about their economy overall. There is also weak evidence that the relative weight placed on unemployment rose somewhat. 12.3.7 United Kingdom Finally, we look at the United Kingdom, the only country in our sample that was not a member of the exchange rate mechanism of the EMS. Figure 12.2h shows that the value of the likelihood function peaks for a breakpoint in May 1987.15Estimates for this shift date are given in the last column of table 12.6. These estimates suggest that British respondents became more optimistic about their economy overall after 1987 but that there was a large and significant increase in their tolerance for unemployment relative to inflation. (The relative weight on unemployment falls from 1.2 before 1987 to 0.3 more recently.) 15. Interestingly, the 1987 breakpoint coincides with the abandonment of the United Kingdom’s medium-term financial strategy.

566

Susan M. Collins and Francesco Giavazzi

Table 12.6

EP = C,

Estimation Results: Denmark and the United Kingdom

+ C, . UT + C, . PT + N C , + C, . UT + C, . PT} + 0 , . C, . PT Denmark

Breakpoint

c 2

May 1976

Jan. 1986

-5.977 ( - ,679) - .522 ( - 3.532) - ,047 ( - .229) 10.908 ( 1.3 13) .120 (.722) - ,731 ( - 3.022)

- 10.883 (2.861) - .351 (-4.349) - .733 (-6.572) - 11.228 (-2.121) - ,225 ( - 1.236) ,191 (.351)

..

C6

,843

Rho

( 10.990)

,629 (5.497)

Jan. 1986, Jan. 1976 (PT only)

United Kingdom, May 1987

11.157 (2.315) - ,379 (-5.110) - ,361 ( - 2.692) - 13.360 (-2.537) - .219 (-1.260) ,287 (.565) - ,471 (-3.346) ,812 (8.908)

24.999 (6.631) - ,585 ( - 6.576) - ,480 (-4.753) 38.482 (2.257) ,141

- 151.240

- 151.750

- 147.020

,644

,665

,695

49 11.11 .52

49 .48 1.06

49 1.05 8.08

( ,544)

1.059 ( - 2.762) -

...

,134 (.844)

- 160.492 ,741 49 1.22 .29

Note; The table shows maximum likelihood estimates, correcting for first-order autocorrelation. “Rho” is the first-order autocorrelation coefficient. The sample is from May 1974 to May 1990. &statistics are given in parentheses. D is a dummy equal to zero before the breakpoint and one from then on. D,is a dummy variable equal to one before 1976 and zero from then on. ’Based on transformed data, using autocorrelation coefficient, rho.

12.3.8 Summary and Discussion These empirical results provide surprisingly strong support for the hypotheses advanced at the beginning of this paper. Shifts in attitudes about unemployment relative to inflation can help explain the “success” of the EMSincreased convergence of monetary policies and inflationary performance among members. To summarize, Italians became more tolerant of unemployment relative to inflation shortly before the EMS came into operation. Unemployment tolerance also increased in France, but the shift appears to have begun somewhat later, with most of the change occurring after the EMS was already in operation. (In both countries, the shift comes after the failure of the Snake.) In contrast-and surprisingly-Germany seems to have experienced a rise in relative tolerance for inflation in the mid-l980s, in the sense that

567

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

expected inflation became a less important determinant of overall assessments of the future economic situation. This was reinforced at the end of the decade, perhaps by German unification. Belgium and the Netherlands-both members of the Snake and perhaps “closer” to Germany-as well as Ireland seem to have had a similar rise in their tolerance for inflation. However, their shifts appear to have occurred earlier than the shifts in Germany. Denmark experienced both a decline in tolerance for inflation in the mid-1970s and a subsequent reversal in the mid-1980s. Last but not least, the United Kingdom has experienced a fall in tolerance for inflation at the end of the 1980s, consistent with its late entry into the exchange rate mechanism of the EMS.

12.4 Attitudes toward Inflation and the Choice of an Optimal Central Banker The previous section has presented evidence of shifts in attitudes toward inflation and unemployment in Europe since the late 1970s. In particular, there appears to have been some convergence in attitudes, in the sense that initially high-inflation countries (such as France and Italy) have become less tolerant of inflation relative to unemployment while attitudes in traditionally low-inflation countries (such as Germany) have shifted in the opposite direction. This section develops a theoretical framework to illustrate how such shifts in private-sector attitudes might make fixed exchange rates more likely. The model is based on the new theories of economic policy that suggest that societies concerned about the stability of both prices and output may solve the dilemma between the lack of credibility of an “activist” central banker and the cost of giving up monetary policy as a stabilization tool by delegating the conduct of monetary policy to an independent central banker whose preferences for output and price stability are slightly more “conservative” than society’s.l 6 More specifically, we follow recent work by Alesina and Grilli (1991) that assumes that citizens elect central bankers by majority rule. They have shown that, if citizens were to vote on which type of central banker to appoint, they would choose one who values price stability more than the median voter.’’ Next we study the consequences of a shift in consumers’ attitudes toward inflation. We ask the following question. Assume that, after the shift in preferences, the first-best outcome (the election of a new central banker) is ruled out. Then there are two options: (1) keep the current central banker, who is no longer optimal since tastes have changed, or (2) force the domestic central 16. For an early statement of this view, see Rogoff (1985b). Empirical evidence can be. found in Grilli, Masciandaro, and Tabellini (1991). 17. The original work by Rogoff (1985b) was not cast in the framework of a voting equilibrium: he simply showed that society’s welfare can be improved by appointing a central banker whose preferences differ from those of society.

568

Susan M. Collins and Francesco Giavazzi

banker to relinquish domestic monetary sovereignty, for example, through a commitment to peg the exchange rate. 12.4.1 Voters’ Preferences and the Optimal Central Banker The starting point is the time-consistency problem illustrated by Kydland and Prescott (1977) and Calvo (1978). Following Barro and Gordon (1983), we assume that the time-consistency problem arises from the central banker’s attempt to steer the economy toward higher output growth. The preferences of the central banker are described by the loss function: L

(3)

=

E[p2 + b(x - /c)~],

where E(.) is the expectation operator, p the inflation rate, and (x - k) the deviation of the level of output from the central banker’s target k. The parameter b measures the weight that the central banker attaches to output fluctuations, relative to fluctuations in the level of inflation. Output is determined by an expectational Phillips curve:

+

x = ( p - Ep) e, (4) where the “natural” rate of output has been assumed to be equal to zero, the elasticity of output with respect to unexpected inflation has been assumed to be equal to one, and e is an i.i.d. real shock with mean zero and variance u : . Expectations are formed-and wages are negotiated-before the central banker sets the inflation rate; the realization of the random shock e is known to the central banker when monetary policy is set but not to wage setters when contracts are signed. It is well known that, in this setup, the time-consistent levels of inflation and output are

b

1

+ be’

Equations (5) and (6) illustrate the time-consistency problem. If the central banker attempts to steer output away from the natural rate, the average rate of inflation is positive (its optimal level in [3] is zero), with no gains in terms of output stabilization. The trade-off between average inflation and the variance of output, u: = $/(1 b)Z,depends on the value of the parameter b. The lower is b in the central banker’s objective function, the lower is inflation (on average), but the higher is the variance of output. If b were equal to zero, (average) inflation would be eliminated, but monetary policy would loose any ability to stabilize output. Suppose now that voters were able to elect a central banker who, during her term of office, could freely pursue her preferred monetary policy-that is, once elected, the central banker cannot be recalled until the end of her term. Voters differ with respect to the relative weight that they attach to inflation and to output stabilization. Voter i’s preferences are described by a loss function

+

569

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

identical to (3) but with a relative weight bi on the two objectives. Under majority rule, the central banker elected will be the one preferred by the median voter, that is, by the voter characterized by the median value of bi: b,. As shown by Alesina and Grilli (1991), the first-order condition that determines the type of central banker chosen by the median voter is

(7)

b* * k2 -

[

(1

+a’b*)’ ] . (b,,, - b*) = 0.

The parameter b*, which characterizes the preferences of the “optimal” central banker from the viewpoint of the median voter (and thus of the central banker who will be elected by majority rule), is a function of the preferences of the median voter, b,, and of the variance of real shocks. Equation (7) shows that, for u: > 0, b* is always positive but smaller than b,: as originally shown by Rogoff (1985b), the median voter has an incentive to appoint a central banker who is more “conservative” than she herself is-that is, a central banker who values fighting inflation more than the median voter does.’* 12.4.2 The Choice to Relinquish Monetary Sovereignty Consider now the effects of a change in voters’ preferences. Let us assume that the distribution of preferences across voters shifts so as to result in a lower value of b,: the median voter becomes relatively more concerned about inflation.I9 As b, falls, so does b*, according to equation (7).*O The obvious outcome of a change in voters’ preferences is that, at the end of the term of office, a new central banker is elected whose preferences reflect the change in voters’ concerns for output and price stability. For the purpose of our discussion, however, we are interested in studying the case when the first-best outcome (the election of a new central banker) is ruled out. We consider two options: (a) keep the current central banker, who is no longer optimal but cannot be removed, or (b) short-circuit the independence of the domestic central banker by signing an international agreement that implies relinquishing domestic monetary sovereignty, for example, a commitment to passively peg the exchange rate to a foreign currency, made credible by a sufficiently high political cost of abandoning the peg. What are the costs and benefits of option b? If the shift in voters’ prefer18. The “independence” of the central banker during her term of office is crucial to the result, as it amounts to a form of precommitment. If the central banker could be recalled before the end of her term, it would be impossible to improve on the time-consistent equilibrium corresponding to b = b,, which, by the first-order condition ( 5 ) , is inferior to the equilibrium corresponding to b = b*. The length of the central banker’s term of office is clearly crucial because, when a new one is elected, b* can change. What is relevant, in the framework of this model, is that the length of office be at least as long as that of wage contracts. 19. In our empirical work, we test for a change in consumers’ preferences assuming that the median and the average voter-the only ones we can observe-coincide. This obviously depends on the distribution of preferences. 20. It is straightforward to show, from (7). that the derivative of b* with respect to b, is positive.

Susan M. Collins and Francesco Giavazzi

570

ences reflects an increased concern for inflation, then replacing the domestic monetary authority (elected at a time when inflation was not perceived as such a serious problem) with a foreign central banker more committed to price stability may be an attractive option. The cost is a higher variance of domestic output since the foreign central banker-to the extent that she cares-will stabilize output in her own country. The correlation between domestic and foreign real shocks will thus be an important factor in the decision to relinquish monetary sovereignty. The choice between options a and b is illustrated in figure 12.3. The parameter b, shown on the horizontal axis, characterizes the preferences of the relevant central banker. On the vertical axis is the value of the median voter’s loss function: L(b, b,) is the loss function under the “old” voter’s preferences (b,); it is minimized at b = b*(b,). A shift in preferences, from bmto b,‘ < bm, implies that the “optimal” central banker will be relatively more concerned about inflation also (b* also falls along with b,). The relevant comparison is between L[b*(b,), b,’]--point A-and the value of the loss function at some other point, say for b = b,, which corresponds to the relative weight that the foreign central banker attaches to price and output fluctuations (point B in the figure). We shall first study the simpler case where bf = 0: the foreign central banker is concerned only about price stability and gives no weight to fluctuations in output. This case is simpler because the covariance between domestic L

I

I I I

I

I I I

I

I I

I

I

I I I

I

bf

I

I b*(bm’c brn)

I I I

I I I I

I I

b*(bm)

Fig. 12.3 Choice to relinquish monetary sovereignty

b

571

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

and foreign real shocks is obviously irrelevant. We shall then ask how the incentive to relinquish monetary sovereignty is affected by the covariance of real shocks when the foreign central banker is also trying to stabilize (her own) level of output. Pegging to a Central Banker Solely Committed to Price Stability As we show in equation (A4) in the appendix, the choice between options a and b can be described by figure 12.4. On the horizontal axis appears b*(b,), which characterizes the preferences of the central banker that were optimal before the shift in preferences. This is shown on the vertical axis: (b, - bm’),The locus through the origin describes the points where the median voter, following the shift in preferences, is indifferent between keeping the domestic central banker and deciding to short-circuit her independence by signing an international agreement that commits her to passively peg to her foreign counterpart. This will happen whenever the shift in preferences is sufficiently large, given how different the two central bankers available are: the domestic one, whose preferences are described by b*(b,), and the foreign one, whose b equals zero. If the concern for inflation was relatively high to start with (thus resulting in a relatively high value of b*), the shift must be large to convince the median voter to “adopt” a foreign central banker who

(brn-brn’)

b*(brn)

Fig. 12.4 Pegging to a central banker committed to price stability

572

Susan M. Collins and Francesco Giavazzi

cares only about price stability. An increase in the variance of domestic real shocks (that for given voters’ preferences raises b*) shifts the indifference locus upward: the shift in preferences must be larger to justify abandoning the home central banker, even if she is no longer optimal. The Correlation between Domestic and Foreign Real Shocks

If the foreign central banker, like the domestic one, were chosen by majority rule, her concern for output stabilization would not be zero, as assumed above. We show in the appendix that, if the correlation between domestic and foreign real shocks is small, the higher is bf (the parameter describing the preferences of the foreign central banker), the less likely is the median voter to decide to give up monetary sovereignty. This is true even if b, happens to match the preferences of the new “optimal” central banker (from the viewpoint of the domestic voter), as long as the covariance of real shocks is small, because the home country imports a positive rate of inflation that abroad is justified by the gains in terms of output stabilization but at home is just a source of inefficiency. This has the following important implication: if the correlation between domestic and foreign real shocks is zero, a foreign banker stubbornly committed to price stability is preferable, even relative to one who happened to attach the “optimal” weight to price and output stabilization. As the correlation between domestic and foreign real shocks rises, the home country starts benefiting from the stabilization independently carried out by the foreign central banker: for a given shift in preferences, the higher the correlation, the more likely it is that the median voter will choose to relinquish monetary sovereignty. In terms of figure 12.4, an increase in the correlation between domestic and foreign real shocks shifts the locus downward, increasing the chances that the domestic central banker will be short-circuited. (A formal proof is provided in the appendix.) 12.5 Concluding Remarks

History has provided us with a number of examples of multiple-country fixed exchange rate regimes that have eventually fallen apart. At the top of the list is the Bretton Woods system, which collapsed in 1971. Early attempts at a European exchange rate system also failed-notably, the Snake that was formed in 1972 with ten members but had dwindled to only five small countries pegging to Germany by 1979, when it was replaced by the EMS. In light of these failures, why has the EMS been so successful in reducing inflation differentials and stabilizing exchange rates among members and, indeed, in expanding its membership? This paper has argued that a key factor in explaining these successes has been a convergence in attitudes toward inflation and unemployment in Europe since the late 1970s. In contrast, analysts of the Bretton Woods collapse point to a growing divergence in attitudes

573

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

toward inflation, and in the resultant policy choices, from the late 1960s. However, the EMS was not initially viewed as a success. In its early years (1979-82), there is little evidence that membership in its exchange rate mechanism ( E M ) forced countries to give up monetary sovereignty. France and Italy both followed quite expansionary monetary policies, in contrast to Germany. Both experienced rising inflation and underwent a series of large exchange rate adjustments. Interestingly, other countries that did not participate in the ERM, such as the United Kingdom (and the United States), did significantly reduce their inflation rates during the early 1980s. However, since 1982, there has been widespread convergence of policy and performance among ERM members, and membership has expanded to include Spain and the United Kingdom. This paper had two tasks. First, it presented some new empirical evidence to support the view that one important reason for the success of fixed exchange rates in the EMS has been a convergence of attitudes toward inflation and unemployment among EMS members. Second, it developed a theoretical framework to illustrate why shifts in attitudes of voters within a given country might lead that country to give up monetary sovereignty by pegging its exchange rate to a “leader.” The theoretical section of the paper can be summarized as follows. Consider a country that has inherited a central bank with a prespecified stance toward fighting inflation-that is, preferences that were optimal given voters’ past preferences toward inflation and unemployment. Now suppose that the median voter becomes more concerned about inflation. There are two possible outcomes. First, the country might engineer a change in the anti-inflation stance of its own central bank. Whether this is possible depends on characteristics of the country in question, such as its political system, the timing of elections, and the freedom of its central bank from political suasion. The second, less preferable option its to tie the hands of the domestic central bank by committing to follow the monetary policy of a central bank that is more committed to fighting inflation. The gain from giving up monetary sovereignty is lower inflation. The cost is a reduced ability to stabilize domestic output. The paper shows that whether the country will choose to become (and remain!) a follower depends on how much domestic preferences have shifted, on how different the domestic and the potential leader’s central banks’ antiinflationary stances are, and on the covariance between domestic output and the leader’s output. The empirical section of the paper uses household survey data to look for shifts in attitudes toward inflation and unemployment in eight European countries during 1974-90. For each country, we regress expectations about the general economic situation on expectations about the behavior of inflation and unemployment and look for shifts in the structural coefficients. Our empirical results provide considerable evidence of such shifts. There are three main findings. First, concern about inflation relative to un-

574

Susan M. Collins and Francesco Giavazzi

employment appears to have increased significantly in both Italy and France during our sample period. Both these countries had high inflation during the 1970s and opted to pull out of the Snake. But both reduced their inflation rates and stabilized their exchange rates during the EMS period. Interestingly, we find that much of the change in attitudes occurred ufer the EMS was already in operation-particularly for France. This could help explain why these countries did not adopt more anti-inflationary policies until the mid- 1980s. Exchange rate realignments were not "politicized" and thereby made costly until after 1982. Before this time, ERM membership arguably required little reduction in monetary sovereignty. Second, we also find that the United Kingdom experienced an increase in concern about inflation in the late 1980s. Perhaps this shift in attitudes was a factor in the recent decision to join the ERM. The final result is a surprise that we find intriguing: a shift in the opposite direction for households in Germany. During the mid-l980s, they appear to have become less concerned about inflation. Interestingly, some of the small countries that stayed with Germany in the Snake-that is, Belgium, the Netherlands, and Denmark-show a similar shift. (Such a shift is also evident in Ireland, which is somewhat more puzzling.) This finding provides a possible explanation for why Germany might be willing to stay in an exchange regime that requires it to accept a higher inflation rate. Our analysis thus points to some key differences between Bretton Woods and the EMS that help explain why one system of fixed exchange rates collapsed while the other has expanded. Consider first the Bretton Woods experience. The early phase (1960-66) saw large inflation differentials, relative to the center (the United States), which had relatively low inflation. However, it was difficult to adjust exchange rates, resulting in large real misalignments. In the second phase (1967-71), exchange rate adjustments did occur, but there was no convergence in inflation rates. In fact, the differential reversed as U.S. inflation rose relative to inflation in most other members. The EMS experience is strikingly different. Although there were also large cumulative inflation differentials (with low-inflation Germany at the center) in the first phase (1979-87), they were in large part offset by exchange rate realignments, and capital controls helped avoid disruptive speculative attacks. In the second phase (1987-91), exchange rate adjustments stopped; however, this was sustainable because of the convergence in inflation rates. Thus, there are two important differences between the two systems. First, the EMS exchange rate mechanism facilitated smooth adjustments of exchange rates in the period before inflation rates converged. Second, a convergence in attitudes toward inflation (and unemployment) appears to have occurred in Europe during the 1980s. This shift in attitudes facilitated-and made sustainable-a convergence in inflationary performance.

575

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

Appendix Relinquishing Monetary Sovereignty to a Foreign Central Banker Solely Committed to Price Stability We first consider the case where the foreign central banker is solely committed to price stability: that is, the relative weight that she assigns to output stabilization is b = bf = 0. After a shift in the median voter’s preferences, from b, to bk, the choice between pegging and keeping the domestic (suboptimal) central banker depends on the difference between the expected values of the loss function at b = b*(b,,,) and b = b, = 0: (‘41)

E{L[b*(b,),

- E{L(O, bL1.

Noting that for any given value of a central banker’s preferences, b, the expected loss for a voter with preferences b, is (A2)

E[L(b, b,)]

=

.5(b2 + b,)[k2 + u;/(l

+ b)2],

and using the median voter’s first-order condition, equation (7) in the text, it can be shown that ( A l ) is equal to

where b* = b*(b,). (A3) is positive for (A4)

(b, - b;) > b*

+

b*(l

+ b*)1 + b*)3] [l + b*(2 (1 + b*)‘

= f ( b * , af).

It can also be shown that both dfldb* and d2fld(b*)2 are positive, thus justifying the graph in figure 12.4. Moreover, dfldcr’ = (dfldb*)(db*/du2) dflda2 is also positive: an increase in the variance of domestic real shocks shifts the indifference locus upward.

+

The Covariance between Domestic and Foreign Real Shocks We now assume that b, > 0, the parameter that characterizes the preferences of the foreign central banker, is positive and smaller than b*(b,). In this case, the choice whether to relinquish monetary sovereignty depends on the sign of (A51

E{L[b*(b,), bL1) - E[Ubf, bL)l.

Note that (A5) can be decomposed as follows:

576

Susan M. Collins and Francesco Giavazzi

The term in the braces corresponds to expression (Al), whose sign depends on the sign of (A4); the term in brackets is equal to

(A6)

-

bfrkj - (1

+ b:) [b,/(l + b,)]’

+ 2 b;[b,/(l + bf)]ud,,

where k, is the foreign output target, uf is the variance of foreign output shocks, and u# is the covariance between the shocks to foreign and domestic output. If shocks are uncorrelated, a positive value of b,(motivated by the fact that the foreign central banker cares about output fluctuations in his own country) reduces the incentive to “adopt” the foreign central banker. As we noted in the text, if the covariance were zero, a foreign central banker who happened to be the ideal one after the shift in preferences would be worse than one stubbornly committed to price stability. The intuition requires remembering (from the first-order condition described in eq. [7] in the text) why the median voter would choose a central banker with a positive b,, thus resulting in a positive average inflation rate: only because a positive value of b,dampens the variance of output. But if foreign output shocks are uncorrelated with domestic shocks, from the viewpoint of home residents a positive value of b, is just a source of inefficiency; it keeps inflation positive with no effects on the variance of domestic output. The third term in the expression shows instead that, the higher the covariance between foreign and domestic output shocks, the higher the likelihood that “adopting” the foreign central banker may be a superior option. This is because the correlation between domestic and foreign shocks allows the home country to benefit from the stabilization independently carried out abroad.

References Alesina, A., and V. Grilli. 1991. The European Central Bank: Reshaping Monetary Politics in Europe. Centre for Economic Policy Research Discussion Paper no. 563. July. Barro, R., and D. Gordon. 1983. A Positive Theory of Monetary Policy in a Natural Rate Model. Journal of Political Economy (July). Calvo, G. A. 1978. On the Time-Consistency of Optimal Policy in a Monetary Economy. Econometrica 46: 1411-28. Collins, S. M. 1988. Inflation and the EMS. In The European Monetary System, ed. F. Giavazzi, S. Micossi, and M. Miller. New York: Cambridge University Press. Emminger, 0. 1977. The D-Mark in the Conjict between Internal and External Equilibrium, 1948-75. Princeton Essays in International Finance, no. 122. Princeton, N.J.: Princeton University, International Finance Section. Fischer, S., and J. Huizinga. 1982. Inflation, Unemployment and Public Opinion Polls. Journal of Money, Credit andBanking 14 (February): 1-19. Giavazzi, F., and A. Giovannini. 1989. Limiting Exchange Rate Flexibility: The European Monetary System. Cambridge, Mass.: MIT Press.

577

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

Grilli, V., D. Masciandaro, and G. Tabellini. 1991. Political and Monetary Institutions and Public Financial Policies in Industrial Countries. Economic Policy 2. Johnson, H. 1973. The Exchange-RateQuestion for a United Europe. In The Economics oflntegration, ed. M. B. Krauss. London: Allen & Unwin. Kydland, F., and E. Prescott. 1977. Rules rather than Discretion: The Inconsistency of Optimal Plans. Journal of Political Economy (June). Mankiw, N. G., J. A. Miron, and D. N. Weil. 1987. The Adjustment of Expectations to a Change in Regime: A Study of the Founding of the Federal Reserve. American Economic Review 77 (June): 358-74. Papadia, F., and V. Basano. 1981. Survey Based Inflationary Expectations for the EEC Countries. European Commission, Economic Papers, No. 1, May. Persson, T., and G. Tabellini. 1990. Macroeconomic Policy, Credibility and Politics. London: Harwood Academic. Rogoff, K. 1985a. Can Exchange Rate Predictability Be Achieved without Monetary Convergence? Evidence from the EMS. European Economic Review (June/July). . 1985b. The Optimal Degree of Commitment to an Intermediate Monetary Target. Quarterly Journal of Economics (November). Triffin, R., ed. 1979. EMS-the Emerging European Monetary System.” Bulletin of the National Bank of Belgium 54( 1). Webber, A. A. 1991. Credibility, Reputation and the European Monetary System. Economic Policy, no. 12.

COIlUtlent

Michele Fratianni

Although Susan Collins and Francesco Giavazzi’s paper does not deal directly with the Bretton Woods system, its relevance for this conference’s theme comes from their assessment of the failure of Bretton Woods and the implications for the construction of a stable, fixed exchange rate area in Europe. The message is that Bretton Woods broke down because it lacked flexibility in exchange rate adjustment and because the participating countries had different views about the desirable rate of inflation. In contrast, “common attitudes toward inflation are a necessary condition for the viability of a fixed exchange rate system,” in particular the European Monetary System (EMS) and its evolution into a full-fledged monetary union. The conclusion of the paper is that the successful reduction of inflation differentials in the EMS stems from “a convergence in attitudes toward inflation and unemployment in Europe since the late 1970s.” In turn, the source of this convergence is that voters in France and Italy became more “conservative” (i.e., more concerned about inflation) in the 1980s, whereas voters in Germany became less “conservative.” While I agree that inflation convergence has taken place because of changes Michele Fratianni is professor of business economics and public policy at the Graduate School of Business, Indiana University.

578

Susan M. Collins and Francesco Giavazzi

in attitudes about the role of inflation,’ I would like to raise some methodological points about the way Collins and Giavazzi reach their conclusions and, at the end, suggest an alternative to the hypothesis of a shift in voter preferences: namely, a shift in central bank attitudes about the role of inflation in the economy and the adoption of medium-term monetary targets aimed at reducing the rate of inflation. The Theoretical Construct My first point concerns the applicability of the scenario motivating the model adopted by Collins and Giavazzi (see Alesina and Grilli 1991). An election takes place, the outcome of which is a shift to the “right”-that is, more concerned with inflation-of the median voter. The model assumes that the existing central banker, who has more liberal views on inflation than the median voter, cannot be dismissed because of central bank independence. The government faces two options. Either it keeps the old central banker and accepts a higher inflation rate than the median voter desires, or it signs an international agreement whereby the exchange rate is pegged to a currency of a country that is credibly committed to a high degree of price stability. This scenario lacks plausibility for two reasons. First, in countries like France and Italy, the governor of the central bank serves at the discretion of the government. Dismissing the governor carries for the government a lower cost than signing and sticking to the above-mentioned international agreement. Second, going back to the formation of the EMS, we have two examples-at opposite ends of the inflation spectrum-of governments signing the EMS agreement that do not fit the motivation underlying the model. The first is Germany, whose chancellor, Helmut Schmidt, short-circuited the Bundesbank, not because of undesirable inflation policies pursued by the bank, but because he wanted to use the EMS to achieve political union in Europe: “I had always regarded the EMS, not only as a mere instrument to harmonize the economic policies of the EC member countries, but also as part of a broader strategy for the political self-determination in Europe. . . . France under Giscard was prepared for the loss of sovereignty that would come at the end of this road; the Bundesbank and many of the German professors of economics, who think of themselves as experts, were not prepared for it (and still are not today)” (Schmidt 1990). Italy, on the other hand, offers another type of illustration that is also hard to reconcile with Collins and Giavazzi’s argument. The Italian government entered the EMS to actually discipline itself, an objective that had eluded the Bank of Italy for a long time. The then governor of the bank, Paolo Baffi, wrote that “the participation of our 1. In fact, empirical evidence suggests that the change in attitudes in the early part of the 1980s is more important in explaining EMS convergence than the popular alternative of German dominance in the EMS (Fratianni and von Hagen 1990).

579

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

country in the EMS implies a commitment to reach in a short time a degree of monetary stability equivalent to the Community average of which we are a part; beyond the exchange rate, this commitment includes public finances, productivity, wages, and prices” (Banca d’Italia 1979, 375).

Household Survey Data and Empirical Testing My second point concerns the use of the European Community household survey data and the empirical procedure. Given the importance of the data for the strategy and conclusions that Collins and Giavazzi reach, a more thorough discussion of what the surveys represent would have been useful. To begin with, the question on price trends (PT) clearly refers to the expected change of the inflation rate and not the expected level of the inflation rate, whereas the questions on general economic conditions (EP) and unemployment (UT)refer to the expected level of the variables over the next twelve months. In contrast, the loss function (1) is written in terms of expected levels of EE PT, and UT. Furthermore, following Collins and Giavazzi’s assumption that EP expresses a utility level dependent on PT and UT, the responses to E e PT, and UT are linked by a rationality criterion. To check this, I arbitrarily looked at the survey data for France and Italy for the period 1974:1-1983: 1 (there was a break in the series after 1983) to see whether the three responses were consistent in the sense that when respondents indicated that PT and UT would deteriorate EP would decline, or vice versa. I found that five out of fifty-fourobservations violated this consistency requirement. Unless respondents are irrational, this finding suggests that at least one more variable is missing in the information set of individuals when expressing their sentiments about future values of EP or, more generally, that the assumed preference function is not empirically valid. One obvious candidate for a missing variable is output growth. Changes in the unemployment rate are an imperfect proxy of output growth, and even more so in European countries where UT has remained relatively high, despite “decent” economic growth. Hysteresis effects, labor market rigidities, and laws that hamper labor mobility are some of the reasons why changes in unemployment rates do not correlate well with output growth (Commission of the European Communities 1984; Lawrence and Schultze 1987). The omission of a potentially important variable, such as output growth, from equation (2) creates the problem that the residual term in the equation captures movements of the omitted variable. Since UT and output growth are related, it follows that UT is not independent of the error term, with the attendant econometric consequences. Finally, supposing for argument’s sake that the missing variable poses no problem in the sense that it is constant over time, how does one interpret the regression results of the paper? I believe that the estimated coefficients have more to do with the relative ability of households to forecast unemployment

580

Susan M. Collins and Francesco Giavazzi

rates and changes in inflation rates than with welfare weights and that the variable EP has more to do with an overall index of combining different forecasts than a welfare index. Some evidence in favor of this proposition comes from Papadia and Basano (1981), who looked at the properties of Pi? They transformed the price trend survey data responses into quantitative values of the expected rate of inflation using a logistic (and linear) function. The transformed series behaves rationally in the (weak) sense that it can forecast the rate of inflation more accurately than purely autoregressive estimates of the inflation rate. I do not know of any comparable work on the unemployment rate. Yet, on the basis of this evidence, I submit an alternative explanation of the estimated values of b-the relative weight on unemployment-in the regressions. Households got better in forecasting inflation rates in the 1980s because the mean and variances were shifting downward in high-inflation France and Italy. In sum, the downward shift in the estimate of b for France and Italy could reflect improvements in forecasting PT relative to Ui? Peculiar Results While the empirical results for France and Italy accord with our intuition, those relating to Germany do not: the estimated b rises sharply for Germany after October 1988. This, by itself, is a stunning result in light of the history of the EMS since the last realignment of January 1987. France and, to a lesser extent, Italy have repeatedly complained about tight German monetary policy while at the same time preventing the deutsche mark from realigning upward. In contrast, the convergence of the b’s among the three countries should have spelled French and Italian acquiescence to German monetary policy. An Alternative Hypothesis

As I mentioned earlier, there is something to the story of a change in attitudes about inflation rates. Figure 12C.1, showing annual percentage changes in the consumption price deflator for EMS countries and a group of non-EMS countries, suggests that such a change can be approximately dated at the end of the 1970s. The failures of active demand management of the 1970s led policymakers of the industrial countries to gradually switch to medium-term monetary targeting aimed at reducing the rate of inflations2Such a switch did not exclusively affect the EMS countries; the non-EMS countries were affected as well.3 Indeed, the two lines are so close as to suggest that there is nothing unique about EMS countries’ inflation rates (Fratianni and von Hagen 2. See Chouraqui and Price (1984). Quantitative monetary targets were adopted by Germany (1974), Italy (1974), Canada (1973, France (1977), Japan (1977), the United States (October 1979), and the United Kingdom (March 1980). 3. The data come from Fratianni and von Hagen (1992, chap. 2). Non-EMS countries consist of Australia, Austria, Canada, Finland, Greece, Japan, Switzerland, the United Kingdom, and the United States. The weights used in adding individual countries’ inflation rates are based on 1982 GDP shares.

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

581

14

Percent

4-

20 1974

1976

1978

1980

1982

1984

1986

1988

Year Qrowth rate

Fig. 12C.1 EMS and non-EMS consumption deflator

l4

1974

1976

1978

1980

1982

1984

1986

1988

1990

Year Fig. 12C.2 EMS v. non-EMS money base growth

1992). Further corroborating evidence comes from figure 12C.2, showing the growth rates of the aggregate EMS and non-EMS monetary base.4 Camen, Genberg, and Salemi (1991) apply an inverse control procedure to infer the 4.The countries and weights are the same as those for the inflation rate.

582

Susan M. Collins and Francesco Giavazzi

objectives of the Bundesbank and the Banque de France and find that the latter gave a much larger weight to inflation after June 1982. Such a shift makes the French central bank behave more like the Bundesbank. In sum, my alternative to the Collins-Giavazzi hypothesis is that the disinflation of the 1980s was to a large extent the result of changes in attitudes of the monetary authorities toward inflation. Clearly, this alternative does not preclude the Collins-Giavazzi hypothesis of the median voter becoming more conservative. While central bank preferences will ultimately have to converge with those of the electorate, I believe that the disinflation of the 1980s was the result of the central bankers leading the charge, with the public trailing right behind.

References Alesina, A., and V. Grilli. 1991. The European Central Bank: Reshaping monetary politics in Europe. Harvard University. Mimeo. Banca d’Italia. 1979. Relazione annuale per I’anno 1978. Rome: Banca d’Italia. Camen, U., H. Genberg, and M. Salemi. 1991. Asymmetric monetary policies? The case of Germany and France. Open Economies Review 2:219-36. Chouraqui, J. C., and R. W. R. Price. 1984. Medium-term financial strategy: the coordination of fiscal and monetary policies. OECD Economic Studies 29-49. Commission of the European Communities. 1984. Annual economic report, 1984-85. European Economy 22(November):5-53. Fratianni, M . , and J. von Hagen. 1990. German dominance in the EMS: The empirical evidence. Open Economics Review 1:67-87. . 1992. The European Monetary System and European Monetary Union.Boulder, Colo.: Westview. Lawrence, R. Z., and C. Schultze. 1987. Overview. In Barriers to European growtha transatlantic view, ed. R. 2. Lawrence and C. Schultz. Washington: Brookings. Papadia, F., and V. Basano. 1981. Survey Based Inflationary Expectations for the EEC Countries. European Commission, Economic Papers, No. I , May. Schmidt, H. 1990. Kampf gegen die Nationalisten. Die Zeit, 31 August.

Comment

Niels Thygesen

This paper by Susan Collins and Francesco Giavazzi is an original and intriguing contribution to our efforts at understanding why the European Monetary System (EMS) has succeeded where the Bretton Woods system failed. The EMS has moved through three stages of increasingly tight exchange-rate management: (1) an initial stage for the first four years since the start in March Niels Thygesen is professor of economics at the University of Copenhagen and senior research fellow at the Centre for European Policy Studies, Brussels.

583

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

1979 during which the system looked like a crawling peg, although devaluations sometimes.did not fully offset excess inflation; (2) a disciplinary stage, lasting also nearly four years, as most non-German participants aimed to squeeze their inflation rates by maintaining as stable exchange rates as possible for their currencies against the deutsche mark; and (3) a third stage, now lasting for more than five years, during which rates have moved only within the fluctuation margins. This process of gradually increasing emphasis on nominal convergence is usually analyzed as the result of changing perceptions among policymakers as to the unemployment-inflation trade-off, combined with a shift in preferences toward a greater weight on the objective of price stability. Both authors have made important contributions to the literature on these explanations of why a number of member states of the European Community have chosen to modify their policies and why they have opted to do so through participation in the EMS. In the present joint paper, the authors advance the hypothesis that the shift in policy was preceded by a shift in voter preferences toward more aversion to inflation; in the absence of an opportunity to replace their domestic central bank leadership with a more “conservative” one to reflect this shift, voters have been content to see monetary sovereignty constrained by pegging to the deutsche mark. Furthermore, the authors devise an ingenious way of using survey data to evaluate empirically if, and when, changes in preferences can be identified. This empirical study is the most novel and commendable part of the paper. Since January 1974, the Directorate-General for Economic and Financial Affairs of the European Commission has published at intervals of four months the results of a questionnaire study of household perceptions of recent trends in the economic performance of the respondents’ country and expectations as to the next twelve months. Two specific questions are asked, namely, with respect to changes in the rate of inflation and in the unemployment percentage. Consumers are also asked to evaluate the general macroeconomic situation of their country. The questionnaire also asks three questions relating to the particular situation of the individual respondents with respect to their financial situation, savings, and intended major purchases. The authors focus on the three first questions and assume that the responses to the general macroeconomic question can be seen as weighted averages of the responses to the two specific questions on inflation and unemployment. They discuss in a careful and well-documented way possible dates for one or more breakpoints in the weights of the two macroeconomic indicators that add up to the general evaluation by consumers and find interesting and surprising results for most of the seven main countries that have participated in the EMS since its start and for the United Kingdom. The main results suggest that voters in France and Italy had revised their preferences in the direction of greater emphasis on lower inflation before, or early, in the EMS experience, hence giving political impetus to the gradual tightening of the system that occurred with some delay,

584

Susan M. Collins and Francesco Giavazzi

and that, in contrast, voters in Germany and the Netherlands shifted their preferences in the opposite direction-toward greater emphasis on employment-at different points in time in the second half of the 1980s. This latter shift is interpreted by the authors to imply growing political support in the anti-inflationary bastions for the increasing inflationary risks implied by an EMS in which exchange rates have become de fact0 frozen. There are three reasons why I find these rationalizations of the evolution of the EMS less than fully convincing. The first is that the evaluation by consumers of the general economic situation may depend on more macroeconomic indicators than simply the inflation and unemployment rates. The authors themselves point out that there is significant autocorrelation in several of the national equations, which could be an indication of the omission of one or more explanatory variables. One likely candidate as omitted variable is an indicator of the sustainability of the government’s economic policy, expressed, say, by the ratio of government debt to GDP-or the rate of change of this ratio-which has been a focal point of the public debate in several EMS countries in the 1980s. Other possible candidates are the current imbalance and an indicator of competitiveness; discussion of external imbalance has also figured prominently in the evaluation of the general performance of the European economies in the 1970s and 1980s. A second reason for questioning the outcome of the empirical analysis is the simplified and linear way in which the two main indicators of economic performance enter the preference of consumers. For example, the authors remark that German consumers appear to have shifted toward greater tolerance of inflation in late 1986. This was at a time when consumer prices were falling slightly as the combined result of a sharp drop in oil prices and an extended period of tighter budgetary policies from 1982 onward. Presumably, German consumers recognized at the time that inflation had been dealt with quite effectively, so that concern could justifiably be expressed primarily with respect to observed rising unemployment. Conversely, French and Italian consumers were well justified in the late 1970s and early 1980s in focusing their concerns on inflation, which remained stubbornly at the double-digit level. The importance of the level of the two indicators disappears when the preferences are expressed only in terms of changes in them. A third reason for questioning the results-not unrelated to the first twois linked to the fact that the empirical study fails to identify shifts in preferences at times when voter dissatisfaction with the general economic situation found expression in rather massive swings of the political balance. In France, the center-right government of President Giscard d’Estaing and Prime Minister Raymond Barre, which had followed a steady policy of squeezing down inflation-although without much success-was ousted from office in 1981 and replaced by a socialist government committed to expansionary policies and ready to take greater inflationary risks. This does not square well with the

585

Attitudes toward Inflation and the Viability of Fixed Exchange Rates

shift toward greater aversion to inflation implied by the consumer survey after October 1979. Equally puzzling is the failure of the survey to pick up the shift in voter preferences expressed in the switch from a Labour to a Conservative government in the United Kingdom in May 1979. Given the major role that the aim of reducing inflation played in the campaign that brought Mrs. Thatcher to power in that election, the empirical results are surprising to an economist, at least. There are also less important questions with respect to the timing of the evolution of preferences in some other countries, although, on the whole, they appear more plausible. The three reasons given for questioning the reliability of the empirical results of the paper should not be seen as a criticism of the authors’ effort to use previously unexploited data in a novel and imaginative way to address an important and striking phenomenon: why a number of European countries have adopted changes in macroeconomic policies, going so far in a noninflationary direction that EC governments have now even agreed to sign a new treaty that recognizes price stability as the primary objective of monetary policy. While I fully recognize the pioneering nature of the empirical analysis, I have more reservations as to the linkage made by the authors to the theoretical literature on commitment to price stability by adopting an increasingly rigid exchange rate regime. It appears to me farfetched to assume that voters have influenced the choice of exchange rate regime in order to bring about a cession of national monetary sovereignty through participation in the EMS. The gradual evolution of the EMS that can be observed in the 1980s has become clear only ex post, and it seems more likely that the central banks and governments have participated in it as a result of their experience of how their economies worked and because the central banks-with their inherent bias toward price stability-have gradually gained the upper hand in policy-making as capital mobility increased and monetary management became more market oriented and less susceptible to national preferences. In other words, the voters did not have to impose a shift in their preferences on their respective central banks; the latter were quite ready to seize the opportunity to conduct policies geared increasingly to low inflation. From this perspective, studies of the evolution of central bank independence and of the decisions to liberalize capital movements and push financial integration seem more relevant to an explanation of changes in the EMS than an analysis of shifts in voter preferences. Studies of the latter may, however, be illuminating for evaluating to what extent such changes have been supported by the voters. The main conclusion of the paper is that the consumer surveys can be interpreted as indicative of such support in the higher-inflation countries. At the same time, the excellent results in achieving a high degree of price stability in the low-inflation countries by the mid- 1980s may have facilitated their acceptance of a system that has entailed slightly higher inflationary risks for them toward the end of the period of observation.

586

Susan M. Collins and Francesco Giavazzi

General Discussion Fred Bergsten reflected on a parallel between Germany’s current position in the EMS and the U.S. position at the end of Bretton Woods. He argued that the Germans are still devoted to fighting inflation but have lost the use of the exchange rate instrument under the existing EMS-hence they have ended their opposition to monetary union. This is similar to the U.S. situation in 1971. The United States wanted to depreciate but in that case felt that it could do so only by breaking the system. Charles Wyplosz disagreed. In his view, Germany is running European monetary policy and has not lost its ability to fight inflation. The German concern with monetary union is over losing its monetary independence to other countries. Albert0 Giovannini argued that, as long as the EMS exists, the option to realign remains. However, with full monetary union, the option is no longer there. This is his explanation for why there is German opposition to monetary union. Charles Wyplosz suggested that the reason for the growing success of the EMS since 1979 was a flow investment by European central banks in the stock of credibility. During the post-Bretton Woods period, different central banks have gone through different learning processes, culminating in the 1980s with the view that credibility was paramount. Sebastian Edwards wondered what the authors’ surveys were actually capturing. He thought that responses by individuals in each country to questions about unemployment and inflation might be strongly influenced by the overall level of optimism or pessimism, which in turn would be colored by exogenous events such as the Falklands War in the United Kingdom, capital controls in France, or the World Cup in Italy. Allan Melfzer reflected on the lessons of Bretton Woods for the EMS. He argued that, during the 1970s, the Germans learned that revaluation did not inflict permanent harm on their export industries, so they learned not to resist adjustment as they had in the 1960s. The French learned that the only way to maintain the EMS system was to disinflate and that, by doing so, there would not be permanent effects on unemployment. In addition to having learned the lesson from Bretton Woods of the need to allow adjustment, the Europeans (especially the Germans) did not want to continue absorbing U.S. inflation, as they had under Bretton Woods and throughout the 1970s. Thus, in 1979, the year the EMS began, the Germans told Paul Volcker at Belgrade that they were no longer willing to support the dollar.

13

Panel Session 11: Implications for International Monetary Reform Barry Eichengreen, chair C. Fred Bergsten Stanley Fischer Ronald I. McKinnon Robert Mundell Martin Feldstein

The Collapse of Bretton Woods: Implications for International Monetary Reform c. Fred Bergsten During the first twenty-five years of postwar monetary history, the world operated an adjustable peg version of a fixed rate system-the Bretton Woods regime. That system began to erode in the early 1960s, and twenty-five or thirty years ago there were already calls for sweeping reform. It soon became clear that Bretton Woods was unable to facilitate the exchange rate changes and other adjustments that were necessary to achieve a stable international economy. The system collapsed at the outset of the early 1970s when the dollar became overvalued by about 20 percent, protectionist pressures rose as a result, and the regime could not cope.’ There is a clear record of failure of that version of fixed exchange rates. For most of the next twelve and a half years, from March 1973 until September 1985, we had a system (or nonsystem) of unmanaged flexibility of exchange rates. It is clear that this system also failed. It permitted the dollar to become overvalued by 40-50 percent, more than twice the misalignment that brought the collapse of Bretton Woods. It failed to keep trade open; proC. Fred Bergsten is director of the Institute for International Economics and chairman of the Competitiveness Policy Council. He served as assistant secretary of the treasury for international affairs during 1977-81. This material is copyright 0 1992, Institute for International Economics, Washington, DC. All rights reserved. 1. The “Triffin dilemma’’ and other liquiditykonfidence issues were an underlying source of difficultyfor the system but clearly did not trigger its collapse. Indeed, the intellectual and policy focus on those issues throughout the 1960s diverted attention from the shortcomings of the adjustment process, which were the primary weakness of Bretton Woods. My account of the collapse is in The Dilemmas ofrhe Dollar (New York: New York University Press, for the Council on Foreign Relations, 1975), 91-93.

587

588

C. Fred Bergsten

tectionism grew throughout the 1980s, and the world trading system is still eroding. It had no meaningful effect on national economic policies and therefore failed to achieve the most rudimentary objective of any international economic system.2 The authorities have been groping for new monetary arrangements for about six years. At the Plaza in September 1985, they clearly recognized that the extant system had failed.3 At the Tokyo Summit in May 1986, they adopted a set of “economic indicators” to guide the adjustment process. At the Louvre in February 1987, they installed a system of reference ranges. The world’s monetary authorities thus decided to proceed with a two-track program, based on reference ranges and indicators, in an effort to find a new regime.4 After some backsliding in 1988-90, when the dollar was permitted to appreciate prematurely, the G7 seem to have reestablished their reference ranges in 1991: a floor was successfully placed under the dollar4eutsche mark rate in February at 1.45: 1, and the dollar’s subsequent rise was capped effectively in July at about 1.85:1. Well before these renewed efforts at the global level, most of the European countries re-created an adjustable peg system among themselves via the European Monetary System (EMS). After numerous initial doubts, the EMS is now widely viewed as a resounding success. Indeed, it has been so successful that it will probably evolve into a full Economic and Monetary Union (EMU) within this decade. We are thus in a transition to a completely new monetary system, as in the late 1960s and early 1970s, with the establishment of fixed rates in Europe and reference ranges globally. The process is evolutionary, and I would guess that we are witnessing a true Hegelian synthesis. The Bretton Woods version of fixed exchange rates was too rigid and would not work. Unmanaged flexibility failed because it permitted massive and costly misalignments. We thus need to devise a system that combines the best features of both previous regimes and avoids the worst of each-an intermediate solution that will provide a more stable and effective basis for the world economy. It is interesting to recall that, when Bretton Woods broke down and the world moved to flexible rates, there was a great deal of interest in intermediate solutions. They were then called wider bands and crawling pegs.5 The com2. See my “Exchange Rate Policy,” in American Economic Policies in the 1980s. ed. Martin Feldstein (Chicago: University of Chicago Press, forthcoming). 3. There were numerous similarities between the Plaza Agreement and the Smithsonian Agreement of December 1971 that sought to pick up the pieces from the collapse of Bretton Woods: international agreement to depreciate the dollar sharply, in order to correct a huge (for the time) U.S. deficit and counter the resultant trade protectionism, and the beginnings of major systemic reform. 4.See Yoichi Funabashi, Managing the Dollar: From the Plaza to Louvre (Washington, D.C.: Institute for International Economics, 1988). 5. A number of the leading proposals were presented in George N. Halm, ed., Approaches to Greater Flexibiliry ufExchunge Rarest The Burgenstock Papers (Princeton, N.J.:Princeton University Press, 1970).

589

Panel Session I1

bination of those two represents a close approximation to what we now call (crawling) target zones. Among the several intermediate possibilities, this is both the most promising and the most feasible. Target zones would represent a natural further evolution of the current reform process, particularly as an extension of the reference ranges implanted since the Louvre. Moving to an effective system of target zones will require five basic changes from the way in which the reference range system was originally constructed at the Louvre, some of which are already evolving. First, the officials must agree to a set of exchange rate relationships that will achieve and maintain equilibrium in national current account positions with economies in internal balance, meaning the fastest possible economic growth without igniting new inflation.6 That sounds trivial, at least in principle, until we recall that the original method for setting reference ranges was to center them on whatever the level of rates was on the day that the G7 were meeting.’ That is obviously a rather arbitrary basis for trying to stabilize exchange rates. It is clear, and was even at the time, that the ranges set at the Louvre were decidedly premature, and several subsequent “rebasings” were soon required. The G7 are learning, however. In 1991, they set the floor and ceiling of the new dollar4eutsche mark range sequentially rather than at one time-the floor in February, the ceiling in July. Since it is politically difficult for governments to agree on exchange rates that differ from where rates are in the market on the day they are meeting, and since any effort to do so could destabilize the markets severely, pragmatic considerations suggest that the authorities should look for a time when market rates are close to long-run equilibrium levelsand then take steps to keep them within a reasonable distance of those levels. At present, with the exception of Japan to a modest extent, there is fairly strong evidence that rates are now reasonably close to equilibrium levels and that the time is thus ripe to systematize the currently informal reference ranges. In October 1991, the Institute for International Economics released a study by Paul Krugman entitled Has the Adjustment Process Worked?8On the basis of a two-day conference held in late 1990, which considered detailed studies of the three largest imbalances of the 1980s (the United States, Japan, Germany), Krugman answered the question with an unequivocal yes. The American deficit dropped from 3.6 percent of GNP in 1987 to 0.7 percent in the first half of 1991, the Japanese surplus fell to 1.1 percent of its GNP in Japan’s fiscal year 1990 (ending March 1991), and the German surplus-propelled of course mainly by unification-will probably disappear this year. Several other recent studies conclude that current rates are reasonably close to 6. A blueprint for doing so is in John Williamson, The Exchange Rate System, rev. ed. (Washington, D.C.: Institute for International Economics, 1985). I . S e e Wendy Dobson, Economic Policy Coordination: Requiem or Prologue? (Washington, D.C.: Institute for International Economics, April 1991). 8 . Paul Krugman, Has the Adjustment Process Worked? (Washington, D.C.: Institute for International Economics, October 1991).

590

C. Fred Bergsten

equilibrium level^.^ Hence, present exchange rate relations may provide a reasonable basis for installing a full regime of target zones. Japan may be an exception because its surplus has risen again sharply in 1991. Moreover, Yoshitomi has indicated that the last $20-$30 billion of the reduction in that surplus in 1990 could not be explained by any model and may have been produced by purely temporary factors. l o MITI has released a survey in which Japan’s major international companies acknowledge that they can compete fully at 120:l. Hence, a yen appreciation of 10-15 percent from the recent level of about 130:1 seems called for before the new ranges are set. The second key change from current procedures is to have arrangements in place that will maintain exchange rates at equilibrium levels. This means that the targets must be real exchange rates, not nominal exchange rates, because inflation differentials have to be offset by currency movements. The rates would of course be stated in nominal terms, but the targets have to be real. It will also be essential to install procedures to change the real rates to offset underlying differences in national economic developments, such as productivity differentials. For example, under a crawling target zone system, I would expect the zone between the yen and the dollar to rise by several percentage points per year (in nominal terms). Japan will probably run lower inflation rates than the United States. Its productivity growth will probably be higher. Its huge creditor position and investment earnings, contrasted to the U.S. debtor position and likelihood of growing interest payments, will probably account for a percentage point or so on the exchange rate. Keeping the rates in equilibrium is thus going to require annual appreciation of the nominal yendollar rate, and the system has to comprehend that. It must of course also be able to change rates whenever there are large shocks, such as a major shift in the price of oil. Third, the target zones must be considerably wider than the Louvre ranges. These ranges have been variously reported as plus or minus 2.5 or 5 percent. That is not large enough, for three reasons. One is that we cannot know with precision the equilibrium level of rates. Another is that it is important to permit exchange rates to move a bit in order to permit continued use of monetary policy to pursue domestic targets, particularly price stability. Moreover, when the currency midpoints have to change, it is desirable that they do so within the ranges to avoid both market disturbances and political problems. In any event, there is no need for a high degree of rate fixity. What is needed are rates that avoid the large misalignments and thus the large disequilibria of the type that have pervaded the last fifteen years. Target zones could achieve that goal. 9. See William R. Cline, “United States Adjustment: Progress, Prognosis and Interpretation,” in International Adjustment and Financing: Lessons of 1985-1991. ed. C. Fred Bergsten (Washington, D.C.: Institute for International Economics, 1991); and John Williamson, Equilibrium Exchange Rarest An Update (Washington, D.C.: Institute for International Economics, forthcoming). 10. See Masaru Yoshitomi, “Surprises and Lessons from Japanese External Adjustment in 1985-90,” in Bergsten, ed., International Adjustment and Financing.

591

Panel Session I1

There is new evidence that the G7 have learned this lesson too. As noted above, the new reference range for the dollar-deutsche mark rate appears to be from 1.45:1 to 1.85:1. This is the equivalent of a zone centered at 1.65:1 with margins of 12 percent on either side. Such a zone will be much more sustainable than the narrow bands adopted at the Louvre. The fourth change that is required is that the participating countries accept a commitment to change their policies when needed to protect the zones. Under the Louvre system, the only pledge was to consult when the rates move a certain degree away from their midpoints. The credibility of the system depends on the willingness of the major countries to change policies when the rates reach the edge of the zone. The authorities may not actually have to make such changes very often if the initial criteria are met correctly. If the officials do have to act, intervention (and associated jawboning) would likely be the first point of departure. Changes in monetary policy would come next. On occasion, changes in fiscal policy would be required. A new analysis by Kathryn Dominguez and Jeffrey Frankel, using official data from the United States and Germany for the first time, suggests that intervention can be quite effective in altering market rates provided that the intervention is publicly announced. The G7 experience indicates that the cost-benefit ratio of recent intervention efforts has indeed improved dramatically: the successful defense of both ends of the new dollar4eutsche mark reference range in 1991 was achieved with very modest levels of activity. Hence, the need to resort to monetary policy, and other instruments that might run counter to domestic goals, is likely to be less than has been thought. Dominguez and Frankel’s findings on intervention underline the need for a fifth and final emendation of the current reference ranges to achieve effective target zones: public announcement of the ranges. Once the authorities establish credibility for the new system, such announcement will promote stabilizing private capital flows and reduce the need for official intervention and other policy changes. History tells us that the only effective efforts to achieve systematic coordination of economic policies have occurred when such efforts have been prompted by an agreed exchange rate mechanism. For all its shortcomings, the Bretton Woods system did work in that respect to a significant extent. The EMS is now working, and an EMU should do even better. There is no historical case where an effort to coordinate macroeconomic policies directly produced significant results. We should learn from the past and move to an intermediate and pragmatic system of target zones for all the major countries. If that can be done effectively, we will have learned the lessons of both Bretton Woods and its tortured aftermath. There is one additional reason for moving in the near future to systemize I 1. Kathryn Dominguez and Jeffrey Frankel, The Efecrs of Foreign-Exchange Intervenrion (Washington, D.C.: Institute for International Economics, forthcoming).

592

C. Fred Bergsten

the current ad hoc international monetary regime: the likely evolution of the European Monetary System into an Economic and Monetary Union. A successful move to EMU will convert Europe from a series of small and mediumsized economies into one large and much less open economy. This change will have several effects: It will tend to increase the extent of currency fluctuations among Europe, America, and Japan-generating greater international financial instability and potential misalignments that would distort trade and add further to the tendencies toward trade protection. It will tempt Europe to practice “benign neglect” from time to time, as the other large and relatively closed economy has done, or at least to try to force the costs of adjustment onto others, as the United States has also done. If it fails to achieve a unified fiscal policy to go with its unified monetary policy, there will be a strong possibility of a Europe-wide repetition of Reaganomics from the early 1980s and the German policy mix of the early 1990s: large fiscal stimulus, very tight money, a sharp appreciation of the currency, large trade deficits, and resultant protectionism. Without a true political master, the European Central Bank will be particularly likely to foster such an outcome. This will be especially true in its early years, as it seeks to prove its fealty to the goal of price stability and to discipline recalcitrant governments into fiscal rectitude. Moreover, achievement of EMU-even without the final step of a single currency, but especially with it-will propel the ecu to a central role in a new multiple reserve currency system. This will both reflect and produce a substantial portfolio adjustment from (mainly) dollars into ecu, reinforcing the likely appreciation of European currencies with attendant trade balance and protectionist problems. This effect would be further accelerated if the EMU pooled Europe’s monetary reserves and attempted to dispose of some of the “excess,” identified by the EC Commission as on the order of $200 billion.’* The policy implication is that the United States and Japan should engage Europe in negotiations on the global monetary system while the latter is working out its regional arrangements-particularly as both of the basic blueprints for EMU, the report of the Delors Commission and Karl Otto Pohl’s design for a Eurofed,I3 totally ignore the external dimension thereof. American strategy in the trade area throughout the postwar period has been to engage Europe 12. See One Market, One Money: An Evaluation of the Potential Benejits and Costs of Forming an Economic and Monetary Union, European Economy no. 44 (Brussels: Commission of the European Communities, October 1990), chap. 7. 13. Report on Economic and Monetary Union in the European Community (prepared by the Committee for the Study of Economic and Monetary Union, April 1989); Karl Otto Pohl (president of the Deutsche Bundesbank), “Basic Features of a European Monetary Order” (lecture organized by Le Monde, Paris, 16 January 1990).

593

Panel Session I1

in global negotiations at each key milestone in its evolution: the Kennedy Round, when the Common Market was created; the Tokyo Round, when it expanded to bring in the United Kingdom and others; and the Uruguay Round, as it moved toward “1992.” A similar approach is needed in the monetary area to avoid the risk that EMU will destabilize global arrangements and that, once its details have been put in place, it will be too late. This should be feasible now that, by successfully placing a floor under the dollar in February 1991 and effectively capping the dollar in July 1991, the G7 seem to be returning at least de facto to reference ranges among the major currencies B la Louvre.

Stanley Fischer The founders of the international economic system who met here in July 1944 aimed to create a system that would promote international growth. They succeeded, even though none of the three institutions that were to run the system-the International Monetary Fund (IMF), the International Bank for Reconstruction and Development (IBRD), and the International Trade Organization (IT0)-operated according to plan. I will start by discussing the role of the international institutions in the world economy and then briefly take up the issues of the convertibility puzzle and the problems of international capital flows in the 1990s.

The Fund, the Bank, and the I T 0 The IMF was supposed to deal primarily with international monetary relations among the industrialized countries. This IMF role was limited even before 1973 and has been more limited since-with the 1976 British program representing the last major operational involvement of the Fund with the industrialized countries. Flexible exchange rates and the mobility of international capital have made the Fund unnecessary to the major countries, and the G7 and the G10 are less unwieldy settings in which to discuss matters of mutual concern. The Fund still plays an informational and monitoring role in the industrialized countries, through its annual Article IV consultations. This informational role would be enhanced if the annual Recent Economic Developments reports Stanley Fischer is professor of economics at the Massachusetts Institute of Technology and a research associate of the National Bureau of Economic Research. The author has benefited from comments on an earlier version of his remarks delivered at the conference, particularly by Fred Bergsten and Leslie Pressnell.

594

Stanley Fischer

on member countries were published. Publication would improve the quality of policy discussion within countries and, ultimately, the quality of economic policy. It should be possible to find some governments that both are strong enough and sufficiently value informed public discussion to agree to the publication of the reports on their countries.’ If a few countries set an example, others will eventually follow. The IMF now operates as an agency through which the industrialized countries deal with developing countries, including Eastern Europe and the former Soviet Union. The Fund’s role in the developing countries was especially important during the debt crisis, and it will be important in the early years of the economic transformation in Eastern Europe. However, there are few serious evaluations of the Fund’s developing country operations, and there is surely much to learn from a careful appraisal of the historical record. The IBRD was set up to promote private capital flows to the developing countries, mainly by providing guarantees. In fact, the World Bank has, with trivial exceptions, not operated as a formal guarantee agency. Since the Bank’s direct borrowing and relending can be viewed as an efficient way of providing ironclad official guarantees, it has to be asked what difference direct guarantees would make. One possible benefit is that greater direct involvement of industrialized country banks and enterprises in the developing countries would increase the efficiency of foreign capital flows. The commercial banks and other potential investors, both in Eastern Europe and in other developing countries, continue to press for public sector guarantees of their investments. In 1988, the Bank Group set up MIGA-the Multilateral Investment Guarantee Agency-to provide insurance against political risks. MIGA is still establishing itself, but, even if and when it does, there remains room for the World Bank to play an enhanced economic insurance function. As the IMF deals increasingly with the developing countries, and the World Bank in the 1980s expanded its operations beyond project lending and into structural adjustment loans, why not merge the institutions? It is easy to see the advantages of a single agency, not least the saving that would come from having only one Board of Directors. There are also major benefits to having two agencies. The Bank certainly, and perhaps also the Fund, is so large as to stretch the span of management’s control. Despite their overlapping responsibilities, the agencies have different tasks; at some point, the Fund may be given back its original role of dealing seriously with the industrialized countries. Most important, a merger would be a mistake as long as the agencies continue to operate with as much secrecy as they do. Each of the agencies is immensely powerful and operates in the developing countries with very few checks or balances. As separate agencies, 1. Publication would affect the frankness of the reports, but the professional quality of the staff can be relied on to ensure that the basic message gets across.

595

Panel Session I1

each provides a necessary check on the activities of the other. Unless a better control mechanism can be invented, the Fund and the Bank should not be merged. The IT0 was stillborn, and the General Agreement on Tariffs and Trade (GATT) is widely viewed as an inferior substitute. Nonetheless, the expansion of international trade is the most striking success of the postwar economy. Deplorable as the increase in nontariff barriers has been, and important as it is to stop the trend toward voluntary trade restrictions, trade has grown more rapidly than output almost every year since the end of the war. The credit for the growth of trade must be shared between the GATT and the steady decline in restrictions on international payments. East European experience has made clear the close link between current account convertibility and trade liberalization. The early postwar literature leaves the impression that this close connection was less well recognized then than now. If it were, the IMF and the IT0 might have been designed as a single agency. As the Uruguay Round negotiations falter, fear of the development of a three-separate-trading-bloc world grows. There is no question that a genuinely successful conclusion to the Uruguay Round talks would be better than a shift of emphasis to regional trading arrangements. It is also clear that the Uruguay Round will be in trouble unless European politicians grasp the nettle of their agricultural protectionism, which harms both many developing country exporters and Eastern Europe. But it is unlikely that three closed regional trading blocs will develop. East Asia’s economic dependence on access to the North American market means that, with the usual shoving and hauling, transpacific trade barriers are likely to continue being reduced and the volume of trade to continue to grow. The real difficulty is with Europe, where many see the completion of the single market as an event that should benefit Europeans, not foreigners. It would be a great pity, not least for Europe, if the forces of protectionism and exclusion ultimately win out in Europe. If they do, restrictions are likely to affect not only international trade but also the flows of international capital and investment between Europe and the rest of the world, to the detriment of all. The decisions lie with European policymakers.

The Convertibility Puzzle In the current East European orthodoxy, current account convertibility at a heavily devalued exchange rate and trade liberalization should come at the start of the reform program. If that strategy had been followed at the end of World War 11, the Europeans would have devalued heavily against the dollar and removed restrictions on trade and current account payments. Why was that not done? First, there was much less faith in the price system then than now and much more reliance on quasi planning-in which eco-

596

Stanley Fischer

nomic policy focuses on quantities of needed inputs, using implicit inputoutput matrices that permit little substitution. Second, and certainly as important, it would have been virtually impossible at the end of the war to ask war-ravaged populations to make further short-run sacrifices to achieve faster adjustment. Of course, adjustment would have had to take place sooner if the United States had not provided financial assistance, including Marshall aid. Third, in the case of Britain, there was the problem of sterling balances. These balances accumulated during the war as Britain drew on the sterling area for resources. By the end of the war, the balances were about 200 percent of exports and 50 percent of GNP, debt indicators that are about the same as those of Mexico in 1989. Today we would say that Britain had a serious debt problem, except that it also had significant external assets in 1945. The standard prescription today would be to devalue, reschedule the debt, and adjust. Why was this not done? Skepticism about relative price changes and the perceived unfairness of requiring further hardships after those of the war have already been noted. In addition, devaluation was strongly opposed by the holders of sterling balances, including India. Britain could, however, have funded these balances and provided purchasing power guarantees. That it chose not to do so must be due partly to its desire to retain sterling’s role as an international currency. The failure to deal decisively with the balances early constrained British policy for the next twenty years; London’s role as an international financial center turns out not to require the use of sterling as an international currency. Which approach to current account convertibility is right, the postwar West European approach or the current East European theory? Or is each right for its times? The East Europeans need the price signals that come from trade liberalization more than the West Europeans did forty years ago. No doubt, too, East European practice will be closer to West European practice than is the current theory, and that will probably be to the good since it should help mitigate declines in physical output. Still, the tentative answer is that the West Europeans adjusted too slowly.

International Capital Flows Growth performance in the developing countries was good in the heyday of the Bretton Woods system. Since 1980, the developing world has grown on two separate tracks, the fast and medium growers of Asia and the slow and negative growers of Latin America and Africa. Much of the responsibility for these differences in performance lies with domestic policymakers, but some rests with the purveyors of international capital. The euphoria over the successful recycling of oil revenues in the 1970s was widely shared by policymakers in the industrialized countries, bankers, the international agencies, and academic economists. Few warning voices were raised about the dangers of the growing debt, even between 1979 and 1982, when exchange rate overvaluation became the norm in the borrowing coun-

597

Panel Session I1

tries. There is much blame to be shared for permitting the debt crisis to develop, just as there will be much blame to share next time there is a debt crisis, as there will be. The international system dealt much too slowly with the debt crisis and imposed too much of the burden on the developing countries. Now, ten years after the debt crisis began, some of the heavily indebted Latin American borrowers are coming back to the markets. Chile and Mexico have earned their way back. But, incredibly, private-sector loans are being extended to some countries that have not yet adjusted their internal policies or dealt with their existing debts. The private international capital markets are often said to have the memories of elephants; their memory is more like that of the crocodile, which is said to be twenty-four hours. The lessons of the debt crisis for the international system need to be drawn and acted on. The international agencies need to ask what measures have to be taken to prevent or at least delay the next debt crisis. At a minimum, the agencies should take a far more public role not only in monitoring but also in evaluating international capital movements. While capital flows to Latin America are beginning to resume, there is little prospect of private capital flowing to sub-Saharan Africa on an appreciable scale. Africa will have to rely on continued large-scale aid and support from the international community, including reductions in its debt burdens. Now, at the start of the 1990s, the problem of capital flows to Eastern Europe is at the top of the international agenda. Over the longer term, the countries of Eastern Europe, and especially the former Soviet Union, should be able to attract large inflows of foreign direct investment-indeed, Hungary is already beginning to do so. But, in the immediate future, the bulk of international capital flowing to these countries will have to continue to come from the public sector, including the Bretton Woods organizations. The two agencies successfully set up nearly a half century ago will have much to do in the next decade in the developing countries and in the reconstruction and development of Eastern Europe and the former Soviet Union. This being the preanniversary of the Bretton Woods conference, we should also ask what the agencies will be doing fifty years from now, at the Bretton Woods centennial. But that would be an academic question.

Bretton Woods, the Marshall Plan, and the Postwar Dollar Standard Ronald I. McKinnon After the final breakdown of the Bretton Woods par value system in 1971-73, the unexpectedly violent fluctuations in untethered relative currency values Ronald I. McKinnon is William D. Eberle Professor of Economics at Stanford University.

598

Ronald I. McKinnon

greatly strengthened the tendency to form regional trading blocs-within which more stable exchange rates can be more easily established. However, exchange rate swings among the major blocs remain as big as ever. Over the past four years, the yeddollar and markldollar exchange rates have moved as much as 25 or 30 percent. From the mid- 1970s onward, this exchange rate uncertainty provoked, and is provoking, a resurgence in interbloc protectionism-mainly in the form of quantitative restrictions rather than tariffs. Indeed, when exchange rates are highly volatile and close to being randomly determined, much of the resulting exchange risk cannot be effectively hedged.’ Consequently, governments tend to offset some of this risk by imposing quantitative restrictions-such as import quotas-on trade between currency areas. Because they insulate the domestic economy from exchange fluctuations with lesser restraint on the volume of trade, quotas are much more efficient than “equivalent” tariffs*whence the proliferation of quota protection for agricultural markets, “voluntary” export restraints in automobiles and steel, market-sharing agreements in textiles and semiconductors, sliding-scale export subsidies, and so on. Largely because of exchange rate instability among trading blocs, in the 1990s the industrial world is lapsing into this rather dangerous mercantilistic rivalry. But need commitments to the General Agreement on Tariffs and Trade (GATT), and to freer global trade based on the most-favored-nation principle, atrophy because of currency instability? History has much to tell us about worldwide monetary standards among countries that were not tightly integrated into regional groupings. Following a period of currency disorder after World War 11, virtually stable par values for exchange rates among all (noncommunist) industrial countries from 1950 to 1970 successfully undergirded the GATT. By the end of the 1960s, quantitative restrictions in trade among the major industrial countries had been largely eliminated, and tariff protection was moderated. The common price level in terms of tradable goods (as measured by WPIs) was virtually stable (see fig. 13.1). Moreover, real output growth from 1950 to about 1973 was higher than seen before or since-what Angus Maddison calls the “golden age” of the world e~onomy.~ After 1971-73, tariff levels continued to drift downward under successive GATT negotiations-but quantitative restrictions among emerging trading blocs began to escalate. Exchange-rate and price-level volatility increased, while real economic growth in the industrial economies slowed sharply. 1 . See Ronald I. McKinnon, “Monetary and Exchange Rate Policies for International Monetary Stability: A Proposal,’’ Journal of Economic Perspectives (Winter 1988): 83-103. 2. See Ronald I. McKinnon and K. C. Fung, “Floating Exchange Rates and the New Protectionism,” in Protectionism and World Welfare, ed. Dominick Salvatore (Cambridge: Cambridge University Press, forthcoming). 3. Angus Maddison, The World Economy in the 20th Century (Paris: Organization for Economic Cooperation and Development, 1989).

599

Panel Session I1

Nominal Exchange Rates fQ51=100Annual data

Im

30 I 1950

I

I955

I

I960

I

1965

I

I970

I

1975

I

1980

I

1985

I! 90

--Yen/$

----DM/$ 400

350-

300250-

200150100-

50

I

I

I

I

I

I

I

30

But this presents a paradox. If the monetary order of ‘‘virtually’’ fixed exchange rates from 1950 to 1970 was so successful, why did it collapse? Why were academic economists-both Keynesians and monetarists-so generally hostile to the fixed rate system well before the final breakdown? The answers are important in understanding whether a common monetary standard across similarly diverse economies is feasible in the 1990s.

600

Ronald I. McKinnon

The Origins of the Fixed Rate Dollar Standard: Bretton Woods or the Marshall Plan? Through common usage, economists refer to the postwar monetary order based on pegged par values for exchange rates as the “Bretton Woods system.” Similarly, the collapse of the commitment to fixed par values in 1971-73 is commonly referred to as the “collapse of Bretton Woods”-as per my own usage in the first sentence of these remarks. But this conveniently plausible shorthand terminology is deceptive. The Articles of Agreement negotiated by Britain and the United States, and then presented to an assemblage of forty countries in the legendary town of Bretton Woods, New Hampshire, in July 1944, were essentially different in spirit from the fixed-rate dollar standard that had evolved by 1950. The postwar monetary order that John Maynard Keynes, the principal British negotiator, and his American counterpart, Harry Dexter White, envisaged in 1944 is summarized by the six rules in rule box 1.4 In interpreting the “spirit of the treaty” of 1944 in rule box 1, let me emphasize just two aspects: 1. Symmetry. The rules were intended to apply to all nations more or less equally, not to establish an asymmetrical key-currency regime. 2 . National macroeconomic autonomy. Each country was to have free rein to determine its own level of aggregate demand and rate of price inflation unconstrained by any international monetary standard (rule 6, box 1). Not only did the negotiators seek to escape from the discipline (fetters?) of the classical gold standard, but they had no intention of reestablishing a world monetary standard with a common price level or “nominal anchor.” In particular, Keynes was adamant that each government have the macroeconomic autonomy to manage its own aggregate demand and to choose its own rate of national price inflation5-whence the concern at Bretton Woods that governments have exchange rate flexibility in the longer run (rule 2, box l), although par values were to be stable in the short run. Changes in official par values were to compensate for differing rates of national price inflation or to help secure appropriate adjustment in “real” exchange rates. In intervals between these discrete changes, economies could remain somewhat insulated from each other by retaining capital controls on the balance of payments supplemented by generous credits from the IMF. Beyond the IMF articles themselves, Keynes’s macro views triumphed in academe as well. The primacy of national macroeconomic autonomy-and 4. A detailed explanation of, and rationale for, each rule is provided in Ronald I. McKinnon, “The Rules of the Game: International Money in Historical Perspective,” Journal of Economic Literature (forthcoming). 5 . See John Williamson, “Keynes and the International Economic Order” (1983), in Political Economy and International Money: Selected Essays of John Williamson, ed. C . Milner (New York: New York University Press, 1987).

601

Panel Session I1

Rule Box 1 THE BRETTON WOODS AGREEMENT IN 1945: THE SPIRIT OF THE TREATY All Countries 1.

Fix a foreign par value for the domestic currency by using gold, or a currency tied to gold, as the numeraue; otherwise. demonetize gold in all private transacting.

11.

In the short run, keep exchange rate within one percent of its par value: but leave the long-run par value unilaterally adjustable at the behest of the country in question.

111.

Free currency convertibility for current-account payments; use. capital controls to dampen currency speculation.

IV.

Use national monies symmetrically in foreign transacting, including with the International Monetary Fund (IMF).

V.

Buffer short-run payments imbalances by drawing on official exchange reserves and IMF credits; sterilize the domestic monetary impact of exchange-market interventions.

VI.

National macroeconomic autonomy: each member government to pursue its own price level and employment objectives unconstrained by a common nominal anchor or price rule.

the consequent need for exchange rate flexibility to secure international adjustment-dominated, and still dominates, postwar textbooks on openeconomy macroeconomics. Beginning with James Meade’s seminal Balance of Payments, one can trace this line of thought through Fritz Machlup, W. M. Corden, Milton Friedman, Harry Johnson, Paul Samuelson, and many other older authors, down to the current generation as reflected in the works of Rudiger Dornbusch and the just-published tract by Paul Krugman Has the Adjustment Process Worked?6 All emphasize the importance of leaving exchange rates flexible ex ante in order more easily to secure adjustment in the balance of trade ex post. Thus, the academic profession “bought” the Keynesian idea of the primacy of national macroeconomic autonomy, and the corresponding need for flexibility in nominal exchange rates, that lay at the heart of the 1944 Bretton Woods Agreement. 6. James E. Meade, The Balance of Puyments (London: Oxford University Press, 1951); Paul R. Krugman, Has the Adjustment Process Worked? (Washington, D.C.: Institute for International Economics, 1991).

602

Ronald I. McKinnon

But, after the Bretton Woods conference, an apparent historical aberration ensued. By 1950, a highly asymmetrical fixed-rate dollar standard had evolved that, to academic observers, seemed unduly rigid. Countries other than the United States found it increasingly awkward to adjust the par values of their exchange rates. Over the next twenty years, very few exchange rate changes among the industrial economies occurred-and these were all quite modest by modem standards. Worse, the United States emerged as the only country with a truly independent monetary policy-and it provided the nominal anchor for a common price level in tradable goods. The American (and world) wholesale price index remained remarkably stable from 1951 to 1969 (see fig. 13.1 above). Other countries were inadvertently caught in a strait jacket-that is, a new and apparently unplanned international monetary standard-where the elbowroom for exercising national macro autonomy was limited. The mixture of written and unwritten rules governing this new standard is laid out in rule box 2. Rule box 2 shows the asymmetrical rules by which the fixed-rate dollar standard actually worked.’ In contrast, rule box 1 displays the symmetrical and more flexible rules of the game as intended by the negotiators at Bretton Woods. This discrepancy was particularly vexing to academic economists in whose textbooks the exchange rate received center stage as an instrumental (or endogenously adjusting) variable. How could such a discrepancy arise? What put the world on a fixed-rate dollar standard with unwritten rules so different from the spirit of the Bretton Woods treaty on which it was ostensibly based? After 1945, one could argue that the Bretton Woods Articles never came into effect! The IMF did nothing to alleviate the festering problem of currency inconvertibility in Western Europe (and Japan) in 1946-47 and the seizing up of intra-Western European trade. Instead, a major historical-institutional event-one that the Bretton Woods negotiators did not anticipate in 1944-gave the industrial economies a tremendous push toward the fixed-rate dollar standard. The Marshall Plan was formally begun in April 1948 with the express purpose of using American financial assistance to restore intra-European trade and financial stability, which were in great disarray.8But not until September 1950 was the monetary centerpiece of this great effort, the European Payments Union (EPU), finally completed for sixteen European countries.g The EPU restored multilateral current-account convertibility among Western European currencies by using the dollar as a unit of account for calculating debit and credit balances for each member and as the fundamental means of settlement. At the end of each month, debtor countries had to use up their 7. The nature of which is discussed in McKinnon, “The Rules of the Game.” 8. See Alan Milward, The Posfwar European Recovery, 1945-51 (London: Methuen, 1951). 9. See Jacob Kaplan and Gunther Schleiminger, The European Payments Union: Financial Diplomacy in the 1950s (Oxford: Clarendon, 1989).

603

Panel Session I1

Rule Box 2 THE FIXED-RATE DOLLAR STANDARD: 1950-1970 Industrial Countries Other Than the United States

I.

Fix a par value for the national currency with the U.S. dollar as the numeraire, and keep exchange rate within 1 percent of this par value indefinitely.

11.

Free currency convertibility for current-account payments; use capital controls to insulate domestic financial markets, but begin liberalization.

111.

Use the dollar as the intervention currency, and keep active official exchange reserves in U.S. Treasury Bonds.

IV.

Subordinate long-run growth in the domestic money supply to the fixed exchange rate and to the prevailing rate of price inflation (in tradable goods) in the United States.

V.

Smooth the short-run domestic monetary impact of fluctuations in international payments by partial sterilization of foreign exchange interventions (Bagehot’s Rule).

VI.

Limit current account imbalances by adjusting national fiscal policy (government net saving) to offset imbalances between private saving and investment. The United States

VII.

Remain passive in the foreign exchanges: practice free trade with neither a balance-of-payments nor an exchange-rate target.

VIII. Keep U.S. capital markets open to foreign governments and private residents as borrowers or depositors.

IX.

Anchor the dollar (world) price level for tradable goods by an independently chosen American monetary policy.

X.

Maintain position as a net international creditor (in dollar denominated assets) by limiting fiscal deficits.

dollar exchange reserves, or draw on a line of credit from the EPU, so that creditor countries were assured that they would be paid in dollars. But, for this to work, each European country had to declare an exuct dollar paritywithout even the 2 percent band permitted by the Bretton Woods Agreement-and then rearrange its internal monetary affairs to maintain this dollar parity as long as possible. As the more financially stable “outsider,” the United States alone had the monetary independence to provide a nominal anchor for the group. To improve the credibility of their domestic monetary stabilization plans, the Westem Europeans could then conveniently lean on the United States after 1950

604

Ronald I. McKinnon

(much like the anchoring role Germany played v i s - h i s the other EMS members after 1979)-whence the fixed-rate dollar standard that underpinned the unprecedented world growth of the 1950s and 1960s and the reduction of protectionist barriers to trade. But the commitment to fixed exchange rates eventually broke down because the unwritten rules of the game necessary to keep the dollar standard going differed too much in spirit from its legal cover, that is, the 1945 Bretton Woods articles, and from the principle of national macroeconomic autonomy. lo When, following the advice of most academic economists, President Nixon devalued the dollar in August of 1971 and continued to inflate the American price level at a higher rate than America’s trading partners would tolerate, he was only exercising the American “right” to exchange flexibility and national macroeconomic autonomy promised in the 1945 Bretton Woods Agreement. However, dollar devaluation violated the unwritten rules (understandings) by which the fixed-rate dollar standard had successfully operated for the previous twenty years. To continue with dollar-based par values for exchange rates after 1970, those rules would have required disinflation of the American economy (coupled with the demonetization of gold)12 in order to provide a stable nominal anchor for the system as a whole. As much as any other, the American economy would have been the principal beneficiary from avoiding the monetary disorder of the 1970s and 1980s. The general lesson is clear enough. To curb interbloc protectionism in the world economy, a global monetary standard is both feasible and desirable. But, to return to some kind of par-value system for exchange rates in the 1990s, the rules should be more explicit and likely more symmetricalL3than those prevailing under the highly successful fixed-rate dollar standard.

Robert Mundell At this “Retrospective on the Bretton Woods System,” organized on the twentieth anniversary of its breakdown, my assignment is to draw lessons from our experience. I shall accordingly discuss (1) the special characteristics of the Bretton Woods “system,” (2) the steps that would have averted its breakdown Robert Mundell is professor of economics at Columbia University. 10. See McKinnon, “The Rules of the Game.” 1 1. See Williamson, “Keynes and the International Economic Order.” 12. See R . Triffin, Gold and the Dollar Crisis (New Haven, Conn.: Yale University Press, 1960). 13. See Ronald I. McKinnon, An International Standard for Monetary Stabilization (WashingInstitute for International Economics, 1984), and “Monetary and Exchange Rate Politon, D.C.: cies for International Monetary Stability.”

605

Panel Session I1

in 1971, (3) the defects of the flexible exchange rate system that succeeded it, and (4) measures that I believe would assist in making the present system more effective.

An Agreement, not a System There never was a “Bretton Woods system. ” The Bretton Woods Agreement accommodated the rest of the world to an international monetary system that already existed. After the Tripartite Agreement among the United States, Britain, and France in 1936, the essential structure of the gold-dollar standard was already determined. This tendency was reinforced by the outbreak of World War 11, the resulting inconvertibility of the European currencies, and the increased dependence on the dollar as the international medium of settlement and standard of value. The Fund’s 1944 gold dollar, equivalent to one-thirtyfifth of an ounce of gold, was adopted as the unit of account of the IMF. The dollar was the only gold-convertible currency in the postwar system. This was an outcome of several factors: the Gulliver-in-Lilliput position of the United States in the immediate postwar world economy; the importance of the dollar in every foreign exchange market; the maldistribution of gold in the world (over two-thirds in the United States); and the link between gold reserves and the money supply in the United States (the U.S. gold reserve ratio was lowered from 40 percent to 25 percent in 1945). The asymmetry of the position of the dollar was compatible with the Agreement because of an enabling clause inserted at U.S. behest. The first sentence of Article IV-4 (b) states, “Each member undertakes . . . to permit within its territories exchange transactions between its currency and the currencies of other members only within the limits prescribed under Section 3 of this Article” (i.e., 1 percent of parity). But this would have required the United States to control (or close) its foreign exchange market when U.S. practice was not to intervene in the market at all. Accordingly, at U.S. prompting, a second sentence was added to Article IV-4 (b): “A member whose monetary authorities, for the settlement of international transactions, in fact freely buy and sell gold within the limits prescribed by the Fund under Section 2 of this Article shall be deemed to be fulfilling this undertaking.” Curiously, it was not until 1949 that the U.S. secretary of the Treasury confirmed, in a letter to the managing director, that the United States was “in fact freely buy[ing] and sell[ing] gold.”’ Only then was U.S. practice brought into conformity with the letter of the treaty. Otherwise-apart from the enabling insertion of the gold clause-the Bretton Woods Agreement failed to anticipate the asymmetrical position of the dollar as the intervention currency. The concept of an intervention currency did not then exist, nor is it implied, in the Agreement. To keep the letter of the Agreement on exchange rates, every country not using the gold clause would be required to intervene in every single exchange market whenever

606

Robert Mundell

exchange transactions threatened to move outside the prescribed exchange rate limits. Such an arrangement was tantamount to doing without a numeraire. A third sentence should have been added to Section IV-3 (b) to accommodate nongold countries. The Fund Agreement had to be stretched, by a bylaw, to permit, as a “multiple exchange practice,” other exchange rates to diverge from the prescribed limits if a country was keeping its own exchange rate fixed, within the required limits, to a convertible currency. This meant in practice that, as long as a country was pegging the dollar within the prescribed limits, it would be absolved from the need to intervene in any other exchange market on its territory. The major countries elected to peg the dollar, and, by so doing, they exempted themselves from the other provisions of Article IV-4 (b). An analogous bylaw relating to multiple-exchange practices proved to be necessary to finesse the problem of exchange rates moving outside the prescribed limits in the case of dependent currencies; if, say, currencies of the franc and sterling areas were pegged to those currencies within the prescribed 1 percent limits and the French and British each pegged their currencies to the dollar within 1 percent limits, swings in the exchange rate between the upper and lower limits of the two dependent currencies could be many times larger than the prescribed limits. Inside the Fund, it was necessary to cope with another problem that resulted from the failure of the architects to anticipate the asymmetrical nature of the actual system. In Article V-3 (a) (i), dealing with the use of the Fund’s resources, it was required that “the member desiring to purchase the currency represents that it is presently needed for making in that currency payments which are consistent with the provisions of the Agreement.” But it was the dollar that was needed for fixing exchange rates. Even when the dollar came under attack in the late 1950s and early 1960s, deficit countries needed to draw dollars from the Fund, aggravating the dollar’s weakness. Theoretically, the IMF was supposed to be a revolving credit system, with unchanged total assets that were always maintained in value. Thus, when a country devalued, it was required to increase the quantity of its currency to the Fund to maintain its gold value. It soon became clear, however, that the inconvertible currencies held by the Fund were of no use for other members to draw. To this extent the Fund became illiquid. When the dollar became weak, the Fund’s liquidity became inadequate. The 1961 General Agreements to Borrow (GAB) reflected this illiquidity without correcting it. The resources of the Fund proved of little use for the United States itself. To summarize, then, although the international monetary system that had developed in the late 1930s and that characterized the postwar period was not anticipated by the architects of the Fund Agreement, that Agreement was stretched to make it conform to the monetary system as it continued to evolve. The system was in effect an anchored dollar standard that broke down in 1971. The Bretton Woods Agreement, with its two amendments, is still in force.

607

Panel Session I1

Could the 1971 Breakdown Have Been Averted? When, on 15 August 1971, President Nixon announced the suspension of the external convertibility of the dollar, many of the major countries reacted by dropping their peg to the dollar. Although the European countries preferred fixed exchange rates, they were reluctant to peg an inconvertible dollar and unable to coordinate actions for a joint European float. For a few months, exchange rates floated. The float came to an end with the new system adopted at the Smithsonian Institution in December 1971. The new arrangements were in effect an unanchored dollar standard because the United States was no longer buying and selling gold at the new $38.00 an ounce price. The formal creation of what amounted to an unanchored paper dollar standard broke a precedent because it imposed obligations on the rest of the world but not on the United States. With no convertibility requirement for the United States, the system broke down within two years. The architects of the Smithsonian Agreement misconceived the major problem of the anchored dollar standard, and they lost a golden opportunity to rectify its defects. The anchored dollar standard broke down because of the undervaluation of gold. An excess demand for gold had developed in the aftermath of World War I1 inflation, the external convertibility of European currencies, and the decision of European countries to accumulate gold reserves at the expense of the United States (as was their right under the system). The situation after World War I1 had much in common with that after World War I. In both cases, wartime inflation had lowered real gold balances relative to trade and output, creating a gold scarcity that was temporarily averted by the use of foreign exchange reserves in lieu of gold. In both cases, there was sufficient gold for an anchored dollar standard, but not enough to fulfill the needs for substantial gold holdings on the part of the rest of the world. And, in both cases, the real value of gold appreciated in the crisis stage. The experiences part company, however, in the manner in which the problem was dealt with. When the crisis emerged in the early 1930s, the gold scarcity was dealt with in the United States and France by deflation, within the context of the prevailing gold parities. (Sterling countries, however, wisely opted out of the deflation in 1931.) The postwar system did not repeat the mistake of deflation and depression. Instead, the market price of gold was decontrolled in 1968, and gold transactions came to a standstill between central banks. That solution, however, had the defect of changing the system from an anchored to an unanchored one. Raising the price of gold was an alternative, advocated by Sir Roy Harrod, Jacques Rueff, and others. A provision was made in Article IV (7) of the Articles of Agreement: “The Fund by a majority of the total voting power may make uniform proportionate changes in the par values of the currencies of all members, provided each such change is approved by every member which has

608

Robert Mundell

ten per cent or more of the total quota.” The voting restriction gave the United States a veto. It is necessary to understand that the gold undervaluation problem involved both stocks and flows. The flow problem had been dealt with by the agreement to create gold-guaranteed special drawing rights (SDRs), enacted in the First Amendment to the Articles of Agreement. The first allocation of the SDRs had been made in 1970, and further regular allocations would have provided for reserve growth over time, permitting the gradual phasing out of the dollar and gold as the principal reserve assets. Of course, dollars would still be needed for working balances as long as the asymmetrical exchange system lasted. But the paper gold SDR had a fair chance of working if the system had been in equilibrium to start with. This was, however, an excess stock demand for gold. There were two ways to meet this problem: (1) a very substantial initial increase in the amount of paper gold or (2) an increase in the price of gold. But a very substantial allocation of SDRs-perhaps as much as $40 billion worth, probably combined with the much-discussed Gold Substitution Fund-was not politically negotiable. Neither was an increase in the price of gold.

Defects of the Unanchored Systems The unanchored dollar standard created at the Smithsonian Institution broke down because of the inflationary monetary policies of the United States. Following the second devaluation of the dollar in February 1973, Europe tried again to organize a joint float. Again, no consensus could be reached, mainly because of the problem of how to settle intra-area balances. In June 1973, the Committee of Twenty abandoned the pursuit of international monetary reform for a regime of flexible exchange rates. The floating exchange rate system shifted the responsibility for inflation from the center country to individual countries. But abandoning the system did not improve matters. The world money supply became highly elastic. Countries adopted more passive monetary policies, accommodating price increases initiated from the side of costs. Within six months of the adoption of flexible exchange rates, the price of oil quadrupled. The soaring oil prices created huge balance-of-payments deficits in oildependent countries and an explosion of liquidity as receipts of oil-rich countries were recycled to deficit countries in the Eurodollar market. The explosion of liquidity, which had been accommodated by easy money on the part of the Federal Reserve, not only ratified the increase in the price of oil but spread it to the entire commodity structure. In 1974, the price of gold hit $200. A few years later, in 1979, the problem was repeated, and the unanchored international monetary system again permitted the supply of liquidity to meet the demand. Inflation under the unanchored flexible exchange rate regime was greater than every previous experience in peacetime.

609

Panel Session I1

In the 1980s, the inflationary policies of the 1970s were reversed; the rate of inflation was brought down to moderate levels by the middle of the decade. Nevertheless, the cost of the inflationary error and its correction was huge, much larger than realized. Macroeconomic stability was undermined as the overshooting downward of the exchange rate raised costs that were never reversed in the appreciation phases; the ratchet effect came into play, raising the “core” rate of inflation. The debt problems of the 1980s have their roots in the gyrating and overshooting exchange rates. In the buildup of the inflation rate, real interest rates became very low and even negative, leading to a huge buildup of debt in the late 1970s by the developing countries. With the rising interest rates brought about by the anti-inflation program, many of the developing countries became insolvent. Growth in the developing countries came to a standstill. The international debt problem was a child of the unanchored flexible exchange rate system. But that child had a twin. The twin was the domestic financial system of the United States. The instability of the level and structure of interest rates played havoc with the banking system and especially the savings-and-loan associations. Bank failures and saving-and-loan bailouts now promise to cost the taxpayer hundreds of billions of dollars. The unsound condition of American financial institutions can be traced directly to the instability of the level and structure of real interest rates associated with the breakdown of the anchored fixed exchange rate system. Not only have unanchored flexible exchange rates been responsible for accommodating monetary excesses, but they can also be blamed for the collapse of fiscal discipline. Under the fixed exchange rate systems of the past-the gold standard, the gold exchange standard, or the anchored dollar standardcountries have been forced to maintain fiscal discipline. Absence of discipline would quickly result in adverse speculation, reserve losses, and a convertibility crisis. But, under the flexible exchange rate system, deficits in most countries have exceeded 3 percent of GNP and in some countries have attained more than 10 percent of GNP. Under flexible exchange rates, deficits can be accommodated, if necessary, by the monetary authorities even if that accommodation results in depreciation of the currency. The gyrations in exchange rates between 1973 and 1988 have not been conducive to stability. Tighter money would have reduced the rate of inflation and the depreciation of the dollar in the late 1970s, reduced the need for excessively tight money in the early 1980s when the dollar was soaring, and eliminated the need for the Plaza Agreement to depreciate the dollar. The excessive swings in the dollar were manifested also in excessive swings in the price of gold, which often serves as an early indicator of incorrect monetary policies. The unanchored regime of flexible exchange rates proved to be even worse than the unanchored fixed exchange rate system because it created spurious fluctuationsin real exchange rates that later had to be reversed.

610

Robert Mundell

Steps to Improve the International Monetary System The G7 countries took a positive step toward restoring stable exchange rates at the Louvre meeting in February 1987; they agreed to try to stabilize rates around “current levels.” Had the G7 not recognized the need for a mechanism to determine the burden of adjustment, this would have been just another exercise in establishing an unanchored fixed exchange rate system. But the Louvre Agreement was an improvement over the Smithsonian system because it tried to meet the problem of assigning responsibility for adjustment. Should strong currencies loosen monetary policies, or should weak currencies tighten? The use of an inflation index was considered as a means of determining how the burden of adjustment should be distributed. If the index indicated excessive deflation in the world economy, it would be necessary for surplus countries to expand and deleterious for deficit countries to contract, whereas, if the index indicated excessive inflation in the world economy, it would be necessary for deficit countries to contract and harmful for surplus countries to expand. At the annual meeting of the International Monetary Fund in September 1987, in Washington, D.C., Secretary of the Treasury James Baker I11 announced his support for such an index, which, he said, “should include gold.” This promising approach was, unfortunately, cut short by the stock market crash, the latter itself a victim of renewed exchange rate uncertainty. In the confused aftermath of the crash, plans were scrapped, and, since that time, international monetary leadership has been lacking. Meanwhile, the monetary ball has passed to Europe. If Europe moves toward a complete monetary union, the character of the international monetary system will be profoundly affected. But the prospect of monetary union in Europe does not reduce the need to reestablish an effective international monetary system. If the European Monetary Union cannot be brought about, the main European countries will be more eager to reform the international monetary system. If, on the other hand, Europe achieves monetary union, the rest of the world will still benefit from the establishment of an international system with or without the new Europe.

Parameters of Reform As at Bretton Woods, any agreement must be consistent with the economic and political parameters of the system. A solution that fails to accommodate the interests of the major countries will not be negotiable. The important decisions have to be compatible with the interests of the G7 countries and should, perhaps, include Russia as a future great economic power. Most of the other countries would accept an international monetary agreement that was negotiated by the G7 or G8. Europe, however, is a question

611

Panel Session I1

mark. If Europe’s agenda excludes the rest of the world, it will not put the same effort into international monetary reform; in the medium-term future, international and European monetary reform are alternatives, not complements. If, for that reason, Europe drags its feet on international monetary reform, the United States, Japan, Russia, and Canada should proceed without Europe, bringing into the discussions developing powers like India, Brazil, Mexico, and others. Fixed exchange rates will not work in a vacuum. It is necessary to have either an anchor or an alternative arrangement that assigns responsibility for adjustment policies between deficit and surplus countries. As already noted, the burden of adjustment (monetary deceleration) should be on deficit countries when there is global inflationary pressure, and the burden of adjustment (monetary acceleration) should be on the surplus countries when there is global deflationary pressure. Perhaps in an ideal world it would be possible to develop a common commodity basket that each country could use for determining price indexes, after which it would use a weighted average of such indexes of prices in national currencies to determine the burden of adjustment. Another possibility, along the lines of numerous proposals for commodity reserve currencies, would be for each country actually to buy and sell such a basket of commodities. Such a proposal would meet a typical criticism of the gold standard, that the real price of gold is not constant. None of these proposals are free from defects, however, and no one has yet come up with a formal plan that is negotiable. In the meantime, it is worth considering workable second-best solutions. Could the special properties of gold be used as a signal for dividing adjustment measures? If gold were stable relative to other commodities, it would be a good signal. Over the very long run, gold has been stable against commodities, or at least more stable than any other single commodity. Gold was not, however, very stable in the 1970s. The soaring gold price was due to a concatenation of several factors: inflationary monetary policies pursued since 1971; correction for undervaluation since 1934; the appreciation that resulted when the gold prohibition was lifted from American citizens in 1974; and the special connection between the price of oil and the price of gold. In recent years, gold seems to have stabilized around a fairly narrow range of three hundred SDRs per ounce. With annual production in the range of fifty million ounces and an outstanding stock of three billion ounces (composed of official stocks, speculative hoards, and jewelry), normal variations in annual production have only a minor effect on price. The main fluctuations in the price of gold now result from changes in inflationary expectations. That gives gold, more than any other single commodity, special properties useful for the international monetary system. Changes in inflationary expectations are as quickly reflected in the price of gold as they are (in the opposite direction) in the price of long-term bonds. On an experimental basis, the United States, Japan, and Europe could es-

612

Robert Mundell

tablish informal reference ranges for the prices of gold in terms of the national currencies. Suppose that a gold parity were set for each country with action points at ranges of, say, seven and a half cents above and below the gold parity. At the lower action point, the country would ease monetary policy; and at the upper point, it could tighten monetary policy. There are two ways of dealing with changes in the long-run equilibrium real price of gold. One approach would be to adjust the informal central gold parity of each currency to compensate for the change in the real price of gold. This procedure has the advantage of simplicity; its disadvantage is that new parities have to be renegotiated periodically and that variable parities increase uncertainty about future monetary policy. An alternative is to establish a Gold Stubilizution Fund to stabilize the real price of gold. Such a fund would use central bank stocks to stabilize the market and operate somewhat like the Gold Pool organized by eight central banks in the 1960s. Each country could contribute part of its gold stocks to the Fund in exchange for gold-value guaranteed SDRs. The Gold Stabilization Fund could exchange gold against SDRs with the member countries and gold against currencies in the private market, supplying gold to the market when it is rising relative to commodities and taking it from the market when it is falling. The central banks and the official institutions (the IMF and the European Monetary Cooperation Fund [EMCF]) hold about 1.1 billion ounces of gold, equivalent to a twenty-two years’ annual supply at current rates of production. Under the Gold Index Plan I am proposing, gold is used only as a guidepost for monetary policy; countries do not actually buy and sell gold. But part of the gold could be used for helping stabilize its real price. The proposed system could be started on a pragmatic and informal basis, with the major countries experimenting with the implications of using the gold points as signals to change monetary policy. As experience with the system develops, countries may elect to narrow the margins. To the extent that the system is successful, more formal arrangements could be made and the Gold Stabilization Fund established under the auspices of the IMF. Most of the other countries would fare better within a framework for anchored fixed exchange rates. The parity system established at Bretton Woods failed to anticipate the problems of the asymmetrical anchored dollar standard, but it is, ironically, better suited to modem conditions than the Fund Agreement after the enactment of the Second Amendment establishing managed flexible exchange rates. For the new members of the IMF, as well as for most of the smaller countries, an anchored parity system would be more conducive to better policy than the unstructured arrangements now in effect. In the not-too-distant future, it would be desirable to consider a Third Amendment to the Articles of Agreement to establish a new framework for exchange rate parities and steps to allow the SDR to evolve into a genuine international currency.

613

Panel Session I1

Lessons of the Bretton Woods Experience Martin Feldstein

The agreement devised at Bretton Woods in 1944 was supposed to create a system of “fixed” exchange rates that could be adjusted when a country experienced a “fundamental disequilibrium” in its balance of payments. Under this system, currencies were supposed to be fully convertible into both dollars and gold. Controls on international capital movements were expected to permit international differences in interest rates and to avoid capital flight when a currency’s devaluation looked likely. The International Monetary Fund was created to manage this system, with the power to authorize exchange rate adjustments and the ability to provide liquidity to member countries that experienced temporary (and presumably not “fundamental”) balance of payments deficits. Although the Bretton Woods Agreement was accepted by the United States and all the other major nations of the non-Soviet world, the system never worked the way that it was designed to do. Several papers at the conference discussed why the Bretton Woods system eventually broke down and was totally abandoned. I want to comment instead on the more basic question of why the world economy never followed the Bretton Woods rules. Such an analysis has useful lessons for anyone who today thinks about changing the current system in which the dollar, the Japanese yen, the German mark, and a number of other currencies float freely with little more than “verbal intervention” by governments. It may also be useful for those who are considering the desirability of shifting from the current European Monetary System to a monetary union with a single currency. After summarizing the differences between actual experience and the Bretton Woods rules, I will discuss three basic reasons why the Bretton Woods rules were never followed: (1) changes in economic conditions, (2) changes in professional opinion about the most appropriate system of international monetary arrangements, and (3) an unwillingness of major countries to accept the constraints imposed by the Bretton Woods Agreement when it conflicted with national interests.

A System That Never Was Although a fundamental principle of the Bretton Woods system was supposed to be the convertibility of national currencies into dollars and gold, the European countries did not accept current account convertibility in practice until 1958 and capital account convertibility until even later. Prior to that time, they argued that their individual shortages of foreign exchange and the fragilMartin Feldstein is the George F. Baker Professor of Economics at Haward University and president of the National Bureau of Economic Research.

614

Martin Feldstein

ity of their exchange rates made it impossible to accept the requirement of convertibility. Actual practice appeared to conform most closely to the system described at Bretton Woods during the nine years between the establishment of convertibility in 1959 and the end of the international Gold Pool in 1968. But even in those years there were significant departures from the Bretton Woods principles. The major countries showed an unwillingness to adjust exchange rates even when there were large and eventually unsustainable trade imbalances. The surplus countries, particularly Germany, were reluctant to revalue their currencies because of the adverse effects on their export industries. The deficit counties also frequently delayed devaluations until exchange crises developed and resorted to periods of overly tight policy aimed at reducing imports before accepting the need for an exchange rate adjustment. In addition, the major countries broke the explicit Bretton Woods rules when it suited their own national interests. Britain had a major devaluation without appropriate IMF authorization. France adopted a multiple exchange rate system. The value of the Canadian dollar was allowed to float. This occurred against a background in which the increase of world trade and of the overseas holding of U.S. dollars was creating a rising probability of a run on the U.S. gold supply. It became increasingly clear that, since the stock of gold was not increasing in proportion to the value of world trade, a revaluation of gold in terms of all currencies might be needed at some time in the future. Foreign holders of dollars would lose in such a revaluation relative to those who had previously converted their dollars into gold. As confidence in the ability to maintain the dollar price of gold declined, the risk of a run on the dollar increased. In addition, those who feared that the dollar would cease to be convertible into gold also had a strong incentive to convert their holdings from dollars into gold. The low rates of interest on Treasury bills during this period provided only a small inducement to stay in dollars. Although various “agreements” were reached among the major counties to avoid such a run, by 1968 the gold-dollar system could no longer be sustained, and the Bretton Woods system became a pure dollar system rather than a gold-dollar system. In 1971, the United States formally closed the gold window and declared that dollars were no longer convertible into gold. By early 1973, the adjustable peg system had disappeared. The current period of floating exchange rates had begun. As several of the papers at the conference noted, the final collapse of the Bretton Woods system reflected a combination of the fundamental flaws in the gold exchange system itself (the difficulty of adjustment to eliminate undesired trade imbalances and the threat of a gold-dollar convertibility crisis) and the specific problems associated with the inflationary monetary and fiscal policy pursued by the United States after 1965 at a time when the dollar was supposed to provide the nominal anchor for all major currencies. While this explains why the Bretton Woods system did not persist, there is

615

Panel Session I1

perhaps a more fundamental question of why the system never worked as its designers had intended. Why the World Economy Never Followed the Bretton Woods Rules Even the most carefully crafted system of state controls and international rules cannot hope to persist in a world in which economic conditions and ideas are changing and in which national governments have both the moral obligation and the political incentive to act in the interests of their own citizens. Although the end of World War I1 seemed like a natural time to think about the future of the international economy, it was also a time of rapid change in economic conditions that made it particularly unlikely that any complex system of international rules drawn up at that time would be suitable for the actual economy as it evolved. The fundamental mistake at Bretton Woods was not in the particular rules of the system but in the very idea that a detailed system of rules could be crafted that would be applicable to a rapidly changing world. Instead of accepting an arrangement in which exchange rates were determined in the market and national governments had responsibility for sound domestic policies, the architects of the system created rules that appeared to be logically attractive but that were inapplicable in practice. It is not at all surprising therefore that the system of detailed international economic rules developed at Bretton Woods never described the operation of the world financial arrangements and had eventually to be totally scrapped. In a dynamic world governed by real political actors, any system of detailed international economic rules cannot last for more than a very short period of time. It was no doubt particularly difficult for the political leaders and economic officials who designed the Bretton Woods system in the final years of World War I1 to anticipate correctly the way that the world economy would evolve in the decades ahead. The economy had been in depression or war for nearly two decades, virtually destroying world trade. Economic controls had become a way of life in both Europe and the United States. The specter of Communism once again hung over the European continent, threatening to substitute state planning for a market economy. The dynamic changes in international banking and finance that would be brought about in the coming decades by changes in telecommunications and in financial theory could not possibly be foreseen. At a previous NBER conference, Guido Carli, the former head of the Italian Central Bank, explained that he and other Europeans were eager at the end of World War I1 to strengthen international trade as a way of preventing national economic planning of the type then being advocated by the Communists in Italy and other Western European nations. Many economists like Carli regarded stable exchange rates as a necessary condition for the expansion of trade. The need for capital controls in such a system was not considered to be

616

Martin Feldstein

technically difficult (because such controls were already in place) or economically disadvantageous. The lack of real exchange rate flexibility was not regarded as important for macroeconomic stabilization by a generation of economists that believed in the power of discretionary Keynesian domestic policies. Economic growth and trade both flourished in the postwar period. Private international capital markets developed in magnitude and character in ways that were never anticipated. Because the international capital markets were able to supply funds to countries with temporary balance of payments problems, the IMF stabilization lending became unnecessary, and the IMF lost its most powerful lever on national economies. In addition, developments in the capital markets made it harder and harder to enforce capital controls and therefore to maintain an adjustable peg system. The thinking of professional economists also changed substantially over the years since Bretton Woods. The original Keynesian pessimism about the prospects for full employment in the postwar period melted quickly in the light of experience. Attitudes changed also about the importance of permitting capital flows and of using nominal exchange rate adjustments to achieve real exchange rate changes. Although professional thinking is always in a state of flux, floating exchange rates are now more generally favored than they had been at the time of Bretton Woods. Economists recognized the difficulty of a system that linked the supply of international reserves to the stock of gold and that could not revalue gold in terms of the dollar without creating runs on the reserve currency. The changing global conditions in product and financial markets and the changing attitudes of economists would have been enough to cause frequent changes in the Bretton Woods system and its eventual abandonment. But in addition the Bretton Woods system failed to operate as it had originally been designed to because the major governments of the world refused to accept the constraints and responsibilities implied by the Bretton Woods Agreement. As I already noted, countries with trade surpluses did not appreciate their currencies, and deficit countries tried to avoid devaluations for too long. Canada floated its currency. France used a multiple exchange rate system. And the United States unilaterally brought the gold convertibility feature to an end by closing the gold window. Under the gold-dollar and dollar standards, the world depended on a low rate of inflation in the United States to achieve low rates of inflation elsewhere. It is important to note therefore that the inflationary policies of the United States after 1965 were pursued despite restrictions that would in principle prevent such inflationary policies: the requirement that there be one dollar of gold for every four dollars of currency and that the United States stand prepared to provide gold for dollars at a fixed rate. When President Lyndon Johnson found that the combination of the Vietnam War and the “great society” programs would be inflationary unless he was prepared to accept an un-

617

Panel Session I1

popular tax increase or higher real interest rates, he chose to disregard his international obligations. Government pursuit of national self-interest is an inevitable and fundamental political fact. So too is the pursuit of political self-interest by politicians. Foreign obligations and the well-being of foreign countries come a distant third, even for a country like the United States that takes pride in its international role. Although detailed rules and a system of sanctions can be used to enforce narrow microeconomic agreements like the GATI' rules on dumping and tariffs, macroeconomics is both too vague and too important to be subject to such a control process. Recent G7 experience with attempts to coordinate monetary and fiscal policies shows just how futile such activities are. There is no unambiguous way to monitor the efforts that countries make to expand or contract aggregate demand and no way to hold a government responsible for promises that must be accepted by a parliament, a congress, or a central bank that it does not control. The basic implication of this is that an international monetary system is fundamentally flawed if it depends on governments to subordinate their national interests in the management of macroeconomic policy in favor of international goals. Similarly, the only sure guardian of a low rate of inflation is a disciplined domestic monetary authority.

Four Lessons The experience with the Bretton Woods Agreement suggests four basic lessons. First, any system of specific rules for controlling international macroeconomic relations is likely to be short lived because of fundamental changes in economic conditions. Second, a realistic international monetary system cannot be based on the idea that governments will subordinate national interests to international cooperation. Even if short-term trades are possible, it is unlikely that a government will accept a current sacrifice of national interests in exchange for the prospect that other nations will make sacrifices that benefit it in the future. Macroeconomic issues are too important to countries and too vague to be subject to a system of effective internationalcontrols and sanctions. Third, if nominal exchange rates are fixed, necessary real exchange rate adjustments require changes in domestic wages and prices that can be slow, painful, and costly to the national economy. It is generally important therefore to permit nominal exchange rates to adjust. Even if some groups of countries constitute an optimal currency union area within which exchange rates should be fixed, the international economy as a whole is definitely not an optimal currency area. Unfortunately, internationalagreements like the Bretton Woods

618

Martin Feldstein

system and the G7 attempts at macroeconomic coordination focus on nominal exchange rates and emphasize stability over adjustment. A system of managed real exchange rates is not likely to be operational in practice. Fourth, no international agreement can guarantee domestic price stability. The gold-dollar system failed to provide low inflation in the world economy when the United States abrogated its obligation to tie the dollar to gold and permitted rising inflation in the United States. The European Monetary System as it operates today will assist participating countries to keep inflation low only as long as the “anchor currency” achieves a low rate of inflation. The proposed European Monetary Union will provide low inflation for the member countries only if the European central bank is sufficiently disciplined. A country can guarantee a low rate of inflation only by the prudent management of its own monetary policy.

14

Epilogue: Three Perspectives on the Bretton Woods System Barry Eichengreen

Historians following too close on the heels of events, it is said, risk getting kicked in the teeth. Twenty years since the collapse of the Bretton Woods system is sufficient distance, one hopes, to assess the operation of the postWorld War 11 exchange rate regime safely. This chapter approaches the history and historiography of Bretton Woods from three perspectives. First, I ask how the questions posed today about the operation of Bretton Woods differ from those asked twenty years ago. Second, I explore how today’s answers to familiar questions differ from the answers offered in the past. Third, I examine the implications of the Bretton Woods experience for international monetary reform. In doing so, I summarize the contributions of the NBER conference papers collected in this volume. 14.1 The Questions

In the 1960s and 1970s, most of the literature on the Bretton Woods system was organized around the “holy trinity” of adjustment, liquidity, and confidence.’ As Obstfeld (chap. 4 in this volume) demonstrates, these three problems were interconnected. The adjustment problem was whether there existed market mechanisms or policy instruments adequate to ensure the maintenance of balance-of-payments equilibrium. And, if they existed, the persistence of U.S. balance-of-payments deficits raised the question of why such mechanisms or instruments did not restore external balance. The liquidity problem Barry Eichengreen is professor of economics at the University of California, Berkeley, and a research associate of the National Bureau of Economic Research. The author thanks Luisa Lambertini for research assistance, Bill Hutchinson for help with data, and Tam Bayoumi for permission to draw on joint work. Michael Bordo and Maury Obstfeld provided exceptionally detailed and insightful comments. 1. See, e.g., Mundell(l969,22-23) and the references therein.

621

622

Barry Eichengreen

was whether the system had the capacity to supply international reserves (gold, reserve currencies, and special drawing rights [SDRs]) in amounts adequate to satisfy governments and central banks. The difficulty of providing adequate liquidity was that rapid balance-of-payments adjustment by the reserve currency countries limited the growth of foreign exchange reserves. In their desperate scramble to obtain reserves, governments might raise interest rates and pursue contractionary policies, condemning the world to another deflationary spiral like that of the 1930s. The confidence problem, closely related to that of liquidity, concerned the sustainability of the reserve structure. Under Bretton Woods, gold reserves were supplemented with foreign exchange (mainly dollars), but, within six years of the restoration of currency convertibility in Europe, the official foreign liabilities of the United States had come to exceed its monetary gold. If confidence in the dollar ebbed, foreign governments might stage a run on U.S. gold reserves, much like a run by depositors on a commercial bank, bringing the entire system crashing down. Adjustment, liquidity, and confidence remained topics of debate at the NBER conference. Yet the way these questions were posed differed significantly from formulations twenty years ago. And, with two decades of hindsight, authors and discussants suggested that the holy trinity needed to be supplemented by other, equally important questions.

14.1.1 How Good Was the Performance of the Bretton Woods System? This question can be approached narrowly, from the perspective of the exchange rate system, or more broadly, from the perspective of global economic conditions of which exchange market outcomes are one part. Narrowly speaking, did the Bretton Woods years constitute a singular quarter century of international monetary stability, the only precedent for which was the classical gold standard of 1880-1913? Or should the Bretton Woods exchange-rate system be characterized as operating for just a short interval, from the restoration of convertibility in Europe in 1958 to President Nixon’s decision to close the gold window in 1971, or perhaps only until the two-tier gold market was established in 1968? More broadly, what was the performance of real and financial variables under Bretton Woods, compared to periods before and since? When drawing comparisons, contemporary observers had foremost in their minds the interwar years, relative to which the 1950s and 1960s were decades of stability and prosperity. To what extent are their comparisons reinforced or modified by a longer time frame that also encompasses the post-Bretton Woods experience? If it is found that economic performance under Bretton Woods compares favorably with other periods, this leads to a pair of related questions.

14.1.2 Was the Stability of the Economic Environment Responsible for the Smooth Operation of Bretton Woods? After World War 11, growth was rapid, and, it is argued, major macroeconomic disturbances were infrequent. This is not to suggest that shocks were

623

Epilogue

absent between 1945 and 1971. But qualitative accounts suggest that they were dwarfed by the oil shocks and inflationary disturbances of the 1970s and the disinflationary shocks and fiscal changes of the 1980s. If this view is correct, then the singular stability of the Bretton Woods system was simply a fortuitous by-product of an exceptionally placid environment. Thus, it is critically important to determine whether economic disturbances-in particular, supply shocks autonomous to the exchange rate system-were in some sense less pervasive under Bretton Woods.

14.1.3 Was the Smooth Operation of Bretton Woods Responsible for the Stability of the Economic Environment? Crediting the stability of the economic environment for the smooth operation of Bretton Woods may be putting the cart before the horse. The stability of the environment was an endogenous variable to which the exchange rate regime could have contributed. The question then becomes what features of Bretton Woods were conducive to economic stability. One hypothesis is that Bretton Woods institutions provided rules that disciplined policymakers and solved the time-consistency problem with which they are typically faced. Another is that Bretton Woods arrangements discouraged behavior inconsistent with international cooperation and stabilized the international system.

14.1.4 What Was the Role of the IMF in the Operation of the Bretton Woods System? The International Monetary Fund (IMF) was a prominent institutional innovation distinguishing the post-World War I1 exchange rate regime from previous international monetary arrangements. It is tempting to hypothesize that the IMF contributed to the smooth operation of the Bretton Woods system. An adequate statement of this hypothesis requires that one specify the channels through which the Fund made its influence felt. Did the IMF enforce the “rules” of the Bretton Woods game forbidding competitive devaluation and unsustainable financial policies? Did the Fund and the conditionality attached to its loans provide a commitment mechanism preventing policymakers from reneging on promises? Or was its role one of surveillance and of disseminating the information required for the harmonization of economic policies internationally?

14.1.5 What Explains the Maintenance of Capital Controls? The architects of Bretton Woods envisaged a system characterized by limited international capital mobility. This was true of American as well as British negotiators: how else can one understand the American proposal, which would otherwise have been rendered infeasible by speculative pressures, that major exchange rate changes be approved before the fact by three-quarters of the members of the Stabilization Fund?*The question is why the framers were 2. For details on the negotiations in conjunction with which these proposals were mooted, see Bordo (chap. 1 in this volume).

624

Barry Eichengreem

so confident of their ability to enforce restrictions on capital mobility. Did they believe that foreign investors, disheartened by losses on foreign bonds in the 1930s, would not be inclined to circumvent capital controls? Or had the effectiveness of wartime controls convinced policymakers that financial market restrictions could be enforced in peacetime as well? Did they have any inkling of the revolution in financial services technologies and financial market structures that would undermine the effectiveness of controls starting in the 1960s?

14.1.6 What Caused the Breakdown of Bretton Woods? There has never existed a consensus on the causes of the breakdown of Bretton Woods. At the NBER conference, at least six distinct explanations were advanced: differences between U.S. and foreign monetary policies, differences between U.S. and foreign fiscal policies, failure of deficit countries to devalue, failure of surplus countries to revalue, a secular decline in the international competitive position of the United States, and flaws in the structure of the system (notably Triffin’s liquidity dilemma). All these hypotheses are familiar from the literature of the 1960s. What differed at the conference was the way in which they were formulated. Discussion was systematized by the use of models of collapsing exchange rate regimes developed since the demise of Bretton Woods (see Salant and Henderson 1978; and Krugman 1979). 14.1.7 What Are the Lessons for International Monetary Reform? Each of the preceding questions has obvious relevance for the architects of Europe’s prospective monetary union and for policymakers contemplating wider international monetary reform. If the stability of exchange rates under Bretton Woods contributed to the stability of other real and financial variables, can Europe expect to reap similar benefits if in the 1990s it fixes its exchange rates once and for all? Should other countries expect to reap the same benefits if they negotiate a wider exchange rate stabilization agreement? Which of the factors helping sustain the Bretton Woods system-conceivably including a placid economic environment, the credibility of policymakers’ commitment to the system, and the convergence of policies across countries-remains a prerequisite for exchange rate stability? What flaws in the structure of the Bretton Woods system must be avoided by the prospective European Monetary Union (EMU) and by wider plans for international monetary reform? 14.2 The Answers

14.2.1 Macroeconomic Performance under Bretton Woods According to the conventional wisdom, stable exchange rates encourage financial market integration (see, e.g., McKinnon, in press). Once the uncertainty associated with exchange rate fluctuations is removed, one should see

625

Epilogue

international financial transactions proliferate, forcing interest rates in different countries to equality and smoothly financing current account imbalances. These are among the benefits that should have been reaped from the Bretton Woods agreement. Marston (chap. 11 in this volume) finds little support for this view. Contrary to the conventional wisdom that exchange risk is relatively low in periods of stable rates, the exchange risk premium, or, more precisely, its average value as measured over relatively long periods, showed little tendency to rise following the collapse of Bretton Woods. Since the exchange risk premium depends on the perceived riskiness of exchange rate changes, not on the frequency of actual changes, Marston’s results suggest that the perceived risk of exchange rate changes was not significantly lower in the 1960s than in succeeding decades. This points to a critical distinction commonly glossed over by analysts of Bretton Woods: rather than a system of fixed exchange rates, it was a system of pegged but adjustable rates whose adjustment was a matter of uncertainty. Pegged exchange rates therefore did little to reduce the risk premium or to promote financial market integrati~n.~ Marston’s analysis of covered interest differentials points similarly to the conclusion that financial market integration was limited under Bretton Woods, Covered interest differentials can exist in an environment of high capital mobility if controls are used to limit capital movements. Marston finds that capital controls, as reflected in the covered interest differential, were more pervasive in the 1960s than following the collapse of Bretton Woods. International capital mobility was correspondingly reduced. These restrictions on international capital movements were one of the distinctive features that differentiated the Bretton Woods years from other periods of exchange rate stability. Using savings-investment correlations across countries A la Feldstein and Horioka (1980) as a measure of capital mobility, Bayoumi (1990) found that these correlations were much lower, indicating higher capital mobility, during the classical gold standard years than since 1965. I reran these same savings-investment regressions for the Bretton Woods years.4 For the Bretton Woods period as a whole (1946-70), the coefficient on savings is unity. It remains unchanged when separate regressions are run for the preconvertible and convertible Bretton Woods subperiods. This contrasts with a point estimate for 1880-1913 of 0.63, indicative of much larger net international capital ROWS.~ 3. Marston also finds, however, that, while the mean of the risk premium was the same during and after Bretton Woods, its standard deviation was larger following the system’s collapse. This suggests that the post-1973 period may have nonetheless been characterized by a relatively large number of short periods when the risk premium was unusually large (although over time these periods canceled out one another’seffect on the mean). 4. Both saving and investment were expressed as ratios of GNP. Data were drawn from the same sources as in Eichengreen (1992). 5. The standard error is 0.31 (Eichengreen 1991, table 1). Obstfeld (chap. 4 in this volume) points out that it may be perilous to base inferences on capital mobility on savings-investment

626

Barry Eichengreen

It does not appear, then, that the pegged rates of the Bretton Woods period were conducive to financial market integration. If anything, the opposite was true, given Bretton Woods’s dependence on capital controls, which were more prevalent than under previous fixed-rate systems. As Marston observes, postWorld War I1 capital controls were a potential source of allocational inefficiency. Thus, there was nothing particularly admirable about financial market performance under Bretton Woods. The story is different when one turns to the behavior of real variables. As documented by Bordo (chap. 1 in this volume), the Bretton Woods period exhibited the most rapid GDP growth of any modern exchange rate regime. Bordo’s comparisons extend back to the classical gold standard and forward to the post-1973 float. He shows that, whereas national income in the G7 countries grew by 4.2 percent per annum during the Bretton Woods period, its growth has averaged only 2.2 percent per annum since 1974. Other real variables reinforce the picture of favorable macroeconomic performance during Bretton Woods. Output variability was lowest in the convertible Bretton Woods period (1959-71). Real exchange rates were least variable during this period. Real interest rates were exceptionally stable. The standard deviation of short-term real interest rates was lowest during the Bretton Woods convertible period. The standard deviation of long-term real rates was lower only during the classical gold standard years. As implied by the previous paragraph, there are important differences between the periods before and after the restoration of convertibility in 1958. Although the average rate of output growth was virtually the same, the performance of most other real variables was significantly better following the restoration of convertibility. Output growth, real exchange rates, and real interest rates were more stable. These findings point to a pair of further questions. Was the restoration of convertibility itself responsible for the very different performance of real variables? Or did the growing stability of real variables permit European countries to restore convertibility? 14.2.2 The Stability of the Economic Environment The stability of economic growth during the convertible Bretton Woods period is incontestable. The issue is the causality: whether macroeconomic stability was responsible for the successes of Bretton Woods, or the converse. The implication of the evidence cited above, that there was nothing exceptional about financial market performance under Bretton Woods, is that responsibility rested ultimately with the economic environment. The Bretton Woods exchange rate system was first and foremost a financial arrangement. If it failed to enhance the performance of financial variables significantly, it ~~

correlations,especially when cross sections of long time periods are used. The equality of savings and investment on average over long periods may simply conflate successive periods of imbalance in opposite directions. Still, the contrast between coefficients of 1.0 and 0.6 would appear to be very strong evidence of a qualitative difference.

627

Epilogue

could hardly have stabilized the real magnitudes on which those financial variables acted. Yet pegged exchange rates were only one element of the Bretton Woods regime, which also included commitments by governments to liberalize international trade, encourage investment, and support the unemployed. High investment stimulated aggregate supply, trade liberalization fueled aggregate demand, and the welfare state bought labor peace. These were the ingredients of the golden age of rapid growth following World War II.6 The Bretton Woods system could have been a necessary precondition for the successful conclusion of this bargain, insofar as stable exchange rates promoted the expansion of international trade and provided a nominal anchor for wage demands. Given the existence of arguments pointing in both directions, determining whether the stability of the economic environment was caused by or followed from the smooth operation of Bretton Woods requires a systematic attempt to distinguish supply from demand disturbances. Supply shocks can be taken to reflect the stability of the underlying economic environment, while demand shocks reflect the stabilizing or destabilizing influence of demand management policy.’ One way to distinguish supply from demand disturbances, which I utilize here, is the structural vector autoregression approach developed by Blanchard and Quah (1989). But, unlike Blanchard and Quah, who examined the time-series behavior of output and unemployment, I follow Bayoumi (1991) by instead considering output and prices, which allows me to interpret the results in terms of the familiar aggregate supply-aggregate demand framework (see also Bayoumi and Eichengreen 1991).8 To distinguish supply from demand disturbances, I impose the identifying restrictions that aggregate demand disturbances have only a temporary effect on output but a permanent effect on prices while aggregate supply disturbances permanently affect both prices and o ~ t p u t . ~ 6. De Long and Eichengreen (in press) elaborate on these points. 7. One can think of exceptions to this association of demand shocks with policy and supply shocks with other factors, e.g., changes in tax policy that shift the aggregate supply curve or changes in households’ rates of time preference that alter demand. It is straightforward to reinterpret statements in the text accordingly. 8. Details on the methodology appear below in the appendix. Subsequent work has gone on to impose further identifying restrictions, so as to distinguish money demand from other aggregate demand disturbances (Galf, in press) and labor supply from other supply disturbances (Shapiro and Watson 1988). 9. It is of course restrictive to assume that disturbances affecting output permanently are “supply disturbances” while those affecting output only temporarily are “demand disturbances.” One can imagine models other than the standard aggregate supply-aggregate demand framework in which demand shocks had permanent effects (namely, the hysteresis model of Blanchard and Summers [1986]). Similarly, one can imagine transitory supply shocks (like an oil-price increase that is reversed) that have only temporary effects on output. But a special advantage of my methodology is that the “overidentifying restriction”-that positive demand shocks raise prices while positive supply shocks reduce them-provides an independent check on the aggregate supply-aggregate demand interpretation. In my framework, even if the long-run aggregate supply curve was positively inclined rather than vertical, so that an aggregate demand shock would raise output permanently, it would also raise prices. In fact, shocks that raise output permanently are found to lower

628

Barry Eichengreen

In this kind of time-series analysis, it is critical to employ estimates of economic activity whose cyclical properties are consistent over time. Pre1913 GNP estimates for the United States have been criticized for possessing cyclical properties inconsistent with the interwar and postwar figures.l0 The U.S. figures are built up from time series on production whose constituents vary over time. For Britain, in contrast, materials exist to estimate GNP annually on both production and expenditure bases. Thus, Feinstein’s (1972) estimates for Britain have not been subjected to the same criticisms as the American figures. I therefore carry out the analysis using British data. I utilize Feinstein’s compromise estimates of British GNP, which are an average of his production- and expenditure-side series. The implicit price deflator-the ratio of nominal to real GNP-is used to measure prices. Results are displayed in figure 14.1. I t They show both supply and demand disturbances to have been moderate during the Bretton Woods years. The largest shocks are clustered during World Wars I and 11, leaving open the possibility that the Bretton Woods period was unexceptional for peacetime. This point is reinforced when the interwar period, one of large disturbances, is excluded from consideration. Even after the sample is partitioned in this manner, the second half of the Bretton Woods period stands out. While the standard deviation of aggregate supply disturbances, shown in table 14.1, is no different during the Bretton Woods years (1946-70) than for the period as a whole, there is a strong contrast between the two Bretton Woods subperiods. Supply disturbances appear to have been particularly pronounced from the end of World War I1 to the restoration of convertibility in Europe in 1958. Thereafter, the opposite was true. The standard deviation of supply shocks during the convertible Bretton Woods subperiod was less than one-third the magnitude of that for the period as a whole and only half as large as during the other period of relative stability, the classical gold standard years. This supports Bordo’s partitioning of the Bretton Woods years into preconvertible and convertible subperiods. The same is true of demand shocks. Demand disturbances were larger in the immediate aftermath of World War I1 than following the restoration of convertibility in 1958. Their standard deviation during the years of Bretton Woods convertibility was less than one-third that for the period as a whole and only half as large as during the classical gold standard years. Insofar as supply shocks can be taken as an indication of the stability of the

prices, which is consistent with my interpretation that such shocks are primarily of the aggregate supply variety. Similarly, shocks that affect output only temporarily affect prices in the same direction, consistent with my interpretation of them as of the aggregate demand variety. 10. On this debate, see Romer (1989) and Balke and Gordon (1989). 1 1 . As noted in n. 9 above, the implicit “overidentifying restriction” was satisfied: positive supply disturbances were associated with a fall in prices, while positive demand disturbances were associated with a rise in prices.

629

Epilogue

ii

1'1

-0.05 -

-

.

I

~. ...... Demand -0.10 -

-supply

I

aI . ,

?

i

Fig. 14.1 Aggregate supply and aggregate demand shocks, 1874-1975

Table 14.1

Standard Deviation of Supply and Demand Disturbances for Great Britain, Various Subperiods

~

Period 1874-191 3 1946-70 1946-58 1959-70

Supply

Demand

,015 .025

.017 ,015

,033 ,008

,020 .008

Source: See text.

environment while demand shocks can be taken to indicate the stabilizing or destabilizing influence of demand management policy, this evidence suggests that both the underlying environment and policy contributed to the stability of output growth between 1959 and 1971. Global economic conditions may have been fortuitous for the operation of a pegged-rate system, but policy also played a part. 14.2.3 Contributions of Bretton Woods to Output Stability A first channel through which the Bretton Woods system could have stabilized output was by influencing the formulation of monetary and fiscal policies. Table 14.2 reproduces evidence from Bordo (chap. 1 in this volume) on money growth rates and supplements it with the growth of real government

630 Table 14.2

Barry Eichengreen Rates of Growth of Money Supply and Real Government Spending (% per annum) ~

~

Money Growth U.S. Gold standard (1881-1913) Mean Standard deviation Grand mean Convergence Interwar (1919-38) Mean Standard deviation Grand mean Convergence Bretton Woods (1946-70) Mean Standard deviation Grand mean Convergence Preconvertible Bretton Woods ( I 94658) Mean Standard deviation Grand mean Convergence Postconvertible Bretton Woods (1959-70) Mean Standard deviation Grand mean Convergence Floating (1974-89) Mean Standard deviation Grand mean Convergence

1.8

Canada Japan France Germany Italy U.K. 0.4 1.4

4.6 5.5

-.O 5.0

.6 2.6

.6 3.2

.3 3.1

.6 8.6 2.0 1.7

1.1 4.7

.5 9.7

6.4 8.5

1.3 10.1

3.6 6.2

.8 4.7

6.3 5.8 10.1 4.2

6.0 4.0

17.3 15.9

11.5 7.5

12.8 6.0

13.3 7.8

3.2 3.2

6.4 8.3 11.4 6.0

5.0 3.9

18.2 18.5

14.7 7.2

17.6 5.6

15.9 10.5

1.7 2.9

7.0 1.5 9.8 2.5

9.4 4.3

14.6 2.5

8.6 6.6

10.9 4.7

12.4 2.0

5.5 2.9

8.6 2.4 9.5 2.7

11.0 5.5

5.7 6.2

8.8 3.4

5.7 4.5

13.4 4.9

13.5 5.6

5.0 1.o 1.o

Real Government Spending Growth Gold standard (1881-1 913) Mean Standard deviation Grand mean Convergence Interwar (1919-38) Mean Standard deviation Grand mean Convergence Bretton Woods (1946-70) Mean Standard deviation

3.2 9.3 4.6 2.0

5.7 13.4

10.3 31.2

1.5 5.4

5.6 9.3

4.5 2.4

2.5 .8

1.3 23.8 3.6 3.8

.6 13.7

10.3 14.6

4.8 22.2

6.5 11.4

6.8 -3.1 47.8 14.2

1.1 17.6

1.8 13.9

7.0 11.2

7.6 18.0

7.9 13.6

8.6 10.6

.2 9.5

631 Table 14.2

Epilogue (continued) Money Growth U.S.

4.7 Grand mean 3.4 Convergence Preconvertible Bretton Woods (1946-58) -2.2 Mean 23.4 Standard deviation 3.1 Grand mean 5.9 Convergence Postconvertible Bretton Woods ( I 959-70) 4.7 Mean 5.1 Standard deviation 6.1 Grand mean 1.8 Convergence Floating (197449) 4.0 Mean 3.0 Standard deviation 4.8 Grand mean 1.6 Convergence

Canada Japan France Germany Italy U.K

-1.6 18.2

2.3 16.1

13.3 25.9

11.1 10.2

5.7 4.2

10.1 3.1

3.4 5.3

5.8 15.0

8.6 8.6

4.8 4.9

5.3 4.6

8.7 4.1

3.3 3.0

2.0 3.3

6.5 5.9

4.8 8.8

8.7-3.9 12.2 10.2

Sources: For money, Bordo (chap. 1 in this volume). For government spending, Mitchell (1975, 1983)

and the references cited therein. (The GNP deflator used to normalize nominal spending is also from Bordo.)

spending.12For the G7 countries, money growth rates were more stable during the years of Bretton Woods convertibility than in any comparable period before or since. Their standard deviation was lower than during both the classical gold standard years and the period since 1973. When stability is measured by the coefficient of variation (the standard deviation divided by the mean), the years of Bretton Woods convertibility stand out even more dramatically. Two conditions must be satisfied before we may conclude that the Bretton Woods system was responsible for the stability of money growth: first, that money growth was stable in the center country, the United States, which retained some monetary autonomy; and, second, that the exchange rate system forced other countries to emulate the U.S. example. Table 14.2 confirms that U.S. money growth was exceptionally stable during the years of Bretton Woods convertibility. But even if U.S. policy was stable, Bordo’s results suggest that Bretton Woods institutions forcing other countries to follow the U.S. example cannot account for monetary stability elsewhere. His measures of convergence do not indicate exceptionally close harmonization of monetary polices during the Bretton Woods convertible years. The average deviation of individual countries’ money growth rates from the subperiod grand mean was smaller during the gold standard years and even during the interwar period. The variability of that deviation has been smaller since 1974. 12. Nominal government expenditure is divided by the GDP deflator.

632

Barry Eichengreen

That other countries were not forced to closely emulate U.S. monetary policies and that their inflation rates consequently differed, two facts documented in Stockman (chap. 6 in this volume), again point to the prevalence of capital controls. Capital controls reconciled exchange rate stability with national monetary autonomy. The degree of monetary autonomy, as measured by the international inflation differential, should not be exaggerated, however. Inflation differentials can be observed even when monetary independence is limited. Even if trade and capital controls are absent and market integration is complete, one should still not expect to see identical inflation rates across countries. Opportunities are limited for international arbitrage in nontraded goods, and the relative price of nontradables tends to be low in low-income countries (the Balassa effect). Thus, the relative price of nontradables will rise most quickly where productivity growth is fastest. Even if arbitrage equalizes the prices of traded goods across countries, one should expect to see the consumer price index rise most quickly in those countries with rapid productivity growth. Even if the Balassa effect exists, it may not have had an economically important effect on inflation differentials across Organization for Economic Cooperation and Development (OECD) countries during the Bretton Woods years. Figure 14.2 arrays Maddison’s (1982) estimates of labor productivity growth between 1950 and 1970 against the rate of CPI inflation.I3It confirms a point made by Obstfeld (chap. 4 in this volume): that the U.S.-Japan inflation differential is consistent with Balassa’s story. But it also makes clear that the power of the relation hinges on the inclusion of Japan. When this one observation is omitted from the sample of industrial countries, the productivity-inflation correlation vanishes. Additional factors besides productivity growth, plausibly including capital controls and divergent monetary policies, thus must be invoked in order to account for differences in inflation performance. l4 Turning from monetary to fiscal policy, the second panel of table 14.2 shows measures of the rate of growth of real government spending.I5 At 4.7 percent, the average annual rate of growth of real government spending under Bretton Woods was virtually the same as during both the classical gold standard years and the post-1973 float. But the variability of the growth of real public spending across countries, as measured in the rows labeled “convergence,” was higher during Bretton Woods than in either of the other periods. 13. Labor productivity is measured as output per man-hour, while inflation is measured as the ratio of Maddison’s CPI in 1970 to 1950. Note that Balassa formulated his effect in terms of the behavior of GDP deflators rather than consumer price indices, which include import prices. The distinction is irrelevant if the law of one price holds for traded goods. 14. One way of reinforcing the point is to observe that, if Germany as well as Japan is removed from the sample, evidence of the Balassa effect is strengthened. Germany pursued an exceptionally restrictive anti-inflationary monetary policy behind the shelter of capital controls, which explains why it is an outlier. 15. Sources of the GDP deflators are described in the appendix in Bordo (chap. 1 in this volume), while data on government expenditure are drawn from Mitchell (1975, 1983) and updated using the national sources he cites.

633

Epilogue

+Finland Japan +/

+France

1

r0 2.51

l.

c r

-c c

2.0

Denmark

++/

/ Canada

,

.,

+ Norway +

Netherlands

+'-

Switzerland Belgium

1

2

+Germany 1

1

3

4

5

Productivity Growth, 1950-70

Fig. 14.2 Inflation and productivity growth, 1950-70

The dispersion of growth rates was greater only during the turbulent interwar years. Once again, the Bretton Woods experience combines very different performances before and after 1958. The growth of government spending was slower and less variable after the restoration of convertibility than before. Still, whether measured by individual country standard deviations or by overall convergence, the stability of fiscal policies was no better under the convertible Bretton Woods regime than it has been under the post-1973 float. If policy instruments were not themselves unusually stable, it still could be that the framework within which policy was formulated maximized policy's stabilizing effects. A simple formulation of this hypothesis runs as follows. An expansionary initiative adopted in response to an incipient decline in economic activity is more likely to raise output and employment rather than wages and costs when it is not expected to persist.16If it is expected to persist, workers will demand higher wages in anticipation of higher prices. The shortrun aggregate supply curve graphed in price-output space will be vertical since higher prices will not alter real wages or profitability. If, in contrast, the expansionary initiative is not expected to persist, workers have less reason to 16. This, of course, is the insight behind the Phelps (1967jFriedman (1968) expectationsaugmented Phillips curve.

634

Barry Eichengreen

worry about the inflation-induced erosion of their real incomes, and they will not be as insistent in demanding compensation in the form of higher nominal wages. The short-run aggregate supply curve graphed in price-output space will be positively sloped since higher prices reduce real production costs and enhance profitability. Demand stimulus will produce additional output and employment rather than higher wages and prices. Policy will retain some capacity to stabilize the macroeconomy. How do these points relate to Bretton Woods? America’s stated policy of pegging the dollar to gold at $35.00 an ounce and foreign governments’ desire to peg their currencies to the dollar at prevailing parities may have been interpreted as commitments to a nominal anchor that reduced the perceived probability of persistent inflation, at least toward the start of the convertible Bretton Woods period, when the stability of declared parties was beyond doubt. Agents perceived that countries could not pursue consistently inflationary policies given their commitment to the maintenance of Bretton Woods. Hence, one-time policy initiatives affecting the price level were more likely to stabilize output and employment and less likely to be neutralized by offsetting changes in wages and costs. Two points about this thesis are worth noting. First, it is quite revisionist relative to the older literature (Mundell 1963; Fleming 1962), which argued that monetary policy is less effective under fixed than floating rates. According to my thesis, monetary policy can be more effective under fixed rates so long as it is formulated in a manner consistent with the maintenance of those rates. The eason is that monetary policy has different supply-side effects depending on the regime within which it is formulated. (Recall that the Mundell-Fleming model takes output and employment as demand determined.) In contrast, it is consistent with a subsequent literature (e.g., Dornbusch and Krugman 1976) that argued that floating rates steepened the shortrun Phillips curve trade-off. Second, this view is quite consistent with growing complaints in the late 1960s on the part of monetary policymakers about the capital account offset to monetary policy. So long as the nominal anchor was credible, the capital account offset was small. Higher prices unaccompanied by higher wages, one consequence of which was higher output and employment, also stimulated money demand. Much of the additional money was willingly held; only a fraction leaked abroad through the balance of payments. Once the nominal anchor began to drag, output and employment did not increase; hence, the demand for money did not rise commensurately, and the monetary injection was offset through the capital account. Evidence on this issue is provided by Alogoskoufis and Smith (1991), who estimate time-series models for inflation in the United Kingdom and the United States spanning the period 1857-1987. They find that an AR(1) process for inflation is an adequate univariate representation for both countries. The coefficients differ significantly across periods, however. For both coun-

635

Epilogue

tries, the coefficient on lagged inflation is larger after World War I than before. Moreover, there is evidence of a further rise in its magnitude coincident with the breakdown of Bretton Woods in the early 1970s. Alogoskoufis and Smith then estimate an expectations-augmented Phillips curve in which the change in wage inflation depends on unemployment and on the lagged change in prices.” The coefficient on lagged inflation in this wage-change equation shows the same tendency to rise during World War I and again following the breakdown of Bretton Woods. This supports the notion, described above, that monetary policies affecting prices produced smaller increases in wages and larger increases in output and employment during the Bretton Woods period, when inflation was not expected to persist, than subsequently. Table 14.3 replicates Alogoskoufis and Smith’s U.S. and U.K. inflation regressions, but for subperiods more precisely distinguishing the Bretton Woods years. It reports comparable regressions for two additional countries: France and Japan.’* For each of the countries considered, the coefficient on lagged inflation is larger after World War I than before. More important for present purposes, for each of the four countries the coefficient on lagged inflation is larger after the Bretton Woods period than during it. Recursive regression estimates similar to those of Alogoskoufis and Smith for the period starting in 1948 show a tendency for the coefficient on inflation persistence to rise between 1970 and 1973, the time of Bretton Woods’s breakdown. These estimates are derived by first running the regression with a minimum of degrees of freedom and then adding the observation for each subsequent year. For Britain and the United States, the tendency for the point estimate to rise when post-1971 data is added is evident whatever year in the second half of the 1940s the sample starts.Ig (Plots of the point estimates are displayed in figs. 14.3 and 14.4.) For France, a post-1971 rise in the coefficient estimate is apparent only if one excludes the immediate post-World War I1 years, a period characterized by unusual inflation persistence. For Japan, in contrast to table 14.3, the recursive regression estimates show little evidence of increasing inflation persistence (see figs. 14.5 and 14.6).20

17. Consumer or retail price indices are used. Wages and prices are lagged before differencing, while both current unemployment (differenced) and lagged unemployment (in levels) are included as regressors. 18. Price data for the United States and the United Kingdom are from the sources used by Alogoskoufis and Smith, while consumer price indices for other countries are from Maddison (1982). Efforts to estimate such an equation for Germany are hindered by a break in the price index in 1923-24. The results in table 14.3 differ from those of Alogoskoufis and Smith by the time periods chosen. Their definition of the Bretton Woods period extends from 1948 to 1967, not from 1946 to 1970 as here and in Bordo’s chapter. Similarly, they do not distinguish the preconvertible and convertible Bretton Woods subperiods. Their data set ends in 1987, mine in 1989. 19. The results are consistent with the earlier analyses of Klein (1975) and Barsky (1987). 20. Interestingly, France and Japan are the two countries for which Obstfeld (chap. 4 in this volume) finds little evidence of price stickiness for the Bretton Woods period as a whole.

Barry Eichengreen

636

Estimates of the Coefficient of Lagged Price Inflation in the Price Equation

Table 14.3

U.S. 1892-1970

Britain

France

Japan

.58

~09) .20 (.22) .59

1892-1 9 13 1914-70

(. 11)

1946-70

.48 ~19) .67

1971-89 Source: See text.

Nore: Standard errors in parentheses.

1.5,

1.0 -

--___

,---_ ,______----*.--

- - - _ _* - - _ _ _

- - - - - _ _ _ _ _ . _ _ _ - - *- ----____

0.5 -

-1.01

'

'

1

"

1955

'

'

I

"

1960

"

I

'

1965

"

"

1

1970

'

"

I

'

1975

'

'

'

I

"

1980

Fig. 14.3 Recursive regression estimate of lagged inflation coefficient, United States

Direct evidence on inflationary expectations would buttress the conclusion that inflationary impulses were not expected to persist to the same extent during as after Bretton Woods. Figure 14.7 therefore shows the result of regressing expected inflation in the United States as measured by the Livingston survey on a constant and on its own lagged value.21The figure, displaying 21. The June surveys for 1946-84 were employed. For details on the Livingston data, see Carlson (1977).

Epilogue

637

,*.--

-0.5-1 .o

,

-Recursive C(2) Estimates

_ _ _ _ _ +- 2 S.E.

, , I

-'t:

-

1

.

1955

5

1

1960

"

1965

~

~

1970

~

1975

8

1980

&

~

,

A

~)

I

'

1985

)

~

Fig. 14.4 Recursive regression estimate of lagged inflation coefficient, Britain

a