The IMF and the Future [1 ed.] 9781134700707, 9780415299879

Citation preview

The IMF and the Future

The International Monetary Fund has been criticized from both the right and the left of the political spectrum with the right arguing that it is too interventionist and creates more problems than it solves and the left on occasion demanding that it be abolished altogether. What seems almost beyond question is that the IMF needs to be reformed. Defining a future role for the IMF will always be a controversial issue, but vital to any consideration will be a measured assessment of how it has operated in the past. This excellent new book from an internationally respected expert on the IMF intends to do just that. Starting with a historical background tracing the evolution of the IMF, the book goes on to cover such themes as: • • • •

The circumstances under which countries tum to the IMF The various aspects of IMF conditionality Institutional issues such as lending facilities and how the Fund is resourced The political economy of IMF issues.

Bringing together an array of articles, the book will undoubtedly be required reading for anyone with a serious interest in development studies, international relations and international economic policy.

Graham Bird is Professor of Economics at the University of Surrey, and Director of the Surrey Centre for International Economic Studies.

Priorities in Development Economics Edited by Paul Mosley

The University of Sheffiekl

Foreign Trade Reforms and Development Strategy

Jean-Marc Fontaine Adjustment and Poverty Options and choices

Frances Stewart Econometrics and Data Analysis for Developing Countries

Chandan Mukherjee, Marc Wuyts and Howard White The Green Revolution in Africa An economic and political analysis

Paul Mosley The IMF and the Future Issues and options facing the Fund

Graham Bird

The IMF and the Future Issues and options facing the Fund

Graham Bird

~~ ~?to~~~~~~~up LONDON AND NEW YORK

First published 2003 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN Simultaneously published in the USA and Canada by Routledge 270 Madison Ave, New York NY 10016

Routledge is an imprint of the Taylor & Francis Group Transferred to Digital Printing 2006

© 2003 Graham Bird Typeset in Goudy by Newgen Imaging Systems (P) Ltd, Chennai, India All rights reserved. No part of this book may be reprinted or reproduced or utilized in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers.

British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging in Publication Data Bird, Graham R. The IMF and the future / Graham Bird. p. em. - (Priorities for development economics) Includes bibliographical references and index. 1. International Monetary Fund. 2. International finance. 3. International Monetary Fund-Developing countries. I. Tide. II. Series. HG3881.5.158 B523 2002 332.1'52--dc21 ISBN10: 0-415-29987-X (hbk) ISBN10: 0-415-40687-0 (pbk) ISBN13: 978-0-415-29987-9 (hbk) ISBN13: 978-0-415-40687-1 (pbk)

2002068742

Contents

List of figures List of tables Preface Acknowledgements

vii ix xi xiv

1 A suitable case for treatment?: understanding the ongoing debate about the IMF

1

2 Borrowing from the IMF: the policy implications of recent empirical research

31

3 IMF lending: how is it affected by economic,

4

political and institutional factors?

44

IMF programmes- do they work? Can they be made to work better?

68

5 The effectiveness of conditionality and the political economy of policy reform: is it simply a matter of political will?

93

6 IMF programmes: is there a conditionality Laffer curve?

113

7 The credibility and signalling effect of IMF programmes

130

8 The IMP's role in mobilizing international capital: is there a catalytic effect?

9 Restructuring the IMP's lending facilities

141 163

10 Resourcing the Fund: direct borrowing from private capital markets

179

vi

Contents

11 Crisis averter, crisis lender, crisis manager: the IMF in

search of a systemic role

194

12 The IMF and developing countries: a review of the evidence and policy options

211

13 Political economy influences within the life cycle of IMF programmes

243

14 The political economy of the SDR: the rise and fall of an international reserve asset

267

Index

291

Figures

5.1 Implementation and effect of the IFI programme 6.1 The conditionality Laffer curve 8.1 Flow changes from IMF and private lenders 10.1 Purchases from the IMF 11.1 Total IMF credit outstanding to members 11.2 IMF liquidity ratio 11.3 Rate of incompletion of IMF agreements

98 122 153 182 196 197

200

Page Intentionally Left Blank

Tables

1.1 1.2 3.1 3.2 3.3 4.1 4.2 4.3 4.4 6.1 6.2 7.1 9.1 12.1 12.2 12.3 12.4 13.1 13.2 14.1 14.2 14.3

Net credit and loans from IMF Standby, EFF and ESAF arrangements Estimation results for the basic models Estimation results for the supplemented models Estimation results for the sub-periods Summary of empirical evaluations of the effect of Fund programmes Countries with number and years of IMF programmes, 1980-96 Fraction of IMF loan actually disbursed under each arrangement, distribution by quartiles Fixed-effects estimation of influences over capital flows IMF conditionality - summary of performance criteria, 1993-9 The completion of IMF programmes Size of IMF lending relation to current account balance of payment deficits Use offacilities between 1989 and 1999 Developing countries: net credit from the IMF, 1982-7 Developing countries: net credit from the IMF, 1988-93 IMF arrangements in effect in the financial years ended 30 April1975-95, by facility type Results of before-and-after tests on 16 developing countries with IMF Programme commenced in 1979-85 Estimation results for two models of IMF lending The implementation of IMF programmes: results of random effects model Holdings of SDRs among participating countries Official development assistance of 07 countries, 1980-94 Holdings of reserve assets (all countries), 1988-94

11 13 54 57 58

72 75 78 84 116 125 135 168 214 215 216 230 250 255 274 279 283

Page Intentionally Left Blank

Preface

Over recent years a great deal of attention has been paid to reforming the international monetary system and designing a new international financial architecture. As part of this, the role of the international financial institutions has been scrutinized. As the premier international financial institution it is the International Monetary Fund (IMF) that has received the closest attention. Significantly, the context of this attention has often been the financial crises that were experienced by a relatively small number of countries in Latin America, East Asia, and elsewhere, rather than the Fund's involvement with a large number of poorer economies. Numerous proposals for reform have been put forward to try and mimimize the risks of future crises. In part these have related to private international capital markets and domestic financial systems, but they have also included proposals for reforming the IMF. The Fund has been much criticized from both the right and the left of the political spectrum. The right has presented it as excessively interventionist and as often creating more problems than it has solved. The left has presented it as advocating policies that have often been detrimental to long run growth and development. Some commentators have gone so far as to advocate that the Fund should be closed down altogether, arguing that with flexible exchange rates, and mobile international capital it is no longer needed. Without going this far, others envisage a narrower and more tightly defined role for the Fund in the future, arguing that mission creep over the years, has extended the Fund's operations well beyond its original mandate. They claim that it has moved into areas particularly structural adjustment- where it is ineffective and where other institutions - particularly the World Bank - are better equipped. At the same time others have sought to extend its role still further by bringing the capital account of the balance of payments more firmly within its sphere of responsibility, and by paying greater attention to domestic financial supervision and regulation. Designing a future role for the IMF should be based on a measured assessment of how it has operated in the past. However, this assessment has often been lacking. It is the central purpose of this book to try and provide it. The book identifies a series of key issues, analyses them, draws on empirical evidence where relevant, and discusses policy options in the light of this analysis and evidence. What is the lay-out of the book?

xii

Preface

Almost all the chapters have been recently published in academic journals. They are organized here to provide a logical and coherent discussion of the IMF at the beginning of the twenty-first century. Chapter 1 presents a brief historical background, tracing IMF evolution over time. It also identifies the central issues that have been raised in recent debates about the Fund's future, and discusses the various opinions that have been offered by commentators on the Fund. Chapter 2 examines the empirical evidence concerning the circumstances under which countries tum to the IMF for financial assistance. Are there particular economic characteristics that are common amongst users of IMF resources and what are the implications for policy? Chapter 3 takes a broader look at IMF lending and investigates the extent to which political and institutional factors as well as economic ones help to explain it. Does the data reveal any evidence of systematic political bias? Crucial to any discussion of the IMF supported programmes, Chapters 4, 5 and 6 examine various aspects of IMF conditionality. Chapter 4 investigates the issue of whether programmes work and whether they could be made to work better. It provides a summary of conventional indicators of performance in terms of macroeconomic variables, but also examines other indicators including the degree of programme completion and the degree to which client countries keep on returning to the Fund for assistance - so called IMF recidivism. Chapter 5 analyses the issue of completion in more depth drawing on the political economy of policy reform. Chapter 6 continues with a similar theme by raising the possibility that conditionality may at some point become excessive, such that there are negative returns to incremental conditionality. It has often been claimed in the literature on the IMF that it is via conditionality that the Fund may perform a catalysing role, encouraging other creditors to lend to countries that negotiate and implement IMF programmes. Chapter 7 discusses the credibility of IMF programmes in this context and the extent to which the Fund's involvement is likely to have a beneficial signalling effect. Chapter 8 builds on this discussion and presents an attempt to test empirically for the existence of a catalytic effect. Various forms of disaggregation are used to distinguish between different types of IMF involvement and different types of capital flow. Chapters 9 and 10 raise two institutional issues, and by this means deal with the way in which the Fund can best fulfil its intended functions. Chapter 9 examines the range of lending facilities. It discusses the way in which these have proliferated over the years and explores the scope for more radical rationalization than has been undertaken by the Fund. Chapter 10 turns to the way in which the Fund is itself resourced. Passing comment on the inadequacies of quotas, the chapter discusses the often vexed question of direct borrowing from private capital markets and links this to the ability of the Fund to deal with capital volatility and financial crises. Chapter 11 returns to the fundamental role of the IMF, and examines whether its focus should be on helping to avert crises, providing resources during crises, or, more generally, managing crisis resolution. Since over the period since the

Preface xiii mid-1970s, the Fund has dealt almost exclusively with developing countries, Chapter 12 provides a review of the Fund's role in developing countries, summarizing both the evidence and the policy options. Any discussion of the IMF that relies purely on economics will be missing something important. Chapters 13 and 14 again emphasize the importance of political economy - a theme that occurs in earlier chapters as well. Chapter 13 demonstrates how political issues arise at the different stages of a country's involvement with the IMF, and shows the complexity of the client-Fund relationship. Often overlooked in discussions of the IMF is the Fund's own reserve asset, the Special Drawing Right (SDR). Still on the theme of political economy, Chapter 14 explains how a combination of economies and politics may be used to account for the evolution of the SDR. There is no concluding chapter that attempts to provide the 'right' answers to the questions that form the basis of the on-going debate about the IMF. Although almost all the chapters in this book contain some discussion of policy, the focus is to provide an empirically informed analysis of the principal issues involved in the debate about reforming the Fund and to present a reasoned assessment of the options for reform, rather that to advocate one particular alternative. Having said that the author's favoured solutions probably do permeate through. Now turning to the acknowledgements, a number of people who should be thanked are not. These are the anonymous referees of the journals in which the chapters of this book originally appeared and whose comments allowed me - or forced me - to improve on earlier drafts. I am also indebted to the publishers of the journals for allowing me to reproduce the articles here. I would in particular like to thank Dane Rowlands for allowing me to include some of our joint work in this volume and Tom Willett for commenting on some of the chapters that are included. An institutional thank you should go to the University of Surrey for granting me a period of sabbatical leave which allowed me to work on this collection and to The Fletcher School of Law and Diplomacy, Tufts University for hosting me during my sabbatical. The students of both institutions who have had me try out my ideas on them should also be thanked. Research on the IMF is very much 'work in progress'. My hope is that this book provides a concise and accessible summary of what that work has achieved up to now and how it may usefully be developed in the future.

Acknowledgements

I need to record a number of debts. The first is to the publishers of the journals in which many of the chapters contained in this book originally appeared and to the referees who commented on them at the time. These journals are as follows: Chapter 1, 'A Suitable Case for Treatment: Understanding the Ongoing Debate About the IMF', Third World Quarterly, 22(5), 823-48 (2001). Chapter 2, 'Borrowing from the IMF: The Policy Implications of Recent Empirical Research', World Development, 24(11), 1753-60, November (1996). Chapter 3, 'IMF Lending: How Is It Affected by Economic, Political and Institutional Factors?', The]ournal ofPoUcy Reform, 4(3), 243-70,2001 (with Dane Rowlands). Chapter 4, 'IMF Programmes: Do They Work? Can They Be Made to Work Better?', World Development, 29(11), 1849-65 (2001). Chapter 5, 'The Effectiveness of Conditionality and the Political Economy of Policy Reform', The Journal of PoUcy Reform, 1, 89-113 (1998). Chapter 6, 'IMF Programmes: Is There a Conditionality Laffer Curve?', World Economics, 2(2), 29-49, April-June (2001). Chapter 10, 'Catalysis or Direct Borrowing: The Role of the IMF in Mobilising Private Capital', The World Economy, 24, 81-98, January (2001). Chapter 11, 'Crisis Averter, Crisis Lender, Crisis Manager: The IMF in Search of a Systemic Role', The World Economy, 22(7), 955-75, September (1999). Chapter 12, 'The IMF and Developing Countries: A Review of the Evidence and Policy Options', International Organisation, 50(3), 477-512, Summer (1996). Chapter 14, 'The Political Economy of the SDR: the Rise and Fall of an International Reserve Asset', Global Governance, 4, 355-79 (1998). A second debt is to Dane Rowlands, with whom I co-authored Chapters 3, 8, 10 and 13, for allowing me to include these papers in this volume. Leaving to one side the intellectual debts I have that are too numerous to mention I should also acknowledge the debt I owe to my family, Heather, Alan, Anne, Simon and Tom. Apart from anything else my kids will only judge this book by seeing whether their names are mentioned. There you are, they are.

1

A suitable case for treatment? Understanding the ongoing debate about the IMF

Although the International Monetary Fund (IMF) has been a much-studied institution since its inception at the Bretton Woods conference in 1944, it is probably fair to say that it has never received such close examination and critical analysis as it has since the mid-1990s. This rush of interest started somewhat artificially with the 50th anniversary of Bretton Woods and the deliberations of the Bretton Woods Commission, but became more compelling as the wider debate about a new international financial architecture gained momentum in response to various economic and financial crises. Accusations that the Fund mishandled the crises in East Asia in 1997, and in Russia in 1998, combined with broader criticism of its systemic role in a world economy dominated by large and volatile private capital flows gave the debate a sharper and more immediate focus. Even large industrial economies now feared that they might not be exempt from the spillover effects of crises in emerging market economies and, in some quarters, the crises were seen as the 'Achilles heel' of global capitalism. In this context, attention has been paid to the part that the IMF might play in helping to avert such crises or to minimize their adverse global effects, should they occur. Moreover, the large loans that were associated with its involvement in crisis countries meant that the Fund itselfbecame short of liquidity. The decision as to whether or not to increase its capital base provided both a further opportunity, as well as the motivation, to reassess its role. Clearly the world economy at the end of the twentieth century was very different from the one that had existed in the middle of the century, when the Bretton Woods institutions were set up; it was natural therefore to consider whether these institutions had adapted appropriately to meet the contemporary challenges the world economy now appeared to face. Ultimately, did the world still need the IMF? If so, did it need it in its current or some modified form? Against this background there has been a spate of reports published seeking to assess the pros and cons of the Fund's operations and to plot out the direction that the reform should take. Indeed, when combined with the academic and popular literature on the subject there has been a veritable avalanche of material produced about the Fund. It is easy to get swept away in this, but in so doing to fail to isolate the principal issues that lie at the heart of this debate and also to

2 The ongoing debate about the IMP fail to distinguish areas where there is consensus from those where there is dissent. The purpose of this chapter is to provide a guide to help understand the ongoing debate about the future of the IME It aims to provide a brief historical background to explain how the Fund's activities have evolved over time, and to isolate the central questions to which this story of evolution gives rise, putting the current debate in context. The next section identifies the principal historical landmarks and traces out the evolution of the debate about the Fund. The following section summarizes the Fund's contemporary sphere of operations. The next section extracts the key issues that emerge from the foregoing discussion and that dominate the current debate. The final section offers a few concluding remarks about the political economy of reforming the IME

A brief historical background1 The Bretton Woods conference in 1944 represented a response to the international financial 'anarchy' that had existed in the 1930s as countries pursued beggar-my-neighbour policies in the form of competitive devaluations and protectionist commercial policy. Against a backdrop of global recession and shrinking world trade there was a belief that an international monetary order could be established that would facilitate global expansion and 'full' employment without having to revert to the rigidities imposed by the gold standard. Of the two principal plans discussed at Bretton Woods, the US plan presented by Harry Dexter White was generally favoured over the more ambitious Keynes Plan put forward by the United Kingdom, which involved a new international currency, bancor, and an International Clearing Union in which countries would hold bancor accounts which would then be used to settle international debts. The White Plan maintained that all that was needed was some form of relatively modest stabilization fund that would provide countries with short-to-medium term financial assistance to tide them over during periods of balance of payments deficit while they corrected their underlying domestic macroeconomic disequilibria. Although some acknowledgement was made of Keynes' concerns about the potentially asymmetrical distribution of the adjustment burden between deficit and surplus countries by including a 'scarce currency clause', which in principle enabled pressure to be exerted on countries which persistently ran surpluses, the Articles of Agreement emerging from Bretton Woods essentially created White's stabilization fund. Borrowing from the IMF did not initially involve 'conditionality'- this was introduced in the 1950s as the United States continued to fear that it might in effect be required to underwrite the Fund's lending operations. By the time the Bretton Woods system was up and running at the end of the 1950s, when the free convertibility of European currencies was introduced, the IMF had assumed the general role of overseeing, or even managing, the Bretton Woods system. This system involved countries pegging the par values of their currencies, and only altering them in the event of a 'fundamental disequilibrium', where domestic and external policy targets became incompatible at the pegged exchange rate.

The ongoing debate about the IMP 3 However, not all currencies were treated equally. The US dollar was made the pivot of the system, with its price being tied to gold. Although in principle involving an adjustable peg system for exchange rates, the reality was that pegs were not adjusted very frequently so that balance of payments correction, in fact, relied heavily on managing domestic aggregate demand. Within the context of the Bretton Woods system the IMF fulfilled four key functions. First, it was an adjustment institution offering advice and some degree of compulsion in terms of appropriate domestic macroeconomic policy. Second, it was a financing institution providing short-to-medium term financial assistance in support of approved policy, designed to eliminate balance of payments deficits. Third, it helped to co-ordinate macroeconomic policy internationally by encouraging countries to design policies in ways that supported pegged exchange rates, in effect supervising a rule-based system of macroeconomic co-ordination. Finally, it acted as a forum for the debate of international monetary reform. Although never completely free from contemporary criticism, during the golden age of the 1950s and 1960s criticisms of the Fund were muted. The Fund had been set up to run a system designed to encourage global economic growth, full employment and expanding world trade and on all counts the world economy seemed to be performing quite satisfactorily. 2 During the 1960s debates involving the Fund focused instead on, first, its potential contribution to creating additional global liquidity and, second, whether it needed to modify its lending operations to accommodate what developing countries presented as their particular and unique problems. 3 In terms of global liquidity the fear was that, within a system involving infrequent exchange rate changes and a relatively inefficient adjustment mechanism, inadequate liquidity would threaten future global economic growth, and that, with the rapidly growing Eurodollar market, a shortage of official reserves would expose the system to the threat of exchange rate crises associated with speculative attacks on currencies. This debate culminated at the end of the 1960s with the introduction of Special Drawing Rights (SDRs) -an international reserve asset that the IMF could itself create as it deemed necessary. 4 In terms of the special interests of developing countries, the Fund relaxed its initial approach, which was that its standby agreements adequately met the needs of all its member countries, and during the 1960s introduced first the Compensatory Financing Facility and then the Buffer Stock Financing Facility, each of which was, in its own way, designed to deal with the problems of export instability and adverse terms of trade movements. Unlike standby arrangements, drawings under these facilities involved only modest or 'low' conditionality, reflecting the view that export instability was largely exogenous and required liquidity rather than adjustment. However, these reforms failed to prevent the collapse of the Bretton Woods system during 1971-3. To what extent was the IMF to blame for the collapse? In a sense it cannot be exonerated from blame. After all, the Fund had been designed to operate the Bretton Woods system, and yet it clearly failed to prevent developments that led to the system's eventual breakdown.5 In essence,

4 The ongoing debate about the IMP it had not prevented currencies from becoming seriously misaligned. Whatever its share of responsibility, the legacy of the collapse of the Bretton Woods system was to call into question the future of the Fund. If it was there to supervise the Bretton Woods system, why was it needed once that system had collapsed? While this question was asked by several commentators, a more measured analysis of the need for international financial reform and of the IMF was proferred, at the time, by the so-called Committee of Twenty (C20). While still emphasizing the importance of stable exchange rates, the C20 recognized that greater exchange rate flexibility could be appropriate. It also envisaged a larger role for the SDR as a reserve asset and an extension of IMF lending into the medium to longer term in association with structural change. As the debate about international financial reform was going on, the world was in effect forced by expediency to adopt generalized flexible exchange rates in 1973 to try and correct persistent currency misalignment, and encountered an enhanced need for international financial intermediation in the context of the quadrupling in the price of oil in 1974, which created large surpluses in the oil exporting countries and large deficits among oil importers. 6 While the Fund responded by creating a temporary Oil Facility designed to recycle oil revenues from oil producers to countries with oil-related balance of payments deficits, the task of recycling petrodollars was largely performed by private international banks. In broader terms, during the 1970s the IMF became marginalized as an institution. Its adjustment role was reduced by the introduction of exchange rate flexibility. Its lending role was overshadowed by international financial intermediation by the private banks. Its macroeconomic co-ordination role seemed less relevant where it was believed that flexible exchange rates would insulate countries from external shocks. Even its role as a forum for international monetary reform was reduced by the trend towards regionalization in Europe and the existence of other fora with overlapping spheres of responsibility. 7 Moreover, with flexible exchange rates and large private capital flows, the issue of the global adequacy of official international reserves became, in many ways, irrelevant. Although the Fund continued to set itself an (over) ambitious systemic target in terms of establishing the SDR as the principal international reserve asset, the reality was that the system evolved more in the direction of multiple reserve currencies; even the Fund's own asset faded into obscurity. There were also changes in the global economic paradigm that had a bearing on the Fund. The Bretton Woods system had been devised by reasonably likeminded Keynesians and the IMF was designed to provide countries with financial assistance in order to help them avoid deflationary policies that would have adverse implications for economic growth and employment. During the 1970s the Keynesian consensus evaporated and by the beginning of the 1980s political leaders in a number of the Fund's major shareholders, who were anxious to reduce the role of government nationally, were at least ambivalent about the role of an intergovernmental organization such as the IMR If the Fund had been

The ongoing debate about the IMP 5 established at a time of Keynesian consensus it became marginalized over a period of time when Keynesianism was under attack. Attempts continued to address the special interests of developing countries which, during the 1970s, became the Fund's only clients, with the opening of a new medium-term lending window - the Extended Fund Facility (EFF) - the establishment of a Trust Fund, financed by IMF gold sales, to help provide concessionary assistance to poor countries through interest rate subsidies, and the liberalization of conditionality. In practice, the EFF was initially relatively little used and seemed to be unpopular both among borrowing countries and the IME Certainly it did not seem to meet the needs of the world's poorest countries. Even so, by the beginning of the 1980s the IMF was lending exclusively to low-income countries. The better-off developing countries seemed able to meet their need for external financing by borrowing from the private international banks. An apparently natural division of labour had emerged between the IMF and private capital markets as the Fund dealt with those countries bypassed by the private banks. Critical analysis of the IMF focused on its involvement in developing countries rather than its eroded systemic role and concentrated on the design and effectiveness of conditionality. 8 Here critics of the Fund argued that, in design, conditionality reflected monetarist modes of thought, focused excessively on the demand side of economies to the exclusion of the supply side, and was insufficiently flexible to respond to country-specific needs. Others complained that conditionality was based on old-fashioned macroeconomics, that it misinterpreted the appropriate role of the state in many developing countries by emphasizing its 'crowding out' effects rather than its 'crowding in' effects, and that it was subject to a fallacy of composition in as much as policies that might be effective when pursued in isolation by one country would be ineffective when pursued simultaneously by a group of countries. New structuralists argued that IMF conditionality based on exchange rate devaluation and monetary contraction was stagflationary. More generally, critics claimed that the empirical evidence suggested that conditionality had little if any significant impact on key macroeconomic variables. The debate over conditionality - for there were counterclaims to each of these accusations - became and remains an enduring feature of most discussions of the IME However, the 1980s saw a number of important changes for the IME In terms of both its clientele and its role there was a radical change in 1982. Faced with a deteriorating global economic environment, which involved sharply higher world interest rates and faltering export performance alongside aspects of domestic economic mismanagement, many highly indebted developing countries largely in Latin America - that had formerly relied on private capital began to encounter severe debt problems and started to tum to the Fund. To the extent that these threatened the solvency of United States and other banks that had made loans to these countries, the debt crisis was perceived by some - although by no means all- observers as threatening to destabilize the international financial system. The Fund's involvement took on a quasi-systemic dimension.

6 The ongoing debate about the IMP While it was applauded in some quarters, others criticized the Fund for what they saw as its mishandling of the Third World debt crisis. The criticisms comprised a number of elements. First, critics argued that the Fund had become more overtly political. According to this view, lending decisions were forced through by creditor countries, in particular the United States, with the clear objective of preventing specific debtor countries from defaulting and thereby damaging the interests of the large commercial banks that still had significant loans outstanding to them. The claim that this was justified in order to ensure systemic stability became increasingly illegitimate as the banks themselves adjusted to the crisis by provisioning, by reducing their developing country exposures through the secondary market, and by expanding other lines of business. Even the Fund's own staff began to complain that it was being politically manoeuvred. 9 Without having to subscribe to this view, aggregate evidence drawn from the mid-1980s suggests that positive net transfers from the Fund to indebted developing countries financed negative transfers with the banks. It requires only a short step to claim that the Fund was bailing out the banks. A second criticism was that the Fund treated the debt problem as if it was one of illiquidity and inadequate cash flow rather than a more deep-rooted and longer term one of structural adjustment. The Fund opted to support new financing which assisted countries in meeting their outstanding debt servicing obligations, but did less to restore medium-term viability to their balance of payments. The adjustment programmes favoured by the Fund called for conventional demandside contraction. Third, critics complained that the Fund failed to devise incentives that positively encouraged adjustment. The additional foreign exchange that adjustment created was largely used to repay commercial creditors abroad. The domestic rate of return was in many cases low. Critics argued that it was this low domestic return which led to problems of non-compliance. Governments found it difficult to muster domestic political support for programmes which seemed to imply heavy domestic costs but no readily discernible benefits. Such critics went on to argue that what was needed was debt relief in order to reduce the domestic adjustment burden and raise the incentive to adjust. Instead, through its policy of'assured financing', the Fund discouraged banks from granting relief and in this way postponed appropriate systemic reform. 10 It was only following the introduction of the Brady Plan, towards the end of the 1980s, that the Latin American debt crisis appeared to ease. Before that the Fund had largely failed to achieve its target of normalizing creditor-debtor relations and restoring country access to sustainable and spontaneous lending. While highly indebted developing countries borrowed heavily from the IMF in the 1980s, most Fund programmes continued to be with the low-income countries of Africa and Asia. Here, more than elsewhere, some critics of the IMF argued that conventional demand-side conditionality was particularly inappropriate. They argued that low foreign trade elasticities limited the effectiveness of devaluation, and that ill-developed financial markets constrained the impact of rising interest rates. Moreover, since fiscal deficits were frequently monetized,

The ongoing debate about the IMF 7 the strengthening of the tax base was often presented by critics as a higher priority than curbs on government expenditure.U Some NGOs went as far as suggesting that the Fund was doing more harm than good in low-income countries because its programmes had a negative effect on economic growth, income distribution and poverty. Capturing the essence of this critique, UNICEF claimed that what was needed was 'adjustment with a human face'. Early in the 1980s, and partially in response to a second big increase in oil prices at the end of the 1970s and a change in the dominant economic paradigm, the Fund had retreated from the EFF and had, in effect, phased out low conditionality lending through reforms to the CFF. The emphasis had returned to conventional macroeconomic stabilization based on controlling aggregate demand through fiscal and monetary policy. Now alongside the appointment of a new Managing Director - Michel Camdessus - in the late 1980s, the Fund offered an institutional response to concerns that it was under-emphasizing the supply side and the social consequences of adjustment. Its rhetoric began to refer more to encouraging economic growth as a 'primary' objective of adjustment programmes and protecting 'vulnerable groups' from the costs of adjustment. Perhaps most notably, the Fund introduced in 1987 the Enhanced Structural Adjustment Facility (ESAF). This was targeted at low-income countries and emphasized a broader range of conditionality which, while still including conventional demand-side measures, also included specific supply-side and microeconomic measures. 13 The move into structural adjustment that the ESAF signalled began both to increase the diversity and sharpen the intensity of the debate surrounding the Fund. It now became more difficult to sustain the demarcation between the Fund and its sister institution the World Bank; a demarcation further eroded by the Bank's move into policy-based structural adjustment starting at the beginning of the 1980s.l 4 As originally envisaged, the division of labour between the Fund and the Bank had been relatively straightforward. The Fund's orientation was towards the short run, the balance of payments, the demand side, the monetary sector and programme support. The Bank's was towards the long run, economic development, the supply side, the real sector and project support. The differences between the two institutions were well encapsulated in Keynes' observation that the Board of the Fund should comprise 'cautious bankers' whereas that of the Bank should comprise 'imaginative expansionists'. In 1966 an internal memorandum had clarified the division of labour by assigning 'primary responsibilities' to each agency. The Fund had jurisdiction 'for exchange rates and restrictive systems, for adjustment of temporary balance of payments disequilibria and for evaluating and assisting members to work out stabilization programs as a sound basis for economic advice'. The Bank's primary responsibility, in contrast, was 'for the composition and appropriateness of development programs and project evaluation, including development priorities'. Less than a decade later the Fund had begun to accept that payments deficits could be of a structural nature, for which correction required longer term

8

The ongoing debate about the IMF

financial support. The EFF had been introduced to fill what was perceived as a gap in the range of the Fund's lending facilities and clouded the distinction between the Fund and the Bank. But it was in the 1980s and in the context of structural adjustment that the overlap became most pronounced. With the introduction of the ESAF (and the SAF shortly before it), the argument that the IMF was moving too far in the direction of becoming a development agency rather than being solely an international financial agency became more frequently heard. While those from the right of the political spectrum claimed that the Fund was evolving in an inappropriate way by offering too much aid to countries that were pursuing ill-designed policies, often under unacceptable political regimes and with little chance of success, those from the left including, as noted earlier, the NGOs - maintained that, to the extent that the Fund had taken on a development role, it performed this role badly by hurting economic growth and failing to alleviate poverty. Even those who had criticized IMF conditionality for being too narrow and for being overly reliant on seeking to control monetary aggregates, now began to draw attention to the dangers of excessive conditionality. Certainly, in spite of its highly concessionary rate of interest, many developing countries remained reluctant to use the ESAF and the track record of those that did was not such as to convince its critics that the facility represented a worthwhile extension of the Fund's operations. During the 1990s a series of reviews of ESAF - including an external assessment - struggled to strengthen it. 15 Its metamorphosis into a lending window which more directly focused on economic growth and poverty reduction culminated with its renaming as the Poverty Reduction and Growth Facility in 1999. To some critics this was a step in the right direction, but to many others it was simply not a legitimate part of the Fund's mandate to become directly involved in issues of poverty and economic growth - areas in which in any case the Fund was supposedly at a comparative disadvantage. But there were other developments during the 1990s that further and perhaps more particularly enlivened the debate about the Fund. Just as the Third World debt crisis during the 1980s had thrust many Latin American economies into the Fund, the fall of Communism at the beginning of the 1990s and the embarkation of formerly socialist economies into a process of transition towards more market-based systems meant that the Fund was confronted with the challenge of assisting many of them. Challenge it was, since the institutional structures necessary to support a market economy were absent. Yet again the debate raged over how well or badly the Fund responded to the challenge. Some commentators suggested that it responded completely inadequately by providing insufficient financial support and offering inappropriate policy advice. Others argued that the Fund simply poured good money after bad without exerting enough pressure on countries - most strategically Russia - to carry out the necessary policy changes. Stories that Fund resources were simply passing through Russia briefly on their way to Swiss bank accounts fanned the increasingly popular view that the IMF was beginning to make too many mistakes, often allowing political objectives of its major shareholders to interfere

The ongoing debate about the IMP 9 with its economic judgement. 16 This perception was hardly helped by the Mexican peso crisis in 1994, nor, more significantly perhaps, by the East Asian financial crisis in 1997-8 and the additional crises that occurred in Russia in 1998 and Brazil in 1999. While in the mid-1990s the Volcker Commission had been - perhaps to a fault - measured and temperate in its analysis and the advice it gave, towards the end of the 1990s commentary on the Fund became much more outspoken and directP Illustrative of this was a lively- if not downright heated - interchange between Joseph Stiglitz and Rudiger Dornbusch in the pages of the New Republic. Stiglitz accused the Fund of seriously misdiagnosing the nature of the East Asian crisis and of continuing to design conditionality as if one size fitted all. He argued that Fund conditionality had a perverse effect and came close to suggesting that the Fund was incompetent and closed to criticism, even when it came from the World Bank. Dornbusch strongly defended the Fund, arguing that the sharp recovery in Korea following the crisis vindicated its advice. But in terms of the published contributions - as is usually the case with contentious issues - criticisms tended to outweigh defences. Among the issues raised were the following. Why had the Fund not seen the East Asian crisis coming? Was it monitoring the wrong variables? What was it missing? Had it pushed financial liberalization too hard? Had it failed to recognize the deficiencies of domestic financial supervision and regulation? Was it a mistake to advocate capital account liberalization alongside domestic financial liberalization? Was the sequencing of liberalization wrong? Had its willingness to provide large amounts of finance to Mexico in 1994 led private capital markets to over-lend elsewhere, believing that they would be effectively baled out by the Fund should things go wrong; did IMF lending therefore involve a moral hazard problem? Why had the Fund recommended fiscal retrenchment in East Asian economies when fiscal deficits were not large, and when private sector saving was high? Had the devaluations favoured by the Fund as well as the breadth and depth of its conditionality triggered further panic and capital outflows, consequently leading to exchange rate overshooting and further economic recession? Was the Fund sending out the wrong signals to private capital markets? Had sharp rises in interest rates exacerbated problems for the financial and corporate sectors? Had the Fund done enough to try and bail in private capital? In designing conditionality, had the Fund stipulated policies beyond those necessary to stabilize the affected economies and well beyond those that it had a 'moral right' to require? Was it encroaching too far into microeconomic issues, into economic structure, into the reform of the financial institutions as well as into aspects of political behaviour? Was this motivated by a desire to serve the interests of richer indusrrialized economies and in particular the United States? Was it acting more as an agent for private capital markets than its client countries? Was the desire to avoid conditionality forcing countries to build up excessive holdings of international reserves? The list of questions goes on. But fundamentally the Fund's role as both an adjustment and a financing institution came under sharp attack. 18

10

The ongoing debate about the IMF

Critics suggested that in spite of its surveillance of economies the Fund had not done enough to avert crises, that its role as a crisis lender had been in some ways counter-productive, and that it had met with only mixed success as a crisis manager. In these circumstances, and against a backdrop that called for a new international financial architecture, it is easy to see why a full reassessment of the Fund's role in the global economy seemed relevant. The Fund's response to the crises of the 1990s, as it had often been beforehand, was to erect new lending windows. As noted above with transition economies in mind, it had established the Systemic Transformation Facility. Now with quotaconstrained demands on its resources, it introduced the Supplemental Reserve Facility, which allowed its lending to individual countries to exceed quota limits, and introduced New Arrangements to Borrow to augment its own resources. In an attempt to provide countries with precautionary finance to help in the contingency of a contagion effect from a crisis, it established the Contingent Credit Line. But were these adequate or even appropriate responses? An important question became whether the Fund's operations had evolved in an incremental and piecemeal way that had resulted in it trying to play a role that was either inappropriate or beyond its means? Had there been excessive 'mission creep'? In the context of this surge of criticism and with the Fund looking for additional resources, a clustering of reports coming from the Council on Foreign Relations, the International Financial Institution Advisory Commission, the Group of 24 and UNCTAD, the Overseas Development Council and the Centre for Economic Policy Research were published in 1999 and 2000.

A brief statistical background The purpose of this section is to provide an indication of the Fund's operations during the 1990s. It therefore sets out the empirical context in which the current debate about reforming the Fund is being conducted. Table 1.1 shows how Fund lending changed throughout the 1990s. Both in terms of net credit - allowing for repayments - and gross disbursements, it is apparent that there are peaks and troughs in Fund lending. These depend essentially on whether large economies such as Mexico or Korea or Russia are drawing resources from the Fund. Thus, the sharp rise in IMF lending in 1995 was associated with the Mexican crisis and its spillover effects, while the subsequent rise in 1997, and more so in 1998, reflected the East Asian financial crisis. Although the IMF is heavily involved in sub-Saharan Africa, drawings from these economies generally represent a very modest fraction of total Fund lending. Indeed, as far as low-income countries, in general, are concerned, in 1998 drawings under the ESAF accounted for only about 10 per cent of total disbursements and just under 2 per cent of net credit provided by the Fund. From the figures provided in Table 1.1 it is easy to see why the current debate about the Fund has concentrated so heavily on its role in emerging economies in crisis. Quantitatively it is these countries that have dominated Fund lending.

1.9 0.2

0.2

1.1

0.733

0.644

31.217 0.217 5.041

-

-

1.3 0.1 0.1 -1.6 1.6 0.5 0.5 1.0

-

-0.2

-

34.503 0.157 5.285

3.374 -0.060 0.253

3.7 2.0 2.0 1.5 0.2

-0.9

0.2 0.7 0.6 0.1

-0.1

-0.4

37.276 0.153 6.634

0.594 -0.014 0.998

2.4 0.5 0.2 1.5 0.3

0.9 0.5 -0.8 0.4 0.4 -1.3

-0.8

-

-

-

-

0.3 -

1994

1993

1992

53.275 0.141 8.342

15.633 -0.015 1.619

0.8 0.6 -1.5 -0.3 0.4 12.9 4.7 -1.3 -2.7 5.5 0.6

12.6

-

-0.1

1995

0.325

0.291

51.824 0.137 8.392

-

-0.8 3.2 0.5

-

0.6 0.1 -1.7 -0.4 0.1 -2.0 3.7

-2.9

-

-0.1

1996

62.703 0.121 8.049

14.355 -0.007 0.179

-0.5 -0.5 5.0 5.7 0.2 -4.0 2.5 0.7 0.4 1.5 0.2

0.8

11.3 11.3

1997

84.961 0.126 8.788

18.811 -0.001 0.374

0.3

5.3

-0.4 -0.3 6.6 7.0 -0.1 2.5 5.5 -0.1 -0.4

8.5

5.2 5.2

1998

69.913 0.122 8.760

-12.856 -0.001 0.193

-3.6

2.1 0.6 -0.9 -3.6

1.7

-0.2 -0.1

1.3

-10.3 -10.3

1999

Notes a Includes net disbursements from programmes under the General Resources Account, Trust Fund and Poverty Reduction and Growth Facility (formerly ESAF). The data are on a transactions basis, with conversion to US dollar values at annual average exchange rates. b Converted to US dollar values at end-of-period exchange rates.

Source: IMF Annual Reports (various).

Disbursements at year-end under:b General resources Account Trust Fund PROF

31.821 0.226 4.499

2.520 -0.069 1.070

Total Net credit provided under: General resources Account Trust Fund PROF

Memorandum

-

2.4 2.4 2.4

-1.0

-

-

-

Countries in transition Central and Eastern Europe Excluding Belarus and Ukraine Russia Transcaucasus and Central Asia

Regional groups Africa Sub-Sahara Asia Excluding China and India Middle East and Europe Western hemisphere

Developing countries

Advanced economies Newly industrialized Asian economies

1991

Table 1.1 Net credit and loans from IMF (US$ billions)•

12

The ongoing debate about the IMF

However, just to focus on its total amount of lending is to misrepresent the Fund's operations. The data in Table 1.2 provide a slightly different perspective. This table focuses on the number of programmes as well as the amount of lending. The data in it provide a snapshot of the Fund's operations towards the end of 1998, which we have already identified as a year when the Fund was lending relatively heavily. It was against this background that the reports mentioned in the previous section were published. A number of features of Table 1.2 are notable. First, there were outstanding arrangements with 62 countries, representing about one-third of the Fund's membership. Second, the arrangements covered both developing countries and countries in transition, ranging from the better-off among them to the very poorest countries in the world. Taking the UNDP's Human Development Index (HOI), the highest ranked recipient of IMF finance was Korea (32), with the lowest ranked being Rwanda (174). Standby and extended arrangements were more common among the better-off developing countries and CITs, while ESAF loans were focused on the poorer countries. Of the 62 arrangements recorded in Table 1.2, 17 were with countries ranked in the top 100, according to the HOI, while the remaining 45 were with countries in the bottom 75. Third, arrangements existed across most regions of the world including Africa, Asia, Latin America and Central and Eastern Europe. Fourth, as suggested by Table 1.1, in quantitative terms, total Fund lending commitments (at nearly SDR49 billion) were concentrated in a relatively small number of countries; arrangements in Korea (SDR15.5 billion), Russia (SDR13.2 billion) and Indonesia (SDR4. 7 billion), for example, accounted for about 68 per cent of the total amount of lending approved; on the other hand, lending to 36 low-income countries under the ESAF amounted to only about the same as the single loan to Indonesia.

The debate about the Fund: the underlying issues The foregoing discussion covering both its historical evolution and the nature of its current operations allows us to place in context the current debate about the IMF. The debate touches on a range of issues and in this section we identify the most important of them and discuss each briefly.

Does the world still need the IMF? With flexible exchange rates and large private capital markets from which countries can borrow to finance balance of payments deficits, why is the Fund needed? Would it not be simpler to scrap it altogether? The counter-argument is that for various, but essentially conventional reasons, private capital markets may 'fail'. There may, for example, be shortcomings in dealing with systemic risk, externalities associated with international lending, problems associated with asymmetrical information and co-ordination and collective action problems. Should

Bosnia and Herzegovina Cape Verde Djibouti El Salvador Estonia Korea 1 Latvia Philippines Thailand Uruguay Zimbabwe EFF arrangements Argentina Aberbaijan Bulgaria Croatia, Republic of Gabon Indonesia Jordan Kazakhstan Moldova Pakistan Panama Peru

Standby arrangements

Member

28 May 1999 19 April 1999 31 March 1999 22 February 2000 16 March 1999 3 December 2000 9 April1999 31 March 2000 19 June 2000 19 March 1999 30 June 1999 3 February 2001 19 December 1999 24 September 2001 11 March 2000 7 March 1999 5 November 2000 8 February 1999 16 July 1999 19 May 1999 19 October 2000 9 December 2000 31 March 1999

4 February 1998 20 December 1996 25 September 1998 12 March 1997 8 November 1995 25 August 1998 9 February 1996 17 July 1996 20 May 1996 20 October 1997 10 December 1997 1 July 1996

Expiration date

29 May 1998 20 February 1998 15 April 1996 23 September 1998 17 December 1997 4 December 1997 10 October 1997 1 April1998 20 August 1997 20 June 1997 1 June 1998

Date of arrangement

Table 1.2 Standby, EFF and ESAF arrangements (as of 30 November 1998)

36.36 2.10 1.95 37.68 16.10 2,175.00 33.00 823.42 700.00 125.00 91.45 15,536.23 2,080.00 17.56 523.02 324.38 49.63 2,566.70 35.52 309.40 97.50 398.06 80.00 139.70

353.16 110.30 4,669.10 238.04 309.40 135.00 454.92 120.00 300.20

4,042.06

Undrawn balance

60.60 2.10 8.25 37.68 16.10 15,500.00 33.00 1,020.79 2,900.00 125.00 130.65 24,414.26 2,080.00 58.50 627.62

19,834.17

Amount approved

3,960 3,641 3,740 4,187 3,284 1,576 2,154 6,104 3,645

8,937 1,670 4,533

1,862 1,270 2,417 4,294 10,656 3,332 2,681 7,104 6,752 2,196

Real GDP per capita (ppp $) 1994

(Continued)

77 120 99 84 93 110 139 45 89

36 106 69

123 162 112 71 32 92 98 59 37 129

Human development index rank

Russian Federation• Ukraine Yemen ESAF arrangements Albania Armenia Azerbaijan Benin Bolivia Burkina Faso Cameroon Central African Republic Chad Congo, Republic of Cote d'Ivoire Ethiopia The Gambia Georgia Ghana Guinea

Member

Table 1.2 (Continued)

25 March 2000 3 September 2001 28 October 2000 12 May 2001 13 February 1999 19 December 1999 27 August 1999 17 September 2002 13 June 1999 19 August 2000 19 July 2001 28 April 1999 27 June 1999 16 March 2001 10 October 1999 28 June 2001 27 February 1999 29 June 1999 12 January 2000

13 May 1998 14 February 1996 20 December 1996 28 August 1996 18 September 1998 14 June 1996 20 August 1997 20 July 1998 1 September 1995 28 June 1996 17 March 1998 11 October 1996 29 June 1998 28 February 1996 30 June 1995 13 January 1997

Expiration date

26 March 1996 4 September 1998 29 October 1997

Date of arrangement

151 164 1,130 700 2,410 1,668 427 939 1,585 1,960 1,103

41.20 8.26 55.58 202.47 58.98 17.18 27.75 68.50 23.60

49.44 49.56 69.48 285.84 88.47 20.61 166.50 164.40 70.80

130 145 170 165 105 132 167

102 103 106 146 113 172 133 2,788 1,737 1,670 1,696 2,598 796 2,120

Human development index rank 67 95 148

Real GDP per capita (ppp $) 1994

4,828 2,718 805

Undrawn balance

7,426.86 1,400.00 87.90 2,301.00 29.42 33.75 23.40 18.12 34.13 6.63 81.06

13,206.57 1,645.55 105.00 4,580.98 35.30 101.25 93.60 27.10 100.96 39.78 162.12

Amount approved

107 80 152 161 171 101 166 127 173 139 174 160 115 149 159 148 143

1,930 3,965 694 694 543 3,766 986 1,580 787 2,154 352 1,596 1,117 656 1,370 805 962

53.75 27.28 54.24 15.27 10.34 27.83 12.60 84.09 9.66 454.92 59.50 89.18 78.00 38.76 43.52 176.75 40.00 21879.29

64.50

25 June 2001 10 April 2000 26 November 1999 31 December 1998 5 August 1999 29 July 2000 24 August 1999 17 March 2001 1 September 1999 19 October 2000 23 June 2001 19 April 2001 23 June 2001 7 November 1999 9 November 2000 28 October 2000 5 December 1998

26 June 1998 11 April1997 27 November 1996 18 October 1995 10 April 1996 30 July 1997 21 June 1996 18 March 1998 12 June 1996 20 October 1997 24 June 1998 20 April 1998 24 June 1998 8 November 1996 10 November 1997 29 October 1997 6 December 1995

Note a Includes amounts under a Supplemental Reserve Facility. EFF = Extended Fund Facility. ESAF = Enhanced Structural Adjustment Facility. Figures may not add up to totals because of rounding.

54.56 81.36 45.81 62.01 33.39 75.60 100.91 57.96 682.83 71.40 107.Dl 96.00 161.59 100.43 264.75 701.68 48829.40

104 156 134

2,729 896 1,404

44.80 75.88 124.63

53.76 91.05 149.55

14 July 2001 17 October 1999 25 April 1999

15 July 1998 18 October 1996 26 April 1996

Source: IMF Treasurer's Department, UNDP, Human Development Report.

Guyana Haiti Kenya Kyrgyz Republic Macedonia, FYR Madagascar Malawi Mali Mongolia Mozambique Nicaragua Niger Pakistan Rwanda Senegal Tajikistan Tanzania Uganda Yemen Zambia Total

16

The ongoing debate about the IMP

the Fund not be playing a role in offsetting these failures and in supplying what are in effect international public goods such as information, conditionality and crisis lending and management? Would private markets left to themselves allocate financing in an efficient and equitable way? Moreover, exclusive reliance on private capital markets would lead to many countries being unable to raise balance of payments financing. Would this be acceptable to the international community? If the Fund did not exist, would we need to establish it? But then again, just as markets may involve elements of market failure, so too the Fund may be subject to bureaucratic failures; what of the public choice approach to the Fund's operations? Fund lending might enable countries to avoid making the policy changes that would in the long run assist their market access. While the currency crises of the 1990s have been taken by some as a clear indicator of the need for the IMF in order to help the world contend with such crises, others suggest that it has been the very presence of the Fund that has contributed to causing the crises in the first place. Just as markets may fail, domestic governments may also fail in the sense of pursuing inappropriate policies that may in tum impose external costs on other countries. It was the pursuit of beggar-thy-neighbour policies in the 1930s that in part justified setting up the Fund. In principle therefore the involvement of the IMF may lead to superior policies. Moreover, signing an agreement with the Fund may be seen as locking governments into these policies in a way that deals with the time consistency problem- the incentive for governments to renege on policy promises if their announcement has positive effects while their implementation carries a political cost. However, critics argue that there is little reason to believe that the IMF is in a position to design appropriate adjustment policy and that the poor completion record of IMP programmes shows that governments are not in practice locked in. Although those who advocate closure are in the minority, the majority view is that, while the world continues to need the IMF, it needs it in a modified form. 19

Should the Fund be an adjustment or a lending institution? Balance of payments deficits can be handled using different intensities of adjustment and financing. What is the optimum mix, and does the Fund facilitate it? If not, does it provide too much or too little financing? A public choice approach might suggest that the Fund should not be lending at all, since this creates perverse incentives. This approach views IMF lending as encouraging countries to pursue policies which increase their access to concessionary financing from the Fund. On the other hand, would the Fund's adjustment programmes be credible with private capital markets if the Fund did not itself provide financial support? Does it have to put its money where its mouth is? The grounds for believing that Fund conditionality alone will catalyse others to lend is weak. But while Fund lending may, for these reasons, need to be positive, how large should it be? If this was an easy question to answer there would be less disagreement each time a quota increase is discussed. 20

The ongoing debate about the IMF

17

What should be the Fund's relationship with private capital markets: how important is moral hazard? As noted earlier, one view is that Fund lending encourages private lenders to take excessive risks by providing what is perceived as a safety net. In the midst of a crisis the Fund bails out private lenders, which then does nothing to alter this perception. To the extent that currency crises reflect excessive capital inflows before crises as much as they do excessive contemporary outflows during them, there is a potential moral hazard problem. However, it is difficult to quantify the importance of this problem and consequently views differ sharply. A counter-view is that the prospect of Fund involvement in no way creates a risk-free environment for all private lenders and that, as part of a broad explanation of private capital flows, the relevance of the Fund is, at most, modest. 21 While the moral hazard argument suggests that the Fund will be lending at precisely the times that private capital is moving out, a related issue is whether the Fund can manage other capital flows more effectively and make them a complement to its own lending. In the aftermath of the Third World debt crisis the Fund certainly exerted significant pressure on commercial banks to continue lending at the same time as it was involved. While circumstances have moved on since then, not least because creditors are more diverse, the question remains of whether the Fund can still beneficially influence private capital flows. Part of the answer could involve easing up on the pursuit of capital account liberalization and supporting countries which wish to tax excessive capital inflows and thereby seek to discourage excessive capital inflows. Other ideas hinge on improving the effectiveness of conditionality, improving surveillance and providing and interpreting additional information- either implicitly or more explicitly by giving countries a credit rating - and by permitting the suspension of debt repayments and supporting the reform of loan contracts to allow for 'collective action clauses' which might then both discourage excessive lending before crises and facilitate crisis management by 'bailing in' private capital as opposed to bailing it out. At a general level many of these ideas seem uncontroversial. The devil tends to be in the detail. How can conditionality be made more effective? What extra information should the Fund collect and publish? Given that the Fund's predictions have not always been accurate in the past, would there not be a danger that by publishing more outspoken judgements on countries the Fund would spook capital markets and create the crises that it is seeking to avoid? Certainly we have no idea of the response elasticities and may even be unsure as to their sign.

Should the Fund be a crisis-averter, crisis-lender or crisis-manager? The Fund versus other financial agencies Given the incidence of financial crises in the 1990s, much of the current debate surrounding the IMF has focused on its role in the context of crises. Systemically the issue breaks down into how best the Fund can try to prevent crises and deal

18 The ongoing debate about the IMP with those that it fails to prevent. Of course, behind this apparently straightforward distinction lies the whole question of what causes crises in the first place. Since there is theoretical and empirical disagreement on this question, it is unsurprising to find disagreement also about how best the Fund should seek to avert crises. Will greater surveillance of economic policy and performance do the trick? But what elements of policy and performance are important? Do countries perhaps need to be encouraged to tum to the Fund at an earlier stage before their problems reach crisis proportions? But then how can they be encouraged to do this, and will the Fund's principal shareholders be happy to see the Fund lending to countries that may still have access to private capital? Will a pre-commitment to provide financial support in the event of a crisis create greater confidence in private markets and make crises less likely or will the negotiations undermine and provoke crises? Some commentators believe that the Fund should evolve into a more fully fledged international lender of last resort (ILLR) providing the kind of confidence to international financial markets that national central banks provide for domestic financial markets. Others argue that this analogy is misplaced since the Fund cannot create money in the same way as central banks can. Without the necessary resources, the Fund's commitment to such a role would not carry credibility, and it may be unrealistic to assume that the Fund will be given them even if we knew what resources would be necessary. To the extent that the world economy requires some type of ILLR facility to provide countries with short-term emergency lending on a sufficient scale to calm markets in the event of a crisis, the question then becomes whether there are other institutions that already exist, or may still need to be created, that could undertake the task better than the Fund. Some commentators favour the BIS in the role. Others argue that, since the contagion effects of crises have tended to be regional, there may be a role for regional financial institutions to play. Perhaps the Fund should instead focus on co-ordinating other creditors, attempting to exert an influence on others to lend rather than lending itself. But again how can it exercise this influence ?22

Should the Fund be involved in long·tenn lending to developing countries? Is there too much overlap with the World Bank? When critics talk about 'mission creep', it is frequently the Fund's operations in low-income countries to which they are referring. In short, has the Fund moved too far in the direction of becoming a development agency? Again, in order to constitute a part of the ongoing 'debate' about the Fund there must be arguments on both sides. One point is that it has become much more difficult, in the context of structural adjustment, to draw a clear distinction between economic development and the balance of payments. Institutional ambiguities may simply reflect this. But even so, much concern has been expressed about the Fund becoming directly involved with issues of poverty, economic growth and debt relief. This concern has found expression in proposals that the Fund should

The ongoing debate about the IMF 19 either discontinue the Poverty Reduction and Growth Facility (PROF) altogether or that the facility should be moved to the World Bank where, it is suggested, the comparative advantage in issues surrounding poverty and economic growth lies. Two fundamental issues are therefore at the heart of the debate about this question. The first is how best to assist low-income countries. What policies do they need, how much economic and structural adjustment should they undertake and how much external financial support is required to support these policies? The second is what are the best institutional arrangements for providing this assistance? While there is some appeal in the argument that issues of poverty and growth are development issues and responsibility for them should therefore be lodged with the World Bank, rather than the IMF, there are complicating factors. How would macroeconomic stabilization be integrated into programmes led by the Bank? The Fund would still need to make an input. Would a shift in the responsibility for the PROF have an effect on the willingness of private capital markets, or, more significantly, bilateral aid donors to lend? What would be the effects on the HIPC initiative? Moreover, the World Bank's record on structural adjustment has not been clearly superior to that of the Fund and there must therefore be some doubts about whether shifting the PROF to the Bank would in itself make a significant difference for low-income countries, which might, in any case, not be completely enthusiastic about allocating a monopoly role in structural adjustment to the Bank any more than they would to the Fund.B

Do Fund-supported programmes work? What future is there for IMP conditionality? There would be less debate about the future of the Fund and its role in developing countries and emerging economies if its programmes had been perceived as successful in the past. This is not the general perception, hence the debate. The difficulty is that there is enough ambiguity in the empirical evidence to allow the Fund to claim that 'on balance' its programmes 'work' and for others to claim that they do not. Evaluating the effects of Fund programmes is an ongoing and complex exercise and there still remain fundamental questions to answer. For example, to what extent does success depend on the extent to which agreed programmes are completed? At present, the empirical answer to this question is unclear, as therefore is also the extent to which the effects of Fund programmes depend on the degree to which they are implemented. However, to criticize IMF conditionality is not necessarily to advocate that it should be abandoned. A more nuanced response is to seek to reform it in order to make it more effective. But how can this be achieved? Should conditionality be narrow or wide-ranging? How much emphasis should be placed on conventional demand-side policies as opposed to structural adjustment? Has conditionality become excessive, with counter-productive implications? Should more emphasis be placed on preconditions rather than conditions that have to be met during the course of a programme, and, if so, what should these preconditions be? Should the Fund seek to restore the distinction between low and high

20

The ongoing debate about the IMF

conditionality lending, allowing countries to gain access to financial assistance with few conditions in circumstances where it can be established that balance of payments problems have been caused by external shocks, where there is a reasonable presumption that the shock will be self-correcting or where appropriate policies will be pursued without direct Fund involvement, and where the need is for additional liquidity from the IMF rather than adjustment advice? Even when high conditionality is appropriate, there is a strong and growing consensus that programmes work best when the underlying policies are 'owned' by the relevant government. How can the Fund go about establishing a stronger sense of ownership? It is interesting that most moves in this direction have occurred in the context of the ESAF, and now the PROF, which is the facility that many critics of the Fund are most anxious to see discontinued. Are they objecting to the trend in conditionality towards greater ownership or to the inappropriateness of the client base of the PROF in terms of the Fund's underlying rationale ?24

Should the Fund's organizational structure be reformed? Transparency and governance As noted earlier, the IMF has typically responded to the deficiencies in its provision by introducing new lending windows, of which there has been a rapid proliferation. A first step in structurally adjusting the Fund's operations might be to rationalize these. One possibility would be to retain a low conditionality facility and a high conditionality facility, with the degree of concessionality depending on the per capita GDP of the borrowing country. But other types of rationalization could be feasible building on the notion that if the Fund is to be remodelled it should have fewer lending windows. Other possibilities would be to give countries some choice about the combination of conditionality and concessionality, allowing them to opt for cheaper/more conditional support or more expensive/less conditional support. Current conditionality could also perhaps be modified to reflect past performance in order to provide a stronger incentive to avoid economic mismanagement and to comply with conditionality. The ease with which new programmes can apparently be arranged may create another moral hazard. Other organizational issues relate to the transparency of the Fund's operations, its governance and exposure to political manipulation. An opaque organizational and decision-making structure may make it unclear as to where the responsibility for mistakes lies. This may also make it more likely that mistakes will be made and less likely that they will be corrected in the future. Greater transparency and exposure to outside judgement could offset any tendency towards inefficiency. Commentators have certainly argued that the Fund has been insufficiently willing to evaluate its decisions and operations and has been resistant to criticism. For the Fund to be an effective financial organization it needs to be flexible and responsive. It is therefore important to avoid an unwieldy organizational structure where decisions are slow to be made; this implies a significant degree of

The ongoing debate about the IMF 21 delegation of decision making. On the other hand, it is also important that all members of the IMF feel a degree of institutional ownership, otherwise the Fund's decisions will carry less credibility and its policies will be less effective because they lack the support of those who believe that their views have been under-represented. There is also the question of the extent to which the Fund's staff or management should be allowed to make decisions independent of political interference. If there are strong arguments for granting central banks independence at least in terms of deciding how to achieve inflationary goals, are there equally strong argument for granting the Fund greater autonomy? Or is the analogy between national central banks and the IMF far from perfect? For starters the Fund faces a less well defined set of objectives and does not exert direct control over many crucial aspects of policy. 25

How should the Fund be financed? The Fund currently relies on periodic increases in quotas alongside ad hoc arrangements such as the General Arrangement to Borrow (GAB) and New Arrangement to Borrow (NAB) in order to increase its lending capacity. Discrete quota increases are, in many ways, unsatisfactory. Although, from a political point of view, they may keep the Fund on a short leash, they do little to ensure that the Fund's lending capacity adjusts smoothly to the claims put upon it. The quota system, according to some observers, encourages the Fund to engage in 'hurry-up lending' shortly before a quota review, designed to justify an increase, and certainly it seems to generate significant political controversy. Would it be better to have quotas linked automatically to some set of indicators of the need for Fund resources? But then what should these indicators be? The need for Fund resources will, of course, depend on the role that the Fund is seeking to perform. If it became an adjustment-only institution, debates about the adequacy of its lending resources would become irrelevant. Other ideas are that the Fund should move away from relying on quota-based subscriptions to provide its resources and instead put greater emphasis on SDR allocations, gold sales and even the revenue from a Tobin tax on international currency transactions. More realistic might be to examine the scope for the Fund to borrow directly from private international capital markets. Up until now this option has been rejected as being inconsistent with the Fund's essential character. But if this is changing in a global environment where capital market volatility has become an increasingly important problem, the idea may be worthy of re-examination. 26

Concluding remarks Has the debate about the Fund led to any consensus about reform? The Reports mentioned earlier in this chapter do share some common ground. They agree first that the Fund should be retained but reformed; second, that its operations

22 The ongoing debate about the IMP should be more transparent and that it should be more accountable; third, that there should be a clearer delineation between the Fund and the World Bank and that the Fund should withdraw from long-term lending to developing countries. Finally, they agree that it should strongly encourage countries to avoid pegged exchange rates. Areas of dissent across the reports relate to the circumstances in which countries should be able to borrow from the Fund and the role of preconditions and conditionality, although the dissent widens when the views of other commentators are taken into account. What is the likelihood of reform? There has been some movement by the Fund over recent years. The image of the Fund as an institution impervious to criticism and incapable of change is misplaced. It has become increasingly open, has shown a willingness to introduce new lending facilities and modify existing ones and is in the process of reviewing both conditionality and quotas. Having said this, it seems unlikely that there will be fundamental reforms. Changes are likely to be marginal, incremental and piecemeal. Fundamental reform is only likely to be seen as appropriate in the midst of a global crisis. But in such circumstances time will not allow reform to be fundamental. Once the crisis passes the perceived need for fundamental reform diminishes. Thus, it seems improbable that conditionality will be abandoned but likely that its focus on monetary policy, fiscal policy and exchange rate policy will be reaffirmed. It seems improbable that the Fund will cease to lend to developing countries altogether and pass this role over completely to the World Bank, but likely that the pursuit of closer institutional cooperation and coordination will continue. It seems improbable that the Fund will withdraw from lending in crisis situations but likely that it will reform the Contingent Credit Line (CCL) and encourage the growth of regional arrangements to augment international liquidity in the event of crises. Moreover, while industrial countries may be persuaded to grant a bigger say to developing and emerging economies it is unlikely that they will relinquish their dominant position in the Fund's decision making. Just as it was the East Asian crisis that heightened interest in a new international financial architecture and a reformed role for the IMF, in the post-crisis recovery the momentum for reform will slacken, at least until the next crisis comes along. 27

Notes This section aims to provide only a brief historical account. The Fund itself has produced a series of volumes which discuss the origins and evolution of the IMF in considerable detail. For example Horsefield (1969) describes the establishment of the IMF and its early years. A useful retrospective view on the Bretton Woods system is to be found in Bordo and Eichengreen (1993), with James (1996) providing a thorough discussion of events since the collapse of that system up until the early 1990s. Bordo and James (2000) focus on the role of the IMF taking an historical perspective and discuss, in slightly more detail, and sometimes with different emphasis, many of

The ongoing debate about the IMF 23

2

3 4 5

6

7

8

9

10

the issues that we cover here. They also provide a useful summary of institutional details that we omit. There is, of course, an underlying issue of causation. Was it the Bretton Woods system that facilitated or even created economic success, or was it economic stability and prosperity that enabled the Bretton Woods system to survive? At the time these problems were expressed largely in terms of export instability and adverse long-run terms of trade movements. For a brief discussion of the origins and evolution of the SDR see Bird (1998). Bordo and James (2000) claim that 'the best way of thinking of the IMF and its functions during the period of the so-called Bretton Woods regime (1945-73) is not so much as an institution, but as the institutional embodiment of a system of rules (the IMF's Articles of Agreement)' (p. 13). Given this interpretation, the collapse of the Bretton Woods system amounted to almost the same thing as the collapse of the 'institutional embodiment'. An international monetary conference held at the Smithsonian Institute in Washington DC in December 1971 had, in effect, tried to salvage the Bretton Woods system by increasing the dollar price of gold, realigning exchange rates and introducing slightly wider bands around the new central values. In effect it was the failure of the Smithsonian Agreement that drove the global economy towards flexible exchange rates. In particular, the Bank for International Settlements (BIS) had a mandate to promote international cooperation on monetary and financial issues with the objective of contributing to international financial stability. Over time a range of other institutional changes, including global economic summits involving G7 countries, began to erode somewhat the position of the IME But in this context it is interesting that during the 1960s the Fund had had to fight hard to ensure that the SDR was incorporated within the IMF and that SDRs should be allocated universally rather than just to industrial economies. Even at that time swap agreements designed to provide additional liquidity to support pegged exchange rates were orchestrated largely outside the Fund. Two volumes on the IMF produced at the beginning of the 1980s (Williamson, 1983; Killick et al., 1984) took conditionality as their principal theme. Defenders of the Fund also based their arguments on a similar agenda (Nowzad, 1981). The question of reforming the IMF, in large measure, came down to the issue of reforming conditionality (Spraos, 1986). Discussion of other aspects of the Fund's operations did take place and was covered in the literature (see, for example, Bird, 1987a) but these other issuessuch as the Fund's lending role, whether there was a moral hazard problem, and alternative ways of financing the Fund, did not excite commentators on the IMF as much as conditionality. Bird ( 1996a) provides a brief survey of the issues raised by the debate over conditionality and cites many of the most important studies. These include Khan and Knight (1982), Goldstein and Montiel (1986), Gylfason (1987), Pastor (1987) Edwards (1989) and Khan (1990). Edwards criticized the Fund's lethargy in keeping up with advances in macroeconomic theory, while Killick (1989) argued that the Fund had gone too far in its attack on the state. Perhaps the most powerful critique came from Finch (1989). However, others argued that the Fund modified conditionality according to the strategic importance and bargaining position of borrowing countries (Stiles, 1990). Claims that Fund involvement not only reflected political imperatives but also resulted in political instability in borrowing countries were also made, although systematic evidence for this appeared to be lacking (Siddell, 1988). Sachs (1989a,b) was among the strongest critics of the IMF in the context of its involvement in the Third World debt crisis of the 1980s, and linked much of his criticism to the related concept of 'debt overhang'. Although not necessarily seeing this as part of the Fund's responsibilities, it was also in the context of the 1980s debt crisis

24 The ongoing debate about the IMF

11

12

13

14

15

16

17

that the idea of international bankruptcy arrangements or similar began to be widely discussed (e.g. Carden, 1988; Cohen, 1989; Sachs, 1989a). Among others, arguments such as these were put forward by Helleiner (1983 ), anxious to ensure that a preoccupation with Latin American debt did not crowd out consideration of the Fund's role in the poorest countries of the world. In anticipation of its 50th anniversary 'celebrations' Oxfam took the lead in criticizing the effects of the Fund's involvement in Africa (Oxfam, 1993 ). This, in spite of the fact that research demonstrated the complex way in which adjustment was interrelated with poverty (Demery and Addison, 1987). However, UNICEF's attack on Fund and Bank conditionality (Cornia et al., 1987) was illustrative of mounting concerns that conventional IMF conditionality had a negative effect on economic growth, the implications of which for poverty could be accentuated by income distributive effects. This coincided with a Washington consensus across both the Fund and the Bank about the appropriate direction of economic policy in developing countries covering macroeconomic stability, microeconomic liberalization and openness. At about the same time, some critics of the Fund claim that it also extended its conditionality on another front by beginning to use 'political conditionality' to cover governancerelated issues (Kapur and Webb, 2000). The erosion was observed and well described early on by Feinberg (1988), with Bird (1993) summarizing the institutional implications to which this gave rise. The overlap in the responsibilities of the IMF and the World Bank initially led to institutional frictions and 'turf wars'. However, under pressure from their principal shareholders, the institutions devised procedures for minimizing conflict and maximizing coordination and cooperation. One initiative was the introduction of Policy Framework Papers that were associated with ESAF lending. These have now been replaced by Poverty Reduction Strategy Papers which are designed to bring together the Fund and the Bank with the government and civil society. While borrowing countries clearly do not want to be confronted with conflicting conditionality, a more open question relates to the extent to which it is desirable to have both of the Bretton Woods institutions singing from exactly the same song sheet; some degree of differentiation may be desirable. Although some commentators have suggested that agreement around Washington consensus policies served to reduce institutional conflict (Junguito, 1996), it is also clear- according to Stiglitz (2000a)- that there was significant disagreement between the Fund and the Bank about the appropriate policy response to the East Asian crisis. The Fund's own internal reviews had generally been quite positive (Schadler et al., 1993 ), although outside academics heavily criticized the evidence upon which this judgement was reached (Killick, 1995). In response the Fund established a panel of independent assessors - a rare departure - to examine the ESAF in detail. More critical of it than the internal evaluations had been, this created some momentum for reform, although the Fund appeared somewhat reluctant to fully endorse the findings of the outside assessors (Botchwey et al., 1998). Again Sachs became a 'thorn in the side' of the Fund by writing a series of articles which were highly critical of the IMF's handling of transition (e.g. Sachs, 1995). Claims that the United States and France had pressured the Fund to make loans to favoured clients were now matched by the claim that the US and European economies had pushed the Fund into making loans to Russia in circumstances where there was no evidence to suggest that the 'agreed' programmes would be implemented. Institutionally the Fund responded to the needs of CITs by temporarily introducing the Systemic Transformation Facility. To many observers this involved relatively light conditionality, a point not missed by the Fund's developing country members. While mentioning many of the issues that were to come to dominate discussion of the Fund, the Bretton Woods Commission (1995) tended to concentrate most on the

The ongoing debate about the IMF 25

18

19

20

21

question of international macroeconomic coordination. Many industrial economies were, at the time, running large fiscal deficits which were perceived to have an internationally destabilizing effect. Although the Mexican crisis had happened and was seen by some as a sign of things to come, there was insufficient consensus around this view to make capital volatility the dominant theme of the Commission's report. While expressing some dissatisfaction with flexible exchange rates, the Bretton Woods Commission remained studiously non-committal about exchange rate policy, choosing instead to focus on fiscal imbalances as a root cause of exchange rate instability. Apart from Stiglitz (2000a) and Dornbusch (2000) many other well known and well respected economists contributed to the debate, offering severe criticism of the Fund. In a sense this reflected the importance attached to the issues discussed but also raised the profile of the debate. Thus, Feldstein (1998), Rodrik (1999) and Radelet and Sachs ( 1998) were all critical of the design of IMF programmes arguing that conventional fiscal contraction had been used in circumstances where it was inappropriate. Via various papers posted on his website Krugman made a similar argument. They were also generally critical of the premature move towards capital account liberalisation. Radelet and Sachs argued that overly contractionary programmes supported by the Fund had encouraged further capital outflows, making the situation worse. Feldstein argued that conditionality, whatever the merits or demerits of its contents, went far beyond what was legitimate for the Fund to require in order to stabilize economies and restore market access, and certainly far beyond what would have been required of industrial economies in similar circumstances - hence the issue of 'moral right'. Bhagwati (1998) as well as Felstein suggested the crisis and IMF conditionality were being used to further the interests of largely US-based private capital and transnational companies. Feldstein (1998), Sachs (1997) and Stiglitz (2000b) also argued that the IMF was bypassing and overriding domestic political processes and interfering with the jurisdiction of sovereign governments. For illustration of claims that the Fund should be scrapped, see, for example, Schultz et al. (1998). For an attempt to provide an analytical justification for the Fund's existence based on market failure and collective action see Bird (1995), Stiglitz (2000b), and Bird and Joyce (2001). While these sources discuss market failure relative to the possibility of bureaucratic failure, Vaubel (1991, 1994 and 1996) provides a more forthright critique of the Fund on public choice grounds. Willett (1999) offers a 'softer' public choice critique of the Fund, which emphasizes appropriate reform rather than closure. Frey ( 1997) also uses a public choice approach to help analyse the Fund. Some commentators suggest that there is an important distinction between 'global' and 'local' public goods. The logical conclusion based on subsidiarity is that regional agencies may be better able than the IMF to deal with some dimensions of market failure. It is in this context that there have been proposals for an Asian Monetary Fund. Bird and Rajan (2000) discuss the arguments for and against such an idea. The optimum mix will differ according to circumstances. There is a compelling argument that temporary and self-reversing disequilibria should be handled via additional liquidity rather than adjustment, something that has implications for IMF conditionality. Rodrik (1996) and Bird and Rowlands (1997) discuss and attempt to estimate the statistical importance of the catalytic effect of IMF lending. The question of the optimal amount of Fund lending depends, to a degree, on the strength of the catalytic effect, but akin to the debate surrounding the global adequacy of international reserves in the context of proposed additional allocations of SDRs, it is impossible to be precise about the optimal amount of IMF lending. While some commentators have suggested that it is impossible to overstate the importance of the moral hazard problem, others claim that there is no evidence to support it. Disagreement over the moral hazard of IMF lending was one issue upon which the minority report of the International Financial Institution Advisory Commission (IFIAC) strongly distanced itself from the main report.

26 The ongoing debate about the IMP 22 One approach is to ask what functions need to be performed in order to achieve a given set of objectives and then contemplate the best institutional arrangements for realizing these. For some commentators this involves reforming the IME For others it involves setting up new institutions (Eichengreen, 1999, provides a succinct review). Thus, Eatwell and Taylor (2000) propose a World Financial Authority that, inter alia, would assume the responsibility of setting prudential and supervisory standards. Others see the Fund tapping into national expertize and selecting and advocating best practice. For some, the Fund's role would be that of standing at the apex of a pyramid of other agencies, some of which might have issue-specific expertize and others of which would be region-specific. Having the Fund exist alongside other agencies with related functions could also weaken its monopoly power (Stiglitz, 2000b). 23 Mosley et al. (1991) offer a comprehensive assessment of World Bank structural adjustment lending and point to a number of deficiencies. Moreover, there is evidence to suggest that, even where structural adjustment has contributed to an improvement in economic performance, this has had little to do with the World Bank (Dollar and Svensson, 1998). Reforms have served to strengthen the ESAF and there is some evidence to suggest that it may have become a more effective tool of IMF lending (Bird and Mosley, 2000, Dicks-Mireaux et al., 2000). 24 As noted earlier, in a large literature dealing with various aspects of the IMF's operations, most attention has been paid to conditionality and the effects of IMF-supported programmes. However, there is probably as much that we do not know as we do. A recent attempt to survey this area drawing on all the research available at the time is provided by Killick (1995a). However, it is not enough to know whether IMF conditionality is effective or not. We also need to know what factors influence effectiveness. Why might it be effective in some cases but not in others? Collier (1997) provides a strong critique of conditionality as it has been exercised by the IMF and the World Bank. 25 Some critics argue that the fact that the IMF is accountable to finance ministries rather than (say) labour ministries also shapes its policy advice in a globally inappropriate way, providing a deflationary bias to an institution which was originally established precisely to avoid deflation and recession. On governance, a worry may be that giving the Fund staff greater autonomy could further expose it to the public choice critique. Again, for this reason, some reformers advocate greater accountability rather than autonomy (Stiglitz, 2000b), although the two are not necessarily mutually exclusive. Helleiner (2000) and others have stressed how developing countries, and in particular the poorest countries of the world, are largely excluded from the decisionmaking processes of global institutions, including the IMF, where the United States frequently carries an effective veto and the Managing Director is by convention a European. The worry is that an institution which advocates good governance and democracy may itself fail to adhere to the basic principles which it expects of others. What can small developing economies do when they disagree with policies adopted by the Fund? Given their relatively poor representation in the Fund's decisionmaking processes the 'voice' mechanism is not available to them. However, neither is the 'exit' mechanism, since countries generally only tum to the Fund when nobody else is prepared to lend to them. Woods (1998) provides a review of governance issues associated with the Bretton Woods institutions. 26 Vaubel ( 1996) provides a critique of the quota system based on the idea of 'hurry up' lending. A more rounded analysis of the quota system is provided by Bird (1987b). Bird and Rowlands (2001) discuss the idea of direct borrowing from private capital markets, while Polak (1996, 1999) advocates a more full-blown 'streamlining' of the Fund's financial structure based on the SDR. 27 Although international organizations have a perhaps surprisingly high mortality rate overall, it is quite low for large organizations. Moreover, the longer an organization survives, the longer it is likely to survive. The IMF is large and has been around a long

The ongoing debate about the IMF 27 time. It Is most unlikely to be scrapped. But what factors affect the susceptibility of organizations to reform (Kapur and Webb, 2000)? Is international monetary reform involving significant changes to the IMF likely to occur (Bird, 1996b)? Policy makers seem to be myopic.

References Bhagwati, J. (1998) 'The Capital Myth', Foreign Affairs, May-June. Bird, G. (1987a) International Financial Policy and Economic Development (London: Macmillan). Bird, G. (1987b) 'Financing the Fund and Reforming Quotas', in G. Bird, International Financial Policy and Economic Development (London: Macmillan). Bird, G. (1993) 'Sisters in Economic Development: The Bretton Woods Institutions and Developing Countries', Journal of International Development, 5(1), 1-25. Bird, G. (1995) IMF Lending to Developing Countries: Issues and Evidence (London: Routledge). Bird, G. (1996a) 'The International Monetary Fund and Developing Countries: A Review of the Evidence and Policy Options', International Organization, Summer, 477-511. Bird, G. (1996b) 'From Bretton Woods to Halifax and Beyond: The Political Economy of International Monetary Reform', The World Economy, 19(2), 149-72. Bird, G. (1998) 'The Political Economy of the SDR: The Rise and Fall of an International Reserve Asset', Global Governance, 4, 355-79. Bird, G. and P. Joyce (2001) 'Remodeling the Multilateral Financial Institutions', Global Governance, 7, 75-93. Bird, G. and P. Mosley (2000) 'Should the IMF Discontinue its Long Term Lending Role in Developing Countries?', paper presented to the Development Studies Association conference, London, November. Bird, G. and R. Rajan (2000) 'Is There a Case for an Asian Monetary Fund?', World Economics, 1(2), 135-44. Bird, G. and D. Rowlands (1997) 'The Catalytic Effect of Lending by the International Financial Institutions', The World Economy, 20(7), 967-91. Bird, G. and D. Rowlands (2001) 'Catalysis or Direct Borrowing: The Role of the IMF in Mobilising Private Capital', The World Economy, 24(1), January, 81-98. Bordo, M. and B. Eichengreen (eds) (1993) A Retrospective on the Bretton Woods System: Lessons for International Monetary Reform (Chicago, IL: University of Chicago Press). Bordo, M. and H. James (2000) 'The International Monetary Fund: The Present Role in Historical Perspective', NBER Working Papers, 7724, NBER, MA, June. Botchwey, K.P. Collier, J.W. Gunning and K. Hamada (1998) Report by a Group of Independent Experts: External Evaluation of the ESAF (Washington, DC: IMF). Bretton Woods Commission (1995) 'Report and Background Papers to the Bretton Woods Commission', in Bretton Woods: Looking to the Future (Washington, DC). Cohen, B.J. (1989) 'Developing Country Debt: A Middle Way', Essays in International Finance, 173 (Princeton, NJ: Princeton University Press). Collier, P. (1997) 'The Failure of Conditionality', in C. Gwin and J. Nelson (eds), Perspectives on Aid and Development, ODC Policy Essay no. 22 (Washington, DC: Overseas Development Council). Corden, M. ( 1988) 'An International Debt Facility', IMF Staff Papers, June.

28 The ongoing debate about the IMP Cornia, G.A., R. Jolly and F. Stewart (eds) (1987) Adjustment with a Human Face: Protecting the Vulnerable and Promoting Growth (Oxford: Oxford University Press). Demery, L. and T. Addison (1987) The Alleviation of Poverty Under Structural Adjustment (Washington, DC: World Bank). Dicks-Mireaux, L.M. Mecagni and S. Schadler (2000) 'Evaluating the Effect of IMF Lending to Low Income Countries', Journal of Development Economics, 61,495-526. Dollar, D. and J. Svensson (1998) 'What Explains the Successes or Failure of Structural Adjustment Programs?', Policy Research Working Paper (Washington, DC: World Bank). Dornbusch, R. (2000) 'Stiglitz Versus the IMF: Another View', New Republic, April. Eatwell, J. and L. Taylor (2000) Global Finance at Risk: The Case for International Regulation (New York: New Press). Edwards, S. (1989) 'The International Monetary Fund and the Developing Countries: A Critical Evaluation', Carnegie Rochester Conference Series on Public Policy, 31, 7-68. Eichengreen, B. (1999) Toward a New International Financial Architecture: A Practical Post Asia Agenda (Washington, DC: Institute for International Economics). Feinberg, R.E. (1988) 'The Changing Relationship Between the World Bank and the International Monetary Fund', International Organization, Summer. Feldstein, M. (1998) 'Refocusing the IMF', Foreign Affairs, 77, 20-3. Finch, D.C. (1989) 'The IMF: The Record and Prospect', Essays in International Finance, 175 (Princeton, NJ: Princeton University). Frey, B. (1997) 'The Public Choice of International Organizations', in D. Muller (ed.), Perspective on Public Choice (Cambridge: Cambridge University Press). Goldstein, M. and P. Montiel, (1986) 'Evaluating Fund Stabilization Programs with Multicountry Data: Some Methodological Pitfalls', IMF Staff Papers, 33, 304-44. Gylfason, T. (1987) 'Credit Policy and Economic Activity in Developing Countries with IMF Stabilization Programs', Studies in International Finance, 60 (Princeton, NJ: Princeton University). Helleiner, O.K. (1983) The IMF and Africa in the 1980s, Essays in International Finance, 152 (Princeton, NJ: Princeton University Press). Helleiner, O.K. (2000) Developing Countries in Global Economic Governance and Negotiations, in D. Nayyar (ed.), Governing Globalization: Issues and Institutions (Helsinki: Wider/Oxford University Press). Horsefield, J.K. (1969) The International Monetary Fund, 1945-1965 (Washington, DC: International Monetary Fund). James, H. (1996) International Monetary Co-operation Since Bretton Woods (New York: Oxford University Press). Junguito, R. (1996) 'IMF-World Bank Policy Advice: The Co-ordination/Cross Conditionality Question', in O.K. Helleiner (ed.), The International Monetary and Financial System (New York: StMartin's Press). Kapur, D. and R. Webb (2000) 'Governance-Related Conditionalities of the International Financial Institutions', 024 Discussion Paper Series 6, Center for International Development, Harvard University, UNCTAD, August. Khan, M. ( 1990) 'The Macroeconomic Effects of Fund-supported Adjustment Programs', IMF Staff Papers, 3 7, 195-231. Khan, M. and M. Knight (1882) 'Some Theoretical and Empirical Issues Relating to Economic Stabilization in Developing Countries', World Development, 10, 709-30.

The ongoing debate about the IMF 29 Killick, T. (1989) A Reaction Too Far: Economic Theory and the Role of the State in Developing Countries (London: Overseas Development Institute). Killick, T. (1995a) IMF Programs in Developing Countries: Design and Effect (London: Routledge). Killick, T. (1995b) 'Can the IMF Help Low Income Countries: Experiences With the Structural Adjustment Facilities', The World Economy, 18(4), 603-16. Killick, T. et ai. ( 1984) The Quest for Economic Stabilization: The IMF and the Third World (London: Gower Overseas Development Institute). Mosley, P., J. Harrigan and J. Toye (1991) Aid and Power: The World Bank and Policy-based Lending, 2 vols (London: Routledge). Nowzad, B. (1981) 'The IMF and its Critics', Essays in International Finance, 146 (Princeton, NJ: Princeton University Press). Oxfam (1993) Africa Make or Break: The Failure of IMP/World Bank Policies (Oxford: Oxfam). Pastor, M. (1987) 'The Effects oflMF Programs in the Third World: Debate and Evidence from Latin America', World Development, 15, 249-62. Polak, J. (1996) 'Should the SDR become the Sole Financing Technique for the IMF?', in M. Mussa, J. Boughton and P. lsard (eds), The Future of SDR in Light of Changes in the International Financial System (Washington, DC: International Monetary Fund) pp. 221-38. Polak, J.J. (1999) 'Streamlining the Financial Structure of the International Monetary Fund', Essays in International Finance, 216 (Princeton, NJ: Princeton University Press). Radelet, S. and J.D. Sachs (1998) 'The East Asian Financial Crisis: Diagnosis, Remedies and Prospects', Brookings Papers on Economic Activity, 11-74. Rodrik, D. (1996) 'Why is there Multilateral Lending?', in M. Bruno and B. Pleskovic (eds), Annual World Bank Conference on Development Economics (Washington, DC: World Bank). Rodrik, D. (1999) 'Governing the Global Economy: Does One Architectural Style Fit All?', mimeographed. Sachs, J.D. (1989) 'Strengthening IMF Programs in Highly Indebted Countries', in C. Gwin and R. Feinberg (eds), The International Monetary Fund in a Multipolar World: Pulling Together (Washington, DC: Overseas Development Council). Sachs, J.D. (1989a) 'New Approaches in the Latin American Debt Crisis', Essays in International Finance, 174 (Princeton, NJ: Princeton University Press). Sachs, J.D. (1989b) 'Conditionality, Debt Relief and the Developing Country Debt Crisis', in J.D. Sachs (ed.), Developing Country Debt and Economic Performance, vol. 1, International Financial System (Chicago, IL: University of Chicago Press). Sachs, J.D. (1995) 'Why has Russia Failed to Stabilize?', in A. Aslund (ed.), Russian Economic Reform at Risk (London: Pinter) pp. 53-63. Sachs, J.D. ( 1997) 'IMF is a Power Unto Itself', The Economist, 11 December. Schalder, S., F. Rozwadowski, S. Tiwara and D.O. Robinson (1993) 'Economic Adjustment in Low Income Countries: Experience Under the Enhanced Structural Adjustment Facility', Occasional Paper 106 (Washington, DC: International Monetary Fund), September. Shultz, G., W. Simon and W. Wriston (1998) 'Who Needs the IMF?', Wall Street]ournal, 3 February. Siddell, S. R. ( 1988) The IMF and Third World Instability: Is There a Connection? (London: Macmillan).

30 The ongoing debate about the IMF Spraos, J. (1986) 'IMF Conditionality: Ineffectual, Inefficient, Mistargeted', Essays in International Finance 166 (Princeton, NJ: Princeton University Press). Stiglitz, J. (2000a) 'What I Learned at The Economic Crisis', New Republic, April. Stiglitz, J. (2000b) 'Globalization and the Logic of International Collective Action: Re-examining the Bretton Woods Institutions', in D. Nayyar (ed.), Governing Globalization: Issues and Institutions (Helsinki: Wider/Oxford University Press). Stiles, K. W. (1990) 'IMF Conditionality: Coercion or Compromise', World Development, 18,959-74. Vaubel, R. (1991) 'The Political Economy of the International Monetary Fund', in R. Vaubel and T.D. Willett (eds), The Political Economy of International Organizations: A Public Choice Perspective (Boulder, CO: Westview Press). Vaubel, R. (1994) 'The Political Economy of the IMF: A Public Choice Analysis', in D. Bandow and J. Vasquez (eds), Perpetuating Poverty: The World Bank, the IMF and the Developing World (Washington, DC: The Cato Institute). Vaubel, R. (1996) 'Bureaucracy at the IMF and the World Bank', The World Economy, 19(2), 195-210. Willett, T.D. ( 1999) 'A Soft Core Public Choice Approach to IMF Reform', mimeographed. Williamson, J. (ed.) (1983) IMF Conditionality (Washington, DC: Institute for International Economics). Woods, N. (1998) 'Governance in International Organizations: The Case for Reform in the Bretton Woods Institutions', in International Monetary and Financial Issues for the 1990s, vol. IX (New York and Geneva: United Nations).

2

Borrowing from the IMF The policy implications of recent empirical research

Introduction There are now a number of studies which attempt to move away from anecdotal evidence on individual member countries' borrowing experiences towards empirical investigation of the factors that are in general associated with countries drawing (or indeed not drawing) resources from the International Monetary Fund (IMF). A range of methodologies and data sets have been used, and there have been differences in the results. In a recent paper, Knight and Santaella ( 1994) provide both a summary of existing work and a useful addition to it, by using Probit analysis to estimate the probability that a financial arrangement will be approved for a given country in a given year. Their analysis seeks to distinguish between those factors which cause countries to request financial assistance from the Fund (the demand side) and those that are associated with the Fund's response (the supply side). A feature of the research is the large data set used: a pooled sample of annual observations on major economic variables for 91 developing countries over 1973-91. In another important study covering 76 developing countries, Conway (1994) uses Probit and Tobit analysis to explain participation in Fund-backed programmes over 1976-86. Relative to the research that has been done on the effects of Fund-backed adjustment programmes, the question of what factors lead to their initial adoption has remained rather neglected over the years. This is unfortunate since, in an important sense, Fund-backed programmes should surely be endeavouring to rectify the circumstances which cause countries to borrow in the first place. After all, the Fund's Articles envisage it providing only temporary support. From the research available it is probably legitimate to claim that we now have a reasonable understanding of the overall effects of Fund-backed programmes.1 But is there a similar degree of consensus about the characteristics of user countries? If there is, what conclusions may be drawn with respect to future policy involving the Fund? Although both studies cited above make a significant contribution towards answering the first of these questions, they side-step the second. At a time when the role of the IMF is under close scrutiny, this is an unfortunate omission. This chapter seeks to address both issues in order to advance the policy debate.

32

Borrowing from the IMF

The section 'Circumstances surrounding Fund arrangements: areas of consensus' puts the Knight-Santaella (KS) and Conway studies in the context of earlier research on the characteristics of user countries including Bird and Orme (1981), Cornelius (1987) and Joyce (1992) and identifies areas where a consensus is emerging. The section 'Areas of dissension and debate' discusses issues where significant disagreement remains. The section 'Policy implications: recidivism, conditionality, growth and graduation' examines the policy implications of some of the more robust findings, and the final section offers some concluding remarks, as well as a brief critical analysis of recent policy developments in the light of the empirical work reviewed.

Circumstances surrounding Fund arrangements: areas of consensus Underlying theory suggests that the demand for Fund credits will vary positively with the size and incidence of balance-of-payments (BoP) deficits and negatively with the availability of alternative means of financing. BoP deficits may themselves result from a number of interdependent causes, including adverse movements in the terms of trade, excessive inflation (which may in turn be the consequence of domestic fiscal and monetary mismanagement or increased import prices), overvalued real exchange rates (which are either caused by excess domestic inflation or by a decline in the sustainable real equilibrium exchange rate) and high levels of external debt. Financing may come from the decumulation of international reserves, or from international borrowing. Given these factors, as well as the demand for Fund credit as a demand for technical assistance, and the notion of the Fund as a lender of last resort, borrowing from the Fund may also be expected to be inversely related to the level of economic development. All studies of the demand for Fund credit have adopted a similar theoretical framework, but they have differed in the precise range of explanatory variables included and the empirical proxies used. Neither KS nor Conway deparr from convention, although the list of the demand-side factors used in the KS study is quite lengthy and incorporates the existence of a previous Fund arrangement. KS also include the domestic investment ratio, arguing that countries with a low investment ratio are more likely to turn to the Fund since this indicates limited access to international capital markets, limited imports of capital and intermediate goods and distorted domestic credit markets. In terms of previous Fund arrangements, they claim that countries that have had arrangements with the Fund in the past are more likely to enter into subsequent arrangements since they are further along their learning curve in terms of negotiating with the Fund. Neither Conway nor KS however, attempt to explain the demand for Fund credits in terms of variations in conditionality (although KS introduce this indirectly in the context of the supply side), or in terms of any difference between the rate of charge on Fund resources and interest rates in private international capital markets. They do not, therefore, examine demand as a function of the implicit price of Fund credit.

Borrowing from the IMP 33 In principle, Fund arrangements could reflect the interplay between the demand for them, as a positive function of the degree of concessionality and a negative function of the degree of conditionality, and the supply of them, as a positive function of both the rate of charge on the use of Fund resources and the degree of conditionality. To the extent that borrowing from the Fund is largely demand-determined, a widening excess of commercial interest rates over and above the rate of charge on Fund resources, along with a perceived relaxation in conditionality, would be expected to result in an increasing number of Fund arrangements. Concessionality and conditionality provide the Fund with mechanisms for managing the demand for its credits, with a public choice approach suggesting that there would be institutional pressures to use such mechanisms to ensure that the demand for resources is sufficient to justify expanding the Fund's size of operations. The results discovered by KS and Conway are in many ways in line with those of earlier studies, allowing us to claim that there is indeed some consensus on the economic determinants of the demand for Fund credits. Taking the results of both bivariate and univariate probit procedures, KS find that the stock of international reserves, the real effective exchange rate, the external debt-service ratio, the growth rate of per capita GOP, the investment ratio, the level of per capita GOP, and the dummies for previous Fund arrangements are all empirically significant and have the 'expected' sign. But, coefficients on the overall balance of payments, the inflation rate, the terms of trade, and non-Fund financial resources are insignificant and have the 'wrong' sign. Conway finds that exogenous factors have a predictable and significant effect on participation. Improved terms of trade lower it, while a rising debt burden increases it. He also finds that variations in the 'international climate' influence participation, as does prior macroeconomic performance, with rapid economic growth in the previous year reducing the time spent in Fund programmes. Participation in a programme in one year is found to make future participation less likely but only after a three-year lag. Meanwhile, a higher ratio of foreign reserves to imports reduces the duration of participation. Which of these results tally with those of earlier studies? Broad consensus exists on the following issues. First, the Fund is a more important source of finance for relatively poor countries, demand for Fund credit is inversely related to economic growth and development which can, as a result, wean countries away from the Fund. Second, the probability that a country will demand credit is, in the short run, positively related to its previous involvement with the Fund, although programmes phased over a longer time span seem to be more effective at reducing future participation. Other studies of IMP lending have drawn attention to the strong tendency for the Fund to have programmes in an enduring set of developing countries, even to the extent that it has a quasi-permanent presence in some of them. 2 Third, the availability of non-Fund sources of external finance is, at the aggregate level, insignificantly associated with the use of Fund credits, although the sign of the coefficient varies across studies (Bird, 1994). This suggests that private external financing is neither clearly a substitute for nor

34 Borrowing from the IMP a complement to borrowing from the IME Fourth, exchange rate overvaluation on the other hand significantly and positively affects the demand for Fund resources. KS found that this is associated with external factors which lower the equilibrium exchange rate rather than with domestic inflation which causes the real exchange rate to appreciate. Fifth, although some studies have failed to identify a significant relationship, the consensus is that international reserve holdings have a significant negative effect on the demand for Fund arrangements. A similar consensus exists that, while all countries turning to the Fund have a necessary BoP need, neither a current account nor an overall BoP deficit provides sufficient motivation for seeking Fund support. The mere existence of a BoP deficit does not, in itself, make it probable that a country will demand resources from the Fund.

Areas of dissension and debate Clearly, there are broad areas of agreement which are reinforced by the KS and Conway studies. Where then are the areas of dissension? The first relates to the terms of trade, for, while KS find them to have an insignificant effect (and the wrong sign), Conway and others have found their impact to be both significant and to work in the way that theory predicts. Detailed empirical studies of the circumstances under which developing countries tum to the Fund have also found adverse movements in the terms of trade to be an important factor (Killick et al., 1992).3 More recent research within the Fund (Santaella, 1995) has also found external factors such as the terms of trade to exert a significant influence over the demand for Fund resources, and the KS result is therefore left as something of an outlier. A second area of dissension relates to the effects of external debt on the demand for Fund credit. While Bird and Orme (1981), and Joyce (1992), discovered a negative but insignificant effect, KS endorse Conway's finding that external debt (particularly short-term debt in the case of Conway's study), is significantly and positively associated with drawings on the Fund. A third question which remains open is the extent to which the demand for financial arrangements with the Fund may be explained by economic factors alone. Many of the early studies based on ordinary least squares (OLS) regression had a low coefficient of determination, and the models had poor predictive qualities (Bird and Orme, 1981; Cornelius, 1987). This feature was also true of the study by Joyce (1992) which used logit analysis. The challenge, therefore, seemed to be to explain the residual variation which was unexplained by economic characteristics. As noted earlier, KS do not attempt to estimate a demand function for Fund credit as such; instead they try to explain Fund arrangements in terms of a confluence of demand-side and supply-side variables. Supply-side constraints could, in principle, explain the tendency that earlier models had to overpredict borrowing from the Fund. It would certainly be interesting to know to what extent the Fund passively responds to the demand for arrangements from its members or

Borrowing from the IMF 35 actively rations its credits. Beyond this, if there is credit rationing, upon what basis does the Fund ration? Unfortunately, it is in its treatment of the supply side that the KS study is least compelling methodologically. The principal difficulty comes from its attempt to proxy ex ante intentions by ex post outcomes. 4 This would be acceptable if there was a close positive correlation between the two. But many studies of the effects of Fund-backed programmes catalogue their poor implementation record (Edwards, 1989; Killick 1995; Conway, 1994) and, in these circumstances, the procedure is illegitimate. Somewhat surprising therefore are the statistically significant results that KS achieve on both government revenue and expenditure, as well as, to a lesser degree, the growth of real domestic credit. More meaningful would have been a detailed quantitative analysis of 'letters of intent' and 'prior actions'; after all, it is a country's willingness to make commitments that may be expected to influence the Fund's willingness to agree a loan. 5 In any event, to discover, as they do, that the Fund is likely to favour exchange rate devaluation, and fiscal and monetary restraint, is no more than would have been expected, given existing research. Even with its methodological problems however, the KS discussion of the supply side remains interesting. Without doubt, the Fund has made important modifications to its conditionality over recent years, paying greater attention to structural problems and the costs of adjustment on vulnerable groups. But there has been a lively debate about the extent to which the hard core of programmes has changed as a consequence (Killick, 1995). Has the reality matched the rhetoric? Implicitly KS provide both qualitative and quantitative insights into this important question. They claim that because of the changes to conditionality, ... the design of programs supported by Fund financial arrangements (has become) much more complex and comprehensive than it was in earlier years ... Nevertheless, since a balance of payments deficit usually reflects an excess of aggregate domestic demand relative to domestic supply, Fund programs ... must emphasise measures of demand restraint. The policy instrument variables included in our empirical model are therefore intended to reflect these essential demand-management elements of a stabilisation program. (Knight and Santaella, 1994: 18)

This implies qualitatively that the hard core of Fund-backed programmes has not, in their view, changed, and that fiscal and monetary restraint combined with exchange rate devaluation remain absolutely central. Had there been a significant change in conditionality, the model used by KS should surely have yielded rather poor results, or, at the very least, should have yielded less good results in more recent years. In fact KS offer some recognition of the intertemporal dimension by including a dummy to reflect the 'clustering' of approvals since 1979, but they do not interpret this as reflecting changes in conditionality over time. They find that their arrangement-supply equation has good overall explanatory power. Although the discovery of empirical support for their model

36

Borrowing from the IMF

is surprising given the evidence on poor implementation, to the extent that it holds, it provides quantitative support for those who argue that the changes in conditionality have been more cosmetic than real. The KS finding that adjustments in government revenues and expenditure and in exchange rates are more significant than those in domestic credit creation is less surprising given that earlier studies of the effects of Fund programmes have also found them to have an insignificant or weak impact on monetary variables. Conway also finds that the effects of Fund programmes are not uniform either across intermediate policy instruments or across ultimate performance targets. In terms of policy, the most significant influence is felt on the budgetary balance where public investment declines. No significant impact is discovered on either the exchange rate (unlike other studies that have found it to be the area where the Fund exerts greatest effect, Killick et al., 1992) or domestic credit creation. This is interesting given the conventional caricature of Fund programmes as focusing on the exchange rate and domestic credit creation. Indeed with other research (Santaella, 1995) also finding little empirical evidence of the Fund exerting an effect on either the exchange rate or monetary aggregates, a reasonable question is, what is left of IMF conditionality? In terms of economic performance, Conway finds two significant effects: an improvement in the current account ratio and a decline in the investment ratio. Both inflation and economic growth decline in the short run, but the effects are not statistically significant. Over the longer run, however, economic growth rises significantly. These results generate plenty of fascinating questions. By what mechanism is the improvement in the current account achieved in the absence of an effect on the exchange rate? It would appear that fiscal contraction must be sufficient to drive down the demand for imports. But again this is surprising given the insignificant effects of programmes on inflation, and the fact that other studies have shown that the current account improvement tends to be shared between import compression and export expansion (Killick et al., 1992). Furthermore, there is an inconsistency between the decline in the investment ratio and the lagged increase in economic growth, given that Conway (1994) acknowledges 'the importance of domestic investment to the future growth prospects of developing economies' (Conway, 1994: 381). Is it that there is a sharp rise in the marginal productivity of capital, and is this sufficient to sustain growth in the face of a falling investment ratio? Returning now to the predictive qualities of the KS and Conway models, how do these compare with previous studies? Early failures to explain drawings on the Fund in terms of purely economic variables led to suggestions that negotiating a Fund arrangement was as much a political event as an economic one. The recidivist nature of Fund borrowing could then be seen as reflecting the idea of a 'threshold'. Countries might resist borrowing from the Fund even if their economic circumstances warranted it because of a perceived political cost. The political cost could be interpreted, however, as a quasi-fixed cost which, once incurred, would not affect further marginal decisions to borrow. The tendency for economic models to overpredict drawings on the Fund was interpreted as revealing this perceived political cost.

Borrowing from the IMF 3 7 By incorporating the supply side in negotiating Fund arrangements, KS implicitly introduce another potential political factor in terms of the Fund's own political preferences. Concentrating explicitly on exclusively economic variables, however, the KS and Conway studies are much more successful in predicting outcomes than earlier research has been. The KS study's predictions with respect to approvals and non-approvals vary according to the use of the bivariate or univariate model, but the correct answer is generated in at least 80 per cent of cases. Conway claims a 90 per cent success rate in predicting participation. The interesting feature is that the KS model is better at predicting non-approvals than approvals, which it tends to underpredict. KS try to explain the tendency of their model to produce 'false negatives' in terms of the willingness of the Fund to use (and members to accept) structural conditionality. This would be relevant either if, in some cases, structural conditionality substituted for (rather than merely complemented) conventional conditionality, or if member countries gained positive utility from Fund conditionality and preferred to have more of it rather than less. These conditions are met neither in principle nor in practice, however, and the KS explanation is therefore unconvincing. A less contrived explanation may perhaps be found in terms of political economy. Some observers have claimed that, particularly in the immediate aftermath of the post-1982 debt crisis, the Fund relaxed its conditionality and became prepared to agree to arrangements in circumstances where it would previously have rejected them (Finch, 1989). The claim was that political pressures were eroding the Fund's financial standards. The KS result could be interpreted as providing some empirical support for this claim, although it should also be noted that empirical studies that have set out to test for it have found little evidence of any systematic political bias (Rowlands, 1994 ). Moreover, the institutional explanation of Fund activity has received mixed empirical support. While Vaubel (1991) has claimed to have found evidence of its existence, not least in the context of the IMF 'hurrying up' lending in advance of quota reviews, others have argued that changes in the degree of concessionality have been principally associated with variations in world interest rates. Moreover, evidence pointing to the substantial continuity in the design of conditionality at the same time as there have been large variations in the availability of external finance from commercial banks and private capital markets suggests that the public choice approach is largely misplaced (Bird, 1995). In any case the 'good results' that KS and Conway achieve on the basis of economic characteristics alone appear to leave less scope for a political explanation of Fund borrowing and lending than was previously thought to exist.

Policy implications: recidivism, conditionality, growth and graduation Although clearly some questions remain to be answered, the results of the recent studies by Knight and Santaella and by Conway, in conjunction with those of earlier studies, provide sufficient foundation upon which to build fairly robust

38 Borrowing from the IMP policy conclusions. Policy should focus on these areas of consensus, while future research is seeking to resolve the remaining areas of disagreement. Attention was drawn earlier to the strong elements of recidivism that have been noted in studies of IMF lending (Goreux, 1989; Killick et al., 1992; Bird, 1995). KS confirm this recidivist tendency, but include previous arrangements with the Fund in their model with an 'expected' positive sign. A positive sign only makes sense however, as an indicator of success in the context of a political explanation of Fund arrangements, and KS reject such an explanation. 'Learning' in terms of negotiating with the Fund, the explanation that KS offer, could perhaps be sustained as a success indicator were the tendency to return to the Fund interspersed with periods when there was no involvement, but this is not what the evidence shows. Indeed, Conway's results suggest that the probability of future referral declines after a three-year lapse. Moreover, countries can 'learn' about negotiating with the Fund without 'doing' it. An equally effective route would be for governments to hire negotiators who had previously worked for the Fund and who were therefore well informed about its operations. Given the Fund's Articles of Agreement, and the notion of Fund financial support as being temporary and revolving, recidivism should more realistically be seen as an indicator of failure. The Fund fails if it does not design and effectively encourage the implementation of programmes that improve economic performance to the extent required to make unnecessary further referrals to the Fund. But why is there a recidivist tendency? Is it that, even though programmes are usually fully implemented, the necessary improvement in macroeconomic performance fails to materialize? Such an explanation would point to deficiencies in the design of conditionality, the sheer size of the initial macroeconomic disequilibria that have to be corrected; or the incidence of exogenous shocks that push economies off course. Or, is it that countries fail to implement the agreed conditions? Is it only the noncompliers that tum out to be recidivists? If so, how can noncompliance be explained? It may be that inadequate attention has been paid to the political economy of implementation in the design of the program. It is certainly overly simplistic to view noncompliance simply as evidence of a 'lack of political will'. Instead, the reasons why governments reach the decisions that they do need to be understood, and attempts made to include these factors in designing policy reform. It may also be that the failure to realize targets reflects overambition in setting them. This could be explained by political as well as by economic factors, with both the country and the Fund sides in negotiations having a mutual desire to strike a deal that will be supported by the Executive Board. Of course, if it is the non-complying countries that are the recidivists, another question is why should the Fund continue to lend to countries that have ignored its advice in the past. To the extent that countries have opted not to implement agreed policies (rather than tried and failed), the Fund could eliminate recidivism and raise the incentive to implement agreed policies by refusing to make new loans. Indeed by offering them, the Fund opens itself to a moral hazard

Borrowing from the IMF 39 criticism that it is effectively encouraging countries to perpetuate their economic problems in order to gain access to Fund finance. The difficulty is in discerning adjustment effort, as distinct from achievement, since the Fund would not wish to penalize countries in which the failure to achieve targets is associated with factors largely beyond the control of the government. While this is an area where there are probably more questions than answers, there is some evidence upon which to draw. First and unambiguously, the majority of programmes which are supported by the Fund are not completed, with most of the non-completion reflecting non-compliance with the quantitative components of the programme. Second, but more ambiguously, there seems to be little statistically significant difference between the macroeconomic performance of non-compliers as compared with compliers (Killick et al., 1992), although Conway finds that compliance does have a significant positive effect on some aspects of future performance. Clearly, where compliance has no significant impact on performance, recidivists would be just as likely to be compliers as noncompliers, and recidivism could be as much a reflection of the content of programme design as the political economy of implementation. But why do so many programmes remain incomplete? Part of the answer may lie in the findings of Knight and Santaella which accentuate the orientation of Fund arrangements towards aggregate demand. The short-term compression of aggregate demand clearly generates greater political and social unrest than the longer term expansion of aggregate supply. Governments may therefore seek to escape these costs as soon as possible either as the current account of the BoP improves or as capital inflows materialize. The attempt to escape however, may be premature in terms of economic fundamentals, with the result that economic performance deteriorates, private capital flows evaporate and the government is forced to tum again to the IME The puzzle is how to break away from this vicious circle of recidivism towards a virtuous one which offers graduation from the Fund. A robust finding across all studies is that the demand for Fund assistance is inversely related to economic growth and development. It is by encouraging economic growth that countries can break away from the habitual use of Fund resources. Fund-supported programmes need, therefore, to have a strong growth orientation. KS capture the growth motivation for borrowing from the Fund by including the investment ratio in the demand side of their model, and suggest that its empirical significance, 'is consistent with the view that the ultimate reason for seeking a Fund arrangement is to improve a country' s longer-term investment and growth prospects' (Knight and Santaella, 1994: 24 ). But do Fund arrangements have this effect? Studies that have concentrated on the near-term effects of Fund programmes have generally concluded that they have little effect on economic growth one way or the other, but that they may have an adverse short-term effect. More recent research (including the studies by Conway, 1994 and by Killick et al., 1992) has tried to allow for lagged effects and has discovered a positive effect after about three years, which is interestingly also the period of time after which Conway finds a negative effect of current programmes on the

40

Borrowing from the IMF

probability of future programmes. This lagged effect on growth may be consistent with a number of explanations. First, the Fund claims that it is by encouraging macroeconomic stabilization that its programmes create the necessary prerequisites for economic growth, and might therefore argue that the evidence now supports this. The difficulty here is that the empirical research on the effects of Fund-backed programmes finds little evidence that they induce macroeconomic stability. Second, to the extent that Fund programmes are associated with policies that reduce aggregate demand relative to aggregate supply and create spare productive capacity rather than reduce inflation, the relaxation of such policies would be expected to result in a temporary increase in the rate of economic growth as this capacity is brought back into use. But where this is the case, there is little reason to believe that the increased rate of economic growth will be sustained unless investment increases. What does seem to be well established empirically, however, is that Fund programmes are associated with a significant decline in investment ratios (Edwards, 1989; Khan, 1990; Killick et al., 1992; Conway, 1994). From a theoretical point of view this is unsurprising. If the purpose of a programme is to strengthen the BoP in the near term through the management of aggregate demand, then the brunt of adjustment will fall on domestic expenditure, of which investment is a component. Indeed short-term political considerations are likely to lead governments to favour policies which reduce investment in preference to consumption and government current expenditure. Similarly, since BoP deficits reflect an excess of domestic investment over domestic saving, unless saving can be increased, payments adjustment will require a decline in investment. But while for a time the adverse effects of falling investment may, to some extent, be offset by increasing the productivity of capital, it remains of concern that investment should fall in circumstances where a key reason for seeking Fund assistance is to increase the rate of economic growth. The apparent dilemma may be overcome by downplaying short-term aggregate demand management, and by extending the time frame of programmes to accommodate an aggregate supply-side approach. This was in part the purpose behind the Fund's move towards structural adjustment lending during the 1980s, but, as KS's empirical evidence implies, the changes in practice have been minimal, if not indiscernible. Research by Khan and Knight ( 1985) has shown how the adverse growth effects of Fund-backed programmes may be reduced by changing their orientation towards supply-side measures. A shift in emphasis of this type would, however, put greater strains on the financing capacity of the Fund which would need to respond either by being more selective in terms of those countries that it supported- effective selectivity would incidentally reduce the degree of recidivismor by securing increases in its own resources. There is evidence that an increase in Fund financing per programme would also encourage greater commitment to implement programmes, with Killick et al. (1992) finding that as the amount of finance provided by the Fund, expressed as a percentage of the initial BoP, increased, so also the chances of completion increased. If this strategy were to be

Borrowing from the IMF 41 successful, the increased demand for Fund resources would only be temporary, as enhanced economic performance would reduce the probability that countries would demand financial support from the Fund in the future. This is an area where success breeds success. Just as it is poor economic performance that traps countries into returning to the Fund again and again, and prevents them from securing financial support from private international capital markets, so economic success would increase their access to private capital markets and would, as a consequence, reduce their demand for resources from the IMF. Until Fund-backed programmes can be made more effective, it is unlikely that they will generate the catalytic effect that the Fund claims to exist but which empirical studies have generally failed to identify (Bird, 1995).

Concluding remarks At the same time as the future of the Fund is being widely debated, a number of relevant and important empirical studies have been published which have advanced our knowledge concerning the factors which determine IMF lending and the effects of Fund-backed programmes. There is little indication however, that this empirical evidence is being used to inform the policy debate. For as long as Fund-backed programmes fail to effectively encourage economic growth as a top priority, many developing countries will remain Fund recidivists and the Fund's resources will not be temporary and revolving as required by its Articles of Agreement. The attempt to reduce recidivism and encourage graduation from the Fund requires a fundamental rethinking of its role in developing countries. This does not imply a departure from the basic idea of the Fund as a BoP agency, but it does imply a stronger commitment to viewing the BoP in a longer term perspective, acknowledging in the design of programmes that the long-term sustainability of the BoP depends upon success in raising the rate of economic growth. Urging the Fund, as the 07 is doing, to concentrate on its 'core concern' of macroeconomic stability and therefore implicitly encouraging it to move away from structural adjustment threatens the small and hard-won improvements in the design of IMF conditionality that have been achieved over recent years. In a situation where the World Bank is not anxious to expand structural adjustment lending, greater emphasis on short-term macroeconomic stabilization through managing aggregate demand will mean that Fund programmes continue to have negative effects on investment, and will do little to help secure the enhanced growth performance which will enable developing countries to reduce their reliance on the Fund for financial support in the future.

Notes 1 Useful surveys are Khan (1990), Killick et al. (1992) and Killick with Malik (1992). 2 Bird (1995) overviews the evidence on this. Goreux (1989) shows how low-income countries in particular find it difficult to disengage themselves from the Fund, having

42 Borrowing from the IMF once turned to it for financial assistance. Some low-income countries have had outstanding credit from the Fund for almost 30 consecutive years. Indeed, Goreux discovered that 21 countries had credit outstanding from the Fund for in excess of 14 years. Such continuous Fund involvement clearly challenges one of the basic tenets of its operations. But is prolonged financial support by the Fund limited to low-income countries? An analysis of the number and time duration of IMF standby arrangements during 1985-90 shows that during this period 47 countries negotiated standbys. Of these, 25 had only one standby, but the remaining 22 had at least two and sometimes three. In some of the cases where only one standby was arranged, drawings on the Fund were also made under other facilities. Mexico, for example, while having only one standby covering the end of 1986, 1987 and the beginning of 1988, also had extended arrangements which covered 1984-5 and 1989-92. Mexico was therefore involved with the Fund throughout almost the entire period. Indeed, of the 25 countries that had only one standby, five also drew resources from the IMF under its other facilities. Of the ten countries that had three standby arrangements during 1985-90, six were lowincome countries, but four were not. This evidence seems to confirm that the image of the Fund coming into a country, offering swift financial support, helping to turn the BoP around, and then getting out, is purely and simply wrong. Moreover, a relationship that spans a number of years is not purely a feature of the Fund's dealings with low-income countries; it also applies to better-off developing countries. Further evidence of a long run and effectively ongoing involvement by the Fund in many developing countries is presented by Killick with Malik (1992), with this being a feature of its dealings with about 70 per cent of those developing countries with which it agreed programmes over 1979-89. In the KS study, the existence of a previous Fund arrangement is included as a dummy variable, and emerges from their estimation with a positive coefficient of 0.7755, a standard error of0.120 and at-ratio of 5.49. 3 Earlier studies of the causes of BoP deficits in part of the period covered by KS also discovered adverse movements in the terms of trade to be significant (Khan and Knight, 1983). 4 KS acknowledge this difficulty but understate it by noting that 'to the extent that these actual values differ from programme targets some estimation problems may arise' (Knight and Santaella, 1994: 20). 5 An interesting issue which KS do not raise is the extent to which the Fund's willingness to make an arrangement depends on a country's previous track record in implementing agreed programmes.

References Bird, Graham (1994) 'The Myths and Realities of IMF Lending', The World Economy, 17 (September). Bird, Graham (1995) IMF Lending to Developing Countries: Issues and Evidence (London: Routledge). Bird, Graham and Timothy Orme (1981) 'An Analysis of Drawings on the IMF by Developing Countries', World Development, 9(6), 563-8. Conway, Patrick (1994) 'IMF lending programs: Participation and Impact', Journal of Development Economics, 45(2)(December). Cornelius, Peter (1987) 'The Demand for IMF Credits by Sub-Saharan African Countries', Economics Letters, 23. Edwards, Sebastian (1989) 'The IMF and Developing Countries: A Critical Analysis', Carnegie-Rochester Conference Series on Public Policy, no. 31 (Washington, DC: N. Holland).

Borrowing from the IMP

43

Finch, David C. (1989) The IMF: The Record and the Prospect, Essays in International Finance, no. 175 (Princeton: Princeton University) September. Goreux, Louis M. (1989) 'The Fund and the Low-income Countries', in Catherine Gwin and Richard E. Feinberg (eds) The International Monetary Fund in a Multipolar World, US-Third World Policy Perspectives, no. 13 (Washington, DC: Overseas Development Council). Joyce, Joseph P. (1992) 'The Economic Characteristics of IMF Program Countries', Economics Letters, 38. Khan, Mohsin S. (1990) 'The Macroeconomic Effects of Fund-supported Adjustment Programs', IMF Staff Papers, 37 (June). Khan, Mohsin S. and Malcolm Knight (1983) 'Determinants of Current Account Balances of Non-oil Developing Countries in the 1970s: An Empirical Analysis', IMF Staff Papers (December). Khan, Mohsin S. and Malcolm Knight (1985) 'Fund Supported Adjustment Programs and Economic Growth', IMF Occasional Paper, no. 41 (Washington, DC: IMF) November. Killick, Tony ( 199 5) IMF Programmes in Developing Countries: Design and Impact (London: Routledge). Killick, Tony with Moazzam Malik (1992) 'Country Experiences with IMF Programmes in the 1980s', The World Economy, 15 (September). Killick, Tony, Moazzam Malik and Marcus Manuel (1992) 'What Can We know About the Effects of IMF Programmes?', The World Economy, 15 (September). Knight, Malcolm and Julio A. Santaella (1994) 'Economic Determinants of Fund Financial Arrangements', IMF Working Paper, WP/94/36 (Washington, DC: IMF) March. Rowlands, Dane (1994) 'Country Conditions and Conditional Lending: IMF Behaviour after Bretton Woods', mimeographed (Ottawa: Carleton University) February. Santaella, Julio A. (1995) 'Four Decades of Fund Arrangements: Macroeconomic Stylized Facts before the Adjustment Program', IMF Working Paper, WP/95/94 (Washington, DC: IMF). Vaubel, Roland (1991) 'The Political Economy of the International Monetary Fund: A Public Choice Analysis', in R. Vaubel and T.D. Willett (eds) The Political Economy of International Organizations (Boulder, CO: Westview Press).

3 IMF lending How is it affected by economic, political and institutional factors?

Introduction The International Monetary Fund (IMF) is the pre-eminent international financial institution. One way or another its activities influence most countries in the world, even those that do not directly draw resources from it. As explained in its Articles of Agreement, the stated purposes of the Fund are framed in terms of promoting international economic cooperation, expansion and stability. With its resources coming primarily from members' subscriptions, the Fund provides financial support to countries in balance of payments need, usually in the context of agreed adjustment programmes which are designed to strengthen the balance of payments. This confers a great deal of power and influence upon the Fund. Over the years, however, questions have been raised about whether or not its decisions are based purely on positive economics. The purpose of this chapter is to test for the presence of political or other bias in the distribution of IMF programmes. The grounds for suspecting bias are strong. Certainly the Governors of the Fund are politicians, and the Executive Directors represent constituencies that have different political interests. It is, therefore, difficult to sustain the argument that the Fund is non-political. But does the political dimension filter through in any systematic fashion into its lending decisions? Is it the case that member countries get treated differently for political reasons, or because they embody other non-economic characteristics that repel or attract the IMP's attention? There is quite a lot of anecdotal material which claims that politics pervades IMF lending, but there has been relatively little systematic analysis of the evidence. The political dimension of IMF lending received renewed attention in the context of large loans to Mexico, Russia and countries in East Asia during the mid-1990s. In a recent study sponsored by the US Congress - the Meltzer Commission (2000)- the criticism is implicitly accepted that '07 governments, particularly the United States, use the IMF as a vehicle to achieve their own political ends', with the recommendation being made that the IMF should not be used as a 'slush fund' to 'satisfy decisions of the 07 finance ministers or other groups of powerful members'. Developing the argument more fully, IMF lending is presented as a way in which the US Treasury attempts to circumvent the

IMP lending 45 wishes of the US Congress. In contrast, some recent studies imply that IMF lending can be satisfactorily explained purely in terms of economics. Which of these arguments is supported by the data? The question of what factors affect the pattern of IMF agreements is an important one. The presence of political or other bias in its lending decisions may undermine the Fund's financial reputation and its capacity to provide clear signals to private markets. The chapter is organized along the following lines. The section 'Concepts and issues: an analytical framework' analyses the way in which political economy considerations may, in principle, exert an influence over IMF arrangements, distinguishing between the various stages at which this influence may be felt. The section also briefly discusses some of the anecdotal evidence. The section 'Review of the empirical literature' goes on to review the systematic evidence regarding the factors that affect IMF lending agreements. The section 'Model specification and estimation' provides and interprets new empirical evidence that more directly attempts to capture some of the political economy factors identified in the second section, as well as including the economic factors emphasized in earlier studies. It thereby attempts to provide a fuller explanation of IMF lending than currently exists. The last section offers a few concluding remarks.

Concepts and issues: an analytical framework IMF arrangements come about as a consequence of a member country first approaching the Fund for financial assistance, and second being able to negotiate a programme of policies with the Fund that is then approved by its Executive Board. In principle, politics could be relevant at each of these stages. The decision to turn to the Fund is a political decision. Some governments will, for political reasons, wish to avoid the Fund at almost all costs. Other governments, on the other hand, may see political benefits from involving the Fund. Once an application has been made, the negotiation and approval of programmes also have political elements. All economic adjustment programmes have distributional consequences and distributional issues lie at the heart of politics. On top of this, the Fund's management may have its own institutional agenda. Furthermore, the Fund's major shareholders, who can exert a disproportionate influence on Executive Board decisions, may be more or less inclined to validate IMF support for some countries than for others. Politics also enters the story once a programme has been negotiated since its effects will have a political dimension, and politics will also influence whether the programme is fully implemented. Indeed the Fund often explains the relatively poor implementation record of Fund-backed programmes in terms of a 'lack of political will'. There is already a substantial literature that examines the political economy of policy reform (Williamson, 1993; Rodrik, 1996) and this chapter therefore focuses more narrowly on the political economy of agreeing to a Fund arrangement in the first place rather than the extent to which the agreed programme is

46 IMF lending implemented. The focus is on the three elements identified above, the role of politics in the demand for and supply of IMF loans and institutional influences on IMF lending. Although there may be political factors influencing both the demand and supply of IMF loans, these are difficult to isolate in practice since what is observed is simply an outcome, in the form of either an IMF programme or no IMF programme. Moreover, there are inherent methodological difficulties in distinguishing between demand and supply factors, as well as between economic and political factors. To observe, for example, that the IMF does not make loans to governments of a particular complexion does not tell us whether it is these governments that choose not to borrow from the IMF or the IMF that chooses not to lend to them. The outcome is the same in either case. To observe that political instability in member countries influences IMF lending to them would not tell us the causes of the political instability. It may be the consequence of economic instability. Our objective in this chapter is more modest. We do not attempt to build a detailed political-economy model of IMF lending. Instead we set out to see whether our ability to explain IMF lending can be improved by systematically adding quasi-political variables to the list of economic variables that has traditionally been used. Does the inclusion of these variables help us statistically to explain IMF lending? And if it does, is there a reasonable explanation that we can offer as to how these variables are exercizing their influence? Apart from the economic situation in which they find themselves and the availability of credit from elsewhere, a country's willingness to tum to the IMF for help and agree a programme may be expected to depend on the perceived loss of national sovereignty associated with IMF conditionality and the size of the gap between the government's preferred policies and those associated with IMF conditionality. Other factors may include the extent to which referral, and the apparent failure it reflects, can be blamed on the previous administration, the extent to which the final programme is expected to lead to political unrest and the extent to which the costs of adjustment can be blamed on the IMF. Disagreement may, of course, exist within governments, with finance ministries favouring involving the Fund while spending ministers oppose it. On the supply side, the institutional structure of the IMF including the composition of its Executive Board, the European nationality of the Managing Director, the distribution of quotas and the need for super majorities in the case of some decisions certainly allows the industrialized countries to be the most influential members and potentially to use IMF lending to serve their own interests. But what are these interests and how might IMF lending be modified to serve them? The 'interests' may be quite specific. Industrialized countries may be expected to favour developing countries that are strategically important, or in which industrialized country banks or pension and mutual funds or multinational companies have a substantial financial involvement. Alternatively, they may be more general, in terms of the preferences of richer countries for 'good governance' and a liberal economic system.

IMF lending 4 7 In seeking to serve these interests one mechanism would work on the basis of rewards and penalties. Developing countries with political and economic systems favoured by industrial countries and in which the latter had direct political and economic interests would be rewarded by receiving larger loans with less conditionality than their economic situation taken alone would suggest. On the other hand, it would be possible for the developing countries with undemocratic political systems, high levels of corruption, and poor civil liberties and human rights, as well as those of little political and economic significance, to be penalized by receiving smaller loans and more conditionality. Fund programmes, of course, could also be used to bribe countries into pursuing policies deemed more desirable by key stakeholders. In this case poor performers would be offered IMF loans as an incentive to reform. To acknowledge that politics will influence the lending behaviour of the Fund does not therefore automatically imply exactly how this political influence will be exercized. There could be a range of political models, which generate different and opposing outcomes. Sophisticated political models indicate that the identification of bias is very complex. In a comprehensive study of IMF and World Bank relations with Soviet bloc countries, Assetto (1988) concludes that there is likely to have been a political dimension to the treatment of these countries by the Bank and Fund, though it was favourable in nature and probably minor in magnitude. Both institutions appeared reluctant to punish the East European countries, and seemed eager to support any reforms which they voluntarily undertook. Their relatively favourable treatment may have occurred as part of an effort to proactively integrate them into the international economy, promote market oriented reform, and possibly even disrupt relations within the Soviet bloc. Stiles (1990) also presents evidence to suggest that IMF conditionality may be varied according to the political influence that the borrowing country can exert with the Fund's principal shareholders. An understanding and analysis of how political preferences are manifested in Fund activities clearly requires a degree of nuance. 1 Beyond this, does the IMF have an institutional identity that may have contributed to the evolution of objectives separate and distinct from those of its member states? Two institutional biases which may exert an influence on IMF lending are considered here. While the IMF and World Bank have consistently proclaimed themselves as being apolitical institutions, there is little doubt that the standard of international behaviour embedded in the Fund's mandate reflects classical liberal principles (Fleming, 1964: 8-9). Consequently there are certain country attributes which - being favourable to the requirements of orthodox economic development - may be more acceptable to the IMF. However, a simple example highlights the problem of separating the views of Fund staff from shareholder power. The political bias of the Fund may well include preferences regarding the broad economic orientation of potential loan recipients. Governments with a 'socialist' perspective may be viewed less favourably by the Fund because staff perceive the policy environment in the state

48 IMP lending to be fundamentally hostile to sound (i.e. orthodox, liberal, market-based) economic development. Apparent political bias may, therefore, have its source in internal Fund culture rather than external pressure or direct voting behaviour from countries such as the United States. The propriety of such biases is also dependent on context. For example, an unsavoury dictatorial regime may be well-positioned to implement and enforce the economic adjustment policies accompanying Fund agreements. Therefore there may be, or may have been, a tendency to favour such countries because of the presumption of better implementation. However, good governance is now receiving growing recognition as a prerequisite for sustainable development. So, while the refusal to grant a loan to a country with a poor record on civil liberties may be interpreted as a violation of the Fund's announced political neutrality, it would be quite reasonable to reject such a loan request if the capacity of the country to implement an agreement were compromised by poor governance, including restrictive policies on civil liberties. In recent years, the Fund and Bank have both acknowledged the importance of good governance in their rhetoric, though whether this has emerged as a guiding factor in determining loan allocations remains an empirical issue, and one to which we return later. The second broad institutional perspective is derived from the public choice approach to institutional behaviour. Vaubel (1994, 1996) presents evidence of increased ('hurry-up') lending immediately prior to regular quota reviews. Thus, there is a clear suggestion that there are temporal issues (cycles and trends), which may affect the overall propensity of the Fund to enter into agreements. The public choice approach also identifies some country-specific characteristics, which may be associated with IMF loan frequencies. First, the distribution of agreements across countries may be affected by institutional structures. For example, there may be a bias towards lending to countries with larger quotas to reduce the ratio of resource costs to loan amounts. Similarly there may be regional differentiation resulting from lending decisions being made within partially isolated geographic departments. A focus on the IMF as an institution also suggests that client-patron relations are likely to develop between specific countries (the clients) and the Fund (the patron). Officials in the Fund and their counterparts within the member country government need to work fairly closely with one another. Each side has an incentive to invest in this relationship in order to ensure that it proceeds as smoothly as possible. The position and promotion prospects of government officials may depend on being able to deliver both economic reform and access to financing. The failure of an IMF programme may mean more to them than the end of policy reform, especially if the Fund is seen as a key external ally for reform-minded elements in an administration. On the other hand, a country's failure to implement a programme successfully may also reflect badly on associated Fund personnel, giving the latter an incentive to maintain clients and attribute programme failures to inherently unpredictable factors beyond the control of either themselves or their counterparts in a country's administration. In this case bureaucratic imperatives predict a high rate of programme repetition.

IMF lending 49 Even countries, which have successfully completed a programme and gone a long way towards correcting their balance of payments deficits may negotiate additional agreements, as both Fund and government officials have an incentive to maintain a successful relationship. 2 The analysis in this section has shown how political and institutional considerations may influence both the demand for and the supply of Fund arrangements. Although anecdotal and specific case study evidence of a political bias in Fund lending is available, it remains controversial. Little analysis of systematic political bias has been undertaken, and most studies of IMP lending continue to emphasize economic factors almost exclusively. As a consequence, little insight has been provided on issues of governance, or the institutional side of political economy explanations. The next section briefly reviews the existing largesample studies, focusing on their indirect implications for the existence of a political dimension in IMP lending, and assessing how well they have performed in estimating the pattern of lending.

Review of the empirical literature Given the amount of attention paid to the IMP, the literature on the determinants of IMP lending is remarkably limited. Bird and Orme (1981), Cornelius (1987), Joyce (1992), Doroodian (1993), Conway (1994), Rowlands (1995), Bird (1995, 1996), Knight and Santaella (1997), and Thacker (1999) are the principal studies. Essentially two methodologies have been used. The first attempts to estimate the demand for Fund credit, while the second analyses involvement in Fund arrangements in the context of a binary choice model. The literature is reviewed in some detail in Bird (1995). There is broad consensus that first, the Fund is a more important source of finance for relatively poor countries; the demand for IMP credit is inversely related to economic growth and development. Second, the chances of current involvement with the Fund are positively related to previous near term involvement; there is strong Fund recidivism. Third, the availability of non-Fund sources of external finance is, at an aggregate level, not significantly associated with IMP lending, with the sign of the coefficient varying across studies (Bird and Rowlands, 1997). Fourth, exchange rate over-valuation significantly and positively affects the use of Fund resources, and fifth, international reserve holdings generally seem to have a negative effect on the demand for Fund resources. A consensus also exists that, while all countries turning to the Fund of necessity have a balance of payments need, neither a current account nor an overall balance of payments deficit provides sufficient motivation for seeking Fund support. Outside these factors, there is more dissension within the existing literature (Bird, 1996). A feature of many of the early studies is their poor overall explanatory power (Bird and Orme, 1981; Cornelius, 1987; Joyce, 1992). Some estimations leave unexplained as much as 75 per cent of the variations in drawings on the Fund. Bird (1995) notes the 'resilience of the low explanatory power of the estimated

50 IMP leruling equations, arguing that "something important" is being omitted'. He presents the 'remaining challenge' as being 'to explain the residual variation which is left unexplained by economic characteristics'. (Bird, 1995: 149). Some aspects of the empirical evidence hint that it may be the political dimension of Fund lending that is being omitted. Thus, Bird and Orme (1981) find that their model based on economic characteristics generates false positives, and they interpret this as being consistent with the idea that there are political costs, which reduce the demand for IMF loans below what the economics would suggest. 3 The recidivism identified in many studies (Conway, 1994; Bird, 1995; Knight and Santaella, 1997) is consistent with a political threshold model where having once met the political costs of turning to the Fund, the marginal cost of further referral falls. Joyce (1992) finds that the ratio of government expenditure to GDP and the growth of domestic credit are positively related to the presence of an IMF agreement. He interprets these results as indicating a greater need to tum to the IMF in an economic downturn. However, in terms of political bias his results imply that more expansionary governments are more likely to end up with an IMF arrangement, suggesting that country need dominates any reluctance on the part of either the Fund or the government to deal with one another. The study by Edwards and Santaella (1993) extends the analysis of country characteristics associated with Fund programmes to include several political variables. Their results provide weak evidence that coup frequency and dictatorial regimes are positively associated with Fund agreements. However, political popularity, political violence and ideology are all found to be statistically insignificant. Unfortunately, a drawback of their analysis is its focus on a relatively small sample of countries, which experienced devaluation episodes between 1954 and 1971. Related studies that deal with devaluation in developing economies, (Klein and Marion, 1997), and with exchange rate fluctuations more generally (Block and Hess, 1997), find that political factors, such as the length of incumbency of the current government, are significant, but they do not relate the exchange rate changes to the existence of IMF programmes. The studies by Conway ( 1994) and Knight and Santaella ( 1997) focus on participation in IMF arrangements but concentrate on economic determinants. Both studies find evidence of inertia in country-Fund relations. As noted in the previous section, this could have a political as well as an economic explanation, although they do not interpret it in this way. Although not beyond criticism methodologically (Bird, 1997), both these studies appear to have a high degree of explanatory power and, therefore, imply that correctly specified economic variables can satisfactorily explain involvement with the Fund. In effect, therefore, they find it unnecessary to include political variables. Of the very few large sample studies that explicitly include political variables, Rowlands (1995) discovers ambiguous results for variables with a political interpretation. With respect to geographic bias, dummy variables indicate some differentiation between regions, with developing countries in Asia and the Western Hemisphere more likely to have an IMF agreement than those in Africa or Europe and the Middle East. Civil and political freedoms in the recipient

IMF lending 51 country seem to have no effect on the allocation of Fund agreements, though there is some weak evidence that the strategic interests of the United States (measured by military loans) has an effect on Fund agreements. Thacker (1999) finds evidence to support the US influence hypothesis, with political proximity to the United States and political movement towards the United States being measured by referring to voting patterns at the UN General Assembly. While political proximity appears statistically significant in the post Cold War period only, movement towards greater congruence with the voting behaviour of the United States appears statistically important for both the Cold War and post Cold War periods. Furthermore, Thacker's sample period is one in which many countries embarked on economic and political reform simultaneously, often in conjunction with IMF and World Bank programmes. Sorting out the direction of causality, or attributing the correlation to overt US influence, is therefore problematic. For example, Thacker concludes that Fund activities have become more strongly influenced by US interests since the end of the Cold War. This result contradicts Killick's suggestion that the end of the Cold War should have liberated institutions such as the Fund from political interference (Killick, 1995). However, some caution is needed in interpreting this result due to the heavy involvement of the IMF in the transition economies. While politics may well play a part in the Fund's dealings with these economies, and may even be central in some cases such as Russia, they also represent a special case in economic terms. Despite these caveats, Thacker's paper provides the best example to date of integrating political and economic variables into the analysis of Fund lending, and provides evidence of a political dimension to the process. One of the interesting questions about these studies is their capacity to correctly predict when countries will, or will not, enter into an agreement with the IME Among the econometric studies comparable to the approach in this chapter, Conway (1994) claims a high accuracy ratio for the 1976-86 period, suggesting that up to 90 per cent of the variation is explained in his model. Unfortunately the correct prediction percentage is not reported. Knight and Santaella (1997) report an overall accuracy rate of 82.3 per cent over the sample period 1973-91. Rowlands (1995) correctly predicts 86.7 per cent of the cases in the 1973-89 period by allowing some coefficient flexibility between the pre- and post-1982 debt crisis sub-periods. Thacker's ( 1999) best results are nearly 85 per cent for the 1985-94 period, with sub-period accuracy rates of just 81 per cent and 88 per cent, respectively, for 1985-9 and 1990-4. While these rates of prediction may appear impressive, it should be noted that over most of these sample periods, only about 20 per cent of countries will actually have an agreement. Therefore a guess of 'no agreement' will be accurate 80 per cent of the time. While adding economic and other variables seems to have improved on this somewhat (and undoubtedly are able to provide greater assurance about the accuracy of prediction) there still appears to be a discouragingly high proportion of cases that defy prediction. Will models of IMF lending that incorporate in a systematic fashion the other political economy variables, such as good governance or institutional preferences,

52 IMF lending improve predictive capacity? The model outlined in the next section is structured to test for these influences as well and to provide a more comprehensive explanation of IMF lending.

Model specification and estimation The principal purpose of this chapter is to discover whether the ability to explain IMF arrangements can be systematically improved by including an array of political and institutional variables. Modelling economic factors is difficult enough, but at least operational measures of the presumed determining factors are reasonably straightforward. Not so in the case of political factors, where broad and imperfect proxies often have to be used. Results based on such estimations, therefore, need to be treated with caution and viewed as indicative rather than definitive. As already noted, methodologically it is difficult to distinguish between demand- and supply-side political factors since what is observed is the outcome of the interaction between them in the form of either a programme or no programme; demand- and supply-side factors jointly determine this outcome. In any case, there is ambiguity about whether individual political variables are best thought of as being related to the demand-side or the supply-side. Will high levels of corruption affect a government's propensity to try to borrow ftom the Fund or the Fund's proclivity to lend? Moreover the expectation of supply-side resistance may influence the demand for IMF loans. The dependent variable for our analysis is a binary indicator of whether or not a country signed a high conditionality StandBy Arrangement (SBA), Extended Fund Facility agreement (EFF) or Enhanced Structural Adjustment Facility (ESAF) agreement in a certain calendar year. Countries that were not eligible to sign such an agreement were excluded from the sample. 4 The independent variables were chosen to reflect both the theoretical priors, as well as the results of earlier studies. In almost all cases these variables were lagged by one year to improve confidence in assigning causality. In terms of economic variables, the right-hand side of the model includes per capita income, with the presumption that wealthier countries are less likely to sign an agreement with the Fund because of their access to private financing. Economic growth is also expected to have a negative coefficient, indicating that countries with good recent growth performance will be less likely to need IMF assistance. Of course rapid economic growth may lead to balance of payments problems (which is controlled for), and the Fund may itself favour poorer and slower growing countries, a result consistent with a needs-based perspective of its operations. The other economic variables included reflect each country's international economic characteristics. Improvements in the ratio of reserves to imports, the percentage increase in reserves, and the current account balance-to-GDP are expected to diminish the likelihood of signing an IMF agreement, and therefore their expected coefficient estimates are negative. A variable measuring the change in the real exchange rate over the past three years is also included. Real

IMF lending 53 exchange rate apprec1atton is generally associated with a worsening of the current account, and may trigger IMF intervention, but continued depreciation over an extended period may be symptomatic of structural balance of payments problems. The sign of the coefficient is therefore ambiguous. A higher debt-service payments-to-exports ratio is expected to increase the chances of needing an IMF programme. The change in the debt-service ratio is included as a second variable to capture the potential effect of a debt-service shock, which should also have a positive correlation with IMF agreements. Other debt variables include the debt-to-GDP ratio, though since the burden of debt is already reflected to a large degree in the debt-service ratio, a high debtto-GOP ratio may not be correlated with the need for IMF agreements. Instead it may indicate the ability to access other sources of lending, leading to a negative correlation with IMF programme presence. The last four economic variables that are included measure country-specific and global measures of potential debt problems. An increase in the arrears-todebt level should be positively correlated with IMF programmes due to their signalling of repayment difficulties and the likely need for rescheduling. 5 A history of debt reschedulings may indicate the propensity to require Fund programmes in support of the rescheduling process. At the global level, a high rate of real international interest rates (real LIBOR), or an increase in the rate, may provoke higher debt servicing problems that require subsequent Fund assistance. Therefore, a measure of both the level and change in real LIBOR is included. A second version of the model is estimated with an additional, forward looking variable to reflect future debt reschedulings. The rationale for its inclusion is that the other debt variables may simply be insufficient for accurately predicting the incidence of debt rescheduling. 6 Since an IMF agreement is almost always a prerequisite for debt reschedulings at either the Paris Club or London Club, a government's desire for debt rescheduling in the near future will require them to first sign an agreement with the Fund. Therefore, this variable is included in one model to see if the anticipation of rescheduling helps to explain the pattern of IMF agreements. Since the dependent variable in our model is binary, a probit model is used to estimate the basic equation. 7 The sample is drawn from a pooled time-series cross-sectional data set. Data availability results in an unbalanced panel of 1041 observations that cover the years 1974-94, and includes 95 different countries. To ensure the comparability of the results, the sample is kept uniform for all models regardless of the variables included in the estimation. The results of the basic estimation and the version with the forward looking rescheduling variable are provided in Table 3.1. In general the model without the future rescheduling variable performs reasonably well. The income, growth and balance of payments variables have the correct negative sign for the estimated coefficient, though only the first of these has a statistically significant estimated coefficient, and only weakly so. The coefficient estimates are strongly significant and correctly signed for the reserves-toimports, reserve change and debt-service-to-exports ratios. The positive and

54 IMF lending Table 3.1 Estimation results for the basic models Variables

Constant only

Basic modell

Basic model 2

Constant

-0.909*** (-20.08)

-0.851*** (-5.75)

-0.794*** ( -5.31)

-0.0802* ( -1.76) -0.00103 ( -0.27) -0.930*** (- 3.93) -0.225*** (- 2.36) -0.0108 ( -1.46) 0.301 *** (2.71)

-0.101** (-2.13) -0.00186 ( -0.49) -0.990*** (-4.14) -0.218** ( -2.27) -0.0124* ( -1.66) 0.77*** (2.58)

1.50*** (3.98) -0.000366 ( -0.98)

1.41 *** (3. 72) -0.000295 ( -0.80)

-0.254* ( -1.78) -0.0191 ( -0.03) 0.387*** (5.42) 0.0231 ( -1.09) 0.0745*** (2.97)

-0.257* ( -1.83) -0.485 (-0.72) 0.338*** (4.64) -0.0279 ( -1.31) 0.0759*** (3.02) 2.49*** (3.54) 1041 82.22 856 189 0.129 -428.2

GNP per capita GDP growth Reserves/imports Change in reserves Current account/GOP Real exchange rate change Debt-service ratio Change in debt-service ratio Debt/GDP Arrears/debt Past reschedulings Real LIBOR Change in real LIBOR Imminent rescheduling Number of observations % correct predictions # of correct predictions # of positive observations R-squared Log Likelihood

1041 81.84 852 189 -442.9

1041 82.71 861 189 0.116 -434.4

Notes: The t-statistics appear in parentheses. Coefficient estimates marked *, ** and *** are statistically significant at the 0.05, 0.025 and 0.01 levels of confidence for a one-tailed test. The dependent variable indicates a signing of a SBA, EFF or ESAF arrangement.

significant coefficient estimate for the accumulated real exchange rate is consistent with the explanation that longer term real depreciation is a symptom of economic difficulties, which increase the probability of IMF agreements. There is no statistically significant effect of the change in the debt-service-to-exports ratio, while the coefficient estimate for the total-debt-to-GDP ratio is negative, but only weakly significant. A history of frequent rescheduling, and an increase in the LIBOR rate are both positively and significantly associated with subsequent IMF agreements, while the LIBOR rate itself and the accumulation of arrears do not have statistically significant estimated coefficients. The model with the future rescheduling variable has similar results, with slightly higher levels of significance for some of the estimated coefficients for explanatory variables such as per capita GNP and the current account balance. In addition, the future rescheduling variable has the expected positive coefficient, which is statistically very significant. Likelihood ratio tests indicate that the two models are statistically distinct, and that restricting the coefficient on future reschedulings to be zero, implicitly the case in the basic model, results in a statistically inferior model. The basic equation used here is comparable to most of the other large-sample equations in terms of accuracy of prediction. In this case the percentage of

IMF lending

55

correct estimations is 82.7 per cent. It should be noted that a model that uses only a constant and no other explanatory variables is accurate in 81.8 per cent of the cases in the sample. An interesting feature of these non-linear estimations, however, is that adding explanatory variables does not necessarily increase the predictive success of the model, even when the sample is held constant and the added variable itself has a statistically significant coefficient estimate. In the estimation with the future rescheduling variable, the correct prediction percentage drops to 82.2, implying that the more parsimonious model has a net increase of 5 of the 1041 cases being predicted more accurately than the more extensive model. How does the inclusion of political and institutional variables affect these results? To investigate political and institutional influences on Fund agreements three additional sets of variables are added to the basic economic model. The first set of variables is chosen partly to take into account the economic interests of key shareholders and comprises US exports to the country and French exports to African countries (where France is presumed to. have greater interests and influence). A second set of variables represent political conditions in potential recipient countries, and includes a variable indicating whether the country can be characterized as socialist, an indicator of whether the government had changed recently, the level and change in civil liberties and the number of coups in the preceding five years. On the one hand, some of these variables are going to reflect primarily the recipient government's willingness to accept an IMF agreement. For example, it has been argued that new governments are likely to be more capable of embarking on a structural adjustment or stabilization programme. The country's socialist orientation, degree of civil liberty and coup frequency, on the other hand, could reflect the preferences of either the recipient or the IMF and its key shareholders. The third set of variables incorporates country relations with the IMF as well as institutional characteristics. The first variable is the number of IMF agreements in the previous five years that had more than 20 per cent of the agreed amount undrawn at expiration. This measure is used by Killick et al. (1992) to identify incomplete or unsuccessful programmes. The next three variables indicate the presence of an imminent quota review (within the next two years), a measure of IMF liquidity (industrial country loans less outstanding loans and purchases, as a proportion of industrial country quotas) and the size of the country's real GDP, which reflects in part its relative economic importance in the world. These three sets of variables are added to the basic economic model discussed previously. In addition, a second version of the model includes three additional variables. One of these is the presence of future reschedulings, as discussed previously. The second is also forward looking; betng an indicator of a government change in the next year that is not a coup. These government changes are, presumably, ones that are more expected, and hence more likely to affect current government decisions. The expectation is that governments that anticipate their

56 IMP lending potential dismissal may be less willing to adopt a Fund programme that may further their unpopularity. The third variable is backward looking; being the number of months under a high conditionality IMF programme in the previous three years. This variable hopefully captures the institutional links that build up between the IMF staff and recipient country politicians and bureaucrats, as well as the fact that continuing IMF involvement may be easier once the initial programme has been accepted, and consequent political costs paid. Of course, the variable may also simply reflect the presence of recidivism, a common dependency that emerges in country-IMP relations. Table 3.2 presents the results for the additional variables. Of the additional non-economic variables, few seem to be statistically significant individually. In the more parsimonious of the two equations, the indicator of socialism has the expected negative sign, suggesting either that these states are less willing to accept the liberal economic programmes of the IMF, or that the IMF and its key shareholders are less inclined to deal with these countries. Worsening civil liberties also appear to be positively associated with signing of IMF programmes, though the estimated coefficient is only weakly significant. If accurate, this result may reflect a government's capacity to survive the political unpopularity presumed to accompany IMF programmes, thus increasing its propensity to sign such agreements. Finally, a history of incomplete IMF agreements seems to be positively and significantly related to IMF agreements. In the supplemented model, socialist governments and those with improved civil liberties remain less likely to sign an IMF agreement, with the estimated coefficient of the latter indicator being more significant, statistically. In addition, US export levels also have a significant coefficient estimate, though with an unexpected negative sign. The additional variables also appear to be statistically important. Future rescheduling again has a strongly significant coefficient estimate, as does the presence of past IMF programmes. The inclusion of this latter variable, which is positively related with current agreement signings, is accompanied by a substantial reduction in the statistical significance of the variable measuring the number of past incomplete agreements. These two variables are clearly related, but the results here suggest that it is the presence of the relationship, not the completion of programmes, which is most important. Finally, the variable indicating a 'normal' change in government in the near future has the expected negative sign, but its level of statistical significance is modest. By and large, the inclusion of these additional variables has little effect on the coefficient estimates of the economic variables, suggesting both robustness and an absence of multicollinearity. The inclusion of these additional variables, however, adds little to the ability to accurately predict the pattern of agreements. The most parsimonious economic model still predicts at least as accurately as any of the other models. For the models presented in Table 3.2, the more parsimonious of the two actually predicts accurately four fewer cases than the simplest economic model presented in Table 3.1. Likelihood ratio tests confirm that these different models are statistically distinct, and higher R-squared statistics accompany the models that use more

IMF lending 57 Table 3.2 Estimation results for the supplemented models Variables

Supplemented model I

Supplemented model 2

Constant GNP per capita GDP growth Reserves/imports Change in reserves Current account/GDP Real exchange rate change Debt-service ratio Change in debt-service ratio Debt/GDP Arrears/debt Past reschedulings Real LIBOR Change in real LIBOR US exports French exports (Africa) Socialist Recent government Civil freedom Change in civil freedom Coup frequency Past incomplete programs Imminent quota review IMF liquidity Gross real GDP Imminent rescheduling Imminent new government Past IMF agreements Number of observations % correct predictions # of correct predictions # of positive observations R-squared Log likelihood

-1.07*** (-3.14) -0.0258 ( -0.50) -0.00115 ( -0.30) -0.943*** (- 3. 70) -0.239*** (-2.43) -0.0146* ( -1.89) 0.303*** (2.75) 1.50*** (3.52) -0.000347 ( -0.87) -0.262* ( -1.70) -0.314 ( -0.46) 0.343*** (4.54) -0.0179 ( -0.71) 0.0851 *** (2.98) -0.00398 ( -1.27) -0.00836 ( -0.65) -0.935** (- 2.21) -0.278 ( -0.99) 0.0340 (0.90) 0.106921 * (1.78) 0.216 (1.21) 0.196*** (4.17) -0.00838 ( -0.06) -0.163 ( -0.36) 0.0834 (0.75)

-0.961514*** (-2.73) -0.0391 ( -0.74) -0.00263 ( -0.67) -0.904*** (- 3.46) -0.264*** (- 2.55) -0.0171** (-2.16) 0.234*** (2.89) 1.59*** (3.66) -0.000288 ( -0.72) -0.307** ( -1.98) -0.0508 ( -0.07) 0.186*** (2.31) -0.036 ( -1.39) 0.0929*** (3.16) -0.813** (- 2.10) -0.0113 (-0.84) -0.828** ( -1.99) -0.392 ( -1.34) 0.0151 (0.378) 0.122** (1.96) 0.326* ( 1.73) 0.0443 (0. 76) -0.0358 ( -0.27) -0.229 ( -0.49) 0.117 (1.03) 2.63*** (3.61) -0.234* (-1.74) 0.0411 *** (5.00) 1041 82.71 861 189 0.185 -394.4

1041 81.46 848 189 0.138 -416.5

Note: See notes to Table 3.1.

explanatory variables. Despite this, the additional variables do not seem to improve the predictive accuracy of the basic equation at all. Finally, the estimations are repeated for two sub-periods before 1990 and after 1989. The 1989-90 break is useful both because it represents the end of the Cold War, and coincides with the World Bank's identification of the importance of governance to development. The results of these estimations are presented in Table 3.3. The models do differ slightly between the full sample and the post-1989 subsample. In the latter, four variables, the indicators of socialism, coups, new governments and IMF liquidity, have to be omitted due to singularities in the data.

58 IMF lending Table 3.3 Estimation results for the sub-periods Variables

Supplemented model pre-1990

Supplemented model post-1989

Constant GNP per capita GDP growth Reserves/imports Change in reserves Current account/GDP Real exchange rate change Debt-service ratio Change in debt-service ratio Debt/GDP Arrears/debt Past reschedulings Real LIBOR Change in real LIBOR US exports French exports (Africa) Socialist Recent government Civil freedom Change in civil freedom Coup frequency Past incomplete programmes Imminent quota review IMF liquidity Gross real GDP Imminent rescheduling Imminent new government Past IMF agreements

-0.994*** ( -2.55) 0.00289 (0.04) 0.000843 (0.18) -0.762*** ( -2.61) -0.302** (- 2.30) -0.0219** ( -2.26) 0.143 (1.16) 1.82*** (3.30) 0.000256 (0.03) -0.397 ( -1.47) -0.227 ( -0.20) 0.209** (2.12) -0.0372 ( -1.31) 0.0871 *** (2.53) -0.00850*** (- 2.05) -0.0106 ( -0.68) -0.730* (-1.71) -0.397 ( -1.26) 0.000687 (0.02) 0.166** (2.23) 0.408** (2.04) 0.0245 (0.37) 0.133 (0.78) -0.517 ( -0.95) 0.0675 (0.43) 2.07*** (2.34) -0.160 ( -1.00) 0.0533*** (5.47) 795 82.01 652 147 0.191 -300.0

1.40 (1.26) -0.0827 ( -0.74) -0.00717 (-0.71) -1.91 *** (- 2.49) -0.324 ( -1.60) 0.0201 ( 1.04) 0.693 (1.60) 2.10** (2.20) -0.000600 ( -0.86) -0.0632 ( -0.221) 0.0437 (0.04) 0.271 (1.45) -0.475 ( -1.52) 0.532*** (2.96) -0.143 (-0.19) -0.0101 ( -0.34)

Number of observations o/o correct predictions # of correct predictions # of positive observations R-squared Log likelihood

-0.0608 ( -0.55) 0.0611 (0.46) 0.103 (0.71) -1.26* ( -1.95) 0.0829 (0.44) 5.93*** (2. 75) -0.276 ( -0.96) -0.00346 ( -0.16) 246 87.80 216 42 0.330 -78.37

Notes: See notes to Table 3.1. Note that coefficient magnitudes and the log likelihood are not directly comparable across the two columns in this table.

For the pre-1990 sub-sample, most of the estimated coefficients for the economic variables change only slightly from the full sample estimates in terms of their level of statistical significance. The magnitude of the coefficients do change somewhat, but these are not strictly comparable due to the different sample. There are some changes in the importance of some of the political and institutional variables. The statistical significance of the coefficient on socialism diminishes notably, while that for the reduction in civil freedom variable increases in statistical significance. In addition, the presence of past coups has

IMF lending

59

a positive and statistically significant estimated coefficient. Future reschedulings and a history of IMF agreements remain strongly and positively associated with current programme signings. The capacity of the model to predict the pattern of agreements remains largely unchanged from the full sample, declining slightly to 82 per cent. The predictive accuracy of the model with only a constant and the base economic model are 81.5 per cent and 82.4 per cent, respectively. For the post-1989 sample period the story is somewhat different. In this case the statistical significance of several coefficient estimates declines markedly, though by and large they are otherwise reasonably comparable to the full sample results for the economic variables. Of the additional variables, the quota review variable has an unexpected negative coefficient estimate that is marginally significant, while past agreements become unimportant statistically. It should be recalled that two of the omitted variables, the indicator of socialism and the presence of new governments, were at least somewhat significant in the full sample. The capacity of the model to predict IMF agreements in the post-1989 period increases significantly. The model with the most explanatory variables is able to predict accurately 87.8 per cent of the cases, or 216 of the 246 cases. This period also coincides with the most accurately predicted model in other studies, specifically the 88.1 per cent accuracy rate for the 353 cases examined in Thacker (1999). 8 However estimating with just a constant or the simplest economic model yields accurate predictions in 82.9 per cent and 86.2 per cent, respectively. Once again, the addition of political and institutional variables, while yielding some modest insights, does not help explain a large number of additional cases. There are two types of lessons that emerge from the estimations. First of all, there do appear to be a large number of economic and non-economic hypotheses about IMF programme patterns that are supported by the econometric analysis. By and large indicators reflecting the severity of economic difficulties are associated with a higher likelihood of signing an agreement. Furthermore, a history of IMF agreements, the anticipation of the need to reschedule, a nonsocialist orientation, and possibly reductions in civil liberties also seem to be correlated with entering into such programmes. Second, despite the statistical support these models lend to a variety of hypothesized factors that affect IMF agreement patterns, and despite the additional capacity of these variables individually and collectively to explain the statistical variance in the models, their inclusion does not generally increase the net predictive accuracy of the estimating equations. Instead, changes in predictive accuracy seem driven largely by the sample. On the one hand, the pre-1990 period conforms less well to any of the estimating models, yielding accuracy rates around 81 per cent. In the post-1989 period, however, accuracy ratios increase to closer to 87 per cent. Minor alterations in the data sample can have significant effects on specific explanatory variables. The absence of robustness regarding the coefficient estimates and significance of certain variables is consistent with the hypothesis that several factors affect the pattern of IMF agreements idiosyncratically rather than systematically. If correct, such an interpretation would be

60

IMF lending

discouraging for those who prefer a more objective and coherent basis for IMF interventions.

Concluding remarks This chapter attempts to identify analytically and empirically the potential sources of systematic bias within the IMF. It sets out to provide a more comprehensive explanation of IMF lending than currently exists in the literature. On the analytical side, the preferences of key shareholders, the interaction of domestic political variables with Fund objectives and requirements and institutional imperatives of the IMF are identified as potential influences that may affect the pattern of IMF agreements. While the absence of suitable proxies prevents an exhaustive testing of the list of possible biases, several variables are included in the empirical estimations in an attempt to test their relative significance. The empirical results are interesting. The variable indicating socialist orientation turns out to have a negative coefficient estimate, which is consistent with an interpretation of a pro-liberal bias at the IMF and a penalty-based incentive system. This is also consistent with the results in Thacker (2000) regarding the interests of the United States. However, interpreting the results for this variable is difficult since it measures only an extreme position, and does not provide much insight into the potential bias for or against moderately socialist non-aligned states. Furthermore there is no way of knowing whether the Fund discriminates against these members, or whether they simply do not wish to adopt Fund programmes. The finding is, therefore, also consistent with explanations which emphasize the gap between a government's and the IMF's policy preferences in determining a country's willingness to turn to the Fund. On governance issues there is reasonable evidence that in the past Fund programmes have been associated with tighter restrictions on civil liberties. Of equal importance, however, is that in the post-1989 period this correlation disappears. Therefore, there are grounds for suspecting that IMF lending is indeed becoming more sensitive to the requirement of good governance. Finally, the tests of the institutional variables are also quite clear. The political economy story emerging from public choice hypotheses receives little support in terms of the importance of quota reviews. Nor does its own liquidity seem to have imposed a constraint on Fund lending. However, there is clearly a marked recidivist tendency. Our general conclusion is that political and institutional factors do systematically add something to our understanding of Fund lending and it is wise not to ignore them. However, it is also wise not to get overexcited about their consequences for prediction. Since IMF agreements are present in only about 20 per cent of the observations, in most large sample econometric studies a prediction of 'no agreement' would be correct about 80 per cent of the time. Even for within sample data our ability to improve on this guess is relatively modest. Using economic variables alone raises our powers of prediction to something generally in the range of 82-83 per cent. Including political and institutional

IMP lending 61 factors does little to raise this success rate. Instead, predictive accuracy is affected far more by the sample. Two further conclusions may be drawn from this. The first is that claims for an apparently high rate of success in explaining IMF lending need to be treated with caution. However, the second conclusion is that there is evidence that political and institutional factors are significantly related to which countries sign IMF agreements. IMF lending is not an exclusively economic phenomenon; there are political and institutional influences at work. The policy consequences of these results speak first to the quality of the signal being transmitted by the Fund regarding the economies of its clients. By introducing political noise into its signal the IMF may run the risk of debasing its economic seal of approval and compromising its ability to stimulate lending from private sources. Even if the Fund serves the strategic and political interests of its more powerful shareholders only infrequently, by doing so it forfeits the claim to political neutrality and weakens the credibility of its argument that programmes are based on economic considerations alone. While the subordination of the Fund to political interests may serve the short-term interests of some shareholders, it should also be noted that their long-term interests might be harmed. The subversion of international institutions and regimes to political imperatives damages their credibility and their ability to evolve to fulfil new roles, especially in times of crisis. While the incorporation of shareholder interests into Fund activity clearly breaches the spirit in which the institution is supposed to operate according to its Articles of Agreement, the same may not be true for basing programmes partially on governance issues. Recent evidence strongly suggests that governance concerns are important determinants of both long-term development and the success of policy reforms. As long as the elements of governance upon which Fund agreements are conditioned reflect their acknowledged economic importance, the integrity of programmes should not be compromised. However, the Fund should not move beyond the professional consensus on governance matters, as the line separating these issues from primarily political issues is often very thin indeed. Turning to institutional imperatives, the empirical results suggest that the Fund develops long-term relationships with many of its clients, and that the maintenance of these relationships is a key determinant of Fund lending. The danger here is clear. Those who think that the Fund is too activist and too wasteful will see this as evidence that the IMP simply pours good money after bad. Others will view the process as creating dependency. Still others will worry about how several consecutive programmes will be perceived by private lenders, who may well ask how they are to determine whether a country adopting an IMF agreement will end up recovering quickly or becoming a ward of the Fund. One possible way of addressing this problem is to make the roles of the different Fund facilities more distinct, and therefore the signals they send more transparent. Finally, large sample investigations of IMF lending such as this one set out to identify those factors that determine which countries sign and do not sign IMF

62

IMF lending

agreements. There may always remain an inherently indeterminate element in this. Legitimate differences of opinion or disagreements between government negotiators and Fund officials may prevent an agreement from being signed. Even in these cases, however, both sides may be better served by making the process more transparent. Private lenders, non-governmental agencies, other governments and financial institutions, and ordinary citizens should not have to guess why some countries have an IMF agreement and others do not. International monetary and financial affairs already have a surfeit of imperfect information that is asymmetrically distributed. International regimes are not intended to exacerbate this imperfection. Large sample estimations do not tell us very much in detail about the role of politics and institutional factors in IMF arrangements. The nuances and subtleties will only be captured by case studies, with a systematic review of these being needed to identify common elements. This having been said, aggregative study does have its place in trying to move beyond anecdotal and piece-meal evidence. It does tell us whether certain reasonably well-defined political and institutional factors generally exert a significant and quantitatively important effect, and it does enable us to assess their importance relative to the economic factors that have conventionally been used. While not telling us all there is to know about the political economy of IMF lending the inclusion of political and institutional variables does tell us something, and rather more than we already knew.

Data definitions GNP per capita: GNP per capita in thousands of $US (source: World Bank). GDP growth: percentage change in GDP (source: World Bank). Reserves/imports: total foreign reserves divided by total imports of goods and services (source: World Bank and IMF). Current account/GDP: current account balance as a percentage of GDP (source: World Bank). Change in reserves: the year on year change in total reserves divided by the previous year's reserves (source: IMF). Real exchange rate: the official number of domestic currency units per $US multiplied by the ratio of the US consumer price index to the country's consumer price index. This number was calculated for the current year and for three years previously (adjusting for changes in base years) and the difference between the two was expressed as a proportion of the value from three years before (source: IMF and World Bank). Debt-service ratio: the ratio of total long-term debt-service payments to exports (source: IMF and World Bank). Change in debt-service ratio: the year-on-year change in the debt-service ratio expressed as a ratio of the previous year's debt-service ratio (source: IMF and World Bank). Debt/GDP: the ratio of total foreign debt to total GDP (both expressed in $US) (source: World Bank).

IMP lending 63 Change in private debt: the year-on-year change in indebtedness to private sources expressed as a ratio of the previous year's level of indebtedness to private sources (source: World Bank). Arrears/debt: the level of arrears to private and official sources expressed as a percentage of total foreign debt (source: World Bank). Past reschedulings: the number of years out of the previous two years in which a country rescheduled some portion of its official or private debt (source: World Bank). Real LIBOR: the London Interbank Offered Rate on US 6 month Treasury Bills (annual average) less the rate of US CPI inflation (source: IMF). Change in real LIBOR: the year-on-year change in real LIBOR (as calculated above) as a ratio of the previous year's real LIBOR (source: IMF). US exports: the value of exports shipped from the United States to the country (in$ thousands) deflated by the US consumer price index (source: IMF). French exports (Africa): the value of exports shipped from France (in $ thousands, deflated by the US consumer price index) multiplied by a binary variable that took the value of 1 if the country was in Africa, and 0 otherwise (source: IMF). Socialism: a qualitative variable that took the value of 0, 0.5 or 1 with 0 implying no clearly stated socialist orientation of the government, 0.5 for those identified as moderately socialist, and 1 for those countries with a strongly socialist orientation (source: based on Europa Yearbook descriptions). Civil freedom: a qualitative variable in which 1 represented the most political freedom and 7 represented the least (source: Freedom House). Change in freedom: a difference between the measure for civil freedom for one year and the measure two years previously, expressed as a ratio of the level from two years before (source: Freedom House). New government: a binary variable that took the value 1 if a substantively new government had taken power within the previous two years (source: descriptions in Europa Yearbook). Coup frequency: the number of years out of the preceding five years in which there had been a coup d'etat in the country {source: descriptions in Europa Yearbook). Quota review: an indicator taking the value of 1 if there was a quota review to occur within the next two years (source: IMF). Failed agreements: the number of agreements in the past five years which were 'incomplete' according to the methodology of Killick et al., that is, agreements with more than 20 per cent of the commitment undrawn by the country at the time of expiry (source: IMF). IMF liquidity: the amount of industrial country quotas at the IMF less outstanding IMF credit, as a ratio of total industrial country quotas (source: IMF). GDP: the gross domestic product of a country expressed in $US (source: World Bank). Future rescheduling: an indicator of whether or not some portion of private or official debt is rescheduled in the upcoming year (source: World Bank).

64

IMF lending

Future government: an indicator of whether the government changed in the upcoming year, but was not associated with a coup (source: Europa Yearbook). Past IMF: an indicator of IMF agreements being present in the previous two years (source: IMF).

Notes For example, Payer (1974, 1983) has been among the more strident critics of the IMF, casting it as little more than a front for the political interests of its principal shareholders in general and the United States in particular. In a similar vein, Polanyi Levitt (1983) claims that the 'IMF is the ultimate guardian of the interests of capitalists and bankers doing international business'. Gisselquist also suggests that developing countries that 'have the approval of major industrial countries are supported with IMF credits, citing South Korea, the Philippines, Chile and South Africa' (Gisselquist, 1981: 224 ). In more restrained fashion, Loxley (1986) examines a number of anecdotal events which, taken together, lead him to conclude that the IMF has been manipulated politically by the United States, though its influence is neither 'unrestrained or unchallenged' (Loxley, 1986: 91). These anecdotal accounts have since received more corroboration by Thacker (2000), who concludes on the basis of a large sample investigation that the political interests of the United States are indeed important determinants of Fund behaviour. In addition to influencing credit agreements with members, it has also been alleged that the United States has used its voting power to affect membership in the Fund in such cases as the expulsion of Czechoslovakia in 19 54 (Kahler, 1990) and the exclusion of Cuba. Other observers, however, have raised questions about some of these specific cases. Williamson (1982) argues that many of the governments that the United States was supposed to be seeking to punish were indeed major recipients of IMF loans, including the communist governments of China, Cambodia, Laos, Romania, Vietnam and Yugoslavia. Williamson also notes that under Prime Minister Manley, Jamaica was the largest borrower (per capita) of IMF resources, and that even Allende's government in Chile was permitted to make drawings on a compensatory financing facility loan from the Fund. Kahler (1990) also suggests that while the United States can clearly influence programmes 'for friends and clients at the margin' its influence at the Fund arises largely because of the support its initiatives receive from other large industrial countries (Kahler, 1990: 110). The areas in which politics may affect Fund activities are broad, and extend beyond the lending decisions examined here. For example, Bird (1998) has shown how a full appreciation of decisions relating to the Fund's Special Drawing Rights can only be achieved if politics and economics are integrated. 2 The institutional theory of the IMF claims that its utility as a unified actor is an increasing function of the size of its loan portfolio and its conditionality. The implication is that the Fund will manipulate its conditionality in the light of developing countries' access to private international finance, relaxing it when private capital markets are keen to lend and tightening it when there is severely impaired access to private capital. Although the evidence on this is only partial and inconclusive, some observers, stressing the time invariant continuity of IMF conditionality, doubt whether there is strong support for the public choice interpretation (Bird, 1995). Nevertheless the approach does highlight the potential ability of the IMF's management to pursue its own institutional agenda with some degree of independence from the principal shareholders. 3 Knight and Santaella (1997) find more false positives as a proportion of actual agreements, though in terms of magnitude there are far more false negatives generated by their model.

IMP lending 65 4 A country was deemed ineligible to sign an IMF agreement for three reasons. Some countries are obviously ineligible because they are either not members or because they have been declared ineligible due to the accumulation of arrears to the Fund. One version of the model was estimated in which those countries ineligible for an agreement due to arrears were included in the sample, on the grounds that the IMF could solve the arrears problem if it so wished by extending a new agreement. Only minor differences were observed as a consequence of the sample change. Countries that were already operating with a previous IMF agreement for the entire calendar year, however, were also considered ineligible. It should be noted that there were six cases of countries signing a second agreement while operating an earlier one, five of which were cases of lower conditionality Structural Adjustment Facility (SAF) programmes. These cases were included in the sample. Except for rare cases of brief transition periods from one agreement to another, countries do not operate with two high conditional agreements, and presumably do not seek to sign such agreements. Again, the models were also estimated in which countries that had only two months to sign an agreement during a calendar year - but did not do so - were also excluded on the grounds that there may have been insufficient time to sign one. Again, the results from these estimations are largely comparable to those reported in the text. 5 The effect of arrears is potentially interesting, since prior to the Brady Plan the IMF was reluctant to lend to countries that were in arrears to creditors. Therefore, one might expect a strong negative effect for arrears in the early part of the sample. However, the prospect of an IMF agreement could persuade a country to pay its arrears in anticipation of financing from the IMF. In this case arrears could partly reflect the IMF's own decisions on lending, as well as a factor in that decision. 6 There is a large literature that attempts to predict such debt-servicing problems, notably Mcfadden et al. (1984) and Hajivassiliou (1992). 7 There is evidence of heteroscedasticity related to per capita GNP, total debt, and the trade-to-GDP ratio, which emerges from estimating the model coefficients jointly with a model for the variance, using maximum likelihood procedures. The results are obviously dependent on the specific variables included as terms in the variance. Deaton (1997: 85-6), however, makes it clear that such a problem need not interfere with the interpretation of results from the uncorrected model. The results reported here are largely consistent with those from the models that are estimated jointly with a separate variance term. 8 To examine more closely the comparability with Thacker's results, further estimations were undertaken using his measure of proximity to the United States. This limited the sample even further since comparable records were not kept prior to 1983. By and large Thacker's results were not replicated in these estimations. Two notable changes were that the change in voting proximity was not found to be statistically significant in any of the equations estimated. On the other hand, the simple measure of proximity was found to have a reasonably significant coefficient estimate, especially in models run for the pre-1990 period. This should not be taken as refutation of Thacker's results. Instead it is simply indicative of one of our principal conclusions; results are sensitive to the specific sample. In one way the results are similar to Thacker's; the predictive accuracy for estimations over the post-1989 period tend to be markedly higher than for the full sample or for the earlier sub-sample.

References Asetto, V. (1988) The Soviet Bloc in the IMF and the IBRD (Boulder: Westview Press). Auvinen, J. (1996) 'IMF Intervention and Political Protest in the Third World: A Conventional Wisdom Redefined', Third World Quarterly, 17(3), 377-400.

66

IMP lending

Bird, G. (1995) IMF Lending to Developing Countries: Issues and Evidence (London: Routledge}. Bird, G. (1996) 'Borrowing from the IMF: The Policy Implications of Recent Empirical Research', World Development, 24(11 ), 1753-60. Bird, G. (1998) 'The Political Economy of the SDR: The Rise and Fall of an International Reserve Asset', Global Governance, 4, 355-79. Bird, G. and T. Orme (1981) 'An Analysis of Drawings on the International Monetary Fund by Developing Countries', World Development, 9, 563-8. Bird, G. and D. Rowlands (1997) 'The Catalytic Effect of Lending by the International Financial Institutions', The World Economy, 20(7), 967-91. Blomberg, S.B. and G.D. Hess (1997) 'Politics and Exchange Rate Forecasts', Journal of International Economics, 43(1-2), 189-205. Conway, P. (1994) 'IMF Lending Programs: Participation and Impact', Journal of Development Economics, 45, 365-91. Cornelius, P. (1987) 'The Demand for IMF Credits by Sub-Saharan African Countries', Economic Letters, 23, 99-102. Deaton, A. (1997) The Analysis of Household Surveys: A Microeconometric Approach to Development Policy (Baltimore: The World Bank/John Hopkins University Press). Doroodian, K. (1993) 'Macroeconomic Performance and Adjustment under Policies Commonly Supported by the International Monetary Fund', Economic Development and Cultural Change, 41(4), 849-64. Edwards, S. and J. Santaella (1993) 'Devaluation Controversies in the Developing Countries: Lessons from the Bretton Woods Era', in M. Bordo and B. Eichengreen (eds) A Retrospective on the Bretton Woods System: Lessons for International Monetary Reform (Chicago: University of Chicago Press and NBER). Feldstein, M. (1998) 'Refocusing the IMF', Foreign Affairs, 77, 0-33. Fleming, J. (1964) The International Monetary Fund: Its Forms and Functions, IMF Pamphlet Series no. 2 (Washington, DC: The IMF). Gisselquist, D. (1981) The Political Economy of International Bank Lending (New York: Praeger). Hajivasiliou, V. (1987) 'The External Debt Repayment Problems of LDCs: An Econometric Model Based on Panel Data', Journal of Econometrics, 36, 205-30. Joyce, J. (1992) 'The Economic Characteristics of IMF Program Countries', Economic Letters, 38, 37-242. Kahler, M. (1990) 'The United States and the International Monetary Fund: Declining Influence or Declining Interest?', in M. Karns and K. Mingst (eds) The United States and Multilateral Institutions: Patterns of Changing Instrumentality and Influence (London: Routledge). Killick, T. (1995) IMF Programs in Developing Countries: Design and Impact (London: Routledge}. Killick, T., M. Malik and M. Manuel (1992) 'What Can We Know About the Effects of IMF Programs', The World Economy, 15(5), 575-98. Klein, M. and N. Marion (1997) 'Explaining the Duration of Exchange Rate Pegs',Journal of Development Economics, 54(2), 387-404. Knight, M. and J. Santaella, (1997) 'Economic Determinants of IMF Financial Arrangements', Journal of Development Economics, 54(2), 405-36. Kwitny, J. (1983) 'US Charged with Bias in IMF Votes', Wall Street Journal, May 18, 39. Loxley, J. (1986) Debt and Disorder: External Financing for Development (Boulder: Westview Press, Inc.).

IMP lending 67 McFadden, D., R. Eckaus, G. Feder, V. Hajivassiliou and S. O'Connell (1984) 'Is There Life After Debt: An Econometric Analysis of the Creditworthiness of Developing Countries', in G. Smith and J. Cuddington (eds) International Debt and the Developing Countries (Washington, DC: The World Bank). Payer, C. (1974) The Debt Trap: The IMF and the Third World (Harmondsworth: Penguin). Polanyi-Levitt, K. (1983) 'Jamaica: Manley's Defeat - Whose Responsibility?', in J. Torrie (ed.) Banking on Poverty: The Global impact of the IMF and World Bank (Toronto: Between the Lines). Rodrik, D. (1996) 'Understanding Economic Policy Reform', Journal of Economic Literature, XXXN( 1) 9-41. Rowlands, D. (1995) 'Political and Economic Determinants of IMF Conditional Credit Allocations: 1973-1989', Norman Paterson School of International Affairs Development, Working Paper Ottawa: NPSIA). Siddell, S. (1988) The IMF and Third-World Political Instability: Is There a Connection? (Houndmills, Hampshire: The Macmillan Press Ltd). Stiles, K. W. (1990) 'IMF Conditionality: Coercion or Compromise?', World Development, 18,959-74. Thacker, S. (1999) 'The High Politics ofiMF Lending', World Politics, 52, 38-75. Vaubel, R. (1994) 'The Political Economy of the IMF: A Public Choice Analysis', in D. Bandow and I. Vasquez (eds) Perpetuating Poverty: The World Bank, the IMF, and the Developing World (Washington, DC: The Cato Institute). Vaubel, R. (1996) 'Bureaucracy at the IMF and the World Bank: A Comparison of the Evidence', The World Economy, 19(2), 195-210. Williamson, J. (1982) The Lending Policies of the IMF (Washington, DC: The Institute for International Economics). Williamson, J. (ed.) (1993) The Political Economy of Policy Reform (Washington, DC: Institute for International Economics).

4

IMF programmes - do they work? Can they be made to work better?

Introduction International Monetary Fund (IMF) conditionality has been a subject of debate ever since it was introduced in the early 1950s. A number of recent reports have claimed that it has become excessive, and some commentators have suggested that there may be a conditionality Laffer curve with increased conditionality being linked to diminishing effectiveness. 1 The Meltzer Commission concluded that IMF programmes are 'unwieldy, highly conflictive, time consuming to negotiate and often ineffectual'. With a new Managing Director in place the IMF has itself undertaken a review of conditionality (IMF, 2001). Central to the debate and its implications for policy is the question of whether conditionality and the programmes that embody it 'work'. If they do, why change things? If they do not, then should conditionality be abandoned in its present form, as recommended by the Meltzer Commission, or reformed in some way, and if so, how? This raises the fundamental question of how to judge whether IMF programmes work. 2 A narrow criterion is to focus on their balance-of-payments effects. After all, the IMF is primarily a balance-of-payments institution. With this focus in mind a recent survey of the cross-country evidence conducted by the IMF (ul Haque and Khan, 1998) claims that its programmes do indeed 'on balance' work. This criterion may be broadened to include the effects of programmes on other additional macroeconomic outcomes such as economic growth and inflation. There is a relatively large literature that assesses conditionality in this way. The literature also encompasses studies that examine the effects of IMF programmes on intermediate policy targets such as fiscal deficits, monetary growth and the exchange rate. Very little attention has been paid, however, to other potential indicators of success or failure, which move beyond the macroeconomic effects of IMF programmes. The purpose of this chapter is to identify these other indicators so that a broader definition of the concept of 'work' may be offered, and to see what the evidence reveals using this definition. The findings may also give a clue as to the direction in which reforms to conditionality should go. The chapter only seeks, however, to be indicative rather than definitive. The layout of the chapter is as follows. The section 'The macroeconomic effects of IMF programmes' undertakes a brief review of the existing literature on

IMF programmes - do they work?

69

the effects of IMF programmes. Rather than being comprehensive, it sets out to establish broad areas of consensus and to summarize what we know. Against this background, the section 'Do IMF programmes work? Other indicators' extends the criteria by which the effects of IMF programmes may be judged. It does this by identifying institutional objectives and examining the degree to which conditionality as currently practised has enabled these objectives to be realized. If the objectives are achieved this is taken to indicate that Fund programmes 'work'. If they are not, then a less positive conclusion seems justified. The section 'Recidivism' draws on the available empirical evidence combining that from the existing literature with the additional evidence presented here to discuss the implications for policy. By isolating possible shortcomings in the design of conditionality it attempts to identify areas for reform. The final section offers a summary and some concluding remarks about the current debate surrounding conditionality in the light of the analysis undertaken in this chapter.

The macroeconomic effects of IMF programmes This section sets out to provide a brief summary of the alternative methodologies that have been used to assess the macroeconomic effects of IMF programmes and the results that have emerged. It cites some of the principal and most frequently quoted studies and by so doing attempts to provide a representative sample of research in this area. There are a number of studies that offer a more detailed critical analysis of the alternative methodologies and a more comprehensive summary of the literature (Killick, 1995; ul Haque and Khan, 1998). But, for our purposes this level of detail is not required. Underlying all research into the effects of IMF programmes is the problem of the counter-factual, that is, what would have happened in the absence of an IMF programme and therefore what effects can be attributed to the programme per se as opposed to other factors. In practice, there is no completely satisfactory means of dealing with this problem. Before-and-after tests implicitly assume that other things remain constant, which they do not. With-without tests assume that it is possible to formulate an accurate view as to what would have happened in the absence of an IMF programmes. One way of doing this is to compare countries that have negotiated Fund programmes with other countries that have not. But there is a problem in finding countries that are similar in all respects apart from the involvement of the Fund. Otherwise some attempt needs to be made to allow for differences between those countries that do and do not have programmes, in terms of their economic circumstances and how policy might evolve. This is what the 'generalized estimator evaluation' approach attempts to do. But the very decision as to whether or not to turn to the IMF itself implies a significant difference in the approach to economic adjustment as well as in the political environment in which adjustment takes place, and this is difficult to capture empirically. This problem may be overcome by endeavouring to simulate the performance of an individual economy under different sets of policies. But simulation analysis encounters another problem since it relies on being able to specify a model

70

IMF programmes- do they work?

that accurately describes how individual economies function. In fact, economies differ, and no one model is likely to provide an accurate description of all countries that might tum to the IMF (Dicks-Mireaux et al., 2000). Furthermore, the response to policies undertaken under the auspices of the Fund may differ from the response when the same policies are pursued independently, and, on top of this, parameter values may be affected by the policies adopted. A case study methodology offers another approach but suffers from problems of generalization. Other difficulties that get in the way of an unambiguous assessment of the effects of IMF programmes relate to the time period over which the effects are monitored and the range of performance indicators that are studied. There is no reason to presume that programme effects follow some linear path over time or that positive effects on some economic variables will be matched by similar positive effects on others. Economic variables may move in different directions over time and better performance in terms of one variable may be offset by worse performance in another. Unless a reasonably sophisticated social welfare function has been specified, it is difficult to say whether things overall have improved or deteriorated. 3 While the search for methodological improvement in econometric analysis of the effects of IMF programmes continues, there is an accumulation of evidence from studies using different methodologies and this has gradually provided an overall picture. Results that are robust across different studies and methodologies may engender greater confidence than those that are methodology-specific, or specific to individual studies. Without undertaking a detailed review, we can make a number of important generalizations. In terms of the key performance indicators, early before-andafter studies found no significant improvement in the current account balance of payments, although both Khan (1990) and Pastor (1987) discovered significant positive effects on the overall balance of payments in more recent research. With-without tests of one form or another have tended to show stronger results for the current account (Gylfason, 1987; Khan, 1990; Doroodian, 1993). Killick (1995) uses a combination of before-and-after tests and case study evidence. His results suggest that both the overall and current account balances strengthen, especially over a three-year period, in part by import compression (rather than import strangulation) but also by relatively large increases in export volume, which rise through time. This performance was secured against a background of deteriorating commodity terms of trade. For inflation before-and-after tests reveal a record that is generally weak, with the inflation rate increasing as frequently as it declines. The results are, however, almost always statistically insignificant. With-without tests and simulation studies suggest a better performance, although significance is at best, low. The demand-reducing effects of IMF programmes seem to be offset by the effects of devaluation and liberalization. On economic growth, although many early studies found little connection, Goldstein and Montiel (1986) found IMF programmes to have a significantly

IMP programmes - do they work?

71

negative effect. Simulation tests by Khan and Knight (1982) also predicted that demand management programmes similar to those supported by the IMF will have negative short-run effects on growth, although subsequent research by them suggests that these effects can be ameliorated by incorporating supply-side measures to protect investment (Khan and Knight, 1985). Conway (1994) finds significant differences between the short-run and long-run effects of IMF programmes on economic growth and investment, as contemporary reductions are followed by lagged increases. Killick (1995) finds a largely neutral association with economic growth, although over a longer time-frame the association is positive (albeit of only limited statistical significance). This is somewhat surprising in light of the apparent negative effect on fixed investment, suggesting that growth has been achieved through increases in the marginal productivity of capital or that it may represent a temporary recovery from recession. Stabilization under the auspices of the Fund is generally achieved by lowering investment rather than by increasing savings. It is investment that carries the main burden of reduced absorption; private and public consumption are apparently little influenced by the negotiation of a programme with the Fund (which has some bearing on the debate over the effects of IMP-backed programmes on the poor). While Hutchison (2001) discovers a significant adverse effect on output growth over one to two years, Przeworski and Vreeland (2000) claim that IMF programmes have an enduring adverse effect on economic growth. Meanwhile, Garuda (2000) suggests that their effect on income distribution depends on the severity of the initial circumstances in which a country turns to the Fund. Table 4.1 provides a summary of the results of the principal studies as presented by ul Haque and Khan (1998). Although one can quibble about whether they have identified all the important studies, the overall picture would be widely acknowledged as reasonable. 4 Another part of the existing literature focuses on the effects of IMF programmes on the policy instruments through which an influence on final outcomes is sought. Edwards (1989) paints a rather bleak picture showing that for 1983-5 there was widespread failure to comply with either fiscal or monetary targets. For example, in no year in the study period did the rate of compliance reach 50 per cent for the target relating to the size of the government deficit relative to GDP. Subsequent research has confirmed that at best only about a half of policy targets are achieved; in general, policy variables seem to be little influenced by IMF programmes with the record being weakest for monetary restraint. Thus, while Killick (1995) finds that IMF programmes exert a significant effect on the real exchange rate and have some impact on certain fiscal variables, although not necessarily the fiscal deficit, neither he nor Conway (1994) finds a significant effect on the rate of credit expansion. What emerges from this brief survey of the existing evidence? On the positive side IMF programmes do seem to be associated with a statistically significant and enduring depreciation in the real exchange rate. Perhaps connected with this, they also appear to be associated with some significant strengthening in the

(1981) (1985)

1968-75 1968-75

29 29

-

+ +

+*

-

+ +

-

+* +* +*

-

-

+

-*

+ 0

+ 0 +* +*

-

-

+* +* +

-

-,+*

-* -,+* +* +*

0 0

+

+ 0 0 0 0 +* + 0 0 -* 0 0 -* 0 0 0 0 +*

-

0 0 0

Growth

Inflation

Current account

Balance of payments

Effects ana

Notes a Direction of change: ( +) indicates positive effect, (-) indicates negative effect, (0) indicates no effect. * Statistically significant at the 5 per cent level.

Source: ul Haque and Khan (1998).

Khan and Knight Khan and Knight

Simulation

88

58 69 73 68 74

68 259 217

1974-81 1973-88 1976-86 1973-92 1986-91

Generalized evaluation

Goldstein and Montiel (1986) Khan (1990) Conway (1994) Bagci and Perraudin (1997) Dicks-Mireaux et al. (2000)

55

23 24 22 18 16 19 12 44 38 14

-

79 31 38 35

Number of countries

12 78 38 32

1963-72 1973-7 1974-9 1980-1 1965-81 1979-85 1983-93

Number of programmes

1970-6 1971-80 1971-82 1977-9

(1978)

Time period

Donovan (1981) Donovan (1982) Loxley (1984) Gylfason (1987)

With-without

Reichmann and Stillson Connors (1979) Killick ( 1984) Zulu and Nsouli (1985) Pastor Killick (1995) Schadler et al. (199 3)

Before-after

Study

Table 4.1 Summary of empirical evaluations of the effect of Fund programmes

IMF programmes- do they work?

73

balance of payments. But, their impact on fiscal and monetary restraint is less significant or not significant at all, and the effects on inflation and economic growth are generally insignificant or significantly negative in the case of economic growth. If an objective of Fund programmes is to reduce inflation and encourage growth it is difficult to argue that they work. The claim that 'on balance' they do, would therefore appear to depend on the weights attached to strengthening the balance of payments as compared with other macroeconomic objectives. Still underpinning all the existing literature is the problem of attributing outcomes to IMF programmes. It may be that the counterfactual is being reasonably well captured and that the coefficients are indeed insignificant. It may also be that the findings of insignificance reflect incorrect specification of the models. The coefficients in the context of better constructed models might be significant. The response to this could be twofold. The first could be to improve the models used to better capture the effects of IMF programmes and no doubt research will continue along these lines. The second, and the one pursued here, is to look for other criteria which sidestep the problem of the counterfactual and judge whether IMF programmes 'work' in other ways.

Do IMF programmes work? Other indicators IMF programmes and the conditionality that they involve represent the mechanism through which the Fund seeks to achieve its institutional objectives. The Fund's Articles of Agreement envisage its assistance as being 'temporary' and 'revolving'. The implied sequence of events is as follows. A country encounters economic problems and turns to the IMF with the 'balance-of-payments need' that, according to the Articles, is a prerequisite for financial assistance from the Fund. A programme of policies is negotiated that the country then implements as set out in the 'letter of intent' it signs. This programme strengthens economic performance in general, but in particular strengthens the balance of payments, thus providing the resources with which the Fund can be repaid. The improvement also eliminates the 'need' for further assistance, allowing the Fund's resources to revolve and to be used in support of other countries. To the extent that this is an accurate representation of the relationship between the Fund and client countries envisaged in the Articles of Agreement, two features of IMF programmes emerge by which they may be judged. First, borrowing from the Fund should be a low-frequency event. Countries should be spending only relatively short periods of time under Fund programmes and should not have a succession of them. Second, since conditionality is the mechanism through which the Fund seeks to strengthen economic performance, it might be assumed that programmes have to be implemented and carried through to completion. In addition to the above, a third criterion is suggested by the Fund's own pronouncements about what its programmes seek to achieve. While, in part, they are designed to achieve macroeconomic stabilization and supply-side

74 IMP programmes - do they work? adjustment, they are also intended to catalyse other supporting capital inflows. In large measure this catalytic effect is seen as being the consequence of the signal that is supposedly transmitted by conditionality; a signal of enhanced credibility in terms of policy reform. How well do IMF programmes work in terms of these objectives? The remainder of this section examines each in turn.

Recidivism Members of the IMF may be divided into a number of categories in terms of their drawings; nonusers, infrequent users and frequent users. Frequency of use may be measured either in terms of the number of programmes that a country has over a specific period of time or in terms of the time spent under Fund programmes. The evidence is that a relatively large number of countries have been fairly frequent users of Fund resources according to both measures (Goreux, 1989; Bird, 1995; Bird et al., 2000). Table 4.2 provides data covering 1980-96 showing both the number of programmes and the number of years spent under programmes. By the first criterion Senegal exhibited the highest degree of recidivism with 11 programmes, closely followed by Madagascar with 10, and Costa Rica, Jamaica, Kenya, Mauritania, Morocco, and Togo with nine each. By the second criterion, Jamaica was the most recidivist nation with 16 years spent in Fund programs, followed by Malawi, Togo, and Uganda with 15 years each, and Cote d'Ivoire and the Philippines with 14 years each. Further empirical support for recidivism comes from studies that seek to explain econometrically the incidence of IMF arrangements. These commonly show that recent arrangements are statistically significant in explaining current arrangements (Conway, 1994; Knight and Santaella, 1997; Bird and Rowlands, 2001a). As noted above, however, not all countries are IMF recidivists. For some, drawing resources from the Fund is a low-frequency event. The recidivists generally appear to cluster within the group of low-income countries. Statistical analysis of their characteristics shows them to have relatively large current account deficits, low reserve holdings, low capital inflows, high rates of programme cancellation, low terms of trade, high debt-service ratios, and high perceived levels of corruption (Bird et al., 2000). The relatively high rate of recidivism and the quasi-permanent involvement of the IMF with some countries is not only directly at odds with one of the Fund's institutional objectives, but it may also carry indirect information concerning the extent to which IMF programmes work on macroeconomic variables. Where recidivism is concentrated among low-income countries the negative effect that studies find IMF programmes to have on investment and economic growth implies a vicious circle of low growth and low income, IMF referral followed by even lower economic growth. Moreover, while crosscountry econometric studies suggest that IMF programmes may strengthen the balance of payments, in the case of IMF recidivists the strengthening is clearly insufficient to eliminate the

IMF programmes - do they work?

75

Table 4.2 Countries with number and years of IMF programmes, 1980-96 Country

Programmes Years

Algeria Angola Argentina Bangladesh Benin Bolivia Botswana Brazil Bulgaria Burkina Faso Burundi Cameroon Central African Rep. Chad Chile Colombia Congo, Dem. Rep. Congo Rep. Costa Rica Cote d'lvoire Dominican Rep. Ecuador Egypt El Salvador Ethiopia Gabon Gambia, The Ghana Guatemala Guinea Guinea-Bissau Haita Honduras Hong Kong Hungary India Indonesia Iran, I R of Israel Jamaica Jordan Kenya Korea Kuwait Lesotho

4 0

Source: Bird et al. (2000).

3

5 0 11 11 8 11 0 6

7 5

3 6 0 5

3

3 3

6

4 8

6 10 6

3 3 0 8

7

7 0 11

3

3

9 8 4

11 14 5 9 10 8 5 9 8 9 4 11 5 10 10 0 9 4 0 0 0 16

7 4

7 3 6 6

7 4 5 2 6

3 0

7 3 0 0 0 9 4 9 4 0 5

7 13 6 0 9

Country

Programmes

Years

Liberia Madagascar Malawi Malaysia Mali Mauritania Mauritius Mexico Morocco Mozambique Namibia Nepal Nicaragua Niger Nigeria Oman Pakistan Panama Papua New Guinea Paraguay Peru Philippines Poland Rwanda Saudi Arabia Senegal Sierra Lone Singapore Somalia South Africa Sri Lanka Sudan Syrian Arab Republic Tanzania Thailand Togo Trinidad and Tobago Tunisia Turkey Uganda United Arab Emirates Uruguay Venezuela Zambia Zimbabwe

5 10 8 0

6 12 15 0 13 13

7 9 4 4 9

7

7

10 10 10 0 8 4 10

3

3

0 8 6

0 9 10 5 0 8 14 6 4 0 13 9 0 10 1 11 6 0 9 5 15 2 6

3 0

3 2

3 0 4

7 4 2 0 11 6 0 6 1

3 3 0 5

3 9 2 2 4 8 0

7 2 6 5

7 15 0 11 5

7 7

76 IMF programmes- do they work? balance-of-payments need which warrants further Fund support. Recidivism, therefore, poses questions about the degree of improvement in the balance of payments associated with IMF programmes. While infrequent users fit the pattern of usage envisaged in the Fund's Articles, the existence of frequent users and sometimes quasi-permanent users suggests that in these cases IMF programmes are failing to work in terms of delivering institutional objectives. Recidivism may also be connected with another criterion upon the basis of which IMF programmes may be assessed, since it more commonly occurs in countries that fail to complete previous programmes. Programme completion

When a country signs an agreement with the IMF, part of the agreement stipulates the size of the loan and the tranches in which it will be disbursed, conditional upon compliance with a range of quantified performance criteria. There is an end date by which it is expected that the programme will be completed. The degree of completion may then be measured by the extent to which the agreed credit is drawn down by this date, or by total purchases as a proportion of the total committed resources. While there is a discrete distinction between completion and non-completion, it is more useful to think of completion as a continuum. Non-completion occurs where less than 100 per cent of the loan is disbursed. Completion has sometimes been used as a proxy for policy implementation (Conway, 1994; Killick, 1995) but there are differences between the two. On the one hand, a country may fully implement the policies agreed in a programme and meet all the required performance criteria but decide not to use all the Fund finance for which it is eligible. Completion, in these circumstances, understates implementation or compliance. On the other hand, another country may fail to implement fully the agreed measures and fail to comply with all the quantified performance criteria but still be permitted to draw on the agreed loan; it may still complete the programme. Completion will then overstate implementation. The IMF may grant a waiver, which allows the country access to the loan in spite of its failure to meet performance criteria. This will usually happen where some unforeseen development outside a government's control has blown the programmea little off course but substantial progress has still been made and the deviation appears temporary. Where the deviation is larger, the initial programme may be modified to accommodate changed circumstances, or cancelled. Another programme may then be put in its place. Additional complications arise if completion is interpreted as reflecting success or failure. In principle non-completion could reflect economic success. A country may, for example, be forced to borrow from the Fund because of a deterioration in its terms of trade. If these subsequently improve the country may opt to discontinue its purchases from the IMF, effectively allowing the agreement to lapse. Alternatively, it may be that the policy measures agreed as a part of the programme tum out to be more effective than was initially assumed,

IMF programmes - do they work?

77

either in strengthening the current account balance of payments or in inducing private capital markets to lend. The agreement may even have been signed as a precautionary measure in the hope that the resources would not be needed. Again, the Fund programme may remain uncompleted because of economic success. More frequently, however, it may be assumed that non-completion signals a breakdown or failure of the programme, with the extent of the breakdown being reflected by the extent to which the IMF loan remains undisbursed. For arrangements where disbursements amount to less than half those initially agreed, Mussa and Savastano (1999) acknowledge that 'mainly these were cases where the programme went off track because policies deviated significantly from those agreed with the Fund and subsequent negotiations failed to reach agreement on a modified programme'. But while they argue that a drawdown of 75 per cent or more is generally indicative of success, Killick (1995) argues that the cut-off point should be slightly higher at 80 per cent. Killick tests this threshold against available case-study evidence and finds that it consistently identifies the programmes that were perceived as breaking down, and distinguishes these from the ones that were seen as successful. It also seems to be generally consistent with the Fund's judgement on eligibility, with countries using less than 80 per cent of the loan becoming ineligible (Schadler et al., 1995). Demonstrating the potential ambiguities involved, Mussa and Savastano (1999) claim that completion of an IMF programme implies that in the Fund's view, the country made substantial and satisfactory progress towards the primary objectives of its adjustment programme ... and that the policies of the authorities were broadly in line with the (often revised) understandings reached with the IMF during the life of the arrangement (p. 17). This is interesting since it suggests that failure to complete the programme reflects unsatisfactory progress, and offers an even stiffer test for success than the one they suggest elsewhere in their paper. It also implies that waivers and modifications are granted when programmes are perceived by the Fund as performing satisfactorily, in spite of a failure to comply with quantified targets, and this reinforces the view that non-completion generally reflects poor implementation of the policies and lack of success. How widespread is non-completion? Table 4.3 presents data for 1973-97 as reported by Mussa and Savastano (1999). A number of noteworthy features stand out. First, non-completion is indeed widespread. Over the full period only about 35 per cent of programmes were fully disbursed. Second, whereas the completion rate was fairly stable over 1973-87 at somewhere between 40 and 45 per cent, it fell quite sharply in the two subsequent sub-periods. In allowing for the bias reported in Table 4.3, the completion rate in the mid-to-late 1990s was very modest at probably below 20 per cent. Third, in the most recent period over 70 per cent of programmes had a final loan disbursement of 75 per cent or less. This compares with about 48 per cent of programmes in 1983-7 and 53 per cent

21.6 23.1 33.3 9.0

Full period (1973-97)c Of which Standby" EFF SAF/ESAF

0.25

15.3 13.4 22.2 18.9

7.1 16.1 15.8 15.1 19.1

0.25%.;;;