The Euroarea and the New EU Member States
Euro-Asian Studies General Editor: Christoph Bluth, Visiting Professor, Cent...
50 downloads
718 Views
1MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
The Euroarea and the New EU Member States
Euro-Asian Studies General Editor: Christoph Bluth, Visiting Professor, Centre for Euro-Asian Studies, University of Reading, and Professor in International Studies, University of Leeds The transition of the countries in Euro-Asia is one of the most important developments affecting the international system since the end of the Cold War. The development of market economies after decades of central planning, the formation of new states and national identities, the creation of new, democratic institutions of states and the reintegration into the world economy pose enormous challenges. Whilst some countries have progressed relatively well and are in the process of joining the European Union, others have experienced several economic and social dislocations, to the point of political disintegration and armed conflicts. The Centre for Euro-Asian Studies at the University of Reading is dedicated to the academic study of the political, economic, social and cultural aspects of this process. This series presents the most recent contributions from leading academics in the field. With an interdisciplinary focus, it seeks to provide a substantial, original and ongoing contribution to our understanding of the region which is of vital importance for academics and of high policy relevance for governments and businesses. Titles include: Yelena Kalyuzhnova and Wladimir Andreff (editors) PRIVATISATION AND STRUCTURAL CHANGE IN TRANSITION ECONOMIES Yelena Kalyuzhnova and Dov Lynch (editors) THE EURO-ASIAN WORLD A Period of Transition Yelena Kalyuzhnova, Amy Myers Jaffe, Dov Lynch and Robin C. Sickles (editors) ENERGY IN THE CASPIAN REGION Present and Future Yelena Kalyuzhnova and Michael Taylor (editors) TRANSITIONAL ECONOMIES Banking, Finance, Institutions Lúcio Vinhas de Souza and Bas van Aarle (editors) THE EUROAREA AND THE NEW EU MEMBER STATES Euro-Asian Studies Series Standing Order ISBN 0-333-80114-8 (outside North America only) You can receive future titles in this series as they are published by placing a standing order. Please contact your bookseller or, in case of difficulty, write to us at the address below with your name and address, the title of the series and the ISBN quoted above. Customer Services Department, Macmillan Distribution Ltd., Houndmills, Basingstoke, Hampshire RG21 6XS, England
The Euroarea and the New EU Member States Edited by
Lúcio Vinhas de Souza Kiel Institute for World Economics, Germany
and
Bas van Aarle Catholic University Leuven, Belgium
Editorial matter and selection © Lúcio Vinhas de Souza and Bas van Aarle 2004 All chapters © Palgrave Macmillan 2004 All rights reserved. No reproduction, copy or transmission of this publication may be made without written permission. No paragraph of this publication may be reproduced, copied or transmitted save with written permission or in accordance with the provisions of the Copyright, Designs and Patents Act 1988, or under the terms of any licence permitting limited copying issued by the Copyright Licensing Agency, 90 Tottenham Court Road, London W1T 4LP. Any person who does any unauthorised act in relation to this publication may be liable to criminal prosecution and civil claims for damages. The authors have asserted their rights to be identified as the authors of this work in accordance with the Copyright, Designs and Patents Act 1988. First published 2004 by PALGRAVE MACMILLAN Houndmills, Basingstoke, Hampshire RG21 6XS and 175 Fifth Avenue, New York, N.Y. 10010 Companies and representatives throughout the world PALGRAVE MACMILLAN is the global academic imprint of the Palgrave Macmillan division of St. Martin’s Press, LLC and of Palgrave Macmillan Ltd. Macmillan® is a registered trademark in the United States, United Kingdom and other countries. Palgrave is a registered trademark in the European Union and other countries. ISBN 1–4039–1519–9 hardback This book is printed on paper suitable for recycling and made from fully managed and sustained forest sources. A catalogue record for this book is available from the British Library. Library of Congress Cataloging-in-Publication Data The Euroarea and the New EU member states / edited by Lúcio Vinhas de Souza and Bas van Aarle. p. cm. – (Euro-Asian studies) Includes bibliographical references and index. ISBN 1–4039–1519–9 1. European Union – Europe, Eastern. 2. Europe, Eastern – Economic policy. 3. Monetary policy – European Union countries. 4. Europe – Economic integration. I. Souza, Lúcio Vinhas de. II. Aarle, Bas van. III. Series. HC241.25.E852E97 2003 337.1 42—dc21 10 9 8 7 6 5 13 12 11 10 09 08
2003053647 4 07
3 06
2 05
Printed and bound in Great Britain by Antony Rowe Ltd, Chippenham and Eastbourne
1 04
Dedicated to my parents, Rômulo and Sônia, to my brother Marco and my sister Mônica. Many, many thanks to Yelena Kalyuzhnova at the University of Reading, as without her support this book wouldn’t have seen the light of the day. And to M., for what could have been. Lúcio Vinhas de Souza With Dutch modesty, I simply dedicate this book to my parents. Bas van Aarle
This page intentionally left blank
Contents List of Tables
x
List of Figures
xii
List of Abbreviations
xiv
Preface
xvi
Foreword
xix
List of Contributors
xxi
Part I 1
2
3
General Policy Issues for the Accession Countries
Introduction – Monetary Integration of Central and East European Accession Countries: The Pros and Cons of Speedy versus More Gradual Strategies Peter Backé and Cezary Wójcik 1 Introduction 2 Participating in the Euroarea: medium- to long-term effects 3 Meeting the requirements of Euroarea entry: short-term implications 4 Conclusions Is Accession to EMU More Justifiable ex post than ex ante? Jarko Fidrmuc 1 Introduction 2 The optimum currency area theory 3 The endogeneity hypothesis of OCA criteria and EMU enlargement 4 Indices of endogenous optimum currency area 5 Conclusions Modelling Alternative Paths to EMU for the Accession Countries Lúcio Vinhas de Souza and Elisabeth Ledrut 1 Introduction 2 Modelling the exchange rate regime vii
3 3 7 15 20 23 23 25 30 34 36
39 39 40
viii
Contents
3 4 5 6
Data and procedures Estimation results Welfare effects of exchange rate regime choices Non-structural estimation of the effects of domestic and foreign shocks 7 Conclusions 4
5
Macroeconomic Adjustment in EU Accession Countries: An Analysis Using a Small Macroeconomic Model Bas van Aarle, Joseph Plasmans and Bruno Merlevede 1 Introduction 2 A small macroeconomic model of Accession Countries 3 Model estimation 4 Model simulation Nominal and Real Forex Regimes and EMU Accession Pieter van Foreest and Casper de Vries 1 Introduction 2 Monetary view on forex rate determination 3 The nominal forex regime and transitional growth 4 The real forex regime and transitional growth 5 Conclusions
43 44 48 50 52 54 54 56 59 70 79 79 80 85 94 98
Part II Country-specific Monetary Policy and Exchange Rate Questions in the Run-up to Monetary Union 6
7
Slovenia’s Monetary and Exchange Rate Framework in the Run-up to ERM II Gonzalo Caprirolo and Vladimir Lavraˇc 1 Introduction 2 Money-based stabilization policy (1991–95) 3 Price and real exchange rate stability dual targeting policy (1996–2001) 4 Exchange-rate-based stabilization policy and accession to ERM II (2001–) 5 Conclusions Monetary Policy in Estonia: The Transmission Mechanism Urmas Sepp, Martti Randveer and Raoul Lättemäe 1 Introduction 2 Main features of the Estonian monetary regime
103 103 106 111 119 126 130 130 131
Contents
3 The core features of the Estonian economy 4 The MTM in Estonia 5 Conclusions 8
9
10
Czech Monetary Policy on the Road to European (Monetary) Union Roman Matoušek and Anita Taci 1 Introduction 2 DIT in transitional economies 3 Methodology and data 4 Asset price reactions to changes in the CNB’s official interest rate before DIT 5 Asset price reactions to changes in the official interest rate in the DIT period 6 Conclusions Poland’s Accession to EMU – Choosing the Exchange Rate Parity Łukasz Rawdanowicz 1 Introduction 2 Concepts of equilibrium exchange rate 3 Empirical estimations 4 FEER calculations 5 BEER estimations 6 What should be the entry exchange rate? 7 Conclusions Monetary and Exchange Rate Strategies in Hungary on the Way to the Euro Attila Csajbók 1 Introduction 2 The recent history of the Hungarian exchange rate and monetary regimes 3 The Euroarea entry date: arguments, prospects and expectations 4 Strategic issues related to ERM II participation 5 Conclusions
References Index
ix
135 151 160
164 164 165 168 172 174 177
180 180 181 185 187 193 196 201
203 203 203 209 219 224 225 237
List of Tables
2.1
Trade integration and business cycles
27
2.2
Intra-industry trade, trade integration and business cycles
30
2.3
Comparison of business cycles of selected countries with that of Germany
34
2.4
Comparison of indices of EOCA of selected countries with those of Germany
36
3.1
Estimated coefficients for the float specification
45
3.2
Estimated coefficients for the peg specification
46
3.3
Coefficients for regime-specific samples
47
3.4
The loss-function outcomes
49
3.5
Overview of effects of temporary shocks by country and regime
51
4.1
A small macroeconomic model of Accession Countries: relations and definitions
57
4.2(a) Estimation results for consumption
60
4.2(b) Estimation results for investment
61
4.2(c) Estimation results for exports
62
4.2(d) Estimation results for imports
63
4.2(e) Estimation results for money demand
64
4.2(f) Estimation results for employment
65
4.2(g) Estimation results for wage inflation
66
4.2(h) Estimation results for inflation
67
A.1
Variables and data sources
78
5.1
Countries in panel
86
5.2
Panel DOLS regressions for nominal forex model
88
5.3
Cross-sectional averages for nominal forex model
89
5.4
Multiple rank regressions for nominal forex model
93
5.5
Panel DOLS regressions for real forex model
95
5.6
Cross-sectional averages for real forex model
96
5.7
Multiple rank regressions for real forex model
97
x
List of Tables xi
6.1 6.2 6.3 7.1 7.2 8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8 8.9 9.1 9.2 9.3 9.4 10.1 10.2 10.3
Macroeconomic indicators for the Slovenian economy (1993–96) Macroeconomic indicators for the Slovenian economy (1998–2000) Contribution of controlled prices to inflation (%) (1997–2002) Selected indicators of the Estonian economy, 1992–2001 GDP and CPI elasticities (in %) for temporary shocks, Estonia Average volatility of PRIBOR Effects of official interest rate on short-term interest rates (01.96–03.97) Asset price reactions to changes in the official repo rate (01.96–03.97) Effects of official interest rate on short-term interest rates (04.97–12.97) Effects of official interest rate on short-term interest rates (01.98–12.01) Asset price reactions to changes in the official interest rate (01.98–12.01) Effects of ECB interest rate on Czech short-term interest rates (01.98–12.01) Asset price reactions to changes in ECB interest rates (01.98–12.01) Confidence intervals of coefficients for official interest rate decisions before and during DIT Geographical and currency structure of Polish exports, 01.95 (% of total) Geographical and currency structure of Polish imports, 01.95 (% of total) FEER calculations for 2002 Sensitivity analysis of FEER calculations for 2002 Shares of Euroarea exports and imports in selected countries’ total exports and imports (%) Shares of value added by economic sector (%) Shares of employment by economic sector (%)
107 116 123 136 159 171 172 173 173 175 176 176 177 177 186 186 191 192 211 213 213
List of Figures 4.1(a) Czech Republic: actual and simulated adjustment 4.1(b) Estonia: actual and simulated adjustment 4.1(c) Hungary: actual and simulated adjustment 4.1(d) Poland: actual and simulated adjustment 4.1(e) Slovenia: actual and simulated adjustment 5.1 Scatter plots for averages of nominal forex model 6.1 Exchange rate and monetary policy, Slovenia 6.2 Exchange rate overshooting, Slovenia 6.3 BoS nominal interest rates (%) 6.4 Exchange rate dynamics during different regimes, Slovenia (year on year; January = 100) 6.5 Monetary aggregates and inflation, Slovenia 7.1 CBA cover (NFA and domestic liabilities of Eesti Pank, bn EEK) 7.2 Volume of EEK money market, CB’s forex window and market (bn EEK) 7.3 Money market rates in Estonia and in the Euroarea 7.4 Forex forward difference and interbank rates (Hansabank quotes) 7.5 Annual growth of Estonian banking sector domestic assets and liabilities vs annual change in banking sector NFA 7.6 Short-term interest rates and retail rates, Estonia 7.7 Estimated share of indexed loans in long-term loans to real sector, Estonia 7.8 Openness of the Estonian economy 7.9 Annual growth of Estonian exports and Finnish and EMU imports 7.10 The interest rate channel: adjustments after the shock, Estonia 7.11 The structure of real sector borrowing Estonia (bn EEK) 7.12 The credit channel: adjustment after the shock, Estonia 7.13 The exchange rate channel: adjustment after the shock, Estonia 8.1 CPI inflation for the Czech Republic and Germany (1993–2002) xii
73 74 75 76 77 91 104 108 110 112 114 133 138 138 140 142 143 145 145 146 152 154 156 158 168
List of Figures
8.2 8.3
9.1 9.2 9.3 9.4 9.5 10.1
10.2 10.3 10.4 10.5 10.6 10.7 10.8
Movement of official interest rates, CNB (January 1996–December 2002) Interest rate differentials between official interest rates (ECB and CNB) and five-year government bond (Germany and the Czech Republic) Measures of REER rates for Poland, 1994–2002 (% change, year on year) Nominal zloty–euro exchange rate vs relative PPP exchange rate, 1994–2001 (% change, year on year) Estimated BEER and actual REER, 1997–2001 (% change, year on year) Peseta–euro exchange rate (daily quotations) Escudo–euro exchange rate (daily quotations) Depreciation of the Hungarian forint against the relevant currency basket and the crawling band, January 1994–February 2003 CPI inflation in Hungary, 1993–2003 Exports plus imports as a percentage of GDP (2001) Cyclical component of industrial output in the Euroarea and in selected Accession Countries Progress in meeting the Maastricht criteria Budget deficit (ESA95 standard) Implied Euroarea entry dates for Hungary Survey of macro-analysts on Hungary’s Euroarea entry date
xiii
169
174 183 187 195 199 200
205 206 212 214 215 216 218 219
List of Abbreviations ADF AEG BEER BoE BoS BS CB CBA CD CEEC CIS CNB CPI CZK DEM DIT DOLS EBRD EC ECB ECM EEK EMU (V)ECM ERM EU EUR FEER FDI GDP HIPC HP IMAD IMF LoLR MF
Augmented Dickey-Fuller Test Augmented Engle–Granger test behavioural equilibrium exchange rate Bank of Estonia Bank of Slovenia Balassa–Samuelson effect central bank currency board arrangement certificate of deposit Central and Eastern European country Commonwealth of Independent States Czech National Bank consumer price index Czech koruna Deutschmark direct inflation targeting dynamic ordinary least squares European Bank for Reconstruction and Development European Commission European Central Bank error correction model Estonian kroon economic and monetary union (vector) error correction model exchange rate mechanism European Union euro fundamental equilibrium exchange rate foreign direct investment gross domestic product harmonized index of consumer prices Hodrick–Prescott filter Institute of Macroeconomic Analysis International Monetary Fund lender of last resort Mundell–Fleming model xiv
List of Abbreviations
MNB MRR MS MT MTM NBP NBR (E)OCA OECD OENB PEP PLZ PPI PPP REER SE SEK SDR SGP UIP USD VAR VAT
xv
Magyar Nemzeti Bank (Central Bank of Hungary) multiple rank regression member state Maastricht Treaty monetary transmission mechanism National Bank of Poland National Bank of Romania (endogenous) optimum currency area Organization for Economic Cooperation and Development Oesterreichische Nationalbank Pre-Accession Economic Programme Polish zloty producer prices index purchasing power parity real effective exchange rate standard error Swedish krona special drawing rights Stability and Growth Pact uncovered interest parity United States of America dollar vector auto-regression value added tax
Preface This collection of papers deals with a subject of utmost relevance for the future of the European integration experience, the current impending enlargement: from 1 May 2004 the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia and Slovenia – hereafter denoted as Eastern European Accession Countries (ACs) – will become fully fledged members of the European Union. One single country, Poland, is responsible for over 50 per cent of the total GDP of this group, and for more than 43 per cent of its joint population. Entry into the Common European Currency Area – the Euroarea – is a stated policy objective of virtually all the ACs. This is the specific subject that will be analysed in this book, from several different perspectives. Some of the chapters deal with pre-EU or EMU entry questions (Chapters 1 to 5), while others deal with prospective policy options that will be relevant on ERM II or Euroarea accession, and are, in this sense, forwardlooking (Chapters 6 to 10). Several of the chapters (Chapters 3, 4, 5 and 8) are based on papers produced for the EU PHARE–ACE project no. P981065-R, ‘Monetary and Exchange Rate Strategies related to the Current European Union’s Enlargement Process’, authored and managed by the two editors. The main conclusions of that project were: (i) the need to further strengthen, during the pre-accession phase, autonomous monetary policies based on direct inflation targeting and flexible exchange rates in countries such as Poland, Hungary and the Czech Republic; (ii) the need to apply inflation targeting strategies that are based on forward-looking and transparent rules, rather than on highly discretionary, backwardlooking and less predictable policies; (iii) the need to avoid considerably overvalued exchange rates on Euroarea entry; (iv) the need to account for the ongoing relative price adjustments consistent with the Balassa– Samuelson effects, while designing exchange rate strategies that are adequate to the individual countries. Chapter 1, by Peter Backé and Cezary Wójcik, provides a clear introduction to the subject, presenting the general EU framework for Euroarea accession for the future member states, and also a brief review of the pros and cons of Euroarea accession itself. After the overview provided in Chapter 1, in Chapter 2 Jarko Fidrmuc analyses in detail the OCA (optimum currency area) issues related to the ACs’ accession to EMU. In particular, aspects of the endogeneity xvi
Preface
xvii
hypothesis of OCA criteria are considered. The main finding is that it is intra-industry trade that induces the convergence of business cycles. This finding confirms the OCA endogeneity hypothesis, but it underlines the role of trade specialization. Furthermore, it estimates a comparable degree of business cycle harmonization for only some ACs (mostly Central European ones) with the EU. In Chapter 3 by Lúcio Vinhas de Souza and Elisabeth Ledrut a small expectations-expanded ‘Mundell–Fleming’ model for the ACs is provided. After constructing and estimating the model, the authors use it to assess the optimality of different exchange rate regimes (a peg and a float) through a simple welfare function. Floating appears the best option for most of the countries in our sample, and this conclusion is robust to changes in the weights of the welfare function. The ‘shockabsorbing’ qualities of the regimes for different types of innovations are furthermore assessed via a VAR (vector auto-regression) model, and here again the float seems to outperform a harder regime in the occurrence of temporary shocks. Similarly, Chapter 4 by Bas van Aarle, Joseph Plasmans and Bruno Merlevede estimates a small macroeconomic ECM (error correction model) for the ACs, which is then used for macroeconomic policy analysis and to simulate the effects of integration of ACs with the EU. In Chapter 5, Pieter van Foreest and Casper de Vries provide evidence, using a panel data set of ACs, current EU members and other non-EU benchmark countries, that the choice of the foreign exchange regime is not of first-order importance for achieving high output growth, as, given the forward-looking nature of the foreign exchange market, exchange rate stability hinges on the current and anticipated coherency of monetary and fiscal policies. In Chapter 6, Gonzalo Caprirolo and Vladimir Lavraˇc provide an extensive description of the history of monetary policy in Slovenia – one of the countries resulting from the break-up of the former Federal Republic of Yugoslavia – and also of the choices that it faces on the eve of EU accession. Chapter 7, written by Urmas Sepp, Martti Randveer and Raoul Lättemäe, deals with questions of fundamental importance for the effectiveness of any active monetary policy, the transmission channels, and, therefore, tries to foresee how Estonia, a currency board country where the main transmission channel is automatic, via the exchange rate and, therefore, in principle deprived of any active monetary policy, would perform within the Euroarea, where the main transmission channels are via interest rate decisions.
xviii
Preface
In Chapter 8, Roman Matoušek and Anita Taci analyse what assets prices can say about the evolving credibility of the monetary policy of the Czech Republic’s Central Bank (CNB), a monetary authority which operates a floating currency cum direct inflation targeting (DIT), using changes in the two-week repo rate on short- and long-maturity market interest rates. The main results are that the CNB’s monetary policy was credible both before and after the introduction of DIT, and that there is already a substantial degree of dependence on Euroarea monetary decisions. Chapter 9 by Łukasz Rawdanowicz deals with the very relevant policy question of the choice of the exchange rate parity on Poland’s entry to EMU. Using a behavioural equilibrium exchange rate (BEER) approach, he finds that the zloty–euro exchange rate in 2002 is not far from the level consistent with the current state of fundamentals, while using the fundamental equilibrium exchange rate (FEER) approach, it might require a reduced depreciation to be in line with the equilibrium level of fundamentals. Chapter 10 by Attila Csajbók presents – similarly to Chapter 6 on Slovenia – a history of the monetary framework of Hungary and discusses forthcoming policy choices of fundamental importance, with an emphasis on the speed of entry into the ERM II and the associated costs and benefits. The general message of the studies in this book is that entering the Euroarea – as much as EU accession itself – represents a substantial challenge for the ACs, but is a development that can also yield substantial gains for those countries, as all the previous enlargement waves have show. Also, there seems to be no analytical reason to treat any of the future member states as anything but ‘normal’ European market economies, while any notion of ‘transition’ gets less and less relevant as time passes. Lúcio Vinhas de Souza and Bas van Aarle Brussels, April 2003
Foreword The European Council, meeting in Copenhagen, in December 2002, reached a historical agreement according to which ten countries – Cyprus, the Czech Republic, Malta, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia and Slovakia – are to become new members of the European Union (EU) by 1 May 2004. These countries are very diverse in many respects. For example, the structure of their economies is different, the importance of their financial systems varies considerably, while their exchange rate regimes range from currency boards to inflation targeting together with exchange rate floating. Enlargement will increase diversity in the EU. Diversity lies at the heart of innovation, progress and vitality. The enlarged Union will be an example – of continental dimensions – of unity with diversity. All ten acceding countries have a GDP per capita well below the EU average, ranging from 30 per cent (Latvia) to 88 per cent (Cyprus). Successful membership will be accompanied by a process of economic catching up, characterized by the increase of productivity to the levels prevailing in the Union. Membership in the EU implies the acceptance of the acquis communautaire. It entails the need to conduct national economic policies in conformity with the principles of an open market economy with free competition, stable prices, sound public finances and monetary conditions and a sustainable balance of payments. All new member states are committed to participate in the Euroarea, in full respect of the criteria and procedures set out in the EU Treaty. This volume deals with some of the challenges and choices facing the new members in their process of integration into the EU. It focuses on the countries of Central and Eastern Europe. It is a very timely addition to the available research literature on this important set of topics.
Vitor Gaspar Frankfurt am Main, 14 April 2003
Vitor Gaspar is the Director of the Dirctorate-General (DG) Research of the European Central Bank. He is the former head of the Research Department of the Portuguese Central Bank and of the DG Studies and xix
xx Foreword
Forecasting of the Portuguese Ministry of Finance. He was also a professor at the FE-UNL in Lisbon, the institution from which he holds a Ph.D. in Economics. His research interests are concentrated on international macroeconomic and monetary issues. He has published widely in these fields.
List of Contributors
Peter Backé is an economist at the Central Bank of Austria, currently in secondment to the European Central Bank. He has published in the fields of fiscal, monetary and exchange questions applied to Eastern Europe. Gonzalo Caprirolo is currently State Undersecretary and Chief Economist at the Ministry of Finance of Slovenia. His research there includes topics such as capital inflows, monetary programming, current account sustainability, fiscal stance and debt sustainability, public sector debt and debt portfolio management, taxation, government accounts classification, the impact of pension reform on fiscal stance, and pensions reform. Before that, he worked for the UNDP as an adviser in the Central Bank of Bolivia, as a consultant for the IMF and at Mexican Ministries of Trade and Finance. He holds an MA from Columbia University (New York, USA) and a Master in Economics from El Colegio de Mexico (Mexico City, Mexico). Attila Csajbók is Deputy Head of the Monetary Analysis Division at the Economics Department of the Magyar Nemzeti Bank (MNB, Central Bank of Hungary). He was involved in various research activities covering topics such as zero-coupon yield curve estimation, extracting information from financial assets prices, public debt management issues, modelling the currency risk premium, and central bank reaction function estimations. More recently, his research interests are primarily issues related to the European integration process, including costs and benefits of Euroarea entry, the estimation of equilibrium real exchange rates and exchange rate regime choice. He has published a number of research papers on these topics. Mr Csajbók holds an M.Phil. in Economics from Cambridge University and an MA in Finance from the Budapest University of Economic Sciences. Casper G. de Vries is Professor of Economics at Erasmus University, Rotterdam, where he teaches monetary economics and financial risk management. He is a fellow and board member of the Tinbergen Institute, he directs the financial stochastics programme at the EURANDOM research institute, and he serves as a member of the EMU Monitor Group. His graduate training was at Purdue University; he has held positions at Texas A&M University, the Catholic University of Leuven, and he has xxi
xxii List of Contributors
been visiting scholar at several European and American research institutes. His research interests are focused on international monetary issues, such as foreign exchange rate determination and exchange rate risk, the issues surrounding the euro, financial markets risk and risk management. He has published widely in leading internationally refereed journals. Jarko Fidrmuc is an economist at the Central Bank of Austria, and formerly held the position of lecturer at the Institute of Higher Studies (IHS) in Vienna. He has worked extensively with subjects related to the integration of Eastern Europe in the EU. He has a Ph.D. in Economics from the University of Vienna. Raoul Lättemäe is an economist at the Bank of Estonia. He acquired his MA in Economics at the Tartu University in Estonia. His main research interests are the synchronization of economic cycles between EU and Accession Countries, the monetary policy transmission mechanism in Estonia and the determinants of sovereign credit ratings. Currently he is responsible for the forecasting and analysis of monetary and credit developments in Estonia. Vladimir Lavraˇc is a senior research fellow at the Institute for Economic Research in Ljubljana. The main areas of his research interest are international finance, international monetary relations and monetary integration process. He coordinated two ACE–PHARE research projects on the inclusion of Central European countries in EMU. He worked as consultant to the European Commission, the World Bank and ICEG. He has published widely, including journal articles, contributions to conferences and chapters in books. Elisabeth Ledrut is an economist at the Central Bank of the Netherlands. She has a degree in Economics from the Erasmus University in the Nertherlands, and has also worked as a consultant for the European Parliament and for EU PHARE projects. Roman Matoušek is a former economist at the Central Bank of the Czech Republic, and currently an associate lecturer at the London Guildhall University. He has a Ph.D. in Economics from the Charles University in Prague, Czech Republic. Bruno Merlevede is a researcher at the Faculty of Economics of the University of Ghent, Belgium. He holds an MA from the Catholic University of Leuven. His research interests concern transition economics and various fields in macroeconomics.
List of Contributors
xxiii
Joseph Plasmans is a professor at the Faculty of Applied Economics of the University of Antwerp (UFSIA), Belgium. He holds a Ph.D. from Tilburg University, the Netherlands. His research is in the areas of econometrics, multi-country modelling, dynamic games, European integration, transition economies, intangible investments and innovation strategies. He has collaborated in many research projects in these fields and (co)authored a large number of studies. He is also affiliated with Tilburg University and CESifo. Łukasz W. Rawdanowicz is an economist at the Centre for Social and Economic Research (CASE), based in Warsaw, Poland. He holds an MA in International Economics from Sussex University (UK) and an MA in Quantitative Methods from Warsaw University (Poland). His main area of interest is applied international macroeconomics. He works with issues related to trade liberalization, currency crises propagation, exchange rate misalignments, and exchange rate regimes. He has also been involved in macroeconomic forecasting for the Polish economy and in analyses of world economic developments. Martti Randveer is Head of the Research Department at the Bank of Estonia. He acquired his MA in Economics at the Tallinn Technical University in Estonia. He has published on nominal and real convergence, sustainability of the currency board, monetary policy transmission in Estonia, inflation and equilibrium exchange rates. He is currently working on issues related to fiscal policy automatic stabilizers and consumer behaviour in Estonia. Urmas Sepp was Head of the Research Department at the Bank of Estonia from 1993 until his death in 2002. Before joining the Central Bank he worked in the Institute of Economy at the Estonian Academy of Sciences, and was a professor of Economics at the Tallinn Technical University. He actively participated in several governmental working committees in the newly independent Republic of Estonia at the beginning of the 1990s. His main research interests were macroeconomic modelling, inflation, economic growth, productivity and monetary policy, and he has published widely on these issues. Anita Taci is an economist at the Chief Economist Office of the European Bank of Reconstruction and Development in London. She has a Ph.D. in Economics from the CERGE-EI, at the Charles University in Prague, Czech Republic.
xxiv List of Contributors
Bas van Aarle is a postdoctoral researcher of the FWO (Fund for Scientific Research, Flanders) at LICOS, Catholic University of Leuven. He has several publications in the fields of fiscal and monetary policies. He has a Ph.D. in Economics from Tilburg University in the Netherlands. He is also affiliated with the University of Nijmegen and CESifo. Pieter W. van Foreest holds a post-doctoral position at Erasmus University, Rotterdam, The Netherlands. He is involved in the project ‘The Risk of Financial Débâcles and the Value-at-Risk Measure’ funded by the Netherlands Organization for Scientific Research. He has a Ph.D. in Economics from the Erasmus University, Rotterdam. His research interests are focused on financial economic issues. Lúcio Vinhas de Souza is currently an economist at the Kiel Institute for World Economics (IfW) in Germany, and worked previously as an Associate Economic Affairs Officer at the United Nations, as a Visiting Fellow at the European Commission and at the ECARES-Free University of Brussels, and as a Visiting Researcher at the Central Banks of Estonia and Germany. He also worked as a consultant for the European Parliament, for the World Bank–Global Development Network, for the USAID and for EU PHARE projects. His most recent research deals mainly with the monetary and fiscal aspects of the accession of Eastern European countries to the EU. He has a Ph.D. in Economics from the Tinbergen Institute in the Netherlands. Cezary Wójcik is an assistant professor at the Faculty of Economics at the University of Warsaw, Poland, and a former economist at the Research Department of the Central Bank of Austria. He has published widely in the fields of monetary and exchange questions applied to Eastern Europe.
Part I General Policy Issues for the Accession Countries
This page intentionally left blank
1 Introduction – Monetary Integration of Central and East European Accession Countries: The Pros and Cons of Speedy versus More Gradual Strategies Peter Backé and Cezary Wójcik
1
Introduction
Monetary integration in the European Union (EU) has advanced very far, based on the Maastricht Treaty which outlined a staged approach towards the creation of an economic and monetary union (EMU). Within this framework, eleven EU member states formed a monetary union in 1999 and the euro was introduced as the single currency in this newly formed monetary area. In the meantime, the Euroarea has been enlarged to twelve countries and, furthermore, the three remaining EU member states which have not yet adopted the single currency have increasingly been attracted to the Euroarea, in particular by the successful introduction of euro coins and banknotes at the beginning of 2002. Against this backdrop, the final destination of monetary policy and integration for the Central and Eastern European Accession Countries (ACs) is obvious: at some point in the future, the ACs are set to join the Euroarea. The European Union has outlined a three-step approach to the monetary integration of ACs. The applicants will first join the EU, then enter the Exchange Rate Mechanism II (ERM II) of the European Union and finally, after fulfilment of the Maastricht convergence criteria, accede to the Euroarea, i.e. participate fully in economic and monetary union. This means that the euro is to be introduced in today’s ACs in a consensual manner, based on the standard convergence examination procedure and
3
4
General Policy Issues for the Accession Countries
not sooner than at least two years after EU accession. The latter aspect results, in particular, from the convergence criteria which include, as a legal requirement for the adoption of the euro, a two-year participation in the ERM II, without a devaluation of the parity rate against the euro during this period (see ECOFIN, 2000). ERM II, in turn, is solely open to EU Member States and participation in this mechanism can thus begin only after accession to the European Union. The EU has made it clear that introduction of the euro as legal tender in an Accession Country without the consent of the European Union is not an appropriate way to move ahead towards full monetary integration with the Euroarea. The main argument is that such a unilateral euroization ‘would run counter to the underlying economic reasoning of EMU in the [EC] Treaty, which foresees the eventual adoption of the Euro as the endpoint of a structured convergence process within a multilateral framework’ (ECOFIN, 2000). In taking this stance, the European Union was reacting to a discussion that gained momentum in 1999 and early 2000 about whether a fast unilateral introduction of the euro in the ACs would be economically more advantageous, in particular for the candidates, than the monetary integration path outlined by the European Union.1 Given the position of the EU, it is evident that such a rapid or even immediate adoption of the euro, if it were to occur, could only be done unilaterally by the Accession Country which would – hypothetically – opt for such an approach. Backé and Wójcik (2002) argue that the institutional considerations that are usually invoked to validate the European Union’s objection to unilateral euroization by ACs are supported by economic considerations: unilateral euroization would bring fewer benefits than joining the Euroarea on the standard pathway (no seigniorage revenues, no lender of last resort function under the former option). At the same time, the adoption of a foreign currency as legal tender, while eliminating the risk of exchange rate crises, does not do away with the balance of payments constraint. In addition, if nominal convergence is not far advanced at the point in time of the introduction of the foreign currency, major swings in real activity may ensue, with negative knock-on effects on financial stability and catching-up perspectives. Another point the EU has underlined is that the Maastricht convergence criteria will not be changed for the ACs and that these criteria will be applied in the same manner as in the convergence examinations so far. This is to ‘ensure equal treatment between future Member States and the current participants in the Euroarea’ (ECOFIN, 2000). Again, the European Union’s emphasis on this point has been a response to a debate on
Introduction
5
whether the Maastricht criteria should be adapted for Accession Countries, mainly to take into account that these countries have embarked on a catching-up path towards average EU income-per-capita levels.2 The ACs have essentially accepted the EU position in the course of the accession negotiations, and the negotiation chapter on economic and monetary union has been concluded with all Accession Countries (except Romania, where it has not yet been opened). Discussions about rapid unilateral euroization or an adaptation of the Maastricht criteria have not impinged upon the basic policy line to stick, in principle, to the three-stage approach proposed by the European Union. Awareness about the risks and costs of a rapid and unilateral introduction of the euro has increased in the ACs, and there is a growing perception in the candidate countries that the predominantly nominal Maastricht convergence criteria form an acceptable set of criteria to qualify for monetary union, in particular if compared with other potential sets of benchmarks which may have included real convergence requirements. This being so, the most important issue for the Accession Countries to decide is whether to aim for an early introduction of the euro two or three years after EU accession or to opt for a more gradual strategy of monetary integration. This chapter reviews the arguments for and against each of these approaches. Thus, the focus is on the question of how speedily to introduce the euro in the ACs, within the standard path laid down in the EU accession negotiations. This issue of what is the appropriate pace of monetary integration is being intensely discussed in the ACs. Several countries (see the respective country studies in this book), notably Slovenia and Estonia, are aiming to join the Euroarea as soon as possible after EU accession. In Hungary, the authorities adopted a joint strategy in 2001 targeting Euroarea entry in 2007 but, after a change in government in 2002, the commitment of the new cabinet to the strategy is somewhat ambivalent. In the case of Poland, a working group consisting of government and central bank representatives elaborated a common strategy document in October 2002 declaring the intention to enter the Euroarea as soon as possible (implicit target date 2007). It is, however, not fully clear how binding this document is. In the other countries, the decision-making process is still evolving. Within this group, the central banks of Slovakia, Latvia and Lithuania are more or less clearly leaning towards going for swift participation in the Euroarea after having joined the European Union, but an official policy statement on the issue (accorded between the government
6
General Policy Issues for the Accession Countries
and the central bank) has not yet been released. In the Czech Republic, the central bank also has a preference for a relatively speedy adoption of the euro (around 2007), while the government is in favour of a more gradual approach.3 What are the issues that arise in assessing the merits and disadvantages of (relatively) fast monetary integration versus a more gradual approach? Joining a monetary union holds considerable potential benefits, but also substantial potential risks if undertaken prematurely. However, there are severe limitations to making an economic cost–benefit analysis of a country’s participation in a monetary union and, even more so, in using cost–benefit analysis for determining the optimal speed towards full monetary integration. This is so mainly for two reasons. First and generally, there is no uniformly accepted basis among economists for assessing the costs and benefits of joining a monetary union. Second and more specifically, in the case of the ACs, there is no satisfactory model to estimate all relevant effects jointly within a unified framework. In this latter respect, the situation in the ACs differs from the state that prevailed in the incumbent EU member states when they did their cost–benefit assessments in the run-up to the creation of the Euroarea.4 Against this backdrop, a pragmatic approach is to focus on the most important effects only and to assess these factors individually. In doing so, the key factors which have to be discussed are the costs and benefits of giving up the monetary and exchange rate instrument, trade and growth gains and credibility effects. Other aspects, for example the role of monetary integration as a potential catalyst to structural reforms, are also relevant but appear, in overall terms, to be less central and are therefore not discussed further.5 In the analysis, a dynamic perspective has to be taken which considers how the effects change over time. This is particularly important if the costs of full monetary integration tend to decrease over time, as structural convergence proceeds. Based on this line of reasoning, what has to be assessed is at what point in time the costs and risks of full monetary integration are sufficiently contained so that they are outweighed by the benefits of participation in the Euroarea. Obviously, the downside of this approach is that it essentially neglects possible interlinkages among the single effects and that it takes a very simplistic line on the aggregation of individual factors. Still, these shortcomings have to be accepted, as there is apparently no other feasible approach at this stage. To round out the analysis, the discussion of the most important medium- to long-term costs and benefits of participating in a monetary union must be complemented by an examination of the short-term
Introduction
7
effects that emerge as ACs move towards meeting the entry conditions for participating in the Euroarea. The chapter is organized as follows. The next section examines the main medium- to long-run effects of a future participation in the Euroarea. In this section both the cost side and the benefit side are explored. Section 3 discusses potential short-term implications of the fulfilment of the Maastricht convergence criteria, with a particular focus on fiscal issues, which appear to be most critical in this context. Section 4 summarizes the main conclusions.
2 Participating in the Euroarea: medium- to long-term effects 2.1 The cost side The diversity of views among economists about the potential effects of joining a monetary union is particularly pronounced for the cost side. The standard approach to assess the costs of the adoption of a foreign currency as a legal tender is, or has until recently been, the optimum currency area theory (OCA theory). The OCA theory considers a common currency optimal for countries which are exposed to mainly symmetric shocks or which have mechanisms in place for adjustment to asymmetric shocks. The latter include, according to the theory, wage and price flexibility, factor mobility and/or fiscal transfers. The smaller the exposure to asymmetric shocks, the less need there is to resort to such adjustment mechanisms. In order to lower the probability of asymmetric shocks, it is crucial that the trade of participating countries is highly integrated and that their exports are well diversified in terms of the structure of exported goods and services, which in turn will contribute to fostering business cycle synchronization.6 However, OCA theory, which was long ‘the organizing framework’ (Eichengreen, 1997) for the analysis of monetary unification has recently met with increasing criticism within the economists’ profession, mostly on three grounds. The first argument is that the OCA criteria are endogenous. Frankel and Rose (1998) maintain that joining a currency union (or a credible fixing of the exchange rate) will eliminate exchange rate uncertainty and reduce currency transaction costs, which will stimulate bilateral trade and hence deepen the economic integration between trade partners. This will foster business cycle synchronization and reduce the exposure to asymmetric
8
General Policy Issues for the Accession Countries
real shocks, which in turn will validate (ex post) the adoption of the common currency. Second, risk-sharing arguments suggest that, under full financial market integration, countries that are exposed to asymmetric shocks may profit from monetary unification. The idea is that using a common currency will facilitate portfolio diversification, which allows countries to adjust more smoothly and at lower costs to asymmetric real shocks, due to mutual claims on each other’s resources. This view was first put forward by Mundell (1973) and has increasingly been echoed in the recent debate on the OCA theory (see McKinnon, 2001; Buiter, 2002). A third proposition is that the exchange rate tends to be a source of shocks rather than a shock absorber, in particular for small open economies. Thus, even if there were a potential for asymmetric real shocks to occur, the exchange rate would either be ineffective as an adjustment tool and/or any beneficial effects from retaining it may be more than offset by the costs caused by nominal exchange rate volatility and, in the worst instance, exchange rate crises (see for example Buiter, 2000). This stance, which can be traced back to Friedman (1968), challenges the view that structural considerations are important for the choice of exchange rate regime – a view that at least implicitly assumes that monetary and exchange rate policy is an effective tool of economic policy. This criticism of the OCA theory also features in the ‘fear of floating’ literature (see Hausmann et al., 1999; Calvo and Reinhart, 2000), which essentially argues that emerging market economies cannot effectively utilize the nominal exchange rate to absorb shocks from abroad – due to credibility deficits, a strong inflation pass-through of exchange rates and/or widespread currency substitution. How valid are these arguments and, consequently, how relevant does the OCA theory remain as a tool for assessing the costs of monetary integration? The endogeneity of the OCA criteria appears to be fairly well established by the recent empirical literature. Still, reliance on endogeneity should not be taken too far when evaluating policy choices about monetary strategies and monetary integration, for at least two reasons. First, the endogeneity proposition may not hold in each and every case. In extremis, the effects may even go into the opposite direction. Krugman (1993) develops a theoretical model which shows that more trade due to the use of a common currency could result in countries becoming more specialized in the goods in which they have a comparative advantage. As a result, the sensitivity of countries to industry-specific shocks could increase and business cycles could become less synchronized. A second
Introduction
9
and probably more important caveat is that it may take a long time for the endogeneity to work its way through the economic system. The experience of the Euroarea since 1999 is a case in point. Gaspar and Mongelli (2001) conclude that ‘looking at the matrix of intra-Euroarea trade, such integration effects have not (yet?) become apparent’. Thus the transition period to the new equilibrium in which the potential for external shocks would become much smaller may well be fairly lengthy. During the intermediate period, the exposure continues to persist (or goes down only very gradually) and adjustment mechanisms remain particularly important. At the same time, it is notoriously difficult to increase an economy’s adjustment capabilities quickly and, thus, to reduce the exposure to shocks in the transition period to the new steady state. The second argument against the traditional OCA theory, relying on risk-sharing considerations, presupposes a complete portfolio diversification in order to be effective. While a common currency removes one obstacle to diversification, there are other factors that make for a home bias. Market segmentation tends to be nurtured by national borders, and the full harmonization of regulations on financial services is an arduous process, as the EU experience shows. Differences in tax laws, difficulties in assessing credit risks adequately (partly due to divergent insolvency laws) and the not-yet-completed consolidation of financial infrastructure (in particular settlement systems) constitute further barriers to full integration in the Euroarea. A recent ECB study shows that diversification in the Euroarea has increased only slowly and in a limited manner since 1999 (see ECB, 2001).7 While Accession Countries have made major strides in aligning their regulations on financial services and capital movements to EU standards, full integration and thus a substantial degree of diversification will only be reached in the medium to longer term (partly also due to transition periods for some ACs, e.g. limitations for domestic pension funds to invest abroad). Risk-sharing arguments, therefore, do not alter the cost–benefit equation substantially at this stage or in the near future. The debate on the merits and costs of retaining – or removing – the monetary and exchange rate policy instruments has led to a perception that in small open economies monetary and exchange rate policies cannot be effectively used to smooth cyclical fluctuations. On the other hand, there are benefits of retaining the exchange rate as a policy instrument to correct major exchange rate misalignments in cases when adjustment through wages and prices would be much more costly due to the presence of rigidities. In other words, there are advantages of an escape option in a period of substantial distress.8 On this line of
10
General Policy Issues for the Accession Countries
reasoning, removing the exchange rate instrument irrevocably therefore presupposes that any major risks and sources of potential exchange rate misalignments are sufficiently contained. The effectiveness of using the exchange rate instrument in exceptional circumstances to facilitate adjustment hinges to a large extent on the consistency and soundness of the overall policy compound a country has pursued. Backé and Wójcik (2002) argue that advanced Accession Countries have established a solid track record in terms of stabilization and reform which may facilitate the effective use of the exchange rate if a major asymmetric real shock hits. Moreover, misalignment risks in ACs should not be underrated. The completion of price liberalization and adjustments of regulated prices but also the upward adjustment on agricultural prices due to the prospective integration into the EU’s common agricultural policy may lead to price–wage spirals. Furthermore, demand-side effects associated with the catching-up process may affect the competitive position of a country, in particular if they lead to additional wage pressure in the tradables sector or if investment shifts to the non-tradables sector (see Wójcik, 2001). This leaves the flipside of the argument, namely that the exchange rate can be a source of shocks. How relevant this issue is for the Accession Countries will be examined below in the discussion of the benefits of monetary integration. The overall conclusion on this issue is that the OCA theory still has some validity in assessing costs and thus a fair weight must be given to OCA-related considerations and conclusions in drawing up an overall cost–benefit equation. What is the empirical picture in the Accession Countries of Central and Eastern Europe with respect to the OCA criteria? First, as regards the susceptibility of ACs to asymmetric shocks, several, though not all, Accession Countries have already achieved a considerable degree of business cycle synchronization, at least in the area of industrial production (see, e.g., Fidrmuc and Schardax, 2000). However, if one goes a step further and assesses the likelihood of asymmetric shocks by examining the correlation of supply and demand shocks between countries of the Euroarea and the Central and Eastern European Accession Countries, a less encouraging picture emerges (see Fidrmuc and Korhonen, 2001; Horvath, 2001b). The latter study, for example, in which shocks are recovered from estimated structural VAR models of output growth and inflation, finds that only Hungary, Estonia and, to a somewhat lesser extent, Poland display positive correlations of demand and supply shocks with the Euroarea in the period 1992/95 to 2000/01. Horvath (2001b)
Introduction
11
examines the correlation of supply and demand shocks of Central European and Baltic Accession Countries with four large EU economies for 1993/95 to 2000 and arrives at somewhat different but not very robust conclusions, with Hungary and Slovenia displaying the relatively highest correlations for both types of shocks. Thus the picture is diverse: Some ACs – Hungary, perhaps also Estonia, Slovenia and Poland – show positive correlations; the others do not. Furthermore, caution is warranted when drawing conclusions from these results, in particular if one considers that the correlations for some Euroarea countries like Greece and Ireland are not encouraging either. As for the other side of the OCA coin, i.e. the functioning of adjustment mechanisms, product and labour markets display considerable variation among candidate countries, and this is particularly true for the wage formation process and wage flexibility. No comprehensive empirical study appears to exist which undertakes an in-depth assessment of the functioning of product and labour markets in all ten Accession Countries.9 In very general terms, it seems to emerge from the limited analytical body available that product markets in ACs tend to function somewhat less efficiently than those of EU countries, while the Accession Countries’ labour markets tend to be more flexible than those of the member states of the European Union (see IMF, 2000). Whether migration is an effective channel in Accession Countries for adjusting to idiosyncratic shocks is rather doubtful (see Fidrmuc, 2002). The same is true for the question of whether fiscal transfers can play a major role in easing asymmetric shocks. It should be noted, however, that these channels do not play a major role within the current Euroarea either. Finally, capital flows may also facilitate adjustment in the short run but capital mobility cannot solve the adjustment problem in the long term, i.e. if there are persistent external imbalances, as there are limits to negative net wealth positions of countries vis-à-vis the rest of the world (see Corden, 1973). In sum, one can differentiate among Accession Countries which have made substantial advances towards meeting the OCA criteria – and, in a few cases, progress appears to be quite similar to that of some southern and non-continental EU member states – while others have moved ahead less. Thus, based on the OCA theory, a diverse picture emerges, with considerable risks for a number of Accession Countries. However, this is only a static snapshot. In a dynamic perspective, the correlation of shocks will probably increase with a further deepening of trade and financial integration in the run-up to membership in the
12
General Policy Issues for the Accession Countries
European Union and beyond. The inclusion into the EU internal market will lower real trade costs and thereby foster trade. Financial integration will be nurtured by improved confidence and reduced uncertainty associated with EU accession. By a similar token, further reforms of product and labour markets, again in the EU membership context, will tend to increase the adjustment capabilities of Accession Countries. On the other hand, EU accession itself may constitute an asymmetric real shock for some candidates, giving rise to adjustment processes as a consequence of the full integration into the EU internal market during the early stages of membership in the European Union.
2.2 The benefit side Moving to the benefit side of monetary integration, there are three major advantages. First, participation in monetary union eliminates the risk of exchange rate crises. This is particularly relevant for cases of sudden shifts in sentiment leading to abrupt stops or reversals in capital flows and thereby to currency crises. Second, monetary integration generates trade and growth gains which are driven by lower transaction costs and reduced uncertainty. Third, the prospect of joining a monetary union can have positive credibility effects. When putting the first benefit, the elimination of the risk of currency crises, into perspective, two points emerge. First, assessing the risk of future exchange rate crises is notoriously difficult, if not impossible. A widely shared view is that the risks of excessive capital inflows and sudden capital flow reversals can be mitigated by sound macroeconomic policies, by avoiding ‘soft’ exchange rate pegs, by measures that strengthen financial institutions’ risk-management capabilities and by supervisory activities concerning the financial sector and the foreign borrowing of the corporate sector. However, despite such measures, significant risks of nominal exchange rate shocks that are unrelated to any change in fundamentals may remain. There are different ways to cope with this risk. One is monetary and exchange rate policy cooperation within the European Union upon accession. More specifically, ERM II can, in principle, contain such risks if it is operated in a way which provides reasonable shelter from speculative attacks that are not related to changes in fundamentals, i.e. if the mechanism puts off ‘unjustified’ capital flow reversals for those economies that are basically healthy in terms of their fundamentals. It could also be considered to complement existing arrangements by establishing an additional financial facility with automatic access for
Introduction
13
non-Euroarea member states of the European Union that have a straight record within intra-EU economic policy coordination and surveillance. Second, joining a monetary union per se does not contain the risk of financial crises other than exchange rate crises. On the one hand, this underlines how essential financial sector soundness and supervision are. On the other hand, it points to the crucial importance of achieving a high degree of nominal convergence, as embodied in the Maastricht convergence criteria, before adopting a common currency. If progress with nominal convergence were not sufficiently advanced, boom–bust cycles could develop (see Backé and Wójcik, 2002). Such cycles are often associated with banking crises emerging in the bust phase and also with a less dynamic GDP-per-capita convergence over the full cycle. The second main benefit of monetary union relates to trade and growth gains. Until recently, these effects were thought to be relatively modest, based on a string of empirical research applying time-series methods. During the last two years, a new strand of papers relying on panel data methods has questioned this view. The debate was kicked off by Rose (2000), who finds that the trade effects of using a common currency are statistically significant and huge: countries with a common currency are found to trade over three times as much as countries using different currencies. Moreover, Rose concludes that the impact of a common currency is an order of magnitude larger than the effect of reducing moderate exchange rate volatility to zero but retaining separate currencies. Frankel and Rose (2000) also find that potential benefits from the use of a common currency on trade to be large and, moreover, that this additional trade has substantial positive effects on growth. Subsequent studies by Rose and Van Wincoop (2001), Melitz (2001) and Persson (2001) look further into the impact of using a common currency on trade and arrive at considerably lower but still large positive effects, with trade expanding, according to most estimates, by 40 per cent to 50 per cent. Still, the currency union effect on trade remains a highly contested issue (see Honohan, 2001; Nitsch, 2002). Vinhas de Souza (2002b), on the other hand, looking at the early Euroarea experience, finds negative or non-significant effects. This may suggest that participation in a monetary union holds potential trade and growth gains, although there is limited knowledge on the issue of to what extent these effects vary among countries participating in a monetary union. And what is unknown is the time profile, i.e. how quickly these effects will materialize, as discussed above for the case of the Euroarea.
14
General Policy Issues for the Accession Countries
The third benefit of joining the Euroarea pertains to credibility effects. The argument is that joining a monetary union solves credibility problems of monetary authorities that stem from the dynamic inconsistency problem and thus eliminates a potential inflationary bias.10 These credibility gains – together with the reduction of the interest risk premium, due to the elimination of exchange rate risk, and with deepening financial market integration – lead to a reduction of real interest rates which in turn stimulates investment and spurs growth. Evidently, the significance and the size of these effects depend on the degree of credibility a country’s policies enjoy in the first place, i.e. before it engages in a monetary unification process. In this context, two aspects that relate to the Accession Countries deserve particular attention. First, most ACs have made substantial headway towards achieving macroeconomic stability. As a result, the credibility of the monetary authorities and the confidence in the national currencies have been on the rise, whereas inflation has been on a firm falling path. It is obvious that the prospects of EU integration have played and will continue to play a fundamental role in this respect. The external constraints that result from fulfilling the conditions for EU accession are helping to solve the commitment problem of monetary and fiscal authorities and constitute an anchor for macroeconomic discipline, but also for institution-building/reinforcement and for structural reforms. In particular, preparing for EU accession has fostered the creation of domestic institutions dedicated to price stability, as legal provisions on central bank independence have been strengthened substantially. Cukierman et al. (2001) as well as Dvorsky (2000) show that the legal independence of central banks in ACs is well developed. Actual membership in the European Union and, in particular, participation in economic policy coordination and surveillance will further enhance the credibility of Accession Countries’ macroeconomic policies and, in general, eliminate any significant inflation bias of monetary policy. A coherent and thoroughly implemented strategy of joining the Euroarea and, subsequently, participation in the monetary union will further add to this, mostly by consolidating the credibility gains reaped at the earlier stages. The implications of this discussion of credibility issues for the speed of monetary integration are not straightforward. In essence, credibility is largely endogenous to the soundness and consistency of the overall economic policy mix over time. Whether the pace of monetary integration has an impact on the quality of the policy compound is a question that
Introduction
15
can hardly be answered ex ante. There may be cases where a speeding up of monetary integration (e.g. setting an ambitious target date) will reinforce a virtuous circle of improving economic fundamentals and credibility. Conversely, if the policies pursued are perceived to be or become inconsistent with the pace of monetary integration intended by the authorities, credibility will most probably suffer. All this suggests that credibility effects of EU and subsequent Euroarea accession are important; however, it is uncertain whether the pace of monetary integration does affect the build-up of credibility and thus the time profile along which the related benefits can be reaped.
3 Meeting the requirements of Euroarea entry: short-term implications The actual timing of Euroarea entry depends on the point in time when the Maastricht convergence criteria are met. The fulfilment of these criteria poses several challenges to ACs. In this chapter the focus is on the most critical issue in this respect, namely the fulfilment of the budget deficit criterion by those ACs which currently record major fiscal imbalances. This point is particularly relevant for the Czech Republic, Hungary, Poland and Slovakia, whose 2002 budget deficits ranged from more than 4 per cent to 8 1/2 per cent of GDP. This aspect should be included in the cost–benefit analysis since fiscal consolidation may imply non-negligible costs – or benefits – in terms of output and employment. Against the backdrop of present imbalances in some ACs, an important issue is whether such costs (or benefits) depend on the speed of budgetary tightening. Upon EU accession, new member states will participate in the European Union’s economic and fiscal policy coordination and surveillance framework (EU Treaty provisions, Stability and Growth Pact). In essence, this implies that, as EU members, today’s ACs will have to avoid excessive public sector deficits which, according to the EU rules, are defined as deficits of more than 3 per cent of GDP. Subsequently, during the first few years of membership in the European Union, further progress will have to be made to ensure that actual deficits remain below the 3 per cent threshold also in times of ‘normal’ economic downturns, i.e. that the cyclically adjusted budget balances are compatible with the 3 per cent threshold for actual deficits also during belowpotential growth periods. (For outright recessions, EU rules include escape clauses.)
16
General Policy Issues for the Accession Countries
For ACs which intend or consider introducing the euro soon after EU accession, fiscal issues are becoming particularly critical, not only because of the fulfilment of the Maastricht fiscal criteria but also because of the much more stringent mechanisms to enforce compliance with fiscal rules for Euroarea participants.11 EU membership not only brings new fiscal policy rules but also has important direct and indirect effects on the fiscal positions of ACs. Backé (2002b) explores this issue in detail and concludes that in the short run, membership in the European Union will presumably add somewhat to the fiscal strains on ACs, mainly due to additional expenditures (contributions to the EU budget and possibly cofinancing of EU structural operations). In the medium run, as positive growth effects of EU membership kick in, the overall impact on the ACs’ fiscal accounts can be expected to be broadly neutral or slightly positive. Some uncertainty prevails, though, on the magnitude of several individual effects, in particular on future public investment needs. In the context of the economic policy dialogue with the European Union, the ACs have tabled Pre-Accession Economic Programmes (PEPs) in 2001 and in 2002 which include fiscal frameworks that extend to 2005 and, in some cases, 2006. Backé et al. (2002) present an analysis of these programmes and, in particular, their consolidation trajectories as well as the risks to these strategies. The main findings can be summarized in four points. First, a comparative analysis of the fiscal strategies of the ACs shows, by and large, an overall picture of fiscal restraint and consolidation: countries with modest deficits intend to preserve their positions or even improve to full balance, while most countries which currently record higher deficits intend to reduce their deficits substantially until 2005. As a consequence, the highly divergent budget balance figures are to become much more similar by the middle of the decade, falling mostly into the moderate deficit range. There are some qualifications to this general assessment: the Czech Republic constitutes a certain exception, aiming only at a moderate reduction of relatively high deficits. Poland intends to consolidate substantially only towards the end of the framework period. Latvia’s deficit is to rise in 2003 and consolidation will set in only in the two years thereafter, resulting in a somewhat higher (but still relatively moderate) deficit/GDP ratio in 2005 compared to 2001 and 2002. In these three country cases, this less ambitious stance is also reflected in rising public debt/GDP levels.
Introduction
17
Second, according to the PEPs, improvements in budget balances are to be achieved through reductions in the expenditure/GDP ratios; these ratios are to be reduced even further than budget deficits, since revenue/GDP ratios are also projected to fall to some extent in most countries, with the exception of the Czech Republic, as a result of tax reforms and, in some cases, reductions in social security contributions. Third, a more detailed analysis of the most important expenditure categories – public investment, subsidies, social and welfare expenditures, public consumption and interest expenditures – shows that consolidation strategies differ in important respects. Public investment expenditures are to be reduced substantially in Hungary, Poland and Slovakia. In particular, Poland and Slovakia’s public investments are to be reduced to very low levels, raising doubts about whether such cuts can be fully sustained in the accession and catching-up context. On the other hand, public investment expenditures relative to GDP are projected to remain comparatively high in the Czech Republic and, to a lesser extent, also in Estonia. Subsidies are projected to fall, as a percentage of GDP, in Bulgaria, the Czech Republic and Hungary, thereby contributing to consolidation, while reductions in other countries will be minor. Nevertheless, subsidies in Hungary remain relatively high. The latter is also true for Poland, and even more so for the Czech Republic. Kopits and Székely (2002) argue that reduced spending in the area of production subsidies could make a substantial contribution to fiscal consolidation in most ACs, provided that the respective reforms and restructuring measures are taken.12 Social and welfare expenditures are very different in size among ACs, being much larger in Central Europe than in Baltic and South Eastern European countries. This is mainly due to different levels of social transfers and benefits, which currently range from 11 per cent to 27 per cent of GDP in the ACs. Poland is the only country which explicitly targets a tangible reduction of overall social and welfare expenditures relative to GDP, from a high initial level, which is apparently to be achieved by keeping real growth of such expenditures below real GDP growth. However, most of the adjustment is to come only in 2004 and 2005. Hungary and Slovenia target more modest reductions of expenditure/GDP ratios in this area.13 In Slovakia, social expenditures are to increase relative to GDP until 2003 before declining again thereafter, so that spending will be marginally lower in 2005 than it was in 2002. The other Central and Eastern European EU Accession Countries aim at broadly unchanged social expenditures relative to GDP between 2002 and 2005.
18
General Policy Issues for the Accession Countries
Backé et al. (2002) argue that the intended reshuffling of the expenditure side of Central and Eastern European budgets cannot spare social and welfare expenditures, in particular in those countries which have to lower their deficits substantially. They suggest that this should primarily be done by keeping real growth of social and welfare expenditures below real GDP growth: after successful disinflation in most ACs, there should be more room for such an approach in the current environment of low inflation in which it may become easier to resist or dismantle (quasi-)automatic indexation pressures in this area. Furthermore, they underline that the scope for spending restraint in the area of social expenditures eventually depends on the societies’ preferences regarding the size of the welfare state. In this respect, it appears that there is a pronounced desire, in particular in Central European countries, for a strong welfare state, which cannot be entirely disregarded when designing fiscal consolidation strategies. Practically all ACs target a reduction of public consumption/GDP ratios during the period 2002 to 2005, though to different degrees. Public sector wage moderation and reduction in public sector employment may, in theory, be an effective tool to achieve consolidation, but in the rather particular Central and Eastern European setting this does not seem to be a major feasible tool for achieving a lasting improvement in public sector balances. As Backé et al. (2002) note, the share of the wage bill in total expenditures will rather have to increase than decrease in most ACs, given the currently low relative wages of public servants and the need to improve administrative capacity to ensure adequate implementation of the EU acquis communautaire. The analysis further shows that consolidation perspectives are determined, to a tangible extent, by interest expenditures for a number of ACs. It also highlights that the interest expenditure effects of fiscal strategies that imply rising public debt/GDP levels (which is the case for the Czech Republic, Latvia and Poland) are by no means negligible. Fourth, structural fiscal balances seem to have worsened recently in a number of ACs, due to discretionary measures. This does not facilitate the tasks ahead until 2005. Relatedly, it appears that the size of fiscal stabilizers is comparatively small in most ACs. This implies that an improved growth performance in coming years will only contribute to a limited extent to a reduction of deficits. In addition, the implementation of fiscal consolidation measures may be delayed in 2003 in some Accession Countries in order not to endanger positive outcomes of EU accession referenda. While there is a clear political logic to such a posture, this may lead to a compression of the brunt of fiscal consolidation
Introduction
19
into a rather short timespan, in particular for countries that currently record comparatively high imbalances. Against the backdrop of the challenges in the fiscal realm that ACs face in the run-up to entry into the Euroarea, two questions seem to be particularly pertinent. First, how will fiscal consolidation affect growth? And second, what are the requirements for a fiscal consolidation to be successful and lasting? The standard view, embedded in the Keynesian literature, envisages that for given monetary policy, a fiscal contraction has a negative impact on growth, i.e. the fiscal multiplier is positive. Moreover, the contractionary effect will be larger for a spending cut than for a tax increase. A recent strand of literature has, however, challenged this view. Starting with a seminal contribution by Giavazzi and Pagano (1990), it has been shown that fiscal contractions can be in fact expansionary, i.e. the fiscal multiplier may be negative. Subsequent studies (see, e.g., Giavazzi and Pagano, 1996; Alesina and Ardagna, 1998; Alesina and Perotti, 1998) have explored under which conditions fiscal contractions may be expansionary. In particular, this literature suggests that fiscal adjustments are typically successful, lasting and expansionary if they are expenditure-based, consisting in particular of spending cuts in social transfers and the government wage bill. There is also some evidence that adjustments tend to be more successful if they are accompanied by agreements on wage moderation and/or currency devaluation. Furthermore, large adjustments are more likely to be successful, lasting and expansionary than small ones, although the composition of the adjustment itself seems to be more important. The findings of this literature, although inspiring and stimulating, do not constitute a widely shared consensus and should be interpreted with due caution (e.g. see Kamps, 2001). For example, the results may suffer from a selection bias, as some of the papers are based on case studies or on a limited number of statistical data; more research is still needed to confirm the robustness of their findings. Nevertheless, it seems fair to conclude on the basis of the new evidence that fiscal adjustments – if not expansionary – may be at least neutral for short-term growth. This, somewhat more moderate, proposition seems to be consistent with the recent experience of EU member countries and transition countries. Purfield (2003) takes a closer look at the fiscal stabilization in a broad set of transition countries from Central and Eastern Europe between 1992 and 2000. She finds that although there is little evidence of expansionary fiscal contractions, budgetary consolidation did not have a significant negative impact on growth in these countries. She
20
General Policy Issues for the Accession Countries
confirms, however, that larger and expenditure-based fiscal policies were more successful and durable in addressing fiscal imbalances. It should be noted, though, that most of the consolidation episodes Purfield (2003) looks at are taken from transition economies which are not candidates for EU accession. Lehment (2002) presents similar findings for EU countries, analysing budgetary balances before and after the start of EMU. He finds no evidence for a trade-off between budget consolidation and growth, at least in a medium-term perspective. In fact, most of the countries that undertook the largest reductions of structural budget deficits have recorded growth rates that were above the EU average and higher than before the consolidation. Thus, if there were short-term costs they were offset by an improved performance thereafter. Combining the analysis of the fiscal strategies of ACs with the recent empirical literature on fiscal contractions, it turns out that the prospects for sustained correction of budget imbalances are somewhat ambivalent. While the general approach to tackle deficits through expenditure cuts bodes well, the specific limitations that appear to exist in Accession Countries with respect to cuts in the areas of the public sector wage bill and of social transfers are less promising. In the Central European setting, fiscal consolidations may therefore tend to be associated with temporary costs in terms of growth and employment. If this is so, strong and sustained political determination will be required for implementing a strategy aiming at a fast entry into the Euroarea. In a less supportive setting, a more gradual approach may be more appropriate.
4
Conclusions
Three conclusions emerge from the preceding analysis. First, the available evidence on the economic costs and benefits of a future participation in the Euroarea is not uniform for all Accession Countries. This implies that, on economic grounds, the appropriate speed towards Euroarea accession may well be different between individual ACs. In general terms, the costs of full monetary integration tend to decrease over time, as structural convergence – driven by the completion of transition and the accession to the European Union – proceeds. Second, from today’s perspective, there is a considerable degree of uncertainty about the optimal date for joining the Euroarea. Results depend on what weights one assigns to individual effects, what probabilities one attaches to future events and with what interest rate one discounts future costs and gains, if they materialize at different points in time. Thus, based on economic reasoning, it is not possible, in most
Introduction
21
cases, to pinpoint a particular optimal target year for Euroarea accession for individual candidate countries, but most probably there will be a range of several years with similar cost–benefit balances. Third, joining a monetary union is also a political-economy issue. As the economics are not sufficiently clear-cut, the decision about the date will, at the end of the day, hinge upon political considerations as well. This, in turn, may tip the balance in favour of a relatively speedy quest for Euroarea participation for a number of Accession Countries, if the political will is strong enough to enact and sustain fiscal consolidation measures. This proviso pertains, in particular, to those ACs which presently record substantial fiscal deficits.
Notes 1. For a review of this discussion see Backé and Wójcik (2002). 2. For the main features of this debate see Backé (2002a). 3. For a selective review of the monetary policy integration strategies of Accession Countries see Moser et al. (2002). 4. For two of these country studies on the costs and benefits of Euroarea accession, pertaining to Austria and Sweden respectively, see Baumgartner et al. (1997) and Calmfors et al. (1997). 5. For a review of the effects not covered in this chapter, see Backé and Wójcik (2002) and Backé (2002a). 6. For a detailed review of the OCA theory see Horvath (2001a). 7. An alternative explanation for home bias is put forward by Obstfeld and Rogoff (2000), who argue that equity portfolios with home biases result from trading costs of goods from which these securities derive. 8. This is also acknowledged by OCA critics like Buiter (2000), who argue that, in an overvaluation situation, ‘generating [the needed] differential rates of inflation [between the domestic economy and abroad] is likely to involve greater resource costs than achieving the same relative price or cost realignment through a change in the nominal exchange rate’. 9. OECD (2000) has a useful analysis of product and labour market issues for the Czech Republic, Hungary and Poland. Two recent publications which cover part of the ground for a larger set or all Accession Countries respectively are Riboud et al. (2002) and the Transition Report 2000 (European Bank for Reconstruction and Development, 2000), the latter containing a concise overview chapter on labour market issues. 10. Clearly, this is only true if the respective country joins a monetary union like the Euroarea, which does not itself suffer from a dynamic inconsistency problem. 11. For Euroarea participants, EU fiscal rules comprise sanctions (including fines) in case of non-compliance, while for EU members that do not yet participate in the Euroarea fiscal discipline is primarily ensured through peer group pressure.
22
General Policy Issues for the Accession Countries
12. Whether net savings would be equally high might be questioned if such reforms were flanked by specific social measures to dampen their impact on the most affected strata of the population. 13. It should be noted, however, that social spending was substantially increased in Hungary from 2001 to 2002, so that the social expenditure/GDP ratio will actually be higher by 2 percentage points in 2005 than it was in 2001.
2 Is Accession to EMU More Justifiable ex post than ex ante? Jarko Fidrmuc
1
Introduction
Soon after accession to the European Union (EU) in 2004, the new EU member states will have to consider a timetable for accession to economic and monetary union (EMU). The adoption of the euro by them will further strengthen the integration of their economies with the other countries of the Euroarea. In sum, they are likely to gain from the reduction of transaction costs, but the loss of monetary sovereignty will create new problems during the catching-up process. Countries participating in a currency area have to face benefits and costs of the common currency. The benefits are directly related to transaction costs in countries’ bilateral trade. Therefore, countries with intensive trade relations are likely to gain relatively more from monetary integration. In addition, Frankel and Rose (1997 and 1998) hypothesize that business cycles are also becoming more similar across countries with close trade links. This hypothesis is supported by cross-section estimations of the relation between the correlation of business cycles and trade intensity among OECD countries between 1959 and 1993. Moreover, Fatás (1997), Artis and Zhang (1995) and Hochreiter and Winckler (1995) show that a common European cycle has been emerging as predicted by the endogeneity hypothesis of OCA criteria. Nevertheless, there remains considerable doubt whether there is a causal relationship between trade links and correlation of business cycles of the involved countries. Kose et al. (2003) find only weak evidence for the hypothesis that increased trade and financial flows have increased the synchronization of business cycles. Kenen (2000) notes that the correlation of business cycles may increase with the intensity of trade links between these countries, but he argues that this does not necessarily 23
24
General Policy Issues for the Accession Countries
mean that asymmetric shocks are reduced as well. Moreover, Hallett and Piscitelli (2001) show that a currency union may increase cyclical convergence, but only if there is already sufficient symmetry in the shocks and institutional structure across the countries. Their findings thus support Krugman’s (1993) discussion of the implications of the US currency union for European monetary union. In Krugman’s view, trade liberalization facilitates increased specialization according to comparative advantage of countries and possibly a divergence of business cycles in EMU. Furthermore, Frankel and Rose’s work lacks a stronger relation to trade structure1 (intra-industry trade) which should also explain the similarity of business cycles, although they use them as arguments. In particular, the effects of trade on convergence of business cycles depend on the degree of industrial specialization which is induced by the integration. Indeed, Helpman (1987) and Hummels and Levinsohn (1995) find that trade specialization plays a lesser role for trade among developed economies. Thus a majority of trade is observed within the same industries (that is, as so-called intra-industry trade). This should imply increasing correlation of business cycles between these countries. Therefore, this chapter tests OCA endogeneity using bilateral levels of intraindustry trade between OECD countries in the 1990s. It is shown that intra-industry trade induces the convergence of business cycles between trading partners, while there is no direct relation between business cycle and trade intensity. As a result, the OCA endogeneity hypothesis is confirmed. However, this result also underlines the role of specialization in trade. Finally, I ask whether the ACs should introduce the euro as soon as possible after accession to the EU or whether they should do so at a later stage. This question is addressed by applying the endogeneity hypothesis of OCA criteria to eight transition economies (the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia and Slovenia) invited to joint the EU in 2004. This chapter applies the relation between the degree of trade integration, the shares of intra-industry trade, and the convergence in business cycles to predict the degree of business cycle harmonization of CEECs with EU countries in the medium term. This approach reflects the Lucas critique in so far as it considers possible structural changes during the accession of the CEECs to the EU and EMU. Alternatively, these predictions can be interpreted as ‘indices of endogenous optimum currency area’ (EOCA indices) similar to those introduced by Bayoumi and Eichengreen (1997).
Accession to EMU – ex post or ex ante Justifiable? 25
The chapter is structured as follows. The next section (2) starts with a short overview of the current discussion on the endogeneity of the OCA criteria. Then it tests the OCA endogeneity hypothesis with respect to trade intensity and intra-industry trade. The sensitivity analysis (3) includes further variables covering similarity in size and preparation for participation in EMU. Section 4 applies the revealed relation between, on the one hand, the correlation of business cycles, and, on the other, the trade variables for the computation of a potential correlation of business cycles (EOCA indices) in the CEECs. Section 5 concludes.
2
The optimum currency area theory
2.1 Endogeneity of optimum currency area criteria The theory of optimum currency areas, which was developed by Mundell (1961), McKinnon (1962), and Kenen (1969), has become particularly popular for analyses of the costs and benefits of monetary integration, in particular with reference to EMU. The basic point of the OCA theory is that countries or regions exposed to symmetric shocks, or possessing mechanisms for the absorption of asymmetric shocks, may find it optimal to adopt a common currency. This literature therefore focuses on assessing the symmetry of output shocks in monetary unions, and/or evaluating the absorption mechanisms, such as labour mobility or fiscal transfers. In general, the stronger any of these linkages between countries participating in a currency area, the more gains may be expected by the participating countries. In particular, the OCA theory discusses the following criteria:2 first, potential gains from the creation of an OCA are determined by the degree of openness. A country where trade within the OCA accounts for a high proportion of domestic output can profit from participating in a currency area. Second, the OCA theory stresses the importance of the similarity of shocks and business cycles. Asymmetric shocks and business cycles raise the need for country-specific adjustment policies; however, in a singlecurrency area, country-specific monetary policy is not possible. Third, Mundell (1961) points to international factor mobility (especially migration) as an alternative adjustment channel. High labour mobility facilitates adjustment to the adverse effects of asymmetric shocks and thus reduces the pressure for exchange rate adjustments. Fourth, Kenen (1969) stresses the importance of product diversification. A country exporting highly diversified products is less vulnerable to sector-specific shocks. Therefore, countries with a large product spectrum are less likely to be
26
General Policy Issues for the Accession Countries
induced to use the exchange rate as an adjustment tool. Fifth, Kenen (1969) also examines fiscal transfers, which can be used to counteract asymmetric shocks in a currency area. Finally, the degree of policy integration and similarity between rates of inflation has been introduced to the OCA theory more recently (see for example Dixit, 2000). On the one hand, differences between rates of inflation result in a loss of competitiveness in high-inflation countries, which calls for external adjustments (see Carlin et al., 2001). On the other hand, a high degree of existing policy integration before the creation of a currency area is likely to result in lower costs for the participating countries. Frankel and Rose (1998) suggest that the first two criteria are endogenous. Closer trade relations result in a convergence of business cycles. Further, similar business cycles create good preconditions for policy integration and the creation of a currency area. However, this view is not universally shared in literature. For example, Krugman (1993) points out that, as countries become more integrated, they specialize more. This view is also supported by Eichengreen (1992) and Bayoumi and Eichengreen (1993). In turn, Eichengreen (2000) accepts the endogeneity of OCA criteria in his discussion of the dollarization in the Latin American countries. Artis (2002) provides a recent survey of this discussion. Furthermore, Kenen (2000) and Hallett and Piscitelli (2001) argue that Frankel and Rose’s results should be interpreted cautiously. Kenen (2000) shows in a framework of the Keynesian model that the correlation between two countries’ output changes increases unambiguously with the intensity of trade links between these countries, but this does not necessarily mean that asymmetric shocks are reduced as well. Barro and Tenreyro (2003) find that common currency reduces the co-movement of shocks to real GDP, which implies that that currency unions lead to greater specialization. Kose and Yi (2001) also do not find any positive relations between business cycle and trade intensity in a standard international business cycle model. Therefore, it is important to keep in mind that it is not trade relations alone that induce the convergence of business cycles in an OCA. Indeed, Frankel and Rose’s hypothesis underlines that bilateral trade is mainly intra-industry trade, although this indicator does not enter directly into their analysis.3
2.2 Trade integration and business cycles If intra-industry trade accounts for a high share of bilateral trade, Frankel and Rose (1998) argue that the simple trade intensity increases the
Accession to EMU – ex post or ex ante Justifiable? 27 Table 2.1
Trade integration and business cycles Industrial production
Constant Trade intensity No. of observations SE Adjusted R2
Real gross domestic product
Exports (1.a)
Imports (1.b)
Total (1.c)
Exports (1.d)
Imports (1.e)
Total (1.f )
0.683 (8.005) 0.084 (5.378) 253 0.288 0.093
0.686 (8.517) 0.084 (5.632) 253 0.286 0.104
0.715 (8.355) 0.091 (5.683) 253 0.286 0.102
0.688 (6.832) 0.086 (4.655) 231 0.331 0.069
0.681 (7.064) 0.083 (4.780) 231 0.329 0.080
0.705 (6.939) 0.090 (4.782) 231 0.330 0.076
Note: The dependent variable is the index of correlation of de-trended indicator of economic activity (fourth difference of logs) between trading partners. Trade intensity is measured as a share of bilateral trade aggregate in total trade aggregates of both countries as indicated by the columns’ headers. The instrumental variables in the two-stage OLS include the log of distance, a dummy for geographic adjacency, a dummy for the Euroarea, the log of aggregate income and the log of income per capita. Heteroscedasticity-robust t -statistics are in parentheses. Adjusted R2 and standard errors of regression (SE) are computed using the second-stage residuals.
convergence of business cycles. Indeed, they report a significant and positive relation between trade intensity and the correlation of business cycles as measured by various indicators of economic activity in a crosssection of OECD countries between 1959 and 1993. For empirical tests, the endogeneity hypothesis of the OCA criteria may be stated as Corr(Q i ,Q j ) = α + β log(TIijT ),
where TIijT =
Tij Ti + Tj
,
(2.1)
where Corr(Q i ,Q j ) stands for the correlation of de-trended (fourth differences of logs) indicator of economic activity4 and TI denotes the natural logarithm of the bilateral trade intensity between countries i and j. Trade intensity may be defined either in relation to exports, imports, or trade turnover (T = X, M, X + M). Table 2.1 reports several specifications of equation (2.1) for OECD countries between 1990 and 1999. However, the OLS regression of bilateral economic activity on trade indicators may be inappropriate. Countries are likely to orient their monetary policy and fix the exchange rates towards their most important trading partners. Bilateral trade might already reflect the adoption of common exchange rate policy
28
General Policy Issues for the Accession Countries
and not vice versa.5 Therefore, the regressions have to be instrumented by exogenous determinants of bilateral trade flows. Such instruments are provided by the so-called ‘gravity model’, including the log of distance between trading partners, a dummy for geographic adjacency and a dummy for the 12 member states of the EU, and the aggregate income as well as the income per capita (in logs) of the included countries. At this stage, trade intensity is revealed to have a significant and positive effect on the correlation of business cycles. This result is robust to the selection of the indicator of economic activity and the particular definition of trade intensities. The business cycles of industrial production seem to be better explained by trade than the business cycles as defined by the correlation of countries’ real GDP. This corresponds to the high share of tradables in the manufacturing industry. However, the adjusted R2 is relatively low for all specifications of equation (2.1). As might be expected, the coefficients estimated for trade intensity indicators are slightly higher in the 1990s than in the previous decades as reported by Frankel and Rose (1998). This could indicate that the role of trade relations has increased recently. 2.3 Intra-industry trade and business cycles However, equation (2.1) applies only bilateral trade to explain the similarity of business cycles, although the similarity of trade structure (e.g. the level of intra-industry trade) may be viewed as a major adjustment force inducing the convergence of business cycles between trading partners. Frankel and Rose (1998), Kalemli-Ozcan et al. (2001), but also Krugman (1993) and Hallett and Piscitelli (2001) use the trade structure arguments in favour as well as against the endogeneity hypothesis of OCA criteria. Therefore, I estimate the relation between the correlation of business cycles, trade integration, and the bilateral level of intra-industry trade as Corr(Q i ,Q j ) = α + β log(TIijT ) + γ IITij ,
(2.2)
where Q and TI are defined in the same way as in the corresponding formulations of equation (2.1) and IITij stands for intra-industry trade as measured by the Grubel–Lloyd indices, which is computed as |X − Mit | , GLIt = 1 − i it (X it + Mit ) i
(2.3)
Accession to EMU – ex post or ex ante Justifiable? 29
where X and M denote the EU’s exports and imports by three-digit SITC commodity groups i (as published by UN), respectively. This level of disaggregation is more detailed than industry production used by Imbs (1999) and Clark and van Wincoop (2001). An index value of 0 shows that there is exclusive inter-industry trade, i.e. a complete specialization in different products for each country, while an index value of 1 indicates exclusive intra-industry trade. When available, data according to Eurostat were taken. Intra-industry trade at the same level of disaggregation between non-EU countries was computed using the UN World Trade Data. Note that the data according to this classification are not available for the earlier periods due to changes in trade statistics. Equation (2.2) is again estimated by two-stage OLS. Note that the selected instrumental variables are also highly correlated with intraindustry trade (see Loertscher and Wolter, 1980; Hummels and Levinsohn, 1995).6 In this specification (see Table 2.2), the coefficients of intra-industry trade are significant if estimated for industrial production, although they are insignificant (but positive) for two specifications applying real GDP. By contrast, the coefficients of bilateral trade intensity are close to zero (indeed, they have signs contrary to those expected in several specifications) and insignificant for both indicators of economic activity.7 The revealed pattern is very robust with respect to the choice of instrumental variables and country sample. This indicates that trade intensities have no direct effect on the correlation of business cycles. Therefore, I drop TIij from estimated equations, Corr(Q i ,Q j ) = α + γ IITij ,
(2.4)
which are reported in the last columns of the two blocks of Table 2.2. The coefficients of intra-industry trade are highly significant in both specifications of equation (2.4). In so far as the countries with close trade links have high shares of intra-industry trade, the endogeneity hypothesis of OCA criteria is confirmed by equations (2.2) and (2.4). However, Table 2.2 shows that the correlation of business cycles of trading partners is not driven by the simple aggregation of shocks, being transferred between the countries via direct trade channels, as argued by Kenen (2000). By contrast to this demand-driven view of OCA endogeneity, equations (2.2) and (2.4) imply that it is the structure of foreign trade and not the direct effect of bilateral trade which induces the synchronization of countries’ business cycles.
30
General Policy Issues for the Accession Countries
Table 2.2
Intra-industry trade, trade integration and business cycles Industrial production Exports Imports (2.a) (2.b)
Total (2.c)
Real gross domestic product
Only IIT Exports Imports (4.a) (2.d) (2.e)
Total (2.f)
Only IIT (4.b)
Constant
0.259 0.468 0.379 0.499 0.444 0.578 0.539 0.476 (1.598) (4.325) (2.576) (11.934) (2.361) (4.381) (3.221) (9.636) Trade intensity -0.085 -0.011 -0.042 -0.011 0.038 0.022 (-1.619) (-0.323) (-0.879) (-0.188) (0.913) (0.422) Intra-industry 0.335 0.207 0.257 0.187 0.195 0.103 0.135 0.175 trade (3.597) (3.043) (3.047) (6.554) (1.812) (1.304) (1.457) (5.324) No. of 253 253 253 253 231 231 231 231 observations SER 0.280 0.282 0.281 0.281 0.329 0.329 0.329 0.329 0.141 0.130 0.132 0.133 0.078 0.080 0.078 0.082 Adjusted R2 Note: The dependent variable is the index of correlation of de-trended indicator of economic activity (fourth difference of logs) between trading partners. Trade intensity is measured as a share of bilateral trade aggregate in total trade aggregates of both countries as indicated by the columns’ headers. The instrumental variables in the two-stage OLS include the log of distance, a dummy for geographic adjacency, a dummy for EC-12, the log of aggregate income and the log of income per capita. Heteroscedasticity-robust t -statistics are in parentheses. Adjusted R2 and standard errors of regression (SER) are computed using the second-stage residuals.
3 The endogeneity hypothesis of OCA criteria and EMU enlargement Since the beginning of the 1990s, all CEECs have aimed at future membership in the European Union. After more than a decade of economic reform, these countries have largely succeeded in adjusting their economies to market principles. As a result, the EU started membership negotiations with five CEECs (the Czech Republic, Estonia, Hungary, Poland and Slovenia) in 1998, which were extended to all ten CEECs in 2000. The Copenhagen summit of the EU invited eight CEECs, which are the focus here, to joint the EU by 2004. As part of this enlargement agenda, several CEECs have already expressed their aspiration to join the Euroarea as soon as possible after accession. Furthermore, several authors discuss the possibility of adopting the euro as legal tender in some CEECs before full membership in the EU. This discussion was started by Bratkowski and Rostowski (2001), but Portes (2001) and Buiter and Grafe (2001) have also addressed this issue. Hochreiter and Wagner (2002) provide a detailed discussion of the arguments for and against so-called euroization.
Accession to EMU – ex post or ex ante Justifiable? 31
By contrast, the European Union, including the eurosystem, has outlined a three-step approach to the monetary integration of the candidate countries from Central and Eastern Europe, which is described in more detail by Kopits (1999). The applicants should first join the EU, then enter the exchange rate mechanism (ERM II) of the European Union and finally, after the fulfilment of the convergence criteria, accede to economic and monetary union.
3.1 Trade integration between EU and CEECs Since the opening up of Eastern Europe, the importance of EU countries for the CEECs’ trade has increased dramatically. As of 2001, the European Union was the most important trading partner of all CEECs. For instance, Poland’s exports to the EU account for nearly 70 per cent of its total exports, and imports are also only slightly lower (61 per cent). Thus Poland belongs with those CEECs mostly integrated with the single market of the European Union. For comparison, the EU accounted for between 50 per cent (Lithuania) and 75 per cent (Hungary) of total exports of the CEECs.8 These export shares are comparable to or even higher than intra-EU shares for nearly all EU member states. On the import side, the predominance of the EU is only slightly weaker. Furthermore, the shares of exports and imports going to and coming from an ‘enlarged EU’, which is the current EU plus all ten CEECs, are even higher. According to this indicator, the enlarged Europe is the most important export market for Slovakia and the Czech Republic, followed by Portugal, the Netherlands and Austria. The CEECs are relatively open economies. Exports account for about one-third of GDP in Hungary, and above 40 per cent in the Czech Republic, Slovakia and Slovenia. Thus, these countries are relatively more open than nearly all EU countries. There are only a few EU countries, including Belgium, the Netherlands and Ireland, which are significantly more open than the smaller CEECs (export shares between 50 per cent and 70 per cent of GDP). Only Poland’s exports are relatively lower at 17 per cent of GDP, but this corresponds to the larger size of the Polish economy. Buiter (2002) notes that the CEECs are also relatively open if we compare their trade to GDP at purchasing power parities. From the point of view of the conventional OCA theory, if intraindustry trade accounts for a high share in trade, then, ceteris paribus, business cycles are expected to become more similar across countries. By contrast, increased bilateral trade intensity may lead to the divergence of business cycles if the increase in trade is mainly due to the increased
32
General Policy Issues for the Accession Countries
specialization as predicted by the alternative view of an OCA. Therefore, intra-industry trade may be used to identify which approach is more appropriate for a particular group of countries. The growth of intra-industry trade, which is observed in intra-EU trade, also dominates recent East–West trade developments. This would increase net gains from the integration of CEECs into the Euroarea. According to Fidrmuc (2001), the shares of intra-industry trade in the EU’s trade with the Czech Republic, Slovenia and Hungary, as computed by Grubel–Lloyd indices for three-digit SITC commodity groups, were already comparable to or even slightly larger than in EU trade with e.g. Spain and Sweden (that is, about 60 per cent) in 1998. Poland and Slovakia report somewhat lower levels of intra-industry trade at about 50 per cent. These levels are comparable to those of Ireland and Portugal. However, the shares of intra-industry trade in EU trade with Estonia, Lithuania and Latvia have still remained slightly above the level of EU intra-industry trade with Greece and Turkey (below 35 per cent).
3.2 Observed convergence of business cycles in the EU and the CEECs There is mixed evidence on the convergence of business cycles in the EU and the CEECs. On the one hand, the level of GDP grew slowly in relation to Western European countries during the period of the central planning system. The divergence of Western and Eastern Europe speeded up in the 1970s and the 1980s. Therefore, the increasing welfare difference between market and central planning economies in Europe was one of the major reasons for the introduction of early reforms in Eastern Europe. Furthermore, there were also few signs of convergence between Central and Eastern European countries in this period. Estrin and Urga (1997) find only limited evidence of convergence in the former Soviet Union, as well as within various groups of Central European command economies. Even more surprisingly, Fidrmuc et al. (1999) conclude that the Czech Republic and Slovakia did not converge, neither between 1950 and 1990, nor within a sub-sample from 1970 to 1990. By contrast, Koˇcenda (2001) finds output convergence between 1991 and 1998 especially within the CEFTA (Czech Republic, Hungary, Poland, Slovakia and Slovenia). Several authors report increasing similarities of business cycles between the EU (mainly Germany) and the CEECs since the economic reforms were introduced. In particular, Boone and Maurel (1998 and 1999) find a significant convergence between business cycles (as measured by unemployment rates) in Germany and selected CEECs (the Czech Republic,
Accession to EMU – ex post or ex ante Justifiable? 33
Hungary, Poland and Slovakia). According to Boone and Maurel (1999), between 55 per cent (Poland) and 86 per cent (Hungary) of the CEECs’ cycles (given by de-trended unemployment) are explained by German shocks. This figure is lower than the estimate for the French–German interdependence of business cycles (91 per cent), but higher than the estimates for the German influence on Spanish (43 per cent) and Italian (18 per cent) business cycles. Therefore, the authors conclude that the benefits from eventually joining the Euroarea could outweigh the costs in the CEECs. Similarly, Korhonen (2003) looks at monthly indicators of industrial production in the Euroarea and nine Accession Countries (excluding Bulgaria) in Central and Eastern Europe. The issue of correlation is assessed with the help of separate VARs for the first difference of Euroarea production and production in each of the Accession Countries. He finds that the most advanced Accession Countries (especially Hungary and perhaps also Slovenia) exhibit a quite high correlation with the Euroarea business cycle. Moreover, correlation seems to be at least as high as in some small current EMU member countries. Indeed, business cycles in several CEECs have become strikingly similar to the business cycle of the EU (as proxied by Germany) since the second half of the 1990s (see Table 2.3). At the beginning of the 1990s, the business cycles in the CEECs were determined by the so-called transitional recession. Therefore, the correlation of business cycles was low between 1991 and 2001. The recovery in these countries has been strongly influenced by growing exports to the EU. As a result, the business cycle of the EU has determined developments in CEECs’ economies since the middle of the 1990s. In particular, the correlation of growth of industrial production between Germany and Hungary (0.83), and Germany and Slovenia (0.86), has been higher than the corresponding correlations of EU countries with Germany on average (0.52) between 1996 and 2001. However, the period of about six years might be too short to conclude that the business cycles have already become similar. In particular, this period corresponds to only about one full business cycle. Moreover, this period was characterized by only few supply and demand shocks (see Fidrmuc and Korhonen, 2001). Actually, the correlations of industrial production in Germany and that in the Czech Republic9 and Slovakia have remained relatively low. In so far as the Czech Republic and Slovakia are quite similar to other CEECs, this indicates that country-specific shocks may still have significant effects on these economies. The difference between the Czech Republic and Slovakia, on the one hand, and
34
General Policy Issues for the Accession Countries Table 2.3 Comparison of business cycles of selected countries with that of Germany Industrial production
Real gross domestic product
1991–2001a
1996–2001a
1991–2001a
Austria Belgium Finland France Greece Ireland Italy Netherlands Portugal Spain
0.79 0.33 0.35 0.86 0.47 0.27 0.58 0.56 0.58 0.80
0.79 0.05 0.85 0.71 0.46 0.12 0.55 −0.11 0.42 0.63
0.46 0.24 −0.55 0.08
0.79 0.05 0.85 0.71
0.19 0.19 0.06 0.12 0.19
−0.03 0.55 −0.11 0.42 0.63
Denmark Sweden UK
0.69 0.38 0.11
0.60 0.13 0.35
−0.02 −0.20 −0.65
0.51 0.34 0.45
Czech Republic Estonia Hungary Latvia Lithuania Poland Slovakia Slovenia
0.35
0.18c
1996–2001a
0.35 0.57c 0.83 0.39c 0.63 0.36c 0.31c 0.86
0.02b 0.43b 0.67b 0.41b 0.08b 0.18b −0.29b 0.34b
Note: The similarity of business cycles is measured by the correlation of de-trended indicator of economic activity (fourth difference of logs). a Approximate time periods; the actual time periods may differ by about one year due to data availability. b Correlation according to Fidrmuc and Korhonen (2001), time period starting in 1995/1996 and ending in 2000. c Data according to WIIW. Sources: IMF, own calculations.
the remaining CEECs, on the other, indicates that asymmetric shocks are still likely in the EU and the CEECs.
4
Indices of endogenous optimum currency area
The revealed trend of unification of business cycles in Europe is not surprising. It fully corresponds to the endogeneity of OCA criteria.
Accession to EMU – ex post or ex ante Justifiable? 35
Therefore, I use equations estimated in the previous section to evaluate the potential correlation of business cycles in Germany and the CEECs given the current integration of these countries and the current level of intra-industry trade. Note that these correlations can be alternatively interpreted as indices of endogenous optimum currency area (EOCA indices) similar to those constructed by Bayoumi and Eichengreen (1997). A comparison of Tables 2.3 and 2.4 shows that the correlations of business cycles in Germany and in other EU countries were on average slightly higher in the 1990s than those predicted by the EOCA indices. However, this is not so surprising. First, the European Union has made significant progress in the coordination of economic policy in member states. As a result of the introduction of the single market in 1992 and the preparations for EMU in this decade, the similarity of business cycles within the EU countries has probably been higher in the 1990s than in the previous decades. Second, Germany was selected as a proxy for the EU (see Bayoumi and Eichengreen, 1993). Using various specifications of equation (2.1), the correlation of industrial production and GDP in Germany and other EU countries is predicted at about 0.37 for both indicators on average. Actually, the corresponding correlations predicted for the CEECs (EOCA indices) are only slightly lower. The Czech Republic, Poland and Hungary could potentially reach correlations as high as 0.35 on average in the medium run, while Slovak and Slovene trade is less oriented towards Germany, resulting in a lower predicted correlation of about 0.24 on average. Similarly, I use equation (2.4) to compute the EOCA indices in Germany and in selected countries, which are even higher than the previous figures (see Table 2.4). In fact, the Czech Republic is predicted to have a higher correlation of industrial production with Germany than all EU countries except for France, although this prediction still remains below the realized levels in several EU countries. The comparison of predicted, or potential, business cycle correlations for selected Western and more advanced Eastern European countries shows small differences between both regions. Further coordination of economic policy in the Czech Republic, Hungary, Slovenia, and to a lesser extent also in Poland and Slovakia, with the EU is likely to result in a fast convergence of business cycles. Thus, these CEECs face favourable preconditions for a fast convergence to the business cycle in the EU (or EMU). This expectation is based on the high openness of the CEECs vis-à-vis the EU and the high shares of intraindustry trade in bilateral relations. Given, first, the potential gains from
36
General Policy Issues for the Accession Countries
Table 2.4 Comparison of indices of EOCA of selected countries with those of Germany Industrial production
Real gross domestic product
Estimation applied
Exports Imports Total Only IIT Exports Imports Total Only IIT (1.a) (1.b) (1.c) (4.a) (1.d) (1.e) (1.f) (4.b)
Austria Belgium Greece Spain Finland France Ireland Italy Netherlands Portugal
0.41 0.44 0.26 0.40 0.28 0.47 0.22 0.44 0.45 0.30
0.44 0.43 0.22 0.38 0.29 0.46 0.30 0.44 0.46 0.30
0.42 0.43 0.24 0.38 0.29 0.46 0.27 0.43 0.44 0.30
0.43 0.43 0.24 0.41 0.34 0.45 0.28 0.39 0.41 0.36
0.40 0.43 0.26 0.39 0.28 0.46 0.21 0.44 0.44 0.29
0.44 0.43 0.22 0.38 0.29 0.46 0.30 0.44 0.45 0.30
0.42 0.43 0.24 0.38 0.28 0.46 0.26 0.43 0.44 0.30
Denmark Sweden UK
0.34 0.36 0.45
0.35 0.34 0.42
0.34 0.38 0.35 0.36 0.43 0.43
0.34 0.35 0.44
0.35 0.34 0.42
0.34 0.37 0.35 0.35 0.43 0.41
Czech Republic Estonia Hungary Latvia Lithuania Poland Slovakia Slovenia
0.36 0.06 0.33 0.09 0.14 0.35 0.24 0.23
0.36 0.06 0.33 0.09 0.12 0.37 0.27 0.23
0.36 0.07 0.33 0.10 0.14 0.36 0.26 0.23
0.35 0.06 0.33 0.08 0.14 0.34 0.24 0.22
0.36 0.06 0.33 0.09 0.12 0.37 0.27 0.23
0.36 0.06 0.32 0.09 0.13 0.35 0.25 0.22
0.43 0.16 0.38 0.21 0.20 0.33 0.34 0.36
0.41 0.41 0.24 0.39 0.32 0.43 0.27 0.37 0.39 0.35
0.41 0.15 0.37 0.21 0.19 0.31 0.33 0.35
Note: Indices of endogenous optimum currency area are computed according to particular specifications of equations (2.1) and (2.4) as indicated by columns’ headers.
an OCA between the current EMU countries and the CEECs, as illustrated by the importance of EU trade in the CEECs, and second, the currently observed convergence of business cycles in both regions (which is partly caused by the first observation), we can expect a tendency for the CEECs to join the EMU in the future. Less convergence is predicted for the remaining CEECs (Estonia, Latvia and Lithuania).
5
Conclusions
This contribution examines the endogeneity hypothesis of OCA criteria originally introduced by Frankel and Rose (1998). On the one hand, this issue has significantly influenced the debate on European monetary
Accession to EMU – ex post or ex ante Justifiable? 37
integration. On the other hand, there is considerable doubt whether there is a causal relationship between trade and business cycles. For example, Krugman (1993) and Eichengreen (1992) argue that integration is likely to support the specialization of participating countries according to comparative advantage. Indeed, Krugman finds empirical support for his arguments in the specialization pattern and business cycles of the US regions. Furthermore, Kenen (2000) demonstrates that the trade links alone do not ensure the convergence of business cycles if countries are not sufficiently similar. With respect to this discussion, it is shown here that there is no direct relation between business cycle and trade intensity if regressions are augmented by additional variables. Nevertheless, intra-industry trade is shown to induce the convergence of business cycles in OECD countries. As a result, the endogeneity hypothesis of OCA criteria is confirmed but with respect to intra-industry trade. This finding is robust with respect to the definition of trade intensity and the selection of the indicators of economic activity for comparison of business cycles. Furthermore, this chapter addresses the controversial issue of the current enlargement agenda. The future enlargement of the Euroarea to Central and Eastern European countries has initiated an intense academic and political discussion. This discussion is characterized by a multitude of different policy proposals, ranging from the immediate adoption of the euro in some countries, to suggestions that the CEECs should not give up exchange rate flexibility in order to support their growth and convergence to the EU. This chapter to the discussion focuses on the ACs invited to joint the EU in 2004. On the one hand, it confirms earlier findings, e.g. that the CEECs have rapidly converged to the EU countries in terms of business cycles and trade integration. In particular, business cycles in several CEECs (Hungary, Slovenia and, to a lesser extent, Poland) are strongly correlated with the business cycle in Germany, in the period since 1993. In this respect, it may seem that Hungary, Slovenia and Poland, but not the Czech Republic and Slovakia, have made headway towards constituting an optimum currency area with the EU. Less convergence is predicted for the remaining CEECs (Estonia, Latvia and Lithuania). On the other hand, this chapter shows that the observation period is still too short to conclude that the business cycles have already become similar. In particular, this period has been characterized by only few supply and demand shocks. Furthermore, the business cycle in the Czech Republic is not correlated with that in Germany. As the Czech Republic is quite similar to other CEECs, this indicates that country-specific shocks
38
General Policy Issues for the Accession Countries
may still have significant effects on these economies. Poland, the largest AC, also shows a comparably low correlation of business cycle with the EU in the 1990s.
Notes 1. Clark and van Wincoop (2001), Kalemli-Ozcan et al. (2001) and Imbs (1999) show that the similarity of production at a relatively high degree of sectoral aggregation has a positive influence on correlation of business cycles among countries. 2. A novel contribution to OCA criteria is provided by Alesina and Barro (2002). 3. Fontagné (1999) discusses the relation between intra-industry trade and the symmetry of shocks in a monetary union. 4. The country sample includes Switzerland, Norway, the USA, Canada, Australia, New Zealand, Turkey and Israel in addition to 14 EU countries (Belgium and Luxembourg are reported as a single region). I use IMF/IFS industrial production and GDP. The quarterly GDP is not available for Greece. Trade intensities were computed for 1997. 5. Rose (2000) and Frankel and Rose (2000) document positive effects of currency unions and negative effects of exchange rate volatility on bilateral trade. Vinhas de Souza (2002b), on the other hand, finds negative trade effects from the early years of EMU for its participating members. 6. The Theil’s measure of multicollinearity is close to or below than 0.1 for all specifications of equation (2.2). This indicates that both explanatory variables are not multicollinear. The Theil’s measure of multicollinearity is defined as the difference between R2 of the particular specifications of equation (2.2) and the incremental contributions to this explanatory power by ITij and IITij , which are computed using the R2 value for the corresponding specifications of equations (2.1) and (2.4), that is as R2(2) − (R2(2) − R2(1) ) − (R2(2) − R24 ). This indicator is zero if all explanatory variables are orthogonal (see, e.g., Judge et al., 1988). 7. This result is similar to the finding by Imbs (1999) that the inclusion of additional variables lowers the size of trade integration effects on the correlation of business cycles. 8. As estimated by gravity models, Fidrmuc and Fidrmuc (2001) show that the trade between the CEECs and the EU, as well as the trade between individual CEECs, has already reached its ‘natural’ level, corresponding to the economic size, the distance between these countries, and the stage of integration. 9. In contrast to our results, Cincibuch and Vavra (2001) show that an alternative measure of similarity in business cycles – standard deviation of percentage changes in relative output in the Czech Republic and Germany – has declined during the reform period, meaning that the symmetry of business cycles has increased.
3 Modelling Alternative Paths to EMU for the Accession Countries Lúcio Vinhas de Souza and Elisabeth Ledrut
1
Introduction
European Union accession and the perspective of Euroarea integration will undoubtedly have a strong effect on the present and future macroeconomic policies of the Central and Eastern European countries that are candidates for EU accession (namely Bulgaria, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia and Slovenia), the so-called Central and Eastern European (CEE) Accession Countries. The question of which exchange rate regime these countries should choose in the run-up to Euroarea accession is particularly interesting in this respect. In this work we will focus on the potential effects on both growth and inflation rates in the CEECs of two extreme types of exchange rate strategies: fixed and flexible.1 The underlying assumption is that the choice of the exchange rate regime is of considerable short-run importance for the integration process of these economies. We use a traditional Mundell–Fleming (MF) model2 (so called from the combination of work done independently by Marcus Fleming and Robert Mundell during the early 1960s: see Fleming, 1962 and Mundell, 1962), expanded with an expectations formation mechanism (see Dornbusch, 1976), to study the effects of alternative exchange rate regimes. Furthermore, we analyse the consequences of the interaction with the Euroarea on the macroeconomic adjustment of individual transition economies. MF models have been criticized for lacking clear micro-foundations: there are no agents in the set-up, determining their actions by openly minimizing a loss function or maximizing a welfare function, which, among other things, makes welfare evaluations based on the model’s results somewhat tricky. Nevertheless, the expanded MF still remains very much the ‘work-horse’ of most macroeconomic modelling used for 39
40
General Policy Issues for the Accession Countries
policy purposes, due to its elegance, simplicity and intuitive policy implications (see Obstfeld, 2000 and Rogoff, 2001). We choose it here because of its small size and low data requirements, which enable individual estimations for all the countries in our sample. Furthermore, its tractability and flexibility and the existence of an established body of literature on its application have influenced our choice.
2
Modelling the exchange rate regime
Following the MF set-up, we assume two regions, a small domestic country and a large foreign economy, the Euroarea. Given our focus on the CEECs, this ‘small-country’ assumption is justified by the fact that the joint GDP of all ACs is around 5 per cent of the EU’s GDP, or a little more than 7 per cent of the Euroarea GDP. These countries are price-takers on international markets (i∗ , the world real interest rate, is exogenously given, as is p∗ , the world price level; and they face a horizontal demand curve), while the effects of the CEECs’ economies on the large Euroarea economy are negligible. The estimated log-linear model will assume the specifications below. All series – except the interest rates – are in natural logarithms, and in deviations from the long-run trend (estimated using a Hodrick–Prescott filter upon the original series using a quarterly penalty parameter λ equal to 1.600). Additionally, due to a question of scale, the national net current account and net financial account were converted from USD into the national currencies using the average nominal quarterly exchange rate. The resulting figures were then divided by real GDP, generating series in terms of output share upon which the HP filtering process was used.3 In equation (3.1), we have the IS schedule for the real goods market, defined as real domestic income in the transition economy (nominal GDP deflated by the CPI), which is assumed to be a function of lagged domestic real GDP, the real interest rate (defined as the nominal interest rate in time t – the annualized lending interest rate series are set to quarterly rates before that – minus the realized CPI inflation rate in time t), the level of real government consumption (the nominal series deflated by the CPI), a competitiveness parameter defined as the real exchange rate, the external balance (defined as the net current account) and an external demand shock (the real GDP of the Euroarea, the most important trade partner of all the CEECs). ∗ Yit = α1 yit−1 − α2 rit + α3 git + α4 cit + α5 bit + α6 yit−1 +μ
(3.1)
Alternative Paths to EMU
41
As indicated above, the competitiveness parameter c is defined as the real effective exchange rate, or the difference of the log nominal exchange rate s and the difference between the domestic and external price levels, given by Cit ≡ sit − pit + pit∗ .
(3.2)
For a peg regime, the REER series above will be estimated with the nominal exchange rate set at t = 0, i.e., its level at the beginning of the sample, or E(s) = 0. In equation (3.3), we have the LM schedule, where the current money stock is a function of the real GDP level, the opportunity cost of holding money (the nominal interest rate) and the inflation level, and, in the case of the fixed regime, the change in international reserves held at the monetary authority (the sum of the reserves in hard currencies and gold at national valuation, converted to domestic currency using the nominal exchange rate). Mit = α7 yit−1 − α8 iit−1 + α9 pit−1 + α10 reit + μ
(3.3)
In equation (3.4), we have the BP schedule, where, in a fixed exchange rate regime, the net external balance is defined as, again, the sum of the net current and financial accounts, as given by the difference of the nominal domestic and external interest rate (net capital flows are, therefore, assumed to be determined by the differential returns), a competitiveness parameter c (the REER series for a fixed exchange rate regime is calculated in the same way as described above), lagged domestic activity and lagged external activity. ∗ ∗ Bit = α11 iit−1 − iit−1 + α12 cit + α13 yit−1 + α14 yit−1 +μ
(3.4)
As free floating is assumed to keep the balance of payments in equilibrium (B = 0), the equation above, in a floating exchange rate regime becomes equation (3.5): ∗ ∗ + α12 cit + α13 yit−1 + α14 yit−1 + μ. E(˙s) = α11 iit−1 − iit−1
(3.5)
We assume rational exchange rate expectations, which, in the absence of uncertainty, implies perfect foresight and therefore, E(˙s) = s˙ .
(3.6)
42
General Policy Issues for the Accession Countries
This not a realistic assumption even for mature market economies, and even less for the CEECs in our sample that are introducing market institutions and new currencies, while being subject, at the same time, to both country-specific and common shocks. Nevertheless, given that we do not have an adequate proxy series for the exchange rate expectations (as expectations are not directly observable), we use the series of the realizations of the nominal exchange rate in time t. In equation (3.7), we have a Phillips curve equation, linking inflation with past and future prices (this may be understood as representing an economy with overlapping wage contracts, some set with backwardlooking expectations concerning prices and some forward-looking: see Bank of England, 1999) and lagged GDP. The term α18 is the change in the exchange rate, which can be understood as indicating the degree of pass-through from exchange rate movements to domestic prices in this economy (this represents a change compared to earlier versions of this model; see Vinhas de Souza and Ledrut, 2002). In a fixed exchange rate regime, this coefficient is, of course, equal to zero. Pit = α15 pit−1 + α16 pit+1 + α17 yit−1 + α18 s + μ
(3.7)
A straightforward way to evaluate the comparative optimality of the two possible exchange rate regimes, as given by the results of our estimations, can be derived from a simple loss function, which enables a policy-maker to compare the welfare effects of the alternative regimes. The loss function is defined as equation (3.8): n n U = β1 (Yt ) − β2 (pt ) . (3.8) t=1
t=1
where Y is the GDP series generated by equation (3.1) and P is the dependent variable of equation (3.7), the Phillips curve relationship, the ‘inflation bias’ of each regime. The βs are the weights assigned by the policy-maker to growth and inflation. With such a model, we will also test the different effects of domestic and external ‘shocks’ to key variables of the CEECs’ economies. For the external shock, an additional equation will be estimated, given by equation (3.9): ∗ Yt∗ = −α19 it−1 + μ,
(3.9)
which gives the effects in external demand from an increase in the Euroarea interest rate.
Alternative Paths to EMU
3
43
Data and procedures
Quarterly data series taken from the IMF/IFS database were used for all ten Central and Eastern European countries in our sample. Quarterly GDP was proxied by industrial production for Romania in the following manner: yearly GDP figures were divided in quarters and regressed on the available quarterly industrial production series. Again for Romania, government consumption was proxied by taking total government expenditures multiplied by the average share of the yearly government consumption in total government expenditures. A similar procedure was used for the missing parts of Polish and Hungarian government consumption series. M1 was used for money. The nominal exchange rate series are the nominal national rates to the euro. The REER series were also taken from the IMF, with the exception of Estonia and Lithuania, which were kindly provided by the domestic central banks, and for Latvia, which was calculated using the nominal exchange rate and CPI series, minus the Euroarea CPI series constructed as indicated below. The sample period goes from 1993:3 until 2001:4, not only to avoid the problems associated with the earlier years of transition, but to ensure a sample period in which all necessary data would be available for all countries, including the newly independent ones. This does not mean that data are actually available for all the countries for the full sample range: some of them only have data for a considerably shorter sample. For the Euroarea, the data were taken from the IMF/IFS series for the period 1993–97 and from the ECB for 1998 onwards. For the 1993–97 period, Euroarea GDP was built by aggregating the national quarterly GDP of the Euroarea member states (excluding Belgium, Greece, Ireland and Luxembourg, which do not produce quarterly GDP series: this implies an average loss of, roughly speaking, 5.25 per cent of Euroarea GDP). GDP-weighted average lending rates were constructed. For the same period, the CPI inflation rates were used for the construction of the – also GDP-weighted – Euroarea inflation rate (the later part of the sample uses the HIPC series produced by Eurostat). Before any estimation, the stationarity of the time series was analysed with augmented Dickey–Fuller tests for both level and first-differenced data. Partial autocorrelation graphs and the original series’ plots were also used as an aid to the diagnostic process. The residuals of the log HP filtered original series are level stationary (with the exception of four series, which are stationary after one differentiation).4
44
General Policy Issues for the Accession Countries
4
Estimation results
The estimation procedure was conducted as follows: first, the two simultaneous equations systems above were estimated by a heteroscedasticityconsistent OLS procedure. Afterwards, the series generated by this procedure were used for the estimation of comparative welfare and the simulation of shocks, using a vector auto-regression (VAR) procedure. 4.1 Estimated coefficients for both versions of the model The estimated coefficients and their standard errors (indicated as SE), plus their significance levels for the float and peg specifications are given below, in Tables 3.1 and 3.2, respectively. The name of the country is indicated in the first row, while the second shows the time sample used in the regression. As can be seen, the coefficients do not have the same values for individual countries in different regimes, but they fall within the intervals defined by their respective standard errors and they tend to have the same signs. Concentrating on the BP schedule, there are some indications that the significance of the coefficients for each specification seems to be related to the actual exchange rate regime followed by the country in question: when the country had a period of actual greater exchange rate flexibility, at least one coefficient was significant. This seems to be confirmed by the estimations of coefficients from regime-specific samples for countries with clearly defined peg and float periods (as some of the samples here are rather short – one with only nine observations – those results must be taken with care) shown in Table 3.3. Nevertheless, there are no systematic indications of this for actual pegs, as only Estonia had any significant variables on its peg BP equation among the three ‘classical’ CBA Baltic countries.5 Another interesting observation is that the coefficient α18 in Table 3.1 has the highest significant positive value for Bulgaria. This will change the welfare and shock-absorbing optimal exchange rate regime for this country only, as compared to the estimated outcomes in earlier versions of this model (see Vinhas de Souza and Ledrut, 2002). We may also observe from Tables 3.1 and 3.2 that the values of the coefficients of the BP schedule in the peg for Hungary and the Czech Republic are rather large (even after the GDP share correction made to this series). This is explained by the fact that those countries were the ones that attracted – by far – the greatest inflows of capital among those
α1 SE α2 SE α3 SE α4 SE α5 SE α6 SE α7 SE α8 SE α9 SE α10 SE α11 SE α12 SE α13 SE α14 SE α15 SE α16 SE α17 SE α18 SE α19 SE
−0.118 0.237 −0.004 0.023 −0.211 0.151 −0.152 0.181 0.198 0.210 −0.032 0.021 0.081 0.169 −0.059∗∗∗ 0.020 0.040 0.187 0.049 0.108 −0.005 0.003 0.027 0.049 0.009 0.026 −0.003 0.003 0.489∗∗∗ 0.045 0.572∗∗∗ 0.057 −0.044 0.031 0.123 0.208 −2.593 1.621
0.391∗∗ 0.220 0.045 0.065 −0.087 0.088 0.180 0.378 0.000 0.000 0.030 0.026 0.861∗∗∗ 0.270 −0.229∗∗∗ 0.081 0.483 1.173 0.055 0.154 0.000 0.026 0.365∗∗ 0.165 0.050 0.079 −0.011 0.009 0.434∗∗∗ 0.136 0.534∗∗∗ 0.135 −0.009 0.031 −0.095 0.075 −3.904∗∗ 1.422
0.125 0.185 −0.040∗∗∗ 0.007 −0.201 0.149 −1.839∗∗∗ 0.476 1.330∗∗∗ 0.377 0.092∗∗ 0.045 0.524∗∗∗ 0.130 −0.010∗∗ 0.004 0.882∗∗∗ 0.037 0.392∗∗∗ 0.072 0.017∗∗∗ 0.005 2.210∗∗∗ 0.356 0.545∗∗∗ 0.196 −0.152∗∗∗ 0.052 0.992∗ 0.067 0.065 0.065 −0.063 0.108 0.902∗ 0.102 −3.573∗∗ 1.596
93:3 01:1
93:3 00:4
94:2 00:2
Estonia
Czech Rep.
Bulgaria
0.335 0.209 −0.009 0.053 −0.486 0.476 −0.546 0.450 0.001∗ 0.000 0.000 0.018 −0.015 0.187 −0.070∗∗ 0.032 0.009 0.372 0.266∗∗∗ 0.099 0.001 0.013 0.300∗∗∗ 0.103 −0.009 0.052 −0.013∗∗∗ 0.005 0.577∗∗∗ 0.116 0.447∗∗∗ 0.101 −0.005 0.037 −0.277∗∗ 0.108 −0.253 2.031
96:2 01:2
Hungary
0.482∗∗ 0.140 0.014 0.009 −0.460∗∗∗ 0.076 −0.078 0.184 −0.169∗∗ 0.084 0.001 0.012 0.251 0.201 −0.036∗∗ 0.016 −0.447 0.420 0.186 0.143 −0.004 0.005 0.580 0.948 0.105 0.069 0.000 0.007 0.393∗∗∗ 0.079 0.619∗∗∗ 0.075 0.005 0.032 −0.189∗∗ 0.085 −1.437 1.716
94:1 01:2
Latvia
Estimated coefficients for the float specification
∗ 1% significance; ∗∗ 5% significance; ∗∗∗ 10% significance.
Y∗
PC
BP (S)
LM
IS
Table 3.1
0.012 0.221 0.016 0.015 −0.143 0.129 −0.098 0.313 0.175 0.222 −0.041 0.031 0.322∗∗ 0.140 −0.012 0.012 0.360 0.235 0.173∗ 0.102 −0.002 0.005 −0.145 0.178 0.165∗∗ 0.083 −0.005 0.012 0.451∗∗∗ 0.042 0.547∗∗∗ 0.054 −0.046 0.029 0.124∗∗∗ 0.064 −3.904∗∗∗ 1.422
93:3 00:4
Lithuania
0.403∗∗ 0.174 0.003 0.033 0.654 0.428 −0.686∗ 0.381 −0.060 0.746 −0.036 0.027 0.167 0.127 −0.097∗∗∗ 0.025 1.486∗∗∗ 0.399 0.350∗∗ 0.167 −0.000 0.011 0.333∗ 0.174 −0.059 0.064 −0.013 0.010 0.548∗∗∗ 0.132 0.467∗∗∗ 0.129 0.009 0.030 −0.158∗∗∗ 0.088 −2.554 1.859
95:2 00:1
Poland
1.152∗∗ 0.049 0.006 0.020 −0.453 0.372 −0.055 0.591 0.001 0.002 −0.131 0.106 −0.018 0.026 −0.015 0.010 0.296 0.206 0.239 0.170 −0.010 0.007 0.482∗∗ 0.212 −0.003 0.021 −0.052 0.042 0.662∗∗∗ 0.116 0.438∗∗∗ 0.110 −0.005 0.010 0.211 0.142 −2.589 1.708
95:4 00:4
Romania
0.324∗ 0.199 −0.002 0.021 −0.280∗∗ 0.114 −0.190 0.214 0.115 0.166 −0.035∗∗ 0.018 0.793∗∗∗ 0.281 −0.089∗∗∗ 0.020 1.050∗ 0.591 −0.176∗∗∗ 0.060 −0.002 0.007 0.413∗∗ 0.110 0.044 0.090 −0.003 0.006 0.338∗∗∗ 0.113 0.469∗∗∗ 0.100 −0.021 0.048 −0.258∗ 0.082 −3.904∗∗∗ 1.422
93:3 00:4
Slovakia
0.535∗∗ 0.240 −0.009∗ 0.005 −0.264 0.067 −0.201 0.242 0.195 0.132 0.001 0.008 −1.411∗∗∗ 0.427 −0.032∗∗∗ 0.010 −0.738 0.546 −0.198∗∗ 0.088 −0.002 0.002 0.380∗∗∗ 0.090 0.096 0.086 0.000 0.004 0.505∗∗∗ 0.084 0.556∗∗∗ 0.084 −0.009 0.053 0.011 0.077 −3.904∗∗∗ 1.422
93:3 00:4
Slovenia
α1 SE α2 SE α3 SE α4 SE α5 SE α6 SE α7 SE α8 SE α9 SE α10 SE α11 SE α12 SE α13 SE α14 SE α15 SE α16 SE α17 SE α18 SE
−0.097 0.205 −0.012∗∗∗ 0.004 −0.052 0.155 −0.136∗∗∗ 0.040 1.116∗∗∗ 0.379 −0.057 0.071 0.524∗∗∗ 0.130 −0.010∗∗ 0.004 0.882∗∗∗ 0.037 0.392∗∗∗ 0.072 0.004 0.007 0.023 0.062 −0.139 0.248 0.005 0.110 0.515∗∗∗ 0.086 0.525∗∗∗ 0.082 −0.105 0.231 −3.573∗∗ 1.596
94:2 00:2
Bulgaria
93:3 01:2
−0.052 0.229 −0.005 0.023 −0.286∗∗ 0.139 −0.094 0.102 0.103 0.205 −0.023 0.018 0.067 0.162 −0.059∗∗∗ 0.020 0.031 0.183 0.045 0.106 −0.046∗ 0.026 −0.266∗∗ 0.133 0.011 0.233 −0.011 0.022 0.489∗∗∗ 0.044 0.572∗∗∗ 0.056 −0.044 0.029 −1.735 1.642
0.422∗∗ 0.218 0.076 0.071 −0.083 0.088 −0.216 0.365 0.000 0.000 0.031 0.025 0.861∗∗∗ 0.270 −0.229∗∗∗ 0.081 0.483 1.173 0.055 0.154 70.615 107.889 683.153 644.008 −203.307 271.151 97.799∗∗∗ 31.083 0.478∗∗∗ 0.132 0.483∗∗∗ 0.130 −0.006 0.032 −3.904∗∗∗ 1.422
Estonia
93:3 00:4
Czech Rep
−0.270 0.233 0.000 0.039 −0.113 0.356 −1.018∗∗∗ 0.266 0.000∗ 0.000 −0.046∗∗ 0.018 −0.015 0.187 −0.070∗∗ 0.032 0.009 0.372 0.266∗∗∗ 0.099 −43.784 68.923 71.617 458.709 417.440 344.457 16.319 36.658 0.499∗∗∗ 0.127 0.480∗∗∗ 0.114 −0.014 0.042 −0.253 2.031
96:2 01:2
Hungary
Estimated coefficients for the peg specification
∗ 1% significance; ∗∗ 5% significance; ∗∗∗ 10% significance.
Y∗
PC
BP (B)
LM
IS
Table 3.2
0.418∗∗∗ 0.138 0.002 0.011 −0.441∗∗∗ 0.073 −0.210∗ 0.126 −0.127 0.084 −0.001 0.011 0.251 0.201 −0.036∗∗ 0.016 −0.447 0.420 0.186 0.143 0.004 0.021 −0.154 0.252 −0.207 0.248 0.027 0.023 0.324∗∗∗ 0.078 0.649∗∗∗ 0.079 −0.030 0.031 −1.437 1.716
94:1 01:2
Latvia
0.015 0.213 −0.005 0.020 −0.135 0.124 −0.254 0.176 0.067 0.221 −0.044 0.030 0.328∗∗ 0.143 −0.012 0.013 0.384∗ 0.239 0.163∗ 0.104 0.009 0.018 −0.077 0.191 −0.062 0.204 0.004 0.030 0.469∗∗∗ 0.043 0.555∗∗∗ 0.055 −0.021 0.026 −2.593∗ 1.621
93:3 01:1
Lithuania
0.181 0.211 −0.013 0.029 0.465 0.423 −0.715∗∗ 0.307 −0.009 0.701 −0.057∗∗ 0.028 0.167 0.127 −0.097∗∗∗ 0.025 1.486∗∗∗ 0.399 0.350∗∗ 0.167 0.022 0.014 0.030 0.150 −0.216∗∗ 0.104 0.005 0.014 0.521∗∗∗ 0.139 0.512∗∗∗ 0.135 0.009 0.031 −2.554 1.859
95:2 00:1
Poland
1.157∗∗∗ 0.049 0.013 0.022 −0.436 0.367 −0.172 0.286 0.001 0.002 −0.181 0.124 −0.018 0.026 −0.015∗ 0.010 0.296 0.206 0.239 0.170 4.225 5.431 29.882 85.736 3.528 14.851 5.168 37.670 0.544∗∗∗ 0.087 0.540∗∗∗ 0.088 −0.004 0.011 −2.589 1.708
95:4 00:4
Romania
0.198 0.184 −0.010 0.017 −0.247∗∗ 0.099 −0.549∗∗ 0.212 −0.027 0.146 −0.020 0.015 0.793∗∗∗ 0.281 −0.089∗∗∗ 0.020 1.050∗ 0.591 −0.176∗∗∗ 0.060 0.066∗∗ 0.027 0.668 0.480 0.396 0.380 0.059∗ 0.032 0.471∗∗∗ 0.120 0.480∗∗∗ 0.115 −0.011 0.055 −3.904∗∗∗ 1.422
93:3 00:4
Slovakia
−0.096 0.212 0.002 0.005 −0.158∗∗∗ 0.053 −0.490∗∗∗ 0.103 0.097 0.097 0.004 0.005 −1.411∗∗∗ 0.427 −0.032∗∗∗ 0.010 −0.738 0.546 −0.198∗∗ 0.088 0.006 0.017 0.126 0.437 −0.465 0.629 0.004 0.025 0.501∗∗∗ 0.078 0.559∗∗∗ 0.080 −0.007 0.051 −3.904∗∗∗ 1.422
93:3 00:4
Slovenia
Table 3.3
Coefficients for regime-specific samples Bufloat
α1 SE α2 SE α3 SE α4 SE α5 SE α6 SE LM α7 SE α8 SE α9 SE α10 SE BP α11 SE α12 SE α13 SE α14 SE PC α15 SE α16 SE α17 SE Y∗ α18 SE IS
Czfloat
Safloat
Bupeg
Czpeg
Sapeg
94:2 97:1 97:4 00:4 98:4 00:4 97:2 00:2 93:3 00:2
93:3 98:3
1.105∗ 0.599 −0.059∗∗∗ 0.017 −0.099 0.635 −2.639∗∗∗ 0.940 0.582 0.638 0.063 0.105 0.320 0.275 −0.002 0.008 0.931∗∗∗ 0.077 0.274∗∗ 0.126 0.035∗∗∗ 0.008 1.406∗∗∗ 0.418 0.263 0.264 −0.239∗∗ 0.078 0.548∗∗∗ 0.194 0.528∗∗∗ 0.130 −0.181 0.617 −4.613∗ 2.806
0.154 0.298 −0.012 0.029 −0.182 0.190 −0.595∗ 0.331 0.052 0.235 −0.019 0.020 1.070∗∗∗ 0.349 −0.074∗∗∗ 0.026 1.242∗ 0.704 −0.156∗∗ 0.064 0.106∗∗ 0.041 0.918 0.659 0.752 0.531 0.058∗ 0.037 0.462∗∗∗ 0.084 0.538∗∗∗ 0.076 −0.034 0.037 −7.412∗∗∗ 2.585
0.257 0.374 0.094 0.149 0.006 0.120 −0.060 0.827 0.001 0.001 0.067 0.067 0.370 0.382 −0.046 0.102 −1.868 1.461 0.076 0.254 0.051 0.062 0.438 0.331 0.061 0.155 0.017 0.025 0.482∗∗ 0.220 0.491∗∗∗ 0.213 −0.013 0.069 −2.489 1.571
0.088 0.254 −0.031 0.035 −0.272∗∗ 0.130 −0.422 0.379 −0.350 0.408 −0.058 0.050 −0.193 0.633 −0.150∗∗∗ 0.042 −0.997 1.323 −0.162 0.336 −0.030∗∗ 0.013 0.942∗∗∗ 0.301 −0.198 0.178 0.029 0.039 0.509 0.363 0.352 0.374 0.018 0.180 −2.015∗∗ 1.006
−0.452∗∗ 0.266 0.205 0.342 −0.055∗∗∗ −0.076 0.012 0.138 −0.158 −0.266∗ 0.105 0.152 −0.288∗∗∗ 1.049 0.068 0.724 2.470∗∗∗ 0.000 0.562 0.000 −0.021 0.049 0.100 0.039 0.552∗∗∗ 0.446 0.120 0.357 −0.015∗∗ −0.971∗∗∗ 0.004 0.242 0.846∗∗∗ −0.142 0.028 2.542 0.585∗∗∗ −0.235 0.140 0.187 0.007∗ 314.615 0.004 305.186 −0.041 1578.601 0.053 1652.058 −0.157 −382.432 0.141 643.051 −0.110 136.718∗ 0.090 75.652 0.402∗∗ 0.478∗∗∗ 0.045 0.201 0.659∗∗∗ 0.477∗∗ 0.063 0.224 −0.020 −0.016 0.045 0.025 −3.183 −5.213∗∗ 2.062 2.303
∗ 1% significance; ∗∗ 5% significance; ∗∗∗ 10% significance.
48
General Policy Issues for the Accession Countries
in our sample, given their positions as ‘early reformers’ (see Chapters 8 and 10). During some periods in our sample, the positive inflow of capital surpassed 30 per cent of the quarterly Czech GDP; after the collapse of its peg regime in 1997, the inflows quickly reversed, reaching as low as −10 per cent of its GDP. The Lucas critique is certainly an important question concerning this work. If we were to assume the coefficients of the fundamental variables to be conditional on the policy choice, as they are derived from the actual data series, it would imply that they were determined by the current exchange rate regime. It would not be possible to derive two sets of series characterizing different regimes from the same datagenerating process. As it turns out, our estimated coefficients are quite similar for all key variables (and all differences fall within the range defined by the standard errors), with the exception of the BP schedule, but this is due to the fact that the BP schedule is generated by a different equation for each regime. We may relate our results to the very interesting recent work by Rudebusch (2002), where this author, testing for the actual empirical and economic relevance of the Lucas critique, estimates that reduced-form specifications – both backwardand forward-looking – are rather insensitive to policy shifts.6 We will, therefore, proceed assuming that the Lucas critique argument is not empirically relevant here, namely, that the coefficients are structurally stable within the used estimation sample, and use the series generated by the estimated coefficients in the welfare comparisons and shock simulations below.
5
Welfare effects of exchange rate regime choices
Results by country of the estimated welfare functions are shown in Table 3.4. The weights given to output and inflation by the policymaker were set to vary between 1.00 and 0.00, 0.75 and 0.25, 0.67 and 0.33, 0.50 and 0.50, 0.33 and 0.67, 0.25 and 0.75 and 0.00 and 1.00. The ‘optimal’ regimes in each combination are indicated in bold italic. As an exchange rate strategy, the float seems to dominate the peg: six out of ten countries are better off with it, and even apparently obvious candidates for a harder regime, due to size or stabilization considerations, like Estonia and Latvia, would seem to fare better under a more flexible regime. The peg seems only to clearly dominate in economies still in need of macro-stabilization, with credibility problems for their monetary
Alternative Paths to EMU Table 3.4
The loss-function outcomes β1 0.75 β2 0.25
β1 0.67 β2 0.33
β1 0.50 β2 0.50
β1 0.33 β2 0.67
Float 0.218 −1.836 Peg −0.136 −0.113 Czech Rep. Float 0.006 0.002 Peg −0.005 −0.005 0.054 0.025 Estonia Float Peg 0.046 0.020 Hungary Float 0.257 0.188 Peg −0.077 −0.055 Latvia Float −0.111 −0.074 Peg −0.211 −0.145 Lithuania Float −0.306 −0.226 Peg −0.242 −0.183 Poland Float 0.246 0.183 Peg −0.413 −0.311 Romania Float −83.131 −62.531 Peg −62.515 −46.973 Slovakia Float −0.137 −0.102 Peg −0.133 −0.097 Slovenia Float 0.124 0.094 Peg 0.019 0.014
−2.493 −0.105 0.001 −0.005 0.016 0.166 −0.048 −0.062 −0.124 −0.201 −0.164 0.162 −0.279 −55.939 −41.999 −0.091 −0.085 0.084
−3.890 −0.089 −0.002 −0.004 −0.004 −0.006 0.119 −0.033 −0.037 −0.079 −0.146 −0.124 0.119 −0.209 −41.930 −31.430 −0.068 −0.061 0.063
−5.286 −0.072 −0.004 −0.004 −0.023 −0.023 0.073 −0.018 −0.012 −0.035 −0.092 −0.085 0.075 −0.139 −27.922 −20.861 −0.044 −0.036 0.042
0.013
0.009
0.006
Country
Bulgaria
49
Regime
β1 1.00 β2 0.00
0.012
β1 0.25 β1 0.00 β2 0.75 β2 1.00
−5.944 −0.065 −0.006 −0.004 −0.033 −0.031 0.051 −0.011 0.000 −0.014 −0.066 −0.066 0.055 −0.107 −21.330 −15.887 −0.033 −0.025 0.033
−7.997 −0.041 −0.010 −0.003 −0.061 −0.057 −0.018 0.011
0.037 0.052 0.014 −0.007 −0.009 −0.005 −0.729 −0.345
0.001
0.012 0.002 0.004 −0.001
and/or fiscal authorities, or with some shaky fundamentals (Bulgaria, Lithuania, Romania and Slovakia). More than that, for most countries, the ‘optimal’ exchange rate strategy is indifferent to changes in the combinations of the parameter weights in the loss function, only switching to the other regime at the extreme end of the distribution (e.g. if a zero weight is attributed to growth in the welfare function, while all weight is given to inflation stabilization: the ‘inflation-nutter’ scenario). The exception to this is Bulgaria, where, due to the extreme pass-through, the switch occurs at the other extreme – a zero weight on inflation – of the distribution. The Czech Republic and Estonia are somewhat exceptions as well, favouring either regime within a credible range of parameters, but with the weight of the distribution skewed towards a float.
50
General Policy Issues for the Accession Countries
6 Non-structural estimation of the effects of domestic and foreign shocks The effects of different shocks under each exchange rate arrangement will be simulated via a non-structural approach, namely, through a VAR (vector auto-regression) procedure upon the arrangement-specific estimated series. In the VAR, three types of shocks are simulated: (i) a domestic fiscal shock (a 1 standard deviation unexpected shock to the government consumption expenditures); (ii) a domestic monetary shock (a 1 standard deviation unexpected shock to the nominal interest rate); (iii) an external monetary shock (a 1 standard deviation unexpected shock to the Euroarea nominal interest rate). The two domestic shocks can be seen in terms of the effects of a nominal convergence process, i.e., as part of an attempt by the country to fulfil the Maastricht criteria, while the last shock can be seen as a measure of the degree of integration and vulnerability of these economies to the Euroarea. In Table 3.5 we present an overview of the effects of the VAR simulated shocks on the variables that we use in our welfare function, GDP and CPI inflation: a plus sign (+) means that the shock (fiscal, monetary domestic or foreign) has a positive initial effect on the variable in question (GDP or prices), a minus sign (−) the opposite. In general, a float regime seems to outperform a harder regime as a ‘shock absorber’ for most countries, if all shocks are weighted equally. The clear exceptions are Bulgaria, Lithuania and Poland, while for Latvia and Slovakia both regimes seem to perform similar cushioning functions, and for Romania shocks have ‘explosive’ effects under both regimes, but less under a float. Most shocks not only have smaller GDP and CPI effects under a float, but the generated series also converge faster to the mean. The most consistent exception to this stylized picture is the external monetary shock: if one were to consider this as the most important or likely type of shock for an economy such as, say, Estonia, then the peg would be a better shock absorber. In Table 3.5 we can also observe what we may call ‘non-Keynesian’,7 or non-‘MF’, results, from monetary policies that are effective under a peg to fiscal ones that are effective under a float, or to expansionary fiscal and monetary contractions. GDP expansions under fiscal contraction were estimated for Bulgaria and Slovenia, and GDP expansions under tighter
Alternative Paths to EMU Table 3.5 Country
Bulgaria Czech Rep. Estonia Hungary Latvia Lithuania Poland Romania Slovakia Slovenia
51
Overview of effects of temporary shocks by country and regime Regime
Peg Float Peg Float Peg Float Peg Float Peg Float Peg Float Peg Float Peg Float Peg Float Peg Float
Fiscal shock
Monetary shock
External shock
GDP
CPI
GDP
CPI
GDP
CPI
+ − − − −− − − − −− −− − − 0 + −− −− −− −− + −−
− −− − + + + − + − − −− − 0 − −− −− − − + +
+ − + − − − − − − − + + + ++ ++ ++ + + ++ +
− ++ + + − − − − − − + + − − ++ ++ + + − +
− + + + + + + + + + − + + + −− −− − − + −
++ − + + − ++ ++ ++ + + − + + + −− −− − − −− ++
domestic monetary conditions were estimated for Bulgaria, the Czech Republic, Lithuania, Poland, Romania, Slovakia and Slovenia (some of those outcomes are regime-dependent). We can see that, in most cases, the same regime that is ‘welfare-optimal’ for a given country is also optimal as a ‘shock absorber’ for that same country. Part of these outcomes can be explained by the less than perfect degree of capital mobility in the countries in our sample during the period in question.8 It is a common result in MF models that, under less than complete capital mobility, both types of policy can be partially effective under both regimes, and that is indeed the case for most of the CEECs. The estimated coefficients that would in principle capture capital mobility in our models are, on average, rather low and several are even negative. A possible explanation for this could be the adverse reaction of capital inflows – especially the short-term ones – observed during the 1997 and 1998 Asian and Russian crises. Contractionary fiscal and monetary policies with observed positive growth effects could be the
52
General Policy Issues for the Accession Countries
sign of a rational expectations channel in operation in some of those countries. On the other hand, the large standard errors, the lack of significance of several coefficients and the average low explanatory power of the BP schedule equation do suggest caution in considering those results. Those caveats are possibly caused by some short-run features of the economies that are present in the limited data series used and captured by the estimated coefficients, like the characteristic reduction of inflation parallel to a resumption of growth after the end of the ‘transitional’ recession, and the reaction to shocks and even episodes of contagion during the sample period.
7
Conclusions
We aimed in this chapter to describe the optimal exchange rate strategy for integration of the CEECs into the common European currency zone. The results from a formal modelling exercise of alternative exchange rate regimes for pre-EMU accession for all Eastern European and Baltic ACs seem to indicate that a float regime would bring about, as a rule, a greater degree of aggregate welfare and would also be a better shock absorber for temporary shocks – if all shocks are weighted equally – for most of the countries. Harder regimes would be indicated for countries with weaker credibility and macroeconomic foundations. The welfare results seem to be robust to changes in the policy-maker’s preferences, as expressed in the weights given to the GDP and inflation parameters of the welfare function. Additionally, in most cases, the same regime that is optimal in terms of welfare (GDP) for a given country is also optimal as a shock absorber for that same country. These results would therefore indicate that some current exchange rate regime choices are not the optimal ones for some of the CEECs, suggesting potential welfare and stabilization gains from a regime switch. From the EU perspective, the practical policy implications seem to be that different regimes should be allowed to remain at least until ERM II entry, instead of trying to impose a single framework, as any unique framework might be welfare-reducing for at least some of the CEECs.
Notes 1. The current pre-entry linkage strategies collapse to, in essence, a peg or a float: the remaining exception to this, Hungary, became a floater within a band in
Alternative Paths to EMU
2.
3. 4. 5.
6. 7.
8.
53
mid-2001 (for works that model this strategy, see Golinelli and Rovelli, 2002 and Bofinger and Wollmershäuser, 2001). On applications of variants of the MF model, see Wdowinski and van Aarle (1998), Plasmans (1999), and Roberts and Tyers (2001). For extensions of Dornbusch-type models with policy rules à la Taylor, see Svensson (1997), Leitemo and Røisland (2000) and Bergvall (2000). Another specification, using net current and financial accounts in log levels, was tested and discarded. The estimated ADF statistics are available upon request. A possible explanation for this may be that, given that the Lithuanian CBA was linked to the USD during the whole sample used and that the Latvian hard peg is linked to the SDR, they effectively behaved as floating currencies towards the euro. During the period in question, the Estonian kroon fluctuated ±2.16 per cent towards the euro, while the litas varied by ±8.86 per cent and the lats by ±13.37 per cent. Those two last values are greater than those shown by the Czech and Slovak korunas during their float periods (namely, ±5.49 per cent and ±3.36 per cent respectively), or the Polish zloty (±6.60 per cent), and closer to the variability showed by the Slovenian tolar (±14.02 per cent). The only currencies showing higher nominal variability are the Bulgarian lev, during its float period (±93.04 per cent) and the Romanian leu (±83.66 per cent). Rudebusch (2002) echoes very similar results obtained much earlier by Taylor (1989). One could explain these ‘non-Keynesian’ outcomes by a situation where a contractionary stance by the central bank or by the government is seen as an indication of a more sustainable policy by the markets (Giavazzi and Pagano, 1990). As a remark, Kamps (2001), in a very interesting paper, attributes those earlier results by Giavazzi and Pagano to, in essence, a ‘statistical artefact’ of their dataset. The predictions of an MF (effective fiscal policy under a peg, effective monetary policy under a float) are derived under the assumption of capital mobility: this implies that, for the expected outcomes to be observed, the coefficient(s) α11 an should be ‘large’. For actual capital mobility indicators for the CEECs, in an index from 0 to 100, where 100 indicates full liberalization (see IMF, 2000), Estonia and Latvia score 97.6, Lithuania 85.7, the Czech Republic 73.7 and Hungary 59.5, while a ‘larger’ economy like Poland scores 55.3, Slovenia 40.5, Bulgaria 35.3, Slovakia 23.7 and Romania, the least liberalized in the group, 12.5. The average, non-GDP-weighted, is 58.14. It must be noted that the index above was computed in 1997 – around the middle of our sample – and that now it is certainly higher, especially among the relative laggards like Bulgaria, Slovenia and Slovakia (but with the possible exception of Romania), given that capital account liberalization is a (pre)requisite for EU membership.
4 Macroeconomic Adjustment in EU Accession Countries: An Analysis Using a Small Macroeconomic Model Bas van Aarle, Joseph Plasmans and Bruno Merlevede
1
Introduction
The Central and Eastern European countries (CEECs) have been in a process of rapid change during recent years. First, there is the still ongoing process of restructuring and transformation from a former planned economy towards a market economy. This process has now progressed so far that the institutional structures in these countries increasingly resemble other market economies. Second, these countries have been accepted as members of the European Union (EU) starting from May 2004 onwards (the accession of Bulgaria and Romania will occur from 2007 onwards). Accession is conditional upon the fulfilment of the Copenhagen criteria laid down by the EU in 1993: (i) stability of institutions guaranteeing democracy, the rule of law, human rights and respect for and protection of minorities; (ii) the existence of a functioning market economy as well as the capacity to cope with competitive pressure and market forces within the Union; (iii) the ability to take on the obligations of membership including adherence to the aims of political, economic and monetary union; (iv) the adjustment of their administrative structures, so that European Community legislation is transposed into national legislations and implemented effectively through appropriate administrative and juridical structures. Because of the approaching accessions and for various other reasons, it has become increasingly important to gain a better insight into
54
Macroeconomic Adjustment in Accession Countries
55
the macroeconomic structures of these countries, their macroeconomic policies and macroeconomic adjustment dynamics. Recent years have witnessed both similar and dissimilar adjustment patterns across countries. Notwithstanding considerable variation across countries in the exact patterns, in most cases a process of (i) gradual growth recovery, (ii) disinflation, (iii) significant capital inflows, (iv) deteriorating current account and fiscal balances, (v) real appreciation and (vi) stagnating employment has been seen since 1995. During recent years the CEECs have implemented a broad range of monetary and exchange rate strategies. A considerable number of studies have been produced on the monetary strategies and monetary transmission in the CEECs. Detailed studies, as in OENB (2001), Vinhas de Souza et al. (1999) and Vinhas de Souza and Ledrut (2002) reveal the large variation in the monetary policy strategies that have been followed and the changes and complications that have resulted. This chapter develops a macroeconomic model of the CEECs to analyse macroeconomic adjustment in the Accession Countries. The model consists of the basic macroeconomic relations that govern macroeconomic adjustment. It gives insight into both the adjustment of the internal balance (output, employment and so on) and the external balance (exports, competitiveness and so on) in ten Accession Countries and also allows us to analyse how domestic macroeconomic policies and integration with the EU affects that adjustment. Cross-country comparisons will allow us to discern similarities and dissimilarities in the adjustment patterns. The models are of relatively small size, given the limited data availability for almost all Accession Countries. When available, the time series are of limited size, the quality of the data is generally not yet optimal, and the statistical concepts used not always fully comparable. On the other hand, the small size has also advantages, given its simple structure, and also because it conforms to the recent general trend to construct small-scale macroeconomic models for policy analysis. Because of a relatively detailed modelling of the monetary sphere, this chapter develops tools for monetary policy analysis during the accession phase. During this phase, countries are also preparing to enter EMU, once having complied with the Maastricht criteria. With EU accession, the countries are expected to participate in the ERM II framework. In their preparation for that, they remain unrestricted in respect of the monetary and exchange rate policies that they implement. There is now a consensus that in the pre-accession phase no regime fits all; hence the broad range of strategies, ranging from a free float to a currency board.
56
General Policy Issues for the Accession Countries
This chapter also extends earlier macroeconomic modelling for CEECs. Without claiming completeness, we note, for example, work by Hall et al. (2000), who survey the main technical and practical problems that macroeconometric models of transition countries face. Golinelli and Rovelli (2002) estimate a small macroeconomic model for the case of Hungary and Poland. The relation between monetary policy and disinflation in these countries is then analysed by simulating alternative interest rate policies. Charemza (1994) develops a macroeconometric model for the CEECs and uses it for forecasting and simulation of the economies of the Czech Republic, Hungary, Lithuania, the Slovak Republic and Poland. Juszczak et al. (1993) and Klein et al. (1999) develop quarterly models for Poland and discuss their main properties. The following structure is followed: Section 2 presents a small loglinear macroeconomic model. Section 3 estimates the model outlined for the ten Accession Countries using quarterly data for the period 1993– 2001. Section 4 uses the model for in-sample simulations and macroeconomic policy experiments. Section 5 summarizes our main results.
2
A small macroeconomic model of Accession Countries
A number of factors are generally assumed to be crucially important in the Accession Countries during the adjustment phase towards entering of the EU and later the EMU: (i) the specific importance of macroeconomic fluctuations in the current EU and other aspects of real and nominal convergence with the EU in general; (ii) the importance of fiscal adjustments that have been and will need to be undertaken; (iii) the possibility of large (and volatile) capital flows, mostly in the form of FDI inflows and short-run speculative capital flows; (iv) the important adjustments in the labour market; (v) the important role of monetary and exchange rate policies during the accession process. In an admittedly stylized manner, these features therefore have a crucial role in our small model. The base of the model that will be estimated in this section for the accession economies consists of a small dynamic open economy AD–AS– LM model with price and wage dynamics. It consists of eight estimated macroeconomic relations and 12 definitions (see Table 4.1). denotes a first difference of a variable; EU variables are indicated by an EU superscript. All variables are in domestic currency, unless otherwise indicated. Equation (4.1) defines real private consumption, RCON, as a function of real disposable income, RYDP, and the real interest rate, RSIN. Real consumption is obtained by deflating private consumption, CON, by
Macroeconomic Adjustment in Accession Countries
57
Table 4.1 A small macroeconomic model of Accession Countries: relations and definitions log(RCON) = α1 + α2 (RSIN) + α3 log(RYDP) log(RINV ) = β1 + β2 (RSIN) + β3 log(RGDP) + β4 log(FDI)
(4.1) (4.2)
log(REXP) = γ1 + γ2 log(REUR) + γ3 log(RGDP EU ) + γ4 log(WTR)
(4.3) (4.4)
log(RIMP) = δ1 + δ2 log(REUR) + δ3 log(RGDP) + δ4 log(OIL) M1 log = ζ1 + ζ2 SIN + ζ3 log(RGDP) PPI RWAG log(EMP) = λ1 + λ2 log(RGDP) + λ3 log PRO log(WAG) = ν1 + ν2 log(PPI) + ν3 log(UNE) + ν4 log(PRO) log(PPI) = η1 + η2 log(WAG) + η3 (EUR ∗ PPI EU ) + η4 log(OIL) + η5 log(RGDP) YDP ≡ GDP − REV + GEX − GCO RSIN ≡ SIN − log(PPI) GDP ≡ CON + INV + EXP − IMP + GCO + CIN REUR ≡
EUR ∗ PPI EU PPI
M1 ≡ MMP ∗ M0 M0 ≡ CBC + CLG − CGD + RES ∗ EUR RES ≡ CUA + CAA CUA ≡ (EXP − IMP) ∗ EUR CAA ≡ FDI + OCF GDP/PPI PRO ≡ EMP LAB ≡ EMP + UNE DEF ≡ REV − GEX
(4.5) (4.6) (4.7) (4.8) (4.9) (4.10) (4.11) (4.12) (4.13) (4.14) (4.15) (4.16) (4.17) (4.18) (4.19) (4.20)
the domestic price level, which is approximated by the producer price index, PPI. Disposable income, YDP, is defined in a relatively crude way in equation (4.9) – for reasons of data availability – using GDP and fiscal balances. The real interest rate is defined in equation (4.10) as the nominal interest rate, SIN, minus inflation. GDP is defined in equation (4.11) as the sum of consumption, investment, INV, exports minus imports EXP– IMP, government consumption, GCO, and inventory accumulation, CIN. Real private investment, RINV, equation (4.2), is assumed to depend on the real interest rate (by a cost-of-capital argument), real output (by an ‘accelerator’ argument), RGDP, and foreign direct investments, FDI. Real exports, REXP, in equation (4.3) depend on competitiveness visà-vis the EU, REUR – defined in equation (4.12) as the nominal euro
58
General Policy Issues for the Accession Countries
exchange rate, EUR times the relative output price level – EU output, RGDP EU and world trade, WTR. Similarly, imports, RIMP, in equation (4.4) depend on competitiveness vis-à-vis the EU, domestic real output and the oil price, OIL. Money market equilibrium is given in equation (4.5). The money supply, M1, in equation (4.13) is determined by the workings of the money multiplier, MMP, on the stock of base money, M0. Base money, equation (4.14), consists of a domestic component – credit of the Central Bank to the banking sector, CBC, and to the government, CLG − CGD – and a foreign component – the foreign exchange reserves, RES. Labour demand, EMP, in equation (4.6) is a function of real output and the real producer wage RWAG corrected for productivity, PRO, which serves as a proxy of (real unit) labour costs. The real wage equals the nominal wage, WAG, deflated by the price level. In equation (4.18) productivity is defined as real GDP per employee. The supply of labour, LAB, is defined in equation (4.19) as the sum of employed and unemployed persons, UNE. Wage inflation according to equation (4.7) is driven by increases in output prices, reflecting wage indexation, the level of unemployment reflecting a Phillips-curve element, and changes in productivity. The last effect could reflect the pressure on wages (and thereby on prices) from the Balassa–Samuelson effect that is often thought to have significant inflationary impacts in Accession Countries and to be an important factor behind the trend real appreciation noticed in many countries. Increases of domestic prices in equation (4.8) are the result of wages increases, increases of foreign prices and increases in oil prices (which are three cost-push arguments) and the level of output (a business-cycle/demand-pull argument).1 The effect of foreign prices proxies the amount of (exchange rate) pass-through in the economy. The balance of payments (defined in euros) is defined in equations (4.15)–(4.17), expressing in equation (4.15) that the sum of the current account, CUA, and capital account, CAA, is matched by a change in foreign exchange reserves. The current account, equation (4.16), equals exports of goods and services minus imports. The capital account, equation (4.17), equals foreign direct investment, FDI, and other capital flows, OCF, which in Accession Countries consist to a large extent of short-run portfolio capital flows. Both FDI and other capital flows remain exogenous in the model, for simplicity. The fiscal deficit, DEF, is defined in equation (4.20) as the difference between total government revenue, REV, and total government spending, GEX, which are both exogenous in the model.
Macroeconomic Adjustment in Accession Countries
59
The model can be used to represent and analyse alternative monetary and exchange rate regimes: (i) a pure floating exchange rate regime, for example, combined with inflation targeting, leaves the exchange rate determined by the financial markets, and short-term interest rates or base money are monetary policy instruments that can be targeted by the monetary authorities (i.e. they become exogenous policy variables in the model); (ii) a managed float of the exchange rate implies that exchange rates adjust according to some predetermined (and possibly announced) path and that interest rates are operated to secure the adjustments (see Bofinger and Wollmershäuser, 2001); (iii) a fixed exchange rate combined with endogenous base money and interest rate adjustment; (iv) a currency board that combines a fixed exchange rate with a fixed monetary base (see in particular Nenovsky and Hristov, 2002, and Lättemäe, (2001), and Pikanni, (2001), for a detailed analysis of currency boards and the experiences in their operating in Bulgaria and Estonia, respectively).
3
Model estimation
Here, the macroeconomic model presented in the previous section is estimated for the following Accession Countries: Bulgaria (BUL), the Czech Republic (CZR), Estonia (EST), Hungary (HUN), Latvia (LAT), Lithuania (LIT), Poland (POL), Romania (ROM), Slovakia (SLO) and Slovenia (SLV). When estimating, we need to take into account a number of aspects: (i) the limited quality of the data (e.g. restricted number of observations); (ii) seasonal patterns in the data; and (iii) non-stationarity of almost all variables. The Appendix provides details on the dataset that is used. Notwithstanding the limited availability and quality of macroeconomic data at a quarterly frequency, it turns out that our dataset generally allows estimation of the above model in the case of the ten Accession Countries. The estimation results for the consumption, investment, exports, imports, money balances, labour demand, wage inflation and price inflation relations are provided in Tables 4.2(a)–(h). These structural equations of the model are estimated in the form of a vector error correction model using one lag and one co-integrating vector in its specification (VECM(1,1)). A VECM is a restricted vector auto-regressive model designed for use with non-stationary series that are co-integrated. The VECM has the co-integration relations built into the specifications
−0.376 [−1.11] 0.680∗∗ [13.88]
CZR
−0.024 [−0.12] 0.820∗∗ [21.87]
EST
0.288 [0.85] 1.149∗∗ [7.11]
HUN
−4.845∗∗ [−6.04] 0.944∗∗ [2.97]
LAT
−0.046∗ [−1.85] 0.952∗∗ [36.80]
LIT
−0.331 [−1.55] 0.967∗∗ [27.96]
POL
−0.726∗∗ [−5.67] 0.935∗∗ [12.96]
ROM
0.135 [0.86] 0.964∗∗ [44.51]
SLO
−0.275∗∗ [−3.22] 0.611∗∗ [17.12]
SLV
Note: t -statistics in parentheses, ∗ (∗∗ ) denotes significance at 10% (5%) levels.
−0.637∗∗ −0.404∗∗ −0.319∗∗ −0.053 −0.933∗∗ −0.074 −0.430∗∗ −0.268∗∗ −1.313∗∗ −0.935∗∗ [−5.74] [−5.32] [−2.85] [−4.62] [−0.20] [−2.58] [−1.38] [2.45] [−2.86] [−3.82] 0.039 0.115 0.164 −0.218 0.376∗∗ 0.108 −0.679∗∗ 0.283∗ 0.091 D(LOG(RCON(−1))) 0.520∗∗ [4.32] [0.30] [0.64] [1.53] [−1.08] [2.16] [0.47] [−3.37] [1.69] [0.41] −0.091 −0.497∗∗ −0.010 0.238 D(SIN(−1)-D(LOG(PPI(−1)))) 0.003 0.251 0.172 −0.334 0.045 0.386∗∗ [0.09] [0.50] [1.26] [1.13] [0.22] [−2.81] [−0.46] [−4.19] [−0.07] [1.37] ∗∗ ∗∗ ∗∗ ∗∗ ∗∗ ∗∗ −0.244 −0.406 0.001 0.440 −0.447 0.113 0.596 −0.165 −0.154 D(LOG(RYDP(−1))) −0.629 [−5.81] [−5.53] [−4.07] [0.00] [2.78] [−2.34] [1.54] [2.97] [−1.62] [−0.76] 0.941 0.925 0.846 0.784 0.750 0.644 0.834 0.681 0.678 0.646 Adj. R2 Sample 94:III–01:III 93:III–01:IV 93:IV–01:IV 93:III–01:IV 95:III–01:III 93:III–01:IV 94:III–01:IV 94:III–01:III 93:III–01:III 93:III–01:IV (29 obs.) (34 obs.) (33 obs.) (34 obs.) (25 obs.) (34 obs.) (30 obs.) (29 obs.) (33 obs.) (34 obs.)
Error correction: CointEq1
LOG(RYDP(−1))
−0.255∗∗ [−11.78] 1.010∗∗ [17.54]
BUL
Estimation results for consumption
Cointegrating Eq: SIN(−1)-D(LOG(PPI(−1)))
D(LOG(RCON))
Table 4.2(a)
−0.410∗ [−1.89] 4.920∗∗ [8.64] 0.375∗∗ [6.01] −2.533 [−1.52] 0.046∗∗ [0.11] −0.098 [−1.55]
1.909∗∗ [3.47] 0.861∗∗ [14.28] −0.085∗∗ [−2.35]
EST
0.079 [0.39] 0.170∗∗ [2.11] −0.214∗∗ [−10.44]
HUN
Note: t -statistics in parentheses, ∗ (∗∗ ) denotes significance at 10% (5%) levels.
−0.313∗∗ −0.398∗∗ −0.182 −0.262∗∗ [−3.60] [−4.78] [−2.81] [−1.37] D(LOG(RINV(−1))) −0.152 −0.214 0.354 −0.273∗ [−1.02] [−1.36] [1.50] [−1.91] 0.561 −0.739∗ 0.304 D(SIN(−1)-D(LOG(PPI(−1)))) −0.158∗∗ [0.46] [−1.82] [0.84] [−2.72]∗∗ ∗∗ −0.005 −0.271 0.254 D(LOG(RGDP(−1))) −0.701 [−2.33] [−0.02] [−1.01] [1.36] 0.015 0.017 D(LOG(FDI(−1)/PPI_EU(−1))) −0.017 0.038∗∗ [−0.48] [2.30] [0.653] [0.92] 0.933 0.956 0.541 0.994 Adj. R2 Sample 94:III–01:III 93:III–01:IV 93:IV–01:IV 93:III–01:IV (29 obs.) (34 obs.) (33 obs.) (34 obs.)
Error Correction: CointEq1
LOG(FDI/PPI_EU)
LOG(RGDP(−1))
Cointegrating Eq: SIN(−1)-D(LOG(PPI(−1)))
BUL
D(LOG(RINV))
CZR
Estimation results for investment
Table 4.2(b)
0.478∗∗ [2.18] 3.209∗∗ [8.40] 0.309∗∗ [4.60]
LIT
2.330 [0.90] 0.946∗∗ [18.60] 0.167∗ [1.77]
POL
0.848∗∗ [2.97] 0.913∗∗ [59.01] 0.155∗∗ [2.21]
ROM
−0.475 [−0.28] 1.017∗∗ [42.10] −0.026 [0.33]
SLO
−0.505∗∗ [−3.94] 1.241∗∗ [18.01] 0.011 [0.94]
SLV
−0.097 −0.102 −0.249∗∗ 0.040 −0.279∗∗ −0.131 [−0.71] [−1.14] [−5.62] [0.35] [−10.99] [−0.45] −0.419∗∗ 0.292∗ 0.472 0.068 −0.075 −0.548∗∗ [−2.82] [1.66] [0.91] [0.57] [−0.52] [−2.07] 1.087 −0.624∗∗ 1.638 −0.249 2.058∗∗ 0.144 [1.60] [−3.24] [1.11] [−1.48] [3.24] [0.48] ∗ −0.812 −0.344 −2.409 −0.463 −0.985 0.240 [−1.08] [−1.19] [−1.56] [−1.92] [−1.37] [0.56] 0.039 0.054 0.041 −0.006 −0.004 −0.013 [0.80] [1.54] [0.58] [−0.10] [−0.28] [1.03] 0.864 0.907 0.786 0.902 0.918 0.646 94:I–01:III 93:III–01:IV 93:III–01:IV 94:III–01:III 93:III–01:III 93:II–01:IV (31 obs.) (34 obs.) (34 obs.) (29 obs.) (33 obs.) (35 obs.)
−3.000∗∗ [−6.86] 2.006∗∗ [7.67] 0.512∗∗ [6.62]
LAT
−0.568∗∗ [2.51] 0.441∗∗ [7.07] 0.801∗∗ [12.52]
CZR
1.067∗∗ [3.24] 0.999∗∗ [5.32] 1.051∗∗ [2.35]
EST
−2.487∗∗ [−3.94] −0.058 [0.08] 2.650∗∗ [3.05]
HUN
0.481∗∗ [8.21] 0.897∗∗ [7.08] −0.048 [−0.40]
LAT
0.718∗ [1.76] 0.830∗∗ [2.24] 0.223 [0.53]
LIT
Note:: t -statistics in parentheses, ∗ (∗∗ ) denotes significance at 10% (5%) levels.
−0.630∗∗ −0.182∗∗ −0.073∗∗ −0.115∗ −0.146 −0.174∗ [−1.66] [−4.53] [−2.71] [−8.13] [−1.67] [−1.61] 0.046 D(LOG(REXP(−1))) −0.203 0.016 −0.131 0.060 0.359∗ [−1.12] [0.10] [−0.84] [0.42] [1.78] [0.34] ∗∗ ∗∗ ∗∗ ∗∗ 0.947 0.826 0.544 0.491∗ D(LOG(REUR(−1))) 0.197 0.518 [1.39] [2.57] [2.06] [2.03] [2.92] [1.76] 0.204 0.807∗∗ 0.917 0.162 1.014∗∗ D(LOG(GDP_EU(−1)/PPI_EU(−1))) 0.823 [1.16] [0.84] [3.02] [0.65] [2.16] [1.40] 0.317 0.613 0.006 D(LOG(WTR(−1))) 0.408 −0.091 1.108∗∗ [0.54] [−0.34] [2.77] [1.17] [1.45] [0.01] 0.901 0.849 0.764 0.871 0.853 0.659 Adj. R2 Sample 93:III–01:III 93:III–01:IV 93:III–01:IV 93:I–01:IV 93:I–01:IV 93:III–01:IV (33 obs.) (34 obs.) (34 obs.) (36 obs.) (36 obs.) (34 obs.)
Error Correction: CointEq1
LOG(WTR(−1))
LOG(GDP_EU(−1)/PPI_EU(−1))
1.560∗∗ [14.43] −0.336 [−1.14] 4.275∗∗ [7.41]
BUL
Estimation results for exports
Cointegrating Eq: LOG(REUR(−1))
D(LOG(REXP))
Table 4.2(c)
−0.143 [−1.61] −0.057 [−0.59] −0.209 [−0.75] −0.100 [−0.24] −0.018 [−0.04] 0.750 93:I–01:IV (36 obs.)
1.780 [1.62] 1.269∗∗ [3.07] 1.455∗∗ [4.24]
POL
−0.239 [−0.93] 0.871∗∗ [14.43] 0.507∗∗ [6.08]
SLO
0.014 [0.02] 1.254∗∗ [9.14] 0.491∗∗ [3.93]
SLV
−0.919∗∗ −0.579∗∗ −0.132 [−3.81] [−3.48] [−0.92] −0.005 −0.143 −0.104 [−0.03] [−0.85] [−0.56] −0.030 0.072 0.434 [−0.12] [0.22] [1.56] −0.740 −0.428 0.392∗ [−1.47] [−1.48] [1.65] −0.503 0.239 0.003 [−0.96] [0.78] [0.01] 0.613 0.732 0.634 93:I–01:III 93:III–01:III 93:I–01:IV (35 obs.) (33 obs.) (36 obs.)
0.698∗∗ [5.02] 0.996∗∗ [7.44] 1.120∗∗ [12.20]
ROM
−0.463∗∗ [−3.79] −0.116 [−0.85] −0.156 [−1.32] −0.336∗∗ [−2.78] −0.025 [−0.21] 0.831 94:III–01:III (29 obs.)
−0.191 [−1.32] 0.073 [0.34] −0.059 [−0.19] 0.208 [1.05] −0.002 [−0.02] 0.772 93:III–01:IV (34 obs.)
−0.208 [−0.47] 1.460∗∗ [7.54] 0.294∗∗ [4.67]
CZR
−0.220∗∗ [−8.48] 0.375∗∗ [2.72] 0.567∗ [1.85] −0.152 [−1.00] −0.113 [−1.47] 0.845 93:III–01:IV (34 obs.)
0.503∗∗ [2.40] 0.793∗∗ [10.61] 0.386∗∗ [2.73]
EST
−0.064 [−0.44] −0.283∗∗ [−0.04] −0.085 [−0.18] 0.376∗∗ [2.13] 0.080 [1.00] 0.848 93:III–01:IV (34 obs.)
−0.311 [−1.17] −0.015 [−0.12] 0.135∗∗ [3.18]
HUN
−0.228∗∗ [−5.92] 0.002 [0.02] −0.167 [−0.80] 0.112 [0.53] −0.229∗∗ [−2.53] 0.921 93:I–01:III (35 obs.)
−1.248∗∗ [−8.94] 0.608∗∗ [3.79] −0.734∗∗ [−5.11]
LAT
Note: t -statistics in parentheses, ∗ (∗∗ ) denotes significance at 10% (5%) levels.
Adj. R2 Sample
D(LOG(OIL(−1)))
D(LOG(RGDP(−1)))
D(LOG(REUR(−1)))
D(LOG(RIMP(−1)))
Error Correction: CointEq1
LOG(OIL(−1))
LOG(RGDP(−1))
−0.178∗ [−1.85] 0.849∗∗ [4.25] 0.522∗∗ [5.84]
BUL
Estimation results for imports
Cointegrating Eq: LOG(REUR(−1))
D(LOG(RIMP))
Table 4.2(d)
−0.076 [−0.74] 0.223 [1.30] 0.039 [0.16] −0.589∗∗ [−2.03] −0.343∗∗ [−2.13] 0.574 93:III–01:IV (34 obs.)
0.205 [1.12] 1.124∗∗ [9.76] 0.074 [0.58]
LIT
−0.034∗∗ [−7.18] −0.187∗ [−1.67] −0.562∗∗ [−2.63] 0.248∗∗ [2.86] 0.007 [0.10] 0.803 93:III–01:IV (34 obs.)
−0.226 [−0.11] 0.689∗∗ [2.00] 1.882 [2.34]
POL
−0.197∗∗ [−7.15] 0.184 [0.65] −0.156 [−0.58] −0.175∗∗ [−2.08] −0.083 [−0.65] 0.815 94:III–01:IV (29 obs.)
−0.173 [−0.72] 0.952∗∗ [4.54] 0.602∗∗ [3.26]
ROM
−0.472∗∗ [−8.18] −0.153 [−1.13] 1.194∗∗ [2.85] −0.465 [−1.15] −0.149∗ [−1.69] 0.800 93:III–01:III (33 obs.)
−0.863∗∗ [−3.98] 1.151∗∗ [15.03] 0.394∗∗ [4.16]
SLO
0.404∗∗ [−2.92] −0.184 [−1.02] −0.6.2∗ [1.75] −0.005 [−0.05] −0.0094 [−1.26] 0.818 93:I–01:IV (36 obs.)
−0.377∗∗ [−5.27] 1.127∗∗ [17.38] 0.205∗∗ [3.59]
SLV
−5.633 [−5.40] 1.933∗∗ [6.62]
CZR
−4.42∗∗ [−3.20] 1.281∗∗ [9.59]
EST
−0.354 [−1.01] 1.013∗∗ [79.02]
HUN
−0.665∗∗ [−5.19] 1.013∗∗ [7.79]
LAT
0.500∗∗ [2.48] 0.965∗∗ [69.04]
LIT
−0.211∗∗ [−0.38] 1.382∗∗ [12.00]
POL
1.824∗∗ [2.08] 0.504∗∗ [9.22]
ROM
−7.537∗∗ [−6.19] 1.139∗∗ [63.63]
SLO
−2.162 [−6.54] 0.925∗∗ [73.77]
SLV
Note: t -statistics in parentheses, ∗ (∗∗ ) denotes significance at 10% (5%) levels.
−0.138∗∗ −0.208∗∗ −0.095 −0.105∗∗ −0.105 0.064∗∗ −0.126∗∗ −0.160∗∗ −0.064 −0.217∗ [−0.55] [−1.86] [−3.20] [−7.01] [−1.20] [−2.32] [−0.80] [7.80] [−7.36] [−5.99] 0.621∗∗ 0.378∗∗ 0.331∗∗ −0.139 −0.318∗∗ −0.277∗ 0.095 D(LOG(M2(−1)/PPI(−1))) −0.11 0.151 0.216∗ [−1.07] [0.88] [1.91] [4.02] [2.62] [2.09] [−0.56] [−2.71] [−1.80] [0.62] 0.488 1.108∗∗ 0.413∗∗ 0.929∗∗ 0.008 −0.555 −0.149∗∗ 0.956∗∗ −0.251∗ D(SIN(−1)) −0.418∗∗ [−4.93] [0.26] [4.76] [2.06] [2.53] [0.05] [−1.08] [−3.60] [2.20] [−1.83] −0.285∗∗ −0.098 −0.149 −0.069 −0.092∗∗ −0.298 0.118 D(LOG(RGDP(−1))) −0.604 −0.024 −0.449∗∗ [−3.88] [−0.12] [−4.22] [−2.89] [−0.49] [−0.71] [−0.52] [−2.09] [−1.37] [1.15] 0.930 0.622 0.797 0.904 0.673 0.708 0.475 0.947 0.824 0.830 Adj. R2 Sample 94:III–01:III 93:III–01:IV 93:III–01:IV 93:III–01:IV 94:I–01:III 93:III–01:IV 93:III–01:IV 94:III–01:III 93:III–01:III 94:II–01:IV (25 obs.) (34 obs.) (34 obs.) (34 obs.) (31 obs.) (34 obs.) (34 obs.) (29 obs.) (33 obs.) (35 obs.)
Error correction: CointEq1
LOG(RGDP(−1))
−0.256∗∗ [−1.13] 0.930∗∗ [1.00]
BUL
Estimation results for money demand
Cointegrating Eq: SIN(−1)
D(LOG(M2/PPI))
Table 4.2(e)
0.480∗∗ [7.05] −1.108∗∗ [−5.95]
−0.098∗∗ [−2.50] −0.183∗ [−1.81]
−0.272∗∗ [−6.26] −0.647∗∗ [−3.42]
EST
0.329∗∗ [5.79] 0.201 [1.59]
HUN
−0.023 [−0.39] −0.038 [−0.24]
LAT
0.055∗∗ [3.61] 0.286∗∗ [8.25]
LIT
0.322∗∗ [4.08]] −0.771∗∗ [−5.26]
POL
0.346∗∗ [4.01] 1.293∗∗ [6.25]
ROM
0.157∗∗ [2.67] 1.12∗∗ [7.47]
SLO
Note: t -statistics in parentheses, ∗ (∗∗ ) denotes significance at 10% (5%) levels.
−0.092∗∗ −0.200∗∗ −0.152∗∗ −0.133∗∗ −0.327∗∗ −0.089∗∗ −0.074∗∗ −0.019∗∗ −0.157∗∗ [−2.25] [−3.95] [−2.92] [−6.40] [−2.59] [−3.48] [−3.36] [−9.32] [−2.94] 0.103 −0.038 0.062 0.197 −0.143 0.132 0.373∗∗ D(LOG(EMP(−1))) 0.053 0.298∗∗ [0.27] [2.77] [0.60] [−0.607] [0.71] [1.39] [−0.88] [1.24] [2.15] 0.005 0.007 0.043 −0.006 0.061 0.018 0.006 −0.072 D(LOG(RGDP(−1))) 0.091∗ [1.92] [0.26] [0.13] [1.26] [−0.28] [1.56] [0.49] [0.14] [−0.89] ∗∗ ∗ 0.020 0.085 0.022 0.051 0.018 0.043 −0.022 −0.334∗∗ D(LOG(RWAG(−1)/PRO(−1))) −0.094 [−2.00] [1.10] [1.84] [0.67] [1.36] [0.71] [1.08] [−0.56] [−2.95] 0.638 0.812 0.595 0.932 0.805 0.598 0.805 0.929 0.765 Adj. R2 Sample 94:III–00:III 93:III–01:IV 93:III-01:IV 93:III–01:IV 93:I–01:III 93:III–01:IV 93:III–01:IV 94:III–01:III 93:III–01:III (25 obs.) (34 obs.) (34 obs.) (34 obs.) (25 obs.) (34 obs.) (34 obs.) (29 obs.) (33 obs.)
Error correction: Coint Eq1:
LOG(RWAG(−1)PRO(−1))
Cointegrating Eq: LOG(RGDP(−1))
BUL
D(LOG(EMP))
CZR
Estimation results for employment
Table 4.2(f)
−0.031∗∗ [−2.61] 0.470∗∗ [4.46] −0.098∗∗ [−2.54] −0.099∗∗ [−2.53] 0.701 93:I–01:IV (36 obs.)
0.375∗∗ [8.39] 0.933∗∗ [37.54]
SLV
−0.409∗∗ [−4.78] 0.326 [3.08] 0.366∗∗ [4.82] 0.102 [0.68] −0.627∗∗ [−6.64] 0.969 94:IV–01:III (28 obs.)
−0.684∗∗ [−3.27] −0.518∗∗ [−4.08] 0.624 [1.44] −0.071∗ [−1.71] 0.063∗ [1.67] 0.996 93:IV–01:IV (33 obs.)
1.693∗∗ [3.87] −0.44∗∗ [6.31] 0.137 [1.55]
CZR
−0.523∗∗ [−3.38] −0.201∗ [−1.95] −0.580∗∗ [2.28] 0.069 [0.69] 0.289∗∗ [2.50] 0.975 93:IV–01:IV (33 obs.)
1.188∗∗ 8.59] −0.078 [−4.07] −1.146∗∗ [−890]
EST
−1.495∗∗ [−5.21] 0.335∗∗ [2.06] −0.124 [−0.458] −0.083 [−0.69] −0.118∗∗ [−1.80] 0.993 94:III–01:IV (30 obs.)
−0.017 [−0.13] 0.031∗∗ [2.47] 0.180∗∗ [2.82]
HUN
LAT
−0.374∗∗ [−2.26] 0.248∗∗ [1.65] 0.254 [1.04] −0.162∗∗ [−2.75] −0.020 [−0.11] 0.725 93:I–01:III (35 obs.)
1.251∗∗ [8.04] −0.022∗∗ [−5.19] 1.128∗∗ 8 [6.95]
Note: t -statistics in parentheses, ∗ (∗∗ ) denotes significance at 10% (5%) levels.
Adj. R2 Sample
D2 (log(pro(−1)))
D(LOG(UNE(−1)))
D2 (LOG(PPI(−1)))
D2 (LOG(WAG(−1)))
Error Correction: CointEq1
D(LOG(PRO(−1)))
LOG(UNE(−1)))
0.369∗∗ [4.21] −0.044 [−0.81] 2.103 [5.74]
BUL
Estimation results for wage inflation
Cointegrating Eq: D(LOG)(PPI(−1)))
D2 (LOG(WAG))
Table 4.2(g)
−0.926∗∗ [−4.03] −0.114 [−0.83] 0.312 [−0.92] −0.045 [−0.48] 0.119 [−1.15] 0.801 93:IV–01:IV (33 obs.)
0.854∗∗ [3.37] −0.053∗∗ [−2.94] 0.519∗∗ [3.70]
LIT
−0.101 [−0.54] −0.346 [−1.58] 2.175∗∗ [2.73] −0.247 [−0.95] −0.105 [−0.70] 0.516 93:IV–01:IV (33 obs.)
1.200 [5.50] −0.007 [−0.50] 1.398∗∗ [13.16]
POL
0.029 [0.59] −0.379∗ [1.75] −0.022 [−0.20] −0.243∗∗ [−2.43] 0.002 [0.03] 0.769 94:IV-01:IV (28 obs.)
0.940∗∗ [2.96] 0.009 [0.89] −1.410∗∗ [−7.34]
ROM
−0.601∗∗ [−6.85] −0.421∗ [−2.78] −0.181 [−1.56] −0.034 [−0.73] −0.227 [−3.85] 0.995 93:IV–01:III (32 obs.)
0.175 [0.55] 0.022∗∗ [8.73] 0.698∗∗ [5.35]
SLO
−0.277∗∗ [−2.10] −0.041 [−0.17] 0.419 [1.08] 0.400∗∗ [3.31] −0.088∗∗ [−2.07] 0.719 93:1–01:IV (36 obs.)
1.206∗∗ [8.53] −0.002∗∗ [−4.03] 0.539∗∗ [9.59]
SLV
0.139∗∗ [2.45] −0.409∗∗ [−2.42] 0.070∗ [1.98] 0.074∗∗
[2.18] 0.017∗∗ [2.02] 0.039∗ [1.85] 0.750 93:IV–01:IV (33 obs.)
−0.867∗∗ [−3.55] 0.280∗∗ [2.31] −0.049 [−0.18] −1.039∗∗
[−3.38] 0.156 [0.43] 0.279 [0.96] 0.627 94:III–01:III (29 obs.)
[−2.73] −0.128∗∗ [−6.04] −0.036∗ [−1.96]
[14.04] 0.010 [0.15] 0.013∗∗ [3.38]
[−0.11] −0.024 [−1.05] 0.001 [0.02] 0.661 93:IV–01:IV (33 obs.)
−0.010 [−0.20] −0.035 [−0.28] 0.086 [1.36] −0.024
[1.17] −0.047 [−061] 0.216 [4.24]
1.876∗∗ [6.59] 1.157
EST
[1.26] 0.040∗ [1.98] 0.117∗∗ [2.57] 0.593 93:III–01:IV (34 obs.)
−0.106∗ [−1.72] −0.127 [−0.89] −0.141 [−1.24] 0.097
[0.28] 0.131∗∗ [2.13] 0.036 [0.63]
2.613∗∗ [6.35] 0.066
HUN
[−1.76] −0.131∗ [−1.65] 0.131 [0.90] 0.631 93:I–01:III (35 obs.)
−0.422∗∗ [−2.52] −0.221 [−1.63] 0.165 [0.87] −0.419∗
[5.43] 0.307∗∗ [4.96] 0.098∗∗ [2.63]
0.793∗∗ [6.27] 1.014
LAT
Note: t -statistics in parentheses, ∗ (∗∗ ) denotes significance at 10% (5%) levels.
Adj. R2 Sample
(LOG(RGDP(−1)))
D2 (LOG(OIL(−1)))
D2 (LOG(EUR(−1)))+ LOG(PPI_EU(−1)))
D2 LOG(WAG(−1)))
D2 (LOG(PPI(−1)))
Error Correction: CointEq1
LOG(RGDP(−1))
D(LOG(OIL(−1)))
D(LOG(EUR(−1)+ LOG(PPI_EU(−1)))
0.556∗∗ [2.71] −0.315∗∗
CZR
−1.896∗∗ [−11.27] 2.483∗∗
BUL
Estimation results for inflation
Cointegrating Eq: D(LOG(WAG)(−1)))
D2 (LOG(PPI))
Table 4.2(h)
[−1.18] −0.027 [−0.77] 0.092 [2.46] 0.507 93:IV–01:IV (33 obs.)
−0.143 [−0.73] 0.106 [0.73] 0.100∗∗ [2.52] −0.051
[−1.55] 0.386∗∗ [7.87] 0.002∗ [1.67]
0.133∗∗ [2.10] −0.171
LIT
[−1.53] −0.011 [−0.68] −0.020 [−1.01] 0.658 93:III–01:IV (34 obs.)
−0.535∗∗ [−4.16] −0.117 [−0.90] 0.070∗∗ [2.07] −0.076
[4.68] 0.087∗∗ [4.39] −0.067∗∗ [−4.03]
−0.273∗∗ [−3.85] 0.487∗∗
POL
[3.26] −0.101∗ [−1.66] 0.058∗ [1.70] 0.890 94:III–01:III (29 obs.)
−0.742∗∗ [−6.51] 0.005 [0.05] −0.091 [−0.57] 0.263∗∗
[1.03] 0.274∗∗ [6.53] −0.003∗ [−1.85]
0.716∗∗ [5.80] 0.076
ROM
[1.02] 0.007 [0.49] 0.002 [0.03] 0.918 93:IV–01:III (32 obs.)
−0.037∗ [−1.95] −0.491∗∗ [−7.11] 0.190 [−1.51] 0.062
[0.57] 0.135 [1.53] 0.049 [10.39]
−4.586∗∗ [−599] −0.268
SLO
[0.46] −0.015 [−0.86] −0.050 [−1.98] 0.705 93:I–01:IV (36 obs.)
−0.018 [−0.23] −0.560∗∗ [−3.06] 0.077 [0.65] 0.051
[−4.86] 0.045∗ [1.98] 0.055∗∗ [3.64]
1.708∗∗ [12.60] −0.615∗∗
SLV
i
68
General Policy Issues for the Accession Countries
so that it restricts the long-run behaviour of the endogenous variables to converge to their co-integrating relationship while considering at the same time the short-run adjustment dynamics towards the long-run equilibrium. The co-integration term is known as the error correction term since the deviation from long-run equilibrium is corrected gradually through a series of partial short-run adjustments. A VECM of a vector of endogenous variables y and exogenous variables x and of lag length 1 is written as yt = α0 + α1 yt−1 +
K k=1
βk xk,t−1 +
K
δk xt−1 + α2 yt−1 + et .
(4.1)
k=1
This VECM(1,1) in most cases is appropriate and yields satisfying estimation results. In the VECM, we can derive immediately the important short-run and long-run elasticities of the endogenous variables. These elasticities are our special interest and displayed in Tables 4.2(a)– (h): the long-run elasticities are grouped in the first part of the tables, which displays the co-integrating equation; the short-run elasticities are found in the second part, together with the error correction term. Constants and time trends, which were also included in the estimation of the VECM, have not been reported in the tables for space considerations. Given the presence of seasonal patterns in most variables, seasonal dummies (also not reported) are included into the structural relations (4.1)–(4.8). The estimated residuals may be seen as random shocks to the endogenous variables, representing factors that cannot be explained by the model. In practically all cases the consumption function – found in Table 4.2(a) – is relatively well estimated: in the long-run co-integrating relation there is in most cases a negative effect from a higher real interest rate and a positive effect from higher real disposable income. In the short-run adjustment dynamics of the error correction part of the VECM these effects are also generally present, although the evidence is somewhat less strong. A strong and significant error correction term suggests a relatively quick adjustment of the adjustment dynamics to the long-run equilibrium. While investment is more volatile than consumption, the simple investment model performs still relatively well according to Table 4.2( b): we find evidence of negative real interest rate effects in some cases and positive effects from real GDP increases, especially in the long-run equilibrium. Also FDI inflows appear to have a significant effect on
Macroeconomic Adjustment in Accession Countries
69
domestic investment: in most cases FDI increases investment; in the case of Estonia and Hungary a negative long-run effect is found. Generally speaking, there are several reasons why the real interest rate channel of monetary policy on consumption and investment may not have worked strongly in Accession Countries until recently: (a) the presence of relatively underdeveloped financial markets, ( b) the negative real interest rates during some part of the sample must be considered as an anomaly. The estimations of the export and import are functions found in Tables 4.2(c) and 4.2(d). We have emphasized in the model integration with the current EU: in both exports and imports the real exchange rate vis-à-vis the EU has been included as a proxy of competitiveness of the export sector and the import-competing sector, respectively. In the case of exports, real GDP of the EU is also expected to exercise an effect. Through these channels the importance of CEE trade integration with the EU is included in the picture. In addition, world trade and oil prices are expected to have some effect, the first on exports, the second on imports. The estimation results confirm that these variables are in most cases important determinants of the trade flows of these countries: the real GDP of the EU and world trade have in practically all cases a positive effect on exports. Also the real exchange rate versus the EU has often the expected effects: a real depreciation tends to increase exports and to reduce imports, and these effects are moreover present in both the short and the long run. Taken together, the model results seem to explain trade flows well. The model contains a standard LM curve that characterizes money market equilibrium. Table 4.2(e) provides the estimation results. The theoretical priors of a negative interest rate elasticity and a positive output elasticity of money demand are observed in practically all cases. Altogether, the estimation results seem to hint at fairly stable money demand functions. Table 4.2(f) estimates the employment function. Real output and unit labour costs were identified by the theoretical model as the factor explaining employment. In the majority of cases the effects of these factors are confirmed but in some cases the estimation results do not fully support the theoretical relation and no clear link with output or wage costs seems to be present. The estimated employment functions of the Czech Republic and Estonia appear to suffer most from such inadequacies. The estimations of the wage adjustments are found in Table 4.2(g). A relatively consistent picture results, in particularly in the long-run
70
General Policy Issues for the Accession Countries
equilibrium. Wage indexation (that is, backward-looking inflation expectations) explains the strong positive effect of prices on wages, suggesting the risks of wage–price spiralling. In most cases a wage-moderating effect from unemployment on wages is found, providing evidence of the existence of a Phillips-curve mechanism in wage adjustment. Productivity tends to induce higher wages, except for Estonia and Romania. As noted in the previous section, a link between wages and productivity may serve as a proxy for the workings of the Balassa–Samuelson effect. The final structural estimation concerns the inflation dynamics; results are displayed in Table 4.2(h). The long-run effects in particular are quite clearly estimated: while there are a number of exceptions, in most cases higher wages, foreign prices, oil prices and domestic real output all tend to increase inflation. The first three factors have earlier been singled out as cost-push factors and the final factor as a demand-pull factor of inflation. The estimation results tend to support the hypothesis that inflation in particular was cost-push driven: wage growth, passthrough of foreign prices and exchange rates (here EU prices and the exchange rate vis-à-vis the euro) and also oil prices are mostly positive and significant.
4
Model simulation
The estimated structural relations yield – together with the set of definitions in the model – a small but concise macroeconomic model that provides an account of the goods, labour and money market of the Accession Countries and some of the main relations with the EU. In most cases the structural relations could be estimated with some degree of acceptability and accuracy. Model simulation is now needed to assess the tracking ability of the estimated models. Therefore, we briefly present the results of our simulations in this section. The models are simulated for the period 1998 : 1–2001 : 4. In the case of Bulgaria and Romania, the dynamic simulations start at 1999:I due to lack of stability of the model if started in 1998, due to the very high real and nominal volatility these countries faced during the period 1996–98. Dynamic simulations are an appropriate (and demanding) way in to assess the forecasting ability of models. A dynamic simulation implies that the simulation model is provided with the adjustment path of the exogenous variables, plus the initial value of the endogenous variables in the model. It answers the question of whether the
Macroeconomic Adjustment in Accession Countries
71
model – given the adjustment of the exogenous variables – would predict adjustment dynamics comparable with those that have been actually observed. In its standard form, the model contains 19 exogenous variables (FDI, EUR, WTR, OIL, LAB, PPI EU , GDP EU , RGDP EU , REV, GEX, GCO, CIN, MMP, CBC, CLG, CGD, OCF, CAA, DEF) and 25 endogenous variables (CON, PPI, RCON, SIN, RSIN, YDP, RYDP, INV, RINV, GDP, RGDP, EXP, REXP, IMP, RIMP, REUR, M1, EMP, WAG, RWAG, PRO, UNE, M0, RES, CUA). In the standard form, the exchange rate and interest rates are assumed to be set according to a predetermined path, leaving money balances and foreign reserves to adjust to any ex ante disequilibrium in money and financial markets. This assumption, close to the managed exchange rate approach by Bofinger and Wollmershäuser (2001), can however, be changed: it can also be assumed that a fixed exchange rate, a monetary targeting policy or currency board is adopted. It should be noted that the outcomes of in-sample simulations with the model are not critically dependent on the assumptions about the monetary regime; it merely concerns the assumptions on which monetary variables are predetermined and which are fully endogenous. With out-of-sample forecasting exercises the assumptions about the monetary policy regime are of course more crucial than in the in-sample dynamic simulations carried out below. Bulgaria: The model for Bulgariahas been estimated over a sample that also includes the high nominal and real volatility in the period 1994–97, a period that has clearly been extraordinary. Notwithstanding these difficulties, the dynamic simulation for the period 1999–2001 is fairly acceptable in most cases, when compared to actual data. The model displays increased variability in real and nominal variables, but generally indicates the same directions as the observed adjustments. Most difficulties are found in predicting prices and employment growth: during 2000 the model underestimates actual inflation and for 2001 overestimates it. Czech Republic: In the case of the Czech Republic, the model has some difficulties in reproducing the observed real GDP growth: first it underestimates growth and then overestimates it, basically because of the predicted adjustment of real consumption. The predictions for net exports, employment, money balances and inflation on the other hand are surprisingly accurate. Estonia: The Estonian model overpredicts the volatity in real GDP growth during 1999 and the volatility in base money and reserves. The high volatility of investment, exports and imports is reproduced by
72
General Policy Issues for the Accession Countries
the model. The predictions by the model of the wage–price nexus are particularly accurate: with Lithuania, the Estonian scores best on this point. Hungary: For Hungary, the estimated model follows the trends of most real and nominal variables, but with a certain lag and some imprecision. Inflation of wages and prices is consistently overestimated. Also the estimated money growth and reserves are too high. The predicted real depreciation is some 10 per cent too high. Latvia: On balance, the Latvian model tracks the observed patterns most of the time, but has serious difficulties with consumption, exports, base money and reserves. Its overprediction of exports also implies an overestimate of foreign reserves and base money growth. The predicted drop in wage growth during 1999 was also not observed in reality. Lithuania: Of all cases studied, the model estimated for Lithuania seems to perform best, as the differences between model predictions and actual values are the lowest here. There appears to be no systematic bias in the prediction errors, and the model appears to be able to deliver a good representation of the Lithuanian economy. Poland: For Poland, the estimated model performs quite well: only in the case of investment, exports, wages and reserves does it produce significant divergences from the observed adjustments. Romania: It is perhaps not surprising that the estimated model of Romania experiences difficulties in replicating the fairly volatile adjustment dynamics in this case. In particular, the predicted dynamics of exports and imports, inflation and interest rates are wrong. Slovakia: Except for overpredicting real GDP and consumption growth (probably because of the underestimation of actual inflation), the model performs quite well in the case of Slovakia. Slovenia: Overestimation of output and consumption growth also occurs in the model for Slovenia. The model also ignores the favourable adjustments of (un)employment during the period 1998–2001. On balance, the model is nevertheless fairly acceptable. All in all, in most cases the simulations of the model provide a fairly good account of the macroeconomic adjustment dynamics (Figures 4.1(a)–(e) present a graphical description of this for selected countries). It is probably the use of the VECM, with its well-defined distinction between long-run equilibrium and short-run adjustment dynamics that makes the model work quite well, and also makes it track the actual data – in most cases – with some degree of accuracy.
Macroeconomic Adjustment in Accession Countries RGDP_CZR (year % ch.) 25
RCON_CZR (year % ch.)
20
30
15 10 5
15 10 5
20 10 0
0 –5
0 93 94 95 96 97 98 99 00 01 Actual Baseline
–10 93 94 95 96 97 98 99 00 01 Actual Baseline
REXP_CZR (year % ch.)
10 5 0 –5
PPI_CZR (year % ch.) 8 7 6 5 4 3 2 1 0
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
M0_CZR (year % ch.)
EMP_CZR (year % ch.)
70 60 50 40 30 20 10 0 –10
93 94 95 96 97 98 99 00 01 Actual Baseline
RIMP_CZR (year % ch.) 25 20 15 10 5 0 –5
25 20 15
RINV_CZR (year % ch.) 40
25 20
93 94 95 96 97 98 99 00 01 Actual Baseline WAG_CZR (year % ch.) 20
1
16
0
12
–1
8 4
–2
CUA_CZR (% GDP) 4 0 –4 –8
0 93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
DEF_CZR (% GDP) 20000
0 –2 –4
12000
–6
4000
16000
8000
0
M1_CZR (bin nc)
UNR_CZR (% LAB) 9
34
6
32
5
200
4
30
EUR_CZR
SIN_CZR (%) 14 13 12 11 10 9 8 7
35 34 93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
36
REUR_CZR 36
8 7
300
37
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
400
RES_CZR (min eur)
4 2
–8
500
93 94 95 96 97 98 99 00 01 Actual Baseline
FIN_CZR (min eur) 2100 1500 1000 500 0
93 94 95 96 97 98 99 00 01 Actual Baseline
–500
93 94 95 96 97 98 99 00 01 Actual Baseline
Figure 4.1(a) Czech Republic: actual and simulated adjustment
73
74
General Policy Issues for the Accession Countries RGDP_EST (year % ch.)
RCON_EST (year % ch.)
RINV_EST (year % ch.)
25
20
20
16
20
15
12
10
10 5 0
8
0
4
–10
0 93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
REXP_EST (year % ch.) 40
RIMP_EST (year % ch.) 60 50 40 30 20 10 0 –10 –20
30 20 10 0 –10
PPI_EST (year % ch.) 30 20 10 0
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline M0_EST (year % ch.)
93 94 95 96 97 98 99 00 01 Actual Baseline
EMP_EST (year % ch.) 2
50
93 94 95 96 97 98 99 00 01 Actual Baseline
40
6
30
–2
20
–4
10
–6
0
–8
–10
–10 93 94 95 96 97 98 99 00 01 Actual Baseline
WAG_EST (year % ch.) 40 30 20 10 0
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
DEF_EST (% GDP)
RES_EST (min eur)
CUA_EST (% GDP) 5
4
0
2
1000
0
800
–2
600
–5 –10
–4
–15
–6 93 94 95 96 97 98 99 00 01 Actual Baseline
–8
M1_EST (min nc) 24000
400 200 93 94 95 96 97 98 99 00 01 Actual Baseline UNR_EST (% LAB)
16000 12000 8000 4000
EUR_EST
SIN_EST (%)
15.8
14
15.6
12
100
8 6
15.0
FIN_EST (min eur) 200
10
15.2
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
15.4
REUR_EST 26 24 22 20 18 16 14 12
14 13 12 11 10 9 8 7 6
20000
93 94 95 96 97 98 99 00 01 Actual Baseline
0
4 93 94 95 96 97 98 99 00 01 Actual Baseline
Figure 4.1(b)
93 94 95 96 97 98 99 00 01 Actual Baseline
Estonia: actual and simulated adjustment
93 94 95 96 97 98 99 00 01 Actual Baseline
Macroeconomic Adjustment in Accession Countries RGDP_HUN (year % ch.) 20 10
RCON_HUN (year % ch.)
RINV_HUN (year % ch.)
20
30
10
20
0
0
–10
–10
–20
10 0 –10
–30 93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
REXP_HUN (year % ch.)
93 94 95 96 97 98 99 00 01 Actual Baseline
RIMP_HUN (year % ch.)
PPI_HUN (year % ch.)
30
40
30
20
30
25
10
20
0
10
–10
0
–20
20 15 10 5
–10 93 94 95 96 97 98 99 00 01 Actual Baseline
0 93 94 95 96 97 98 99 00 01 Actual Baseline
M0_HUN (year % ch.)
93 94 95 96 97 98 99 00 01 Actual Baseline
EMP_HUN (year % ch.)
WAG_HUN (year % ch.)
8
30
24
20
4
10
0
20
0
–4
18 16
–10
22
–8 93 94 95 96 97 98 99 00 01 Actual Baseline
14 93 94 95 96 97 98 99 00 01 Actual Baseline
CUA_HUN (% GDP)
DEF_HUN (% GDP) 0
5
–8
–10
–12
–15 93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
M1_HUN (bln nc)
2500 2000
150 140 130
10 9
1500 1000
120 110
8 93 94 95 96 97 98 99 00 01 Actual Baseline EUR_HUN
REUR_HUN 160
13 12 11
3000
93 94 95 96 97 98 99 00 01 Actual Baseline
UNR_HUN (% LAB) 14
3500
RES_HUN (min eur) 18000 16000 14000 12000 10000 8000 6000
–4
0 –5
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
SIN_HUN (%)
FIN_HUN (min eur)
32
240
28
200
2000
24 1000
20
160
16
120
0
12 93 94 95 96 97 98 99 00 01 Actual Baseline
Figure 4.1(c)
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
Hungary: actual and simulated adjustment
75
76
General Policy Issues for the Accession Countries RGDP_POL (year % ch.)
RCON_POL (year % ch.)
30 25 20 15 10 5 0
RINV_POL (year % ch.)
20
80
15
60
10
40
5
20 0
0
–20 93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline REXP_POL (year % ch.)
93 94 95 96 97 98 99 00 01 Actual Baseline
RIMP_POL (year % ch.)
PP1_POL (year % ch.)
30 30 20
30 25 20 15 10 5 0
20
10
10
0
0
–10
–10 93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
M0_POL (year % ch.)
93 94 95 96 97 98 99 00 01 Actual Baseline
EMP_POL (year % ch.)
WAG_POL (year % ch.) 60 50 40 30 20 10 0 –10
2 60
0
40 –2
20
–4
0 –20
–6 93 94 95 96 97 98 99 00 01 Actual Baseline CUA_POL (% GDP)
DEF_POL (% GDP) 0 –1 –2 –3 –4 –5 –6 –7 –8
8 4 0 –4 –8 93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
RES_POL (min eur) 30000 25000 20000 15000 10000 5000
93 94 95 96 97 98 99 00 01 Actual Baseline
M1_POL (min nc)
UNR_POL (% LAB) 18 17 16 15 14 13 12 11 10
100000 80000 60000 40000 20000
EUR_POL
3.0 2.8 2.6 2.4
3.0 2.5 2.0 93 94 95 96 97 98 99 00 01 Actual Baseline
Figure 4.1(d)
93 94 95 96 97 98 99 00 01 Actual Baseline
SIN_POL (%)
FIN_POL (min eur) 4000
26 24 22 20 18 16 14
3.5
REUR_POL 3.2
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
4.0
93 94 95 96 97 98 99 00 01 Actual Baseline
3000 2000 1000 0 –1000 93 94 95 96 97 98 99 00 01 Actual Baseline
Poland: actual and simulated adjustment
93 94 95 96 97 98 99 00 01 Actual Baseline
Macroeconomic Adjustment in Accession Countries RGDP_SLV (year % ch.)
RCON_SLV (year % ch.) 30
20 15 10 5
40 30 20
20
10 0 –10 –20
10
0 –5
0
REXP_SLV (year % ch.)
RIMP_SLV (year % ch.)
10
10 5 0
0 –10
M0_SLV (year % ch.)
EMP_SLV (year % ch.)
40 20 0
WAG_SLV (year % ch.) 35 30 25 20 15 10 5 0
3 2 1 0 –1 –2 –3 –4
80 60
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
100
PPI_SLV (year % ch.) 35 30 25 20 15 10 5 0
30 25 20 15
20
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
30
RINV_SLV (year % ch.)
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline CUA_SLV (% GDP)
93 94 95 96 97 98 99 00 01 Actual Baseline
DEF_SLV (% GDP)
RES_SLV (min eur) 6000
0
–6 –8
0 –2 –4 –6 –8
–10
–10
–2 –4
5000 4000 3000 2000 1000 93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline M1_SLV (bln nc)
UNR_SLV (% LAB)
REUR_SLV 168
17
400
164
16
160
300
15
156
200
14
152
100
13
148 93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline EUR_SLV
FIN_SLV (min eur)
SIN_SLV (% ) 800
200
600
40
180
400
30
160
200
20
0
140 10 93 94 95 96 97 98 99 00 01 Actual Baseline
Figure 4.1(e)
–200 93 94 95 96 97 98 99 00 01 Actual Baseline
93 94 95 96 97 98 99 00 01 Actual Baseline
Slovenia: actual and simulated adjustment
77
78
General Policy Issues for the Accession Countries
Notes 1. A similar modelling of the wage–price nexus (4.7)–(4.8) is found e.g. in Welfe (2002), who estimates both equations for the case of Poland.
Appendix: data sources and definitions The variables shown in Table 4A.1 have been used in the analysis. Table A.1
Variables and data sources
Variable Name
Units
CON RCON PPI SIN YDP
Private consumption Real consumption Producer price index Money market interest rate Disposable income
RYDP INV
Real disposable income Gross fixed capital formation Gross domestic product
IMF IFS line 96F . . . ZF and nat.stat.off. calculated as RCON≡CON/PPI IMF IFS line 63 . . . ZF IMF IFS line 60B . . . ZF calculated as YDP ≡ GDP − REV + GEX − GCO mln/bln n.c., quarterly calculated as RYDP ≡ YDP/PPI mln/bln n.c., quarterly IMF IFS line 93E . . . ZF and nat.stat.off.
GDP RGDP EXP REXP EUR REUR WTR IMP RIMP OIL M2 EMP WAG UNE REV GEX GCO CIN RES CUA CAA FDI OCF MMP M0 CBC CLG CGD LAB
Source
mln/bln n.c., quarterly mln/bln n.c., quarterly 1995=100 % mln/bln n.c., quarterly
mln/bln n.c., quarterly calculated as GDP ≡ CON + INV + EXP − IMP + CIN + GCO Real gross domestic product mln/bln n.c., quarterly calculated as RGDP ≡ GDP/PPI Exports of goods and mln/bln n.c., quarterly IMF IFS line 90C . . . ZF and nat.stat.off. services Real exports mln/bln n.c., quarterly calculated as REXP ≡ EXP/PPI Exchange rate vs euro per.avg calculated from IMF IFS line . . . RF.ZF Real exchange rate vs euro per.avg calculated as REUR ≡ EUR∗ PPI E U /PPI World trade bln US$ calculated from IMF IFS Imports of goods and mln/bln n.c., quarterly IMF IFS line 98C . . . ZF and nat.stat.off. services Real imports mln/bln n.c., quarterly calculated as RIMP ≡ IMP/PPI Oil price $ per barrel IMF IFS line Money, M2 mln/bln n.c IMF IFS line 35 . . . ZF and nat.stat.off. Employment 1000 IMF IFS line 67E . . . ZF and nat.stat.off. Wages 1995 = 100 IMF IFS line 65 . . . ZF and nat.stat.off. Unemployment 1000 IMF IFS line 67C . . . ZF and nat.stat.off. Government revenue mln/bln n.c., quarterly IMF IFS line 81 . . . ZF and nat.stat.off. Government expenditure mln/bln n.c., quarterly IMF IFS line 82 . . . ZF and nat.stat.off. Government consumption mln/bln n.c., quarterly IMF IFS line 91F . . . ZF and nat.stat.off. Change in inventories mln/bln n.c., quarterly IMF IFS line 93L . . . ZF and nat.stat.off. Foreign exchange reserves mln Euro calculated from IMF IFS line .1L.DZF Trade balance mln Euro, quarterly calculated as CUA ≡ (EXP − IMP)/EUR Capital account mln Euro, quarterly calculated from IMF IFS line 78BJDZF Foreign direct investment mln Euro, quarterly calculated from IMF IFS line 78BEDZF Other capital flows mln Euro, quarterly calculated as OCF ≡ CAA − FDI Money multiplier calculated as MMP ≡ M2/M0 Base money, M0 mln/bln n.c., quarterly calculated as M0 ≡ CBC + CLG − CGD + RES mln/bln n.c., quarterly IMF IFS line 12E . . . ZF Central Bank credit to banks Central Bank lending to mln/bln n.c., quarterly IMF IFS line 12A . . . ZF govt. Deposits govt. at CB mln/bln n.c., quarterly IMF IFS line 16D . . . ZF Labour force 1000 calculated as LAB ≡ EMP + UNE
5 Nominal and Real Forex Regimes and EMU Accession Pieter van Foreest and Casper de Vries
1
Introduction
Is the choice of the foreign exchange (forex) regime important for real growth? The objective of this chapter is to investigate this issue geared towards the inception of the EMU and the transition of the CEECs during the 1990s. We find that the choice of the nominal forex regime is not of first-order importance for achieving high and stable real growth. The empirical evidence is that the output growth was unrelated to the amount of nominal forex variability, and the type of currency arrangement in place. The fact that real growth is insulated from the nominal forex variability and the nominal forex system is analogous to the wellestablished relative insensitivity of the trade account to nominal forex uncertainty; see for instance Bacchetta and van Wincoop (2000). Within the monetary theory of exchange rates, the nominal forex regime irrelevance result can be seen as an implication of the forwardlooking nature of financial markets. Since the nominal spot forex rate is determined by the discounted sum of all future expected fundamentals, the stability of the nominal forex rate hinges on the coherence of current and anticipated monetary and fiscal policies. Seen in this way, a flexible nominal forex system can be very stable, if the monetary and fiscal policies are coherent with the market’s nominal forex rate valuation. Conversely, a managed float or currency peg can be quite unstable. Some recent empirical studies provide weak evidence that the choice of nominal forex regime matters for the behaviour of macroeconomic fundamentals, including Edwards (1996, 1998), Ghosh et al. (1997, 2002), Reinhart and Rogoff (2002) and Vinhas de Souza (2002). The empirical results are mixed, however, in the sense that there is no agreement on the regime-specific effects. In our opinion, the weakness and inconsistency 79
80
General Policy Issues for the Accession Countries
of the existing empirical evidence does support the nominal forex regime irrelevancy result. This result is important for the countries aiming to join the EMU given that it is a requirement to keep the nominal forex rate within a narrow band versus the euro for at least two years before entry. Prudent finetuning of policies is a prerequisite for such a currency system. A second important EMU entry requirement is a euro inflation target; that is, the domestic inflation rate must stay close to that of the Euroarea as a whole, with ‘close’ being defined as within 1.5 percentage points. In other words, the potential entrant is in effect required to ‘peg’ both its nominal and its real euro exchange rate to the Euroarea. In view of our analysis, the question arises whether or not the adoption of a real forex peg matters for output growth. We investigate this issue empirically and find that the choice of a Euroarea inflation target (i.e. real euro peg) involves some important policy trade-offs. A premature attempt to peg the real forex to the Euroarea would be undesirable, as it would require a contraction of output to countervail the real forex rate appreciations caused by the BS effect. To avoid a situation of delayed Euroarea entry, the current EMU members must be prepared to accept some temporary flexibility of the real forex rates of the ACs. The rest of this chapter is organized as follows. In Section 2, we first investigate what economic theory has to say on the importance of the choice of the forex regime for real growth. Subsequently, we turn to the empirical analysis of the issue in Sections 3 and 4. Section 5 concludes.
2
Monetary view on forex rate determination
In this section we investigate what economic theory has to say aboutthe importance of the choice of the forex regime for the behaviour of the macroeconomic fundamentals, and in particular real growth. We present the monetary model of the forex and discuss how it characterizes different monetary policy regimes. We obtain the forex regime irrelevancy result from the forward solution of the monetary model of the forex rate.
2.1 The monetary forex rate model As is well known, the monetary model of the forex rate consists of two building blocks: the quantity equation and the purchasing power parity
Forex Regimes and EMU Accession
81
(PPP) supposition. Country i’s quantity equation at time t in logarithmic format reads mi,t − pi,t = τ yi,t − λri,t ,
(5.1)
where mi,t is log money demand, pi,t is the log price level, yi,t is the log output, and ri,t is the nominal interest rate. Plausible parameter restrictions are a positive income elasticity τ > 0, and a negative interest semi-elasticity λ > 0. When money demand and supply are balanced, equation (5.1) describes money market equilibrium. Then domestic prices are determined by pi,t = βmi,t + λri,t + τ yi,t + εi,t ,
(5.2)
where we generalized the model by allowing for additive noise εi,t and by introducing the money demand elasticity β = ∂pi,t /∂mi,t . The quintessence of the monetarist theory is the money neutrality hypothesis β = 1. The coefficients of the quantity equation (5.2) are restricted to be identical across countries. The theoretical reason for this restriction is that the monetary quantity equation is structural and not countryspecific. Another reason is that countries do not often change their monetary regimes, so that per country, time-series estimates usually fail to deliver meaningful results. But across countries, monetary regimes do differ substantially, and hence yield valuable information. A panel with cross-country coefficient restrictions can exploit this variation to obtain meaningful parameter estimates. Under the common coefficients assumption, the relative quantity equation of a country vis-à-vis the benchmark country is ˜ i,t + λ˜ri,t + τ y˜ i,t + ε˜ i,t , p˜ i,t = β m
(5.3a)
where x˜ i,t ≡ xi,t − xbenchmark,t denotes deviations of country i versus the benchmark country. The model predicts that the international price differential or real forex rate p˜ i,t is determined by three fundamental eco˜ i,t , the interest nomic factors, respectively the relative money supply m ˜ ˜ rate differential ri,t , and the output differential yi,t . The international quantity equation (5.3a) is the first building block of the monetary forex model. The second building block is the PPP supposition. In absolute form, the PPP supposition is that goods sell for the same price in two countries. Formally, let si,t denote the log
82
General Policy Issues for the Accession Countries
nominal exchange rate (units of local currency per unit foreign currency), then absolute PPP holds if si,t = p˜ i,t . In our empirical study, we allow for (persistent) deviations from absolute PPP by postulating that s˜ i,t = p˜ i,t + η˜ i,t ,
(5.3b)
where η˜ i,t is the deviation from absolute PPP. Substitution of equation (5.3b) into (5.3a) and rearranging renders the monetary model of the (flexible) nominal forex rate ∗ ˜ i,t + λ˜ri,t − τ y˜ i,t + ε˜ i,t si,t = β m ,
(5.4)
∗ = ε + η . The model predicts that the nominal forex s where ε˜ i,t i,t i,t i,t equals the real forex rate p˜ i,t , which in its turn depends on the three fun˜ i,t , r˜i,t , and y˜ i,t . The (composite) residual damental economic factors, m ∗ captures omitted variables like transportation costs, etc. ε˜ i,t
2.2 The monetary regime characteristics The monetary forex model, equation (5.4), describes the stance of monetary policy of a particular country i vis-à-vis the benchmark country at a particular time t. In order to characterize empirically the different monetary policy regimes and the trade-offs represented by the different regimes, we assume that each country in the sample has essentially operated one particular regime. Under the assumption of stable monetary regimes, the salient features of a monetary regime are captured by the mean vector and covariance matrix of the growth rates of the three fundamental macroeconomic factors. The cross-section provides the overview of the alternative regimes. To ˜ i,t + demonstrate this concisely, we define the ‘financial’ variable fi,t ≡ β m ˜ λ˜ri,t and the ‘real’ variable gi,t ≡ τ yi,t , so that, in first differences, the monetary forex rate model (5.4) reads ∗ si,t = fi,t − gi,t + ˜εi,t ,
(5.5)
where xi,t ≡ xi,t − xi,t−1 . In addition, we introduce shorthand for the unconditional moments, the expected value μx,i ≡ E{xi,t }, the variance 2 ≡ var{x } and the covariance σ σx,i i,t xu,i ≡ cov{xi,t , ui,t }. Model (5.5) imposes the following structure on the unconditional moments of the
Forex Regimes and EMU Accession
83
forex changes: μs,i = μf ,i − μg,i + Ei
(5.6a)
2 2 σs,i = σf2,i + σg,i − 2σfg,i + Ui
(5.6b)
Since the monetary forex rate model (5.5) contains an unobserved ∗ , the moment decompositions (5.6) contain the unidenresidual ˜εi,t tified components Ei and Ui . In the empirical analysis, Ei , and Ui are just residuals. The variance decomposition (5.6b) highlights that the magnitude of 2 the forex volatility σs,i may or may not depend on the levels of one of its components, such as the financial volatility component σf2,i or 2 2 the real volatility component σg,i . For instance, it may happen that σs,i 2 are high (low). One extreme itself is low (high), while both σf2,i and σg,i possibility is that the financial and real variable move in lockstep (i.e., 2 − 2σ 2 = 0) and the omitted fi,t = gi,t for all t, so that σf2,i + σg,i fg,i ∗ factors are constant (i.e., ˜εi,t = 0 for all t, so that Ei = Ui = 0). It then 2 = 0. This particular currency follows that si,t = 0 for all t, such that σs,i system can be classified as a ‘hard float’ or ‘fear of floating’ regime, in which the de facto regime is ‘fixed’, while the de jure regime is ‘flexible’; see Calvo and Reinhart (2000). In fact, the fear of floating is exactly what the Stability and Growth Pact (SGP) is all about. The SGP should ensure that each member country i follows a fiscal policy that yields a real growth gi,t , which is consistent with the single monetary policy yielding financial growth fi,t .
2.3 Asset view on nominal forex determination We investigate the relation between the domestic monetary regime specifics and the forex regime characteristics by solving the monetary forex model forward. Suppose that the uncovered interest parity (UIP) holds r˜i,t = Et {si,t+1 − si,t },
(5.7)
where Et {si,t+1 } is country i’s time t expected forex rate for time t + 1. For ∗ = 0 for all i and t, and for notational consimplicity we assume that ˜εi,t ˜ i,t − τ y˜ i,t . venience we define the forex rate’s fundamental value zi,t ≡ β m
84
General Policy Issues for the Accession Countries
The monetary forex model (5.4) simplifies to
si,t =
1 λ zi,t + Et {si,t+1 }. 1+λ 1−λ
When this equation is solved forward, one obtains1 si,t =
∞ 1 Et {zi,t+j } . 1+λ (1 + 1/λ)j
(5.8)
j=0
The forward solution (5.8) states that the current forex rate si,t is proportional to the expected discounted value of the fundamental process {zi,t+j }. This forward-looking solution of the monetary forex model forms the basis of the asset view on the forex rate determination: all the available information about future changes in the fundamental process is directly and completely incorporated in the current rate, si,t . To provide a simple demonstration of the asset view on the forex rate determination, suppose that the composite fundamental follows a driftless random walk zi,t = zi,t−1 + ξi,t ,
(5.9)
where ξi,t is white noise with variance σξ2,i > 0. Under this randomwalk assumption, all the information about the future fundamental values is fully incorporated in the current fundamental value, that is Et {zi,t+j } = zi,t , for all j; with moment restrictions Et {zi,t+j } = 0 for all j, and var{zi,t } = σξ2,i . By (5.8), the solution is si,t = zi,t . The random-walk assumption has the counterfactual implication that the interest differential is always zero. When the driftless random-walk model (5.9) is substituted into the UIP condition (5.7), one finds that r˜i,t = Et {si,t+1 } = 0 for all t.2 Nevertheless, it is also part of the folk wisdom on exchange rate economics that the forex rates approximately follow driftless random walks. Anyway, the result below does not depend critically on the driftless random-walk assumption. PROPOSITION: Under the assumption that the composite fundamental (zi,t ) follows the random walk (5.9), the fundamental shock (ξi,t ) is transferred one for one to the nominal forex rate (si,t ), since by (5.8) the solution is si,t = zi,t . For this reason the nominal forex rate volatility equals the composite
Forex Regimes and EMU Accession
85
fundamental volatility
var{si,t } = var{zi,t } = σξ2,i . 2 + τ 2σ 2 − ˜ i,t − τ y˜ i,t , we have var{zi,t } = β 2 σm,i But, since zi,t ≡ β m y,i βτ σmy,i , so that as before we conclude that growth and fluctuations in the financial and real sector are to a first order unrelated to the adoption of a particular forex regime. Due to the forward-looking nature of the forex market, forex rate stability hinges on the coherence of the current and anticipated behaviour of the fiscal authorities (partially) controlling the output differential (˜yi,t ) and monetary authorities (partially) controlling ˜ i,t ). the relative money stock (m As the examples of Switzerland and the Netherlands have shown, the official flexible forex regime can be very stable (‘fear of floating’) when monetary and fiscal policies are coherent. Conversely, officially announced managed floats, crawling bands, or fixed rate systems can be very unstable, a phenomenon called ‘fear of pegging’ by von Hagen and Zhou (2002). This ‘fear of pegging’ behaviour is illustrated by the lively history of the United Kingdom. For a thorough theoretical evaluation of the link between the sustainability of pegs and fiscal discipline, see Canzoneri et al. (2001), and references therein. For a detailed discussion of the trade-offs involving the selection of the forex regime, see Frankel (1999).
3
The nominal forex regime and transitional growth
The international monetary theory presented above implies that the choice of forex regime can be more or less unrelated to the growth and fluctuations in the financial and real sector; what matters is the coherence of the current and anticipated fiscal and monetary policies. What do the data have to say on this matter? To this end, Subsection 3.1 first estimates the monetary model of the exchange rate using data for 40 countries during the 1990s. Subsection 3.2 characterizes the regime specifics by estimating the sample moments of the per-country variables. Finally, in order to disentangle the regime-specific cross-sectional variation of the fundamentals in relation to the forex (regime), Subsection 3.3 analyses the links across the country-specific moment estimates. We focus on the cases of the CEEC transition and the EMU inception during the 1990s. For this reason we choose Germany as the benchmark country. Our panel dataset covers 40 countries over the period
86
General Policy Issues for the Accession Countries
Table 5.1 (at) (be) (fi) (fr) (it) (ir) (lx) (nl) (pt) (sp)
Countries in panel
Austria+ Belgium+ Finland+ France+ Italy+ Ireland+ Luxemburg+ Netherlands+ Portugal+ Spain+
(bg) (cz) (es) (hg) (la) (li) (pl) (rm) (sk) (sn)
Bulgaria# Czech Rep.◦ Estonia+ Hungary◦ Latvia# Lithuania# Poland◦ Romania# Slovak Rep.◦ Slovenia◦
(au) (ca) (dk) (gr) (no) (ns) (se) (sw) (uk) (us)
Australia∗ Canada∗ Denmark+ Greece+ New Zealand∗ Norway∗ Sweden∗ Switzerland◦ UK∗ USA∗
in is jp ko ma mx ph sg th tk
India∗ Israel∗ Japan∗ Korea∗ Malaysia∗ Mexico∗ Philippines∗ Singapore∗ Thailand∗ Turkey#
Notes: + (de facto) peg to DM; ◦ (de facto) crawling or moving band around DM; # (de facto) freely falling to DM; * (de facto) managed or freely floating to DM. Source: Reinhart and Rogoff (2002).
1993:4–1999:3 (at most 24 quarterly observations per country).3 We subdivide the panel into four groups: CEEC, EMU, WEST and REST. The exact composition of the country groupings is given in Table 5.1. Clearly, the CEEC group contains the Central and Eastern European countries, while EMU comprises all current EMU members (with exception of the benchmark Germany). The WEST panel contains Western industrialized non-EMU countries, while the REST panel contains other (less) industrialized non-EMU countries. Table 5.1 also reports Reinhart and Rogoff’s de facto classification of each country’s currency regime versus Germany at the beginning of the 1990s.4
3.1 The international monetary structure We first estimate the coefficients of model (5.4). The non-stationarity of the variables permits estimation in levels, but in order to obtain standard errors, we employ the panel version of Stock and Watson’s dynamic OLS (DOLS) procedure.5 The standard DOLS procedure involves a regression of the level of the endogenous variable on the levels of the explanatory variables, the leads and lags of the first differences of the exogenous variables, and a constant. Accordingly, the empirical counterpart of model (5.4) is ˜ i,t + λ˜ri,t − τ y˜ i,t si,t = c + β m ˜ i,t−1 + ai,2 ˜yi,t−1 + ai,3 r˜i,t−1 + ai,1 m ˜ i,t+1 + ai,5 ˜yi,t+1 + ai,6 r˜i,t+1 + εi,t , + ai,4 m
(5.10a)
Forex Regimes and EMU Accession
87
for i = 1, . . . , N, t = 1, . . . , T . By including the leads and lags of the first differences of the per-country explanatory variables, the empirical specification accounts for cross-country differences in (transitional) short-run ˜ and y˜ are taken in deviation from dynamics. In addition, the data for s, m their mean, so that the empirical model (5.10a) indirectly accounts for fixed country effects. To allow for deterministic drift and seasonal components and cross-sectional heteroscedasticity in the panel residual εi,t , we work with the following decomposition: εi,t = da,t + dq,t + ei,t ,
(5.10b)
with year dummy da,t (equal to unity in year a), seasonal quarter dummy dq,t (equal to unity in quarter q), and we assume that ei,t = σi ζi,t , with ζi,t being Gaussian noise. To compute heteroscedasticity-consistent standard errors, we estimate (5.10) by means of GLS. An important issue is whether the estimated model is structural, in a Lucas (1976) sense (see also Chapter 3 on this issue). The model is unlikely to be structural when panel estimates are very different for different datasets, or when the co-integrating vector does not apply to the panel. To examine the robustness of the estimates, we repeat the estimation procedure for various groups of countries and conduct panel co-integration tests. Our main interest lies in the long-run coefficients (β, τ and λ). To save space we do not report other coefficient estimates. The coefficient estimates and co-integration test results are reported in Table 5.2. The monetary homogeneity hypothesis β = 1 holds up quite well in the different panels, except when the data are restricted to the EMU; see second column of Table 5.2. But this is not so surprising, since convergence between the EMU countries in anticipation of monetary unification in 1998 gives insufficient variation in the money stock data to obtain a reliable estimate. By contrast, the monetary hyper-expansions in the CEECs during the 1990s are very conducive to producing a reliable estimate of β. Notably, the Bulgarian monetary hyper-expansion during the years 1996–97 is influential. When Bulgaria’s data are included, the relative money shocks of the CEECs appear to be completely absorbed by the nominal forex rate; compare third row with the sixth row of Table 5.2. The estimates for the other two long-run coefficients, λ and τ , are plausible. Across the board, the estimates for the interest semi-elasticity λ are positive, and those for the income elasticity τ are negative.
88
General Policy Issues for the Accession Countries
Table 5.2
Panel DOLS regressions for nominal forex model
Panel ALL (862 observations) WEST + REST (437 observations) CEEC (218 observations) EMU (207 observations) ALL − Bulgaria (841 observations) CEEC+Bulgaria (197 observations)
c 0.018 (0.005) (3.513) 0.023 (0.011) (2.014) 0.029 (0.040) (0.723) 0.051 (0.007) (7.391) 0.008 (0.005) (1.552) −0.028 (0.029) (−0.960)
β 0.765 (0.027) (28.589) 0.779 (0.018) (42.845) 0.996 (0.043) (23.397) −0.100 (0.022) (−4.529) 0.634 (0.043) (14.836) 0.847 (0.045) (18.754)
λ 0.921 (0.318) (2.896) 1.451 (0.273) (5.316) 0.542 (1.135) (0.477) 2.904 (0.332) (8.739) 2.055 (0.231) (8.881) 3.286 (0.572) (5.742)
τ −0.667 (0.055) (−12.170) −0.595 (0.060) (−9.994) −1.177 (0.180) (−6.523) −0.039 (0.042) (−0.923) −0.413 (0.063) (−6.527) −0.684 (0.154) (−4.446)
AEG −0.178 (0.041) (−4.365) −0.315 (0.055) (−5.775) −0.176 (0.086) (−2.050) −0.216 (0.035) (−6.192) −0.162 (0.028) (−5.809) −0.147 (0.063) (−2.341)
4:II 1999:III; co-integration model (7.10); pooled co-integration Standard errors (first row) and t -statistics (second row).
To see whether the long-run model applies to a panel, we run the pooled augmented Engle–Granger (AEG) test regression for panel residuals6 ei,t = γ ei,s−1 + γ1 ei,t−1 + γ2 ei,t−2 + ui,t ,
(5.11)
for i = 1, . . . , N, t = 1, . . . , T . It is assumed that ui,t is white noise. The null of the AEG test is a unit root H0 : ei,t ∼ I(1), which corresponds to parameter restriction γ = 0. Under conventional conditions, the asymptotic critical t-value equals −3.74 at the 5 per cent level.7 The last column in Table 5.2 gives the pooled co-integration test results. Convincing evidence for co-integration is found in all panels, except for the CEEC panel. This result actually confirms that the CEECs have been in transition (i.e. their policies were not structural). 3.2 The characterization of monetary regimes In order to characterize the regime specifics empirically, we estimate the monetary forex model’s moment decomposition (5.6). We use the
Forex Regimes and EMU Accession
89
long-run coefficient estimates from the total panel of 40 countries (see first row of Table 5.2). On the basis of the empirical variance decomposition, we compute the fundamentals’ sample correlation ρfg,i = σfg,i σf−1 σ −1 . As the number of countries is large, it is impractica,i g,i ble to report all the per-country estimates. We decided to summarize the results for the various country groupings by reporting the cross-sectional mean and standard deviation of each of the per-country estimates in Table 5.3. There are pronounced differences between the various regions. In general, the mean and the variance estimates are remarkably low in the EMU group (see last column of Table 5.3), while they are substantial in the CEEC group (see third column of Table 5.3), whereas the WEST + REST group takes a middle position. As the numbers in parentheses show, the cross-country differences of regime characteristics are very small within the EMU group, while they are very large within the CEEC group. This finding is important for the cross-sectional analysis of policy regimes.
Table 5.3
Cross-sectional averages for nominal forex model
ALL μs μf μg E σs2 σf2 σg2 σfg U rfg
1.18 (3.84) 0.92 (2.80) 0.29 (0.49) −0.03 (1.95) 56.49 (150.36) 33.48 (31.13) 5.09 (7.17) −1.24 (14.64) 19.16 (130.51) 0.02 (0.22)
WEST + REST 0.80 (3.39) 0.65 (2.89) 0.31 (0.41) −0.17 (1.13) 53.01 (55.49) 30.49 (10.03) 3.28 (3.13) −1.30 (5.15) 20.54 (54.00) −0.04 (0.21)
CEEC 3.24 (5.72) 2.78 (3.07) 0.11 (0.57) 0.34 (3.34) 116.27 (290.41) 52.29 (55.97) 10.57 (12.17) −3.70 (29.18) 57.11 (252.48) 0.12 (0.24)
EMU −0.11 (0.23) −0.38 (1.05) 0.41 (0.56) −0.14 (1.53) 3.64 (4.89) 20.65 (16.07) 3.21 (3.09) 1.33 (2.82) −21.54 (21.57) 0.04 (0.21)
ALL − Bulgaria 0.80 (3.04) 0.74 (2.56) 0.32 (0.46) −0.25 (1.39) 33.92 (47.99) 28.98 (12.74) 4.12 (3.84) 0.93 (5.18) −0.11 (47.36) 0.04 (0.21)
CEEC − Bulgaria 1.83 (3.79) 2.17 (2.52) 0.22 (0.49) −0.56 (1.84) 25.14 (38.12) 34.87 (10.46) 7.00 (4.88) 5.43 (4.53) −22.16 (31.93) 0.18 (0.15)
Notes: Numeraire: Germany; period: 1994:III–1999:III; moment decompositions (7.6); crosssectional standard deviations in parentheses.
90
General Policy Issues for the Accession Countries
High cross-sectional variation is conducive to producing a reliable characterization of policy regimes across countries.
3.3 The links between specific regimes How are the characteristics of the domestic policy regime related to the specifics of the forex regime in place, and vice versa? To provide an answer to this question we run a multiple rank regression (MRR) analysis across the country-specific moment estimates. To emphasize that we are studying (two-way) relationships, we repeat the MRR for each of the country-specific moment estimates. In this way, we uncover the partial links across the regime specifics of the representative country versus Germany during the 1990s. To determine whether particular regime specifics are affected by the choice of forex regime, we include a ‘fixed’ regime dummy in the MRR analysis. We decided to analyse the links across the rank-ordered moments instead of using the raw moments. To underpin our choice of the ordinal above the cardinal association measure, we present in Figure 5.1 the cross-country scatter plots for raw and rank-ordered fundamental average values vis-à-vis the mean forex return. At the top of each of the scatter plots we report the respective correlation estimate. It can be seen that the choice of the metric underlying the association measure is of major importance for the outcomes. For example, take a look at the top panels in Figure 5.1. Pearson’s sample correlation for the pair [μs,i , μf ,i ] is a substantial 0.92. In contrast, when the data are rank-ordered first (see right top panel), the rank correlation estimate is much lower, namely 0.69. None the less, both the ordinal and cardinal association measures point at a clear positive association between μs,i , and μf ,i . But it can happen that the association measure switches sign. From the lower panels in Figure 5.1 it can be seen that for the pair [μs,i , μg,i ] the Pearson’s sample correlation is significantly negative at −0.39, while the Spearman’s sample correlation has an insignificant positive value of 0.06. Overall, we observe that Pearson’s sample correlation is very sensitive to the few ‘hyper-inflation’ or ‘freely falling’ episodes in our sample, while Spearman’s sample correlation is not. Rather than using cardinal scales and to exclude the ‘freely falling’ episodes from the cross-country analysis (as is proposed by e.g. Reinhart and Rogoff, 2002), we decided to include these outstanding episodes in the cross-country analysis and to change the scales to ordinal. To run the MRR, we stacked the expected values (μs,i , μf ,i , μg,i ), the 2 , σ 2 , σ 2 ), and the correlation coefficient (ρ variances (σs,i fg,i ) into an f ,i g,i
Forex Regimes and EMU Accession Average financial growth (μg)
Rank average financial growth (μg)
Pearson’s correlation = 0.9217
Spearman’s correlation = 0.6869 tk 1 pl nn bg 0.9 (es) mx sn (lr) 0.8 pn kois hg bg 0.7 ll ln ma (gr) 0.6 (tr) sk sg th cz 0.5 la (sp) (pt) 0.4 uk au (nl) ns (sw) 0.3 us oa (be) (lx) 0.2 no (at) 0.1 (se) jp (ll) 0 (gk) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 15 Rank average forex change (μg)
tk
10 nn
8 6 4 2
(es)
sn
91
pl
mx ng
5 10 Average forex change (μg) Average real growth (μg) Pearson’s correlation = – 0.3915 (tr) 10 hg
8 6
so ph sk
4
pl mx
(es) tnsn 2 no(gr) 0
rm 0
Figure 5.1
5 10 Average forex change (μg)
Rank average real growth (μg) Spearman’s correlation = 0.06341 (lf) 1 hg jn ko ma 0.9 ph sg (ii) pi 0.8 (at) (se) (sk) mx 0.7 us (lx) (sp) (dk) 0.6 ca is cz ii 0.5 (sw) (tr) (es)(ll) 0.4 (be) (pl) (nl) tn 0.3 tk au ns (be) sn tk 0.2 uk lo jp no 0.1 (gr) nn bg bg 0 15 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Rank average forex change (μg)
Scatter plots for averages of nominal forex model
Notes: Numeraire: Germany; period: 1994:III–1999:III; the identifiers of the countries are given in Table 5.1. In parentheses countries that peg to the DM (or euro).
(N ×7) matrix Q. Then the matrix Q is rank-ordered column for column, giving an (N ×7) rank-ordered matrix Q r . Let qij and qijr be the typical elements of the matrices Q and Q r , respectively. If qij is the smallest element in the column j of the matrix Q, then qijr = 1; if qij is the second smallest value in column j of the matrix Q, then qijr = 2; etc. The MRR analysis involves OLS regressions of a specific column of Q r on all other columns of Q r , a constant aj , and a ‘fixed’ forex regime dummy Difix , that is, r qi,j = aj + bj Difix +
6
r αj,k qi,k + φi,j ,
(5.12)
k
for k = j, i = 1, . . . , N, and j = 1, . . . , 7. We assume that the residual φi,j is white noise. We test the null hypotheses that the partial relations
92
General Policy Issues for the Accession Countries
between columns j and k of the matrix Q r are absent, H0 : αj,k = 0, and that there is no ‘fixed’ regime effect, H0 : bj = 0. We add the volatilities to the mean regressions in order to control for the premiums associated with these risk factors. Similarly, the variance regressions are augmented with mean values in order to control for possible feedback from the premiums to the risk factors. This type of mean-variance regression procedure is theoretically consistent when the gross discrete growth rates of the fundamentals are log-normal distributed and the representative agent has a power utility function, see e.g. Hodrick (1989). We include the forex regime dummy in order to measure a level effect of the ‘fixed’ regime. Accordingly, the dummy Difix equals unity if country i has a de facto peg to the DEM; otherwise it equals zero. We rely on Reinhart and Rogoff’s de facto classification of forex regimes at the beginning of the 1990s (see Table 5.1), except that we reclassified Sweden and Switzerland as having de facto pegs to the DEM. Eventually, the ‘fixed’ regime is assigned to 15 countries: the ten initial members of the EMU plus Estonia, Denmark, Greece, Sweden and Switzerland. We present the MRR results in Table 5.4. The MRR analysis shows that the partial effects between nominal mean values, μs,i and μf ,i , are quite substantial, namely 0.75 and 0.55; see the first and second columns in Table 5.4. These highly significant values indicate that countries with low (high) average financial growth tended to experience small (sizeable) depreciations, and vice versa. The regression for μf ,i also shows that there are (weakly) significant positive mean-variance effects in the financial process. Countries with unstable markets were more likely to have high financial growth rates than countries with stable markets, and vice versa. The third column of Table 5.4 presents the regression results for the average real growth rate. Interestingly, none of the slope coefficients is significant. The average real growth rates are independent of the financial and forex changes, and vice versa. We do find evidence for a link between real and financial volatility. In particular, we obtained significant estimates of 0.32 for the coefficient on σg2 in the MRR for σf2 , as well as a significant estimate of 0.39 for the coefficient on σf2 in the MRR for σg2 ; see fifth and sixth columns of Table 5.4. Economies with more (un)stable output markets had a greater probability of having (un)stable financial markets, and vice versa. Probably the most striking result of our empirical study is the flipside of the above results. Influences of the forex rate variables on the real growth rate are apparently absent. The ‘fixed’ regime dummy (Difix ), the
Forex Regimes and EMU Accession Table 5.4
Multiple rank regressions for nominal forex model
μs μs μf μg σs2 σf2 σg2 ρfg Difix c
93
– – 0.75 (0.19) −0.02 (0.13) 0.16 (0.19) −0.22 (0.12) 0.09 (0.17) −0.01 (0.10) 3.12 (3.84) 4.11 (6.70)
μf 0.55 (0.10) – – 0.06 (0.09) −0.09 (0.11) 0.26 (0.11) 0.16 (0.12) −0.02 (0.10) −5.24 (2.76) 3.55 (5.06)
μg
σs2
−0.03 0.12 (0.24) (0.13) 0.15 −0.09 (0.21) (0.12) – 0.09 – (0.11) 0.20 – (0.25) – −0.37 0.08 (0.24) (0.11) 0.16 0.07 (0.13) (0.11) 0.26 −0.26 (0.17) (0.10) 3.20 −16.78 (6.18) (2.28) 11.84 26.76 (12.05) (3.13)
σf2 −0.24 (0.11) 0.38 (0.15) −0.22 (0.11) 0.10 (0.14) – – 0.32 (0.14) 0.19 (0.12) −4.78 (4.34) 11.17 (6.23)
σg2 0.11 (0.23) 0.29 (0.25) 0.11 (0.09) 0.12 (0.16) 0.39 (0.13) – – −0.04 (0.14) 5.22 (4.87) −1.53 (8.08)
ρfg −0.02 (0.17) −0.04 (0.23) 0.23 (0.14) −0.55 (0.25) 0.29 (0.15) −0.05 (0.17) – – −7.14 (5.21) 25.86 (7.67)
Notes: Numeraire: Germany; period: 1994:III–1999:III; multiple rank regression model (5.12); standard errors in parentheses. Bold-faced values are significant at 5% level.
average forex return (μs ), as well as the forex volatility (σs2 ) do not contribute significantly to the explanation of the cross-country variation in the averages and variances of real growth rate (μg and σg2 ); see third and sixth columns of Table 5.4. On the basis of these results we conclude that the choice of forex regime is not of first-order importance for achieving high and stable real growth. The countries that adopted a peg to the DEM had a significantly lower nominal forex volatility; see fourth column of Table 5.4. Thus, as must be true almost by definition, we find that countries with de facto pegs to the DEM tended to have more stable nominal forex rates. In the MRR for σs2 we obtained insignificant estimates for the coefficients on σf2,i and 2 , while we obtained a significant estimate of −0.55 for the coefficient σg,i on ρfg,i . This result is consistent with the theory on forex set out in the previous section. It supports the view that nominal forex rate stability hinges on the coherence of fiscal and monetary policies.
94
General Policy Issues for the Accession Countries
4
The real forex regime and transitional growth
The empirical results presented above support the forex regime irrelevance hypothesis in the sense that the specifics of the output growth rates of the representative country versus Germany during the 1990s were unrelated to its nominal forex rate characteristics, including the type of currency regime in place. In addition, the empirical analysis showed that the sustainability of pegs hinges on fiscal and monetary discipline. This result is important for the countries aiming to join the EMU. It is an EMU entry requirement to keep the forex within a narrow band versus the euro for at least two years before entry. Prudent fine-tuning of policies is a prerequisite for such a currency system. A second important EMU entry requirement is a euro inflation target; that is, the domestic inflation rate must stay close to that of the Euroarea a whole, with ‘close’ being defined as within 1.5 percentage points. In other words, the potential entrant is in effect required to ‘peg’ both its nominal and its real euro exchange rate to the Euroarea. In view of our analysis, the question arises whether or not the adoption of a real forex peg matters for output growth. In this section we investigate this issue empirically using the same strategy as described in Section 3, except that we now focus on the real forex rate ( p˜ i,t ) instead of the nominal forex rate (s) as our left-hand variable. An additional empirical analysis of the real forex rate would be superfluous in case the PPP holds empirically, i.e. when the nominal and real forex are the same. But if PPP fails in the short run, the real forex rate is likely to play a different role from the nominal forex rate. One plausible cause of PPP deviations is the fact that the nominal and real forex rates apply to different relative price levels. The nominal forex rate solely concerns the relative price of the internationally traded goods, while the real forex rate reflects the relative price of all goods, both traded and nontraded. When the sector inflation rates differ from each other within and between countries, the PPP does not hold. In fact, given that the prices of traded goods equalize across countries while the wages equalize across the sectors (but not across countries), the prices of the non-traded goods will rise faster in economies with relatively high labour productivity growth in the traded sector, and cause apparent international inflation differences; see Balassa (1964) and Samuelson (1964).
4.1 The international monetary structure The pooled DOLS estimates of the coefficients of the monetary real forex model (5.3a) are shown in Table 5.5. The parameter estimates for the real
Forex Regimes and EMU Accession
95
and nominal forex rates are quite similar, as one can see by comparing Tables 5.2 and 5.5. However, the real forex rate reacts more slowly to temporary deviations from the co-integration relationship than the nominal forex rate. The evidence for panel co-integration is weaker in the case of the real forex rate; compare the last columns of Tables 5.2 and 5.5. The AEG test results are more in favour of panel co-integration when the Bulgarian hyperinflation episode is excluded from the panel regression. Overall, the estimation results are acceptable and appear robust to the proposed change of the left-hand variable. 4.2 The characterization of monetary regimes We compute the moments of the decompositions (5.6) on the basis of the estimates for the complete panel (see first row of Table 5.5). Table 5.6 reports the cross-country mean and standard deviations of the sample moments. As before, the moment estimates are large in the CEEC panel (see third column of Table 5.6), while they are small in the EMU panel (see fourth column of Table 5.6). In comparison to the nominal forex Table 5.5
Panel DOLS regressions for real forex model
Panel ALL (862 observations) WEST + REST (437 observations) CEEC (218 observations) EMU (207 observations) ALL − Bulgaria (841 observations) CEEC − Bulgaria (197 observations)
c
β
λ
τ
AEG
0.000 (0.003) (0.160) 0.005 (0.004) (1.220) −0.019 (0.021) (−0.910) 0.004 (0.003) (1.378) −0.002 (0.003) (−0.833) −0.107 (0.031) (−3.395)
0.783 (0.028) (28.290) 0.793 (0.034) (23.534) 0.908 (0.093) (9.804) 0.017 (0.011) (1.646) 0.726 (0.014) (50.563) 0.713 (0.034) (21.104)
1.402 (0.236) (5.951) 1.107 (0.311) (3.562) 1.171 (1.030) (1.138) 0.981 (0.178) (5.515) 1.809 (0.145) (12.462) 2.566 (0.453) (5.669)
−0.978 (0.047) (−20.697) −0.909 (0.044) (−20.600) −1.305 (0.234) (−5.577) −0.047 (0.014) (−3.245) −0.878 (0.028) (−30.996) −1.045 (0.193) (−5.410)
−0.127 (0.035) (−3.652) −0.181 (0.056) (−3.225) −0.030 (0.035) (−0.860) −0.292 (0.062) (−4.700) −0.142 (0.033) (−4.275) −0.085 (0.049) (−1.735)
Notes: Numeraire: Germany; period: 1994:II–1999:III; co-integration model (5.10) and co-integration test (5.11), where the real forex rate replaces the nominal forex rate. In parentheses: standard errors (first row) and T-statistics (second row).
96
General Policy Issues for the Accession Countries
Table 5.6
Cross-sectional averages for real forex model
ALL WEST + REST CEEC μp μf μg E σp2 σf2 σg2 σfg U rfg
2.12 (4.05) 0.92 (2.84) 0.42 (0.72) 0.79 (2.32) 39.27 (166.15) 38.30 (45.78) 10.92 (15.40) −3.20 (28.59) −6.76 (138.32) 0.02 (0.23)
EMU ALL − Bulgaria
1.43 5.50 0.14 (3.33) (5.44) (0.19) 0.67 2.72 −0.40 (2.96) (3.12) (1.07) 0.45 0.17 0.61 (0.60) (0.84) (0.82) 0.30 2.61 −0.07 (1.36) (3.37) (1.74) 22.95 111.04 0.13 (93.20) (305.41) (0.09) 33.92 63.60 21.78 (15.71) (84.96) (16.86) 7.05 22.70 6.90 (6.71) (26.14) (6.64) −2.55 −9.69 2.00 (9.19) (57.18) (4.23) −15.47 34.44 −30.54 (85.66) (253.34) (23.54) −0.04 0.11 0.04 (0.22) (0.26) (0.21)
1.72 (3.17) 0.73 (2.61) 0.46 (0.67) 0.53 (1.66) 15.21 (67.66) 31.50 (15.83) 8.86 (8.24) 1.11 (8.77) −26.25 (63.53) 0.03 (0.22)
CEEC − Bulgaria 4.11 (3.39) 2.10 (2.57) 0.32 (0.72) 1.68 (1.77) 14.78 (26.11) 36.93 (10.86) 15.04 (10.48) 8.26 (7.27) −45.45 (19.84) 0.18 (0.15)
Notes: Numeraire: Germany; period: 1994:III–1999:III; moment decompositions (5.6), where the real forex rate replaces the nominal forex rate; cross-sectional standard deviations in parentheses.
rate, the real forex rate’s average growth rate is higher, while its variance is lower. These discrepancies between the moments of both forex rates underpin that the short-run PPP supposition has little empirical validity.
4.3 Links across regimes To uncover the partial links across the specific regime of the representative country versus that of Germany during the 1990s, we run the cross-moment rank regressions (5.12) for the real forex specification. The MMR estimates are reported in Table 5.7. Comparing the estimates presented in Tables 5.4 and 5.7, we find that the MRR results are quite different depending on whether we focus on the nominal or real rate. This finding is important for policy-makers, since it indicates that the choice of a real euro peg involved different policy trade-offs than the choice of a nominal euro peg. We next examine and discuss these empirical trade-offs in more detail.
Forex Regimes and EMU Accession Table 5.7
Multiple rank regressions for real forex model μp
μp μf μg σp2 σf2 σg2 ρfg Difix c
97
– – 0.45 (0.14) −0.18 (0.08) 0.54 (0.12) −0.11 (0.08) 0.00 (0.09) 0.10 (0.06) −0.43 (2.88) 4.20 (4.19)
μf 0.66 (0.13) – – 0.25 (0.08) 0.03 (0.17) 0.26 (0.11) 0.04 (0.09) −0.13 (0.08) 0.39 (2.14) −2.23 (3.20)
μg −0.62 (0.28) 0.59 (0.17) – – −0.01 (0.33) −0.40 (0.20) 0.25 (0.17) 0.29 (0.11) −2.95 (3.24) 19.85 (5.21)
σp2 0.56 (0.16) 0.02 (0.13) 0.00 (0.10) – – 0.02 (0.08) 0.25 (0.12) 0.00 (0.10) −5.61 (2.66) 5.03 (4.61)
σf2 −0.27 (0.26) 0.43 (0.16) −0.28 (0.11) 0.04 (0.19) – – 0.33 (0.16) 0.20 (0.13) −7.08 (3.83) 14.05 (5.53)
σg2 0.00 (0.21) 0.06 (0.16) 0.17 (0.11) 0.59 (0.19) 0.32 (0.14) – – −0.10 (0.11) 7.51 (3.35) −3.77 (6.00)
ρfg 0.42 (0.26) −0.37 (0.24) 0.34 (0.16) 0.01 (0.37) 0.33 (0.16) −0.17 (0.20) – – 3.89 (4.07) 7.68 (5.18)
Notes: Numeraire: Germany; period: 1994:III–1999:III; multiple rank regression model (5.12); standard errors in parentheses. Bold-faced values are significant at 5% level.
Contrary to its nominal counterpart, the real forex rate appears relevant for output growth. The partial effects between the average output growth (μg,i ) and the average inflation differential (μp,i ) are negative and quite large, respectively −0.18 and −0.62; see first and third columns of Table 5.7. At the same time μf ,i is positively related to both μg,i and μp,i . Thus the joint contribution of the pair [μp,i , μf ,i ] to the explanation of the variation in μg,i is small, confirming the absence of forex effects in Table 5.4. In addition, there are significant partial links between μg,i , σf2 and ρf ,g . Countries with stable and coherent financial growth rates were more likely to have high average output growth rates, and vice versa. The MRR analysis for the real forex model also uncovers a significant positive relation between the output volatility (σg2 ) and the inflation volatility (σp2 ). Thus countries with stable real DEM exchange rates were more likely to have stable output growth rates, and vice versa. On top of that, output volatility has been significantly higher in the countries with high inflation volatility and a nominal peg to the DEM. Altogether,
98
General Policy Issues for the Accession Countries
the evidence suggests that the adoption of a Euroarea inflation target did matter for real growth.
5
Conclusions
This chapter claims that the choice of nominal forex regime is not of firstorder importance for achieving high real growth. This claim is based on empirical evidence that the nominal forex return does not help explain economic growth, nor does a de facto ‘fixed’ regime dummy. A priori, the monetary forex model allows for both possibilities. The model certainly does not imply that the choice of nominal forex regime is important for economic growth. The result that economic growth is insulated from the nominal forex rate variable carries a very positive message for policymakers, supporting the view that the choice of nominal forex regime is irrelevant. Policy can focus on providing coherent fiscal and monetary policy, since it is this coherence that may be conducive to growth stability. The choice of a euro inflation target (i.e. real euro peg) involves some important policy trade-offs. We found a strong negative inflation–growth relationship, i.e. high (low) inflation countries tend to have low (high) output growth, and vice versa. In addition, we detected a strong positive inflation–money relationship, i.e. high (low) inflation countries tend to have large (small) growth rates of the broad money stock, and vice versa. These empirical findings are important for the Accession Countries that have to decide when and how to peg their real forex rates to the Euroarea. A premature attempt to peg the real forex rate to the Euroarea would be undesirable. It would require contraction of output to countervail the real forex rate appreciations caused by labour productivity catch-up in the local traded sector, i.e. the Balassa–Samuelson effect. To avoid such an entry scenario, the incumbent EMU members must be prepared to accept some flexibility of the real forex rates of the CEECs for some years.
Notes 1. We do not necessarily rule out the bubble solution. The bubble component does not affect our results qualitatively. 2. To break away from this implication, suppose that a non-zero risk premium enters the UIP condition (5.7). Generally, the risk premium is a function of the fundamentals, such that the qualitative results carry over. 3. Most of these data were obtained from International Financial Statistics (IFS). The forex rate S is the national currency per US dollar (lines AE and AG IFS). Deutschmark exchange rates are derived using the triangle arbitrage rule.
Forex Regimes and EMU Accession
4. 5. 6. 7.
99
The aggregate price level P is the Consumer Price Index (line 64 IFS). The output Y is industrial production (lines 66 IFS), with the exception of the CEECs. For Hungary, Poland, Romania and Slovenia, we used GDP in historic market prices, while for the other CEECs we deflated GDP in current market prices by the CPI. Money M equals money plus quasi-money (lines 34 and 35 IFS); when unavailable the money data were obtained from the national bank. The interest R is lending rate (lines 60P IFS), or when unavailable the deposit rate (lines 60L IFS). See Reinhart and Rogoff (2002). For alternative de facto classifications of forex regimes, see von Hagen and Zhou (2002) and Ghosh et al. (2002). See Stock and Watson (1993). See Engle and Granger (1987). The choice of critical values is determined by six factors, for which we assume the following: (i) there are three non-stationary regressors; (ii) the residual is stationary; (iii) the number of countries is fixed; (iv) time expands forever; (v) non-stochastic regressors are excluded; and (vi) the residuals are perfectly correlated across countries. Under these assumptions, the asymptotic critical values are those reported in Table 20.2 in Davidson and MacKinnon (1993).
This page intentionally left blank
Part II Country-specific Monetary Policy and Exchange Rate Questions in the Run-up to Monetary Union
This page intentionally left blank
6 Slovenia’s Monetary and Exchange Rate Framework in the Run-up to ERM II Gonzalo Caprirolo and Vladimir Lavraˇc
1
Introduction
The monetary policy framework in Slovenia has been modified three times since independence in 1991. First, price stabilization was pursued within a framework that relied on monetary targeting (1991–95). After a single-digit inflation level was achieved, the stability of the currency measured in terms of both prices and the real exchange rate was pursued by means of dual targeting of both base money and the exchange rate (1996–2001) although formally monetary aggregates were used as intermediate and operating targets. The last change in the monetary framework, aiming at addressing the persistence of inflation and the EU accession requirements, rests on a framework that uses the exchange rate as a nominal anchor for reducing inflation (2001–). Two important considerations have guided policy formulation and implementation since independence: the relatively high degree of discretion in conducting monetary policy and the cost dimension. Monetary policy targets (M1) were not announced until 1997 and, from then onwards, wide target ranges for the new M3 intermediate target were used until recently. The last change in the monetary policy framework that took place in 2001 consists of targeting an unannounced predetermined exchange rate depreciation path. Cost minimization of implementing monetary policy, what could be called ‘fear of paying’, has been a permanent and critical consideration even at the expense of suppressing price signals in the economy, even at times when response to shocks would have required movements in the interest rates. This is reflected in the use of direct monetary policy instruments (acquiescence with collective 103
104 Country-specific Monetary Policy and Exchange Rates Freely falling depreciation below compatible depreciation. High currency risk. Money-based stabilization policy. Passive crawing, actual depreciation lower than PPPcompatible depreciation. Capital controls in place. Dual targeting price and real exchange rate stability.
35 30
Year on year (%)
25 20
Active crawling actual depreciation below PPP-compatible depreciation. Exchangerate-based stabilization policy.
15 10 5 0 1993
1994
1995
1996
1997
PPP-compatible exchange rate change
Figure 6.1
1998
1999
2000
actual exchange rate change
2001
2002
inflation
Exchange rate and monetary policy, Slovenia
interbank agreement on interest rates), pervasiveness of capital controls, use of indirect policy instruments that target volume of money rather than its price (tap sale of securities and volume auction) and recurrent sterilization at negative real interest rates. The monetary policy framework has evolved, but it basically relies on the exchange rate transmission mechanism of monetary policy, while the interest transmission channel remained obstructed. This hampered the use of interest rates in defence of the exchange rate and to convey monetary policy stance. The exchange rate regime, a de jure managed floating regime since independence, has been de facto modified in accordance with the main policy objectives that guided monetary policy in different periods (see Figure 6.1). In the period 1991–95 the exchange rate path (nominal and real) followed the trajectory depicted by the overshooting model of Dornbusch. The authorities at that time could not do much to prevent fundamentals-driven exchange rate movements. In 1996, after singledigit inflation was reached, the exchange rate regime shifted to a de facto passive crawling exchange rate regime in which the monetary authorities aimed at closing ex post the domestic–foreign inflation differential, and at curbing exchange rate volatility. As a result of the policy implemented during the period 1996–2001, the real value of the currency measured
Slovenia’s Monetary and Exchange Rate Framework
105
in terms of the CPI-deflated real effective exchange rate remained fairly constant but inflation lingered on at a high single-digit sticky level. The last change in the exchange rate regime took place in order to address the challenges of the prospective membership of Slovenia in the EU and simultaneously to tackle the problem of persistence of inflation. In 2001 the exchange rate regime shifted to a de facto active crawling peg in which the exchange rate closely follows an unannounced depreciation path. The design of Slovenia’s monetary policy has been influenced by four important events: (i) independence in 1991 and inherited hyperinflation, which influenced the choice of a monetary aggregate as the nominal anchor for price stabilization; (ii) an interest rate shock in 1995, driven by de-indexation of demand deposits and changes in the methodology of calculating indexation of financial contracts; (iii) Slovenia’s regaining access to international financial markets in 1996 and the preemptive capital controls policy from 1995 onwards to reduce the cost of implementation of monetary policy; (iv) accession negotiations with the EU in the fields of monetary and exchange rate policy and prospective EU membership that led to the dismantling of capital controls, changes in the central bank law (clearly defining price stability as the primary policy objective), an enhancing of transparency and a more decisive stance to reduce inflation to EU levels. The main question for the future is whether the recently modified policy framework will have to be modified once more before Slovenia joins the ERM II, for example, changing the existing regime to one that also relies on monetary policy implementation through the interest rate channel, which is less vulnerable to unintended depreciation of the exchange rate. Similarly, consideration should be given to shifting to a new regime that would be less discretionary and more transparent, in order to anchor inflationary expectations, and that would respond to the requirement of preparing the financial sector to the ECB policy operating environment where monetary policy is implemented and transmitted primarily through the interest channel and not through the exchange rate channel in a mostly non-indexed environment. A shift in the strategy to a de facto compliance with the ERM II requirements before joining the EU, as done in Hungary (see Chapter 10) could potentially reduce the risks of postponing entry to the above-mentioned exchange rate regime and of delaying the adoption of the euro. In any case, such a policy change in the monetary policy framework would have to occur at the time of joining the ERM II at the latest. The remainder of this chapter is divided into three sections, presenting the monetary and exchange rate policy frameworks during the three periods.
106 Country-specific Monetary Policy and Exchange Rates
2
Money-based stabilization policy (1991–95)
Slovenia inherited hyperinflation from the former Yugoslavia. The main policy task after independence and the introduction of the new currency (the tolar) in 1991 was to achieve price stabilization. The authorities decided on a money-based stabilization policy. The intermediate target was M1, while base money was the operating target. The strategy was successful in bringing down inflation to a high single-digit level by 1995. Fiscal policy to a large extent contributed to this outcome as the general government budget even registered surpluses throughout the period (1992–96). While a de jure managed floating exchange rate regime has been in place since independence in 1991, the description that seems best to fit the actual exchange rate developments and changes in the exchange rate regime is the one of a freely falling exchange rate regime during the price stabilization period (1991–95) and then, from 1996 on, that of a crawling exchange rate regime (Reinhart and Rogoff, 2002). However, it could also be argued, based on the fact that exchange rate intervention had already started in the second half of 1992 to prevent real exchange rate appreciation (Mencinger, 2001), that an early classification of the exchange rate regime as a de facto crawling peg is more suitable. In our view, given the fact that the nominal and real exchange rate paths followed the stylized trajectories of the Dornbusch (1976) overshooting model during the period of price stabilization (1991–95) until the inflation rate reached the high but sticky single-digit level, it seems that the description that best fits the exchange rate regime during the period 1991–95 is one of freely falling. This characterization also seems appropriate given the fact that the monetary authorities could not do much to influence the exchange rate trajectory (BoS, 1996) as determined by the fundamentals, and an intensive exchange rate intervention would have been in conflict with the stabilization objective pursued by the authorities. The operating environment in which monetary policy was implemented during the first half of the decade had the dominant features of a economy closed to capital inflows due to the high currency premium, lack of access to international financial markets, and the freely falling exchange rate phase required a money-based policy to achieve price stabilization. Monetary developments in this period were driven mainly by domestic residents’ regaining confidence in the currency and current account developments under an overshot exchange rate. Despite the specific features of an economy in transition and the shock resulting from the loss of the internal market of the former Yugoslavia,
Slovenia’s Monetary and Exchange Rate Framework
107
Table 6.1 Macroeconomic indicators for the Slovenian economy (1993–96)
Inflation (annul average %) Base money (average growth %) M1 (average growth %) Basic interest rate (TOM %) REER (CPI, 1995=100) Private consumption (real growth %) Real growth of GDP (%) Exports (real growth %) Imports (real growth %) Current account balance (% GDP) Memorandum item: real wage growth
1993
1994
1995
1996
32.9 80.4 86.9 22.4 87.0 13.9 1.9 2.1 13.0 1.5 13.3
21.0 46.8 55.8 18.7 90.2 5.1 4.9 10.0 5.9 4.0 3.6
13.5 48.2 40.4 8.2 100.0 9.1 4.1 1.1 11.3 −0.6 4.4
9.9 13.1 25.7 9.7 96.7 2.1 3.5 3.6 2.1 0.2 4.9
Sources: Bank of Slovenia Monthly Bulletin, various issues and IMAD Spring and Autumn reports, various issues.
it is possible to argue that the economy during the money-based price stabilization period exhibited (Table 6.1), to a large extent, features and empirical regularities similar to other money-based programmes in chronic inflation economies observed by Calvo and Vegh (1999). 1. Inflation was reduced sharply after one year. 2. There was an initial increase in domestic real interest rates in 1992. Since they were indexed to inflation, rates declined with it in 1993. However, the decline slowed down up to 1994, partially offsetting the excess money demand. In 1995, changes in indexation led to a faster reduction of interest rates and a disequilibrium in the money market. 3. The real exchange rate appreciated (particularly in 1995) to correct the money-market disequilibria. This led to an increase in imports which in turn contributed to the current account deterioration in 1995. 4. There was contraction in economic activity in the first year but it is difficult to assess the contribution of the stabilization policy as the economy suffered multiple shocks. However, real growth turned positive as early as 1993, earlier than in other programmes. 5. Contrary to other programmes, in which the inflation rate slowly converged to the rate of growth of money supply, in Slovenia the reverse happened, which seems to reflect strong money demand for transaction purposes. Also contrary to other programmes, further deterioration of the current account was avoided, and reversed at the end of 1995 by a sharp depreciation of the tolar, triggered by an
80
120
70 60
110
50 100
40 30
90
20 10 0 1993
80 1994 Exchange rate
Figure 6.2
REER index 1995 = 100
Inflation and exchange rate, year on year (%)
108 Country-specific Monetary Policy and Exchange Rates
1995 Inflation
1996 REER
Exchange rate overshooting, Slovenia
increase in money demand resulting from the elimination of indexation on one-month demand deposits and changes in indexation calculation methodology. As said above, the movements of the exchange rate during the period of price stabilization followed broadly the characteristics of a freely falling exchange rate regime and the stylized trajectories of the Dornbusch overshooting model (see Figure 6.2).1 This characterization seems the most appropriate for the whole period of stabilization and not only for the first two years after independence (Reinhart and Rogoff, 2002).2 The arguments in favour of characterizing the exchange regime as a freely falling regime are threefold: (i) although inflation in Slovenia was above 40 per cent until April 1993, single-digit inflation was reached only in the last quarter of 1995; (ii) the policy priority was to achieve price stabilization (BoS, 1996); thus reverting the real appreciation of the exchange rate would have undermined the price stabilization process and would have been difficult to implement; and (iii) while Slovenia did not experience a currency crisis, the effects of introducing a new currency are similar. As the evidence shows, the exchange rate (nominal and real) followed an overshooting trajectory during the stabilization period. The freely falling episode started with the initial setting of the nominal exchange rate of the tolar against the DEM at a rate that implied a 16 per cent real devaluation, compared to the December 1989 rate, and then it continued with nominal and real exchange rate overshooting
Slovenia’s Monetary and Exchange Rate Framework
109
trajectories, whereby the exchange rate sought a clearing level for the foreign exchange market.3 The observed exchange rate trajectory during the period was broadly consistent with the monetary authorities’ policy of controlling money supply, a gradual increase of confidence in the currency and the existence of relative price rigidity. The perception that the exchange rate might have overshot its long-run value seems to have occurred in early 1994, when the last sharp upward swing of the monthly depreciation rate was observed. Afterwards, the annual depreciation rate started falling smoothly. At that time, individuals became willing to hold tolars at any domestic interest rate. The expected exchange rate appreciation (the adjustment from the overshooting), combined with tight control of domestic credit, corrected the money demand imbalance (the former by raising money demand at any given interest rate, the latter by reducing its supply). It can be said that a new exchange rate equilibrium was reached in 1995. The growth rate of base money, the operating target of BoS, in 1995 finally reached a high, single-digit, steady-state growth rate characteristic of a post-stabilization period (1996–2002). The increase in money demand, as evidenced in the gradual repatriation of foreign currency deposits by households in the period 1992–95, particularly in 1994–95, stabilized as the high growth rates of total deposits in the banking system decreased after 1995. The unequivocal sign of the end of the overshooting and probably the end of the stabilization period was the first nominal appreciation of the tolar, from March until June in 1995, and the underlying appreciation of the RER. Another stylized fact of this period is that the monetary authorities influenced interest rates only indirectly, through the decline of inflation anchored in the money-based stabilization policy, given the huge volatility of interest rates (e.g. interbank rate4 ). An important factor explaining interest rate volatility and the lack of a direct link between policy intentions and interest rates (nominal and real) developments is indexation. Interest rates in the economy were tied to the movements of the indexation factor of financial contracts, the so-called basic interest rate of the economy (TOM) calculated by BoS. Until 2002 all financial contracts denominated in tolars with maturity beyond 30 days were indexed to the TOM, equal to the previous monthly inflation rate until June 1995.5 Interest rates were either tied to TOM or to the exchange rate (annualized BoS’s end of month DEM and euro exchange rate growth) and exhibited substantial volatility. Interest rates, by being tied to past inflation, were always positive in real terms, with the exception of a
110 Country-specific Monetary Policy and Exchange Rates
35 30 25 20 15 10 5 0 1993
1994
1995
1996
1997
Inflation (year on year) Bill with warrants Figure 6.3
1998
1999
60-day bill 60-day bill (EUR)
2000
2001
270-day bill TOM
BoS nominal interest rates (%)
period in 1995 when the methodology of estimating TOM changed and when indexation of demand deposits with maturities of less than 30 days was abolished. The monetary authorities did not influence directly the nominal and real level of interest rates in the economy, which is explained by the BoS monetary target and by a cost minimization policy (‘fear of paying’). In fact, the interest rate on central bank bills (CB bills) used for controlling base money (foreign currency bills or bills with warrants indexed to either the exchange rate or exchange rate and inflation), when expressed in tolar nominal terms, was negative in real terms (see Figure 6.3). The BoS attracted banks’ demand for CB bills, particularly foreigncurrency-denominated bills that were the main sterilization instrument during the period, by granting banks access to its standing credit facilities and to repo operations using CB bills as collateral, and by requiring their use to meet reserve requirements. Such an approach on the one hand delivered lower costs of monetary policy implementation, but on the other hand hindered the functioning of the interbank market. Moreover, the central bank also refrained from using interest rates to respond to shocks. In particular, the BoS did not raise interest rates to offset the shock caused by the above-mentioned changes in indexation in 1995.
Slovenia’s Monetary and Exchange Rate Framework
111
It responded instead by imposing capital restrictions and administrative measures from February 1995 onwards (as a consequence of high sterilization costs in 1994, see Caprirolo and Lavraˇc, 2001). The BoS influenced the general level of interest rates only by means of direct instruments. Also in 1995, in order to reduce high lending rates resulting from aggressive pricing behaviour of smaller banks to attract deposits, the BoS brokered an interbank agreement capping deposit interest rates. Such an agreement lasted until March 1999, when it was formally abolished. The lack of interest rate signals in the economy and the obstruction of the interest rate channel is explained to a large extent by the monetary targeting policy and is primarily due to cost considerations (reliance on CB bills which absorbed liquidity at non-market-clearing rates). The interbank agreement on interest rates was also an important element in this regard. From the operational environment’s point of view, the main impediment was the widespread indexation of financial contracts.
3 Price and real exchange rate stability dual targeting policy (1996–2001) In 1996 the BoS de facto shifted to a different monetary policy framework, whose primary goals were price stability and external equilibrium. This dual intermediate target policy framework relied on the control of money growth (first M1, later M3) to achieve price stability, and on managing the exchange rate to close the domestic–foreign price differential to preserve the stability of the real exchange rate. Capital controls, which were progressively introduced since February 1995 to reduce the cost of implementing monetary policy, and in a pre-emptive attempt to discourage capital inflows (Bole, 1994), enabled a dual target policy until late 2001.6 The monetary policy framework resembled the one described by Bofinger and Wollmershäuser (2001) in which the monetary authority would pursue the accomplishment of internal equilibrium by influencing interest rates and external equilibrium by managing the exchange rate. However, the BoS used monetary aggregates – and not interest rates – to preserve the internal equilibrium, while it used exchange rate intervention to maintain the external equilibrium. The operating targets were base money and the exchange rate (SIT/DEM and then SIT/euro). To control base money, the BoS resorted to sterilization of excess liquidity CB bills sold on tap. Their interest rates were set administratively on non-market-clearing levels (for instance, the key 60-day bill rate was negative in real terms from the second half of 1999 until late 2000).
112 Country-specific Monetary Policy and Exchange Rates Freely falling
Passive crawling
165.0
105.0
155.0
104.0
145.0
103.0
135.0
102.0
125.0
101.0
115.0
100.0 99.0
105.0 ×
×
× 98.0
95.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1992
1993
1994
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
1995
1997
1996
1998
1999
Active crawling 107.0 106.0 105.0 104.0 103.0 102.0 101.0 100.0 95.0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 2000
2001
2002
Figure 6.4 Exchange rate dynamics during different regimes, Slovenia (year on year; January = 100)
Until 1997 the control of base money by means of sterilization was implemented with foreign currency bills. After 1997 sterilization was implemented with tolar-denominated CB bills (non-indexed 60-day bills and indexed to inflation 270-day bills), while foreign currency bills were used as collateral in open market operations and for meeting reserve requirements. The price of the CB bills during the period did not reflect the BoS policy stance. Thus, indirect instruments withdrew excess liquidity without conveying price signals to the economy. The TOM continued to be the reference interest rate in the economy throughout the period. The exchange rate was managed as a de facto passive crawling peg regime, in which exchange rate movements aimed at eliminating the domestic–foreign price differential. The correction of price divergences took place mostly in the second half of the year (especially during the 1996–98 period: see Figure 6.4). Exchange rate intervention was also
Slovenia’s Monetary and Exchange Rate Framework
113
used to reduce exchange rate volatility. During the years 1999–2000 the pattern of depreciation of the exchange rate, although still closing the inflation differential, was not the same. It was affected by the shock produced by the introduction of VAT, which triggered an unintended depreciation of the exchange rate in 1999, and by a deterioration in the terms of trade that led the BoS to curb depreciation at the end of 1999, displaying a more aggressive exchange rate policy stance to correct the current account deficit in 2000. As a consequence, the CPI-deflated real exchange rate depreciated in 2000. Foreign exchange intervention at the beginning of 1996 was first implemented by means of two cumbersome instruments, the so-called ‘triple offer’ and ‘purchase of foreign exchange with the right-to-sell’, and from 1997 onwards by means of a special agreement between the BoS and the commercial banks (the so-called ‘club’), according to which BoS set the parameters for foreign currency transactions between banks and their clients. The ‘club’ agreement was modified in 1999, 2000 and 2001.7 Two important features inherited from the earlier stabilization period which obstructed the interest channel of monetary policy continued to characterize the operating financial environment during the period under consideration here: the indexation of financial contracts to TOM throughout the period and the ceiling on deposit rates of commercial banks until 1999. In particular, the traumatic experience of the aftermath of the de-indexation of deposits with maturities below 30 days in 1995 seems to have strongly influenced monetary authorities’ attitude towards a further de-indexation of financial contracts. As financial deepening continued, the relationship between the targeted monetary aggregates and inflation became unstable. After 1996, average growth rates of different monetary aggregates stabilized and appeared not to contribute to further reductions in inflation (see Figure 6.5). The first clear reaction to this development was the change of intermediate target from M1 to M3, although the BoS kept base money as the operational target. During the reduction of inflation that culminated with the lowest annual rate in June 1999 (4.3 per cent), annual growth rates of M1 and base money exhibited rising trends, while M3 annual growth rate started to decline only in the second half of 1998. The behaviour of M3 was influenced by the underlying change in its composition towards the tolar component (the share of foreign currency in M3 declined from 33.1 per cent in 1996 : 1 to 23.8 per cent in 1999 : 2), and by the lower depreciation of the exchange rate during 1997–98. After the VAT shock, the annual growth of M3 did not decline further, which can be partly
114 Country-specific Monetary Policy and Exchange Rates
100 90 Year on year (%)
80 70 60 50 40 30 20 10 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Inflation Figure 6.5
M0
M1
M3
Monetary aggregates and inflation, Slovenia
explained by the reversal in the composition of M3 towards the foreign currency component. The stabilization of the growth rates of monetary aggregates in 1997–99, and the relatively little progress in inflation reduction, particularly after 1999, also indicates that further reduction in inflation by monetary targeting would have required a substantial additional monetary tightening. To enhance credibility and transparency in its policy, the BoS started to announce its targets publicly for the first time when it changed its intermediate target to M3 in 1997. The target was set in terms of a relatively wide range of the last quarter annual growth of M3 for the year. The range was 8 per cent during 1997–99, later reduced to 6 per cent in 2000–02. This relatively wide margin for the target provided a satisfactory degree of discretion for the BoS to alternate between the objectives of reducing inflation and maintaining external competitiveness. Nevertheless, this dual targeting framework, instrumented via control of quantity of money within a wide range and an unannounced depreciation path of the exchange rate, could not serve as an anchor to inflationary expectations. The monetary stance was not visible and the short-term connection between the monetary aggregates and inflation was not obvious. Moreover, the policy stance was to a large extent interpreted as conflicting, particularly because the exchange rate accommodated various price disturbances. The main weaknesses of the framework, besides the lack of a visible anchor for inflationary expectations required for reducing inflation, was the control of base money via CB bills sold at
Slovenia’s Monetary and Exchange Rate Framework
115
non-market-clearing rates. Thus, in the short term, the CB virtually relied on the exchange rate as the only price-information-giving instrument. The obstructed interest rate channel of monetary policy further aggravated this. Such a framework, even when supplemented with capital controls, was highly vulnerable to any unintended currency depreciation driven by foreign currency outflows (e.g. loss of confidence in the currency or a current account deficit). This vulnerability was latent even if BoS could resort to the ‘club’ bank agreement on foreign exchange transactions to curb the depreciation of the exchange rate, because the central bank, in the event of a sudden increase in domestic demand for foreign currency or in a crisis of confidence in the currency, could only act reactively to reduce depreciation of the exchange rate and never proactively. However, this dual targeting framework, supported by capital controls, did allow a gradual reduction of inflation until mid-1999, just before the VAT introduction. The policy mix, as captured by a depreciation of the exchange rate lower than the one required to close the domestic–foreign price differential, highlights the critical importance of the exchange rate in reducing inflation (see Figure 6.1). In fact, the annual average depreciation rates of the exchange rate in 1997 and 1998 despite the intervention were the lowest registered so far (below 2.5 per cent annually). The importance of the exchange rate in inflation dynamics is further highlighted if one remembers, as said above, that the average annual growth of money aggregates remained stable, and that the share of controlled prices in the CPI basket declined significantly from 22.4 per cent in 1996 to 14.3 per cent in 1999 – albeit their contribution to inflation was the highest in seven years (EBRD, 2002).8 Capital controls contributed to achieving the relatively constant real exchange rate (using the level of 1995 as the basis) and to lower inflation during 1996–99 primarily by reducing the cost of implementing monetary policy. However, it can be argued that similar policy outcomes could have been achieved by simply relying on actively targeting the domestic–foreign price differential (or the uncovered interest differential) in a de facto active crawling peg regime, instead of using capital controls extensively which, in fact, was the strategy pursued since 2001. This alternative strategy could have been even more suitable taking into account that during the mentioned period FDI inflows were low and privatizations of large public enterprises had not yet started to occur. The best characterization of the exchange rate regime in this period is one of a de facto passive crawling peg.9 The BoS did not announce the rate and pattern of the exchange rate crawl, which was determined on the basis of ex post developments in the domestic–foreign inflation
116 Country-specific Monetary Policy and Exchange Rates Table 6.2
Macroeconomic indicators for the Slovenian economy (1998–2000) Real growth rates (%) Government Exports Private Gross fixed Imports GDP CAB Inflation consumption cons. capital $US mil. (year on formation year, %)
1998 : 1 1998 : 2 1998 : 3 1998 : 4 1999 : 1 1999 : 2 1999 : 3 1999 : 4 2000 : 1 2000 : 2
0.8 −2.7 9.8 −4.2 2.1 7.1 −6.7 3.0 1.1 2.8
−0.6 2.1 3.5 2.0 −5.5 2.3 2.0 3.9 2.2 3.3
1.0 −0.4 2.5 −1.2 2.4 5.1 −3.7 2.4 −0.7 2.1
12.2 −11.4 14.2 10.4 −3.1 21.1 −17.8 9.3 −1.4 3.6
8.2 −5.3 7.1 4.7 −2.2 8.9 −8.3 5.8 1.1 −4.0
6.0 −74.9 2.5 −29.6 3.3 161.4 3.4 −52.2 2.9 −50.9 7.4 −493.7 4.3 18.2 5.0 −256.2 6.3 −169.0 3.6 −89.0
3.1 1.8 0.2 1.4 1.8 1.0 3.5 1.9 2.6 1.7
Source: Statistics Office of Republic of Slovenia. National Accounts, various issues. Bank of Slovenia Monthly Bulletin, various issues.
differential (the latter particularly influenced the exchange rate movements in the second half of the year). Similar indications of the existence of a passive crawling peg are the facts that the actual change in the exchange rate was below the required change needed to close the inflation differential, particularly in the period 1996–98, and that a shift to a more proactive policy stance took place in 2000. The robustness of this monetary framework was tested when VAT was introduced ( July 1999). As a consequence of inflationary expectations concerning the eventual price impact of VAT (Table 6.2), aggregate demand increased in the second quarter of the year in anticipation of the shock (e.g. the annual real growth rate of consumption was 6 per cent, the highest rate so far since 1995), the current account turned into a substantial deficit, while the exchange rate depreciated significantly (the annual rate of depreciation jumped above the annual inflation rate, a situation observed only at the end of 1995 and during the stabilization process), reversing for the second time the pace of disinflation in the economy (the first time this occurred was in 1995). The BoS did not act pre-emptively to quash inflationary expectations and to dampen the strong domestic demand which triggered a sharp depreciation of the tolar. The Bank could have increased interest rates pre-emptively to offset the depreciation of the currency driven by inflationary expectations and the jump of domestic demand in the
Slovenia’s Monetary and Exchange Rate Framework
117
second quarter of 1999 in anticipation of VAT introduction, but it didn’t.10 In particular, it could have raised the interest rates on CB bills: actually, the key rate of the 60-day CB bill turned negative in real terms in the last quarter of 1999 and remained so until the last quarter of 2000 (Figure 6.3). The BoS also did not increase the interest rates on the rest of its instruments.11 The resulting high depreciation rate in 1999 and the reversion in the declining annual growth rate of M3 also point to the accommodation of the shock (IMF, 2001). Even if the BoS had acted preemptively to quash inflationary expectations by raising interest rates, it is questionable whether such a move would have been transmitted by the banking system to the economy, given the indexation and interest ceiling on deposit rates. This highlights once more the weakness of the monetary framework, particularly its limited responsiveness to an unintended depreciation of the exchange rate. The similar developments in 1995 and 1999 indicate that the exchange rate influences prices faster than monetary aggregates, and that there is a strong pass-through from exchange rate to inflation.12 Regarding the size of this pass-through in Slovenia, empirical analysis indicates that it ranges between 0.8 and 1 (Coricelli et al., 2002, estimate the long-run or equilibrium pass-through effect to be about 1). This high pass-through might be explained by the fact that the exchange rate is the most visible policy price variable in the economy, the one that reflects most clearly current developments and because it is the price that the central bank can influence directly and effectively. It clearly indicates that whatever strategy of reduction of inflation is to be implemented has to rely on weakening this strong link by either: (i) lowering the depreciation rate in the case of the current exchange rate regime, or (ii) allowing the exchange rate to float or simply fixing the exchange rate if the exchange rate regime is to be changed. In a crawling peg or floating exchange rate regime it is necessary to develop a strong interest rate transmission mechanism of monetary policy to defend the currency in case of shocks (primarily domestic driven), to convey policy stance and to influence expectations. When evaluating monetary policy during this period, it is also important to assess whether the BoS gave more priority to the internal or external equilibrium objectives. It is possible to argue that it gave higher priority to the internal target, attributing inflation persistence at high single-digit levels to the inherent vulnerability of the framework to unintended depreciation and to the obstructed interest rate transmission channel. However, the fundamental question to address is whether, within the limitations of the existing framework,the BoS could have
118 Country-specific Monetary Policy and Exchange Rates
been more assertive in lowering inflation by reducing the speed of the exchange rate crawl if, for example, the existence of a Balassa–Samuelson (BS) effect in Slovenia was recognized. Most empirical studies have acknowledged the existence of a BS effect in EU candidate countries. Empirical research on the presence of the BS effect in Slovenia has been also examined in several studies based on time-series and panel co-integration techniques. Most of them find that relative price developments depend on relative productivity developments. The main difference in Slovenia seems to be the period in which the BS effect is present. While some pioneer studies like Rother (2000), using sample data for 1993–98, found that productivity differentials explain about 1.5 per cent higher equilibrium inflation in Slovenia than in Germany, recent studies based on time-series analysis indicate that the effect of the productivity differential on the Slovenia–Germany inflation differential over more extensive period (1991–2001) is negative (Egert, 2002), or lower (1993–2001) at 0.7 per cent (Žumer, 2002). The difference primarily concerns the first half of the 1990s, when the structural changes and labour market developments that accompanied the transition process made it difficult to account for any early presence of a BS effect. However, it seems to intensify in the period 1995–2001, explaining an inflation differential that ranges from 0.8 per cent (Egert, 2002) to 1.4 per cent (Žumer, 2002) and that it should remain between 1 per cent and 2 per cent in the future, given the trends in productivity growth (IMAD, 2002). If it is accepted that relative price developments depend on relative productivity developments in Slovenia at least in the second half of the 1990s, it is possible to assess whether the inflation objective or the external equilibrium had priority, by comparing the depreciation rate differential between the BS-consistent exchange rate and the actual exchange rate, on the one hand, and by comparing the resulting differential to the actual movement of the CPI-deflated REER, on the other. It can then be shown that even for a 0 per cent BS effect on inflation differentials, the external equilibrium objective had priority in the year 1996 and in 1999–2001 as the REER depreciated in those years. This could even be the case for 2002 if a BS effect on inflation differential of 1 per cent is accepted. It is also clear that, with or without the BS effect, the inflation objective had priority in 1997 and 1998, as the actual depreciation of the exchange rate was lower than the one consistent with a large BS effect. Similarly, the REER seems to have appreciated in both years above the real exchange rate, consistent with a BS effect on prices of about 1 per cent, particularly in 1997.
Slovenia’s Monetary and Exchange Rate Framework
119
Thus, once the presence of a BS effect on prices is recognized, at least for the period 1999–2002, it is possible to conclude that the central bank could have exercised a more decisive stance towards lowering the depreciation rate of the tolar and thereby lowering inflation. However, the central bank’s preferences seem to have shifted towards external equilibrium at the expense of inflation, as it reached a rate of 8.9 per cent at the end of 2000.
4 Exchange-rate-based stabilization policy and accession to ERM II (2001–) Recent changes in the monetary policy framework are deeply rooted in the accession negotiation process to the EU. This process already triggered the lifting of most capital controls in 1999 (credit operations), leaving restrictions mainly on short-term inflows (portfolio investments with maturity below six months), which were removed later, in January 2002. Similarly, the central bank law had to be amended, particularly with regard to strengthening the final objective of the central bank in terms of price stability. The negotiation process also resulted in an increase of transparency, required by the need to align monetary instruments with those used by the ECB. The impact of the accession process has not only influenced monetary policy directly, due to Slovenia’s bilateral negotiations with the EU on the subject, but also indirectly, as other Central European countries have taken decisive steps to lower inflation in order to join the EMU as soon as possible. The sharpening of the policy stances in other candidate countries resulted in comparatively lower inflation rates than in Slovenia, particularly at the end of 2002, which put additional pressure on Slovenian policy-makers. The central bank decided to change its monetary policy framework in 2001 as a result of the frustration with persistent inflation, the changes resulting from EU negotiations and the challenges of joining the ERM II, and later the Euroarea. The new framework had to address simultaneously the operating environment of a small, open economy, practically for the first time, and the requirement of lowering inflation in order to join ERM II as soon as Slovenia became an EU member. In designing the new policy framework, the central bank explicitly recognized the limited controllability of M3 and its impact on attaining the final objective of price stability, the critical importance of the exchange rate channel, both in terms of inflation and external equilibrium, and the negligible significance of the interest rates channel (BoS, 2001).
120 Country-specific Monetary Policy and Exchange Rates
An aspect that deserves particular attention is the interest rate channel. Among the reasons explaining its weaknesses, the BoS includes the widespread use of indexation of financial contracts, the presence of the interbank agreement on interest rates until the end of 1999 and the existence of ‘structural’ excess liquidity in the money market. The other important factor, as discussed previously, is the implementation of monetary policy (sterilization) by means of non-market instruments. The use of these instruments contributes to perpetuating excess liquidity and also does not convey the policy stance. In particular, a 270-day CB bill was auctioned for the first time only in November 2001, and this instrument remains the only one that is currently auctioned. Even here the CB sets in advance the volume and the highest interest rate, thus in effect capping interest rates. The new policy framework introduced at the beginning of 2002 may best be described as an informal system of inflation targeting (see also the Czech experience in Chapter 8). The BoS aims at achieving a mediumterm inflation projection – not a binding inflation target – by relying on the so-called ‘two-pillar’ approach, similar to the approach employed by the ECB. The first pillar, consistent with BoS’s previous policy emphasis, is control of broad money and its components (M3). The second is a set of indicators (external equilibrium and its determinants, wages and controlled prices) which may be used to justify deviations from the previously stated and now projected reference values for M3 growth.13 The weakness of the new framework is that it still provides a high degree of discretion to the CB. Moreover, as the broad target range of M3 has been abandoned but not replaced, there is no anchor for inflationary expectations. However, given the fact that the overall environment resulting from the EU accession compels prioritization of the reduction of inflation, and that the central bank formally recognized the role of the exchange rate as the main instrumental variable, the central bank has changed its policy to a de facto exchange-rate-based stabilization policy (ERBS), in which the exchange rate is the main anchor of monetary policy and the operating target.14 Since the exchange rate closely follows an unannounced depreciation target path from 2001 onwards, the de facto passive crawling regime in place since 1996 has evolved into a de facto active crawling regime (see Figures 6.1 and 6.4). The new policy is implemented via a gradual reduction of the depreciation rate of the exchange rate. Since 2001 the central bank has intervened in the foreign exchange market in the framework of a modified agreement by which all banks now trade in foreign currency with third parties within narrow bands around a base rate set by the central
Slovenia’s Monetary and Exchange Rate Framework
121
bank. In exchange, banks get unrestricted access to tolar liquidity based on a seven-day foreign currency swap standing facility (i.e., the swap rate is set administratively). One of the main differences in this type of intervention compared to the previous period is that the central bank is now permanently present in the foreign exchange market. The BoS resorted to the use of a short-term foreign currency swap standing facility instead of outright transactions in foreign currency because this instrument, in combination with prudential regulation (a so-called ‘liquidity ladder’), restricts credit expansion (i.e., restrains granting longterm loans on the basis of short-term securities). This option, which on the one hand contributes to monetary control, on the other has the drawback of hindering interbank activity as banks, instead of lending to each other in the interbank market, can simply sell foreign currency to BoS to obtain tolar liquidity (interbank rates are still negative in real terms since the middle of 1999). What is even more problematic is the piling up of foreign currency on seven-day swaps (of about 7 per cent of GDP in mid-2002), which creates enormous vulnerability in the case of a currency crisis. This is a possible reason for the BoS to offer longer maturity swaps (270 days) since the end of 2002. Originally the rate of swap or, more properly, the annualized rate of a seven-day-forward contract, was administratively set to give signals regarding the future dynamics on the exchange rate and simultaneously to contribute to closing the domestic–foreign interest rate differential. At the time of its introduction in 2001 the forward rate was set at a level consistent with the annualized monthly depreciation rate of the exchange rate (4.5 per cent). However, since the domestic–foreign inflation differential has remained wide, while interest rates have declined in the EU, this has created tensions between the objective of closing the interest rate differential – which requires a high depreciation rate – and the objective of lowering inflation – which requires a lower depreciation rate. As a consequence, the level of the annualized forward rate, which in principle should be the unbiased prediction of the future spot rate, has remained unchanged, and so has departed from the actual annualized exchange rate depreciation, which has declined. This difference between the swap rate (forward rate) and the actual depreciation rate of the exchange rate has become a de facto tax on all foreign currency transactions, particularly in the banking system. While this is appealing from the point of view of lowering the costs of monetary policy (or discouraging interest-sensitive capital inflows by contributing to close the interest differential), on the other hand it has a distortionary effect because, like other taxes, the implicit tax, when possible, is
122 Country-specific Monetary Policy and Exchange Rates
transferred to third parties (in this case, clients), either via lower deposit rates or via higher lending rates. Additionally, the de-coupling of the movements of the forward rate and the effective future spot rate resulting from the actual depreciation of the exchange rate means that they provide two conflicting indications of the future depreciation path of the exchange rate. The conflicting signals in a monetary policy framework that already lacks an explicit anchor for expectations can undermine the very objective it pursues, namely lowering inflation. Despite the policy changes introduced in 2001, the monetary framework remains unbalanced and vulnerable because it still relies basically on the exchange rate to simultaneously preserve internal and external equilibrium. Control of base money by means of sterilizing the liquidity created through the swap or forward standing facility is still targeted to ensure low costs of monetary policy, instead of conveying interest rate signals that should be transmitted throughout a (as yet non-existent) yield curve to the economy, to influence inflation expectations and domestic demand. Thus the monetary framework and the inflation objective are still vulnerable to unintended depreciation of the exchange rate, triggered by currency crisis or other shocks, which is magnified by the substantial amount of foreign currency accumulated in very short-term swaps. This framework, which has been designed to face the challenges of an open economy environment by relying on closing the interest rate differential with the ‘help’ of the implicit tax on foreign exchange transactions, can successfully deter interest-sensitive capital inflows, and in fact it has done so, by significantly reducing loans from abroad in 2001–2002, but it cannot cope with non-interest-rate-sensitive capital inflows, such as FDI. FDI inflows, to a large extent due to the relatively slow privatization agenda, have started to pour into the economy, creating an excess liquidity in the banking system that is not sterilized at market-clearing rates.15 In fact, a rate of sterilization of monetized foreign exchange inflows of around 60 per cent under a fixed interest rate since June 2002 (via 60-day and 270-day bills) could not, once again, deter the reverting upward trend of M3 in late 2002. This explains why, for the first time, a one-off offer of a 360-day bill at ‘acceptable’ market interest rates was made available to banks at the end of 2002: to mop up excess liquidity that otherwise could not have been sterilized at the previous non-market rates. A positive and critical development in 2002, which, has not yet been properly incorporated in the monetary policy, is that from July 2002 onwards financial instruments with maturity of less than one year are no longer indexed. This important development happened as
Slovenia’s Monetary and Exchange Rate Framework
123
a result of changes in the accounting standards concerning revalorization of capital. The event is of outmost importance because it has added enormous transparency to the financial system and to its products, at least to those with maturity up to one year. This unintended change has started to unclog the interest rate channel of monetary policy, with potential positive effects for conducting monetary policy and adding resilience to any monetary policy framework. In fact, the key central bank interest rate on 60-day bills, which was negative for the most part of 2000 and for the first quarter of 2002, turned gradually to positive by June 2002 and started unexpectedly to influence other short-term rates. In view of the importance of this policy development, monetary authorities should actively pursue de-indexation of instruments with maturity longer than one year, to continue building an interest rate transmission channel and to dispel persistent inflationary expectations. A well-functioning interest rate transmission channel would enhance transparency and the resilience of the monetary policy framework. Similarly, the overdue de-indexation of financial contracts would contribute to preparing financial institutions for the EU environment in which inflation risk is not fully hedged and where monetary policy is transmitted primarily through interest rates. While it is still premature to make an assessment concerning the effectiveness of the new exchange rate stabilization policy in terms of reducing inflation, it can be said that the trend of reducing inflation observed in 2002, in spite of shocks produced by changes in tax rates and increases in controlled prices, is clearly consistent with the declining depreciation path observed before the shocks in 1995 and 1999. This framework should lead to a similar reduction in inflation in the near future, even with higher-than-average increases in administered prices
Table 6.3 Contribution of controlled prices to inflation (%) (1997–2002)
1997 1998 1999 2000 2001 2002
Inflation
Ministry of Economy
IMAD
9.4 6.5 8.0 8.9 7.0 7.2
20.4 16.5 13.7 14.0 12.2 11.8
20.4 16.5 13.7 14.0 13.4 13.0
Source: Information from the Ministry of Economy and IMAD estimates.
124 Country-specific Monetary Policy and Exchange Rates
than those observed in 2002 (such as those experienced in 1997–98, which accompanied the reduction in inflation until 1999: see Table 6.3). For such an outcome three considerations are important: (i) the rate of exchange rate depreciation should reach similar and sustained lower levels as in 1997–98, which were lower than those observed in 2002; (ii) the monetary authority should resist the temptation to accommodate shocks via the exchange rate; and (iii) the monetary authority should proceed with de-indexation of financial contracts and stand ready to use interest rates to offset currency shocks in order not to repeat the 1995 and 1999 experiences. Could the inflation in 2002 have been lower than it actually was? The answer can be affirmative, given the current account surplus of 2002, which indicates a higher-than-necessary depreciation of the exchange rate. Since the rate of the crawling was not announced, this could have been implemented by reducing the currency depreciation rate. As to the issue of how and with what effects, the options were two: to reduce the depreciation path either slightly or considerably (e.g. via a discrete change in the path). The policy choice would probably have different impacts on the inflation rate and on the evolution of the real exchange rate. In view of evidence which suggests that the intensity of the pass-through has accelerated in time during the last two years (three to five months), it is likely that a considerable reduction in the rate of depreciation could rather have had larger immediate effects on lowering inflation, and would also have signalled a more decisive stance towards the reduction of inflation, therefore enhancing credibility. The impact of the alternative scenarios on the real exchange rate is more difficult to discern. Taking into account the empirical evidence of a high pass-through in both cases, the effect should lead to the same appreciation of the real exchange rate but with different dynamics. Nevertheless, a critical factor influencing the outcome is credibility. In the absence of an announced crawling rate and of another anchor for expectations, the only way to establish credibility in the policy is by a credible reduction of the depreciation rate, but even in this case it may not be easy. From this perspective the eventual real appreciation of the exchange rate typically observed in exchange-rate-based stabilization programmes should be lower under a sharp reduction of the speed of exchange rate depreciation than in the case of a gradual reduction, because of the absence of an anchor of expectations and the room for discretion. From this point of view, the current policy based on a gradual reduction of the unannounced crawling rate can be interpreted as temporary and reversible and can, as such, perpetuate inflation inertia.
Slovenia’s Monetary and Exchange Rate Framework
125
Another important issue to address is whether the Slovenian economy, given the persistence of a high single-digit inflation level and a central bank following a de facto ERBS, would experience the short-term expansionary effects of other exchange-rate-based stabilization programmes (Calvo and Vegh, 1999; Fischer et al., 2002). The discussion is not only relevant in terms of the potential danger inherent in such programmes – given the fact that ERM II provides a default exit if things go wrong: namely, joining ERM II instead of accepting a reversal in inflation – but particularly in terms of the expected developments in some key variables, especially inflation. A key issue in this regard is the timing of joining ERM II and the speed at which inflation will be brought down to an ‘acceptable’ level. This issue is particularly relevant taking into account possible delays in joining ERM II, based on arguments such as that a certain rate of inflation has not been reached yet or that the level of inflation achieved is still not stable. Despite the effectiveness of the ERBS in bringing down inflation, empirical evidence has been found that the inflation rate does not converge to the rate of depreciation. This behaviour is explained by the fact that inflation in the tradable goods sector slows at a much faster rate than in the non-tradable goods sector, explained by, among other factors, inflation inertia, built-in expectations and the productivity differential between the tradable and non-tradable sectors. Thus the very exchange rate strategy chosen by the BoS, compared with other EU candidate countries central banks’ strategies, implies that the inflation rate in Slovenia would remain higher than in other candidate countries. Therefore, a relatively higher inflation (or a lower speed of inflation convergence) in view of the timeframe for accession should not be used as an excuse to delay entry in ERM II, thus endangering the achievement of the major macroeconomic strategic goal of joining EMU as soon as possible. This consideration is even more relevant taking into account the behaviour of inflation differentials in existing EU member countries before and after joining the EMU, and given that Slovenia has achieved a similar degree of real convergence as some of the existing EU members. In fact, existing EMU countries reduced sharply the inflation rates before complying with the Maastricht criteria (1996), while afterwards the inflation rates diverged again (1998). Given that the chosen ERBS strategy in Slovenia, compared to other strategies followed in Central Europe, will deliver a slower pace of disinflation, it is also important for policy-makers to resist the temptation of turning to heterodox policy instruments, in particular the control of
126 Country-specific Monetary Policy and Exchange Rates
administered prices, since such an approach can eventually undo the very objective of the policy. It could hinder confidence, as it can be rightly anticipated that at a certain point in time prices under control would have to catch up with their normal dynamics, while in the meantime they would create losses, increase public sector debt and create fiscal risks. Given the lack of an explicit monetary policy anchor, probably a formal commitment to an early entry to ERM II immediately after joining EU would serve as a monetary anchor, as the monetary authorities would have to abide by the date of entry. This strategy requires a direct involvement of the government in setting the ERM II entry date to be a binding target for the central bank, to enhance credibility. Alternatively, the monetary authorities could speed up the reduction in inflation by changing the exchange rate regime as some other Central European Accession Countries have done, particularly given the historically supportive fiscal stance in Slovenia, with the collateral advantage of enabling an early entrance to the Euroarea, if a de facto compliance with the ERM II is recognized by the EU as formally satisfying the ERM II requirement.
5
Conclusions
This chapter identified three monetary and exchange rate policy regimes in Slovenia since independence in 1991: money-based stabilization policy (1991–95), price and real exchange rate stability dual targeting policy (1996–2001), and exchange-rate-based stabilization policy and accession to ERM II (2001–). The exchange rate regime, a de jure managed floating rate since independence, has been de facto modified in accordance with the main policy objectives that guided monetary policy in different periods. In the 1991–95 period the exchange rate regime can be characterized as a freely falling regime in which the exchange rate (nominal and real) followed the path depicted by the Dornbusch overshooting model. In 1996, after single-digit inflation was reached, the exchange rate regime shifted to a de facto passive crawling exchange rate regime. The last change in 2001 to a de facto active crawling exchange rate regime took place to simultaneously address the challenges of the impending membership of Slovenia in the EU and to tackle the problem of persistent inflation. The main characteristic of the implementation of monetary policy throughout the three different periods, which was termed ‘fear of paying’, is the preference of the monetary authorities to use non-market arrangements for pricing monetary policy instruments, including capital
Slovenia’s Monetary and Exchange Rate Framework
127
controls, in order to minimize costs of implementing monetary policy. This has resulted in a policy framework that is vulnerable to exchange rate shocks, as the interest rate channel of monetary policy still remains blocked. In particular, the lack of use of interest rates to defend the currency in 1995 and 1999 resulted in reversals of the disinflation trends that preceded both shocks. Given the critical importance of joining the Euroarea as soon as possible, the monetary authorities should pursue further de-indexation of financial contracts and stand ready to use interest rates to defend the currency, ensuring that the inflation reduction trend would not revert once more. An early entry to the EMU, in addition to the advantages brought by a single currency, will result in an enhanced and transparent monetary policy framework, able to cope with the challenges of a small open economy, which will eventually contribute to eliminating existing distortions when facing an open financial market. Shifting to a balanced conduct of monetary policy that also relies on the interest rate channel in a de-indexed economy environment would contribute to preparing the financial sector for the ECB policy-operating environment, where monetary policy is implemented and transmitted primarily through the interest rate, and not through the exchange rate. In addition, a de-indexed environment will also contribute to financial institutions’ learning to manage inflation risk. Also, the announcement of a self-imposed target entry date to ERM II at the earliest possible date by the government could serve as the missing anchor for monetary policy. In view of the strategy chosen, the authorities should be aware of the trade-off between the speed of convergence of inflation and an early joining of the ERM II. The experience of existing EU members with respect to inflation before and after Euroarea entry should be taken into account. From this perspective, authorities should resist the temptation of introducing additional anchors to the unannounced crawling exchange rate (like administered price controls), as this would result in accumulation of debt and fiscal risk and in an eventual reversal in the objective of reducing inflation.
Notes 1. This finding seems to challenge for Slovenia the view, valid for other ‘transition’ countries, of an apparent lack of links between the evolution of the nominal and real exchange rates at the beginning of the transition (Grafe and Wyplosz, 1997).
128 Country-specific Monetary Policy and Exchange Rates 2. The original ‘freely falling regime’ characterization of Reinhart and Rogoff (2002) applies to episodes in which 12-month inflation is above 40 per cent, or to currency crises marked by a transition from a fixed or quasi-fixed regime to a managed or independently floating regime, typically characterized by exchange rate overshooting. 3. According to Mencinger (2001) the initial nominal exchange rate was set at a level to match a similar level of the Yugoslav dinar real exchange rate of 1988. That enabled partial convertibility and the growth of foreign exchange reserves during 1998 and 1999. 4. Despite this, the interbank interest rate remained positive, reflecting a lower degree of distortions in the money market during this period. 5. The tolar indexation clause (TOM for ‘temeljna obrestna mera’) is the annual interest rate, calculated by BoS and used for preserving the value of financial liabilities and assets in domestic currency. It is calculated in the following way: from August 1995 to January 1996, average of the previous three months’ inflation (from June until August 1995 indexation was based on the average of the previous three months’ inflation, while from August 1995 it was based on so-called ‘R’, which was equal to the previous month’s inflation rate); from February 1996 to November 1996, four months, from December 1996, six months, and from May 1997, 12 months. Financial liabilities in domestic currency, with maturity less than 30 days, are not indexed from September 1995. Since July 2002 financial liabilities and assets in domestic currency, with maturity less than one year, are no longer indexed. Financial liabilities and assets in domestic currency, with maturity exceeding one year, are still TOM-indexed. 6. Capital controls were mostly dismantled in 1999, although restrictions on foreign portfolio investments of less than six months’ maturity continued until January 2002, when they were finally removed. 7. The terms and conditions of the 1997 agreement, which was amended by an annex to the agreement signed at the end of 1999, laid down mutual rights and obligations in the following cases: (1) intervention buying rate; (2) intervention selling rate; (3) determining maximum margin between buying and selling rates applied by the banks and (4) temporary purchase of foreign exchange; and (5) temporary sale of foreign exchange for the period of two months. The Bank of Slovenia remunerated the banks, signatories of the agreement, for opportunity costs associated with contractual obligations (BoS, 2001). 8. The share of controlled prices in the CPI basket reached its peak in 1995 (22.5 per cent), then it decreased to 20.4 per cent, 17.0 per cent, 14.3 per cent and 13.7 per cent in the years 1996 to 2000 respectively. The contribution of controlled prices to average inflation was 20.4 per cent in 1997, 16.5 per cent in 1998 and 13.7 per cent in 1999. In the last three years (2000–02) the contribution was 14.0 per cent, 13.4 per cent and 13.0 per cent respectively (IMAD, 2002). 9. Bofinger and Wollmershäuser (2001) distinguish between active and passive crawling peg regimes. 10. The use of interest rates to defend the currency would not have major fiscal implications, as for example those mentioned in Lahiri and Vegh (2000) because the government’s debt service is mainly indexed to inflation or the exchange rate, and the government fiscal stance is quite solid in terms of both
Slovenia’s Monetary and Exchange Rate Framework
11.
12.
13.
14. 15.
129
deficit and debt levels. Furthermore, defending the currency would have had a positive fiscal effect. The only rate that increased during 1999 was the 28-day repo rate, as a consequence of the strong demand for tolars. The increase responded to a liquidity crunch, as reflected in the movements of the interbank market rate, which become positive in real terms for the first time since the first quarter of 1998, and remained so for almost three quarters in 1999, when they became negative again. They remain negative. The full pass-through effect is estimated to happen in three quarters. It was uniformly distributed during the period 1996–99, reaching its peak in five to six months. In the period that followed, it seems to have peaked earlier, in three to five months, and even earlier in 2000. The M3 target set at the beginning of 2001 was overshot (23.9 per cent, well above the 11–17 per cent band), as well as the projection of M3 for 2002 (22.7 per cent, again well above the 12–18 per cent band). In part resulting from the change in policy, the volatility of base money has increased. Given the impact of FDI inflows on monetary conditions, there have been some statements favouring postponing privatization (of, for example, banks) until after Slovenia joins the EU.
7 Monetary Policy in Estonia: The Transmission Mechanism Urmas Sepp, Martti Randveer and Raoul Lättemäe
1
Introduction
The main aim of this chapter is to present a stylized review of the aspects that influence the monetary transmission mechanism (MTM) in Estonia. In addition, the different transmission channels are analysed with a small-scale macro-model. The MTM identifies the channels by which monetary policy measures, such as changes in money supply or central bank interest rates, influence domestic demand and inflation. The traditional view is that monetary policy affects liquidity and interest rates in financial markets, and these ultimately influence the real sector via their impact on consumption and investment decisions. Understanding those mechanisms is essential for understanding the impact of a monetary authority’s decisions and actions on the economy. However, there is no consensus about exactly how monetary policy transmits from central bank instruments to monetary policy targets. The reason is that the MTM consists of many different and partially overlapping channels that are difficult to separate empirically. In addition, unobservable expectations play an important role in private decisions, thus affecting the way equilibrium is reached in financial markets. Whereas there exists a huge literature on monetary transmission in general, research on the MTM under a currency board is rather limited. This chapter reports the results of research conducted at the Bank of Estonia that attempts to fill this gap, in the context of the Estonian currency board. In particular, this chapter reviews different aspects of the Estonian monetary system and transmission mechanism (see also Lättemäe et al., 2002; Sepp and Randveer, 2002b; Lättemäe, 2001). In addition, a small-scale macro-model that has been recently built to study 130
Monetary Policy and Transmission in Estonia
131
different aspects of the Estonian monetary regime is reviewed in full here (see Sepp and Randveer, 2002a, 2002b). Some additional conclusions are drawn from the monetary transmission model compiled by Pikkani (2001) and also from different versions of the Bank of Estonia’s macroeconomic models (Basdevant and Kaasik, 2002; Sepp et al., 1999). The first part of the chapter gives a short overview of the Estonian monetary regime – the fixed exchange rate that is supported by the currency board arrangement (CBA). The second part overviews some stylized aspects that can influence the monetary transmission process in Estonia. Those are the depth of the markets, the financial structure and the relationships between financial and real sectors. Additionally, a small empirical macro-model, used to analyse the transmission mechanism in Estonia, is presented in detail. Finally, the third part presents some empirical results, based on this model, as well as on prior research. The main aim of this final part is to explain the impact and relevance of the various transmission channels in the Estonian economy.
2
Main features of the Estonian monetary regime
2.1 Currency board arrangement For a decade, Estonian monetary policy has been based on a CBA. This is a specific type of fixed exchange rate arrangement, governed by strict rules. Those rules include a strong, explicit (usually written into law) commitment to a fixed exchange rate and the requirement that domestic currency is issued only against foreign exchange (IMF, 1996). Therefore, a CBA has to stand ready to exchange domestic currency for the reserve currency at a specified and fixed exchange rate, and all issued currency has to be backed by foreign reserves or gold. Those strict rules constrain the scope for issuing unbacked monetary liabilities, ensuring that the board does not run out of foreign reserves to maintain the parity (IMF, 1996). A CBA, therefore, sets clear constraints on an independent monetary policy. In addition, the currency board cannot extend credit to the fiscal authorities. The fiscal regime, therefore, is subordinated to the monetary regime, and a hard budget constraint is imposed on the politicians (Hanke, 2002). In Estonia, the CBA was introduced together with the monetary reform in 1992, after independence from the former Soviet Union. The so-called ‘Law on the security for the Estonian kroon’ sets the legal framework for the CBA principles: (i) Kroon exchange rate fixed with respect to the DEM;1 (ii) Prohibition to devalue the Estonian kroon;
132 Country-Specific Monetary Policy and Exchange Rates
(iii) Requirement for full backing of issued kroons (i.e. currency in circulation and deposits in the Bank of Estonia) with convertible foreign currencies and gold; (iv) Guarantee for full convertibility of kroons at the official rate; (v) Requirement that the Bank of Estonia issues kroons only against corresponding change in foreign reserves. As we can see, the ‘Law on the security for the Estonian kroon’ enforces the CBA in a quite orthodox way. However, the currency board is not separated institutionally from the central bank. According to the ‘Central Bank Law’, the Bank of Estonia is responsible for typical central bank duties, i.e. maintaining the stability of the legal tender, conducting banking and monetary policy (as long as it does not contradict the above-mentioned principles), promoting financial stability, implementing banking supervision (only until 2002), collecting statistics, conducting research, and so on. In our view, the institutional unification of the CBA principles and other central bank duties does not weaken the enforcement of the CBA principles in Estonia, as the most fundamental ones are guaranteed by the law. In implementing its tasks, the Bank of Estonia is independent of all governmental agencies. Any lending to the government by the Bank of Estonia is prohibited and the Bank is not liable for the state’s financial obligations. Although the scope for possible discretionary monetary policy is limited only by the amount of excess reserves of the central bank, the Bank of Estonia has followed the rather orthodox rules-based principles of the Estonian CBA in practice. Most of the time, the only channel for base money issuance are transactions at the central bank forex window, ensuring that money supply adjusts only according to changes in its foreign reserves (see Figure 7.1). The few rare exceptions, when the central bank has injected additional liquidity into financial sector, have taken place in specific situations, and were carried out to avoid excessive systemic risk in the Estonian financial system. So there is no active monetary policy in Estonia, but discretionary actions in extreme situations are not explicitly ruled out either. There is no central bank policy rate or other operational monetary policy targets in Estonia. Therefore, there exist neither discretionary monetary policy instruments (albeit during 1993–2000 the Bank of Estonia issued small amounts of central bank CDs) nor any explicit lender of last resort (LoLR) facility. The only relevant monetary instrument, in addition to the forex window, has been reserve requirement for banks. Those have been somewhat higher in Estonia (currently standing at
Monetary Policy and Transmission in Estonia
133
Bn EEK
16 14 12 10 8 6 4 2 0
Figure 7.1
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Net foreing assets Domestic liabilities CBA cover (NFA and domestic liabilities of Eesti Pank, bn EEK)
Source: Eesti Pank.
13 per cent), compared to other Central European, ‘transition’ countries, in order to promote stronger financial discipline, strengthen the financial sector and compensate for the lacking LoLR facility (see Section 2.2 for a longer discussion on the evolution of the Estonian monetary framework). This highlights an important characteristic of the Estonian monetary regime: all the necessary adjustments are left to the market, under the anchoring role of the exchange rate. Both market interest rates and liquidity adjust according to economic developments, external financing and arbitrage conditions. Consequently, the key role in achieving and maintaining sufficient liquidity buffers rests with the financial system itself. 2.2 Main developments of the monetary policy framework, 1992–2001 The CBA principles were laid down through the unlimited forex window facility in 1992 (the bid–ask spreads were set at 0.5 per cent). In addition, a compulsory (non-averaged, non-remunerated) reserve requirement of 10 per cent of the liabilities of the banking sector was established. The then unsophisticated financial system and the simplicity of the CBA features permitted maintaining the monetary policy framework largely unchanged during the following four years. The only important development was the introduction of the Bank of Estonia’s CDs in the spring of 1993: however, the main aim of this instrument was to foster the development of the money market, and not to move towards discretion. The first major set of reforms in the monetary policy operational framework was carried out in 1996, when the averaging of reserve
134 Country-Specific Monetary Policy and Exchange Rates
requirements was introduced, the spreads in the forex window on DEM transactions were abolished and remuneration of excess reserves was introduced (the latter is also called ‘the deposit facility’ in the Estonian framework). These steps aimed to improve liquidity management in the banking sector, as well as accommodate market-based principles in the framework. The next set of changes was carried out as early as 1997, caused by concerns about excessive monetary developments. More concretely, the increasing capital inflows, as well as financial sector overborrowing from foreign markets during 1996–97, fostered domestic demand, which led to a rapidly deteriorating current account deficit. As the CBA sets clear limits on using monetary tools for implementing restrictive policies, the ‘stabilization package’ contained not only an increase in the compulsory reserves, but also greater prudential capital adequacy requirements, and also the creation of a ‘Stabilisation Reserve Fund’ from fiscal surplus. In the monetary framework, this meant the widening of the reserve base and the increase in banks’ reserve balances with the Bank of Estonia from 10 to 13 per cent, through the introduction of those additional liquidity requirements. During 1999–2000 the concerns about possible distortions caused by relatively high and uncompensated reserve requirements, as well as the need to start preparing the operational convergence of the Estonian monetary framework towards the eurosystem, led to new major changes in the monetary framework, through the further development of the Bank of Estonia’s rules-based facilities. As a first step, the remuneration of required reserves was introduced in 1999. This aimed to reduce structural deviations from the eurosystem, as well as to signal that higher liquidity buffers in the financial system were still a prerequisite to cope with possible volatility. As a next major step among these changes, the partial fulfilling of reserve requirements with high-quality (at least with AA−/Aa3 credit rating), euro-denominated foreign securities of national governments of advanced Western countries and of supranational institutions was allowed in January 2001 (initially up to 25 per cent of the reserve requirement, and from July up to 50 per cent). That meant lowering the rate of required reserves with the central bank (in national currency), while simultaneously raising required reserves in high-rated liquid foreign assets (in anchor currency). Such treatment of eligible assets in fulfilling reserve requirements allows the banking system to use its liquid assets more efficiently, as well as ensuring that there are sufficient liquidity buffers in the financial system. This reform is also forward-looking, as
Monetary Policy and Transmission in Estonia
135
it induces the banking system to start accumulating and operating with the assets which form the basis of the eurosystem’s monetary operations (it is important to note that all permitted securities belong to so-called ‘Tier 1’ assets of the eurosystem).
3
The core features of the Estonian economy
The structure of the financial system and the linkages between the financial and real sectors are important aspects in evaluating how monetary signals affect the real economy. The first determines the transmission from monetary indicators carrying the signals to other financial sector variables, especially into longer-term retail rates and into assets prices. The second determines the relationship between financial sector variables and the spending decisions of households and firms. In this section we give an overview of the different structural aspects that influence the transmission mechanism in Estonia. In addition, we will present a small-scale macro-model that has been recently used in Sepp and Randveer (2002a, 2002b) to study different aspects of the Estonian monetary regime. Some additional conclusions are drawn from the monetary transmission model by Pikkani (2001), and also from different versions of the Bank of Estonia’s macro-models (Basdevant and Kaasik, 2002; Sepp et al., 1999). The model that is used here is a typical small-scale macro-model with the standard features – a Keynesian short run and a neoclassical long run. The model also contains price-setting rigidities (partly due to market imperfection, which allows staggered price setting) and open economy features. The model also incorporates relevant features of the Estonian economy. First of all, the Estonian economy is very open and small (see Table 7.1). Therefore it is strongly influenced by the external environment. This dependence is also reflected in the model – economic growth is, to a large extent, determined by foreign demand, and domestic inflation is a function of imported inflation. Second, as economic liberalization was already complete by the mid-1990s, the processes in the Estonian economy are essentially market-driven. Third, the development of the Estonian economy is strongly influenced by the convergence process towards the EU. This is evident in nominal (inflation, interest rates) and real variables (income level). Keeping in mind the specific nature of the Estonian monetary arrangement and its importance to the whole economic system (see Lättemäe et al., 2002), it was important to model the monetary system so as to incorporate the main elements of the CBA. However, in this
−9.0 89.8 1.3 69.6 29.3 −0.7 32.6 6.5 5.6
−14.2
1076
a Before 1996 figures also reflect loans to financial institutions.
2.2
55.4 19
1993
1992 −2.0 30 47.7 −7.2 75.3 34.8 +1.3 23.2 7.6 5.0
1994 4.6 32 29.0 −4.4 72.0 38.0 −1.3 17.2 9.7 5.2
1995
Selected indicators of the Estonian economy, 1992–2001
Real GDP growth (%) GDP PPP compared to EU average CPI Curr. acc. balance (% of GDP) Exports (% of GDP) Banking sector assets (% of GDP) Fiscal deficit (% of GDP) Interest rates of real sector loansa Unemployment (%) Central gov. debt (% of GDP)
Table 7.1
4.0 34 23.1 −9.2 67.1 43.8 −1.9 13.1 10.0 6.2
1996 10.4 37 11.2 −12.1 78.4 63.4 +2.2 11.9 9.7 5.2
1997 5.0 38 8.2 −9.2 79.7 55.7 −0.3 13.1 9.9 4.3
1998 −0.7 37 3.3 −4.7 77.2 61.7 −4.7 10.8 12.3 4.6
1999 6.9 38 4.0 −6.0 95.4 67.7 −0.4 9.5 13.7 3.2
2000
5.4 40 5.8 −6.2 91.7 72.0 +0.4 8.7 12.6 2.7
2001p
Monetary Policy and Transmission in Estonia
137
respect we made a typical simplification – the CBA is modelled as a completely credible fixed exchange rate regime. In this way we omitted the political and institutional considerations that in fact distinguish a CBA from a completely credible regime (in principle it is possible that there are nominal exchange rate changes even under a CBA: see Batiz and Sy, 2000). The model’s parameters are a result of a mixed procedure of statistical estimation (for 1995–2000) and the calibration of selected values. The equations are estimated individually. The final aim was to construct a macro-model whose simulations would adequately reproduce past developments of the Estonian economy, while, on the other hand, the behaviour of its variables (including transmission schemes) should also correspond to intuitive views of the inner workings of the Estonian economy.
3.1 Money markets and short-term interest rates The first link in the MTM – the transmission of monetary signals within the financial sector – is broadly determined by the depth of financial intermediation and by the structure of the financial sector. However, in Estonia, this question cannot be analysed in the narrow context of domestic financial markets alone. The Estonian financial sector shortterm borrowing and lending also include foreign exchange markets and foreign-market-related activities. According to the CBA principles, the Bank of Estonia offers to credit institutions unlimited forex purchase and sale facilities for all major currencies. This facility is the key element of the liquidity system. Moreover, as there are no bid–offer spreads for euro transactions, the credit institutions are free to move their liquidity across borders without transaction costs. Consequently, the most active market segment in short-term finance is the forex market, which in Estonia is more liquid than the domestic money market2 and so the transaction volumes are larger (see Figure 7.2). This particular set-up has linked the liquidity management of the Estonian financial sector to the much deeper euromoney and capital markets. This has been a deliberate policy, as it is impossible (or at least not cost-effective) to build up full-scale, sophisticated financial markets in such a small economy as Estonia. In addition, the financial sector in Estonia is highly concentrated, with a couple of larger institutions making up the majority of the sector, which also hinders the development of deep domestic money markets.
40 35 30 25 20 15 10 5 0 01
.9 04 7 .9 07 7 .9 10 7 .9 01 7 .9 04 8 .9 07 8 .9 10 8 .9 01 8 .9 04 9 .9 07 9 .9 10 9 .9 01 9 .0 04 0 .0 07 0 .0 10 0 .0 01 0 .0 04 1 .0 07 1 . 10 01 .0 01 1 .0 04 2 . 07 02 .0 2
Bn EEK
138 Country-Specific Monetary Policy and Exchange Rates
Turnover of central bank‘s forex window Turnover of Estonian kroon money market Turnover of Estonian forex market (spot and forward, all sectors) Intrebank forex forward and swap market
Figure 7.2
Volume of EEK money market, CB’s forex window and market (bn EEK)
01 . 04 96 . 07 96 . 10 96 . 01 96 . 04 97 . 07 97 . 10 97 . 01 97 . 04 98 . 07 98 . 10 98 . 01 98 . 04 99 . 07 99 . 10 99 . 01 99 . 04 00 . 07 00 . 10 00 . 01 00 . 04 01 . 07 01 . 10 01 . 01 01 . 04 02 . 07 02 .0 2
20 18 16 14 12 10 8 6 4 2 0
3M DEM LIBOR/3M EURIBOR
Figure 7.3
3M TALIBOR
Money market rates in Estonia and in the Euroarea
Source: Eesti Pank.
Domestic money markets in Estonia are therefore rather small and closely linked to foreign markets. As under a CBA, monetary policy signals are external to the system (and not set by the central bank), and the role of foreign monetary signals is dominant. This is clearly shown by the Estonian interbank money market rates, as they closely follow the EURIBOR rates, except during the turbulent 1997–98 period (see Figure 7.3).3 The dependence on foreign financial markets allows us to model the short-term interest rates as a UIP condition with a risk premium (see
Monetary Policy and Transmission in Estonia
139
equation 7.1): d(is ) = α1 ECM1,t−1 + α2 d(rm2) + α3 (i∗ )
+ β1 α4 d Et−1,t+j−1 (e)FW ,
(7.1)
where ECM1 = (is − i∗ + f1 (t)); α1 , f1 (t) < 0; α2 , α3 > 0; β1 = 1, if d(Et−1,t+j−1 (e)FW ) > 2SDFW , otherwise β1 = 0.4 The dynamics of the short-term interest rates (is ) is split into its shortrun components and a long-term convergence term, which are linked by the error correction mechanism (ECM). The short-run component is specified by the UIP (α3 d(i∗ ) + β1 α4 d(Et−1,t+j−1 (e)FW )), where eFW is the forward nominal exchange rate. In addition, the short-run dynamics of the interest rates is influenced by the money supply (rm2). The inclusion of this in equation (7.1) is based on the conventional specification of the LM curve (were money supply is negatively related to the interest rate). The long-run component (i∗ + f1 (t)) is equal to the foreign interest rate i∗ plus the domestic risk premium. In the model, the foreign interest rate is specified as three-month EURIBOR. The risk premium is approximated by a declining function of time. This approximation is not a standard one, as the representation of the risk premium as a random walk would be more conventional. However, our approach is supported by developments in the Estonian money market: during the last ten years the domestic risk premium has steadily decreased (Pikkani, 2001), which marks the progress in stabilization, transition to a market economy and accession to the EU. As alternative proxies, real sector indebtedness and the share of foreign ownership in the financial sector are used in some models as indicators for the long-run risk premium in Estonia. It is important to note that the currency risk component of the domestic risk premium is typically lower under CBA, as the arrangement is by definition ‘the ultimate fix’. On the other hand, this proposition is valid only when market participants believe in the arrangement. If the credibility in macroeconomic policies is low, then the domestic risk premium can still be high and volatile. Moreover, when markets are not perfect, domestic liquidity conditions and term structure, as well as regional or global risks, can affect the domestic risk premium or imply devaluation expectations. For example, during the Asian and Russian crises, the interest rate equation above was conditional on the speculative pressure.5 The speculative pressure against the kroon in 1997–98 reflected a growing uncertainty among foreign investors with regard to the sustainability of
140 Country-Specific Monetary Policy and Exchange Rates
01
.9 04 7 .9 07 7 .9 10 7 .9 01 7 .9 04 8 .9 07 8 .9 10 8 .9 01 8 .9 04 9 .9 07 9 .9 10 9 .9 01 9 .0 04 0 .0 07 0 .0 10 0 .0 01 0 .0 04 1 .0 07 1 .0 10 1 .0 01 1 . 04 02 . 07 02 .0 2
20 18 16 14 12 10 8 6 4 2 0
3M DEM LIBOR/3M EURIBOR 3M interbank rate (Hansabank) 3M EUR/EEK (3M DEM/EEK) Forward difference (Hansabank) Figure 7.4
Forex forward difference and interbank rates (Hansabank quotes)
Source: Reuters.
the fixed exchange rate policy. This was caused by the deepening uncertainty about emerging markets in general, and by the economic situation in Estonia in particular. The corresponding increase in Estonian interest rates was an automatic response of the CBA to speculative attacks and deteriorating capital flows, as the monetary authority did not intervene in the money or forex markets. Therefore, the interest rate in equation (7.1) is conditional on the speculative pressure. The devaluation expectations in the model are approximated by the forex market forward points, which show the difference between the spot and the forward exchange rates. Under ordinary conditions (i.e., with a forward point difference below two standard deviations, SDs), the exchange rate expectations do not have a direct impact on the interest rate in the equation, so β1 = 0. During the speculative attack, the devaluation expectations have a definite effect on the interest rate and β1 = 1. In the model, the structure of the – forward-looking – exchange rate expectations follows the logic of the so-called ‘first-generation’ models of speculative attacks (see equation 7.2). In those models, countries suffer attacks when they run unsustainable monetary and fiscal policies (Peria, 2002). Therefore, it is assumed that a significant decrease in foreign demand worsens the current account balance and may cause a balance of payments problem. This, in turn, increases the possibility that the authorities may have to devalue, in order to improve the current account balance. More specifically, the expected changes in the foreign effective demand6 (Et,t+j (d X )) determine the devaluation expectations in the
Monetary Policy and Transmission in Estonia
141
model. In addition, exchange rate expectations also include some inertia, as lagged expectations (Et−1,t (e)FW ) are also included in this equation. ⎞ ⎛ α5 Et,t+j (d X ) + α6 Et−i,t (e)fw ⎠ Et,t+1 (e)FW = 2 + β2 ⎝ j=0,1
⎛
+(1 + β2 ) ⎝
⎞ X
α7 Et,t+j (d )α8 Et−i,t (e)
fw ⎠
,
(7.2)
j=0,1
where α5 , α7 < 0; α6 , α8 > 0; β2 = 1, if Et,t+1 (d X ) < − SDdX , otherwise β2 = 0. Similarly to the interest rate equation (7.1), the specification of the exchange rate expectations is conditional. In accordance with the empirics, the break-point for switching the nature of expectations is a decrease in effective foreign demand by more than one standard deviation of that variable, SDdX . In this case β2 = 1 and α6 ≈ α8 +1. In this case, the impact of the change in external demand on the exchange rate expectations is considerable. If the drop in foreign demand is smaller, then β2 = 0, and the impact is minor. 3.2 Foreign capital flows and money supply Estonian banks have been active in international markets since the early times of the CBA. Initially, those activities covered mainly depositing (i.e., creating foreign assets as additional liquidity buffer, given the missing LoLR facility: Lepik, 1999). Since 1995–96, borrowing by banks from foreign capital markets increased significantly, as markets’ risk perceptions of Estonia lowered.7 An early full liberalization of capital flows (in 1994) not only provided external competition, but also speeded up the integration of the banking sector with the international financial system. If we exclude the turbulent periods from the Asian and Russian crises in 1997–98, the banks’ foreign borrowing has become a clear substitute for domestic deposits, when the domestic credit demand is high (see Figure 7.5). The opposite is also possible – banks’ foreign reserves and foreign investments (i.e., NFA) are the main way to absorb excess liquidity, when growth in deposits exceeds credit demand. This is exactly in line with a CBA’s adjustment mechanism, which ensures that market forces determine and limit the expansion of the money supply through foreign capital flows (Hanke and Schuler, 1994). To characterize those CBA features, money supply is modelled as demand and capital flows determined (equation 7.3). First, it is assumed
% 75
22.5
60
18.0
45
13.5
30
9.0
15
4.5 0.0
–15
–4.5
–30
–9.0
01 . 05 94 . 09 94 . 01 94 . 05 95 . 09 95 . 01 95 . 05 96 . 09 96 .9 01 6 . 05 97 . 09 97 . 01 97 . 05 98 . 09 98 . 01 98 . 05 99 . 09 99 . 01 99 . 05 00 . 09 00 . 01 00 . 05 01 . 09 01 . 01 01 . 05 02 .0 2
0
Bn EEK
142 Country-Specific Monetary Policy and Exchange Rates
Domestic assets (annual growth, left axis) Domestic liabilities (annual growth, left axis) Net foreign assets (annual change in billions of EEK, right axis)
Figure 7.5 Annual growth of Estonian banking sector domestic assets and liabilities vs annual change in banking sector NFA
that money demand is mainly driven by the transaction motive and money supply is dependent on GDP ( y).8 The trend f2 (t) in the long-run equation reflects a rate of ‘natural’ financial deepening. d(rm2) = α9 ECM2,t−1 + d(y) + α10 d(MG(−1)),
(7.3)
where ECM2 = rm2 − (y + f2 (t)); f2 (t) > 0; α9 < 0; MG = i − i∗ − Et,t+j (e)FW . In addition, the money supply in equation (7.3) is dependent on devaluation expectations and interest rate arbitrage conditions. If the interest rate differential (i.e. the difference between the domestic and foreign interest rates) is higher than the devaluation expectations (MG > 0), then, according to the interest rate arbitrage condition, there will be a capital inflow and increase in money supply. In the opposite case (MG < 0), which is typical for the speculative pressure, the capital outflow reduces the money supply. This is in line with the above-mentioned adjustment mechanism, which ensures that the arbitrage conditions and foreign capital flows determine changes in the money supply.
3.3 Longer-term interest rates The developments in the financial sector highlight that foreign interest rates signals transmit widely into Estonian retail rates (see Figure 7.6).
Monetary Policy and Transmission in Estonia
143
% 20 18 16 14 12 10 8 6 4 2 0
12 .9 03 6 .9 06 7 .9 09 7 .9 12 7 .9 03 7 .9 06 8 .9 09 8 .9 12 8 .9 03 8 .9 06 9 .9 09 9 .9 12 9 .9 03 9 .0 06 0 .0 09 0 .0 12 0 .0 03 0 .0 06 1 .0 09 1 .0 12 1 .0 03 1 .0 2
% 20 18 16 14 12 10 8 6 4 2 0
TALIBOR 3M EURIBOR 3M Figure 7.6
Corporate bank loans, long-term Households’ bank loans, loans-term
Short-term interest rates and retail rates, Estonia
Source: Eesti Pank.
Moreover, this effect has become increasingly evident since 1999–2000. This trend is consistent with the role of foreign markets and foreign capital flows under the Estonian CBA. However, one of the most critical questions in the Estonian transmission process lies in the relative role of domestic and foreign money markets in determining the retail rates in Estonia. That is, it determines to what extent the transmission from foreign rates into longer-term interest rates takes place conventionally through the domestic money market or directly from foreign markets to retail rates. This question arises because foreign money markets are more important for liquidity management in Estonia than domestic money markets. The direct impact from foreign interest rates into banks’ retail rates is presumably also due to the close substitutability between domestic retail deposits and foreign borrowing as funds for banks’ lending. On the other hand, the short-run interest rate arbitrage evidently takes place via the money market, whereas the arbitrage between longer-term interest rates is less clear – the capitalization of the bond market is too low in Estonia to allow extensive arbitrage through longer maturities in the bond market.9 In addition, the small size of the domestic financial markets and lack of government securities evidently translate into the lack of a typical yield curve and well-established term structure – the relevant maturities
144 Country-Specific Monetary Policy and Exchange Rates
in Estonian money markets are short-term (less than three months) and there is practically no trading in longer maturities. Hence the direct transmission from domestic money market rates to other domestic financial prices is difficult to capture, as the information chain can be slightly different from the traditional one. A factor that adds to these difficulties is the absence of a government securities market, which usually provides a benchmark for the domestic yield curve. Partly due to restrictions imposed by the currency board, but even more due to successful economic reforms, Estonian fiscal deficits and government lending over the last ten years have remained modest by European standards. Consequently, there has been no need to develop a domestic government bond market, and against this background a government securities market would probably stay thin. Given the key role of the financial sector in financial intermediation, a representative yield curve may, for example, be derived from bank lending rates (see Bank of Estonia, 2001). Empirical analysis shows that the Estonian credit curve dynamics (e.g. the changes in the spread of long- and short-term rates), particularly on a sector level, include some information about ex ante developments in the real sector. However, the interpretation of these ‘credit curve’ signals is somewhat complicated, as shifts in the credit curve are subject to a variety of demand- and supply-related factors (e.g. changes in risk assessments, cost of funds, competition, etc.). In spite of this, we have modelled longer-term interest rates in a conventional way, with long-term rates determined by the domestic short-term interest rates (see Fuhrer and Moore, 1995). As an alternative, it is possible to reduce the model to one interest rate, as was done in prior models of monetary transmission in Estonia: Pikkani (2001) used only the general average retail interest rate, and assumed that the transmission from foreign rates into Estonian retail rates takes place rather directly into the entire spectrum of Estonian interest rates. Here, a distinction between short-term and long-term rates was assumed in order to reflect shocks caused by speculative pressures and to separate the particular impacts of different interest rates on demand. In addition to the mechanisms referred to, foreign rates have considerable impact on real sector credit rates, due to the extensive use of interest rate indexing in Estonia (see Figure 7.7). The long-term retail credit rates are usually linked to the six-month EURIBOR or six-month DEM LIBOR, as the latter reflects changes in the price of financial sector credit resources. Thus, at least to some extent, foreign interest rate signals are automatically passed into the Estonian real sector.
Monetary Policy and Transmission in Estonia
145
12 .9 6 03 .9 06 7 .9 09 7 .9 12 7 .9 03 7 .9 06 8 .9 09 8 .9 12 8 .9 03 8 .9 06 9 .9 09 9 .9 12 9 .9 03 9 .0 06 0 .0 09 0 .0 12 0 .0 03 0 .0 06 1 .0 09 1 .0 12 1 .0 03 1 .0 06 2 .0 2
% 100 90 80 70 60 50 40 30 20 10 0
Households
Figure 7.7 Estonia
Enterprises
Estimated share of indexed loans in long-term loans to real sector,
% 120 110 100 90 80 70 60 50
Figure 7.8
2
1
.0 05
1
.0 09
0
.0 05
0
.0 09
9
.0 05
9
.9 09
.9
8
Exports/GDP
05
8
.9 09
7
.9 05
7
.9 09
6
.9 05
6
.9 09
5
.9 05
5
.9 09
4
.9 05
4
.9 09
3
.9 05
.9 09
05
.9
3
40
Imports/GDP
Openness of the Estonian economy
3.4 Aggregate demand and supply As the Estonian economy is very open and small, real sector developments are strongly influenced by the external environment. Exports represent more than 100 per cent of GDP and are the basis of Estonian economic growth (see Figure 7.8). This dependence is also reflected in the model, as economic growth is to a large extent determined by foreign effective demand. The equation for domestic demand, which is equal to GDP (y), is a modification of a traditional IS curve for the open economy, where domestic demand is
146 Country-Specific Monetary Policy and Exchange Rates
driven by exports (rx, as a proxy for income from export activities) both in the short and in the long-run (see equation 7.4). Similar elements are applied on the IS curve on all the Bank of Estonia’s different macromodels (see Basdevant and Kaasik, 2002; Sepp et al., 1999). long
d(y) = α11 ECM3,t−1 + α12 d(rx) + α13 is,t−2 + α14 it−3 ,
(7.4)
where ECM3 = y − (α15 rx + α16 ); α11 , α13 , α14 < 0; α12 , α15 , α16 > 0. The dynamic part of the equation includes an error correction mechanism as well as the lagged impact of short- and long-term interest rates (correspondingly is and ilong ).10 The equation for exports is an extension of an imperfect substitutes model (see equation 7.5). The short-run dynamics of exports is modelled by the changes in effective nominal exchange rate (NEER) and external income, which is approximated by Finnish GDP (y FIN ). The ECM in the export specification reflects convergence to the long-run equilibrium. d(rx) = α17 ECM4,t−1 + α18 d y FIN + α19 d(NEER),
(7.5)
where ECM4 = rx − y EU + α20 rk + α21 REER ; α17 , α21 < 0; α18 , α19 , α20 > 0. The use of Finnish GDP as a proxy for external demand in the short run is due to the fact that Finland is, by far, the most important trading partner for Estonia (see Figure 7.9). The proximity and strong trade relations are main reasons for the quick and dynamic response of Estonian exports to Finnish demand.
01
.9 04 6 .9 07 6 .9 10 6 .9 01 6 .9 04 7 .9 07 7 .9 10 7 . 01 97 .9 04 8 .9 07 8 .9 10 8 .9 01 8 .9 04 9 .9 07 9 .9 10 9 . 01 99 .0 04 0 .0 07 0 .0 10 0 .0 01 0 .0 04 1 .0 07 1 .0 10 1 .0 01 1 .0 04 2 .0 2
% 70 60 50 40 30 20 10 0 –10 –20 –30
Estonia exports
Figure 7.9
Finland imports
EMU imports
Annual growth of Estonian exports and Finnish and EMU imports
Monetary Policy and Transmission in Estonia
147
This argument is supported by Kaasik et al. (2002), who found that the relatively high correlation between the business cycles in Finland and in Estonia (higher than the correlation between Estonia and the EU) is the result of close trade relations. Therefore, in the long run foreign demand is reflected by the EU aggregate, while in the short run the demand dynamics are approximated by the Finnish GDP. The long-run relationship of equation (7.5) includes both demand-side and supply-side factors. The main supply-side variable in the equation, affecting export growth, is the capital stock (rk), which is dependent on investments. This marks the fact that, in the long run, exports are constrained by the production and technological capacity of the economy. In addition, FDI in the export sector has generated positive externalities (e.g. technological spillovers, managerial and marketing know-how etc.) and increased the competitiveness of the export sector (and the economy). FDI has also helped in reorienting exports from the unstable CIS countries to Western markets, underpinning sustainable growth. On the other hand, exports are also determined in the long run by demand-side factors, namely by the real effective exchange rate (REER) and by EU demand. The REER is here as a proxy for price competitiveness – REER appreciation means a decline in competitiveness. This is consistent with Randveer and Rell (2002), who show that the REER is an influential determinant of Estonian exports. From the income side, the GDP of EU (y EU ) is used as the export demand factor, as nearly two-thirds of Estonian exports are directed into the European Union. Similarly to the export, the specification of the import equation (7.6) is in line with the conventional set-up. The basis for the specification is the imperfect substitutes model: d(rm) = α22 ECM5,t−1 + α23 d(y),
(7.6)
where ECM5 = r m − (α24 y + α25 REER); α22 > 0; α23 , α24 , α25 > 0. Accordingly, imports (r m) are determined by the income (proxied by the GDP, y) and by the REER. The REER impact contains two aspects. First and indirectly, the real appreciation, in the case of a fixed exchange rate in a developing economy, often reflects an increase in income and the subsequent increase in consumption. It is reasonable to assume that the income growth will shift the demand to more sophisticated goods. In these segments, imports have a strong market position, and as a result, the growth of imports outpaces the rise of income. Second, the real appreciation produces a substitution effect. A real appreciation reduces the price competitiveness of the domestic goods. As a result, imports will
148 Country-Specific Monetary Policy and Exchange Rates
replace part of them. However, in practice the magnitude of substitution is restricted. Due to the smallness of the Estonian economy, the range of goods that are produced for the domestic market is limited, and thus the possibility for such substitution is constrained. In equation (7.6), the income elasticity in the long run is smaller than the income effect in the short run. This effect is related to the cyclical development of the economy, as well as to the fact that, due to subcontracting, a significant proportion of Estonian exports is very importintensive. This means that an export-led GDP increase also sharply increases imports. Additionally, a rise in GDP is usually accompanied by an increase in bank lending, which tends to fuel investment-related imports. The joint effect of these factors is that the variation of imports is larger than that of income. The main representation of the supply side is given by a Cobb–Douglas production function (7.7) with conventional properties, which has been drawn from Rõõm (2001): y s = α26 + α27 (l + h) + (1 − α27 )rk + f3 (t),
(7.7)
where α26 , α27 , f3 (t) > 0. Rõõm (2001) shows that such a Cobb–Douglas production function reflects the supply side of the Estonian economy quite adequately. The use of a Cobb–Douglas production function is also justified on the grounds of its simplicity, which is crucial in the case of short time series. According to equation (7.7) output is explained by three exogenous processes: employment (l), a restructuring parameter (h),11 and the Hicks-neutral technical progress (f3 (t)). The only endogenous explanatory variable for the supply is capital (rk), which is dependent on investment. Investments (rn, see equation 7.8), on the other hand, are positively related to the output and the growth rate of the stock of loans: d(rn) = α28 ECM6,t−1 + α29 d 2 (rl),
(7.8)
where ECM6 = rn − (α30 d(rl) + α31 y + α32 ), α28 < 0; α29 , α30 , α31 > 0. The output level (y) indicates the range of investment possibilities, while the stock of loans (rl) reflects the availability of resources for financing investment projects. In addition, the changes in output are positively related to expectations – a rise in expected growth in turn increases investments. Therefore, the change in the stock of loans in the model has both a short-term as well as a long-term impact on the economy.
Monetary Policy and Transmission in Estonia
149
3.5 Inflation As Estonia is a small and open economy, its price level is strongly influenced by import prices. It is important to note that inflation will remain somewhat higher in Estonia as compared to the advanced economies. The main reason for this is higher productivity growth in Estonia due to the real convergence, which also yields convergence of the structure and price level. This effect is believed to cause a one and a half to two and a half percentage points differential, when compared to inflation in advanced economies (Randveer, 2000). The import deflator is the main channel for the transmission of the external price signals in the model, and import prices are predominantly relevant in domestic price formation. The transmission scheme is simple – the import price deflator, which is dependent on foreign prices as well as on the exchange rate, influences the producer price inflation (due to the imported production inputs). The latter affects the tradables inflation and thus the CPI. Formally the CPI is a weighted average of tradables and non-tradables inflation. Demand pressures enter into the model through the output gap from the producer price inflation. The import price deflator (π M in equation 7.9) is is a function of NEER changes and income convergence (y R ).12 Import prices are set following a pricing-to-market behaviour in the model: d π M = α33 ECM7,t−1 + α34 d NEER ,
(7.9)
where ECM7 = π M − (y R + NEER + α35 ); α33 < 0; α34 , α35 > 0. Prices depend on the income level and on income convergence. On the other hand, the long-run import deflator depends on the exchange rate pass-through, which is also in line with conventional belief. The coefficients for exchange rate pass-through are roughly in line with Campa and Goldberg’s (2000) study of a large sample of OECD countries. They report about a 60 per cent pass-through in the short run and over 80 per cent in the long run, whereas the elasticities in our model are about 50 and 100 respectively. Slightly higher pass-through can be explained, given the higher share of energy and raw materials in Estonian imports, as imported energy and materials have pass-through elasticities close to one. Also Calvo and Reinhard (2000) claim that in ‘transition’ economies the pass-through from the exchange rate to inflation is generally higher than in more developed countries. Due to the openness, import prices determine producer price inflation (PPI) in Estonia (π R , see equation 7.10). In addition, Sepp and
150 Country-Specific Monetary Policy and Exchange Rates
Rell (2001) showed that producer prices in Estonia are primarily influenced by supply-side behaviour. Also, producer prices are partially driven by demand, which is proxied by the lagged output gap (ˆy ): d π P = α36 ECM8,t−1 + α37 Et,t+j d π M + α38 yˆ t−1 ,
(7.10)
j=0
where ECM8 = π P − π M ; α36 < 0; α37 , α38 > 0. This equation, taken from Sepp and Rell (2001), is basically a ‘new hybrid Phillips curve’. The import prices affect producer price via expectations (Et,t+i (d(π M ))). Producers tend to slightly overreact to expected imported inflation, as the estimated value for j=0 α37 = 1.08 > 1, i.e., a marginal inflation bias. The long-run relation here is determined by import prices: a fast convergence towards the long run in the equation obviously dampens the imported inflation bias. The import and producer prices determine the tradables inflation (π TR , see equation 7.11). This set-up is also consistent with Sepp and Rell (2001): d π TR = α39 ECM9,t−1 + α40 d π M + (1 − α40 )d π P ,
(7.11)
where ECM9,t−1 = π TR − α41 π P ; α39 < 0; α40 , α41 > 0. Tradables prices are based on producer prices in the long run. This reflects the supply-side price formation, typical for imperfect markets. The significance of the supply side is underlined by the fact that the tradables inflation is higher than producer price inflation, as α27 > 1. In the short run the tradables inflation is a weighted average of the PPI and import prices deflator. The interpretation of the PPI’s role is the same in the long run: the import deflator represents the direct contribution of imports due to their share in the consumption bundle. Due to technical reasons, the non-tradables inflation (π NT, equation 7.12) reflects first of all the impact of administrative price changes (dummy DA ). The reason for such an impact is obvious, as regulated prices form the major part of the non-tradables: d π NT = α42 DA + f4 (t),
(7.12)
where α42 > 0, f4 (t) < 0. The only way to bring the prices of the non-tradables into line with the common price level (or with the underlying non-tradable inflation
Monetary Policy and Transmission in Estonia
151
f3 (t))13 is to apply administrative action. Although this process is in fact an error correction process, it cannot be included to the model as an econometric component. This is because the correction of ‘errors’, or occurrence of administrative measures, depends on the political circumstances (among others, on the election cycle). In some cases the error correction might be significantly delayed. Therefore the timing of the administrative ‘error correction’ process is in principle different from the logic of an ECM, according to which the deviations in the previous period are adjusted.
4
The MTM in Estonia
The monetary policy transmission channels are typically divided into three groups: (i) interest rate, (ii) asset price (i.e. exchange rate and equity prices) and (iii) credit channels.14 Our model incorporates three channels: interest rate, credit and exchange rate. The model allows for several simplifications regarding these, which will be explained below. This is mostly due to the fact that the previous work on the monetary transmission channels in Estonia is rather limited, and we were unable to estimate several transmission channels. To explain transmission, we study exogenous shocks to variables that work through concrete channels. We are interested in the impacts of the transmission processes as well as in the comparative effectiveness of the different channels. The size of the studied shocks was one standard deviation of first-differenced series for the period 1995:1–2000:4. 4.1 The interest rate channel The traditional interest rate channel is supposed to work through real interest rates, influencing consumers’ and investors’ decisions. However, extracting information from market yields can be extremely difficult when the financial markets are shallow and the usual yield-curve instruments are non-existent, as in Estonia. Moreover, the determination of real interest rates in a ‘transition’ country with a high and volatile inflation is a complicated issue, as the link between inflation expectations and actual inflation is obscure and changing. Therefore, it is empirically rather difficult to distinguish the explicit effects of real rates on economic activity. In addition, until 1997, inflation substantially exceeded the level of nominal interest rates, so that ex post real interest rates were clearly negative. Assuming that inflation expectations are in line with
152 Country-Specific Monetary Policy and Exchange Rates
realized inflation, one would conclude that the ex ante real rates were also negative. On the other hand, an estimate using a structural VAR (see St-Amant, 1996) shows that most of the interest rate volatility is caused by nominal variation in expectations, whereas real interest rates have been very stable during the 90s (Lättemäe, 2002). Such outcomes make modelling a plausible link between interest rate movements and real sector behaviour rather difficult. Here it is important to notice that, considering the behaviour of inflation expectations, there is only slight evidence of the existence of forward-looking ‘rationality’ in Estonian inflation expectations (Sepp and Rell, 2001). This type of ‘rationality’ is only found in the corporate sector, whereas Sepp and Rell (2001) did not find any evidence of a rational, forward-looking behaviour in household, survey-based, inflation expectations. According to them, backward-looking future inflation rate expectations described households’ expectations best, whereas the actual headline inflation declined from almost 80 per cent to below 10 per cent. According to our model simulations, the effect of the interest rate channel is conventional. A rise in short-term rate starts a usual Keynesian-type restrictive transmission (iS ↑ → y↓ → yˆ ↓: see Figure 7.10). In addition to the effect on domestic demand there is also a marginal supply-side effect in the model, which is indicated by the decline of exports. This effect exists as capital stock (and therefore investments) is one explanatory variable for exports in the long run, and investments are negatively related to interest rate. Exports converge to the pre-shock
0.0025
0.1 0.05 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
–0.05 –0.1
–0.0025
–0.15 –0.2
–0.005
–0.25 –0.3
–0.0075 y
Figure 7.10
p
rx
tsb
The interest rate channel: adjustments after the shock, Estonia
Notes: π and tsb: change from base; y and rx as: percentage change from base; y and π on the left axes and tsb and rx on the right axes.15
Monetary Policy and Transmission in Estonia
153
trajectory as the impact of the error correction process begins to prevail in later stages. As can also be seen from Figure 7.10, the disinflationary effect of the interest rate shock is limited. This is due to the fact that in the model, inflation depends predominantly on import prices, which are determined by the producer currency pricing (and thus by the changes in NEER). The interest rate also has a marginal impact on the foreign trade balance, due to REER changes. The effect on other variables is insignificant. These results are also broadly in line with prior findings (see Pikkani, 2001). However, these results should be treated with some caution, as the model is based on historical figures. In addition, there have been several structural shifts during the sample period that may make the results less plausible for today’s environment. Due to the continuous increase in financial intermediation, as well as to the increased relevance of the euro in Estonians trade flows, the importance of the interest rate channel has grown during recent years. One must also note that the model does not incorporate possible indirect impacts from foreign interest rates. As, again, Estonia is a very small and open economy, its economic growth is mainly export-driven. Changes in foreign interest rates, therefore, may have strong indirect impacts through changes in external demand. It is also important to note that the impacts of the shocks to foreign and domestic interest rates show, to some extent, different adjustment paths in our estimations. The reason is the different dynamics of the interest rate margin ( MG). In the case of a domestic shock, MG will change; a move in foreign interest rates, however, does not affect it significantly. A change in MG causes interest rate arbitrage and, consequently, capital in- or outflow, with corresponding moves in the money supply (see Section 3). However, the adjustment to foreign and domestic shocks is not fundamentally diverse.
4.2 The credit channel The credit channel emphasizes informational problems in financial markets. Credit channel effects can appear as a change in the external finance premium or as a change in credit availability for the real sector. There are two basic transmission channels that arise as a result of informational problems – the balance-sheet and the bank-lending channels (Mishkin, 1996). The balance-sheet channel works due to the effects of monetary policy on firms’ and households’ balance sheets, which cause moral hazard and adverse selection problems in the retail
70
% 35
60
30
50
25
40
20
30
15
20
10
10
5
0
0
06 .9 7 09 .9 7 12 .9 7 03 .9 8 06 .9 8 09 .9 8 12 .9 8 03 .9 9 06 .9 9 09 .9 9 12 .9 9 03 .0 0 06 .0 0 09 .0 0 12 .0 0 03 .0 1 06 .0 1 09 .0 1 12 .0 1 03 .0 2
Bn EEK
154 Country-Specific Monetary Policy and Exchange Rates
* 2002-I – Preliminary
Figure 7.11
Borrowing from foreign sector (excl. FDI-related borrowing) Leasing and factoring Borrowing from banking sector Year-on-year growth (right scale)
The structure of real sector borrowing Estonia (bn EEK)
borrowing market. The bank-lending channel stresses the fact that banks are financial intermediaries designed to solve informational problems in credit markets. Unfortunately, we do not have data on individual loan contracts or data series of real sector balance sheets to explicitly identify the possible credit channel effects in Estonia. Instead, we specified assumptions that underpin the bank-lending channel and tested their relevance for Estonia. The bank-loan channel is supposed to work when bank loans are not perfectly substitutable with some alternative form of external financing, or when the banking sector is vastly dependent on domestic liabilities.16 Considering first the former, the only significant competitor for banks’ intermediated finance is direct borrowing from foreign capital markets. The domestic corporate bond markets and equity markets are too small and shallow to be relevant. In addition, the domestic leasing companies are almost fully owned and financed by the banking sector. The amount of real sector foreign borrowing (i.e., equity-related capital is excluded) makes up about one-third of the total domestic credit into the real sector (see Figure 7.11). There is also some evidence from the period of the Russian crisis that a decrease in domestic bank lending after the crisis was partly compensated with foreign financing ( Vesilind and Liiv, 2001). However, foreign capital markets are accessible to a limited group of larger enterprises, but the majority of the companies in Estonia are rather small. Those enterprises are more dependent on the banking sector in financing their activities. This means that they have a limited ability to substitute bank financing with other sources of credit, when the domestic
Monetary Policy and Transmission in Estonia
155
credit supply is restricted. The bank-lending effects can, therefore, be present for smaller enterprises and for households. On the other hand, the ability of the Estonian banking sector under a CBA to attract funds from abroad means that, when the foreign (re-)financing conditions are loose, domestic credit supply may follow the credit demand at the given interest rate level. Thus a contraction in domestic deposits does not necessarily lead to a contraction in credit supply in Estonia, as banks can attract foreign liabilities (or, in the short term, use their foreign reserves) instead of domestic deposits to fulfil the credit demand. However, a tightening in the overall monetary environment or deteriorating capital flows may still mean that credit rationing effects can be present from the bank lending side. This conclusion is also supported by Pikkani (2001), who finds evidence of strong credit rationing effects in Estonia during the Asian and Russian crises. The functioning of the credit channel has been approximated in a very simplified way. First of all, the balance-sheet channel, which works through changes in firms’ net worth, has been left aside. Second, bank lending was modelled as a function of credit demand and money supply. We did not take into account possible changes in the bank’s behaviour (e.g. credit rationing). Therefore, we have explicitly modelled the bank credits as a stock of real sector loans (i.e. the central and local government loans and financial institutions loans are not included in the loan stock: see equation 7.13): d(rl)
= α43 ECM10,t−1 + α44 d(il) + α45 d(rm2) +α46 d( yt−1 ) + α47 d(rl)t−1 ,
(7.13)
where ECM10 = rl − (rm2 + α48 ); α43 , α45 , α46 , α47 > 0; α44 < 0. This equation is similar to the model used by Pikkani (2001). To capture the effect of the credit demand, the average lending rate (il) and real GDP ( y) are used. il reflects the price of credit, real GDP ( y) proxies the growth expectations in the backward-looking mode. The effect of credit supply is given by money supply: this ties credit extension to base money. As the money supply is dependent on capital flows (which result from interest rate arbitrage), bank lending is also influenced by international capital flows. Lättemäe (2001) and Pikkani (2001) found that the credit channel has a relevant role in Estonia. They argue that during the Asian and Russian crises banks did not raise interest rates as much as could have been expected, but rather used credit rationing. Therefore, the impact from the increase in banks’ loan portfolio is also simulated.
156 Country-Specific Monetary Policy and Exchange Rates 0.2 0.15 0.1 0.05 0 1
3
5
7
9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43
–0.05 rx
Figure 7.12
y
i
tsb
0.0008 0.0007 0.0006 0.0005 0.0004 0.0003 0.0002 0.0001 0 –0.0001 –0.0002
π
The credit channel: adjustment after the shock, Estonia
Note: rx and y on the left axis; tsb, π and i on the right axis.
As can be seen from Figure 7.12, the increase in banks’ loan portfolio will temporarily increase real exports and GDP. Therefore, the credit channel has a similar effect to the interest rate channel. Both shocks – in interest rates and in the stock of credit – run partly through the supplyside, as they have an effect on the investment level, and thus on the capital stock. Credit growth will increase investment, and hence capital formation. The increase in the capital stock will have effects through the supply and foreign trade channels. This increase in the capital stock will shift the production potential and the long-run growth path. The rise in GDP also has a disinflationary impact through the price channel. In the long run the impact of the increase in banks’ loan portfolio does not have permanent effects on real variables. The increase in the capital stock is also reflected in the rise of exports, and this, in turn, supports GDP growth, and through the domestic income channel, increases the demand and inflationary pressures. Altogether, the rise in banks’ loan portfolio reduces the inflation rate through the credit channel. The impact of the credits on nominal variables (shortterm interest rates and CPI inflation) is negligible. Among the nominal variables the highest variation is in the trade balance: this is due to the credit extension impact through the REER channel. 4.3 The exchange rate channel According to the classification of Mishkin (1996), the exchange rate channel is part of the assets prices channel. It works mainly through exchange rate effects on net exports. Additionally, the exchange rate channel may include interest rate effects, as the exchange and interest rates are linked through interest rate parity.
Monetary Policy and Transmission in Estonia
157
In addition to exchange rate, equity prices also belong to the assets prices channel. The equity price channel affects the real sector through two processes, the wealth effect and Tobin’s q. However, we have not included equity prices in our model. This is due to the fact that Estonian financial markets are shallow, and therefore bank deposits are still the most common savings instruments in Estonia. A reliable estimation of the assets prices channel in Estonia is also difficult to produce. The importance of the exchange rate channel in Estonia comes from the openness of the Estonian economy. The exchange rate channel transmits the impact of both nominal and real exchange rates. Of course, one could argue that a CBA is an extreme form of fixed exchange rate regime, and, therefore, by definition the nominal exchange rate dynamics are excluded. In fact, in some episodes the kroon NEER depends crucially on appreciation/depreciation of the currencies unpegged to the kroon’s anchor currency (DEM, now the euro); hence exchange rate fluctuations stem from the fluctuations of the anchor currency. On a trade basis, the most important such floating currencies for Estonia are the USD and the SEK (the Swedish crown). It is still important to note that the relevance of the euro as trading currency has increased substantially – currently about 70 per cent of all foreign trade is carried out in euros, while in the mid-1990s DEM-based trade flows made up only about 20–30 per cent of the total. With the use of our model it is possible to simulate the impact of exchange rate fluctuations of all the main unpegged currencies used in external transactions. Here we present the results of a change in the EUR/USD exchange rate, as, after the introduction of the euro, the USD is the most important floating currency in Estonian foreign trade. The model simulation of a 1 per cent appreciation of the USD shows that, for a fixed regime, the nominal exchange rate channel is highly influential (Figure 7.13). The nominal depreciation of the kroon affects the economy through several variables, and its impact in the model matches this intuition. First, the nominal depreciation directly supports exports by increased price competitiveness. The export growth will in turn boost GDP, which starts the domestic income channel. Growing income creates some extra demand pressure and inflation, but this effect is minor. Depreciation also has another channel to increase inflation: due to the relatively high exchange rate pass-through, depreciation has an impact on import prices, accelerating domestic inflation. This effect is also marginal. On the other hand, nominal depreciation will cause a real depreciation (in spite of the inflationary effects of the nominal depreciation) and start the
158 Country-Specific Monetary Policy and Exchange Rates 1.4 1.2 1 0.8 0.6 0.4 0.2 0 –0.2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
–0.4 –0.6 –0.8 rx
Figure 7.13
y
i
π
tsb
The exchange rate channel: adjustment after the shock, Estonia
Note: π, i and tsb as the change from base; y and rx as percentage change from base.
real exchange rate channel. An REER depreciation means, ceteris paribus, improved (price) competitiveness and higher export growth. All in all, the final effect of a USD appreciation (or kroon depreciation) is positive. The extent of the nominal depreciation of the Estonian kroon outweighs the impact of higher inflation, leading to a real exchange rate depreciation. This is consistent with Randveer and Rell (2002), who show that the REER is an influential determinant of Estonian exports. From the income side, the EU GDP (y EU ) is the main factor, as nearly two-thirds of Estonian exports are directed to the EU. 4.4 The relative importance of the channels Comparing the performance of transmission channels, one could conclude that the exchange rate channel outperforms the others (see Table 7.2). This result is similar to that of other ‘transition’ countries (for Slovakia, Kuijs, 2002 shows that there is a strong impact of the exchange rate on prices, and virtually no effect from interest rates: in Chapter 6, a similar conclusion was reached for Slovenia). The outcome is also logical within the framework of a two-sector economy paradigm,17 and it is also intuitive, based on the realities of the Estonian economy. Taking into account its openness, one should expect the exchange rate channel to prevail. The interest rate channel acts primarily through the demand for non-tradables, and as the size of the non-tradables sector is relatively small, the impact of interest rates on aggregate measures (as GDP) should be limited. On the other hand, the exchange rate is an influential determinant of the tradables sector output. As the tradables sector is relatively large in Estonia, one would expect
Monetary Policy and Transmission in Estonia
159
Table 7.2 GDP and CPI elasticities (in %) for temporary shocks, Estonia∗
Interest rate Exchange rate Credits
GDP CPI GDP CPI GDP CPI
Impact elasticity
Total elasticity
0.03 0.01 0.16 0.13 0.04 0.00
0.06 0.03 0.28 0.46 0.16 0.05
∗ Impact elasticity describes the direct and immediate effect; total elasticity reflects the overall impact of the shock during adjustment.
the impact of the exchange rate also to be crucial for the development of the whole economy. The prominence of the exchange rate channel is also determined by the price-setting process and its consequences. Inflation in Estonia is predominantly influenced by import prices, while the role of demand pressure is secondary. Due to the prevailing producer currency pricing, import prices are crucially influenced by the nominal exchange rate. In addition, aggregate demand includes substantial exchange rate effects, as the aggregate demand for Estonian goods and services is directly related to the foreign demand, which is sensitive to the NEER. The low effectiveness of the interest rate channel stems from the institutional structure and fairly low levels of financial intermediation in Estonia during the estimation period. According to Kangur et al. (1999) corporate sector investments are mostly financed by internal sources and FDI. The share of domestic external financing (e.g. bank loans) has been of secondary importance. During 1994–98, bank loans to the corporate sector amounted to about 20 per cent of their investments. Additionally, as already mentioned, foreign interest rates may have some indirect impacts on Estonia through foreign demand. Another factor that may influence the transmission of monetary signals into the real sector is that the degree of financial intermediation was relatively low at the beginning of ‘transition’ and is still relatively low, when compared to advanced economies. Further convergence of financial deepening towards advanced countries’ levels implies several structural changes in real sector financing schemes. Those changes, affecting real sector balance sheets (increased freedom to borrow – in other words, to expand the liabilities’ side of their balance sheets), may
160 Country-Specific Monetary Policy and Exchange Rates
have been more relevant in such ‘transition’ countries than changes in interest rates (Kamin et al., 1998). This proposition can also hold when we consider real sector saving decisions. In low-income economies, households’ propensity to save can be more dependent on the income and wealth effects than on the interest rates. However, these results should be treated with some caution, as the model estimates are based on historical figures. The comparative relevance of different channels has been almost continuously shifting in favour of the interest rate channel. This process has been particularly intensive during the last years, for several reasons. First, the 1998 Russian crisis fostered the reorientation of trade towards Western countries, while reducing the relevance of the USD as trading currency. Moreover, our sample period includes figures from 1998 onwards, when the Russian rouble was devalued, causing problems in some sectors of the Estonian economy (especially parts of the food industry, whose exports were then more Russian-oriented). In addition, the initial introduction of the euro in 1999 reduced further the share of floating currencies in the currency composition of Estonian foreign trade, hence naturally reducing the relevance of exchange rate fluctuations. Second, the relevance of interest rates has increased during the years, as financial intermediation has continuously deepened. For example, bank lending to residents accounted to about 15 per cent of GDP in 1995, whereas in 2001 it had increased to more than 45 per cent. Therefore, recent empirical results have found stronger links between interest rates and real sector behaviour than prior works. The credit channel is, according to model estimates, also less effective, if compared to the exchange rate channel. In a way, the credit channel has the same effects as the interest rate channel. Both shocks – in interest rates and in the stock of credits – run partly through the supply channel. But unlike interest rates, the credit channel has clear GDP effects, while its impact on inflation is negligible. As mentioned previously, the credit channel was important during the Asian and Russian crises, but in other periods its importance has not been that clear.
5
Conclusions
One of the key issues in assessing monetary policy is the analysis of the MTM. This mechanism is influenced by several factors – the set-up and consistency of macroeconomic policy, including the choice of monetary regime and the structure of the economy – especially the financial
Monetary Policy and Transmission in Estonia
161
sector – and the intra-sector linkages are just a few of them. The effectiveness of different transmission channels is highly dependent on the evolution of financial systems and market structures and its evaluation is a challenging task even in advanced economies. In Estonia, the monetary system is based on a CBA whose anchor currency is the euro. The features of this CBA are an important aspect of the MTM in Estonia. There is no active monetary policy. Price stability in Estonia is linked to the anchoring role of the exchange rate, and all the necessary adjustments are left to the market. Furthermore, although the Estonian CBA is sometimes regarded as a CBA-like system, its set-up is rather orthodox. Under a fixed exchange rate and free capital mobility, Estonian monetary conditions are therefore closely linked to monetary policy in the EU: in addition to changes in the Estonian risk premium, the interest and exchange rate developments in Europe – and therefore also the monetary policy actions of the ECB – directly influence the Estonian monetary environment. The continuously increasing integration of the Estonian financial sector with foreign markets during the last ten years, as well as the particular set-up of the Estonian monetary framework, have strengthened those links substantially. In fact, the role of foreign money, and forex forward and swap markets, are more important for liquidity management in Estonia than domestic money markets. As a result, the transmission from EU interest rates into Estonian money market and retail rates is evidently strong. The effects of these monetary signals in Estonian enterprises and households depend on several aspects, from the substantial structural changes in Estonian during the last decade to specific factors influencing households’ and enterprises’ everyday decisions. To illustrate this, we presented a set of simulations using a small-scale macro-model. There are several important aspects to be kept in mind when discussing those results. Namely, they are based on ex post data, i.e., they do not take into account structural changes that have taken place in the economic environment in recent years. Our analysis showed that the exchange rate channel was the most relevant in Estonia during the model estimation period (1995–2000). Considering the smallness and openness of Estonia, this result seems reasonable. However, as a consequence of substantial structural changes, the relative relevance of the exchange rate channel has most probably been decreasing. Several factors point to such a shift. First, the reorientation of trade towards EU markets that reduced the relevance of the USD as trading currency and second, the initial introduction of the euro in 1999.
162 Country-Specific Monetary Policy and Exchange Rates
The estimated effects of the interest rate channel are conventional: a deteriorating domestic demand and disinflationary pressures. According to the estimations, its impact is more modest than the impact of the exchange rate channel. However, the relative importance of the interest rate channel has substantially increased during the last years, as several factors have diminished the relevance of the exchange rate channel. The lower impact of the interest rate channel in the estimations stems also from the fairly low levels of financial intermediation and negative real interest rates at the beginning of the estimation period. In addition, the model does not account for the indirect impact of foreign interest rates on Estonian economic activity through external demand. As the Estonian economic growth is mainly export-determined, changes in foreign interest rates may have strong indirect impacts through changes in external demand, in addition to direct impacts through Estonian interest rates. Finally, according to our simulations, the credit channel has clear effects on the economic activity in Estonia, while its impact on inflation is negligible. The credit channel has similar effects as the interest rate channel. Both shocks run partly through the supply side, as they have an impact on the level of investments and on the capital stock. However, the credit channel has been most important during the Asian and Russian crises, when Estonian banks faced constraints on obtaining additional capital from foreign markets. During other periods, its relevance has not been that clear.
Notes 1. The exact exchange rate of the kroon (1 DEM = 8 EEK) was set by a separate decree. Since 1999, the kroon has been fixed to the euro (1 EUR = 15.64664 EEK). 2. The Estonian kroon money market comprises short-term interbank deposits/ loans, corporate debt securities and forex forwards and swaps instruments. No treasury bills are issued by the state. 3. Except during the Asian and Russian crises, when the domestic risk premium increased. 4. In equation (7.1), Greek letters denote parameters, Latin ones denote variables, and the small cap letters denote values in logarithms (except the interest rate). d(·) is the difference operator. 5. The Asian crisis in 1997 resulted in a sharp increase in the speculative positions in foreign currency swap and forward markets in Estonia (see Lättemäe et al., 2002; Lepik, 1999). As a result, forward quotations grew, also pulling the money market rates. The foreign currency swap and forward markets also experienced a similar rise during the Russian crisis in 1998. However,
Monetary Policy and Transmission in Estonia
6. 7.
8. 9.
10.
11.
12.
13. 14.
15. 16.
17.
163
unlike October 1997, the 1998 episode was not followed by a notable increase in trading volumes, as the speculative pressures were smaller. Therefore, pressure on the currency was felt mainly through high forward rates (Lepik, 1999). Foreign effective demand is here the trade-weighted average GDP of Estonia’s main trade partners. Long-term foreign capital has mostly been attracted to subordinated liabilities, issued long-term securities or as long-term borrowing in DEM or euros, where the interest rate has usually been indexed to three- to six-month DEM LIBOR or three- to six-month EURIBOR (plus risk premium). The empirical research in this field has not been able to identify other motives (e.g. speculative demand) for money demand. Although it is interesting to note that the difference between Estonian and Euroarea nominal retail rates for corporate borrowing is currently about 2 per cent, which is close to the differences at the inflation rates: hence the real rates are similar. It should be noted that, given the difficulties with the proper estimation of inflation expectations, we use nominal interest rates rather than real interest rates in our IS-curve specification. Such an approach is supported both by prior practice in Estonian macro-models (see Sepp et al., 1999) and by difficulties in estimating real interest rates (Rell, 1999). The indicator h is added to the production function in order to describe the total productivity change generated from labour reallocation, h is higher, when labour moves into more productive sectors. According to Rõõm (2001), the inclusion of h helps to correct the general measure of employment, expressing the labour-augmenting technological progress generated from sectorial restructuring. The relationship between income and price level is estimated in IMF, World Economic Outlook (2000), Kravis (1986), Hansson and Helliwell (1990) and Randveer (2000). See Sepp et al. (2000). For a discussion on standard approaches to monetary transmission see Mishkin (1996), Benhabib and Farmer (1999), Romer and Romer (1990), Bernanke and Gertler (1995), Bernanke and Blinder (1988) and Kashyap and Stein (1993), to name a few. For ‘transition’ economies, see BIS (1996). π is annualized inflation, y is real GDP, rx is real exports and tsb is current account balance. There are two different views of the credit channel. The first suggests that banks are dependent on deposits, and any contraction in deposits will prompt banks to shrink their balance sheets. The second suggests that there are significant numbers of bank-dependent firms that cannot replace bank lending with other sources of finance (Cecchetti, 1999). According to Leitemo and Røisland (1999), the significance of the monetary policy transmission channels are different for traded and non-traded sectors. In a two-sector model, there are basically two explicit channels for the transmission of shocks: interest rate and exchange rate channels. It seems plausible that the tradables sector output is more sensitive to REER changes, whereas the non-tradables sector output is more affected by the real interest rate, through domestic demand.
8 Czech Monetary Policy on the Road to European (Monetary) Union Roman Matoušek and Anita Taci
1
Introduction
The current economic and political developments in advanced Central and Eastern European countries (CEECs) have unambiguously confirmed that these countries have a strong commitment to become members of the European Union (EU). Many CEECs have perceived EU membership as an intermediate goal on the road to EMU, considered as a final objective for those countries. There is a general consensus that EU membership is not a sufficient condition for EMU membership. One of the important obligations for joining EMU is fulfilment of nominal convergence criteria known as the Maastricht criteria, although some countries became members of EMU without achieving the full set of these nominal criteria. However, one should not expect that the new members of the EU will be allowed such discretion from the current EMU members. This chapter studies whether a current level of relatively low inflation and interest rates, which are two of five macroeconomic criteria required for EMU membership, are sustainable during the convergence process. We test how the introduction of direct inflation targeting (DIT) has increased the transparency and credibility of the Czech National Bank’s (CNB) monetary policy and whether DIT could act as a stabilizing anchor in the economy. The methodological approach is based on several recent studies that measure the effect of announcements of interest rate changes on interest rates along the yield curve (Hardy, 1998; Muller and Zelmer, 1999; Haldane and Read, 2000). In addition, we test a cross-border effect on the Czech financial market by the European Central Bank (ECB). 164
Czech Monetary Policy
165
Since DIT has become the prevailing monetary strategy in many countries, there have been several studies analysing its pros and cons. Some empirical studies have shown that in countries applying DIT, the transparency and credibility of monetary policy have increased (Svensson, 1995). There has been little analysis, to date, of the impact of DIT in transition economies. The focus of our analysis is on examining the effect of changes in the two-week repo rate (the official interest rate) on short- and longmaturity market interest rates. Such an analysis allows us to test the two hypotheses about monetary policy transparency and credibility discussed, e.g., in (Haldane and Read (2000) and Muller and Zelmer (1999). Haldane and Read (2000) show that movements at the short end of the yield curve in response to changes in official rates reflect the market’s degree of uncertainty about the central bank’s monetary policy reaction function, while movements at the long end of the yield curve reflect market assessment of the central bank’s credibility. The argument relies on information asymmetry and the existence of a stationary stochastic equilibrium with full knowledge of the authorities’ reaction function. Only then will the market reaction be insignificant. This chapter is organized as follows. Section 2 discusses the nominal convergence in terms of the Maastricht criteria and evolution of the conduct of monetary policy by the Czech National Bank (CNB) since 1996. We start by discussing the appropriateness of DIT in ‘transition’ economies. We focus on the prerequisites that need to be fulfilled for successful implementation of DIT. We then examine whether DIT in the Czech Republic fulfils the two basic prerequisites of fiscal discipline and a single monetary anchor. Section 3 deals with the methodological issues for testing our nexus between transparency and credibility, versus the market uncertainty reflected in the market reactions to the CNB’s monetary decisions. Our estimation results are presented in Sections 4 and 5. Finally, we summarize our results and examine whether or not our hypotheses about the transparency and credibility of the CNB to sustain the current disinflation trajectory hold.
2
DIT in transitional economies
The appropriateness of DIT for ‘transition’ economies is often questioned. There is no consensus about which framework of monetary policy is most appropriate. The recent economic literature suggests that DIT might be the right choice when disinflation becomes a primary
166 Country-specific Monetary Policy and Exchange Rates
goal of stabilization policy. For example, Orlowski (2000) argues that DIT might be appropriate for the leading Accession Countries (i.e., the Czech Republic, Poland and Hungary) in order to help reduce inflation to the EU level. But there are questions about whether they can meet the necessary preconditions for a DIT framework. First, there should be no fiscal policy dominance, i.e., monetary policy should not be subordinated to accommodate fiscal policy. Second, there should be no other nominal monetary anchors (Krzak and Ettl, 1999). This unequivocal condition holds true only in the strict version of inflation strategy. Nevertheless, the fiscal condition can be difficult to fulfil in such economies. If there is a lack of fiscal discipline, the inflationary pressure arising from the fiscal side might jeopardize the achievement of the inflation target. As for additional policy targets, Svensson (1995) argues that inflation targeting is not consistent with a fixed exchange rate regime, but it is compatible for example with a target of full employment and financial stability. Central banks, in the initial phase of the economic transition, did not possess the necessary credibility that would contribute to their effort to bring down inflation expectations. One way of reducing inflationary expectations is to peg the domestic currency (fixed exchange rate regime). But a successful fixed exchange rate regime requires fiscal discipline and sufficient foreign reserves. This was also a challenge for many transition economies. Several countries have, nevertheless, opted for currency boards (Estonia, Bulgaria, Bosnia and Herzegovina). The Central and East European countries (CEECs) have used a variety of monetary policies. After abolishing other nominal anchors, the Czech Republic and Poland have adopted inflation targeting since the beginning of 1998 and 1999 respectively. Hungary has done so only since June 2001, after abandoning a prolonged reliance on an exchange rate target (see Chapter 10). The question remains whether DIT is an appropriate form of monetary policy under a flexible exchange rate regime in transition economies. Some economists have argued that DIT can bring down inflation and reduce inflation volatility, which targeting exchange rates cannot (Svensson, 1997). In practice, it has been difficult for CEECs to forecast future inflation and to estimate the lags in the response of the economy to changes in monetary policy (due to insufficient time series of observations). But theory predicts that DIT can increase both transparency and credibility in transition economies.
Czech Monetary Policy
167
2.1 DIT in the Czech Republic In January 1998, the CNB became the first central bank of a transition country to adopt inflation targeting. The shift to DIT was precipitated by the forced abandonment of the fixed exchange rate regime in May 1997. As for the prerequisites of DIT that have already been mentioned, the fiscal position was broadly sound on that occasion. In addition, the CNB already had a relatively high degree of independence and there were no explicit nominal anchors. Since the Czech economy is still in the process of price deregulation, the CNB adopted a net inflation index (NII)1 as the target variable instead of the consumer price index (CPI) in the first stage of DIT. The economic rationale behind this was that the CNB is only accountable for deregulated or market prices. At the end of 1999, the Czech CPI included prices of 754 items, 18 per cent of which were regulated or administered. Net inflation in the Czech Republic declined until the second quarter of 1997. However, it rose after the April/May 1997 crisis in the Czech capital markets (which was induced by corporate governance and liquidity problems in the Czech financial sector). The financial crisis led to the decision to float the Czech currency. A strong initial depreciation of the Czech currency renewed inflationary expectations. Higher inflationary expectations persisted until the second quarter of 1998 and coincided with the financial crisis in emerging markets. The CNB responded to the currency attack by raising interest rates during 1997. It took approximately 15 months to bring down the official interest rate to pre-crisis level. The CNB adopted an inflation target to show its commitment to reaching low, sustainable inflation over a long-term horizon. The strong determination of the CNB to strengthen macroeconomic stability and keep inflation low was declared in April 2001. The CNB, in agreement with the government, has announced an inflation target for annual CPI growth in the period January 2002–December 2005. The trajectory of the inflation target is forecast to be falling from in the band 3–5 per cent in January 2002 to 2–4 per cent in December 2005. The institutional framework of the CNB’s inflation targeting regime has many of the characteristics deemed necessary to increase transparency (see Kuttner and Posen, 2001). For example, the CNB has published numerical long-run goal for monetary policy, publishes an inflation report, together with the central bank’s forecast and an ex post evaluation of monetary policy and the minutes of the banking board meetings. The latest steps taken by the CNB, i.e. to target headline inflation (CPI) directly, even increased the transparency of monetary
168 Country-specific Monetary Policy and Exchange Rates
% 15 12 9 6 3 0 01.03.93
01.03.96 Czech CPI
Figure 8.1
01.03.99
01.03.02
German CPI
CPI inflation for the Czech Republic and Germany (1993–2002)
Source: CNB.
policy since it is well understood by the public. The inflation differential in terms of the CPI between the Czech Republic and Germany has dramatically decreased over the last nine years (see Figure 8.1).
3
Methodology and data
The focus of this study is to investigate the effect of changes in the CNB’s official interest rate (the two weeks’ rate, indicated by 2WREPO) and ECB2 official interest rate on short and long-term interest rate contracts. This analysis allows us to examine hypotheses about the monetary policy transparency as discussed in Haldane and Read (2000). They propose that in a transparent monetary policy regime, short-term interest rates anticipate to some degree changes in official rates. Also, in a credible monetary regime, a change in official (short-term) rates does not lead to a change in long-term rates, which reflects expectations about inflation over the longer term. Indeed, a rise in official (short-term) rates may lead to a fall in longterm rates if monetary policy results in lower expectations of inflation in the medium term. Thus if financial markets are efficient, assets prices reflect all available information and react only to unanticipated events that might affect their fundamental values. This argument relies on information asymmetry and the existence of a stationary stochastic
Czech Monetary Policy
169
% 60 50 40 30 20 10 0 1.2.96
1.10.97
1.21.98
2.1.99
REPO2W (middle line) Lombard rate (upper line) Figure 8.2 2002)
2.10.2000 31.01.02 Discount rate (lower line)
Movement of official interest rates, CNB (January 1996–December
Source: CNB.
equilibrium with full knowledge of the authority’s reaction function. Only then will the market reaction be insignificant. 3.1 CNB’s monetary instruments The CNB uses several monetary tools that impact on the financial market. Repo tenders set up by the CNB are the most important monetary policy instruments. The basic duration of these operations is two weeks. With respect to the systemic liquidity surplus in the Czech banking sector, repo tenders are used for absorbing liquidity. Other official interest rates, such as the Lombard and discount facilities, have, we believe, only a ‘psychological’ role. Their spread is to some extent only a corridor for the short-term yield (see Figure 8.2). An additional powerful monetary instrument – compulsory reserves – has recently been significantly reduced and plays now a marginal role only.3 We therefore study the impact of the two-week repo rate on short- and long-term interest rates. 3.2 Methodological approach and asset price data Our study follows the method applied by Haldane and Read (2000), taking into account the specific characteristics of the Czech financial markets. First, the following equation is estimated: it,j = C + βj (L)it,j + β4 itc + εt ,
(8.1)
170 Country-specific Monetary Policy and Exchange Rates
where j and t represent indices for the maturity of market interest rates and date respectively. The dependent variable it,j is the daily change in market interest rates, βj is a polynomial in the lag operator (L). The independent variable itc is a dummy variable of the official interest rate, which is closely watched by the market.4 The coefficient β4 measures the market reaction to changes in the official interest rate. If changes in the official interest rate are perfectly anticipated, then the coefficient β4 is equal to zero. This is consistent with a transparent monetary policy. The magnitude of the coefficient β4 can also be interpreted as a rate of surprise for the market. We apply equation (8.1) to Prague interbank offer rates (PRIBOR) of one to 12 months’ maturity. The CNB announces any change to the official interest rate at 1:00 p.m. on the day before the rate becomes effective. PRIBOR is quoted on the same day at 11:00 a.m. and therefore cannot react on the day of an announcement. A different equation is applied to test the reaction of long-term interest rates to monetary policy announcements. This is because trading in long-term interest rates instruments continues after the CNB announces changes in the official rate and the long-term rates are quoted at the close of business. Thus, long-term interest rates can react to the CNB announcements on the same date. In order to capture this market feature, the coefficient β4 in regression (8.1) must be lagged one day, i.e. the day when the change is announced but not yet effective to test for the reaction of long-term interest rates to official rate changes. So, the following equation is to be tested: c it,j = C + βj (L)it,j + β4 it−1 + εt .
(8.2)
The following dependent variables are used: daily differences in swap interest rates of one to ten years’ maturity (Swap1Y to Swap10Y) and daily differences in government benchmark bonds of two and five years’ maturity (GB2Y and GB5Y). Because of a lack of data for swap interest rates from January 1996 to August 1997, bond price indices (GBPI, ˇ of government bonds and CBPI, of corporate bonds) from the Ceská spoˇritelna (the Czech Savings Bank) are used. They were created to track price movements in the government and corporate markets. They serve as benchmarks for highly liquid instruments. To adjust standard errors for autocorrelation and heteroscedasticity the Newey–West method is used (Mishkin, 1990). The analysis of the effect of changes in official rates on the short and long end of the yield curve during the DIT period is extended.
Czech Monetary Policy
171
The coefficient β4 captures the movement of the yield curve in the time period from t − 1 to t − 5, i.e., itc is lagged t − 5, t − 4, t − 3, t − 2 and t − 1 days. The regression (8.3) is the extension of equations (8.1) and (8.2) for an additional dependent variable itecb – a dummy variable for repo rate in the ECB/Bundesbank.5 This variable allows us to analyse the cross-border effect on the Czech financial market. It is assumed that the long end of the yield curve will exhibit the effect of monetary policy changes abroad, mainly in the Euroarea, on the domestic market. To find out the nexus between the CNB and ECB policy is an additional important dimension of our study, i.e., to analyse interest rates adjustment connected with convergence towards the EU. Thus the following equation is tested: it,j = C + βj (L)it,j + β4 itecb + εt .
(8.3)
3.3 Data sample The data sample consists of the daily data on short-term yields from 2 January 1996 to 31 December 2001. There is a complete data sample for the short-term yields, i.e., one, three, six and 12 months’ PRIBOR, and also for the exchange rate, GBPI and CBPI. The swap interest rates data series starts as of 25 September 1996; the government bond data start as of 8 October 1997. Because of high volatility in the dependent variables (Table 8.1), caused by both transitional effects and exogenous factors, we divide the sample into three sub-samples. The first period includes data from 2 January 1996 to 31 March 1997, which is the period predating the financial crises. The second period, from 1 April 1997 to 31 December 1997, then covers an interval of financial turbulence that the Czech financial markets had to withstand. This sub-sample contains 250 observations. From April 1997 to December 1997 there were 30 changes in the official repo rate. The high number of changes may reflect the authorities’ concern to ensure a ‘soft landing’ for
Table 8.1
Pre-crisis Crisis DIT
Average volatility of PRIBOR PRIBOR 1M
PRIBOR 3M
PRIBOR 6M
PRIBOR 12M
0.017 0.85 0.015
0.023 0.566 0.015
0.025 0.42 0.016
0.027 0.32 0.014
172 Country-specific Monetary Policy and Exchange Rates Table 8.2 Effects of official interest rate on short-term interest rates (01.96–03.97) Maturity PRIBOR1M PRIBOR3M PRIBOR6M PRIBOR12M
β4
R2
D–W
1.04∗∗ 1.18∗∗ 1.19∗∗ 1.218∗∗
0.67 0.59 0.54 0.48
1.70 1.65 1.63 1.62
∗∗ indicates significance at 1%, ∗ at 5%, + at 10%.
the economy while preventing another speculative attack on the Czech koruna. The third period, the period of DIT, includes data from 2 January 1998 to 31 December 2001. The number of observations is 1,016 and there were 22 changes in the official interest rate.
4 Asset price reactions to changes in the CNB’s official interest rate before DIT 4.1 The effect of official interest rate changes in the pre-crisis period First, the regressions are estimated for the period before DIT and the financial crises, from 1 January 1996 to 31 March 1997. During this period, the CNB used a monetary aggregate as an intermediate target, and the exchange rate as a nominal anchor. The CNB changed its official interest rate only five times during this period. Table 8.2 shows that the coefficients of changes in the CNB’s official repo rate have large values of about 1, and are significant at the 1 per cent level for money market interest rates (PRIBOR) with maturities of one to 12 months. This strong reaction in short-term market interest rates of 104 per cent to 122 per cent of the change in the official interest rate is consistent with the hypothesis of a low level of transparency in the Czech National Bank’s monetary policy reaction function. Due to the unavailability of data on bond yields and swap interest rates for this period, the reactions of log-differences in a price index of government bonds (GBPI) and of a price index of corporate bonds (CBPI) constructed by the Czech Savings Bank are analysed instead.6 The results are shown in Table 8.3. The estimated coefficient β4 for the impact of changes in the official repo rate are negative7 and close to zero for both bond price indices. For the government bond price index, the coefficient β4 is statistically significant at the 1 per cent level.
Czech Monetary Policy
173
Table 8.3 Asset price reactions to changes in the official repo rate (01.96–03.97) Maturity GBPI CBPI CZK/EUR
β4
R2
D–W
−0.013∗∗ −0.015 −0.002
0.188 0.13 0.006
2.05 2.06 1.98
∗∗ indicates significance at 1%, ∗ at 5%, + at 10%.
Table 8.4 Effects of official interest rate on short-term interest rates (04.97–12.97) Maturity PRIBOR1M PRIBOR3M PRIBOR6M PRIBOR12M
β4
R2
D–W
−0.13 0.07 0.027 −0.006
0.05 0.15 0.16 0.19
0.05 2.05 2.01 1.98
∗∗ indicates significance at 1%, ∗ at 5%, + at 10%.
The effect of changes in the official repo rate on log-differences of the exchange rate is also investigated. The estimated coefficient is low and not significant in this first period, i.e. changes in the official repo rate had only a marginal effect on the exchange rate. This is probably because this period was characterized by a fixed exchange rate within a narrow band of ±7.5 per cent.
4.2 The effect of official interest rate changes in the period of turbulence In the turbulent period of the financial market crisis, from April 1997 to December 1997, the coefficients for changes in the official interest rate at the short-term contracts are not significant, and their magnitude is lower than in the previous period, as Table 8.4 shows. The effect of official interest rate changes on the long-term bond and swap rates is also estimated. The magnitudes of coefficients are low and not statistically significant, except for the two-year government bond yield, whose coefficient was statistically significant at the 5 per cent level.
174 Country-specific Monetary Policy and Exchange Rates
5 Asset price reactions to changes in the official interest rate in the DIT period As said in the introduction, EMU membership for new EU members is not guaranteed automatically. It will be determined, apart from other prerequisites, by the fulfilment of the Maastricht criteria. In other words, nominal convergence will be regarded as a necessary, but not sufficient, condition for entry into EMU for any country. Two of three monetary criteria, i.e., price stability, convergence of interest rates and exchange rate regime stability are discussed here. This study is not aimed at discussing whether the Czech Republic is in a position to fully cope with these criteria, but rather whether the present stage of convergence is sustainable in the long to medium run before admission. Figure 8.1 shows the CPI inflation rates in Germany and the Czech Republic. Figure 8.3 displays the official interest rates (Repo) differentials between the ECB and the CNB, as well as the interest rate differentials in the five-year government bonds in Germany and the Czech Republic. The path of both the CPI inflation and the interest rates
% 10
% 7
9
6
8
5
7
4
6
3
5 2
4
1
3 2
0
1
–1 –2
0 1.1.99
18.08.99 GB5Y
3.4.00
11.1.00
10.07.01 03.01.02 24.10.02
Official interest rates (right scale)
Figure 8.3 Interest rate differentials between official interest rates (ECB and CNB) and five-year government bond (Germany and the Czech Republic) Source: CNB.
Czech Monetary Policy
175
differentials confirms the convergence trend. But the question remains whether this trend is sustainable.
5.1 The effect of the CNB’s official interest rate changes on the DIT A similar methodological approach, as applied in the previous section, is used to test the credibility of the continuous convergence. This time the DIT period is analysed, i.e. 1 January 1998–31 December 2001. In this period, the analysis is extended by estimating the market reactions five, four, three, two and one day before the official interest rate becomes effective. This is done to disclose the dynamic behaviour of the market in terms of the surprise and credibility of monetary policy. In order to capture a symmetric view, β4 is also estimated for t + 1, t + 2, t + 3, t + 4 and t + 5 days.8 The econometric results show that the coefficients for changes in the CNB’s official repo rate are lower than in the first period (see Table 8.5), and are all statistically significant at a 1 percent level. These results for the short end of the yield curve are consistent with the hypothesis that the transparency of the CNB has increased comparing to the pre-DIT and pre-crisis periods. Table 8.5 shows that only one- and two-month PRIBOR reacted, even before an official interest rate changed. The coefficients are lower compared with the same coefficients in time t. The analysis of the sustainability of the long end of the yield curve, i.e. sustainability of the present long-term spot interest rates, shows that coefficients are statistically significant at the 1 per cent level for swap rates of one, two and five years’ maturity. As for the swap rate of ten years’ maturity, the coefficient is statistically significant at the 5 per cent level. The coefficient of the two-year government bond is statistically significant at the 5 per cent level for t, t − 3, t − 5. The coefficient of the Table 8.5 Effects of official interest rate on short-term interest rates (01.98–12.01) Maturity PRIBOR1M PRIBOR3M PRIBOR6M PRIBOR12M
β4
β4 (−1)
β4 (−2)
β4 (−3)
β4 (−4)
β4 (−5)
−0.21∗∗ −0.15∗∗ −0.12∗∗ −0.096∗∗
−0.04 −0.015 −0.012 −0.008
−0.05∗ −0.04∗∗ −0.02 −0.023
−0.01 −0.02 −0.03 −0.03
0.00 0.003 0.023 0.032
0.01 0.016 0.024 0.02
∗∗ indicates significance at 1%, ∗ at 5%.
176 Country-specific Monetary Policy and Exchange Rates Table 8.6 Asset price reactions to changes in the official interest rate (01.98–12.01) β4
β4 (−1)
β4 (−2)
β4 (−3)
β4 (−4)
β4 (−5)
−0.026 −0.03 −0.023 −0.005 −0.082∗ −0.049+ 0.065+
−0.005 0.002 0.007 0.02 −0.025 −0.04∗∗ 0.004
−0.076∗∗ −0.047∗∗ −0.058∗∗ −0.056∗ −0.05 0.026 0.009
−0.01 −0.021 −0.017 −0.028 −0.056∗ −0.03 −0.017
0.01 0.02 0.031+ 0.036+ −0.009 0.003 0.033
0.02 0.017 0.028 0.022 0.06∗ 0.02 0.01
Maturity SWAP1Y SWAP2Y SWAP5Y SWAP10Y GB2Y GB5Y CZK/EUR
∗∗ indicates significance at 1%, ∗ 5%, + 10% levels.
Table 8.7 Effects of ECB interest rate on Czech short-term interest rates (01.98–12.01) Maturity PRIBOR1M PRIBOR3M PRIBOR6M PRIBOR12M
β4
β4 (−1)
β4 (−2)
β4 (−3)
β4 (−4)
β4 (−5)
−0.023 −0.041 −0.039 −0.04
−0.029 −0.06 −0.055 −0.07
−0.001 −0.02 −0.011 −0.004
0.015∗∗ 0.023+ 0.028∗ 0.032∗
0.023∗∗ 0.038∗∗ 0.031∗∗ 0.023+
0.01 0.022∗∗ 0.021∗∗ 0.03∗∗
∗∗ indicates significance at 1%, ∗ 5%, + 10% levels.
five-year government bond is significant at the 10 per cent and 1 per cent levels for t and t − 1 respectively (Table 8.6). Changes in the official interest rate have a statistically significant effect only at the 10 per cent level on the exchange rate and the time t. 5.2 The effect of ECB monetary policy decisions on the Czech market The second part of the analysis within the period of DIT is focused on a cross-border effect. The results presented in Tables 8.7 and 8.8 indicate that the ECB monetary decisions have had effects on the Czech interest rates and asset prices. As for the short end of the yield curve, the coefficients are statistically significant at the 1 per cent level for PRIBOR1 up to PRIBOR6 but four days before the interest rate was changed. The same is valid for t − 3; however, the coefficient of PRIBOR1 is only significant at the 1 per cent level.
Czech Monetary Policy
177
Table 8.8 Asset price reactions to changes in ECB interest rates (01.98–12.01) Maturity SWAP1Y SWAP2Y SWAP5Y SWAP10Y GB2Y GB5Y CZK/EUR
β4
β4 (−1)
β4 (−2)
β4 (−3)
β4 (−4)
β4 (−5)
−.067 −0.047 −0.032 −0.012 −0.031 −0.057+ −0.000
−0.055 −0.035 −0.021 −0.028 0.026+ −0.032 −0.025
0.001 0.019 0.021 0.026∗ 0.008 0.003 −0.015
−0.002 0.048∗ 0.04∗ 0.039∗ 0.026+ 0.032+ 0.022
0.06+ 0.028 0.027 0.038∗ 0.037∗ 0.026∗ 0.004
0.037∗ 0.027 0.029 0.02 0.029+ 0.028 0.041
∗∗ significant at 1%, ∗ 5%, + 10% levels.
Table 8.9 Confidence intervals of coefficients for official interest rate decisions before and during DIT
1-month PRIBOR 3-month PRIBOR 6-month PRIBOR 12-month PRIBOR
January 1996–March 1997
January 1998–December 2001
[0.995, 1.084] [1.106, 1.267] [1.267, 1.118] [1.135, 1.304]
[0.414, 0.0062] [0.310, −0.01] [0.272, −0.032] [0.240, −0.048]
It is necessary also to analyse the reaction of the long end of the yield curve. The coefficients are mostly significant for t − 3 and t − 4. As for the exchange rate, no statistically significant cross-border effect can be traced (Table 8.8).
6
Conclusions
At short maturities, the coefficients for changes in the official repo rate are lower in the DIT period than in the pre-crisis period. Table 8.9 shows the 95 per cent confidence intervals for the coefficients in each period for maturities of one to 12 months. As we can see, the confidence intervals do not overlap. The results are significant at the 5 per cent level. This implies that the hypothesis of no increase in the transparency of monetary policy with the introduction of DIT can be rejected at the 5 per cent level. During the period of financial market crisis, we found that shortterm interest rates did not react significantly to changes in the official repo rate. The absence of a significant response to official interest rate
178 Country-specific Monetary Policy and Exchange Rates
decisions on the day of the change in rates is consistent with the hypothesis that the market anticipated these changes, and thus that monetary policy was transparent during that period. However, since short-term market interest rates were very volatile in the financial crisis period, an alternative explanation is that the properties of the time series of interest rates (e.g. heteroscedasticity) might lead to greater problems in the econometric estimation during that period, and perhaps a bias in the estimated coefficients and their standard errors. It is also possible that short-term interest rates exhibited a delayed reaction to monetary policy decisions, or that they were affected to a greater extent by other factors, which might have become more important in the financial market crisis. We found that bond yields and interest rate swap of maturities of two years and longer did react significantly (at 1 and 5 per cent levels) to official interest rate decisions in the DIT period. The price index for government bonds considered here showed a small reaction in the pre-DIT and pre-crisis period.9 This is consistent with the hypothesis that monetary policy was credible both before and after the introduction of DIT. But the absence of a significant reaction of long-term yields on the days of changes in official interest rates might also be due to other factors. In particular, there might be a delayed transmission of changes in official interest rates along the yield curve. In addition, we found a statistically significant reaction of the Czech interest rate and assets prices to the ECB’s official interest rate. This confirms our hypothesis about the partial dependence of the Czech money market on the Euroarea, even before entry into the EU. To sum up, our conclusion is that DIT is a potentially effective framework for, at least, the larger, floating currency acceding countries, once the credibility of the monetary authority has been established. Additionally, as in the case of real integration – i.e., in terms of trade and FDI flows – the financial integration of the Czech Republic with the European Union, and, more specifically, the Euroarea, already seems to be a reality.
Notes 1. The NII is defined as the rate of increase in consumer prices, excluding administered and regulated prices and the impact of indirect taxes. 2. The effect of the ECB is analysed only for the period 1998–2001. 3. The Czech National Bank reduced reserves from 11.5 per cent to 2 per cent. 4. The independent variable itc has the value of 1 if the CNB changes the official interest rate and the value of 0 otherwise.
Czech Monetary Policy
179
5. The cross-border effect is analysed only for the period January 1998–December 2001. For the period preceding the establishment of the ECB we apply the Bundesbank rates. ˇ 6. Both bond price indices were established by the Ceská spoˇritelna in January 1994, as described in Section 3. 7. Note that a fall in the price indices corresponds to an increase in yields. 8. Our test does not confirm any reaction of the market after the changes in official interest rates. Therefore this test is not included here. 9. But note that this is a coefficient for log-differences in a price index, rather than a yield.
9 Poland’s Accession to EMU – Choosing the Exchange Rate Parity Łukasz Rawdanowicz
1
Introduction
As its monetary and exchange rate framework for Euroarea accession, Poland, after some indecision (see Vinhas de Souza and Ledrut, 2002), has finally chosen a free floating exchange rate regime and DIT. These arrangements seem to be fixed until accession to the Euroarea, despite some proposals of unilateral euroization or introduction of a currency board arrangement (Bratkowski and Rostowski, 2001; Rostowski, 2001). Thus, Poland is expected to follow a standard path of joining EMU and it will have to stay in ERM II for at least two years. The core principle of the standard ERM II is the maintenance of the exchange rate within the ±15 per cent fluctuation band without devaluation of the central parity. In this case, a nominal zloty–euro parity will have to be chosen upon entry to ERM II and then the Euroarea. The initial ERM II parity may be the irrevocable exchange rate for entry into EMU, although changes, after a multilateral decision process, are possible. In 1999, countries that took part in the stage 3 of EMU establishment fixed their currencies to the euro at their ERM parities. They were initially set before 1979 and then devalued on several occasions. In the case of the UK, which joined the ERM in 1990, the parity was chosen based on the purchasing power parity criterion (MacDonald, 2000). In all cases, the exchange rates were set so as to reflect some ‘equilibrium’ conditions. This principle should also apply to Poland and other prospective EMU members. However, in talking about equilibrium exchange rate, the corresponding conditions must be clearly defined. The existing literature offers various approaches to defining equilibrium exchange rates. They differ in economic interpretation and empirical estimations. 180
Poland and Exchange Rate Parity
181
This chapter deals with the choice of the exchange rate parity upon Poland’s entry to ERM II and EMU. In the quest for an equilibrium exchange rate for Poland, estimations of fundamental and behavioural equilibrium exchange rates are undertaken. The results are discussed in terms of their sensitivity to the assumptions used and the models’ specifications. Their economic interpretation is also provided. The discussion on parity choice goes beyond model-based considerations. In principle, issues around the consequences of choosing the particular nominal exchange rate, political bargains and reactions of financial markets are addressed. This chapter is organized as follows. Section 2 surveys theoretical concepts of equilibrium exchange rates. Section 3 deals with empirical estimations of fundamental and behavioural equilibrium exchange rates for Poland and discusses problems of their application. Section 4 summarizes the theoretical and empirical considerations and draws practical guidelines for setting the zloty–euro parity. Section 5 concludes.
2
Concepts of equilibrium exchange rate
The estimation of equilibrium exchange rates has attracted considerable theoretical and empirical attention – for instance Williamson (1994), Montiel (1997), Clark and MacDonald (1998), MacDonald (2000), and Isard et al. (2001). Generally, the three most popular approaches to assessing the equilibrium exchange rate are identified. These are: (i) purchasing power parity (PPP); (ii) fundamental equilibrium exchange rate (FEER); and (iii) behavioural equilibrium exchange rate (BEER). All these concepts will be briefly discussed in what follows. 2.1 PPP According to PPP, a nominal exchange rate of any two currencies should reflect closely the relative purchasing powers of the two monetary units represented by national price levels (Isard et al., 2001). The strong version of PPP requires that the nominal exchange rate and price ratio should be 1 : 1. As changes in a nominal exchange rate mirror changes in relative price levels between the two countries, the real exchange rate should be constant or at least mean-reverting (in the weak version of PPP). The weak PPP hypothesis has been rejected to hold in the short run, though some econometric evidence of its long-run properties has been found, especially in the case of panel tests (see for instance Isard et al., 2001 or Bayoumi and MacDonald, 1998). Empirical tests fail, however, to prove the strong version of PPP.
182 Country-specific Monetary Policy and Exchange Rates
One refinement introduced to the PPP approach was due to incorporation of the BS effect (Harrod, 1993; Balassa, 1964; Samuelson, 1964). Because of differences in relative productivity (tradable sector vs nontradable sector) between two countries and the ensuing differences in relative prices, the real exchange rate tends to deviate from the PPP path. A country with high productivity growth in the tradable sector has higher inflation in non-tradable goods (a sector with low productivity). Consequently, this country’s currency appreciates in real terms vs the currency of a country with lower relative productivity (i.e., with lower relative inflation). PPP is the most straightforward approach, but it raises many objections. First, PPP as a measure of an equilibrium exchange rate is incomplete. The relative PPP is based on changes in the price levels. Thus the assessment of exchange rate would require choosing some base period as equilibrium (Bayoumi et al., 1994). Second, it fails to take into explicit account major changes in economic policies or in the economic structure. It also does not allow real variables to affect the equilibrium exchange rate (MacDonald, 2000). Finally, this approach is sensitive to the chosen price indicator – different price indices may yield quite different results (Isard et al., 2001) – see Figure 9.1. Consequently, Williamson (1994) and MacDonald (2000) claim that PPP is not a good metric to measure currency misalignment. The former researcher stated strongly that the PPP criterion should be rejected not just as a conceptually incorrect basis on which to estimate the equilibrium exchange rate, but also as not even providing a useful empirical first approximation.
2.2 FEER The notion of fundamental equilibrium exchange rate (FEER), popularized by Williamson (1985), is based on the idea of internal and external macroeconomic balance. The former is defined in terms of output at the full employment level coupled with low and sustainable inflation, and the latter in terms of a sustainable and desired net flow of capital between countries that are internally balanced (Clark and MacDonald, 1998). The FEER shows the exchange rate that would prevail under ‘ideal economic conditions’ that is, a rate consistent with internal and external equilibrium. Thus, this approach should be viewed as normative. It simply boils down to the calibration of the exchange rate at a set of well-defined economic conditions (Clark and MacDonald, 1998). In this context, the FEER is a comparative static, partial equilibrium approach.
Poland and Exchange Rate Parity
183
10.0 5.0 0.0 –5.0 –10.0 –15.0 –20.0 19 94 : 19 1 94 : 19 3 95 : 19 1 95 : 19 3 96 : 19 1 96 : 19 3 97 : 19 1 97 : 19 3 98 : 19 1 98 : 19 3 99 : 19 1 99 : 20 3 00 : 20 1 00 : 20 3 01 : 20 1 01 : 20 3 02 : 20 1 02 :3
–25.0
REER_OECD
Figure 9.1 year)
REER2
REER2_PPI
Measures of REER rates for Poland, 1994–2002 (% change, year on
Notes: 1. OECD REER – based on 40-currency basket deflated with consumer prices. 2. REER2 – based on the euro and USD deflated with consumer prices. 3. REER2_PPI – based on the euro and USD deflated with producer prices. Source: OECD, author’s calculations based on NBP and ECB data.
The FEER is obtained in the balance of payments framework, where the current account balance (CA) is squared with the capital account balance (KA):1 CA ≡ −KA.
(9.1)
Assuming that the ‘sustainable’ current account balance is determined by domestic and foreign demand at full employment and the real effective exchange rate, the solution for FEER can be found by solving the following model: CA(FEER∗ , Yd∗ , Yf∗ ) = −KA∗ ,
(9.2)
where Y is domestic (d) and foreign (f ) demand; asterisks denote the potential/desired level. Thus, in order to calculate the FEER one would have to know a current account model, estimates of potential output in the home country and abroad as well as the estimate of equilibrium capital flows.
184 Country-specific Monetary Policy and Exchange Rates
The FEER approach does not refer explicitly to the theory of exchange rate determination, but rather states an equilibrium position. As pointed out by MacDonald (2000), this equilibrium position should be viewed as a ‘statistical’ one. Given its stock–flow inconsistency, it cannot represent true steady-state equilibrium. Wren-Lewis (1992) noted that the FEER approach implicitly assumes a convergence of the actual real effective exchange rate to its FEER value. In this context, a medium-run current account theory of exchange rate determination is embedded in this approach. It is simply assumed that any divergence in real exchanges will be eliminated. However, the adjustment process is not described and the concept explains explicitly only the equilibrium position (MacDonald, 2000). The FEER method facilitates simple and transparent calculations with a sensitivity analysis of adopted assumptions. However, this approach disregards changes in policies that affect potential output as well as considerations of asset market equilibrium. The latter omission, as Bayoumi et al. (1994) stressed, leads to an implicit assumption that over the medium term interest rates will settle at their equilibrium. This assumption seems to be a very restrictive one and constraining on monetary policy. The calculated FEER can be used for an assessment of the total exchange rate misalignment, i.e. the misalignment resulting from the departure of macroeconomic variables from their fundamental equilibrium levels (defined in terms of the internal and external balance). Thus this approach makes it impossible to distinguish exchange rate misalignment between random/transitory factors and those stemming form misaligned fundamentals. In a sense, the FEER points to the ideal situation with implicit equilibrium in all markets. Bayoumi et al. (1994) and Isard and Faruqee (1998) clearly stress that plausible estimates of FEER may vary quite substantially. In addition, as Bayoumi et al. (1994) point out, the underlying economic conditions that affect a country’s FEER are subject to changes and thus the computed FEER will not be constant over time.
2.3 BEER The behavioural equilibrium exchange rate (BEER) seeks relations between macroeconomic fundamentals and the exchange rate. Therefore, it can be treated as a theory of exchange rate determination. The estimation of the BEER is usually done in a single-equation model where explanatory variables (fundamentals) are chosen based on beliefs
Poland and Exchange Rate Parity
185
concerning the determinants of the exchange rate (for example, the balance of payments theory, the BS effect, the UIP condition, PPP, etc.). For instance, Baffes et al. (1997) employed terms of trade, an indicator of openness (measured as imports plus exports over nominal GDP), resource balance to GDP (trade balance over GDP in constant prices) and investment share, whereas Clark and MacDonald (1998) used the difference in real interest rates, relative government debt, relative ratios of tradables and non-tradables prices, and net foreign assets. The estimated BEER provides information about the current misalignment. The latter term means a misalignment stemming from transitory and random effects, i.e. factors not treated as ‘fundamental’ determinants of the exchange rate (MacDonald, 2000). The BEER method also makes it possible to calculate a ‘fundamental’ equilibrium exchange rate and in turn total misalignment (as in the notion of FEER). This requires choosing the equilibrium levels of the fundamental variables. Having done such calculations, it can be estimated to what extent the exchange rate misalignment results from the transitory factors and to what extent from misaligned fundamentals. On the practical side, highly demanding data requirements – with regard to both data coverage and length of time series (usually annual or quarterly) – are the main drawbacks of the BEER approach. It is very often the case (especially for transition and developing countries) that the data shortcomings make BEER estimations questionable. However, the data shortcomings can be partially circumvented by application of panel techniques.
3
Empirical estimations
Before turning to empirical estimations, definitions of the real exchange rate will be briefly discussed. Empirical models of equilibrium exchange rates usually employ the real effective exchange rate (REER) – i.e. the weighted nominal exchange rate against currencies of main trading partners deflated by selected price indices. The common practice is to use the geographical structure of a country’s trade as a proxy for weights in the REER. However, the geographical structure does not have to correspond closely to the currency structure of trade transactions as indicated by invoices. This is certainly the case for Poland (see Tables 9.1 and 9.2). The share of trade transactions invoiced in USDs is significantly higher than the actual share of exports/imports to/from the USA. Thus, using the trade structure for weighting REER may introduce a bias.
186 Country-specific Monetary Policy and Exchange Rates
Table 9.1 Geographical and currency structure of Polish exports, 01.95 (% of total)
EU15 UK USA EUR∗ USD PLN GBP
1995
1996
1997
1998
1999
2000
2001
70.0 4.0 2.7 45.3 49.1 0.0 2.8
66.2 3.9 2.3 44.6 49.4 0.0 2.3
64.0 3.8 2.6 45.1 48.5 0.0 2.3
68.3 3.9 2.7 52.3 40.0 2.7 2.1
70.5 4.0 2.8 54.8 36.2 4.2 2.2
69.9 4.5 3.1 55.8 36.2 3.5 2.1
69.2 2.4 2.4 58.2 33.8 4.1 2.1
Note: ∗ Sum of all EU12 currencies. Source: Central Statistical Office (CSO) and National Bank of Poland (NBP).
Table 9.2 Geographical and currency structure of Polish imports, 01.95 (% of total)
EU15 UK USA EUR∗ USD PLN GBP
1995
1996
1997
1998
1999
2000
2001
64.6 5.2 3.9 50.8 41.0 0.0 3.2
63.9 5.9 4.4 52.7 39.5 0.0 3.4
63.8 5.5 4.5 54.6 38.0 0.0 3.3
65.6 4.9 3.8 59.7 32.3 1.5 2.8
64.9 4.6 3.6 58.5 32.2 3.5 2.3
61.2 4.4 4.4 55.8 34.8 3.9 2.0
61.4 4.2 3.4 58.4 32.1 4.6 2.0
Note: ∗ Sum of all EU12 currencies. Source: CSO and NBP.
The selection of a particular price index also affects the value of REER. Given that REERs are usually used as an indicator of a country’s competitiveness, prices, which cover mainly tradables, are more appropriate in this respect. It seems that producer price index (PPI) or unit labour costs (ULC) are better suited for this purpose than the consumer price index (CPI), which is most commonly used to compute the REER. As shown in Figure 9.1, inferences with regard to the zloty appreciation differ quite substantially among different measures of REER. The appreciation of the zloty in 1995, 1998 and 2001 was not that substantial, when measured by the REER based on the PPI, as compared to the one based on the CPI.
Poland and Exchange Rate Parity
187
30 25 20 15 10 5 0 1994
1995
1996
1997
1998
1999
2000
2001
–5 –10 EURO
CPI
PPI
CPI_goods
Figure 9.2 Nominal zloty–euro exchange rate vs relative PPP exchange rate, 1994–2001 (% change, year on year) Notes: 1. Changes refer to annual averages. 2. Euro – % change in nominal zloty–euro exchange rate (euro – synthetic). 3. CPI – % change in relative consumer prices between Poland and the EU15. 4. PPI – % change in relative producer prices between Poland and the EU12. 5. CPI_goods – % change in relative prices of consumer goods between Poland and the EU12. Source: NBP and ECB.
For the purpose of all estimations in this chapter, the REER is constructed as an euro–dollar basket deflated by the corresponding consumer prices. The nominal exchange rate is defined as a unit of domestic currency (the zloty) per one unit of foreign currency (the euro and the USD). Thus, an increase in the REER means a depreciation of the zloty. No formal inferences on equilibrium exchange rates based on the PPP model will be pursed in this chapter. Only an illustration of different dynamics of prices and the nominal zloty–euro exchange rate in terms of relative PPP will be presented. Figure 9.2 shows that changes in the nominal zloty–euro exchange rate do not correspond closely to changes in prices. Discrepancies are larger in the case of consumer prices than in the case of producer prices. This could be indicative of the BS effect.
4
FEER calculations
In order to conduct the FEER calculation for Poland one would have to know the balance of payments model, assumptions on potential output growth in Poland and its main trading partners, and the sum of equilibrium capital flows. These three steps will be dealt with in turn.
188 Country-specific Monetary Policy and Exchange Rates
4.1 The balance of payments model The simplest trade equations define relations between real exports/ imports on the one side and real foreign/domestic demand and REER on the other. Foreign demand is usually proxied by GDP in main trading partners, and domestic demand with GDP of the country under investigation. Unfortunately, in the case of Poland it is difficult to arrive at reliable trade elasticities. They are very sensitive to the model specification and the choice of particular variables. The estimated trade equations relate volume of exports with REER and real GDP in the EU2 (Poland’s main trade partner, see Table 9.1), and volume of imports with REER and Polish real GDP. Various definitions of REERs were tested. In the case of exports (and also of imports, depending on the model specification) the price elasticity was at odds with theoretical expectations. A negative relation of REER and export volume was found (i.e., that the depreciation – a higher REER value – would lead to export contraction). Most probably, the main factor behind the distortion was the Russian crisis. In its aftermath, trade volumes contracted significantly and the zloty depreciated. In the course of 1999 and 2000, the zloty gradually appreciated again and trade volumes increased. The distortion was so big that a simple application of dummy variables did not alter the findings – regardless of the econometric techniques used (VAR, ECM, and single-equation models). All equations were estimated using quarterly data. Apparently, more complex trade equations (for instance, using tradeweighted foreign demand based on world GDP growth, instead of only EU GDP growth, or relating import volume to domestic demand and exports, rather then simply to GDP) could improve the statistical and theoretical properties of these estimations, but would make them more difficult to apply in the case of FEER calculations. Also the short lag structure could be a problem; however, lack of data availability makes it impossible to test for higher lags. In addition, there are reasons to expect that trade elasticities were changing during the transition period in Poland. This could be attributable, among other things, to shifts in the commodity and geographical structure of Polish trade. 4.2 Potential output Potential output is usually defined as the maximal output given the endowment constraints of an economy. Although a theoretical aspect of this notion seems to be clear, its operational side is far from easy. As in the case of the equilibrium exchange rate, potential output is an
Poland and Exchange Rate Parity
189
unobservable variable and its final estimate depends to a large extent on judgemental assumptions or estimations. For instance, the IMF does not pursue a standardized approach for all countries, but rather bases its estimates on the knowledge of country-specific features. One of the methods of estimating the output gap focuses on estimation of a production function (De Masi, 1997). The production function approach aims at identifying specific factors contributing to output growth. Linking supply of production factors (labour, capital and total factor productivity) with output facilitates calculations of the output in a situation where the ratio of utilization is at a potential level. As pointed out by De Masi (1997), this approach is the middle ground between fully structural models and purely statistical measures such as the HP filter. The fully structural models, where the variables under investigation are endogenized, are better theoretically grounded, but are difficult to estimate. Given practical difficulties with estimation of either production function or structural models, purely technical methods of time-series smoothing – like the HP filter – are most commonly used. Such approaches, however, are atheoretical and sensitive to the selected parameters of the methods employed and the time window. The structural or semi-structural estimation of potential output in Poland is a research topic on its own and, therefore, no formal attempts to calculate it are pursued in this chapter.
4.3 Equilibrium capital flows The estimation of equilibrium capital flows is based on the national account identity that squares the capital account balance (which must be equal to the current account balance) and the difference between domestic investment and savings. The most straightforward approach is to set the investment and savings levels consistent with potential output. However, it is difficult to find the criteria needed to select such levels. Williamson (1994) attempted to approximate these variables using investment demands over the cycle, demographic effects on saving behaviour, as well as judgemental criteria of sustainability and consistency. The most common approach, however, focuses on the estimation of the saving–investment norm (Isard and Faruqee, 1998). Using historic data for a panel of countries, the current account balance is regressed on saving–investment determinants. The determinants could include the stage of development (proxied by income per capita), demographic
190 Country-specific Monetary Policy and Exchange Rates
structure (dependency ratio), fiscal position, output gap, and world interest rates (Isard and Faruqee, 1998). Having estimated the coefficients of the saving–investment norm, the equilibrium capital flows (or the equilibrium current account balance) can be calibrated at the ‘equilibrium’ levels of the determinants. The problem is, however, that although the saving–investment norm has been estimated for various countries (for instance, Chinn and Prasad, 2000 analysed developing countries, excluding emerging markets, and Doisy and Herve, 2001 focused on CEECs), there have been no attempts to calibrate the determinants at equilibrium levels. Therefore, no formal analysis of the equilibrium current/capital account balance was undertaken for the purpose of this chapter.
4.4 Results In the face of the aforementioned problems with collecting all the information needed to calculate the FEER, back-of-the-envelope calculations will be undertaken. Their aim is to demonstrate sensitivity of results to the adopted assumptions, rather than to provide precise estimates of FEER based on formal analyses of its determinants. Given this reservation, we try to calculate the FEER for Poland in 2002. Our FEER model adopted the following assumptions. Trade elasticities were calibrated using various trade equation estimations undertaken for Poland and long-run trade elasticities estimated for the G-7 countries in Hooper et al. (1998). As imports and exports in the balance of payments are nominal variables, deflators had to be chosen so as to translate export/import volumes estimated in trade equations into nominal values needed for equalization of the current and financial accounts in the balance of payments framework. For the sake of simplicity, the prices of foreign trade (denominated in USD) were proxied by the product of the CPI in the EU and changes in the USD–euro exchange rate. The rationale behind such approximation is that Polish exporters and importers are believed to be price takers in the international markets and the major bulk of imports/exports are invoiced in euros (see Tables 9.1 and 9.2), whereas the data in the Polish balance of payments are denominated in USD. In order to incorporate other items of the current account into the balance of payments, estimated imports and exports covered not only merchandise trade, but also trade in services and unclassified current transactions (the proxy for cross-border trade and trade in services). As inflows and outflows of unclassified current transactions are not available – only net value – the net item was added to exports. Other
Poland and Exchange Rate Parity
191
assumptions – those on potential output in Poland and the Euroarea, the corresponding price indices, sustainable capital flows as well as income, transfers and errors and omissions needed to close the balance of payments accounting – were chosen based on different imposed coefficients (see Table 9.4). Equilibrium capital flows were set so as to equal approximately to 2 per cent, 3 per cent and 4 per cent of GDP in the subsequent variants 1, 2 and 3. Table 9.3 demonstrates possible FEER values for alternative scenarios. They differ only with regard to assumptions for Poland, i.e. potential GDP, corresponding inflation and sustainable capital inflows. Variant 2 is a baseline scenario and is believed to be most probable. As is clearly visible, the differences among variants are not that large. The REER must depreciate in order to reach the equilibrium in all variants, but the level of misalignment is not high – between 3.7 per cent and 6.9 per cent. Given the assumptions on prices, this requires a depreciation of the nominal exchange rate. The implied nominal zloty–euro exchange rates, as presented in Table 9.3, are higher than the actual exchange rate for 2002 – 3.85. Therefore, these results suggest a nominal overvaluation
Table 9.3
FEER calculations for 2002
GDP in the EU12, % change CPI in the EU12, % change CPI in the US, % change USD/EUR GDP in Poland, % change CPI in Poland, % change Capital account, USD bn Other BoP flows, USD bn FEER, % change EUR/PLN(REER = EUR + USD)∗ EUR/PLN(REER = EUR)∗∗
Variant 1
Variant 2
Variant 3
2.7 2.0 2.5 1.00 3.0 3.0 3.5 1.3 6.9 4.08 3.96
2.7 2.0 2.5 1.00 4.0 3.5 5.5 1.3 5.3 4.04 3.92
2.7 2.0 2.5 1.00 5.0 4.0 7.5 1.3 3.7 4.00 3.88
Notes: 1 All calculations based on constant trade elasticities (see Table 9.4). Annual data. 2 Other balance of payments (BoP) flows comprise: income, transfers, and errors and omissions (forecast value for 2002). ∗ Nominal zloty–euro exchange rate based on the REER comprising the euro and the USD; ∗∗ Nominal zloty–euro exchange rate based on the REER comprising only the euro. Source: Author’s calculations.
192 Country-specific Monetary Policy and Exchange Rates
of the zloty in 2002 by approximately 5.7–3.7 per cent. In the case of calculations based on REER comprising only the euro, the misalignment is lower (2.8–0.7 per cent). In order to demonstrate in a better way the sensitivity of the obtained results, the differences in nominal exchange rate stemming from a change in only one assumption as compared to variant 2 are computed and presented in Table 9.4. The results indicate that FEER calculations are very sensitive to the trade equations’ parameters. The estimated equilibrium capital flows and the USD–euro exchange rate play a significant role too. It should be stressed that the calculations are biased to a large extent by the equilibrium or disequilibrium of the USD–euro exchange rate, a factor that is exogenous to the Polish economy. The results presented in Table 9.4 allow us to draw general conclusions on the potential bias of the calculated FEER given the ceteris paribus assumption. The higher the output in Poland, the higher the inflation, the USD–euro exchange rate and the elasticities of import and export incomes, export and import prices (in absolute values: see Table 9.4). The more depreciated the zloty–euro exchange rate, the greater is this effect. Table 9.4
Sensitivity analysis of FEER calculations for 2002 Variant 2 assumption
Alternative assumption
Variant 2 PLN/EUR
Alternative PLN/EUR
GDP in Poland, % change CPI in Poland, % change Capital account, USD bn USD/EUR
4.0
5.0
4.04
4.09
1.1
3.5
4.5
4.04
4.08
1.0
5.5
7.5
4.04
3.93
−2.7
1.00
0.95
4.04
3.96
−1.9
Export price elasticity Export income elasticity Import price elasticity Import income elasticity
0.90
1.90
4.04
3.96
−1.9
1.60
3.40
4.04
3.95
−2.4
−0.60
−1.60
4.04
3.95
−2.1
1.40
2.60
4.04
4.16
2.8
Source: Author’s calculations.
% change in PLN/EUR
Poland and Exchange Rate Parity
193
While interpreting the results (Tables 9.3 and 9.4), it must be underlined that the nominal zloty–euro exchange rate consistent with the FEER is calculated on the basis of annual changes in the variables used in the model, with the REER being no exception. Thus, they are dependent on the nominal zloty–euro exchange rate in 2001, which was used for calculations as a base period. It stood at 3.67 zlotys per euro and, as will be discussed later (see Section 6), this value may be judged as too high. This finding is consistent with the estimates of the FEER for Poland in 2001 made by Baude et al. (2002). They assessed the overvaluation of the real effective exchange rate of the zloty at 6 per cent. Therefore, using any higher values for the reference zloty–euro exchange rate would result in a higher nominal FEER. On the other hand, the employed USD–euro exchange rate seems rather strong as compared to the actual value in 2002 (94.6 USD/EUR). Consequently, the results are biased upwards (i.e. too depreciated). This is indicated by the FEER value based on the REER comprising only the zloty–euro exchange rate. These considerations, however, do not affect the assessment of the REER misalignment. Thus inferences on the zloty misalignment should also take into account issues of the global consistency and the assessment of the USD–euro exchange rate misalignment.
5
BEER estimations
The estimated BEER model of this chapter draws on models by Baffes et al. (1997), Clark and MacDonald (1998), and MacDonald (2001). It can be described as follows: BEER = f ( prod, tot, rir),
(9.3)
where the explanatory variables are: relative productivity (total labour productivity in Poland and in the Euroarea), Polish terms of trade, and difference between real interest rates in Poland and the Euroarea (3M WIBOR and 3M EURIBOR – synthetic, OECD data). The relative labour productivity refers to the PPP notion of competitiveness and could also proxy the BS effect, although no distinction between labour productivity in the tradable and non-tradable sectors is made. This is consistent with the assumption that labour productivity growth in the latter sector is equal in the Euroarea and Poland. The higher labour productivity in Poland relative to the EU12, the stronger the zloty. Thus the expected sign on this variable should be negative. Terms of trade stand
194 Country-specific Monetary Policy and Exchange Rates
for commodities price shocks and should be negatively correlated with the REER. Finally, differences in real interest rates refer to the notion of uncovered interest rate parity. The variables of the model were tested for stationarity.3 The REER and real interest rate differential turned out to be I(1) variables, though in the former case there is marginal evidence in favour of the trend stationarity hypothesis. The relative productivity variable seems to be I(2) based on ADF test, but the Phillips–Person test proves I(1) properties. On the other hand, terms of trade was found to be I(0). Given the quarterly data for the period from the first quarter of 1995 to the second quarter of 2002, based on the Johansen method, one cointegration vector was found for the VAR(2) system. It has the following form (asymptotic standard errors in parentheses): reer = 1.471 −1.322∗ prod (0.23782)
−1.028∗ tot (0.31186)
−2.221∗ rir , (0.20328)
(9.4)
where reer is the log of the REER, prod the log of relative productivity, tot the log of Polish terms of trade, and rir the log of differences between Polish and the Euroarea’s (synthetic) real interest rates. The signs of the coefficients turned out as expected in equation (9.3). The results should be treated with caution, as some statistical properties of the variables and the system may raise objections with regard to econometric inferences for the co-integration analysis. Before discussion of the results, the reservation must be made that the above equation does not claim to be a perfect model of exchange rate determination in Poland. Its parsimonious specification may suffer from the omitted variable problem and consequently the obtained coefficients may be biased. For instance, the model could be augmented by variables like net foreign assets, long-term interest rates, budget deficit or FDI inflows. The augmentation of the model and formal testing of omitted variable problem is, however, limited by the length of the time series. Also the employment of current variables could be discussed in more detail. For example, more attention could be devoted to the disparity between Polish and euro-zone interest rates. The reference to interest rates in other emerging markets rather than to those in the EMU12 may be more indicative of zloty exchange rate changes. In the run-up to the EMU the possibility of a ‘convergence play’ should also be taken into account. In addition, the results may be biased due to structural changes: the evolution of restrictions on capital flows or the Russian crisis serves as a good example of this. The short time series once again
Poland and Exchange Rate Parity
195
poses an impediment for formal testing of such events. The potential solution is to employ panel models, for instance for all CEECs that are to become EU/EMU members soon. As indicated in the survey by MacDonald (1998), panel estimations tend to yield better results in a economic and statistical sense than single-country estimations (e.g. as proved in Chinn, 2001 and Chinn and Johnston, 1996). It should be stressed that the obtained equation indicates only the current misalignment. No attempt was made to calibrate long-run values for the explanatory variables. Therefore, the results point to a trend value for REER rather than equilibrium in the notion of the FEER approach. Figure 9.3 shows that in 1997 and 1998 the estimated BEER and actual REER were moving in opposite directions. In 1997, there was a significant increase in interest rates in Poland, accompanied by falling inflation. This factor caused the appreciation of the estimated BEER, whereas the actual REER depreciated. In 1998, the trends were reversed. Starting from 1999, the estimated and actual exchange rates tend to move together quite closely. The strong appreciation of the zloty in 2001 is not fully explained by the model, which points to a more moderate strengthening. Thus, if the model used here is the correct one, the 2001 appreciation can be treated as caused by transitory/speculative factors. The results for 2002 are more in line with the actual values. The BEER tends to indicate a slightly more appreciated exchange rate than the 10.0 5.0 0.0 –5.0 –10.0 –15.0 –20.0
19
97 19 :1 97 19 :2 97 19 :3 97 19 :4 98 19 :1 98 19 :2 98 19 :3 98 19 :4 99 19 :1 99 19 :2 99 19 :3 99 20 :4 00 20 :1 00 20 :2 00 20 :3 00 20 :4 01 20 :1 01 20 :2 01 20 :3 01 20 :4 02 20 :1 02 :2
–25.0
BEER
Figure 9.3
REER
Estimated BEER and actual REER, 1997–2001 (% change, year on year)
Note: The REER is based on euro-dollar currency basket deflated with consumer prices. Source: Author’s calculations.
196 Country-specific Monetary Policy and Exchange Rates
actual one, but this comparison is biased due to a base effect in the case of actual REER.
6
What should be the entry exchange rate?
Having surveyed the theoretical concepts of the equilibrium exchange rate and attempted some empirical calculations, we turn to a discussion of the practical guidelines for setting the euro parity. We start by stressing the fact that equilibrium exchange rate concepts refer to real (effective) exchange rates and upon ERM II/EMU entry a nominal exchange rate will be set. This differentiation highlights important conceptual issues surrounding equilibrium exchange rates. In this context, the interactions between nominal exchange rate and prices should be thoroughly investigated. Thus, in assessing the misalignment, the focus should not be placed merely on the nominal exchange rate, but also on price developments. In a perfect world with immediate adjustments, changes in the nominal exchange rate would induce corresponding changes in prices of tradables and there would not be deterioration in competitiveness. However, such a textbook scenario is not the case in reality. Price adjustments to exchange rate changes are believed to be slow, and real exchange rates are driven primarily by nominal exchange rates (at least in the short and medium run). Apparently, the speed of the exchange rate passthrough to domestic prices differs among countries and this issue requires formal testing in order to draw any profound conclusions on the consequences of choosing a particular exchange rate. Theoretical and empirical evidence (more pervasive pricing-to-market effect in economies with monopolistic competition markets, and higher share of non-tradables in the structure of the economy and consumption) suggests a weaker pass-through in developed economies as opposed to developing countries. The bottom line of the estimations pursued in this chapter is that it is difficult to provide precise and reliable empirical estimates of an unobservable variable like the equilibrium exchange rate. Empirical estimations are intrinsically uncertain. The estimates are sensitive to the adopted assumptions and model specifications. Besides, each concept of the equilibrium exchange rate has a different interpretation and conveys slightly different information for policy-makers. Given the above considerations, it is important to decide if model-based equilibrium exchange rates are still useful indicators for selecting the euro parity, which is supposed to be the equilibrium exchange rate.
Poland and Exchange Rate Parity
197
The problem with the FEER concept is that it states the equilibrium exchange rate only for ideal conditions and does not provide any information on the ‘appropriate’ exchange rate for the current economic situation. The concept says nothing about how the equilibrium can be achieved, though the adjustment is implicitly embedded in this approach. It would be naïve to expect that, in the real world, moving the exchange rate (in particular the nominal one) to its equilibrium level would be enough to achieve the internal and external equilibrium for an entire economy. There are at least two reasons for this. First, changing the nominal exchange rate may induce the price adjustments mentioned before, and real effects will be muted. This effect depends on the price stickiness and the mobility in factors of production across the economy. Second, the change in the exchange rate may enable potential growth and full employment to be reached, but not necessarily, as this is not the only factor determining equilibrium in the economy. For instance, the exchange rate change is rather unlikely by itself to force changes in fiscal policy: to cut a high budget deficit structural reforms are also required, not just the depreciation of the domestic currency. In the context of setting nominal parity, it is better, however, to think about the ‘equilibrium exchange rate’ as a level of exchange rate that is consistent with other macro-variables for a given point in time and not necessarily as a steady-state exchange rate in a sense of sustainable equilibrium. In this respect, the BEER approach is better suited. It also solves to some extent the problem of endogeneity by using VAR models. However, given the uncertainty about the exact form of the exchange rate determination model, the current assessment of misalignment based on the BEER model may indicate a misspecification problem, and not the misalignment stemming from random effects (i.e., everything that is not explained by the ‘true’ model). The aim of equilibrium exchange rate estimations is the assessment of the exchange rate misalignment with regard to its equilibrium value (i.e., if it is under- or overvalued). Nevertheless, as pointed by Isard and Faruqee (1998), a complete assessment of exchange rate misalignment should not be based simply on model estimates, but take into account a broader range of macroeconomic issues, such as policy mix and structural factors. It is reasonable to expect that not all short- and medium-run misalignments are destabilizing and need a correcting policy action. Some deviations from equilibrium exchange rates may be due to cyclical fluctuations in macroeconomic variables and should not be treated as harmful to the economy.
198 Country-specific Monetary Policy and Exchange Rates
At this point it is important to understand the consequences of choosing an under-/overvalued exchange rate. Unfortunately, the answer is not an easy one, as it would require a detailed identification of relations between the exchange rate and other macro-variables (such as output growth, employment, inflation, etc.). According to a conventional macroeconomic analysis, an overvalued exchange rate spurs recessionary effects and the undervalued currency expansionary ones. However, empirical research does not provide such clear-cut evidence. The extent to which the exchange rate is misaligned seems to be more important than the fact of misalignment per se. This hypothesis is the key finding of the empirical work by Collins and Razin (1998). They discovered that only very high overvaluation leads to slower GDP growth, and medium and high undervaluation to higher growth. This very fact, coupled with the intrinsic uncertainty of equilibrium exchange rate estimates, gives support to the conjecture that the range of ‘optimal’ exchange rates at which Poland can switch to the euro is quite wide. When analysing the consequences of setting the nominal exchange rate it would also be interesting to investigate the microeconomic and structural effects. Kowalski (2002) notes asymmetric sectoral impacts of the zloty devaluation in the run-up to EMU. He argues that domestic industries with the smallest shares in domestic and foreign sectoral consumption could benefit most from a devaluation. Given the structure of the Polish industry, these are branches characterized by relatively low labour productivity. At this point, it is important to differentiate between the short- and long-run consequences of the entry rate choice. In the long run the choice of nominal entry exchange rate will not be significant. Equilibrium will be achieved via appropriate adjustment in prices and wages. This conjecture is valid only if no changes in the structure of the economy and wealth effects take place. Otherwise, there could be long-term effects as well. Against this background, we would like to stress once again that correcting the current exchange rate misalignment does not solve the problem of how to reach the equilibrium. This also refers to the issue of hysteresis in exchange rates discussed by Bayoumi et al. (1994). The hypothesis states that the current misalignment and the adjustment process impact on the final value of the equilibrium exchange rate. So it is important not only to know where we are standing now, but also how to get to the equilibrium. Understanding this problem is very important in the context of joining EMU. Accession Countries will have to meet the Maastricht criteria, so
Poland and Exchange Rate Parity
199
while devising the path to approach the equilibrium exchange rate, other macro-objectives than simply correcting the nominal exchange rate misalignment will have to be taken into account (for instance, inflation, interest rates and debt targets). Also the reaction of financial markets should be considered. As Reluga and Szczurek (2002) point out, it is possible that the market exchange rate will converge quite rapidly to the announced nominal exchange rate parity if this announcement is fully credible. Such conclusions are based on the Krugman (1991) and Ichikawa et al. (1990) models of a credible exchange rate band. Thus, the credibility of announcement may break the relation between the nominal exchange rate and its fundamental determinants. The nominal convergence of exchange rates was clearly visible in the case of so-called ‘non-core’ countries before their accession to EMU. Figures 9.4 and 9.5 depict these phenomena in the cases of Spain and Portugal. In order to demonstrate the possible policy options of dealing with exchange rate misalignment we will consider the case of the overvalued zloty before entry to ERM II (as it could be assessed based on FEER calculations). Polish authorities, if convinced about the zloty overvaluation, may set the ERM II parity at a depreciated zloty–euro exchange rate. Given the credibility of this announcement, the nominal exchange rate should converge gradually to this level and thus the real exchange rate may also follow suit. However, the degree of the parity depreciation will be crucial for credibility and ensuing price adjustments. The bigger the depreciation, the larger the potential increase in inflation. A possible hike in inflation could spur monetary tightening in order to secure 172 170 168 166 164 162 160 –4
Figure 9.4
–3
–2 Years to EMU
Peseta–euro exchange rate (daily quotations)
Source: Reluga and Szczurek (2002).
–1
200 Country-specific Monetary Policy and Exchange Rates 204 202 200 198 196 194 192 –4
Figure 9.5
–3
–2 Years to EMU
–1
Escudo–euro exchange rate (daily quotations)
Source: Reluga and Szczurek (2002).
meeting the Maastricht inflation criterion. Interest rate hikes could, in addition, make it more difficult to fulfil the interest rate criterion. The nominal depreciation of the zloty could also significantly increase costs of foreign debt servicing (denominated in foreign currencies). Such a move is certainly not desirable for Poland, as it currently has a very high central government budget deficit (estimated at 5.1 per cent of GDP for 2002, and forecast at 4.9 per cent of GDP in 2003), which will have to be reduced when Poland becomes an EU member. Moreover, the ensuing higher interest rates (in the event of inflation pick-up) would additionally increase the burden of domestic debt servicing. Also the impact on the private sector repayments of loans denominated in foreign currencies should be accounted for. Thus, for these reasons the scope for setting a too-depreciated parity is limited. When setting the parity exchange rate it is also important to analyse the political economy of this choice. The ERM II parity and the ultimate fixing rate in EMU must be agreed upon by the ECB and Polish authorities. On the one hand, the Polish government will have incentives to depreciate the nominal exchange rate (based on the belief that, at least in the short run, this will be beneficial to the Polish economy), and on the other hand the ECB will be insisting on appreciation of the zloty in order to prevent a loss in competitiveness of its present member states (also in the short run). Given the relative size of the economies, the Polish side will be more interested in this bargain as Poland may potentially gain or lose relatively more then the eurozone countries. Finally, the analysis of selecting euro parity should be put into a time perspective. All empirical estimations and the quantitative assessments undertaken in this chapter referred to 2002 and recent years. They would
Poland and Exchange Rate Parity
201
be helpful in setting the euro parity in the very near future, but Poland’s entry to ERM II is rather unlikely to happen that soon. According to various official statements, 2004–2005 seems to be the earliest possible date. Thus, model-based calculations should be repeated before making binding decisions on the euro parity. This, however, does not invalidate all the aforementioned qualitative considerations.
7
Conclusions
Upon entry to ERM II and then to EMU, Poland will have to choose the exchange rate parity. The nominal zloty–euro exchange rate should be selected based on some concept of equilibrium exchange rate. Two out of three most common approaches to estimating the equilibrium exchange rate (fundamental and behavioural equilibrium exchange rates) were pursued here for Poland. According to these estimations, the zloty–euro exchange rate in 2002 is not far from the level consistent with the current state of fundamentals (as indicated by BEER), and requires some limited depreciation to be in line with the equilibrium level of fundamentals (as indicated by FEER). The possible FEERs for 2002 range between 3.88 and 4.08 zlotys per euro, depending on the variant and REER definition. Because the zloty exchange rate in 2001 (deemed as too appreciated – based on our BEER estimation and Baude et al.’s, 2002 FEER assessment) was used as a reference value, this range could be biased downwards. The results should be treated with caution, as they were demonstrated to be sensitive to the adopted assumptions and model specifications. In addition, they do not take into account the global consistency and equilibrium in the USD–euro exchange rate. Because the consequences of exchange rate misalignment depend primarily on the degree of this misalignment and because there is intrinsic uncertainty about equilibrium exchange rate estimates, the range of ‘optimal’ exchange rates at which Poland can switch to the euro is quiet wide. In qualitative terms, the lower band of this range could be approximated by the estimated BEER, and the upper band by the FEER (given that output growth is below potential). In addition, the scope for depreciation of the nominal zloty–euro exchange rate is limited by the ensuing costs to the economy, the requirements to meet the Maastricht criteria, and political bargaining. Given that the need to set the euro parity in Poland is not immediate (2004–2005 seems the earliest date of entry to ERM II), there is time to refine and update empirical research on equilibrium exchange rates. In particular, there is scope for strengthening empirical analyses
202 Country-specific Monetary Policy and Exchange Rates
of the models’ underlying assumptions. These exercises would not only contribute to more reliable estimates of equilibrium exchange rates for Poland, but also augment the empirical evidence on the functioning of the entire economy, and thus facilitate the conduct of more informed economic policy. Finally, it must be highlighted that, in the long run, the competitiveness of the Polish economy will depend on the micro-efficiency and flexibility of the markets and macroeconomic policies (in particular fiscal and structural), and not on the nominal zloty–euro parity.
Notes 1. Formally, according to the IMF’s Balance of Payments Manual, it is the financial account that comprises capital flows and not the capital account. 2. In cases where long enough time series are not available for the Euroarea, they are approximated by the corresponding variables for the EU. 3. Detailed results are available in Rawdanowicz (2002).
10 Monetary and Exchange Rate Strategies in Hungary on the Way to the Euro Attila Csajbók
1
Introduction
European integration is a top policy priority in Hungary. The country is set to join the EU in 2004 and aims at becoming a full member of the Euroarea relatively early. In order to achieve this, further progress in nominal convergence is necessary, in which the monetary and exchange rate strategies chosen by the authorities will certainly play an important role. This chapter aims at shedding some light on the past, the present and the future of Hungary’s monetary and exchange rate arrangements. Section 2 describes how the evolution of exchange rate regimes since the beginning of the transition process has led to the current arrangement, that is the ±15 per cent ‘shadow ERM II’ combined with inflation targeting. Section 3 reviews the economic arguments for an early Euroarea entry, assesses the progress to date in nominal convergence, and presents the authorities’ view as well as market expectations about the Euroarea entry date. Section 4 deals with the policy dilemmas related to ERM II participation, touching upon the issues of timing, central parity and bandwidth. Section 5 concludes.
2 The recent history of the Hungarian exchange rate and monetary regimes Ever since the transition process started, the exchange rate has played a key role in macroeconomic policy in Hungary. The prominent role of the exchange rate became all the more pronounced as trade reorientation took place and a rapid increase in the country’s openness began. The history of monetary and exchange rate strategies in Hungary can 203
204 Country-specific Monetary Policy and Exchange Rates
be divided into three different phases: a transitional phase (1989–94), a consolidation phase (1995–2000), and nominal convergence phase (2001–), each of which has been characterized by special challenges the country had to face in those periods. The 1990s were characterized by tight exchange rate management in the form of an adjustable peg in the first half of the decade and a narrow-band crawling peg regime operating between 1995 and 2001. The primary aim of these regimes was to try to maintain external balance. Reducing inflation was a policy goal as well, but only if it did not jeopardize the primary goal, that of external stability. Finding a balance between these potentially conflicting goals was not an easy exercise for monetary policy, especially when fiscal policy did not behave in a supportive manner. 2.1 The adjustable peg (1989–94) The early phase of economic transformation was characterized by a transitionary recession, rising fiscal imbalances and large-scale changes in relative prices. Against this background, the main aim of monetary policy was to provide a nominal anchor in a form of pegged exchange rate. The so-called adjustable peg regime involved frequent devaluations, which, however, didn’t compensate fully for the inflation differential. This policy to some extent limited the upsurge in inflation experienced in other transitional countries. Inflation fluctuated in the range of 18–35 per cent during the 1990–94 period at the cost of deteriorating international competitiveness. 2.2 The crawling peg and ‘sustainable disinflation’ (1995–2001) Mounting macroeconomic imbalances, in particular the serious twin deficits that had accumulated by 1994 and the threat of a prospective currency crisis, forced some measures from both fiscal and exchange rate policies. Together with the implementation of a serious fiscal adjustment package, the forint’s central parity was devalued by 9 per cent and the adjustable peg regime was replaced by a narrow-band (±2.25 per cent) crawling peg with a pre-announced devaluation initially of 25 per cent per annum. Over the ensuing years, as competitiveness recovered, the rate of devaluation was repeatedly reduced (to 2.4 per cent per annum in 2001). As a result, inflation has decreased from around 30 per cent in 1995 to 9–11 per cent in 1999. Nevertheless, reductions in the rate of devaluation were always made in a cautious manner, taking into account fiscal developments and keeping a close
Logarithm of forint per basket (inverted scale)
Monetary and Exchange Rate Strategies in Hungary
205
4.7
4.7
4.8
4.8
4.9
4.9 5.0
5.0 5.1
Asian crisis
Rusian crisis
Brazilian crisis
Nasdaq fall
WTC
5.1
5.2
5.2
5.3
5.3
5.4
5.4
5.5
5.5
5.6
5.6
5.7
5.7
5.8 5.8 01.94 07.94 01.95 08.95 03.96 09.96 04.97 11.97 06.98 12.98 07.99 01.00 08.00 02.01 08.01 03.02 09.02 Date
Figure 10.1 Depreciation of the Hungarian forint against the relevant currency basket and the crawling band, January 1994–February 2003 Notes: Composition of the basket: 50% DEM + 50% USD for August 1993–May 1994; 70% ECU + 30% USD for May 1994–December 1996; 70% DEM + 30% USD for January 1997– December 1998; 70% EUR + 30% USD for January–December 1999; 100% EUR since 2000. Source: MNB.
eye on external balance. The term ‘sustainable disinflation’ was coined around this period and it captured the preferences of policy-makers perfectly well. In the first years of the crawling peg regime, this type of opportunistic disinflation was understandable and probably justifiable as the country had just escaped from a fully fledged currency crisis. Partly as a result of the macroeconomic stabilization and the increased predictability provided by the pre-announced path of the exchange rate, in the 1995–98 period a significant upsurge of FDI inflows took place. This resulted in a sizeable increase in total factor productivity, especially in the tradables sector, bringing with it an increase in competitiveness, higher growth and further trade integration with the EU. Current account deficits were within the sustainable range and, after a temporary slippage before the 1998 elections, fiscal policy was no longer on an expansionary track. It was in this situation that adverse inflationary developments started to emerge. After a gradual decline, inflation became stuck at around
206 Country-specific Monetary Policy and Exchange Rates
% 35
% 35
30
30
25
25
20
20
15
15
10
10
5
5
0 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2003 2004 Year Figure 10.2
CPI inflation in Hungary, 1993–2003
Source: MNB.
10 per cent in 1999–2000 and the success of opportunistic disinflation came to an end. The primary cause of disinflation coming to a halt were supply-side factors, especially the increase in the world price of oil and a regional shock in unprocessed food prices. Coming from the supply side, these factors initially did not trigger a reaction from monetary policy. However, after a while these adverse developments started to feed into inflation expectations and there was a danger of substantial inflation inertia building up. The increased rigidity of inflation expectations implied that further disinflation might become more costly. There was a need for tightened monetary conditions. By that time, a large part of capital controls had been abolished; therefore domestic interest rate policy was not available to solve the problem. The disinflationary impetus should have come from the side of the nominal exchange rate. However, by 2001 the rate of pre-announced devaluation was so small (at 2.4 per cent per annum) that a further decrease would not have been enough to break the inflation inertia. It became obvious that the narrow-band crawling peg exchange rate regime limited the room for manoeuvre of monetary policy. At the same
Monetary and Exchange Rate Strategies in Hungary
207
time, a further incentive for more activist monetary policy has emerged, as the Hungarian parliament modified the Central Bank Act, strengthening safeguards of operational independence and clearly defining the achievement and maintenance of price stability as the single goal of monetary policy. Therefore, in May 2001, the exchange rate band was widened from ±2.25 per cent to ±15 per cent, which was followed by a quick nominal appreciation. At the same time, all remaining capital controls were abolished.
2.3 ‘Shadow ERM II’ and inflation targeting (2001–) With the exchange rate band widened to ±15 per cent and the subsequent abandoning of the crawling peg, Hungary has established an exchange rate mechanism quite close to ERM II. The major remaining difference is the absence of exchange rate policy coordination and the unilateral commitment of the central bank to defend the band. Soon after band widening, the Hungarian central bank (MNB) introduced inflation targeting as a new monetary policy framework and set disinflation targets for 2001 and 2002. They were determined and communicated as the start of a disinflation path that would make early Euroarea entry possible. This move carried the risk that inflation targeting and a fixed exchange rate band, even if it is relatively wide, may conflict with each other in the long run. Nevertheless, there are two reasons which may explain the MNB’s move. First, after such a long history of narrow exchange rate targeting, a band of ±15 per cent seemed almost like a float. There was a need for a new nominal anchor as the new wide band implied much greater exchange rate volatility. In the previous regime the pre-announced rate of devaluation could provide an orientation for exchange rate and inflation expectations, at least in the short term. It was clear that this kind of short-term predictability was over in the new wide band. Inflation targets, if they turned out to be credible, could provide a new anchor for expectations, and possibly for a longer run than the rate of crawl in the previous regime. Second, the ‘shadow ERM II-cum-IT’ regime was intended as a temporary arrangement until ERM II proper, and never meant to be an arrangement for the long run. At the start of the regime, there was a chance that the room for manoeuvre provided by the ±15 per cent band would be sufficient for monetary policy to achieve a disinflation path which would make early Euroarea entry possible. Importantly, fiscal policy was supportive of this disinflation path not only in current terms but
208 Country-specific Monetary Policy and Exchange Rates
looking ahead as well, which was reflected in a two-year budget and a medium-term economic programme. Thus the prospects for policy coordination at the start of the new arrangement looked very favourable, which meant that the exchange rate band was not likely to conflict with the disinflation goals during the temporary period until Euroarea entry was secured. However, the adjustment of the corporate sector to a stronger nominal exchange rate proved to be slower than expected and, as a result, private sector wage dynamics remained robust in both 2001 and 2002. In addition, election year 2002 brought about a fiscal expansion beyond the wildest expectations. These developments meant that an increasingly stronger exchange rate was necessary to achieve the inflation targets. By the end of 2002, the forint’s exchange rate was within a one percentage point distance from the upper edge of the band. The exchange rate reached the upper edge of the band at a time when Hungary and the other candidate countries had successfully closed the accession negotiations with the European Union. This has increased the probability of quick Euroarea entry and probably reduced the risk premia Hungary had to face in the past. As the government communicated that the fiscal expansion will be significantly reversed in 2003, markets had the impression that the authorities would do anything needed to achieve quick Euroarea entry. That required further progress with disinflation, which in turn implied providing more room for monetary policy, that is, a re-valuation of the central parity or a temporary switch to a floating regime. Markets therefore started to speculate heavily against the strong edge of the band and in midJanuary, 2003 these tendencies culminated in a fully fledged speculative attack. In Hungary a joint decision by the government and the central bank is needed to change the exchange rate regime. The government strongly insisted on maintaining the current band, primarily driven by concerns about a further potential squeeze on the export sector. In the end, the MNB succeeded in defending the band by rather aggressive interest rate cuts. The consistency of the exchange rate band and the inflation targeting framework was at least temporarily restored when the MNB admitted that it would most probably overshoot its end-2003 target, but the 2004 target was still achievable. The market’s reading of the authorities’ behaviour during and after the speculative attack was that exchange rate stability (curbing further appreciation) was preferred even if the price was that Euroarea entry could take place somewhat later than expected. This revealed preference
Monetary and Exchange Rate Strategies in Hungary
209
has, of course, fed into market expectations of the entry date and was reflected in an increase of long bond yields. The main lesson from the recent speculative attack for Hungary and the other Accession Countries facing participation in ERM II is that even a wide band can quickly prove to be insufficient to keep an economy on a desired disinflation path if fiscal policy is not supportive. Countries approaching ERM II from a floating regime will surely take into account the Hungarian ‘shadow ERM II’ experience and the episode of the speculative attack when deciding about when and how to join this arrangement. For Hungary itself, the speculative attack has drawn attention to an additional challenge: that of managing a smooth transition from the current exchange rate band to ERM II.
3 The Euroarea entry date: arguments, prospects and expectations It is quite clear that the choice of monetary and exchange rate strategies during the run-up to the euro crucially depend on when Euroarea entry is supposed to take place. It is a key variable that governs fiscal and disinflation paths, options for ERM II entry date and central parity as well as the final conversion rate itself. Any analysis of exchange rate strategies should therefore start with assessing the potential Euroarea entry date. In the following sections I look at the economic arguments for early Euroarea entry, the progress Hungary has so far made in nominal convergence, the official views of the MNB and the government on the Euroarea entry date as well as market expectations about the latter. 3.1 Arguments in favour of early Euroarea entry The MNB has long been proposing that the country should aim at early Euroarea participation. The motivation for this is based on a detailed cost–benefit analysis, which revealed significant net gains in growth if Hungary adopted the euro.1 According to the MNB’s estimates, adopting the euro would increase the country’s long-term growth rate by 0.6–0.9 percentage points, thus speeding up real convergence significantly. Postponing Euroarea entry means that Hungary can start reaping these benefits only with a certain delay. In support of this conclusion, model estimations of the gains accrued to Portugal – a country with some similarities to the Eastern European countries – from Euroarea entry also point to substantial GDP gains for the country (see Pereira, 1999).
210 Country-specific Monetary Policy and Exchange Rates
On the other hand, earlier estimations (see Gaspar and Pereira, 1995) show that those gains would have occurred regardless of EMU (or even EU membership), as long as sustainable macro-policies were followed. On balance, the conclusion defended here is that Hungary should participate in the Euroarea as early as possible, unless a quick entry entails extra costs offsetting the gains. The question is to what extent this move would leave Hungary prone to potentially destabilizing shocks. A careful assessing of traditional OCA criteria shows that Hungary and the Euroarea constitute an optimal currency area. The major sources of the growth gains if Hungary enters the monetary union are higher external trade and increased capital investment as a result of the disappearing exchange rate risk and the more stable business environment the euro is expected to bring about. Recent international empirical research suggests that a currency union with major trading partners will boost a country’s external trade, leading to higher growth via various externalities (such as technology and know-how transfer). Following the lines of Frankel and Rose (2000) and Rose and van Wincoop (2001), the MNB’s cost–benefit analysis used a methodology that has recently gained popularity to gauge such effects on external trade and growth. The approach is based on gravity models and large panel data to estimate the effect of a currency union. The results for Hungary showed that the adoption of the euro may raise GDP growth by 0.55–0.76 of a percentage point over the long run, via the expansion of external trade (on the other hand, Vinhas de Souza, 2002, finds no trade gains from early participation in the Euroarea for its member states: actually, the opposite outcome was observed, as one can also see from the figures in Table 10.1). Domestic interest rates currently contain a risk premium component to compensate non-resident investors for the uncertainty about movements in the exchange rate. The switch to the euro will remove this premium from domestic nominal rates, causing real rates to be lower. A lower level of real interest rates will in turn encourage domestic investment. Furthermore, once Hungary is in the Euroarea, investment can grow unhindered by current account deficits, as depreciation will cease to be a threat to non-resident investors. Thus the adoption of a common currency will help the country maintain macroeconomic equilibrium even in the face of a higher current account deficit, thanks to the removal of a major restraint on investment growth. More buoyant investment will, over the longer term, boost economic growth and accelerate convergence towards the average income level within the EU. To quantify this additional growth, the MNB’s cost–benefit analysis first estimated
Monetary and Exchange Rate Strategies in Hungary
211
Table 10.1 Shares of Euroarea exports and imports in selected countries’ total exports and imports (%)
Hungary Portugal∗ Czech Rep. Belgium Spain Poland Netherlands Austria France Italy Germany Ireland Finland
Exports to Euroarea
Imports from Euroarea
1995
2000
1995
2000
58 65 42 67 62 60 60 58 52 49 44 44 32
70 68 62 62 60 60 57 55 50 46 43 38 33
66 71 61 40 56 56 53 55 52 45 49 35 19
69 62 57 56 54 54 52 49 48 41 37 31 21
Portugal∗ Austria Belgium Czech Rep. Spain Hungary Poland France Italy Germany Netherlands Finland Ireland
∗ 1999 data. Countries are ranked on the basis of 2000 export and import shares
vis-à-vis EMU. Source: OECD database.
the size of the expected reduction in the risk premium following the euro changeover, using information contained in the forint yield curve. Using the decrease in real rates obtained in this way, two different exogenous growth models were employed to estimate the effect on long-term growth in numerical terms. The findings suggest that the common currency will raise the rate of GDP growth by 0.08–0.13 of a percentage point over the longer term via lower real rates of interest. Although they are difficult to quantify, and are not included in the MNB’s cost–benefit analysis, the gains arising from importing a credible monetary policy should also be borne in mind. Hungary still has a relatively high inflation rate, which must be reduced whether the country enters or stays out of the Euroarea. The MNB has a clear mandate and is doing its best to achieve this, but even if it succeeds, preserving credibility could be difficult and costly in a country with a rather poor inflation track record in the past 10 to 15 years. In the light of this, it is not surprising either that public support for the euro in Hungary is among the strongest in the region.2 In such circumstances, it may be welfare-improving to give up monetary independence and import monetary policy from an institution which has already established a sound anti-inflationary reputation.
212 Country-specific Monetary Policy and Exchange Rates
% 250 200 150 100
Figure 10.3
USA
Japan
Poland
UK
Greece
Italy
France
Spain
Denmark
Germany
Finland
Portugal
Austria
Sweden
Czech Rep.
Hungary
Belgium
Netherlands
Ireland
0
Luxembourg
50
Exports plus imports as a percentage of GDP (2001)
Sources: OECD, MNB.
Assessment of the traditional OCA criteria shows that the loss of monetary independence as a stabilization tool does not seem to be a significant cost. Hungary’s openness and trade integration with the Euroarea is already very high, either in terms of quantitative measures such as the share of Euroarea trade in GDP (see Figure 10.3), or on the basis of more qualitative indicators such as the share of intra-industry trade or the penetration of high value-added markets. The structure of the Hungarian economy, i.e. the shares of different sectors in employment and GDP, is broadly similar to the Euroarea average. In other words, the Hungarian economy has made significant progress in structural convergence towards the Euroarea over the past decade. Thus sector-specific shocks tend to have a similar effect in the Euroarea as in Hungary. Hence, common monetary policy responses to such sectoral shocks at the Euroarea level also appear to be optimal in the case of Hungary. There exists a certain degree of asymmetry regarding a few specific industries within manufacturing, but this does not exceed the asymmetry experienced by the current smaller Euroarea members. There is also some evidence that since the mid-1990s the business cycle in Hungary has become largely synchronized with that in the EU, at least as much as the business cycles of peripheral Euroarea members. Provided that this trend is permanent, the anticyclical monetary policy of the
Monetary and Exchange Rate Strategies in Hungary
Table 10.2
213
Shares of value added by economic sector (%) Hungary
Agriculture, hunting and forestry; fishing Manufacturing (including energy) Construction Wholesale and retail trade; repairs; transport, hotels and restaurants Financial intermediation; real estate, renting and business activities Other service activities
Poland
Czech Rep.
4.4
6.4
5.3
28.0
33.4
4.8 22.3
EMU12 weighted average
EMU min
EMU max
2.9
0.8
8.9
36.8
23.2
15.2
32.8
7.9 28.1
4.6 25.2
5.5 21.3
4.2 17.7
7.9 28.3
20.9
8.3
18.0
26.1
17.9
38.6
19.2
15.9
10.1
21.0
17.0
23.9
EMU min
EMU max
Note: 1999 data, except for Hungary (2001), and Ireland (1998). Source: OECD Annual National Accounts (SNAV) database, CSO.
Table 10.3
Shares of employment by economic sector (%) Hungary
Agriculture, hunting and forestry; fishing Manufacturing (including energy) Construction Wholesale and retail trade; repairs; transport, hotels and restaurants Financial intermediation; real estate, renting and business activities Other service activities
Poland
Czech Rep.
EMU12 weighted average
5.9
18.8
5.1
5.3
1.9
18.1
28.1
23.8
30.2
20.4
13.9
23.7
6.4 24.5
7.0 23.4
9.3 25.6
7.3 25.0
6.0 22.3
10.5 28.2
7.2
7.7
10.6
13.0
7.0
23.5
27.9
19.3
19.2
29.1
22.0
36.3
Note: 1999 data, except for Ireland (1997), Portugal (1998), Hungary, Poland and the Czech Republic (2000). Source: OECD Annual National Accounts (SNAV) database, Czech and Polish 2001 Regular Reports, CSO.
214 Country-specific Monetary Policy and Exchange Rates
0.1 0.05 0 –0.05 –0.1
Czech Republic
2001.I.
2001.III.
2000.I.
1999.I.
1999.III.
1998.I.
1997.III.
1997.III.
1997.I.
1996.I.
1996.III.
1995.I.
Hungary
1995.III.
1994.I.
1994.III.
1993.I.
EMU
1993.III.
1992.I.
1992.III.
1991.I.
1991.III.
–0.15
Poland
Figure 10.4 Cyclical component∗ of industrial output in the Euroarea∗∗ and in selected Accession Countries ∗ Natural logarithm of the deviation from trend. ∗∗ An aggregate recalculated for earlier periods using the current composition of EMU.
Source: MNB.
European Central Bank will have the desirable cyclical smoothing effect on the Hungarian economy. These stylized facts point toward a low probability of asymmetric shocks hitting Hungary. On the other hand, being an emerging market with a fully liberalized capital account, keeping a national currency in some occasions may be a potential source of asymmetric shocks instead of a stabilization device. It is well documented that the nominal exchange rate may sometimes behave in sharp contrast to what the economic ‘fundamentals’ would warrant, even in advanced industrial countries. This is all the more so in small emerging economies, where sudden changes in investor sentiment, often induced by financial ‘contagion’, may result in speculative attacks and sharp moves of the nominal exchange rate. The prospect of future Euroarea entry to some extent may insulate Hungary from such adverse emerging market phenomena, but the full elimination of this type of asymmetric shocks will only take place when the euro is introduced. Once Hungary is in the Euroarea, the expected speed-up of convergence and the absence of contagion will ensure that the probability of asymmetric shocks is even lower than today. This implies that the policy
Monetary and Exchange Rate Strategies in Hungary
215
response of the ECB to Euroarea shocks will be in most cases optimal for Hungary as well. Should an asymmetric shock nevertheless occur, the burden of stabilization will be on fiscal policy alone, although the relative flexibility of the Hungarian labour market compared to that in continental Europe makes this task somewhat easier. 3.2 The progress in nominal convergence and official views about the entry date With early Euroarea entry, Hungary can shorten the period in which it is exposed to volatile capital flows and start to realize the additional growth the common currency is expected to bring about. So far there has been a broad consensus between the MNB and the government on the desirability of an early adoption of the euro. This was reflected in the fact that Hungary’s medium-term economic strategy, submitted to the EU in the form of the 2002 Pre-Accession Economic Programme (PEP), was designed to be consistent with Euroarea participation starting as early as 2007, although this was not explicitly stated as a target date. However, it is clear that early Euroarea entry requires further efforts, as inflation has not yet reached the level required to meet the Maastricht criterion and the recent expansionary period in fiscal policy can in fact be viewed as a divergence from a path towards balancing the budget. Hungary almost satisfies the government debt/GDP criterion and is also very close to the critical level in terms of the long-term interest rate Deviation from the long-term interest rate criterion Percentage points
Percentage points
Deviation from the inflation criterion 8 7 6 5 4 3 2 1 0 Greece 1996 % 0 –1 –2 –3 –4 –5 –6 –7 –8 –9 –10
Portugal 1994
Spain 1994
Hungary 2002
Hungary Greece 2002 1996
Figure 10.5
Spain 1994
Portugal 1994
Greece 1996
Ireland 1994
Budget balance/GDP
Ireland 1994
9 8 7 6 5 4 3 2 1 0 –1
% 120 100 80 60 40 20 0
Portugal 1994
Hungary 2002
Ireland 1994
Public debt/GDP
Greece 1996
Ireland 1994
Progress in meeting the Maastricht criteria
Sources: Eurostat, MNB.
Spain 1994
Portugal 1994
Spain 1994
Hungary 2002
216 Country-specific Monetary Policy and Exchange Rates
10 9 8
% of GDP
7 6 5 4 3 2 1 0 1997 Figure 10.6
1998
1999
2000
2001
2002∗ 2003∗∗ 2004∗∗ 2005∗∗
Budget deficit (ESA95 standard)
∗ Preliminary data. ∗∗ Hungarian PEP, 2002.
criterion, although both by a narrow margin only. This suggests that if recent adverse fiscal developments are not quickly reversed, the meeting of these criteria may soon become problematic as well. The progress in nominal convergence in Hungary is in many respects similar to that of the ‘peripheral’ Euroarea members five years before their entry. The only exception is the budget deficit, which was quite close to the 3 per cent of GDP level required by Maastricht in 2000, but deteriorated significantly in 2001–2002. In fact, although the fiscal path envisaged in the 2002 PEP shows the commitment of the Hungarian government to early Euroarea entry, currently it seems that fiscal policy presents a major uncertainty to meeting the Maastricht criteria on time. The budget deficit in 2002, measured according to EU standards, widened to a staggering 9.6 per cent of GDP. This is a much higher figure compared even to the 6 per cent put down in the 2002 PEP only a few months before year end. A fiscal loosening of this magnitude makes the credibility of the commitment to early Euroarea entry fragile. The government is communicating that the necessary fiscal adjustment will take place in 2003. So far the market’s attitude to this communication is positive and the consensus forecast for the budget deficit in 2003 is only slightly higher
Monetary and Exchange Rate Strategies in Hungary
217
than that in the PEP. However, should any serious divergence from the announced fiscal path occur, it could easily trigger an adverse revision in market expectations about the Euroarea entry date. Moreover, Hungary is not alone with this problem among the candidate countries. It seems that in many Accession Countries fiscal policy is strongly influenced by the political cycle, which puts certain limits on its efficiency as a stabilization device. Strong adherence to the Stability and Growth Pact after EU accession is essential to prove that these countries are ready to establish the fiscal discipline necessary to maintain stability within the monetary union. Otherwise financial markets may easily become sceptical about the willingness to enter the Euroarea early. Partly as a result of fiscal developments, recent public announcements of MNB officials have suggested that the Euroarea entry date the central bank thinks of as ‘realistic’ has shifted somewhat since the submission of the 2002 PEP, from 2007 to 2008. The government has not so far explicitly set a target entry date, although the finance minister indicated that a ‘realistic’ date is somewhere between 2007 and 2009.
3.3 Market expectations about the entry date There is limited direct evidence on the market’s view about the most likely Euroarea entry date. It is possible, however, to gauge these expectations indirectly, making use of information in the price of financial market instruments. One way to do so is to compare implied forward interest rates, derived from zero-coupon yield curves in Hungary and in the Euroarea. This approach makes use of the fact that after adopting the euro, Hungarian nominal interest rates will differ from Euroarea nominal rates by only a small default risk premium. Since implied forward rates are indicative of the market’s expectation of future short interest rates, the observed differential of one-year implied forwards in, say, 2009 depends on the probability the market attaches to scenarios in which Hungary is already a full member of the EMU by that year.3 Thus, using implied forward differentials, it is possible to recover the implied probability of Euroarea entry by any given future year. Using further assumptions (about the latest possible Euroarea entry date) it is possible to give point estimates of market expectations about the timing of full EMU participation. Figure 10.7 shows the estimates of the expected entry date. According to these, by the second quarter of 2002 the market had started to expect a Euroarea entry date somewhat before the beginning of 2008.
218 Country-specific Monetary Policy and Exchange Rates
2006
2006
Expected entry date Figure 10.7
03.03
2007 12.02
2007 09.02
2008
06.02
2008
03.02
2009
12.01
2009
09.01
2010
06.01
2010
03.01
2011
12.00
2011
10.00
2012
07.00
2012
04.00
2013
01.00
2013
HP trend
Implied Euroarea entry dates for Hungary
This is roughly in line with the schedule suggested by the MNB and the government, and indicates that the medium-term economic programme focusing on early Euroarea entry had acquired credibility. Meanwhile, the most recent evolution of these estimates shows that as a response to the behaviour of the authorities during and after the 2003 speculative attack, the market has shifted the expected Euroarea entry date by almost a full year to around mid-2008. The issue of the timing of Hungary’s Euroarea entry started to get attention in public debate around the middle of 2001. That was the time when the central bank introduced the inflation targeting regime and argued that the path of inflation targets should be set so that they allow the earliest possible (meaning, at that time, 2006 or 2007) Euroarea entry. Inflation dropped significantly in the last quarter of 2001, safely below the first year-end target set by the MNB. Although this was only partly attributable to the new monetary regime, it may have convinced market participants that the early euro agenda can be taken seriously. More recently, the speculative attack and the response to it by the government and the MNB seems to have changed the optimistic view of the market. The revealing of authorities’ preferences for keeping the existing exchange rate band suggested that disinflation and the road to the euro may take longer than previously expected. It is useful to compare the estimation of the expected entry date presented here with survey information. Data availability is a problem here
Monetary and Exchange Rate Strategies in Hungary
219
60 50 40 30 20 10 0 2007
2008 January 2003
Figure 10.8
2009
2010
February 2003
Survey of macro-analysts on Hungary’s Euroarea entry date∗
∗ Percentage of respondents indicating the given years as the most probable entry date.
Source: Reuters.
as regular surveys with such focus are in short supply. The time-series dimension of such surveys is especially limited, i.e. it is hard to tell how market sentiment about the Euroarea entry date has evolved over the past few years. In January 2003, Reuters launched a new monthly poll of the expected Euroarea entry date among macro-analysts focusing on Hungary. Although so far it has only two observations, it is still interesting, since it shows very well the shift in market expectations that took place after the January speculative attack. In just one month, the percentage of respondents indicating 2007 as the most probable entry date has dropped from 50 to 30 per cent, while that of 2009 or even 2010 increased significantly (see Figure 10.8).
4
Strategic issues related to ERM II participation
Since official statements from European authorities suggest that unilateral euroization would be considered as a breach of the Maastricht Treaty, the Accession Countries’ way to the euro will necessarily lead through participation in ERM II. Accession Countries must make three basic decisions regarding ERM II entry: when, at what rate and in which system (exchange rate band or currency board) they should join.
220 Country-specific Monetary Policy and Exchange Rates
4.1 When to enter? The earliest opportunity for joining ERM II is right after EU accession. Nevertheless, it is not obligatory to start participating in ERM II as soon as this. The only strict requirement is to spend at least two years in the mechanism before joining the Euroarea. Therefore the target Euroarea entry date is a key variable influencing the start of ERM II participation. If an Accession Country is determined to join the Euroarea as soon as possible, it cannot delay ERM II entry much beyond EU accession. However, if EU accession takes place as envisaged in May 2004 and the target Euroarea entry date is 2008 or beyond, the country is not in a rush to enter ERM II for a while. Therefore the decision about ERM II entry depends on the prospects of Euroarea entry (i.e. whether it can be done quickly or will take more time) and on the advantages of ERM II itself compared to the exchange rate arrangements the Accession Countries are operating. If the possible date of Euroarea entry is distant or uncertain and the country operates an alternative regime that works well, there is limited incentive to join ERM II. This particular ‘if it ain’t broke, don’t fix it’ strategy may be relevant in the case of Accession Countries operating successful inflation targeting regimes combined with (managed) floating, like the Czech Republic and Poland. Another example of this strategy could be Sweden, also a successful inflation targeter, and a country which has so far tried to delay ERM II entry and will probably minimize the time spent in the mechanism if it decides to enter the Euroarea. The reason why a managed float-cum-inflation targeting may appear superior to ERM II for these countries is that fixed regimes in the past have often proved to be unsustainable arrangements for the long run. In a recent paper, Begg et al. (2002) argue that these lessons (such as the ERM and Asian crises in the 1990s) should be taken seriously into account when the Accession Countries decide about their ERM II entry. Fixed exchange rates, in the absence of capital controls, are an open invitation for capital inflows, especially in catching-up economies. Large amounts of speculative inflow carry the risk of sudden reversals when investor sentiment changes. Such reversals, often triggered by mere ‘contagion’ and not explained by fundamentals, can evolve into full-scale currency crises. Currency crises may spread into the banking sector and have severe economic consequences which can derail progress towards Euroarea entry. Sticking to flexible exchange rates may to some extent limit the size of
Monetary and Exchange Rate Strategies in Hungary
221
short-term speculative inflows, and therefore may turn out to be the preferred option in some Accession Countries if they are not aiming at the quickest possible Euroarea entry. Unlike Poland and the Czech Republic (see Chapters 8 and 9), Hungary currently operates a fixed exchange rate regime and the authorities are committed to maintaining it. After the huge fiscal slippage in 2002, the chances of early Euroarea entry (meaning 2007 or 2008) are fragile and depend to a large extent on fiscal performance in 2003. When decision time arrives in 2004, the authorities’ behaviour will probably be influenced by the prospects for early Euroarea entry. If fiscal adjustment is safely on track and early entry is possible, then there will be a strong incentive to enter ERM II quickly in order to meet the Maastricht criteria on time. However, if it turns out that the necessary fiscal adjustment did not take place and there is little chance of meeting the fiscal Maastricht criterion in time for early Euroarea entry, then the incentives for ERM II entry may become more ambiguous. The authorities in this situation will face a policy dilemma. On the one hand, the multilateral nature of ERM II and the external audit associated with the system may be an effective mechanism to enforce the necessary fiscal discipline. On the other hand, the same multilateral nature means that the flexibility of exchange rate policy is even more constrained than in a ‘shadow ERM II’. That is because in ERM II even intra-marginal interventions are a common concern and in case realigments in any direction become necessary in the future, they will first have to be negotiated with the eurosystem and partner countries participating in ERM II. Therefore the arguments for quick ERM II entry in this scenario are not clear-cut. Needless to say, with ERM II being a multilateral exchange rate arrangement, if domestic fiscal developments are not firmly in control by the time the country could join the mechanism, the other EU member states may turn out to be reluctant to let Hungary participate in ERM II.
4.2 At what rate to enter? The optimal ERM II central parity, just like the timing, depends on the expected date of Euroarea entry. If the country enters ERM II with a commitment and a fair chance of quick Euroarea entry, then it makes sense to communicate the ERM II central parity as the prospective final euro conversion rate. If such a communication is credible, it may help to avoid future speculation for a realignment. But in this case the central parity should be chosen for
222 Country-specific Monetary Policy and Exchange Rates
the long term, and any potential effects of Euroarea membership on the equilibrium real exchange rate should be already taken into account when deciding about the ERM II central parity. A candidate country should ‘choose’ the final conversion rate so that its real exchange rate at entry is close to its long-run equilibrium level. One might argue that the nominal entry rate does not matter as the real exchange rate would sooner or later adjust to its long-run equilibrium through domestic prices. While this argument is certainly true for the long run, entering the Euroarea at a misaligned real rate would involve temporary welfare losses. Joining at too strong a rate would cause output to dip below potential temporarily while the inflation differential vis-à-vis the Euroarea may be lower than the equilibrium level warranted by BS-type real appreciation. A too weak entry rate, on the other hand, would widen the inflation differential temporarily while causing an output gain, possibly for a couple of years. The trade-off between output and inflation due to a misaligned conversion rate, and consequently its effect on welfare, depend on price rigidities prevailing at the time and after joining the Euroarea. Since it is unclear how pricing behaviour will change after the introduction of the euro, in order to avoid unpredictable welfare losses, candidate countries should try to join at their equilibrium real exchange rates. The major difficulty here is that EMU entry itself is likely to bring about effects that will move the equilibrium real exchange rate. If these movements are large, even in the short run, they should be taken into account when choosing the final conversion rate. If, for example, there are reasons to think that Euroarea membership could in itself quickly appreciate the equilibrium real exchange rate, as could be the case if real interest rates drop permanently and the current account constraint becomes more relaxed, policy-makers may decide to let this effect appear in the nominal exchange rate rather than in inflation. A different scenario is when the expected Euroarea entry date is distant or highly uncertain at the start of ERM II participation. In this case, the central parity will be viewed by the market as a temporary one, leaving open the possibility of future realignments. In such a situation it is impossible credibly to communicate that the central parity may serve as a final conversion rate. It also means that the ERM II central parity can be closer to a medium-term equilibrium exchange rate and should not necessarily incorporate the potential shocks expected after full Euroarea membership. Equilibrium real exchange rates are notoriously difficult to estimate, either for the medium term or for the long term (see Chapter 9). For
Monetary and Exchange Rate Strategies in Hungary
223
Accession Countries, getting reliable estimates of current misalignment is crucial, even before ERM II entry. In estimating equilibrium real exchange rate paths, a number of alternative approaches should be used, so that policy-makers get a well-founded picture of current and future developments. Moreover, the potential effect of Euroarea entry on the equilibrium real exchange rate should also be properly modelled. Here the recent experience of current Euroarea members may be of help.
4.3 The choice of the ERM II bandwidth In principle, there are two basic options for Accession Countries to enter ERM II: either with a ±15 per cent band or a currency board. However, the acceptance of a currency board as an ERM II arrangement is not automatic; it will be assessed on a case-by-case basis. Currency boards which are in operation for a longer period will probably pass this assessment, while it could be rather problematic to enter ERM II with a newly and unilaterally established currency board. Currency boards are rather rigid systems, which carry the risk of much higher volatility in real variables (see Vinhas de Souza, 2002b). Meanwhile, being more reversible, they do not provide the advantages of full euroization, i.e. if the credibility of the system deteriorates, risk premia can be relatively high. Another disadvantage vis-à-vis a more flexible regime is the limited control over the course of inflation. In particular, even if inflation stabilizes at a relatively low level, monetary policy cannot be used to suppress the BS-type inflation differential. Therefore the control over the timing of meeting the Maastricht inflation criterion and, consequently, Euroarea entry is limited in a currency board. It is also possible to negotiate an exchange rate band narrower than ±15 per cent. Such a narrow band is currently in operation in Denmark. However, the narrower the band, the more strongly the concerns about volatile future capital flows to Accession Countries apply. Long-term sustainability is an even bigger problem with a narrow band. Moreover, the handling of any remaining asymmetric shocks is much more a burden on fiscal policy than in a wider-band regime, at a time when it has to go through an adjustment anyway. In addition to this, the relative flexibility of monetary policy provided by a wider band may prove useful in handling inflationary shocks. Therefore, if a floating currency Accession Country chooses to enter ERM II with an exchange rate band, it should go for the maximum width (on the other hand,
224 Country-specific Monetary Policy and Exchange Rates
a CBA country with established credibility, Estonia, seems to have reached a decision to enter the ERM II with a 0 per cent band; see Chapter 7).
5
Conclusions
Hungary has a long history of exchange rate targeting, but the motivation behind exchange rate policy has changed over different phases of the economic transformation process. Currently the medium-term focus of macroeconomic policy is nominal convergence to the Euroarea. Although no official target date has been set, the authorities try to keep the early Euroarea entry (meaning 2007–2008) option on the cards. The country has been operating a ±15 per cent ‘shadow ERM II’ exchange rate mechanism since 2001, combined with direct inflation targeting. Such a combination requires close coordination of fiscal and monetary policies. Indicators of nominal convergence have been encouraging up to 2001, but the significant fiscal expansion that took place in election year 2002, if not quickly reversed, may pose a serious risk for an early Euroarea entry scenario. The extent of progress in fiscal adjustment in 2003 will probably be a key determinant in the timing of ERM II entry.
Notes 1. See Csajbók and Csermely (2002). 2. The European Commission’s ‘Candidate Countries Eurobarometer 2002’ survey showed 70 per cent support for the euro in Hungary, the third highest in the region after Slovenia (85 per cent) and Poland (71 per cent). 3. Formally, FSt,T , the observed one-year forward interest differential for year T , observed in t can be decomposed as the following: FSt,T = (1 − Probt (GMUT ))∗ SpreadTNon-GMU + Probt (GMUT )∗ SpreadTGMU where Probt (GMUT ) is the probability at time t that the market attaches to scenarios in which Hungary is a full member of EMU by year T , SpreadTNon-GMU is the expected interest rate differential if Hungary is not in the Euroarea by year T , while SpreadTGMU is the expected interest rate differential once Hungary is in the Euroarea, i.e. the expected default risk premium. Since FSt,T is observable, with plausible assumptions about SpreadTNon-GMU and SpreadTGMU , one can calculate Probt (GMUT ) for any year T .
References Alesina, A. and S. Ardagna, ‘Tales of Fiscal Adjustments’, Economic Policy, No. 27 (1998), pp. 487–545. Alesina, A. and R.J. Barro, ‘Currency Unions’, Quarterly Journal of Economics, 117(2) (2002), pp. 409–36. Alesina, A., R. Perotti and J. Tavares, ‘The Political Economy of Fiscal Consolidations’, Brookings Papers on Economic Activity (1998). Artis, M.J., ‘Reflections of the Optimal Currency Area (OCA) Criteria in the Light of EMU’, mimeo, European University Institute, Florence (2002). Artis, M.J. and W. Zhang, ‘International Business Cycle and the ERM: Is There a European Business Cycle?’, Discussion Paper No. 1191, CEPR, London (1995). Bacchetta, P. and E. van Wincoop, ‘Does Exchange Rate Stability Increase Trade and Welfare?’, American Economic Review, 90 (2000), pp. 1093–109. Backé, P., ‘The Monetary Integration of EU Accession Countries from Central and Eastern Europe’, mimeo, Bratislava (2002a). Backé, P., ‘Fiscal Effects of EU Membership for Central European and Baltic EU Accession Countries’, Focus on Transition, 7(2) (2002b), pp. 151–64. Backé, P. and C. Wójcik, ‘Unilateral Euroization: A Suitable Road Towards Joining the Euroarea For Central and Eastern European EU Accession Countries?’, mimeo, Estonia (2002). Backé, P., W. Köhler-Töglhofer and F. Schardax, ‘Fiscal Developments in Central and Eastern European EU Accession Countries – An Overview’, mimeo, Vienna (2002). Baffes, J., I.A. Elbadawi and S.A. O’Connell, ‘Single-Equation Estimation of the Equilibrium Real Exchange Rate’, World Bank Country Economics Department Paper No. 1800 (1997). Balassa, B., ‘The Purchasing Power Parity Doctrine: A Reappraisal’, Journal of Political Economy, 72 (1964), pp. 584–96. Bank of England, Economic Models at the Bank of England, Bank of England, United Kingdom (1999). Bank of Slovenia, The Medium-term Monetary Policy Framework, Ljubljana (2001). Barro, R.J. and S. Tenreyro, ‘Economic Effects of Currency Unions’, Working Paper No. 9435, NBER, Cambridge (2003). Basdevant, O. and Ü. Kaasik, ‘The Core of a Macro-economic Model for Estonia’, Working Papers of Eesti Pank No. 6 (2002). Batiz, L.A.R. and A.N.R. Sy, ‘Currency Boards, Credibility, and Macroeconomic Behavior’, IMF WP/00/97 (2000). Baude, J.V., C. Coudert and C. Couharde, ‘Exchange Rate Regimes and Sustainable Parities for CEECs in the Prospect of joining the EMU’, mimeo (2002). Baumgartner, J. et al., Auswirkungen der Wirtschafts- und Währungsunion, Österreichisches Institut für Wirtschaftsforschung, Vienna (1997).
225
226 References Bayoumi, T. and B. Eichengreen, ‘Shocking Aspects of European Monetary Unification’, in F. Torres and F. Giavazzi (eds), Growth and Adjustment in the European Monetary Union, Cambridge University Press (1993), pp. 193–230. Bayoumi, T. and B. Eichengreen, ‘Ever Closer to Heaven? An Optimum-Currency Area Index for European Countries’, European Economic Review, 41(3–5), April (1997), pp. 761–70. Bayoumi, T. and R. MacDonald, ‘Deviations of Exchange Rates from PPP: A Story Featuring Two Monetary Unions’, IMF Working Paper No. WP/98/69, May (1998). Bayoumi, T., P. Clark, S. Symansky and M. Taylor, ‘The Robustness of Equilibrium Exchange Rate Calculations to Alternative Assumptions and Methodologies’, in J. Williamson (ed.), Estimating Equilibrium Exchange Rates, Institute for International Economics, Washington (1994), pp. 19–59. Begg, D., B. Eichengreen, L. Halpern, J. von Hagen and C. Wyplosz, ‘Sustainable Regimes of Capital Movements in Accession Countries’, CEPR Policy Paper No. 10 (2002). Benhabib, J. and R.E.A. Farmer, ‘The Monetary Transmission Mechanism’, EUI Working Paper ECO No. 99/35 (1999). Bergvall, A., ‘Exchange Rate Regimes and Macroeconomic Stability: The Case of Sweden, 1972–1996’, mimeo, Uppsala University, Sweden (2000). Bernanke, B.S. and A.S. Blinder, ‘Credit, Money, and Aggregate Demand’, American Economic Review, 78(2) (1988) pp. 435–9. Bernanke, B.S. and M. Gertler, ‘Inside the Black Box: the Credit Channel of Monetary Policy Transmission’, NBER Working Paper No. 5146 (1995). BIS, ‘The Transmission of Monetary Policy in Emerging Market Economies’, BIS Policy Papers No. 3 (1996). Bofinger, P. and T. Wollmerhäuser, ‘Is There a Third Way to EMU for the EU Accession Countries?’, Economic Systems, 25(3) (2001), pp. 253–74. —————, ‘Managed Floating: Understanding the New International Monetary Order’, Discussion Paper No. 3064, CEPR, London (2001). Bole, V., ‘Sterilization in a Small Open Economy: The Case of Slovenia’, Working Paper No. 1, EIPF, Ljubljana (1994). Boone, L. and M. Maurel, ‘Economic Convergence of the CEECs with the EU’, Discussion Paper No. 2118, CEPR, London (1998). Boone, L. and M. Maurel, ‘An Optimal Currency Area Perspective of the EU Enlargement to the CEECs’, Discussion Paper No. 2119, CEPR London (1999). Bratkowski, A. and J. Rostowski, ‘Unilateral adoption of the Euro by EU Applicant Countries: the Macroeconomic Aspects’, in The Ten Years of Transition, Edward Elgar (2001), pp. 125–47. Buiter, W.H., ‘Optimal Currency Areas: Why does the Exchange Rate Regime Matter?’, CEPR Discussion Paper No. 2366, January (2000). Buiter, W.H., ‘Anchor, Float or Abandon Ship: Exchange Rate Regimes for Accession Countries’, CEPR Discussion Paper No. 3184, January (2002). Buiter, W.H. and C. Grafe, ‘Central Banking and the Choice of Currency Regime in Accession Countries’, SUERF, Vienna (2001). Calmfors, L., H. Flam, N. Gottfries, J. Haaland, M. Matlary, M. Jerneck, R. Lindahl, Ch. Nordh Berntsson, E. Rabinowicz, and A. Vredin, EMU – A Swedish Perspective, Kluwer, Boston (1997).
References
227
Calvo, G.A. and C. Reinhart, ‘The Fear of Floating’, NBER Working Paper No. 7993 (2000). Calvo, G.A. and C.A. Vegh, ‘Exchange Rate-based Stabilisation under Imperfect Credibility’, in A. Worgotter and H. Frisch (eds), Open-economy Macroeconomics, Macmillan Press, London (1993), pp. 3–28. Calvo, G.A. and C.A. Vegh, ‘Inflation Stabilisation and BOP Crises in Developing Countries’, Working Paper No. 6925, NBER, Cambridge, MA (1999). Campa, J.M. and L.S. Goldberg, ‘Exchange Rate Pass-Through into Import Prices: A Macro or Micro Phenomenon?’ Federal Reserve Bank of New Work, Staff Reports No. 149 (2000). Canzoneri, M.B., R.E. Cumby and B.T. Diba, ‘Fiscal Discipline and Exchange Rate Systems’, The Economic Journal, 111 (2001), pp. 667–90. Canzoneri, M.B., R.E. Cumby and B.T. Diba, ‘Relative Productivity and the Real Exchange Rate in the Long Run: Evidence for a Panel of OECD Countries’, CEPR Discussion Papers No. 1464 (1996). Caprirolo, G. and V. Lavraˇc, ‘Managing Capital Inflows in Slovenia: Experience and Options’, Working Paper No. 3/2001, ICEG, Budapest (2001). Carlin, W., A. Glyn and J. van Reenen, ‘Export Market Performance of OECD Countries: An Empirical Examination of the Role of Cost Competitivnes’, Economic Journal, 111(1) (2001), pp. 128–62. Cecchetti, S.G., ‘Legal Structure, Financial Structure, and the Monetary Policy Transmission Mechanism’, Federal Reserve Board of New York, Economic Policy Review, July (1999) pp. 9–28. Charemza, W., ‘LAM models for East European economies: General description’, Discussion Paper, University of Gdansk (1994). Chinn, M.D. and E.S. Prasad (2000), ‘Medium-Term Determinants of Current Accounts in Industrial and Developing Countries: An Empirical Exploration’, IMF Working Paper No. WP/00/46 (2000). Chinn, M.D., ‘Sectoral Productivity, Government Spending and Real Exchange Rates: Empirical Evidence for OECD Countries’, in R. MacDonald and Stein (eds), Equilibrium Exchange Rates, Kluwer (2001). Chinn, M.D. and L. Johnston, ‘Real Exchange Rate Level, Productivity and Demand Shocks: Evidence from a Panel of 14 Countries’, NBER Discussion Paper No. 5709 (1996). Cincibuch, M. and D. Vavra, ‘Toward the European Monetary Union: A Need for Exchange Rate Flexibility?’, Eastern European Economics, 39(6) (2001), pp. 23–64. Clark, T.E. and E. van Wincoop, ‘Borders and Business Cycles’, Journal of International Economics, 55(1) (2001), pp. 59–85. Clark, P.B. and R. MacDonald, ‘Exchange Rates and Economic Fundamentals: A Methodological Comparison of BEERs and FEERs’, IMF Working Paper No. WP/98/67 (1998). Collins, S.M. and O. Razin, ‘Real Exchange rate misalignments and growth’, NBER Discussion Paper No. 6174, September (1998). Corden, M., ‘The Adjustment Problem’, in L. Krause and W. Salant (eds), European Monetary Unification and Its Meaning for the United States, The Brookings Institution, Washington, DC (1973). Coricelli, F., B. Jazbec and I. Masten, ‘Exchange Rate Pass-through in Candidate Countries’, mimeo, Budapest (2002).
228 References Csajbók, A. and Á. Csermely (eds), ‘Adopting the Euro in Hungary: Expected Costs, Benefits and Timing’, MNB Occasional Papers No. 24 (2002). Cukierman, A., P. Geoffrey and B. Neyapti, ‘Central Bank Reform, Liberalization and Inflation in Transition Economies – An International Perspective’, Tel-Aviv University Working Paper (2001). Davidson, R. and J.G. MacKinnon, Estimation and Inference in Econometrics, Oxford University Press, New York (1993). de Masi, P.R., ‘IMF Estimates of Potential Output: Theory and Practice’, IMF Working Paper WP/97/177 (1997). Dixit, A., ‘A Repeated Game Model of Monetary Union’, Economic Journal, 110, October (2000), pp. 759–80. Doisy, N. and K. Herve, ‘The Medium and Long Dynamics of the Current Account Positions in the Central and Eastern European Countries: What are the Implications for their Accession to the European Union and the Euroarea?’, mimeo (2001). Dornbusch, R., ‘Expectations and Exchange Rate Dynamics’, Journal of Political Economy, M1304 (1976), pp. 1161–76. Dvorsky, S., ‘Measuring Central Bank Independence in Selected Transition Countries’, Focus on Transition, 5(2) (2000). EBRD, Transition Report, London (several years). ECB, The Euro bond market, Frankfurt am Main, July (2001). ECOFIN Statement, ‘Exchange-Rate Strategies for Accession Countries’, 7 November (2000). Edwards, S., ‘The Determinants of the Choice between Fixed and Flexible Exchange Rate Regimes’, NBER Working Paper 5756 (1996). Edwards, S., ‘Exchange Rate Anchors and Inflation: A Political Economy Approach’, in S.C.W. Eijffinger and H.P. Huizinga (eds), Positive Political Economy, Cambridge University Press (1998), pp. 187–216. Egert, B., ‘Investigating the Balassa–Samuelson Hypothesis in Transition: Do We Understand What We See?’, Discussion Paper No. 6/2002, Bank of Finland, Helsinki (2002). Eichengreen, B., ‘Should the Maastricht Treaty be Saved?’, Princeton Studies in International Finance, No. 74, Princeton University (1992). Eichengreen, B., European Monetary Unification: Theory, Practice and Analysis, MIT Press, Cambridge MA (1997). Eichengreen, B., ‘When to Dollarize’, mimeo. University of California, Berkeley (2000). Engle, R.F. and C.W.J. Granger, ‘Co-integration and Error Correction: Representation, Estimation and Testing’, Econometrica, 55 (1987), pp. 251–76. Estrin, S. and G. Urga, ‘Convergence in Output in Transition Economies: Central and Eastern Europe, 1970–95’, Discussion Paper No. 1616, CEPR, London (1997). Fatás, A., ‘EMU: Countries or Regions?’, European Economic Review, No. 41 (1997), pp. 753–60. Fidrmuc, J. and J. Fidrmuc, ‘Disintegration and Time’, Working Paper No. B24– 2001 (2001). Fidrmuc, J. and F. Schardax, ‘The Pre-Ins Ante Portas – EMU Enlargement, Optimum Currency Area, and Nominal Convergence’, Focus on Transition, 5(2) (2000).
References
229
Fidrmuc, J., J. Horvath and J. Fidrmuc, ‘Stability of Monetary Unions: Lessons from the Breakup of Czechoslovakia’, Journal of Comparative Economics, 27(4) (1999), pp. 753–81. Fidrmuc, J., ‘Intraindustry Trade between the EU and the CEECs: The Evidence of the First Decade of Transition’, Focus on Transition, 6(1) (2001), pp. 65–78. Fidrmuc, J., ‘Migration and Adjustment to Shocks in Transition Economies’, mimeo, Vienna (2002). Fidrmuc, J. and I. Korhonen, ‘Similarity of Supply and Demand Shocks between the Euroarea and the CEECs’, BOFIT Discussion Paper No. 14, Bank of Finland, Institute for Economies in Transition, Helsinki (2001). Fischer, S., R. Sahay and C.A. Vegh, ‘Modern Hyper- and High Inflations’, Working Paper No. 8930, NBER, Cambridge, MA (2002). Fleming, M., ‘Domestic Financial Policies under Fixed and under Floating Exchange Rates’, IMF Staff Papers No. (1962), pp. 369–80. Fontagné, L., ‘Endogenous Symmetry of Shocks in a Monetary Union’, Open Economies Review, 10 (1999), pp. 263–87. Frankel, J.A., ‘No Single Currency Regime is Right for All Countries or All Times’, Working Paper No. 7338, NBER, Cambridge (1999). Frankel, J.A. and A.K. Rose, ‘Is EMU More Justifiable Ex Post than Ex Ante?’, European Economic Review, No. 41 (1997), pp. 753–60. Frankel, J.A. and A.K. Rose, ‘The Endogeneity of the Optimum Currency Area Criteria’, Economic Journal, No. 108, July (1998), pp. 1009–25. Frankel, J.A. and A.K. Rose, ‘Estimating the Effect of Currency Unions on Trade and Output’, Working Paper No. 7857, NBER, Cambridge (2000). Friedman, M., ‘The Role of Monetary Policy’, American Economic Review, 59 (1968). Fuhrer, J. and G.R. Moore, ‘Monetary Policy Trade-offs and the Correlation between Nominal Interest Rates and Real Output’, American Economic Review, 85(1) (1995) pp. 219–39. Gaspar, V. and F. Mongelli, ‘Monetary Unification and the Single Market’, mimeo (2001). Gaspar, V. and A. Pereira, ‘The Impact of Financial Integration and Unilateral Public Transfers on Investment and Growth in EC CapitalImporting Economies’, Journal of Development Economics, No. 46 (1995), pp. 43–66. Ghosh, A.R., A.M. Gulde, J.D. Ostry and H.C. Wolf, ‘Does the Nominal Exchange Rate Regime Matter?’, NBER Working Paper 5874 (1997). Ghosh, A.R., A.M. Gulde and H.C. Wolf, ‘Exchange Rate Regimes: Classification and Consequences’, in A. Ghosh, A. Gulde and H. Wolf (eds), Exchange Rate Regimes, Choices and Consequences, MIT Press (2002). Giavazzi, F. and M. Pagano, ‘Can Severe Fiscal Contractions be Expansionary? Tales of Two Small European Countries’, NBER Working Paper W3372 (1990). Giavazzi, F. and M. Pagano, ‘Non-Keynesian Effects of Fiscal Policy Changes: International Evidence and the Swedish Experience’, Swedish Economic Policy Review (1996), pp. 67–103. Golinelli, R. and R. Rovelli, ‘Painless Disinflation? Monetary Policy Rules in Hungary, 1991–1999’, The Economics of Transition, 10(1) (2002), pp. 1–27. Grafe, C. and C. Wyplosz, ‘A Model of Real Exchange Rate Determination in Transition Economies’, in M.I. Blejer and M. Škreb (eds), Balance of Payments,
230 References Exchange Rates and Competitiveness in Transition Economies, Kluwer, Dordrecht (1997). Haldane, A. and V. Read, ‘Monetary Policy Surprises and the Yield Curve’, Working Paper, Bank of England (2000). Hall, S., G. Myzon and A. Welfe, ‘Modelling Economies in Transition: An Introduction’, Economic Modelling, 17(3) (2000) pp. 339–57. Hallett, A.H. and L. Piscitelli, ‘The Endogenous Optimal Currency Area Hypothesis: Will a Single Currency Induce Convergence in Europe?’, mimeo, University of Durham (2001). Hanke, S.H., ‘Currency Boards’, The Annals of the American Academy, 579, January (2002), pp. 87–105. Hanke, S.H. and K. Schuler, Currency Boards for Developing Countries: A Handbook, San Francisco, ICS Press (1994). Hansson, A. and J.F. Helliwell, ‘The Evolution of Income and Competitiveness in the North Pacific Rim’, Transactions of the Royal Society of Canada, Series 1, Vol. 1 (1990). Hardy, D., ‘Anticipation and Surprises in Central Bank Interest Rate Policy: The Case of the Bundesbank’, IMF Staff Papers, 45 (1998), International Monetary Fund, Washington, DC, pp. 647–71. Harrod, R.J., International Economics, London, James Nisbet and Cambridge University Press (1993). Hausmann, R., M. Gavin, C. Pages-Serra and E. Stein, ‘Financial Turmoil and the Choice of Exchange Rate Regime’, Inter-American Development Bank Working Paper 400 (1999). Helpman, E., ‘Imperfect Competition and International Trade: Evidence from Fourteen Industrial Countries’, Journal of Japanese and International Economies, No. 1 (1987), pp. 62–81. Hochreiter, E. and H. Wagner, ‘The Road to the Euro – Exchange Rate Arrangements in European Transition Economies’, Annals of the American Academy of Political and Social Science, No. 579 (2002), pp. 168–82. Hochreiter, E. and G. Winckler, ‘The Advantages of Tying Austria’s Hands: The Success of the Hard Currency Strategy’, European Journal of Political Economy, No. 11 (1995), pp. 83–111. Honohan, P., ‘Discussion’, Economic Policy, 33 (2001). Horvath, J., ‘The Optimum Currency Area Theory: A Review’, Central European University Budapest, mimeo (2001a). Horvath, J., ‘Supply and Demand Shocks in Europe: Large-4 EU Members’, Visegrad-5 and Baltic-3 Countries, Central European University Budapest, mimeo (2001b). Hodrick, R.J., ‘Risk, Uncertainty and Exchange Rates’, Journal of Monetary Economics, 23 (1989), pp. 433–59. Hooper, P., K. Johnson and J. Marquez, ‘Trade Elasticities for G-7 Countries’, International Finance Discussion Papers No. 609, Board of Governors of the Federal Reserve System, April (1998). Hummels, D. and J. Levinsohn, ‘Monopolistic Competition and International Trade: Reconsidering the Evidence’, Quarterly Journal of Economics, 110(3) (1995), pp. 799–836. Ichikawa, M., M. Miller and A. Sutherland, ‘Entering a Preannounced Currency Band’, Economics Letters, 34 (1990), pp. 363–8.
References
231
IMAD, ‘Spring Report 2002: Medium-term Forecasts for 2004–2006 and Implementation of the SEDS’, Ljubljana (2002). Imbs, J., ‘Co-Fluctuations’, Discussion Paper No. 2267, CEPR, London (1999). IMF, World Economic Outlook, various years. IMF, Currency Board Arrangements: Issues, Experiences and Implications for FundSupported Programs (1996). IMF, ‘Republic of Slovenia: Article IV Consultation – Staff Report’, IMF Country Report 01/76, Washington, DC (2001). IMF, ‘The effectiveness of Fiscal Policy in Stimulating Economic Activity – A Review of the Literature’, IMF SM/00/66 (2000). Isard, P. and H. Faruqee (eds), ‘Exchange Rate Assessment. Extensions of the Macroeconomic Balance Approach’, IMF Occasional Paper No. 167, December (1998). Isard, P., H. Faruqee, G.R. Kincaid and M. Fetherston, ‘Methodology for Current Account and Exchange Rate Assessments’, IMF Occasional Paper No. 209, December (2001). Judge, G.G., R.C. Hill, W.E. Griffiths, H. Lütkepohl and T. Lee, Introduction to the Theory of Econometrics, New York, Wiley (1998). Juszczak, G., M. Kazmierska, N., Lapinska-Sobczak, N. and W. Welfe, ‘Quarterly Model of the Polish Economy in Transition (with Special Emphasis on Financial Flows)’, Economic Modelling, 10(2) (1993), pp. 127–49. Kaasik, Ü., P. Luikmel and M. Randveer, ‘The Relationship between the Estonian Business Cycle and the Business Cycles of the Main Trading Partners’, mimeo, Eesti Pank (2002). Kalemli-Ozcan, S., B.E. Sørensen and O. Yosha, ‘Economic Integration, Industrial Specialization, and the Asymmetry of Macroeconomic Fluctuations’, Journal of International Economics, 55(1) (2001), pp. 107–37. Kamin, S., P. Turner and J. Van’t Dack, ‘The Transmission Mechanism of Monetary Policy in Emerging Market Economies: an Overview’, BIS Policy Papers (1998), pp. 5–64. Kamps, C., ‘Fiscal Consolidation in Europe: Pre- and Post-Maastricht’, Kiel Working Paper No. 1028 (2001). Kangur, A., T. Rajasalu and M. Randveer, ‘Kapitali Liikumine ja Ettevõtluse Rahastamine’, Eesti Panga Toimetised no. 4 (1999). Kashyap, A.K. and J.C. Stein, ‘Monetary Policy and Bank Lending’, NBER Working Paper Series, Working Paper No. 4317 (1993). Kenen, P.B., ‘The Theory of Optimum Currency Areas: An Eclectic View’, in R.A. Mundell and A.K. Swoboda (eds), Monetary Problems of the International Economy, Chicago, USA, University of Chicago Press (1969). Kenen, P.B., ‘Currency Areas, Policy Domains, and the Institutionalization of Fixed Exchange Rates’, mimeo, Princeton University (2000). Klein, L., A. Welfe and W. Welfe, ‘Principles of Macroeconometric Modeling’, Advanced Textbooks in Economics, No. 36, Amsterdam: Elsevier Science BV (1999). Koˇcenda, E., ‘Macroeconomic Convergence in Transition Countries’, Journal of Comparative Economics, 29(1) (2001), pp. 1–23. Kopits, G., ‘Implications of EMU for Exchange Rate Policy in Central and Eastern Europe’, Working Paper 99/9, IMF, Washington (1999). Kopits, G. and I. Székely, ‘Fiscal Policy Challenges of EU Accession for Central European Accession Countries’, mimeo, Vienna (2002).
232 References Korhonen, I., ‘Some Empirical Tests on the Integration of Economic Activity between the Euroarea and the Accession Countries: A Note’, Economics of Transition, 11(1) (2003), pp. 1–20. Kose, M.A. and K. Yi, ‘International Trade and Business Cycles: Is Vertical Specialization the Missing Link?’, American Economic Review, 91(2) (2001), pp. 371–5. Kose, M.A., E.S. Prasad and M.E. Terrones, ‘How does Globalization Affect the Synchronization of Business Cycles?’, forthcoming in American Economic Review (2003). Kowalski, P., ‘Polish accession to the EMU: do levels of the central parity and the ultimate conversion rate matter?’, mimeo (2002). Kravis, I.B., ‘The Three Faces of the International Comparison Project’, Research Observer, No. 1, January (1986). Krugman, P., ‘Target Zones and Exchange Rate Dynamics’, The Quarterly Journal of Economics, 106(3) (1991), pp. 669–82. Krugman, P., ‘Lessons of Massachusetts for EMU’, in F. Torres and F. Giavazzi (eds), Adjustment and Growth in the European Monetary Union, Cambridge University Press and CEPR, UK (1993), pp. 241–61. Krzak, M. and H. Ettl, ‘Is Direct Inflation Targeting an Alternative for Central Europe? The Case of the Czech Republic and Poland’, in Focus on Transition, No. 1 (1999), pp. 28–59. Kuijs, L., ‘Monetary Policy Transmission Mechanisms and Inflation in the Slovak Republic’, IMF WP/02/80 (2002). Kuttner, K. and A. Posen, ‘Does Talk Matter after All? Inflation Targeting and Central Bank Behavior’, mimeo (2001). Lahiri, A. and C.A. Vegh, ‘Delaying the Inevitable: Optimal Interest Rate Policy and BOP Crises’, Working Paper No. 7734, National Bureau of Economic Research, Cambridge MA (2000). Lättemäe, R., ‘Monetary Transmission Mechanism in Estonia – Some Theoretical Considerations and Stylized Aspects’, Working Papers of Eesti Pank No. 4 (2001). Lättemäe, R., ‘Decomposing Estonian Nominal Interest Rates into Expected Inflation and ex ante Real Interest Rates using a Structural VAR Methodology’, mimeo, Eesti Pank (2002). Lättemäe, R., M. Randveer and U. Sepp, ‘The History and Sustainability of the Estonian Currency Board Arrangement’, in The Choice of Exchange Rate Regime and Monetary Policy Rule in Accession to EU and in Entry to EMU, Eesti Pank, Tallinn (2002). Lehment, H., ‘European Fiscal Polices Under the Stability Pact – Some First Insights’, Kiel Working Paper No. 1098 (2002). Leitemo, K. and Ø. Røisland, ‘Effects on the Traded and Non-Traded Sectors’, University of Oslo Memorandum No. 04/99 (1999). Leitemo, K. and Ø. Røisland, ‘The Choice of Monetary Policy Regime for Small Open Economies’, Working Paper No. 2000/05, Central Bank of Norway, Oslo (2000). Lepik, I., ‘Basic Features of the Estonian Currency Board: Convertibility and Liquidity Management’, International Workshop on Currency Boards, Hong Kong, 9 October (1999). Loertscher, R. and F. Wolter, ‘Determinants of Intra-Industry Trade: Among Countries and across Industries’, Weltwirtschaftliches Archiv, 116(2) (1980), pp. 280–93.
References
233
Lucas, R., ‘Econometric Policy Evaluation: A Critique’, in K. Brunner and A. Meltzer (eds), The Phillips Curve and Labor Markets, Carnegie-Rochester Conference Series on Public Policy (1976), pp. 19–46. MacDonald, R., ‘What Do We Really Know About Real Exchange Rates?’, Oesterreichische Nationalbank Working Paper No. 28 (1998). MacDonald, R., ‘Concepts to Calculate Equilibrium Exchange Rates: An Overview’, Economic Research Group of the Deutsche Bundesbank Discussion Paper No. 3/00, July (2000). MacDonald, R., ‘Modelling the long-run real effective exchange rate of the New Zealand Dollar’, Discussion Paper DP2002/02, October (2001). Matousek, R., ‘Market Reactions to the CNB’s Interest Rate Decision’, Prague, mimeo (2001). McKinnon, R.I., ‘Optimum Currency Areas’, American Economic Review, No. 53, September (2001), pp. 717–25. Melitz, J., ‘Geography, Trade and Currency Union’, Discussion Paper Series No. 2987, CEPR (2001). Mencinger, J., ‘How the Tolar was Created’, Banking Newsletter, 50 Years special issue, Ljubljana (2001). Mishkin, F., ‘The Channels of Monetary Transmission: Lessons for Monetary Policy’, NBER Working Paper 5464 (1996). Mishkin, F., ‘Does Correcting for Heteroscedastisicity Help?’, Economics Letters, 34 (1990), pp. 351–6. Montiel, P., ‘The Theory of the Long-Run Equilibrium Real Exchange Rate’, mimeo, World Bank (1997). Moser, G., W. Pointner and P. Backé, ‘Exchange Rate Strategies of Candidate Countries on their way to EMU and their Implications for the Euroarea’, Focus on Austria, No. 2 (2002). Muller, P. and M. Zelmer, ‘Greater Transparency in Monetary Policy: Impact on Financial Markets’, Bank of Canada Technical Report No. 86 (1999), August. Mundell, R., ‘A Theory of Optimum Currency Areas’, American Economic Review, No. 51, September (1961), pp. 657–65. Mundell, R., ‘The Appropriate Use of Monetary and Fiscal Policy for Internal and External Stability’, IMF Staff Papers No. 9 (1962), pp. 70–9. Mundell, R., ‘Uncommon Arguments for Common Currencies’, in H. Johnson and A. Swoboda (eds), The Economics of Common Currencies, Allen and Unwin, St Leonards (1973). NBP, Balance of Payments Statistics, http://www.nbp.pl/en/statistics/index.html Nenovsky, N. and K. Hristov, ‘The New Currency Boards and Discretion: Empirical Evidence from Bulgaria’, Economic Systems, 25(1) (2002), pp. 55–72. Nitsch, V., ‘Honey, I Shrunk the Currency Union Effect on Trade’, World Economy, 25(4), April (2002). Obstfeld, M., ‘International Macroeconomics: Beyond the Mundell–Fleming Model’, IMF, 1st Annual Research Conference, Mundell–Fleming Lecture (2000). Obstfeld, M. and K. Rogoff, ‘The Six Major Puzzles in International Macroeconomics: Is There a Common Cause?’, National Bureau of Economic Research Working Paper No. 7777, July (2000). OECD, Main Economic Indicators, various issues. OENB, Focus on Transition, various issues, Vienna, Austria (2001). Orlowski, L.T., ‘Direct Inflation Targeting in Central Europe’, Post-Soviet Geography and Economic, 41 (2000), pp. 134–54.
234 References Pereira, A., ‘O Impacto do Euro no Comportamento a Longo Prazo da Economia Portuguesa’, Boletim Económico, Banco de Portugal, Lisboa, Junho (1999), pp. 43–52. Peria, M.S.M., ‘A Regime-Switching Approach to the Study of Speculative Attacks: A Focus on EMS Crises’, Empirical Economics, 27(2) (2002) pp. 299–334. Persson, T., ‘Currency Unions and Trade: How Large is the Treatment Effect?’, Economic Policy, October (2001). Pikkani, R., ‘Monetary Transmission Mechanism in Estonia – Empirical Model’, Working Papers of Eesti Pank No. 5 (2001). Plasmans, J., ‘Towards a Strategic Multi-Country Model for Trade and Employment in the Baltics’, mimeo, UFSIA, Antwerp, Belgium (1999). Portes, R., ‘The Euro and the International Financial System’, mimeo, 2001. Purfield, C., ‘Fiscal Adjustment in Transition Countries: Evidence from the 1990s’, IMF Working Paper WP/03/36 (2003). Randveer, M., ‘Tulutaseme konvergents Euroopa Liidu ja liituda soovivate riikide vahel’, Eesti Panga Toimetised No. 6 (2000). Randveer, M. and M. Rell, ‘The Relationship between Competitiveness and Real Exchange Rate’, in Relationship between Real Exchange Rate and Competitiveness in the Baltic Countries, Eesti Pank, Tallinn (2002). Rawdanowicz, L.W., ‘Poland’s Accession to EMU – Choosing the Exchange Rate Parity’, CASE Studies & Analyses, No. 247, December (2002). Reinhart, C.M. and K. Rogoff, ‘The Modern History of Exchange Rate Arrangements: A Reinterpretation’, Working Paper 8963, NBER, Cambridge, MA (2002). Reinhart, C.M. and C.A. Vegh, ‘Do Exchange Rate Based Stabilisations Carry the Seeds of their Own Destruction?’, Washington, mimeo (1995). Rell, M., ‘Ex Post reaalintresside mõõtmine’ (unpublished research report, Eesti Pank, 1999). Reluga, M. and M. Szczurek, ‘Financial Markets and the Process of d/PLN Parity Setting’, mimeo, Warsaw (2002). Riboud, M., C. Sánchez-Páramo and C. Silva-Jauregui, ‘Does Eurosclerosis Matter? Institutional Reform and Labor Market Performance in Central and Eastern European Countries in the 1990s’, World Bank Social Protection Discussion Paper No. 202, Washington, DC (2002). Roberts, I. and R. Tyers, ‘China’s Exchange Rate Policy: The Case for Greater Flexibility’, mimeo, Reserve Bank of Australia, Canberra (2001). Rogoff, K., ‘Dornbusch’s Overshooting Model after Twenty-Five Years’, IMF, 2nd Annual Research Conference, Mundell–Fleming Lecture (2001). Romer, C.D. and D.H. Romer, ‘New Evidence on the Monetary Transmission Mechanism’, Brookings Papers on Economic Activity, 1 (1990). Rõõm, M., ‘Potential Output Estimates for Central and East European Countries Using Production Function Method’, Working Papers of Eesti Pank No. 2 (2001). Rose, A.K., ‘One Money, One Market: Estimating the Effect of Common Currencies on Trade’, Economic Policy: A European Forum, April (2000), pp. 7–45. Rose, A.K. and E. van Wincoop, ‘National Money as a Barrier to Trade: The Real Case for Monetary Union’, American Economic Review, 91(2) (2001), pp. 386–90. Rostowski, J., ‘The Eastern Enlargement of the EU and the Case for Unilateral Euroization’, in M. Blejer and M. Skreb (eds), Financial Vulnerability and Exchange Rate Regimes. Emerging Markets Experience, MIT Press, USA (2001).
References
235
Rother, C., ‘The Impact of Productivity Differentials on Inflation and the Real Exchange Rate: An Estimation of the Balassa–Samuelson Effect in Slovenia’, IMF Country Report, Republic of Slovenia, Selected Issues, 00/56, Washington, DC (2000). Rudebusch, G., ‘Assessing the Lucas Critique in Monetary Policy Models’, mimeo (2002). Samuelson, P., ‘Theoretical Notes on Trade Problems’, Review of Economics and Statistics, 46 (1964), pp. 145–54. Sepp, U. and M. Randveer, ‘The Aspects of the Sustainability of the Estonian Currency Board Arrangement’, Working Papers of Eesti Pank No. 5 (2002a). Sepp, U. and M. Randveer, ‘Currency Board Arrangement versus Alternative Exchange Rate Regimes in Estonia’, in The Choice of Exchange Rate Regimes and Monetary Policy Rule in Accession to EU and in Entry to EMU, Eesti Pank, Tallinn (2002b). Sepp, U. and M. Rell, ‘Ootused Eesti Inflatsiooni Võrrandites’, Eesti Panga Toimetised No. 1, 2001. Sepp, U., R. Pikkani and M. Rell, ‘Eesti Majanduse Makromudel’, Eesti Panga Toimetised No. 2 (1999). Sepp, U., M. Randveer, A. Vesilind and Ü. Kaasik, ‘Estonian Inflation Model’, Working Papers of Eesti Pank No. 1 (2000). St-Amant, P., ‘Decomposing U.S. Nominal Interest Rates into Expected Inflation and Ex Ante Real Interest Rates Using Structural VAR Methodology’, Bank of Canada, Working Paper No. 96-2 (1996). Stock, J.H. and M.W. Watson, ‘A Simple Estimator of Cointegrating Vectors in Higher Order Integrated Systems’, Econometrica, 61 (1993), pp. 783–820. Svensson, L., ‘Inflation Forecast Targeting: Implementing and Monitoring Inflation Targets’, European Economic Review, 49 (1997), pp. 1111–46. Svensson, L., ‘The Swedish Experience of an Inflation Target’, CEPR Discussion Paper No. 1103 (1995). Taylor, J., ‘Monetary Policy and the Stability of Macroeconomic Relationships’, Journal of Applied Econometrics, No. 4 (1989), pp. 161–78. Vesilind, A. and T. Liiv, ‘MB kapitalikonto struktuur. Kapitalivoogusid mõjutavad tegurid’ (unpublished research report, Eesti Pank, 2001). Vinhas de Souza, L., ‘Trade and Monetary Integration in Large, Mature Economies: A Primer on the European Monetary Union’, Kiel Institute for World Economics, Working Papers Series No. 1137, Germany (2002b). Vinhas de Souza, L., ‘Integrated Monetary and Exchange Rate Frameworks: Are There Empirical Differences?’, Working Paper 2002/2, Bank of Estonia, Tallinn (2002a). Vinhas de Souza, L. and E. Ledrut, ‘Alternative Paths towards EMU: Lessons from a Mundell–Fleming Model for the Accession Countries’, Kiel Institute for World Economics Working Papers Series No. 1132, Germany (2002). Vinhas de Souza, L., A. de Groot, H. van Eden, G. Romijn and E. Ledrut, ‘EMU and Enlargement: A Review of Policy Issues’, Economic Affairs Series, Working Paper ECON 117 EN, Directorate General for Research, European Parliament, Luxembourg (1999). von Hagen, J. and J. Zhou, ‘De Facto and Official Exchange rate Regimes in Transition Economies’, ZEI Working Paper B02-13 (2002).
236 References Wdowinski, P. and B. van Aarle, ‘EMU and its Effects: The Case of Poland. A Study for Fixed and Flexible Exchange Rates’, mimeo (1998). Welfe, A., ‘Modeling Inflation in Poland’, Economic Modelling, 17(3) (2002), pp. 375–85. Williamson, J., ‘Estimates of FEERs’, in J. Williamson (ed.), Estimating Equilibrium Exchange Rates, Institute for International Economics, Washington (1994), pp. 177–243. Williamson, J. (1985), ‘The Exchange Rate System’, Policy Analyses in International Economics, Vol. 5, Institute for International Economics, Washington (1985). Wójcik, C., ‘Benefits, Costs and Risks of Unilateral Euroization in Poland’, in M. Stierle and T. Birringer (eds), Economics of Transition: Theory, Experiences and EU Enlargement, INFER, Verlag für Wissenschaft und Forschung, Berlin (2001). Wren-Lewis, S., ‘On the Analytical Foundations of the Fundamental Equilibrium Exchange Rate’, in C.P. Hargreaves (ed.), Macroeconomic Modelling of the Long Run, Edward Elgar, UK (1992). Žumer, T., ‘Estimation of the Balassa–Samuelson Effect in Slovenia’, Bank of Slovenia, Ljubljana (2002).
Index accession negotiations, 5, 105, 208 AEG, 88, 95 Asian crisis, 162 asymmetric shocks, 7, 8, 10, 11 balance of payments, 4, 41, 140, 183, 185, 188, 190, 191 BP schedule, 41, 44, 48, 52 Bulgaria, 17, 33, 39, 44–46, 49–51, 53, 55, 60, 71, 72, 86, 166, 233 business cycles, 8, 23–38, 147, 212 capital account, 53, 59, 183, 189, 190, 202, 214 capital controls, 104, 105, 115, 119, 127, 206, 207, 220 capital inflows, 12, 51, 56, 106, 111, 121, 122, 134, 191, 220 capital mobility, 11, 51, 53, 161 consumption, 17, 18, 40, 43, 50, 57, 58, 60, 69, 70, 72, 73, 79, 107, 116, 130, 147, 150, 196, 198 Copenhagen criteria, 55 crawling peg, 105, 106, 112, 115, 116, 117, 128, 204–207 credibility, 6, 8, 12, 14, 15, 48, 52, 114, 124, 126, 139, 164–166, 175, 178, 199, 211, 216, 218, 223 credit channel, 153–156, 160, 162, 163 currency board arrangement (CBA), 131 current account, 40, 56, 59, 106, 107, 113, 115, 116, 124, 134, 140, 163, 183, 184, 189, 190, 210, 222 Czech Republic, 6, 31, 39, 44, 49, 51, 53, 72, 165–168, 174, 178, 213, 220, 221 default risk, 217, 224 demand shocks, 10, 11, 33, 37, 40 Dickey–Fuller test, 43 discount facilities, 169 disinflation, 18, 56, 57, 116, 125, 127, 165, 204–209, 218 DIT, 164–167, 170–172, 174–178, 180
ECB, 9, 43, 105, 119, 120, 127, 161, 164, 171, 174, 176–179, 183, 187, 200, 215, 228 economic and monetary union EMU, 3 employment, 15, 18, 20, 56, 66, 70, 72, 148, 163, 166, 182, 183, 194, 197, 198, 212, 213 endogenous optimum currency area, 24 equilibrium exchange rate, 222 ERM II, 3, 4, 12, 31, 52, 56, 103, 105, 119, 125–127, 180, 181, 199–201, 203, 207, 209, 219–224 Estonia, 5, 10, 11, 17, 24, 30, 32, 34, 36, 37, 39, 43–46, 48–51, 53, 60, 70, 71, 75, 86, 92, 130–132, 134, 135, 137–141, 143, 144, 146, 147, 149–151, 153–155, 157–159, 161, 162, 166, 224, 225, 232, 234, 235 EU accession, 4, 5, 12, 14, 15, 17, 18, 20, 39, 103, 120, 217, 220 EURIBOR, 138, 139, 144, 163, 193 Euroarea, 3–7, 9–11, 13–16, 19–21, 80, 94, 98, 119, 126, 127, 138, 163, 171, 178, 180, 191, 193, 202, 203, 207–212, 214–224 euroization, 4, 5, 30, 180, 219, 223 exchange rate channel, 105, 119, 156–158, 159, 160–162 exchange rate expectations, 41, 42, 140, 141 exchange rate pass-through, 149, 157, 196 exchange rate regime, 8, 104–106, 108, 115, 117, 126, 137, 157, 166, 167, 174, 180, 206, 208, 221 exchange rate strategies, 39, 203, 209 exchange rate targeting, 207, 224 exports, 7, 27, 31, 56, 58–60, 70, 72, 73, 79, 146–148, 152, 156–158, 160, 163, 185, 186, 188, 190, 211 external balance, 40, 41, 56, 204 external equilibrium, 111, 117, 118, 119, 120, 122, 182, 197
237
238 Index FDI, 57–59, 69, 70, 115, 122, 129, 147, 159, 178, 194, 205 fear of floating, 8, 83, 85 fear of pegging, 85 financial crisis, 13, 167, 172, 178 financial system, 123, 132–134, 135, 141 fiscal adjustment, 19, 204, 216, 221, 224 fiscal policy dominance, 166 fixed exchange rate, 60, 72, 131, 140, 147, 161, 166, 221 Fleming, M., 229 fundamental equilibrium exchange rate (FEER), 181, 182 Hodrick–Prescott filter, 40 Hungary, 5, 10, 11, 15, 17, 21, 22, 39, 44–46, 48, 49, 51–53, 57, 60, 70, 73, 86, 99, 105, 166, 203, 206–219, 221, 224, 228, 229 import prices, 149, 150, 153, 157, 159, 192 imports, 27, 58–60, 70, 72, 73, 79, 107, 146–150, 185, 186, 188, 190, 211, 212 indexation, 18, 59, 71, 105, 107–111, 113, 117, 120, 128 inflation bias, 14, 150 inflation expectations, 71, 122, 151, 152, 206, 207 inflation targeting, 120, 164, 166, 167, 208, 218, 220 inflationary bias, 14 interest rate arbitrage, 142, 143, 153 interest rate channel, 70, 105, 111, 115, 120, 123, 127, 151–153, 156, 159, 160, 162 internal balance, 56 internal market, 12, 106 intra-industry trade, 212 investment, 10, 14, 16, 17, 58–60, 69, 70, 72, 73, 130, 148, 156, 185, 189, 190, 210 IS curve, 145, 146 labour market, 21, 57, 118, 215 Latvia, 5, 16, 18, 24, 39, 43, 45, 46, 48–51, 53, 60, 86 Lithuania, 5, 24, 39, 43, 45, 46, 49–51, 53, 57, 60, 73, 86
Lucas critique, 24, 48 Maastricht criteria, 5, 50, 56, 125, 164, 165, 174, 198, 201, 215, 216, 221 Maastricht Treaty, 3, 219, 228 macroeconomic adjustment, 39, 56 macroeconomic modelling, 39 managed float, 60, 79 misalignment, 10, 182, 184, 185, 191–193, 195–199, 201, 223 monetary integration, 3–6, 8, 10, 12, 14, 15, 20, 23, 25, 31, 37 monetary policy framework, 103, 104, 105, 111, 119, 122, 123, 127, 133, 207 monetary regime, 131, 133, 135, 160, 168, 218 monetary targeting, 103, 111, 114 monetary transmission, 130, 131, 135, 144, 163 monetary union, 3, 5–7, 12–14, 21, 210 money market, 70, 71, 107, 120, 128, 133, 137–139, 143, 144, 161, 162, 172, 178 money supply, 59, 107, 109, 130, 132, 139, 141, 142, 153, 155 Mundell, 25 nominal anchor, 103, 105, 172, 204, 207 nominal convergence, 4, 13, 50, 57, 164, 165, 174, 199, 203, 204, 209, 215, 216, 224 optimum currency area, 25 overshooting model, 104, 106, 108, 126 pass-through, 8, 42, 49, 59, 71, 117, 124, 129, 149, 196 Phillips curve, 42, 150 Poland, 5, 10, 11, 15–18, 21, 24, 39, 45, 46, 49–51, 53, 57, 60, 73, 79, 86, 99, 166, 180, 181, 183, 185–195, 198, 200–202, 211, 213, 220, 221, 224, 232, 236 PRIBOR, 170–173, 175–177 price stability, 14, 105, 111, 119, 161, 174, 207
Index price–wage spiral, 10 production function, 148, 163, 189 purchasing power parity (PPP), 81, 181 real convergence, 5, 125, 209 real effective exchange rate REER, 147 real interest rates, 14, 104, 107, 151, 152, 162, 163, 185, 193, 194, 210, 222 realignments, 222 repo rate, 129, 165, 169, 171–173, 175, 177 reserve requirement, 132, 133 risk premium, 98, 138, 139, 161, 163, 210, 211 Romania, 5, 39, 43, 45, 46, 49–51, 53, 55, 60, 71, 73, 86, 99 Russian crisis, 51, 154, 160, 188, 194 Slovakia, 5, 15, 17, 24, 39, 45, 46, 49–51, 53, 60, 73, 158 Slovenia, 5, 11, 17, 24, 39, 45, 46, 49–51, 53, 60, 73, 86, 99, 103–108, 112, 114, 116–119, 125–129, 158, 224–227, 231, 235, 236
239
speculative attack, 140, 172, 208, 209, 218, 219 Stability and Growth Pact (SGP), 83 sterilization, 104, 110–112, 120, 122 swap rates, 173, 175 symmetric shocks, 7 transaction costs, 7, 12, 137 transmission mechanism, 104, 117, 130, 131, 135 transparency, 105, 114, 119, 123, 164–168, 172, 175, 177 uncertainty, 7, 12, 16, 20, 41, 79, 139, 140, 165, 197, 198, 201, 210, 216 VAR, 10, 44, 50, 152, 188, 197, 232, 235 wage flexibility, 11 yield curve, 122, 143, 144, 164, 165, 170, 171, 175–178, 211