MONEY AND FINANCE IN THE MIDDLE EAST: MISSED OPPORTUNITIES OR FUTURE PROSPECTS?
i
RESEARCH IN MIDDLE EAST ECONOMICS Series Editor: Jennifer Olmsted Volumes 1 and 2: Volume 3:
Volume 4:
Volume 5:
Research in Middle East Economics Edited by K. Pfeifer The Economics of Middle East Peace: A Reassessment Edited by S. Roy The Economics of Women and Work in the Middle East and North Africa Edited by M. Cinar Food, Agriculture, and Economic Policy in the Middle East and North Africa Edited by H. Lofgren
ii
RESEARCH IN MIDDLE EAST ECONOMICS VOLUME 6
MONEY AND FINANCE IN THE MIDDLE EAST: MISSED OPPORTUNITIES OR FUTURE PROSPECTS? EDITED BY
SIMON NEAIME Department of Economics, American University of Beirut, Beirut-Lebanon
NORA ANN COLTON Department of Economics, Drew University, Madison, USA
2005
Amsterdam – Boston – Heidelberg – London – New York – Oxford Paris – San Diego – San Francisco – Singapore – Sydney – Tokyo iii
ELSEVIER B.V. Radarweg 29 P.O. Box 211 1000 AE Amsterdam The Netherlands
ELSEVIER Inc. 525 B Street, Suite 1900 San Diego CA 92101-4495 USA
ELSEVIER Ltd The Boulevard, Langford Lane Kidlington Oxford OX5 1GB UK
ELSEVIER Ltd 84 Theobalds Road London WC1X 8RR UK
r 2005 Elsevier Ltd. All rights reserved. This work is protected under copyright by Elsevier Ltd, and the following terms and conditions apply to its use: Photocopying Single photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for non-profit educational classroom use. Permissions may be sought directly from Elsevier’s Rights Department in Oxford, UK: phone (+44) 1865 843830, fax (+44) 1865 853333, e-mail:
[email protected]. Requests may also be completed on-line via the Elsevier homepage (http://www.elsevier.com/locate/permissions). In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; phone: (+1) (978) 7508400, fax: (+1) (978) 7504744, and in the UK through the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, London W1P 0LP, UK; phone: (+44) 20 7631 5555; fax: (+44) 20 7631 5500. Other countries may have a local reprographic rights agency for payments. Derivative Works Tables of contents may be reproduced for internal circulation, but permission of the Publisher is required for external resale or distribution of such material. Permission of the Publisher is required for all other derivative works, including compilations and translations. Electronic Storage or Usage Permission of the Publisher is required to store or use electronically any material contained in this work, including any chapter or part of a chapter. Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher. Address permissions requests to: Elsevier’s Rights Department, at the fax and e-mail addresses noted above. Notice No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made. First edition 2005 British Library Cataloguing in Publication Data A catalogue record is available from the British Library. ISBN: 0-7623-1216-5 ISSN: 1094-5334 (Series)
8 The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed in The Netherlands.
Working together to grow libraries in developing countries www.elsevier.com | www.bookaid.org | www.sabre.org
iv
CONTENTS LIST OF CONTRIBUTORS
vii
LIST OF REFEREES
ix
INTRODUCTION Simon Neaime and Nora Ann Colton
xi
PART I: SAVINGS, INVESTMENTS AND PORTFOLIO AND FDI FLOWS PORTFOLIO DIVERSIFICATION AND FINANCIAL INTEGRATION OF MENA STOCK MARKETS Simon Neaime
3
FOREIGN DIRECT INVESTMENT IN TUNISIA: ROLE OF THE FREE TRADE AGREEMENT WITH EUROPEAN UNION Mohamed Abdelbasset Chemingui and Nora Ann Colton
21
INSTITUTIONAL INVESTORS, CONTRACTUAL SAVINGS AND CAPITAL MARKET DEVELOPMENT IN EGYPT, JORDAN, MOROCCO, AND TUNISIA Wafik Grais and Dimitri Vittas
43
FINANCIAL LIBERALIZATION, SAVINGS, INVESTMENT, AND GROWTH IN MENA COUNTRIES Lahcen Achy
67
v
vi
CONTENTS
DETERMINANTS OF PRIVATE SAVINGS IN THE MIDDLE EAST AND NORTH AFRICA Kıvılcım Metin O¨zcan and Yusuf Ziya O¨zcan
95
PART II: COMMERCIAL BANKING PROFITABILITY AND RISK MANAGEMENT IN BANKING: A COMPARATIVE ANALYSIS OF EGYPT AND LEBANON Sam Hakim and Simon Neaime
117
AN ANALYSIS OF MERGERS AND ACQUISITIONS IN THE TURKISH BANKING SECTOR Ays-e Mumcu and E. U¨nal Zenginobuz
133
PART III: MONETARY, FINANCIAL AND EXCHANGE RATE POLICIES HYSTERESIS IN CURRENCY SUBSTITUTION: THE MIDDLE EAST AND NORTH AFRICA Wassim N. Shahin and Fadi G. Freiha
165
CHOOSING AN APPROPRIATE PALESTINIAN MONETARY REGIME Moyara Ruehsen
183
DOLLARIZATION AND ITS LONG-RUN DETERMINANTS IN TURKEY Irfan Civcir
201
FISCAL MONETARY AND EXCHANGE RATE POLICIES IN A SMALL OPEN ECONOMY Simon Neaime
233
LIST OF CONTRIBUTORS Lahcen Achy
INSEA, Rabat Institutes, Rabat, Morocco
Mohamed Abdelbasset Quantitative Methods and Modeling Chemingui Department (QM&MD), Kuwait Institute for Scientific Research, Safat, Kuwait Irfan Civcir
Faculty of Political Sciences, Ankara University, Ankara, Turkey
Nora Ann Colton
Department of Economics, Drew University, Madision, NJ, USA
Fadi G. Freiha
Indevco SARL, Ajaltoun, Lebanon
Sam Hakim
Graziadio Graduate School of Business, Pepperdine University, Malibu, CA, USA
Simon Neaime
Department of Economics and The Institute of Financial Economics, American University of Beirut, Beirut, Lebanon
Kıvılcım Metin O¨zcan
Department of Economics, Bilkent University, Ankara, Turkey
Yusuf Ziya O¨zcan
Department of Sociology, Middle East Technical University, Ankara, Turkey
Moyara Ruehsen
Monterey Institute of International Studies, Monterey, CA, USA
Wassim N. Shahin
Lebanese American University, Byblos, Lebanon
Wafik Grais and Dimitri Vittas
Financial Sector Network, World Bank, Washington, DC, USA
Ays- e Mumcu and E. U¨nal Zenginobuz
Department of Economics, Bog˘azic- i University, I˙stanbul, Turkey
vii
This page intentionally left blank
viii
LIST OF REFEREES Sebouh Aintablian
Department of Economics, American University of Beirut, Lebanon
Varouj Aivazian
University of Toronto, Toronto, Canada
Paul Beaumont
Department of Economics, Florida State University
Jeffrey B. Nugent
Department of Economics, University of Southern California, CA, USA
Mine Cinar
Department of Economics, Loyola University, Chicago, USA
Cyrus Bina
Department of Economics, University of Minnesota, USA
Charles Chittle
Bowling Green State University, USA
Donald Cole
Department of Economics, Drew University, New York, USA
Nora Colton
Department of Economics, Drew University, New York, USA
Fred Curtis
Department of Economics, Drew University, New York, USA
Edwin Dickens
Department of Economics, Drew University, New York, USA
Sam Hakim
Graziadio School of Business, Pepperdine University, California, USA
Walid Hejazi
University of Toronto, Toronto, Canada
Dorene Isenberg
Department of Economics, Drew University, USA
Edwin Isenberg
Department of Economics, Drew University, USA
Lynda Khalaf
Universite Laval, Quebec, Canada ix
x
LIST OF REFEREES
Atif Kubursi
McMaster University, Hamilton, Canada
Arman Mansoorian
York University, Toronto, Canada
Leo Michellis
Ryerson Polytechnic University, Toronto, Canada
Simon Neaime
Department of Economics, American University of Beirut, Lebanon
Christopher Nicol
University of Regina, Saskatchewan, Canada
Siddieq Noorzoy
Bowling Green State University, USA
Jennifer Olmsted
Department of Economics, Drew University, New York, USA
Ugo Panizza
Department of Economics, American University of Beirut, Lebanon
Chris Paraskevopoulos Department of Economics, Atkinson College, Toronto, Canada John Paschakis
Atkinson College, North York, Canada
George Zestos
Christopher Newport University, Virginia, USA
INTRODUCTION The Middle East and North Africa (MENA) has witnessed much change over the past few decades. It has experienced the difficulties of post-colonial independence marred by a number of wars. This political instability coupled with too much government control and regulation caused much of the international financial and business community to shun this region for emerging superstars of Southeast Asia. Fund managers estimated that out of a total of US$65 billion of capital floated into emerging markets in the peak year of 1993, only 0.3% trickled to Arab markets. Yet, this region is rich in human capital and natural resources. It is the home of 6% of the world’s population with a wealth of highly skilled workers and a gross domestic product (GDP) of over US$600 billion. The region also has some of the world’s largest oil reserves. This volume sets out to offer a reappraisal of the performance of international financial arrangements in the MENA region over the past two decades. Reappraisal offers much insight into how countries in the MENA region have been attempting to restructure their financial markets over the past two decades in hopes of attracting and utilizing international capital more effectively. It also proffers an analysis of the future prospects of international capital flows in this region as well as examining a number of case studies for insights into how financial markets operate in the Middle East. The Asian currency crisis of 1997 has caused many international organizations as well as governments to re-examine the notion prior to this crisis that financial liberalization is beneficial for all countries and economies. Prior to this crisis, MENA states had considerable pressure put on them particularly by international organizations, such as the World Bank and the International Monetary Fund to follow the Asian example. Consequently, it is important to assess the impact that financial liberalization has had on the region in light of the number of financial crises that have taken place worldwide over the past few years. Although financial liberalization has had many positive aspects, it has not been without risks. It xi
xii
INTRODUCTION
is important for policy makers, academics, as well as international financial institutions to understand what the costs and benefits have been to this important region. The financial crisis that erupted in East Asia in 1997 has led to sharp declines in the currencies, stock markets, and other asset prices of many emerging Asian economies threatening these countries’ financial systems. In addition to its severe effects in Asia, the crisis has put pressure on emerging markets outside the region contributing to fast declines in their stock markets and GDP growth rates. One of the reasons that the crisis escalated so quickly stems from weaknesses in the Asian financial systems and, to a lesser extent, governance. A combination of inadequate financial sector supervision, poor assessment and management of financial risk, and the maintenance of fixed exchange rates led banks and corporations to borrow large amounts of short-term unhedged international capital. Although private sector expenditure and financing decisions led to the crisis, it was accentuated by governance issues, notably government involvement in the private sector and, lack of transparency in corporate and fiscal accounting. While many emerging markets suffered backlashes from Southeast Asia, the MENA countries were among the few where the repercussions were minimal. Indeed, MENA financial markets serve as good examples of newly emerging capital markets with significant growth potentials. The question that we wish to address in the wake of the recent East Asian crisis is whether the MENA region has the potential to assume a leading role in attracting foreign capital as East Asia previously did. If so, under what conditions? What regulatory and other measures are necessary to forestall a possible repeat of the East Asian crisis? The Asian crisis unfolded against the backdrop of several decades of outstanding growth performance. While economists recognize the influence of financial factors on corporate investment decisions and economic growth, this relationship is not at all straightforward or well understood. For example, countries like Jordan, Kuwait, and Bahrain have managed to achieve high growth rates while maintaining, until relatively recently, substantial government controls on their financial systems. Other countries like Egypt and Lebanon, for example, achieved higher growth rates with significantly liberalized financial systems. In the light of the East Asian crisis, one can argue that with a welldeveloped financial and institutional infrastructures (i.e. adequate prudential regulation in banking, bankruptcy laws, accounting systems, and laws and regulations ensuring transparency of the financial and non-financial sectors) direct controls of the financial markets are unwise. Otherwise, in the
Introduction
xiii
absence of such measures government controlled financial systems might be considered a second best alternative to avoid disruptions to economic growth. This matter needs to be thoroughly assessed. With the above in mind, we argue that while financial development has been conducive to economic growth in the MENA region, the development of systems to oversee and regulate financial systems has lagged behind policies promoting domestic and external liberalization. This lag, which adds to an economy’s ‘‘systemic risk’’ can make some MENA economies very vulnerable to external sources of risk, such as large inflows of capital and undermine their ability to sustain growth, as has been demonstrated by the recent East Asian crisis. The increased degree of openness is mainly due to the factor endowments of the region, and the increasing trend towards trade and financial globalization has rendered the tasks of policy makers in the MENA region rather difficult. While developed countries with relatively large financial markets can better absorb and neutralize the effects of external financial shocks, this may be more difficult in the case of developing economies, which are usually smaller in size. Policy makers and academics in the MENA region have to primarily identify the sensitivity and vulnerability of their economies to various external financial shocks and then try to design effective strategies that are capable of coping with the impact of these shocks, in order to minimize their adverse implications. MENA countries should improve their macroeconomic and financial policy coordination to better cope with the fast changing global financial environment. This can be achieved, for example, through enhancing intraregional financial and economic integration. The establishment of a MENA free trade zone will not only stimulate and enhance growth, but will also enhance intra-MENA trade and financial linkages, reducing considerably the reliance on foreign trade and international short-term capital. And to better insulate their economies from the impact of negative financial shocks, MENA countries should embark as soon as possible into a wide integration process of their goods and capital markets. One immediate benefit would be to reduce the magnitude and impact of external financial shocks, similar to those emanating from East Asia, on the MENA region. These economies will be able to better absorb the negative impacts of external shocks as one large economic and financial entity rather than having to withstand the impact of these shocks on an individual country basis. A direct consequence of an integrated MENA capital market will be to dampen the impact of world interest rates fluctuations on MENA countries with high levels of foreign debt. A larger financial market in the MENA
xiv
INTRODUCTION
would lower the cost of raising capital and would allow MENA firms to rely more on the local market rather than tapping world markets to raise capital increasing by this their vulnerability to fluctuations in foreign interest rates. A lower level of the cost of raising capital will subsequently translate into more investment and GDP growth rates in the MENA region. This volume contains three main themes. The first theme relates to financial developments in the MENA region, emphasizing the role of stock markets and portfolio flows; foreign direct investments (FDI); and private and public savings in the growth and development experience of the region. Again, we see the notion that financial liberalization has many benefits as well as risks not only for the countries involved, but also for international investors as well, is echoed throughout these chapters. For the countries of this region, we see the promise of enhanced growth and development through more developed financial markets that can facilitate the important relationship between investors and savers. For the international investor, we see a region that is posed to offer returns that are strongly correlated to risk. In the developing economies of the MENA region commercial banks have played a prominent role in economic growth since capital markets are still underdeveloped and still lack the appropriate mechanisms to channel effectively and efficiently funds from surplus units to deficit units. The next theme of the volume relates to the role commercial banks have played in channeling funds from savings to investments, and their role in the financing for development experience of some MENA economies. Finally, and on a broader level, the conduct of monetary policy in some MENA countries is highlighted with some emphasis on exchange rate policies and the use of some exchange rate regimes and their impact on the economies of the region. Neaime’s chapter on portfolio diversification and financial integration of the MENA stock markets is an important example of the risks and returns that are available for international investors in the MENA region. This chapter examines the MENA stock markets and explores whether these markets can offer international investors unique opportunities in a diversified international portfolio. Neaime finds, using Johansen Cointegration Tests, that the Gulf Cooperation Council (GCC) equity markets are the best markets in the MENA region for fulfilling this requirement. However, other emerging MENA stock markets like those of Turkey, Egypt, Morocco, and Jordan may be important future markets to consider as they integrate into world financial markets. Simon also notes that the MENA region may not be a place for investors attempting to counter trends in the U.S. or U.K. stock markets as these markets are very influential on this
Introduction
xv
region. The results of Neaime’s work are very insightful into how investors might think about the MENA region and these maturing stock markets. The next chapter examines the timely question of whether Tunisia could expect an increase in FDI flows in response to the establishment of a free trade area (FTA) with the European Union. Chemingui and Colton assert that the partnership agreement between Tunisia and the European Union has the potential of playing a catalytic role in increasing the openness of the Tunisian economy and attracting FDI. The empirical analysis shows that the investment potential created by the partnership agreement, as well as the growth of FDI inflows to Tunisia, are conditioned by the increase in the capital productivity. In this case, Tunisia could expect an important welfare gain. In Chapter 3 Grais and Vittas examine the linkage between equity markets, contractual savings, and institutional investors in Egypt, Jordan, Morocco, and Tunisia. This chapter builds on the previous chapters by identifying pre-conditions for maximizing the contribution of contractual savings to capital market development. Although the study does not find it necessary to have contractual savings and institutional investors for the development of equity and bond markets, with the pre-conditions in place they can have a significant impact on market development. Achy tackles the question of whether financial liberalization fuels economic growth by stimulating savings and investment. The purpose of this chapter is three-folds: (1) to review the literature on the rationale for financial repression; (2) to examine the theoretical and empirical literature on the links between financial liberalization, savings, and investment; and (3) to assess empirically the effect of financial reforms on economic performance in the specific case of MENA countries. The coefficients on financial development indicators as well as on financial liberalization index indicate a negative impact of financial depth on private investment in the five MENA countries investigated. Although these findings challenge other such studies, the author tries to explain the results as a bias toward the development of mortgage and consumer-credit markets. This bias has thus distorted private credit allocation away from firms and toward households. By distorting private credit allocation in favor of households at the expense of lending to firms, financial liberalization may not only have reduced private savings but tended also to reduce available loans for the business sector. Regarding the relationship between financial depth and economic growth, financial depth indicators fail to explain the growth experience in the five MENA countries under investigation – Egypt, Jordan, Morocco, Tunisia, and Turkey.
xvi
INTRODUCTION
In the final chapter under this theme, private savings again emerge as an important issue for understanding financial markets in the chapter by Kivilcim and Ozcan. This chapter investigates the relationship between private savings and a broad range of macroeconomic aggregates in the MENA region over the period 1981–1994. Unlike the findings of Achy in the previous chapter, financial depth measure suggests that countries with deeper financial systems will tend to have higher private savings. Private credit and real interest rates capture the severity of the borrowing constraints and the degree of financial repression. Inflation also captures the macroeconomic volatility and has a positive impact on savings. Under banking in the MENA region, Hakim and Neaime’s chapter echoes the theme that liberal banking markets operate most efficiently with regulation and efficient capitalization. The authors focus on two important leaders in the MENA region in terms of banking reform – Lebanon and Egypt. Examining the impact of liquidity, credit, and capital on bank profitability, they demonstrate the importance of risk management practices and enforcement of banking regulations. Sedar and Zenginobuz explore the impact of mergers and acquisitions in the banking industry in Turkey. This chapter is very important to the volume as the restructuring and consolidation of the banking systems within this region has been a characteristic of the past decade. The authors perform comparative static exercises regarding the impact of mergers among domestic as well as with foreign banks on equilibrium outcomes. The authors proffer a number of lessons in this study regarding mergers in the banking sector. As for whether mergers will occur or not, the nature of the merged entity is of critical importance. An example is two fringe banks merging to become a dominant bank; this indicates not only an improvement in the cost structure but also enhanced market power. Thus even in cases where there is no incentive for fringe firms to merge to remain a fringe firm, they will merge if through merging they can exercise market power. In addition to examining the different structures of banking mergers, the study also examines the welfare impact of this phenomenon. Under the broader theme of monetary and exchange rate developments in the MENA region comes the chapter by Shahin and Freiha. Egypt and Lebanon emerge as important examples for examining the impact of recent economic and financial reforms. In this chapter the issue of currency substitution is explored. The chapter sheds some additional observations on the irreversibility of the dollarization phenomenon that has existed in many developing countries. Historically, these two countries have experienced dollarization; however, recent economic and political stability would lead
Introduction
xvii
one to conclude that the factors for this dollarization are no longer valid. Yet, both Lebanon and Egypt are found to continue to experience dollarization. For Lebanon, the authors argue that despite the decline in inflationary expectations, the expectations of currency depreciation, and an increase in real interest rate differentials between domestic and foreign currencies, dollarization did not decline by the anticipated amount. For Egypt, the authors find that unlike many Latin American countries, currency substitution was successfully reversed for a period when the government managed to peg the value of the Egyptian pound to the dollar, yet returned when this system was altered. Ruehsen takes on the task of examining the appropriate monetary regime for the Palestinian Authority and an eventual Palestinian state. The question of whether or not there should be a Palestinian independent currency has been discussed on and off since the mid-1990s. This chapter examines several viable monetary regimes including the introduction of a Palestinian currency operating under a managed float, a Palestinian currency operating under a currency board, monetary union with Jordan, the status quo which permits the Jordanian dinar, Israeli shekel and US$ as legal tender, and finally, dollarization coupled with the introduction of Palestinian coins. Each of these options is compared on the basis of whether it enhances macroeconomic stability, provides the benefits of seigniorage, deters inflation, stimulates investment, and encourages fiscal and monetary discipline. The chapter by Civcir also tackles the issue of dollarization with Turkey as case study. This chapter focuses on Turkey, a country that has experienced much economic upheaval over the past decade. The author explains the dollarization process in Turkey using an extended portfolio model where dollarization is determined by the relative rates of return of domestic and foreign currencies denominated assets, expected change in the exchange rate, exchange rate risk, and credibility of current economic policies. The chapter concludes that interest rate differential and the expected exchange rates are the dominant variables in determining dollarization. The last chapter tackles the questions of fiscal and exchange rate sustainability with Lebanon as case study. Neaime’s chapter on Lebanon examines the long-run relationship between the nominal exchange rate and the inflation rate for evaluating monetary policy in a small open economy framework. The results and estimation techniques are of significant value for understanding the small open economy scenario and the workings of the Lebanese monetary and fiscal policies. The conclusion of this study is that
xviii
INTRODUCTION
the Lebanese Monetary Authority, while successful in containing inflationary pressures by adopting a monetary policy rule with the nominal exchange rate, as target should have adopted a real exchange rate-targeting policy to dampen the effects of its current policy on interest rates, foreign debt, and the growth in GDP. This volume not only demonstrates the complexity of financial arrangements that exist in the MENA region, but the dynamic nature of this sector. Although, historically the banking and financial systems of the MENA region have been immature, they are adapting and changing to a world where liberalization is taking place at a fast pace. Hopefully this volume can serve as a starting point for understanding not only what has happened in the past, but the road forward for financial liberalization in the MENA region. Simon Neaime Nora Ann Colton Editors
PART I: SAVINGS, INVESTMENTS AND PORTFOLIO AND FDI FLOWS
1
This page intentionally left blank
2
PORTFOLIO DIVERSIFICATION AND FINANCIAL INTEGRATION OF MENA STOCK MARKETS Simon Neaime ABSTRACT This chapter studies the properties and characteristics of the Middle East and North African (MENA) stock markets, and the prospects and implications of enhanced financial liberalization in the region. It also explores whether these markets can offer international investors unique risk and returns characteristics to diversify international and regional portfolios. Johansen co-integration tests reveal that the Gulf Cooperation Council equity markets still offer international investors the portfolio diversification potentials mainly through mutual funds, while other emerging MENA stock markets like those of Turkey, Egypt, Morocco, and to a lesser extent Jordan have matured, and are now integrated with the world financial markets.
1. INTRODUCTION Over the last decade, the empirical finance literature has been concerned with the financial integration of the world major stock markets (see e.g. Eun Money and Finance in the Middle East: Missed Opportunities or Future Prospects? Research in Middle East Economics, Volume 6, 3–20 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1094-5334/doi:10.1016/S1094-5334(05)06001-2
3
4
SIMON NEAIME
& Shim, 1989; Kim & Wadhwani, 1990; Joen & Von Furstenberg, 1990; Kasa, 1992; Arshanapalli & Doukas, 1993). Recently, there has been a shift in attention to the emerging markets of developing countries (Chowdhry, 1994; Bekaert & Harvey, 1997; De Santis & Imrohoroglu, 1997; Darrat & Hakim, 1997; Hakim & Neaime, 2000). The new focus stems from the fact that these markets present portfolio and fund managers, a new possibility to enhance and optimize their portfolios. For example, Bekaert (1993), and Bekaert and Harvey (1997) have found that stock market returns in emerging markets were high and predictable but lacked strong correlation with major markets. As emerging markets mature, they are likely to become increasingly sensitive to the volatility of stock markets elsewhere. Their increasing degree of integration with world markets will diminish their ability to enhance and diversify international portfolios. Middle East and North African (MENA) member countries do not treat international, or even intra-regional capital investments uniformly. Financial integration remains a distant goal, and has been overtaken by smaller more functional arrangements, such as the Gulf Cooperation Council (GCC).1 This regional integration is important because the MENA stock markets as a group may be able to offer investment opportunities not possible by one individual MENA market. Specifically, GCC countries have traditionally discriminated against non-GCC investors, but the rest of the regional markets, particularly those of Amman, Cairo, Casablanca, and Turkey are largely open to MENA investors. These markets offer capitalrich GCC equity investors the unique diversification benefits associated with optimum portfolios with a balanced mix of domestic and international securities. The concept of stock market integration is broad. This chapter investigates how MENA equity markets display different co-movements over time. Greater degrees of co-movements generally reflect greater stock market integration (see Eun & Shim, 1989; Engel & Susmel, 1993; Cheung & Mak, 1992). This study reveals for the first time issues relating to regional and international integration and segmentation of seven emerging MENA equity markets. For example, in the context of the MENA region our research will address the question of whether the MENA capital market is segmented from the rest of the world. Questions of market integration are of concern to both MENA equity investors and companies in the region that make capital budgeting decisions. Specifically, if segmentation exists and a firm is forced to raise capital locally, then its cost of capital is likely to be higher than that of a company with unrestricted access to the regional and international capital markets. Therefore, one would expect the restriction to the local
Portfolio Diversification and Financial Integration of MENA Stock Markets
5
capital market to raise a firm’s marginal cost of capital. This chapter will measure the extent to which capital market segmentation imposes restrictions on the free mobility of capital in the MENA countries and the incremental costs a firm is forced to bear (Korajczyk, 1996). Moreover, within the MENA countries, our results identify (1) markets, which are regionally and/or internationally integrated, and (2) diversification potentials these equity markets offer investors. Overall, these findings are expected to contribute to the understanding of the investment potentials offered by the equity markets in the MENA countries. In many ways, the MENA stock markets remain somewhat unsophisticated. Better regulation and more transparency would help to strengthen their market fundamentals. However, as these countries liberalize their financial markets, their returns are likely to correlate more with the world markets. This would raise the question of whether their volatilities will suffer. This chapter constitutes the first attempt at studying a broad number of stock markets in the MENA region (seven markets). Previous studies covered shorter periods of time and a smaller number of markets in this region. These include: (1) El-Erian and Kumar (1995), which covered five markets over the period 1992–1994; (2) Darrat and Hakim (1997) covering three markets and the period 1996–1997; (3) Hakim and Neaime (2000) during the period 1995–1998; (4) Butler and Malaikah (1992) covering two markets and the period 1985–1990. The chapter will also test for the existence of co-integration to determine whether the MENA countries are either (a) segmented or (b) linked to the U.S. stock market (S&P 500 Index), the U.K. (FTSE100), and the Paris (CAC30) stock markets, respectively. Should significant co-integration exist between MENA and other stock markets, one would conclude that inefficiencies in security prices in one market may affect movements in another country’s market. The rest of the chapter is divided as follows. Section 2 overviews the main characteristics and properties of the MENA stock markets. Section 3 presents an empirical model to test whether MENA stock markets are integrated with each other. In the same section, we explore the interrelations these markets exhibit with respect to their counterparts in the U.S., the U.K., and France using Johansen co-integration techniques. Section 4 employs a vector error-correction model (VECM), Granger-Causality Tests,
6
SIMON NEAIME
and impulse response functions to further explore the long- and short-run dynamics of the stock market return series. Section 5 summarizes and concludes the chapter with some policy implications.
2. MENA STOCK MARKETS Stock market developments in the MENA region have been disappointing during the last decade. The GCC market capitalization has remained relatively low and did not grow by as much as in its MENA counterparts. In Saudi Arabia market capitalization has grown only by 30% in between 1989 and 1999. Similar growth performances are observed in Kuwait (see Table 1). This can be attributed to the 1991 Gulf War and to the fact that these markets have remained closed and fairly illiquid. The share of ownership of the respective governments in these markets has rendered the role of private Table 1.
Characteristics of MENA Stock Markets.
Number of Companies Listed
Bahrain Egypt Jordan Morocco Saudi Arabia Turkey Kuwait Total MENA Iran Greece Israel
Market Capitalization (US Billion)
Volume Traded (US Million)
Turnover Ratio (%)
1989
1989
1989
1999
1989
1999
1999
— 510 106 41 59 50 — 796
41 989 152 55 73 285 66b 1595
— 1.71 2.16 0.62 40.90a 6.78 11.31 18.09
7.15 32.83 5.82 13.69 60.44 112.71 20.00 232.64
— 558 — 91 7660 18.80 652 538 3.32 16 3294 38.81 6194a 12,670 6.60 798 68,802 8.50 14.2 18.00b — 1571.2 93,522 76.03
169a 119 262
242b 281 644
6.55a 6.37 8.22
14.87b 204.21 63.82
741a 1389b 549 155,256 3909 16,446
8.84 11.61 2.10
1999 12.82 4.29 10.83 4.16 4.77 1.64 127b 38.51 10.70 1.32 3.88
Source: World Bank Group, IFC, Arab Monetary Fund 2000. (1) Year-end totals, excluding listed investment funds wherever possible. (2) Year-end total market values of listed domestic companies. (3) Year-end total value traded of listed domestic company shares. Calculated by dividing the value of total shares traded by market capitalization for the year. a 1995. b 1998.
Portfolio Diversification and Financial Integration of MENA Stock Markets
7
firms rather limited in scope. This scenario is quite different when one looks at the remaining MENA stock markets. Record market capitalization growth rates can be noted in Egypt, Morocco, Turkey, and to a lesser extent in Jordan over the same period. This is due to (1) the massive privatization plans introduced in Morocco, Egypt, and Turkey; (2) the extensive sale of government assets to private firms; and (3) the considerable efforts devoted recently toward enhancing the efficiency, depth, and liquidity of the three stock markets. MENA performance relative to other Middle Eastern stock markets has also been disappointing and has cast doubt on the performance of MENA stock markets during the last decade. While liquidity ratios have remarkably improved in between 1989 and 1999 in Morocco, Turkey, and Egypt, they have remained relatively low in Jordan, Saudi Arabia, and Kuwait. The turnover ratio has declined in Egypt, Jordan, Morocco, and Turkey. This can be attributed to the considerable increase in the market capitalization in each one of these markets. Table 2 indicates that while GCC stock markets are fully accessible to GCC investors they have remained relatively closed to international foreign investors; even non-GCC MENA countries face restrictions on portfolio investment in these stock markets. Various restrictions still exist in the face
Table 2. Bahrain
Egypt Jordan Kuwait Morocco Saudi Arabia
Turkey
Accessibility of MENA Stock Markets to Foreign and Other Arab Investors. Open to GCC nationals Foreigners residing in Bahrain for at least 3 years may own up to 1% of the capital of a single company Foreigners can trade shares in only four of the 41 listed companies Unrestricted access to foreign investors Repatriation of capital and dividends allowed Foreign investors can hold up to 50% of a company’s capital Repatriation of capital and dividends allowed Open to GCC nationals Non-Kuwait residents are allowed to own shares through mutual funds only Unrestricted access to foreign investors Repatriation of capital and dividends are allowed Open only to GCC nationals who can own up to 25% of listed companies other than banks Shares traded over the counter through banks Opened recently to foreign investors through mutual funds only Open to foreign investors with 49% ownership constraint in any listed or unlisted companies
Source: Author’s estimates.
8
SIMON NEAIME
of MENA portfolio flows and the removal of these restrictions is expected to (1) improve and enhance growth and liquidity in these markets; and (2) reduce the costs of raising capital in the local market. Table 2 also indicates that international investors have complete access to the Morocco, Cairo, and Istanbul stock markets, and a little less access to the Jordanian stock market. Although the open access to foreign investors has contributed significantly to the growth performances of MENA stock markets, this is expected to gradually lower their diversification potentials that used to be offered to international investors. Increased financial integration within the MENA region is expected to bring considerable benefits to MENA investors. A more liquid capital market offers lower borrowing costs for MENA firms wishing to raise funds locally. Moreover, international financial institutions will be willing to diversify their portfolios by tapping the MENA financial markets, which will benefit from portfolio capital inflows, if the convergence of asset returns in the world markets leads international investors to increase their MENA markets holdings in order to diversify across countries with a wide range of risk and returns.
3. DATA AND EMPIRICAL MODEL Our data consists of weekly closing price series up to December 2000 for the MENA stock markets of Bahrain, Kuwait, Saudi Arabia, Jordan, Egypt, Morocco, and Turkey. For the world main financial markets we use the U.S., the U.K., and the French stock markets. This chapter divides the MENA stock markets into two groups. The first consists of the GCC member countries: Bahrain, Kuwait, and Saudi Arabia. The second includes Egypt, Jordan, Morocco, and Turkey. Compounded week-to-week returns are calculated as the natural logarithmic differences in prices: lnðPt =Pt 1 Þ: Older stock markets in the MENA region, as for example Amman and Istanbul, are tracked as early as 1990, while Morocco is tracked since 1992. More recent additions like Saudi Arabia and Egypt are analyzed since 1993, Bahrain since 1995, and Kuwait since 1994 due to the Gulf War.
3.1. Unit Root Tests for Stationarity We start by testing the existence of a long-run relationship between (a) the MENA stock markets as a group (Egypt, Jordan, Turkey, and Morocco);
Portfolio Diversification and Financial Integration of MENA Stock Markets
9
(b) the GCC stock markets as a group (Saudi Arabia, Bahrain, and Kuwait); and (c) between both MENA and GCC markets on one hand and the U.K., the U.S., and the French stock markets on the other. For this purpose the Johansen (1991; 1995) co-integration test will be used after establishing nonstationarity of the series by applying both the Phillips–Perron (PP) and the Augmented Dickey–Fuller (ADF) Unit Root Tests.2 It is common for timeseries data to demonstrate signs of non-stationarity; typically both the mean and variance of macro-economic variables trend upward over time. In any case, tests for non-stationarity are carried out as a preliminary step to explore the possibility of a significant long-run relationship between the variables concerned (i.e. co-integration tests). The following regressions are carried out: DPt ¼ b1 þ b2 Pt
1
þ
k X
di DPt
i
þ t
(1)
i¼1
where D is the first-difference operator; bi and di are constant parameters; and t is a stationary stochastic process. The number of lags (k) will be determined based on the Akaike information criterion (AIC). To determine the order of integration of the series, model (1) is modified to include second differences on lagged first and k lags of second differences. That is, D2 Pt ¼ l1 DPt
1
þ
k X
mi D2 Pt
i
þ 1t
(2)
i¼1
where D2 Pt ¼ DPt DPt 1 ; li and mi are constant parameters; and 1t is a stationary stochastic process. The k-lagged-difference terms are included so that the error terms t and 1t in both the equations are serially independent. To test for stationarity, the ADF and PP tests are applied to Eqs. (1) and (2) and the results are summarized in Table 3. The null hypotheses are b2 ¼ 0; and l1 ¼ 0; respectively; that is, a unit root exists in Pt and DPt 1 implying that the series are non-stationary. The PP and ADF tests results indicate that the MENA and the world stock markets series are non-stationary in the levels (Eq. (1)). However, unit roots in the first differences of the stock prices (Eq. (2)) are rejected at the 1% significance level, suggesting that returns (or stock price changes) are stationary. We conclude that weekly stock prices in the world and the MENA regions are I(1). That is, the first-differenced series do not exhibit a unit root (i.e. the series are stationary). The ADF and PP tests indicate that each variable achieves stationarity only if converted to first-differences.
10
SIMON NEAIME
Table 3. PP (L) Levels Bahrain Egypt Jordan Kuwait Morocco Saudi Arabia Turkey MENA index Paris U.K. U.S.
0.50 1.63 1.23 0.09 0.17 2.06 2.23 0.89 0.09 1.06 1.76
Unit Root Test Results. PP (L) First Difference
(2) (4) (4) (2) (2) (5) (5) (4) (3) (2) (2)
12.32 10.41 16.37 7.00 7.05 10.01 15.71 11.82 15.63 17.38 18.27
ADF (L) Levels
(2)** (4)** (4)** (2)** (2)** (5)** (5)** (4)** (3)** (2)** (2)**
0.39 1.89 1.27 0.09 0.25 1.54 2.53 1.18 0.17 0.85 1.35
(2) (4) (4) (2) (2) (5) (5) (4) (3) (2) (2)
ADF (L) First Difference 5.40 6.32 8.76 4.77 4.61 3.40 7.58 5.90 8.69 9.21 11.20
(2)** (4)** (4)** (2)** (2)** (5)** (5)** (4)** (3)** (3)** (2)**
Source: Author’s estimates. (1) The MENA index is a market capitalization-based weighted index for Egypt, Jordan, Turkey, and Morocco. (2) **Indicates rejection of the null hypothesis of non-stationarity at the 1% significance level. (3) (L) represents the number of lag lengths.
Thus, each variable is integrated of the first-order I(1). Since the DPt series are stationary, they are an I(0) stochastic processes, which means the Pt series are I(1) time series; essentially they are random walks (i.e. non-stationary stochastic processes).3
3.2. Co-integration Tests Our next task is to check whether the series are co-integrated. Specifically, having established the presence of a unit root in the levels of each variable, we need to test whether the series in each country have different unit roots (non-cointegrated), or share the same unit root (co-integrated). Co-integrated variables, if disturbed, will not drift apart from each other and thus possess a long-run equilibrium relationship. Testing for the existence of cointegration among economic variables has been widely used in the empirical literature to study economic inter-relationships. Its existence would imply that the two series would never drift too far apart. A non-stationary variable, by definition, tends to wander extensively over time, but a pair of non-stationary variables may have the property that a particular linear combination would keep them together; that is, they do not drift too far
Portfolio Diversification and Financial Integration of MENA Stock Markets
11
apart. Under this scenario, the two variables are said to be co-integrated, or possess a long-run (equilibrium) relationship.4 The Johansen (1991; 1995) Efficient maximum likelihood test is used to examine the existence of a long-term relationship between the individual MENA subgroups (MENA and GCC) on the one hand and between the MENA and GCC, and the world markets on the other, at the 5% and 1% levels of significance, respectively.5 It is applied using alternative lag lengths in the vector autoregression (VAR). More specifically, consider a VAR of order z: X t ¼ A1 X t
1
þ þ Az X t
z
þ t
(3)
where Xt is our y-vector of the non-stationary I(1) weekly stock market price series and t is a vector of innovations. We can rewrite the VAR as DX t ¼ yX t
1
þ
z 1 X
li DX t
i
þ t
(4)
i¼1
Pz P where y ¼ zi¼1 Ai I i ; and li ¼ j¼iþ1 Aj : Granger’s Representation Theorem asserts that if the coefficient matrix y has reduced rank roy, then there exist y r matrices o and O each with rank r such that y ¼ oO0 and O0 X t is stationary. The r is the number of cointegrating relations (the co-integrating rank) and each column of O is the co-integrating vector. The elements of o are known as the adjustment parameters in the VECM. The Johansen method is to estimate y matrix in an unrestricted form, then test whether we can reject the restrictions implied by the reduced rank of y: We report the values of r (the number of co-integrating relations) in Tables 4–8. The Likelihood ratio test in Table 4 indicates one co-integrating vector at the 5% significance level between the GCC stock markets. This is not surprising since GCC countries have made substantial efforts to integrate their financial markets and have removed all the barriers to the flow of capital between member countries.6 The scenario is not quite similar for the remaining MENA stock markets. While individual MENA countries have devoted significant efforts toward opening and liberalizing their financial markets, they still seem to be segregated as a group casting doubt on the question of whether MENA stock markets are regionally integrated. Table 5 indicates no co-integrating vector at both the levels of significance. While GCC markets appear not to be integrated with the world major stock markets (Table 6), the remaining MENA financial markets have matured and are now integrated with the world financial markets (Table 7). Thus, stock
12
SIMON NEAIME
Table 4.
Johansen Co-integration Likelihood Ratio Test for the GCC Stock Markets.
Hypothesis
Test Statistics
Null
Alternative
r¼0 rp1 rp2
rX1 rX2 r¼3
34.97 10.24 1.86
Critical Values 5%
1%
29.68 15.41 3.76
35.65 20.04 6.65
Source: Author’s estimates. (1) The test is based on the trace of the stochastic matrix. (2) The test allows for a linear deterministic trend in the data. (3) r represents the number of co-integrating vectors. Maximum lag 4 years in VAR. Indicates rejection of the null hypothesis of no cointegration at the 5% significance level. The asymptotic critical values are from Osterwald-Lenum (1992).
Table 5.
Johansen Co-integration Likelihood Ratio Test for the MENA Stock Markets.
Hypothesis
Test Statistics
Null
Alternative
r¼0 rp1 rp2 rp3
rX1 rX2 rX3 r¼4
43.30 19.84 5.57 0.03
Critical Values 5%
1%
47.21 29.68 15.41 3.76
54.46 35.65 20.04 6.65
Note: See ‘Note’ in Table 4. Source: Author’s estimates.
Table 6.
Johansen Co-integration Likelihood Ratio Test for the GCC Stock Markets and the World.
Hypothesis
Test Statistics
Null
Alternative
r¼0 rp1 rp2 rp3 rp4 rp5
rX1 rX2 rX3 rX4 rX5 r¼6
Note: See ‘Note’ in Table 4. Source: Author’s estimates.
92.53 61.96 40.75 23.03 10.25 1.53
Critical Values 5%
1%
94.15 68.52 47.21 29.68 15.41 3.76
103.18 76.07 54.46 35.65 20.04 6.65
13
Portfolio Diversification and Financial Integration of MENA Stock Markets
Table 7.
Johansen Co-integration Likelihood Ratio Test for the MENA Stock Market Index and the World.
Hypothesis
Test Statistics
Null
Alternative
r¼0 rp1 rp2 rp3
rX1 rX2 rX3 r¼4
56.96 30.83 14.87 7.093742
Critical Values 5%
1%
53.12 34.91 19.96 9.24
60.16 41.07 24.60 12.97
Note: See ‘Note’ in Table 4. Source: Author’s estimates. Indicates significance at 5% level.
Table 8.
Johansen Co-integration Likelihood Ratio Test for the MENA Index and the GCC Markets.
Hypothesis
Test Statistics
Null
Alternative
r¼0 rp1 rp2 rp3
rX1 rX2 rX3 r¼4
47.01 20.24 5.09 1.50
Critical Values 5%
1%
54.64 34.55 18.17 3.74
61.24 40.09 23.46 6.40
Note: See ‘Note’ in Table 4. Source: Author’s estimates.
markets can enhance and diversify the portfolios of the remaining MENA region investors through mainly mutual funds. This is also true for international investors seeking diversification in emerging markets. The MENA stock markets offer the equity reach GCC countries diversification potentials not offered by other regional financial markets. Table 8 indicates no cointegrating vectors between all the MENA stock markets (GCC and nonGCC) providing more robust evidence against regional financial integration.
4. VECTOR ERROR-CORRECTION MODEL After establishing that the MENA countries as a group and the world main stock markets are co-integrated, which means there exists a long-run relationship between the two groups, we next employ a VECM to tie the
14
SIMON NEAIME
short-run behavior of each series to its long-run values. The VECM was first introduced by Sargan (1984) and later popularized by Engle and Granger (1987). Engle and Granger have shown that a system of co-integrated variables can be represented by a dynamic error-correction model (ECM) by invoking the Granger’s Representation Theorem. We introduce the lagged residuals (called the error-correction (EC) term) to the model with stationary variables obtained from the long-run relationship. The coefficient on this EC term reflects the adjustment of the dependent variable in the short-run to its long-run position. Engle and Granger’s two-step method is used to estimate the following multivariate error-correction model DPt ¼ a0 þ a1 EC t
1
þ a2 DPt
1þ
3 X
am DPt
m
þ t
(5)
m¼1
where D denotes the first differences, and t is the error term with the usual properties. Regression (5) relates the change in Pt (MENA index)7 to its first lag and that of Pt ; where Pt represents the series of the world financial markets. In this equation DPnt 1 and DPt m capture the short-run disturbances in Pt, whereas the error-correction term ECt–1 captures the adjustment toward the long-run equilibrium. Model (5) assumes a unidirectional relationship from the world stock markets to stock markets in the MENA region. To examine the possibility of a bi-directional relationship between the respective series, the error-correction model is re-estimated after switching the place of DPt and DPnt DPt ¼ a0 þ a1 EC t
1
þ a2 DPt
1
þ
3 X
am DPt
m
þ 1;t
(6)
m¼1
To examine the short-run dynamics of the series we also perform GrangerCausality Tests by running models (5) and (6) without the error-correction term resulting in standard Granger-Causality tests, which are summarized in Table 9. The Granger-Causality tests suggest that the four variables are significantly related in the short run. Specifically, the null hypothesis that the U.S. stock market does not granger cause MENA is soundly rejected at the 1% level of significance. Also the null hypothesis that the U.K. stock market does granger cause MENA is rejected at the 15% level of significance. This suggests that there exist strong unidirectional linkages flowing from the world major stock markets (U.S. and U.K.) to the MENA region. However, the Paris stock market appears not to have any significant impact on the MENA region, but the reverse hypothesis that MENA does not
15
Portfolio Diversification and Financial Integration of MENA Stock Markets
Table 9.
Return Dynamics between MENA Stock Markets and the World.
Short-Run Granger Causality Null Hypothesis P does not granger cause MENA MENA does not granger cause P U.S. does not granger cause MENA MENA does not granger cause U.S. FT does not granger cause MENA MENA does not granger cause FT U.S. does not granger cause P P does not granger cause U.S. FT does not granger cause P P does not granger cause FT FT does not granger cause U.S. U.S. does not granger cause FT Vector-Error Correction Model Long-run VECM VECM equation (5) VECM equation (6) U.S.-dependent variable U.K.-dependent variable Paris dependent variable
Observed 317 317 317 379 379 429
F-Statistics 1.03 5.49 4.12 1.55 1.86 0.75 4.28 2.55 37.21 2.16 18.87 8.29
No. of weeks
F-statistics
317 317
4.48**
Probability 0.35 0.004 0.017 0.21 0.15 0.47 0.014 0.078 1.8E-15 0.11 1.4E-08 0.00029
27.97** 2.95* 31.17**
Source: Author’s estimates. *, **Significance at the 5% and 1% level, respectively.
granger cause the Paris market cannot be rejected. The highly significant Fstatistics on the VECM are consistent with our earlier findings. Specifically, the highly significant F-statistics in both models (5) and (6) provide an additional support for the existence of a strong co-integrating relationship between the MENA markets and the world in the long run. We next turn to explore how the MENA markets would react to shocks in the world financial markets. For this purpose, impulse response functions trace the effect of a one standard deviation shock to one of the innovations on current and future values of the endogenous variable. In other words, a shock to the jth variable directly affects the j-variable, and is also transmitted to all of the endogenous variables through the dynamic structure of the VECM. More specifically, a change in t in Eq. (5) will immediately change the value of the current MENA index series. It will also impact all future values of the MENA index and the change in the world indices since the lagged MENA index
16
SIMON NEAIME
appears in both Eqs. (5) and (6). The impulse response functions shed light on the dynamics of the variables included in the VECM system as a result of shocks to either one of these variables. The tests show that shocks to both the U.S. (S&P) and U.K. (FT) stock markets affect significantly the MENA index but not the other way round (see Fig. 1). Specifically, a one standard deviation
Response to One S.D. Innovations ± 2 S.E. Response of MENA to US
Response of MENA to FT
Response of MENA to P
Response of MENA to MENA
0.06
0.06
0.06
0.06
0.04
0.04
0.04
0.04
0.02
0.02
0.02
0.02
0.00
0.00
0.00
0.00
-0.02
-0.02
-0.02 2
4
6
8
10
2
Response of US to US
4
6
8
10
-0.02 2
Response of US to FT
4
6
8
2
10
0.020
0.020
0.020
0.020
0.015
0.015
0.015
0.015
0.010
0.010
0.010
0.010
0.005
0.005
0.005
0.005
0.000
0.000
0.000
0.000
-0.005
-0.005
-0.005
-0.005
-0.010
-0.010 2
4
6
8
10
-0.010 2
Response of FT to US
4
6
8
10
4
6
8
10
Response of FT to P
2
0.020
0.020
0.020
0.015
0.015
0.015
0.015
0.010
0.010
0.010
0.010
0.005
0.005
0.005
0.005
0.000
0.000
0.000
0.000
-0.005
-0.005
-0.005
-0.005
-0.010 2
4
6
8
10
-0.010 2
Response of P to US
4
6
8
10
4
6
8
10
2
Response of P to P
0.020
0.020
0.020
0.015
0.015
0.015
0.015
0.010
0.010
0.010
0.010
0.005
0.005
0.005
0.005
0.000
0.000
0.000
0.000
-0.005 2
4
6
8
10
-0.005 2
Fig. 1.
4
6
8
10
4
6
8
10
4
6
8
10
Response of P to MENA
0.020
-0.005
10
-0.010 2
Response of P to FT
8
Response of FT to MENA
0.020
-0.010
6
-0.010 2
Response of FT to FT
4
Response of US to MENA
Response of US to P
-0.005 2
4
6
8
Impulse response functions.
10
2
4
6
8
10
Portfolio Diversification and Financial Integration of MENA Stock Markets
17
positive shock to the S&P seems to affect significantly the MENA stock markets permanently and for a period larger than 10 weeks. The effects of the U.K. market on the MENA markets seem to die out after 8 weeks but that of Paris (P) seem to be insignificant. This can be attributed to the fact that cultural, financial, and economic relations between the U.S. and the U.K. are much more important with Egypt, Turkey, and Jordan than they are with Morocco, the only country with important ties with France. Since Morocco’s share in total MENA market capitalization is much smaller than its remaining counterparts, the French stock market seems to affect insignificantly the MENA markets.
5. CONCLUSIONS AND POLICY IMPLICATIONS This chapter highlighted some important aspects of financial integration in the MENA region, and between MENA and the rest of the world. After exploring the main characteristics of the MENA financial markets this chapter used a dynamic model to empirically study the implications of financial integration both at the regional and international levels. The issue of financial market integration has received considerable attention in the finance literature after portfolio managers realized that emerging financial markets offer diversification potentials not offered by more mature markets. Our empirical results confirmed that while the stock markets of Egypt, Turkey, Jordan, and Morocco have matured and are co-integrated with the world financial markets, evidence of regional financial integration is still weak except among the GCC stock markets. Although the GCC stock markets appear to be segregated from the rest of the world, they can still offer diversification potentials to international and regional investors through mainly mutual funds. Next, we turned our attention to the examination of linkages and spillover effects among MENA stock markets and between them as a group and the world markets. Our findings showed that the main stock markets of the MENA region reacted to their world counterparts but smaller GCC markets were isolated from the rest of the world. This finding took us a step further in confirming that the non-GCC MENA stock markets were maturing, and becoming integrated with the world stock markets. While integration is generally a goal of any emerging market, it offers little reward to international investors seeking diversification. If all stock markets were fully integrated, investors will not find the diversification benefits they desire by tapping into emerging markets. Our results suggest that the stock markets of
18
SIMON NEAIME
Istanbul, Cairo, Jordan, and Casablanca appear already integrated with the rest of the world markets. In addition to this long-run linkage, there is also strong evidence of important short-run granger-causality effects flowing unidirectionally from the world main stock markets to the MENA markets, and shocks to the U.S. and U.K. markets appear to significantly affect the MENA region and for long periods of time. The removal of the restrictions and barriers to the flow of capital into the GCC region’s financial markets is expected to improve and enhance economic growth and development in these countries. The finance literature has shown that there exists a strong link between financial liberalization and economic growth. Increased inter-MENA liberalization will not only increase allocative efficiency within MENA, as the experience of the developed economies shows, but will also provide MENA investors with greater opportunities to diversify their portfolios and reduce risks. Increased financial liberalization within the MENA region is expected to enhance regional intermediation of resources through close integration of financial markets and increased access of MENA residents to the region’s financial markets to finance investment. In addition, MENA investors will have access to a variety of risks-adjusted rates of return to enhance the efficiency of portfolio allocation and diversification, which will foster the efficiency of MENA’s financial markets. Increased liberalization within the MENA region is expected to attract important portfolio and direct investments to the region.
NOTES 1. GCC countries are mainly the MENA oil-producing countries: Bahrain, Oman, Qatar, Kuwait, Saudi Arabia, and the United Arab Emirates. In this chapter the GCC countries are Bahrain, Kuwait, and Saudi Arabia due to the lack of data for the rest of the GCC countries. 2. Co-integration tests involve two steps. First, each time series is examined to determine its order of integration. This involves Unit Root Tests based on the work of Fuller (1976), Dickey and Fuller (1979; 1981), and Phillips and Perron (1988). These tests involve the calculation of t-statistics for r ¼ 1 in ordinary least square (OLS) regressions of the form: X t ¼ rX t 1 þ ut (1), where Xt is the variable of interest and ut is independent and identically distributed (iid), Nð0; s2 Þ: The t-statistic does not have the standard t-distribution; critical values under the null that r ¼ 1 are found in Table 8.5.2 of Fuller (1976). If ut violates the iid assumption, modifications to Eq. (1) must be implemented in conducting the test. The ADF (p) test supplements Eq. (1) with p-lagged changes in the dependent variable as additional regressors. Alternatively, the PP test involves the estimation of Eq. (1), coupled with a
Portfolio Diversification and Financial Integration of MENA Stock Markets
19
non-parametric correction of the t-statistic for general forms of autocorrelation in the errors. Eq. (1) is often expressed in an alternative form as DX t ¼ ðr 1ÞX t 1 þ ut ¼ dX t 1 þ ut ; where d ¼ ðr 1Þ and D is the first-difference operator. This equation is equivalent to Eq. (1); however, now the null hypothesis is that d ¼ 0. This is the line followed in the chapter. 3. A time series is integrated of order d (denoted as I(d)) if it must be differenced d times in order to induce stationarity. 4. Examples of possibly co-integrated economic variables are short- and long-term interest rates; prices and wages; prices and money supply; and consumption and income. Equations estimated with stationary variables but without regard to the underlying co-integration are also inappropriate due to the model mis-specification (e.g. an omitted-variable bias). 5. Studies by Cheung and Lai (1993) and Gonzalo (1994), among others, provide ample evidence supporting the use of the Johansen approach over alternative tests. 6. Since 1997, Bahrain and Kuwait have linked their stock markets by allowing the cross-listing of local stocks. Financial markets that are located in the same geographic area and have identical cohorts of investors are bound to have stock markets which react to various shocks in the same way. In addition when a stock is cross-listed in more than two markets, then a shock in one market is likely to be transmitted to the other because investors will tend to react to various financial shocks in a similar way. 7. The MENA index is a market capitalization-based weighted index constructed for Egypt, Jordan, Turkey, and Morocco.
REFERENCES Arshanapalli, B., & Doukas, J. (1993). International stock market linkages: evidence from the pre- and post-October 1987 period. Journal of Banking and Finance, 17, 193–208. Bekaert, G. (1993). Market integration and investment barriers in emerging equity markets. In: S. Claessens & S. Gooptu (Eds), Portfolio investment in developing countries. Washington, DC: The World Bank. Bekaert, G., & Harvey, C. R. (1997). Emerging equity market volatility. Journal of Financial Economics, 43, 403–444. Bulter, K. C., & Malaikah, J. (1992). Efficiency and inefficiency in thinly traded stock markets: Kuwait and Saudi Arabia. Journal of Banking and Finance, 16, 197–210. Cheung, Y. W., & Lai, M. (1993). Do gold market returns have long memory? Financial Review, 12(2), 182–194. Cheung, Y., & Mak, S. (1992). A study of international transmission of stock market fluctuations between the developed markets and the Asia-Pacific markets. Applied Financial Economics, 2, 1–5. Chowdhry, A. (1994). Stock market interdependence: evidence from the Asian NIEs. Journal of Macroeconomics, 16(4), 129–651. Darrat, A. F., & Hakim, S. R. (1997). Price linkages, efficiency, and integration of emerging stock markets in the Middle East. Paper presented at the ERF Fifth Annual Conference on Regional Trade, Finance and Labor Markets in Transition, Beirut, Lebanon, 7–9 October.
20
SIMON NEAIME
De Santis, G., & Imrohoroglu, S. (1997). Stock returns and volatility in emerging financial markets. Journal of International Money and Finance, 16(4), 561–579. Dickey, D., & Fuller, W. (1979). Distribution of the estimators for autoregressive time series with a unit root. Journal of the American Statistical Association, 74, 427–431. Dickey, D., & Fuller, W. (1981). Likelihood ratio statistics for autoregressive time series with a unit root. Econometrica, 49, 1057–1072. El-Erian, M., & Kumar, M. (1995). Emerging equity markets in Middle Eastern countries. International Monetary Fund Staff Papers, 42(2), 313–343. Engel, R., & Susmel, R. (1993). Common volatility in international equity markets. Journal of Business and Economic Statistics, 11(2), 167–176. Engle, R. F., & Granger, C. W. (1987). Granger: co-integration and error-correction representation, estimation, and testing. Econometrica, 55, 251–276. Eun, C., & Shim, S. (1989). International transmission of stock market movements. Journal of Financial and Quantitative Analysis, 24, 241–256. Fuller, W. (1976). Introduction to statistical time series. New York: Wiley. Gonzalo, J. (1994). Five alternative methods of estimating long-run equilibrium relationships. Journal of Econometrics, 60, 203–223. Hakim, S., & Neaime, S. (2000). Perspectives on the integration of financial markets. In: C. C. Paraskevopoulos, et al. (Eds), Global financial instability and economic development. Toronto, Canada: APF Press. Joen, B. N., & Von Furstenberg, G. M. (1990). Growing international co-movement in stock price indexes. Quarterly Review of Economics and Business, 30, 15–30. Johansen, S. (1991). Estimation and hypothesis testing of cointegration vectors in gaussian vector autoregressive models. Econometrica, 59, 1551–1580. Johansen, S. (1995). Likelihood-based inference in cointegrated vector autoregressive models (Vol. 14, pp. 45–46). Oxford: Oxford University Press. Kasa, K. (1992). Common stochastic trends in international stock markets. Journal of Monetary Economics, 29, 95–124. Kim, S. W., & Wadhwani, S. (1990). Transmission of volatility between stock markets. Review of Financial Studies, 3, 5–33. Korajczyk, R. (1996). A measure of stock market integration for developed and emerging markets. The World Bank Economic Review, 10(2), 12–33. Osterwald-Lenum, M. (1992). A note with quantiles of the asymptotic distribution of the maximum likelihood cointegration rank test statistics. Oxford Bulletin of Economics and Statistics, 54, 461–472. Phillips, P. C., & Perron, P. (1988). Testing for a unit root in time series regression. Biometrika, 57, 335–346. Sargan, D. (1984). Wages and prices in the United Kingdom: a study in econometric methodology. In: D. F. Hendry, et al. (Eds), Quantitative economics and econometric analysis. Oxford: Basil Blackwell.
FOREIGN DIRECT INVESTMENT IN TUNISIA: ROLE OF THE FREE TRADE AGREEMENT WITH EUROPEAN UNION Mohamed Abdelbasset Chemingui and Nora Ann Colton ABSTRACT The aim of this study is to determine whether Tunisia could expect an increase in Foreign Direct Investment (FDI) flows in response to the establishment of a Free Trade Agreement (FTA) with the European Union (EU). While the conditions necessary to stimulate the flow of FDI have received considerable attention from economists in recent years, the relationship between trade policy and FDI has not been the subject of in-depth research. The study finds that the partnership agreement between Tunisia and the EU can play a catalytic role in increasing not only the openness of the Tunisian economy, but, subsequently, increasing FDI to Tunisia.
1. INTRODUCTION The 1950s marked the beginning of European integration, and the foundation for what would become a trend of coordinating economic policies. Money and Finance in the Middle East: Missed Opportunities or Future Prospects? Research in Middle East Economics, Volume 6, 21–41 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1094-5334/doi:10.1016/S1094-5334(05)06002-4
21
22 MOHAMED ABDELBASSET CHEMINGUI AND NORA ANN COLTON
Since that time, a great number of regional integration agreements (RIAs) have been created between developed and developing countries. Between 1947 and 1994, the General Agreement on Tariffs and Trade (GATT) secretariat registered a total of 108 RIAs (Al-Khalidi, 1998). This number represents nearly 90% of the total number of countries belonging to the World Trade Organization. There are three primary types of regional agreements. The first is a North–North model of integration, which assembles developed or industrialized countries. The European Union (EU) offers the best example of this type of agreement. The second type is a North–South arrangement, which groups countries of different levels of development. The North American Free Trade Agreement (NAFTA) links Canada, the United States and Mexico to form this type. The third type, a South–South model of economic integration, is established between developing countries. One example is MERCOSUR, which constitutes a Free Trade Agreement (FTA) between a conglomerate of Latin-American countries. The emergence of continental FTAs in industrialized parts of the globe became a reality in the world economy at the end of the 1980s. This ‘‘Continental Regionalism’’ constrains the trade policy options of many lessdeveloped economies, which find themselves compelled to join a trade bloc (Kebadjian, 1995). As Mahjoub (1996) explained, it is impossible for developing countries to remain outside the three great trade blocs (North America, Europe and South Asia) unless they agree to remain outside the international flow of goods and capital, and potentially sacrifice their economic prosperity. Although some regional integration agreements (RIAs) have been the result of political considerations, most are used to accomplish economic objectives. According to Blomstrom and Kokko (1997a,b), regional integration brings short-term benefits by increasing inter-regional trade and investment. In the long run, the establishment of a larger regional market should encourage competition, allocate resources more efficiently, and ultimately provide positive and varied externalities to signatory countries. This intra-regional trade has been credited with promoting economic growth and development. The benefits that developing countries gain from economic integration were demonstrated by the success of the ‘‘Four Asian Dragons’’ marketeconomy policy. This perceived success in Asia convinced many countries of the south, as well as international organizations such as the World Bank/ International Monetary Fund, that integration agreements could be key to successful growth and development. Thus, as noted by Burniaux and
Foreign Direct Investment in Tunisia
23
Waelbroeck (1995), few countries still rely on a trade policy of import substitution for growth and development. The possibility of the establishment of an FTA between Tunisia and the EU by 2010 poses challenges as well opportunities. In addition to the typical challenges associated with creating an FTA between countries with unequal development levels (North–South), the specific trade relations between Tunisia and the EU must be considered. Indeed, while the European Market is almost completely open to Tunisian industrial products, the Tunisian Market has been strongly protected against the entry of European products. Some of the compelling reasons for the establishment of an FTA with the EU could be to attract Foreign Direct Investment (FDI) to Tunisia in addition to allowing European products into Tunisia. If FDI is an outgrowth of trade liberalization, the key issue is to define the role that FDI could play in the economic development of a country like Tunisia. For example, the implementation of an integration policy should be evaluated in terms of its effectiveness in increasing FDI to recoup the adjustment costs of trade liberalization. Therefore, the aim of this chapter is to determine whether or not Tunisia could expect an increase in FDI inflows as a result of the establishment of an FTA with the EU. While the conditions to stimulate the flow of FDI have received considerable attention from economists in recent years, the relationship between trade policy and FDI has not been the subject of in-depth research. The analysis will be conducted as follows. Section 2 examines the benefits of trade liberalization and FDI flows in terms of productivity. Section 3 focuses on the relationship between trade liberalization and FDI in general, as well as on the role that regional agreements could play in its attraction. In Section 4, the evolution of FDI flows to Tunisia is reviewed. Section 5 develops a computable general equilibrium (CGE) model, to measure the impact of trade liberalization on the volume of FDI going to Tunisia. Section 6 presents the results of this model and Section 7 draws conclusions.
2. TRADE LIBERALIZATION, FDI FLOWS AND PRODUCTIVITY International trade economists have long argued that a bilateral and outward-looking trade regime is the best strategy for a small open economy that takes international prices as given. An outward-looking trade regime increases welfare by leading to an optimum allocation of resources in
24 MOHAMED ABDELBASSET CHEMINGUI AND NORA ANN COLTON
production. The literature points to three main channels through which an open trade policy affects growth. These channels are investment, productivity and government policy. Trade liberalization creates open trade regimes that increase exposure to a worldwide stock of productivity enhancing knowledge, which can generate growth. Ben-David and Loewy (1995) state that the growth impact of free trade derives from the premise that: y (1) knowledge may be characterized as a non-rivalries public good, which in many cases is non-excludable, and (2) trade flows facilitate the diffusion of knowledge among countries y Heightened trade will, in general, lead to greater diffusion and faster knowledge growth and hence, faster per capita income growth.
The main trade-related factor that can explain the increase in productivity is access to foreign intermediates and capital goods, as well as access to world markets. Firms then have the opportunity to buy intermediate goods and equipment that allow them to improve their productivity (Djankov & Hoekman, 1997). The study by Djankov and Hoekman (1997) claims to have established a clear link between trade liberalization and productivity. Although the analysis is limited to the level of the firm, the results indicate that trade is an important source of total factor productivity (TFP) growth. The study finds that shifts in the pattern of imports of intermediate goods and the reorientation of export production towards global markets are positively correlated with TFP. There is also an important relationship between FDI and productivity. Rutherford and Tarr (1997) attempt to estimate the impact of trade liberalization on the accumulation rate of capital using a general equilibrium model. They assume that the available capital stock of each country is optimal and chosen only for its return. Another source notes that if the return on capital increases after liberalization, then agents will be encouraged to invest until the marginal productivity of capital returns to its level prior to the reform (Dessus & Suwa-Eisenmann, 2000). This approach presents evidence that there is a potential for new investments to be made given the implementation of an FTA, as long as the agreement leads to an increase in the productivity of physical capital. The study used a multiple regression model that explained the creation of fixed capital. This model places the growth rate of capital within a set target, which is determined by capital productivity and the domestic savings rate. The results of applying this econometric analysis to the case study of Egypt demonstrated that capital returns have a significant and positive impact on the growth of physical capital (Dessus & Suwa-Eisenmann, 2000).
Foreign Direct Investment in Tunisia
25
3. LINKAGES BETWEEN FDI AND FTAS 3.1. Developing Countries’ Motivations to Integrate As underlined by Lawrence (1996), the motives behind the recent integration agreements differ radically between developed and developing economies. In fact, contrary to the agreements of the 1930s, 1950s and 1960s, recent agreements are aimed at providing their members with better participation in the world economy. The recent commitments of developing countries to FTAs indicate that these countries are willing to adopt a strategy of opening their economies in order to promote exports and attract FDI, rather than substitute imports with overprotected, and often unprofitable, local production – a strategy that was popular in the developing world during the 1960s and 1970s. Through regional integration, less-developed countries are attempting to become more attractive to the exporting enterprises of their partner countries. There are many reasons why it is beneficial for developing countries to attract FDI. Some of these include the (1) creation of jobs, (2) transfer of technology and knowledge, (3) improvement of a country balance of payments and (4) development of down- and upstream activities. For more developed countries, FTAs are seen as an opportunity to improve the competitiveness of certain sectors by taking advantage of different factor endowments and fiscal exemptions. The establishment of NAFTA, for example, was mainly motivated by certain large US companies, especially firms with high labor costs, who believed that the establishment of an FTA with Mexico would eventually provide them with a competitive advantage. Such a redeployment of activities could not occur if tariffs and non-tariff barriers were in place that would affect capital flows and finished product trade. Although consumers may also benefit from lower prices, it appears that RIAs are generally motivated and defended by ‘‘Big Business,’’ usually multinational corporations, as they appear to be the main beneficiaries. The intensification of international competition has encouraged the creation of many of the recent RIAs. These agreements come at a time when access to new markets has become important to secure the success of companies. Economies can no longer limit their activities to the local level due to the rapid expansion of technological progress. As companies develop new technology, they rush to cover the fixed costs related to this innovation before competitors can access it. FTAs increase trade and, consequently, inflate profits so that companies can cover these costs.
26 MOHAMED ABDELBASSET CHEMINGUI AND NORA ANN COLTON
Analysis of FTAs needs to go beyond examining the static effects of RIAs (trade creation and trade diversion), and examine the dynamic effects as well. The latter were considered a key to the success of various RIAs and became the primary motivation for developing countries to create FTAs. The first expected dynamic effect of an FTA is an increased FDI flow. This can be beneficial in many ways. Foreign firms bring capital into a country by taking part in privatization, by introducing technology and by their knowledge of markets. The transfer of technology connected with international trade represents another dynamic aspect of RIAs. These transfers of technology could follow various channels: the promotion of exports and imported inputs lead to a better quality of finished product and introduce know-how, and direct investment is a bearer of new technologies (Dessus & Suwa-Eisenmann, 2000). To understand the impact of an RIA in attracting FDI, it is important to understand the relationship that may exist between trade policy and FDI flows.
3.2. The Relationship between Trade and FDI Most of the empirical studies on the relationship between FDI and trade attempt to discover if trade and FDI are substitutable, reciprocal or complementary. Any discussion of the implications of the relationship between trade and FDI for economic growth and development must first establish a working hypothesis about the nature of this relationship. As one of the objectives of this study is to examine the impact of an agreement with the EU on the volume of FDI inflows to Tunisia, it is important to identify how trade liberalization could affect the level or the degree of attraction of FDI by Tunisia. Trade policies may have an influence on FDI promotion in various ways. If a country imposes high duties or other obstacles to trade, FDI may increase in order for companies to serve the local market. Thus, these investments are motivated by the idea that the additional cost linked to the production on the foreign market is more than compensated for by the fact that there are fewer risks than if a company were to export to the country. On the other hand, FTAs may cause firms to seek out low-wage regions leading to more FDI. Although some receiving countries intentionally resort to high duties to attract investments, the advantages stemming from this policy seem to be limited. The FDI attracted by protected markets generally takes the form of independent production units oriented towards the local market, which are
Foreign Direct Investment in Tunisia
27
not competitive in the export field. Indeed, the high duties levied on raw materials and imported intermediate products can reduce competitiveness on the international scale, especially if the local inputs are expensive and poor quality. In order to circumvent these negative effects, receiving countries often establish regimes of ‘‘duties discount’’ for the foreign inputs of export-oriented companies. This is one of the key incentives offered to foreign investors, particularly in the export industry. The size of the local market as well as the technological structure of a firm are important elements in the decision of a multinational company to invest in a host country. By removing the internal obstacles to trade, an FTA or tariff union gives companies the opportunity to sell their products in an integrated market from one or more production sites, and consequently to profit from any economies of scale that might exist. This could clearly influence investment flows, at least during the period when companies restructure their production activities. The most recent theoretical and empirical analysis on FDI tends to study trade and the movement of capital as substitutive modes to foreign market service. This point of view on the relationship between trade and the mobility of production factors confirms that tariff barriers and other restrictions to imports encourage FDI. This means that the generalized reduction of tariffs entails a reduction and even a repatriation of FDI to their original country or to countries, which are still protected by high tariffs. The same relationship could exist between exports and non-tariff barriers because these require the establishment of subsidiary firms abroad, or the granting of licenses to local producers to sell their products in the foreign market. RIAs, which reduce or remove the tariff and non-tariff barriers, could make exporting products easier and more profitable in financial terms, and discourage FDI. Since reductions in tariff and non-tariff barriers are the main characteristics of RIAs, local and foreign investment in a given country can be affected in many ways by regional integration. In analyzing the inter-regional FDI flows, it is expected that these flows will decrease as trade liberalization makes exportation from the original country relatively more attractive than FDI. However, there will obviously be a need to change the structures of regional production, and consequently more flows of investment will go from one member country to another. Thus, inter-regional FDI could increase between some member countries in response to the emergence of new investment opportunities, while decreasing between other member countries, following the absence of redeployment of firms.
28 MOHAMED ABDELBASSET CHEMINGUI AND NORA ANN COLTON
The size of investments made by a company abroad also depends on the relative power it has among the member states of the agreement. FDI flows tend to be relatively limited if the companies that can benefit the most from the new agreement are already evolved in the profitable production sectors. This potential impact on the inter-regional FDI flows is known as ‘‘Investment Diversion’’ (Kindleberger, 1966). Despite the argument to the contrary, analysis suggests that there are many reasons to expect growth in FDI flows in a country following the creation of an inter-regional agreement. FDI flows coming from non-member countries could certainly increase if the protection level is only reduced for member countries. The flows of foreign capital could also increase if they were initially limited by the size of each single market. In this way, the common integrated market would be large enough to take over the fixed costs of the establishment of new foreign subsidiary companies. This phenomenon has been termed ‘‘Investment Creation,’’ since investment is created in response to the effect of trade diversion as a result of regional integration (Kindleberger, 1966). However, it is possible to envision a situation where regional integration may lead to a reduction of the FDI coming in from countries that are not members. In a more specific manner, if the initial stock of FDI of non-member countries is organized in horizontal branches in most of the countries of the region, it is not certain that such structures may remain optimal after the establishment of the FTA. Therefore, the strongly defended arguments in favor of the growth of FDI flows coming from non-members countries seems to be ambiguous and irrelevant. The integrated region could compensate for the diminished FDI flows by consolidating the branches of the country-members, so that the whole region may be supplied by a small number of branches located in only a few of the country-members where economic conditions are thought the best. In this case, similar to the case of diversion of investment demonstrated by Kindleberger, some member countries could expect disinvestment operations, as foreign firms would center their production in a few of the other member countries. Currently, studies are not able to ascertain the consequences of RIAs on the flow of FDI coming from member countries. However, if it is assumed that the FDI from within the region and that coming from the Rest of the World (ROW) are substitutable, then some conclusions can be drawn. The most probable effect of an RIA will be to increase intra-regional investment in comparison with investment coming from the ROW. This wide study has analyzed the relationship between trade policy and FDI attraction and
Foreign Direct Investment in Tunisia
29
shown how the ability of an economy to attract FDI could be influenced by the changes in the trade policy. Another aspect that must be noted is the effect of the variation of FDI on export volume of the receiving country. This aspect is more interesting when we study cases of developing countries that are seeking to expand their exports and decrease the deficit in their balance of payments. Many studies on a wide range of sectors have concluded that there is a positive correlation between the total flow of FDI and the total exports of the receiving country (Hill, 1990; Hummel & Stern, 1994). This point of view originated in the traditional theory concerning FDI, which is based on the idea that it is possible to use goods produced by foreign companies in the host country to replace the domestically produced exports that are usually sold in foreign markets. It is important to keep in mind that not all firms produce goods that can be exported. Some goods are non-tradable because of transport costs, and many firms also sell services rather than goods, often requiring a local presence to do so. For these firms, there is no option of exporting and no direct trade effect in either the home or host country. In addition, not only may FDI growth increase export activity, but also it is believed to reduce the imports of the receiving country. This can be substantiated by the hypothesis that FDI and the imports of the receiving country are substitutable. An important theoretical article (Mundell, 1957) argues that the free circulation of capital (i.e. production factor mobility) could substitute for free trade of goods and services. This relationship is at the core of Mundell’s analysis. Second, as noted earlier, obstacles against importation were set up by governments to encourage FDI from foreign investors attempting to avoid duties. The results were that the local production replaced imports. However, it must be taken into consideration that while imports of finished goods decrease as a result of these barriers to trade imports of equipment goods and intermediary consumption goods increased.
4. FDI IN TUNISIA: STRUCTURE AND DETERMINANTS 4.1. Structure of FDI Inflows to Tunisia At present, the FDI that is flowing into Tunisia is concentrated primarily in three sectors: energy (i.e. oil search equipment and the construction of oil pipelines), tourism and textiles.
30 MOHAMED ABDELBASSET CHEMINGUI AND NORA ANN COLTON
Table 1.
Trend in FDI in Tunisia Breakdown by Beneficiary Sector (in Millions TD).
Total Energy Tourism and real estate Financial institutions Manufacturing industries Other sectors
1992
1993
1994
1995
1996
1997
1998
1999
516.1 89.4 2.5 0.6 3 4.5
659.9 93.3 1.1 1.1 2.3 2.3
542.8 91.1 3.1 0 2.6 3.2
305.1 80.4 9.2 0 7.9 2.4
272.5 61.2 17.8 0 18.2 2.8
402.9 67.3 5.7 0 21.3 5.7
759.9 26.6 3.2 0 68.9 1.3
437.2 44.6 8.5 0 45.3 1.7
Source: Our calculations on the basis of Central Bank of Tunisia’s data.
The energy sector accounted for 90% of the total volume of FDI during the period 1992–1995, but that figure dropped to 45% in the 4 years following the application of the FTA with the EU, beginning in 1996. In contrast, the growth of the tourism sector more than doubled during these two time periods, growing from 3.2% to 7.1%. FDI also increased in the manufacturing sector, from 3.4% of total FDI during the first period to 46% in the second. The increase in FDI in these latter two sectors was due not only to a decline of FDI in the energy sector, but also to the implementation of an intense privatization program. Manufacturing companies were the first targets of the privatization of industries in Tunisia. As a result, their share of total FDI inflows increased rapidly. Later, the privatization of services, especially in the tourism sector, resulted in the sale of state-owned companies, and ultimately increased FDI in this sector as well. Finally, export-oriented investors who were attracted by generous incentives and an abundant labor force undertook greenfield investment in Tunisian industries such as mechanics, electric, electronic and textiles. Table 1 shows the distribution of FDI in Tunisia, broken down by sector, during the last decade. Rather than leading to increases, in FDI, privatization and the establishment of an FTA have been linked to considerable fluctuations in FDI inflows to Tunisia. In 1994 and 1998, FDI increased dramatically. The increase in 1994 can be explained by two factors. First, construction of the Mediterranean gas pipeline, which sends Algerian gas to Italy, began in this year. Second, the realization that large direct investments were made on the Miskar Oil Site, managed by British Gas (a foreign firm that invests in Tunisia), caused the volume of total FDI to increase.
Foreign Direct Investment in Tunisia
31
In 1998, foreign participation increased sharply with the privatization of two cement factories purchased by foreign companies. This privatization totaled nearly US$400 million, which represented more than the sum of the benefits stemming from privatization during the 1987–1997 decade and more than the two-thirds of all of the FDI registered in 1998. The data suggests that Tunisia has not benefited significantly over time from the establishment of an FTA with the EU or from the expansion of world FDI flows. If privatization operations or the investments in the energy sector are not taken into consideration, Tunisia has experienced a significant decline in terms of FDI flows in industrialized products since it signed the FTA with the EU in 1995. In fact, the FDI/GDP ratio (in which GDP is the gross domestic product) lost 1.3% points between 1992–1995 and 1996–1999. However, it is still too soon to evaluate the real impact of this agreement on the amount of FDI, since it was signed on March 1998 (although Tunisia started its application 2 years before).
4.2. Investment Incentives in Tunisia The importance of FDI for the improvement and growth of a developing country like Tunisia has become the dominant paradigm among some development economists. These individuals believe that only the private sector can assure reduced unemployment rates and a higher quality of life; consequently, a country needs to ensure that its private sector is strong. However, a combination of poor mobilization of national savings and an increasing need for technological resources results in weak domestic private investment. Again, it is believed that if a developing country switches from relying on import substitution, to encouraging FDI, it will strengthen its economic condition and be better equipped to provide for its citizens. The benefits of this change will be able to cushion the shock associated with the liberalization process. Tunisia established a code in the beginning of the 1970s to attract investment. This code, later modified in the mid-1990s and titled the Investment Incentives Code, was created in order to cope with the new objectives of Tunisian economic policy. It was founded on the notion of freedom of investment and offered specific advantages to both domestic and foreign investors in four major ways: (1) Reinvested profits become tax free (within the limit of 35% of the taxed revenues and profits).
32 MOHAMED ABDELBASSET CHEMINGUI AND NORA ANN COLTON
(2) A custom duty exemption exists for equipment goods which cannot be found locally. (3) The value-added tax (VAT) is limited to 10% on the import of equipment goods. (4) Investors have the opportunity to benefit from a ‘‘repayment-schedule’’ regime or decreasing ‘‘Paying-Off of Capital Depreciation’’ for all production equipment and other materials whose period of use exceeds 7 years (Ministry of Economic Development, 1994). In addition to the measures taken to attract investments, advantages are granted to specific types of investments, especially those that are either partially or completely export oriented. Regarding the partially exporting firms, these measures consist of complete tax exemption on profits linked to exports during the first 10 years and, starting in the 11th year, an exemption within a limit of 50% for an unlimited period of time (Ministry of Economy, 1994). Export-oriented companies receive the following benefits: (1) A complete tax exemption for reinvested profits and revenues. (2) A total exemption on duties and taxes concerning equipment goods, including the goods’ shipping equipment, raw materials, semi-products and services necessary for production. (3) The right to sell up to 20% of their products on the local market. Commercialized products remain under the taxes and duties in force (Ministry of Economy, 1994). Enticements are also given for investments that target specific sectors of the economy. For example, benefits are given to encourage development in poorer geographic areas, agricultural industry, environmental protection and technological advancements.
5. THE GENERAL EQUILIBRIUM MODEL The following paragraphs are not intended to describe precisely the characteristics of the model employed here, which contains around 6000 equations (see Beghin, Dessus, Roland-Holst, & van der Mensbrughhe, 1996). Instead, the discussion is intended to describe, in non-mathematical terms, the model’s main hypotheses, mechanisms and the statistical information used for Tunisia.
Foreign Direct Investment in Tunisia
33
The model assumes that the stock of capital invested in Tunisia is determined by the savings rate and capital rental rate. As trade liberalization raises the productivity of capital, the capital stock increases until the marginal product is equal to the initial rental rate. This additional investment is the result of trade liberalization. Although this kind of analysis concerning the impact of an FTA has been the object of several studies, very few studies have used a dynamic CGE model. In addition, prices are endogenous in each market (goods, factors) and equalize supplies and demands in order to obtain equilibrium. The equilibrium is general in the sense that it concerns all markets simultaneously. The model uses the information contained in Tunisia’s social accounts matrix for 1992 (Chemingui & Dessus, 1999). It considers 57 economic sectors and five types of employment, the latter being distinguished by their levels of qualification and geographic mobility. The model also takes into account three types of capital: physical capital, reserves of natural resources (crude oil, phosphates) and land. Finally, the model distinguishes two trading partners for Tunisia: the EU and the ROW. The model is dynamic and is resolved recursively each 3 years from 1992 to 2010. Production is modeled using nested constant elasticity of substitution (CES) functions, which describe the substitution and complement relations among the various inputs. It is assumed that constant returns to scale QJ;exist and producers are cost-minimizers. Output results from two composite goods: intermediate consumption and value-added plus energy. The intermediate aggregate is obtained by combining all products in fixed proportions (Leontieff structure). The value-added and energy components are decomposed into two parts: aggregate labor and capital plus energy. Labor demand is broken down into five categories, as mentioned above. Within each segment, labor is totally mobile and completely employed. The composite capital/energy factor is desegregated into capital and energy. Demand for physical capital makes a distinction between old and new capital. The model thus integrates the notion of vintage capital to distinguish the process of allocating capital already installed from that resulting from contemporary investment (putty/semiputty production function). New capital can be allocated more flexibly than old capital since it substitutes for other types of capital more easily. Accelerating investment therefore strengthens the capacity for the productive sector to adjust to changes in relative prices. Finally, the energy aggregate comprises two types of energy, oil/gas and electricity, which are substitutes.
34 MOHAMED ABDELBASSET CHEMINGUI AND NORA ANN COLTON
Income from labor is allocated between the various households using a standardized fixed-coefficient distribution matrix. Income from capital is allocated in the same way between households, companies and foreign investors. Companies pay the tax on this income and save the remainder. Household demand is derived from maximizing the utility function (following the ELES system; Lluch, 1973) specific to each household, and subject to the constraints of available income and the consumer price vector. Household utility is a positive function of consumption of various products and savings. Income elasticities are differentiated by product and by household, and vary from 0.75 for staple products for the richest households to 1.20 for services. The calibration of the model determines a per capita subsistence minimum for each product (the aggregate consumption of a product grows with population), while the remaining demand is derived through an optimization process. Government and investment demands are desegregated into sectoral demands once their total value is determined according to fixed-coefficient functions. The model assumes imperfect substitution among goods originating from different geographic areas. Import demand results from a CES aggregation function of domestic and imported goods (Armington, 1969). Export supply is symmetrically modeled as a constant elasticity of transformation (CET) function. Producers decide to allocate their output to domestic or foreign markets depending on relative prices. At the second stage, importers (exporters) choose the optimal choice of demand (supply) across regions, again as a function of the relative import (export) prices and the degree of substitution across regions. Substitution elasticity between domestic and imported products is set at 2.2 and at 5.0 between imported products, depending on whether they are from the EU or the ROW. The elasticity of transformation between products intended for the domestic market and products for export is 5.0 and 8.0 between the different destinations for export products. The small country assumption holds in this analysis. In other words, Tunisia is unable to change world prices; thus, its import and export prices are exogenous. Capital transfers are also exogenous, and determine the trade balance. The equilibrium condition on the balance of payments is combined with other closure rules so that the model can be solved for each period. First, consider the government budget. Its surplus/deficit is exogenous and the household income tax schedule shifts in order to achieve the predetermined net government position. Second, investment is savings driven, with savings originating from households, enterprises, the government and abroad. The sequential dynamic path of the model results from this closure rule.
35
Foreign Direct Investment in Tunisia
A change in savings influences capital accumulation in the following period. Finally, exogenously determined growth rates are assumed for other factors that affect the growth path of the economy, such as population, labor supply and TFP. Agents are assumed to be myopic and to base their decisions on static expectations. The model considers a large set of policy instruments, some of which have been mentioned previously: production subsidies (by activity), consumption subsidies (by product), VAT (by activity), other indirect taxes (by activity), tariff barriers (by imported product and by origin), non-tariff barriers (by imported product and by origin), direct taxes (by household) and taxes on corporate profits. The model also describes the tariff policy implemented by the EU for Tunisian exports, and tariff quota policies applied by Tunisia and the EU.
6. ACCUMULATION AND FDI Following Dessus and Suwa-Eisenmann (2000), we estimated the impact of capital productivity on capital stock using the following model and data from 1977 to 1999: lnðK tþ1 =K t Þ ¼ C þ ð1
lÞ lnðK t =K t 1 Þ þ mlnRt
1
þ ZlnS t þ U t
where Kt, is the initial capital stock; R, the capital productivity (measured by the marginal productivity of physical capital); S, the domestic saving rate (obtained from the INS); U, endogenous variable, which helps to take into account the adjustment time limits. Using an ordinary least squares (OLS) estimator, we found the following results (in parenthesis are the T-statistics): lnðK tþ1 =K t Þ ¼ 0:14 þ 0:89 lnðK t =K t 1 Þ þ 0:11 ln Rt ð1:82Þ
ð2:60Þ
ð2:55Þ
Adjusted R2 ¼ 0:67; DW ¼
1
þ 0:08 ln S t ð18Þ
1:78
where DW stands for Durbin–Watson. Similar to what Dessus and SuwaEisenmann (2000) obtained, our results show that the observed returns of capital have a significant and positive impact on the growth of physical capital. 6.1. Simulations Results Several assumptions have been made in order to define a plausible development path for Tunisia up to 2010, in the absence of new reforms. The definition of this benchmark is intended merely to provide a baseline to
36 MOHAMED ABDELBASSET CHEMINGUI AND NORA ANN COLTON
which alternative policy scenarios can be compared in order to isolate the impact of a specific policy. The sensitivity analysis conducted by Chemingui and Dessus (1999) suggests that the choice for exogenous variables, within a realistic confidence interval, have no major consequences. Although differing hypotheses concerning growth could have been made, the variations in the different economic aggregates with respect to the baseline scenario after policy shocks seem uninfluenced by these a priori choices. 6.1.1. Growth Hypotheses In order to construct a baseline scenario, the values of a number of variables need to be set. The rate of growth in GDP is set in order to estimate a growth rate for TFP compatible with this development. In constructing the baseline scenario, a figure is defined for the rate of growth in the economy. TFP will then be endogenous. When simulating alternative policies, the previously estimated TFP becomes exogenous and GDP endogenous. We have chosen a figure of 5.7% for the average annual GDP growth rate between 1998 and 2010, in accordance with the forecasts of the ninth social and economic development plan (Ministry of Economic Development, 1998). Over the same period, the rural population is assumed to grow at an average annual rate of 1.0%, while the annual growth rate of the urban population is set at 1.8%. Labor market supply growth will be estimated at 0.9% yearly in rural areas and 2.0% in urban areas. 6.1.2. Economic Policies Implemented in the Baseline Scenario We assume here that the government continues with its policy of fiscal stabilization. Budget spending (excluding investment) will increase in real terms by only 1.5% annually up to 2010. In the baseline scenario, public savings are endogenous. In the alternative scenarios, they are exogenous (and remain at their baseline level), and are obtained by endogenous shifting of the VAT vector. The inclusion of the other changes in economic policy reflects the formal promises of the Tunisian government that were given to the international community. These policies in Tunisia are represented by the signing of two major agreements in 1994 and 1995. First, the multilateral GATT stipulates that consolidated tariffs imposed on imports of agricultural produce and domestic subsidies on agriculture must be phased out over a certain time period. Second, the bilateral partnership agreement signed with the EU provides for the complete abolition of tariff barriers on industrial products by 2010, following a 12-year transition period. Our baseline simulation therefore incorporates these commitments made by Tunisia to its foreign
37
Foreign Direct Investment in Tunisia
partners. Agricultural tariffs with all partners are reduced by 24% over the period from 1995 to 2004 and agricultural subsidies are reduced by 13% between 1995 and 2004. Regarding the EU partnership agreement, tariffs on European industrial products are progressively reduced to zero between 1998 and 2010, and the EU slightly reduces its preferential tariff quotas applicable to Tunisian exports of beverages, citrus fruit, and vegetables between 1997 and 2001. 6.1.3. Major Results Table 2 reports the macroeconomic results of the baseline and an alternative policy scenario. We can observe that the signing of the GATT and partnership agreements with the EU further integrates Tunisia into the system of international division of labor. Exports grow in volume at 8.1% per year, and imports at 6.2%. Due to the preference granted to European industrial Table 2.
Macroeconomics Results. 1992 TND
Annual Growth Rate (%) Baseline scenario (2010) FDI scenario (2010)
Real GDP Output Agriculture and food products Industrial products Services Private consumption Investment Public expenditure Exports To EU To ROW Imports From EU From ROW VAT revenue Tariff revenue Physical capital stock Real rural available income Real urban available income GDP deflator
12.31 27.17 6.35 11.88 8.93 9.82 3.65 2.19 4.23 3.33 0.89 6.10 4.53 1.57 0.90 1.13 24.62 0.776 1.397 1.00
5.75 5.85 4.01 6.22 6.44 5.65 5.11 1.50 8.13 7.13 10.47 6.25 6.94 3.64 5.75 0.68 6.87 4.6 4.1 0.03
6.12 6.20 4.18 6.74 6.65 6.00 5.86 1.50 8.40 7.70 10.49 6.62 7.42 4.10 5.74 0.15 7.40 5.3 4.6 0.05
Note: For the year 1992, these macroeconomic aggregates are expressed in billions of 1992 TND. For the 2010 scenarios, all the variables presented are annual growth rate for the period 1992–2010. TND, Tunisian Dinar.
38 MOHAMED ABDELBASSET CHEMINGUI AND NORA ANN COLTON
products by Tunisia, the ROW’s share of the market is approximately halved for industrial products. However, without further incentives to substitute one source of agricultural imports for another, the share of agricultural imports from the ROW remains stable at around 60%. The gains in competitiveness that allow Tunisia to increase export market share are not due to real depreciation, given that the price of value added remains unchanged and the cut in capital revenue is offset by the rise in real wages. Instead, these gains are due to a combination of factors that benefit the industrial sector, including a reduction in prices for imported input products and a lessening of the distortion of international trade (other than in agriculture). However, while the industrial sector encounters fewer constraints than the agricultural sector (i.e. limits on land suitable for cultivation), it is also more exposed to international competition, forcing it to make greater efforts to adapt. As was, already emphasized above, the effort devoted to factor reallocation and to enhancing competitiveness is concentrated in the industrial sector, which experiences a much more severe external shock than agriculture or food processing. One way to measure this redistribution effort is to examine the change in composition of the production vectors in both factor reallocation and competition enhancement. This calculation indicates a level nine times higher in industry than in agriculture and food processing. An increasingly high percentage of mobile production factors (physical capital and casual labor) are captured by industry, which acquires more substantial commercial outlets, especially abroad. Consequently, labor income rises more swiftly in the non-agricultural than in the agricultural sectors. 6.2. FDI Scenario A second simulation incorporates not only the various trade liberalization commitments made by Tunisia, but also the different modes of financing new investments. This simulation helps to identify the accumulation process, which could be compatible with the capital productivity, observed after the establishment of an FTA with the EU (we used here the capital productivity observed in the baseline scenario to estimate the new physical capital stock). According to our model, there are four possible types of financing available for these new investments: household savings, company savings, government savings and the savings of Tunisia’s two trading partners (the EU and the ROW), in the form of FDI. The decrease in industrial product prices, following the tariff reduction, and the high increase in the household
Foreign Direct Investment in Tunisia
39
debt rate are very likely to reduce household savings. In addition, considering the continuation of a program to stabilize the public budget, it would be difficult for the government to release additional savings. Thus, foreign savings would be the main source of finance for new investments, although slight local participation could be possible. Foreign investment helps to relieve a constraint on Tunisia’s balance of payments by allowing more imports into the country, while keeping the level of exports constant (Dessus & Suwa-Eisenmann, 2000). This will lead to an additional increase of factor productivity on the global level, resulting in an increase of the available capital stock. In this simulation, where new investment is financed only by foreign savings, available income increases by 4.6% for urban households and by 5.3% for rural households. The rate of annual growth of new investment used in this simulation is estimated using the econometric model on the basis of the growth of the TFP observed in the baseline scenario. In the long run, the new equilibrium will correspond to an accumulation rate of 5.8% between 1992 and 2010. The GDP increases at a rate of 6.1% and the investment rate reaches 29% of GDP.
7. CONCLUSION FDI has played an important role throughout Tunisia’s development, as indicated by the high proportion of foreign assets, sales and employment in the economy. The Tunisian case study shows that FDI has also had a significant impact on the quality of Tunisian development by examining the country’s diversification, its technological transfers, its ability to produce skilled workers and managers, its contribution to foreign exchange reserves, and the improvement of productivity, financial structures and management transparency. FDI will continue to be a determining factor in Tunisia’s progress towards becoming an advanced country, although it will encounter many challenges as well as opportunities that will affect FDI attraction. The FTA between Tunisia and the EU may present new opportunities for the Tunisian economy. One important benefit that Tunisia expects from this agreement is the attraction of larger foreign investment inflows, both from the EU and from other regions, which would cause an increase in European financial and technological assistance. In addition, FDI is another frequently mentioned advantage, although the arguments behind these expectations are not well detailed and the trade agreement barely mentions the intention to
40 MOHAMED ABDELBASSET CHEMINGUI AND NORA ANN COLTON
extend liberalization to investment. The main argument underlying this optimism is probably the permanent nature of duty-free access to the EU market. The partnership agreement between Tunisia and the EU has the potential of playing a catalytic role in increasing the openness of the Tunisian economy and attracting FDI. It will encourage investment by causing a decrease in imported inputs price and improve the allocation of resources by increasing the total efficiency of production factors. The empirical analysis shows that the investment potential and growth of FDI inflows to Tunisia that was created by the partnership agreement are conditioned by the existence of important capital productivity. If this is the case, then Tunisia should expect important welfare gains.
ACKNOWLEDGMENTS An earlier version of this chapter was presented at the workshop on FDI in the Arab World in Kuwait in March 2000. Acknowledgments are due to D. Tarr, S. Dessus, B. Ben Redjeb, K. Neymarc, I. Limam and one anonymous referee for help, suggestions and comments.
REFERENCES Al-Khalidi, T. (1998). Arab Economic Integration: Requirements and New Implications under Global changes. Paper presented at the International Conference on New Economic Developments and their Impact on Arab Economies organized by the Arab Planning Institute, 3–5 June, Tunis. Armington, P. (1969). A theory of demand for products distinguished by place of production. IMF Staff Papers, 16(1), 159–178. Beghin, J., Dessus, S., Roland-Holst, D., & van der Mensbrughhe, D. (1996). General modeling of trade and the environment. OECD Development Centre Technical Paper No. 116, Paris. Ben-David, D., & Loewy, M. B. (1995). Free trade and growth. CEPR Discussion Paper No. 1183. Blomstrom, M., & Kokko, A. (1997a). Regional integration and foreign direct investment: A conceptual framework and three cases. Policy Research Working Paper No. 1750. The World Bank, Washington, DC. Blomstrom, M., & Kokko, A. (1997b). How foreign investment affects countries. Policy Research Working Paper No. 1745. The World Bank, Washington, DC. Burniaux, J. M., & Waelbroeck, J. (1995). Le cycle de l’Uruguay et sa proble´matique. Revue Economique, 46(3), 691–700.
Foreign Direct Investment in Tunisia
41
Chemingui, M. A., & Dessus, S. (1999). La libe´ralisation de l’agriculture tunisienne et l’Union Europe´enne: une vue prospective. OECD Development Centre Technical Paper No. 144, Paris. Dessus, S., & Suwa-Eisenmann, A. (2000). Regional integration and internal reforms in the Mediterranean area. OECD Development Centre studies, Paris. Djankov, S., & Hoekman, B. (1997). Trade reorientation and productivity growth in Bulgarian enterprises. Policy Research Working Paper No. 1707. The World Bank, Washington, DC. Hill, H. (1990). Foreign direct investment and East Asian economic development. Asian Pacific Economic Literature, 4, 21–58. Hummel, D. L., & Stern, R. M. (1994). Evolving patterns of North American Merchandise Trade and Foreign Direct Investment, 1960–1990. World Economy, 17(1), 5–29. Kebadjian, G. (1995). Le libre-e´change Euro-Maghre´bin: une e´valuation macro-e´conomique. Revue Tiers-Monde. Kindleberger, C. P. (1966). European integration and the international corporation. Columbia Journal of World Business, 1, 65–73. Lawrence, R. Z. (1996). Preferential trading arrangements: The traditional and the new. Egyptian Center for Economic Studies Working Paper No. 06. Lluch, C. (1973). The extended linear expenditure system. European Economic Review, 4. Mahjoub, A. (1996). La zone de libre-e´change entre la Tunisie et l’Union Europe´enne. Presented at the Seminar organized by the OECD )Les migrations, le libre-e´change et l’inte´gration re´gionale dans le bassin Me´diterrane´en* , Athe`nes, 31 octobre–1 novembre. Ministry of Economic Development (1994). The Tunisian Code for Incentives to Investment. Tunis. Ministry of Economic Development (1998). Neuvie`me Plan de De´veloppement Economique et Sociale, (1997–2001). April, Tunis. Mundell, R. (1957). International trade and factor mobility. American Economic Review, 67, 321–335. Rutherford, T., & Tarr, D. (1997). Regional trading arrangements for Chili: Do the results differ with a dynamic model. Paper presented at the International Conference on CGE Modeling for Economic Development, Paris, 4–5 September.
This page intentionally left blank
42
INSTITUTIONAL INVESTORS, CONTRACTUAL SAVINGS AND CAPITAL MARKET DEVELOPMENT IN EGYPT, JORDAN, MOROCCO, AND TUNISIA$ Wafik Grais and Dimitri Vittas ABSTRACT This chapter looks at the development of contractual savings and institutional investors in Egypt, Jordan, Morocco, and Tunisia (EJMT), and their links with the development of equity markets. The chapter identifies four major potential contributions of contractual savings to capital market development as well as ‘‘impact pre-conditions’’ that can help them obtain. It concludes that contractual savings and institutional investors are neither necessary nor sufficient for the development of equity and bond markets. Nevertheless with certain conditions in place they can have a large impact. The presence of these conditions in EJMT are assessed.
$
The findings, interpretations, and conclusions expressed in this chapter are entirely those of the authors. They do not necessarily represent the views of the World Bank, its Executive Directors, or the countries they represent.
Money and Finance in the Middle East: Missed Opportunities or Future Prospects? Research in Middle East Economics, Volume 6, 43–65 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1094-5334/doi:10.1016/S1094-5334(05)06003-6
43
44
WAFIK GRAIS AND DIMITRI VITTAS
1. INTRODUCTION Egypt, Jordan, Morocco, and Tunisia (EJMT) have been reasonably successful in mobilizing long-term savings in comparison to most countries in Latin America and Eastern Europe. They have also been quite successful in mobilizing short-term deposits. The latter is primarily linked to the maintenance of low inflation and reasonable macrostability, while the former is associated with the operation of partially funded pension systems. However, in the area of banking, where lending portfolios suffer from large volumes of nonperforming loans, EJMT appear to have been less successful in utilizing the mobilized contractual savings. This is reflected in an underdevelopment of equity markets and in low investment returns by their institutional investors. It is also reflected in underdevelopment of the insurance industry. This chapter looks at the development of contractual savings and institutional investors in EJMT and their links with the development of equity markets. Their experience is contrasted with that of countries in Latin America, Eastern Europe, East Asia, and Southern Europe. The experience of the most successful countries in promoting contractual savings and institutional investors is also reviewed and provides the context of global developments in this area. The structure of this chapter is as follows. Section 2 reviews the explosive growth of equity markets and contractual savings around the world. Section 3 summarizes recent research findings in relation to the impact of equity market development on growth and the links between contractual savings and capital market development. Section 4 discusses the various economic and capital market effects of contractual savings and institutional investors. Section 5 offers a more detailed comparative analysis of the experience of EJMT, while the last section concludes with some suggestions about desirable policy reforms.
2. EXPLOSIVE GROWTH The last quarter of the twentieth century witnessed a parallel explosion of equity markets and institutional investors. The latter was associated with a huge increase in contractual savings especially pension funds, although in the last decade or so mutual funds experienced very rapid growth in many countries. Total equity market capitalization as measured by the International Finance Corporation (IFC) Emerging Markets Database soared from US$2.7
Institutional Investors, Contractual Savings and Capital Market Development
45
trillion in 1980 to US$9.4 trillion in 1990 and US$23.5 trillion in 1997 (IFC, 1987; 1998). At the same time, the annual volume of trading increased even faster from US$0.8 trillion in 1980 to US$5.5 trillion in 1990 and US$19.5 trillion in 1997. The turnover ratio, the ratio of trading to market capitalization, jumped from 30% in 1980 to 59% in 1990 and 83% in 1997. The average annual rates of growth over the whole 17-year period amounted to 14% for market capitalization and 21% for the volume of trading. The persistent growth of equity markets was paralleled by an equally persistent growth of institutional investors, including: pension funds, insurance companies, and mutual funds. These investors collectively mobilized huge financial resources in a growing number of countries, and surpassed, in terms of financial power, the traditional importance of commercial and investment banks. In 1980, there were probably no more than 10 countries in the world where institutional investors controlled resources exceeding 50% of gross domestic product (GDP). The countries included the United States, the United Kingdom, Switzerland, the Netherlands, Singapore, South Africa, and, perhaps, a few more Anglo-American and Scandinavian countries. Seventeen years later, nearly all Organization for Economic Co-operation and Development (OECD) countries had institutional investors with resources exceeding 50% of GDP. Among developing countries, Chile and Malaysia joined this group. In Chile, the total assets of institutional investors grew from less than 1% of GDP in 1980 to over 60% in 1997. But of greater significance, is the fact that in countries like the United Kingdom, the Netherlands, Switzerland, the United States, and South Africa the total assets of institutional investors now exceed 150% of GDP. The size of institutional investor assets relative to national income fluctuates from year to year depending on the phase of the economic cycle, the performance of the equity and bond markets, and the asset allocation policies of institutional investors. As stock markets tend to be at or near their peak during periods of economic recession, the ratio of institutional investor assets to GDP oscillates in a counter-cyclical way. However, these cyclical fluctuations do not obscure the long-term trend and the persistent increase in the relative importance of contractual savings and institutional investors.
3. RECENT RESEARCH FINDINGS Interest in the growth of institutional investors, contractual savings, and capital markets is motivated by the potential contribution that they can
46
WAFIK GRAIS AND DIMITRI VITTAS
make to economic growth and development and the dynamic interaction that unfolds over time between markets and institutional investors. For much of the 1970s and 1980s, national governments and multilateral institutions, including the World Bank Group, promoted the establishment and development of securities markets around the world. Policies fostering the development of securities markets have been advocated on the premise that securities markets create a more balanced financial system, encourage greater competition and efficiency, stimulate innovation, and promote higher transparency and accountability. In the beginning, the emphasis was placed on creating markets and encouraging companies to list themselves, often by providing fiscal and other incentives. Of course, it was realized that merely having companies listed in a market that lacked liquidity, transparency, and integrity would not achieve the desired economic and financial benefits. However, the creation of markets was the first step in the long process of capital market and financial sector development. The establishment of stock exchanges in over 100 countries around the world testifies to the impact of this policy. More recently, the emphasis has shifted to the challenge of increasing the liquidity of markets, enhancing their transparency, and ensuring their integrity. It is, in fact, increasingly underscored that what matters for achieving the economic benefits of securities markets is not the number of listed companies and total market capitalization, but rather the level of market liquidity in an environment of transparency and integrity. Research undertaken in recent years at the World Bank by Ross Levine and others underscores the existence of a clear and positive relationship between market liquidity and economic growth. Countries with liquid equity markets, measured by the annual volume of trading in relation to GDP, grow at a significantly higher rate than countries with illiquid markets (Levine & Zervos, 1998). The same research also finds that countries where both stock markets and banks are well developed have a higher growth rate than those where only banks or only stock markets are well developed. In contrast, countries where neither markets nor banks are developed achieve the lowest rates of growth.1 A third research finding in recent years concerns the complementary nature of banks and equity markets in developing countries. Research using firm level data for many countries finds that companies operate with high leverage, in other words a high level of debt over equity, in countries where equity markets are less developed. As equity markets are established and grow, leverage increases, companies raise equity but they also increase their bank loans by a larger amount in order to finance their plans for expansion.
Institutional Investors, Contractual Savings and Capital Market Development
47
It is only in high-income countries, where securities markets are highly developed, that firm leverage declines. In these countries, companies substitute equity for debt (Demirguc-Kunt & Maksimovic, 1998). The recent findings from economic and financial research provide strong empirical support for policies that pursue the development of securities markets. They not only show that markets and banks are complementary to each other, but also that market liquidity is of great importance and that it is the support given to private sector investment that promotes economic growth and efficiency. The link between the development of securities markets and contractual savings and institutional investors stems from the positive influence that institutional investors can exert to enhance the liquidity, transparency, and integrity of markets. In fact, more recent World Bank research points to a strong link between contractual savings and equity market development.2 Data from OECD countries suggest a strong and significant correlation between trading volumes and contractual saving assets and a strong link between market capitalization and contractual savings. Rather interestingly, there is a positive correlation between equity markets and contractual savings even in countries where pension funds and insurance companies follow conservative investment policies and neither invest nor trade heavily in equities. This suggests that contractual savings may have a direct as well as an indirect impact on capital market development. The latter probably stems from the greater attractiveness of capital markets for international investors in countries where contractual savings are large, prospective deficits of public pension schemes are small, and governments are able to finance their long-term borrowing needs by issuing bonds without recourse to inflationary short-term borrowing from commercial banks. The last piece of research that yields interesting results concerns the behavior of international investors and its impact on market volatility. Past research tended to show that international investors, especially international mutual funds, do not exacerbate financial crises that from time to time affect stock markets in developing countries. However, a recent study of the behavior of international mutual funds in Latin America found that, after the Russian market collapse of August 1998, both retail investors in these mutual funds and their managers unloaded their holdings in an indiscriminate and contagious fashion (Kaminsky, Lyons, & Schmukler, 1999). The findings of this study highlight the importance of large domestic institutional investors for greater market stability and lower price volatility in emerging markets. The greater insulation of the Chilean and South African equity markets from the adverse impact of the international financial crisis of 1998,
48
WAFIK GRAIS AND DIMITRI VITTAS
which is attributed, at least in part, to the strong presence of domestic institutional investors, provides additional support for this thesis.
4. ECONOMIC AND CAPITAL MARKETS EFFECTS Institutional investors and contractual savings have four main potential economic effects. Firstly, the funded pension schemes are usually associated with large contractual savings, and their existence implies smaller promised benefits from unfunded public pension pillars. As a result, public pension pillars entail both lower deficits and lower payroll taxes, both of which have positive implications for economic growth. In fact, they stabilize expectations and improve the investment climate. Secondly, they imply lower payroll taxes, and funded pension schemes that tend to cause fewer distortions in the functioning of labor markets. They remove some of the incentives for participation in the informal labor market, which tends to operate with lower capital and thus suffers from lower productivity. To the extent that they encourage more workers to participate in the formal labor market, contractual savings promote higher levels of productivity and higher growth.3 Thirdly, contractual savings may be associated with a higher level of national saving, although the evidence on this is unclear. Arguments can be made that a higher level of efficient contractual saving may, in fact, lead to a fall in the rate of saving. This would happen, for instance, if the income effect of higher investment returns were stronger than the substitution effect. Evidence from OECD countries shows that countries with large contractual saving sectors have lower national and personal saving rates than countries with small contractual savings sectors. The fourth potential effect is the most certain and in many respects the most important. Contractual savings cause a significant qualitative change in the composition of savings in favor of long-term financial savings. In doing so, they result in a flatter yield curve and a lower cost of capital for long-term investment projects and housing. They may also cause a reduction in the risk premium facing the country in the international financial markets, therefore lowering the cost of finance for governments as well as the corporate and household sectors. But in addition to these cost effects, contractual savings, and institutional investors more generally, have a major potential effect on the functioning of financial markets, which results in a lower cost of capital, greater availability of instruments, and higher economic efficiency.
Institutional Investors, Contractual Savings and Capital Market Development
49
Contractual savings and institutional investors entail substantial long-term benefits for the development of financial markets. Due to their long-term orientation, their special financial needs and because professional investors manage them, they stimulate financial innovation. This acts as a countervailing force to existing commercial and investment banks that exerts pressure for greater market integrity and modernized trading facilities. This strengthens corporate governance, and encourages more robust financial regulation with positive demonstration effects for other parts of the financial sector, such as banking and insurance.4 The benefits of institutional investors for the development of capital markets are not, however, automatic but depend on a number of ‘‘impact’’ pre-conditions. These include the attainment of critical mass, the adoption of conducive regulations, the pursuit of optimizing policies, and the prevalence of pluralistic structures. When they first emerge, institutional investors control few resources and carry little weight on the evolution of market practices. It is only when they reach critical mass that they can influence market development. Conducive regulations are also important since in countries where pension funds and insurance companies are required to invest in nontraded government securities, the direct impact on capital market development is negligible. But even quantitative restrictions expressed as upper limits on holdings of various asset classes are inhibiting, especially if they are binding. However, in many countries, quantitative controls are nonbinding. This is because pension funds and insurance companies do not pursue optimizing investment policies but rather conservative ones, investing predominantly in government bonds. This practice has long prevailed in continental Europe and has inhibited the potential beneficial impact that institutional investors can have on capital market development. Finally, the existence of a pluralistic structure is important because it stimulates competition and innovation and allows changes in policies and approach to be effected more smoothly than in the case of nonpluralistic structures. In many countries, where national provident funds or social security corporations have dominated the institutional investor sector, the impact on capital market development has been constrained by the inability of such dominant institutions to change their policies without exposure to major adverse price effects and without causing market instability. The experience of many countries around the world suggests that absent these important ‘‘impact’’ pre-conditions, institutional investors will not make a major direct contribution to capital market development. Moreover, the growth of capital markets in East Asia and Southern Europe, in the
50
WAFIK GRAIS AND DIMITRI VITTAS
absence of strong domestic institutional investors, suggests that other factors, such as new technology, globalization, and regional integration, may also play an important part in capital market development. Thus, pension funds and institutional investors are neither necessary nor sufficient for the development of equity and bond markets. Nevertheless, if they are subject to conducive regulation, adopt optimizing policies, operate in a pluralistic structure, and when they reach critical mass, institutional investors can have a large impact.
5. COMPARATIVE EXPERIENCE OF EJMT 5.1. Macroeconomic Backdrop EJMT have used the 1990s to strive for macroeconomic stability and begin positioning their economies on a path to higher and more sustainable growth. All four economies brought inflation down, contained their budget and current account deficits, as well as reduced their foreign debt burdens and increased their foreign exchange reserves. This occurring at the same time social indicators improved. The emphasis on macroeconomic and social stability was associated with modest rates of growth for countries where development needs remain considerable and population growth high though declining (see Table 1). This strategy may have been related to one or several of the following factors: (a) the observation of the wrenching experiences of Central and Eastern Europe as well as the Former Soviet Union (FSU), (b) the acknowledgment of the poor institutional infrastructure to cope with more rapid liberalization, or (c) the apprehension of drifting into unmanageable and costly social and political upheavals. The persisting constrained role of the private sector in the economies of EJMT, their continued weak attractiveness for foreign direct investment (FDI), and slow and hesitant integration into the world economy, reflect the development strategy that is being pursued. The strategy also shows up in the ability of these economies to generate significant bank deposits, and respectable contractual savings even though they do not reap the full benefit due to poor investment allocation. EJMT are now at the juncture where they need to capitalize on their achievement in the 1990s in order to see their economies take off to higher levels of sustainable growth where contractual savings will play a critical role. The synergy of macroeconomic and social stability with the development of contractual savings and capital markets is a fundamental pillar of
Selected Macroeconomic and Social Indicators. Egypt
1995/1996
CPI inflation (%) Budget deficit (% of GDP) Current account balance (% of GDP) ForEx reserve in months of imports Total extended debit (billion US$) Extended debit service in % of experience of GNFS GDP at market prices growth (%) Population growth (%)
Jordan
1998/1999
1995
1999
Average 1994/ 1995
1999
1995
1998
0.7 4.8 0.8
4.7 2.5 4.1
3.1 2.8 3
2.6 9.4 20.9
2.7 11.3 19.2
3.8 4.2 1.9
2.8 7.8 8.9
0.6 7.3 0
15.7 31 12.9
12.8 28.2 9.8
1.4 6.95 26.4
7.1 7.9 23.4
4 22.4 31.1
5.8 17.5 25.1
5 2.1
6 1.9
3.9
0.6
1.9 1.6
0.7
120
49 47 67.23
GDI (million US$) FDI in FDI to LMI (%) Private investment % of GDFI
734 3.04 62.3
1076 0.63 68.6
21.98 84.66
19.54 77.38
Source: World Bank Group – World Development Indicators. CPI: consumer price index; LMI: linear matrix inequality.
41
38 0.16 44.6 113.6
27 11.5 6.48
5.6 4.8 3.15
Tunisia
7 1.3 0.3
Infant mortality rate per 1000 Adult illiteracy (%) Population 1999
M1/GDP M2/GDP
Morocco
99
49 52 28.2
2.23 1.8 69 54
5 1.7 28 31.5 9.46
310 0.18
165 0.68 68.3
322 0.19 67.9
76 0.31 50.5
650 0.38 50.8
31.1 104.8
46.3 63.5
52.4 71.4
20.45 46.2
23.22 51.77
Institutional Investors, Contractual Savings and Capital Market Development
Table 1.
51
52
WAFIK GRAIS AND DIMITRI VITTAS
sustainable growth. On one hand, the existence of secure and credible pension regimes, insurance markets, and financial access to housing ownership can contribute significantly to the stabilization of consumers and investors expectations. They allow for the expansion of activity, growth, and development. On the other hand, macroeconomic and social stability, as well as a level playing field at the micro-level, will allow for higher and safer rates of return for institutional investors and attract the interest of FDI. EJMT are on their way to achieving this type of capitalization in spite of the problems of being emerging markets. However, a more dynamic approach better coordinated within each country is now needed. Significant privatization has taken place in all four countries bringing with it increased FDI. Jordan Telecom (JTC), second and beyond global system for mobile communication (GSM) licenses in Egypt and Morocco as well as their advanced preparation or launch of the privatization of the main telecom operator, cement privatization in Tunisia attest to the fact that these economies have moved from plans to deeds. In addition, Egypt, Morocco, and Tunisia are taking steps to revamp the banking sectors by increasing the independence of their central banks, reforming banking laws, testing bank privatization with smaller banks, and restructuring or merging others.
5.2. Contractual Saving and Institutional Investors EJMT lag considerably behind the more advanced OECD countries in the development of contractual savings and institutional investors, as well as in capital markets, however their records vis-a`-vis developing and transitioning countries have been somewhat mixed. Egypt, and to a lesser extent Jordan and Morocco, have been able to mobilize a large volume of contractual savings. The total assets of institutional investors in Egypt amount to nearly 40% of GDP. This puts Egypt ahead of most Latin American, Eastern European, and Asian countries, and on par with the countries of Southern Europe. Jordan and Morocco have total institutional investor assets in the region of 20% of GDP, while Tunisia lags further behind at less than 15% (Table 2). A feature of contractual savings and institutional investor assets in EJMT is their concentration in centrally managed public sector institutions. Except for Morocco, the presence of private pension funds and insurance companies is weak. This has resulted in poor asset allocation policies and low investment returns. Coupled with a less advanced integration of international capital and financial markets, this has resulted in a low
53
Institutional Investors, Contractual Savings and Capital Market Development
development of capital markets, especially in terms of liquidity and trading volumes. Countries vary considerably in the relative importance of their contractual savings and institutional investors. In the vast majority of developing countries in Africa, Asia, and Latin America, as well as the transitioning countries of Eastern Europe, the total assets of pension funds, insurance Table 2.
Pension Institutions, Contractual Savings and Institutional Investors (% of GDP). Pension Institutions
Contractual Savings
Institutional Investors
1993
1997
1993
1997
1993
1997
Egypt Jordan Morocco Tunisia
28.7 15.5 8.8 10.2
34.0 19.8 11.7 9.6
32.2 18.5 19.2 13.8
37.7 23.0 23.3 13.2
32.2 18.5 19.2 14.5
37.7 23.0 26.8 16.6
Argentina Brazil Chile
0.0 NA 38.1
3.0 9.0 45.0
2.0 NA 48.2
5.0 11.6 58.7
2.0 NA 51.1
6.8 23.7 62.7
0.0 0.0 0.0
1.3 0.7 0.0
6.0 3.2 0.6
8.7 4.9 0.9
9.1 3.8 1.2
11.7 6.7 1.3
Japan Malaysia Singapore Sri Lanka
31.9 50.7 56.2 15.0
31.0 55.1 55.0 18.5
73.3 61.0 67.8 16.1
74.1 67.4 71.1 21.0
84.2 83.0 NA 16.1
83.0 87.4 NA 21.3
France Germany Italy Portugal Spain
0.0 2.8 4.1 5.7 2.9
0.0 3.1 5.2 10.8 2.1
35.9 29.1 18.4 12.3 21.3
55.9 36.9 20.4 21.6 24.5
78.3 42.5 25.5 25.5 38.7
94.5 61.8 39.2 32.8 59.4
93.3 40.5 78.0 76.0 52.3
117.6 44.0 100.9 81.7 68.7
132.4 112.8 141.9 152.7 90.2
186.2 128.7 181.7 178.1 111.0
163.2 115.5 158.5 173.3 121.1
206.1 130.7 205.2 203.0 161.4
Czech Republic Hungary Poland
The Netherlands South Africa Switzerland United Kingdom United States
Note: Pension institutions: pension funds and social security institutions. Contractual savings: pension institutions and insurance companies. Institutional investors: contractual savings and mutual funds. Source: Own calculations based on national data.
54
WAFIK GRAIS AND DIMITRI VITTAS
companies, and mutual funds correspond to less than 25% of GDP, in many cases less than 10% of GDP. In most continental European OECD countries and a few countries in Asia and Latin America (such as Chile, Japan, Malaysia, and Singapore) these assets range between 25% and 100%, while in a small number of countries, that include the United Kingdom, the Netherlands, Switzerland, the United States, and South Africa, they exceed 100% of GDP, in some cases by a very wide margin. 5.2.1. Pension Institutions In addition to differences in the total assets of institutional investors, there are also considerable differences across countries in the composition of these assets. In most countries, institutional investors are dominated by pension institutions, either private pension funds or in a few cases national provident funds and partially funded social security institutions. However, there are several Western European countries, such as France, Germany, and Italy, where pension institutions are either totally absent or too weak. In these countries the main types of institutional investors are insurance companies and mutual funds. In the case of France, the dominant mutual funds are money market mutual funds, which are short-term in nature and make little contribution to the development of capital and bond markets. EJMT base their social pension systems on the so-called scaled-premium approach.5 This approach generates long-term savings that rise with the maturity of the system so long as the funding ratio is kept at a reasonable high level. However, most countries that start with a scaled-premium approach let their social pension systems unfold over time into unfunded schemes. This is caused by three factors. Firstly, benefits are increased without commensurate increases in contributions. Secondly, political opposition delays the increase in contribution rates that is needed from time to time to ensure the financial balance of the system. And thirdly, the investment performance of the social security institutions suffers from their use as captive sources for financing government spending. In Jordan, the majority of accumulated contractual savings are represented by the financial resources of the Social Security Corporation. The Corporation is financed by a 15% contribution rate. It has benefited from both a low-dependency ratio (a small number of beneficiaries compared to the number of contributors) and from a positive, though moderate, real rate of return. Egypt has applied a much higher contribution rate of 26% in its social insurance system, which explains the much higher volume of mobilization of contractual savings. Annual net flows into the system are very large as both
Institutional Investors, Contractual Savings and Capital Market Development
Table 3.
Current and Projected Demographic. 2000
Egypt Jordan Morocco Tunisia
55
2040
Y
A
O
DDR
Y
A
O
DDR
45.6 50.9 43.3 41.8
47.6 44.4 50.1 49.9
6.8 4.7 6.6 8.3
14.3 10.6 13.3 16.6
28.6 29.9 28.2 27.2
55.2 55.4 53.5 53.2
16.2 14.7 18.2 19.6
29.4 26.4 34.1 36.8
Source: World Bank. Y, young age; A, active age; O, old age; DDR, demographic dependency ratio.
the system and the covered population are young. However, the benefits of the large annual net inflows have been offset by high negative real returns earned in the late 1980s and early 1990s.6 Morocco and Tunisia have social pension systems with much lower contribution rates and more mature structures. These systems already face financial pressures and have limited accumulated resources. However, both countries operate other social funds that have surplus resources, in particular the funds for family allowances. All four countries have young populations which may explain the apparent lack of concern about the long-term sustainability of the system (Table 3). In fact, the social security systems may benefit in the medium term from expansion of coverage, which at present is low and tends to cover the formal labor force. However, to play their full part in providing secure longterm benefits and promoting economic growth, all systems would require some fundamental reforms. Although EJMT presently mobilize more long-term contractual savings than the countries of Latin America (except Chile and Brazil) and Eastern Europe, they have one important disadvantage compared to some of these countries. Fundamental pension reforms and greater integration with world markets, created through a greater openness of the domestic markets to foreign strategic investors, promise greater efficiency and faster economic growth for such reforming countries as Argentina, Brazil, Chile, and Mexico in Latin America or the Czech Republic, Hungary, and Poland in Eastern Europe. Thus, although their starting point is more advanced, continuation of present policies are likely to cause EJMT to lose considerable ground.
56
WAFIK GRAIS AND DIMITRI VITTAS
5.2.2. Insurance The previous section is more relevant when attention is focused on recent developments in the insurance sector. In most countries around the world, including East Asia, Latin America, and Eastern Europe, the insurance industry has experienced considerable growth in recent years, but it has stagnated in EJMT. In large part, this may be attributed to the underdevelopment of life insurance. Whereas in many countries around the world the fraction of life premiums out of total premiums has been growing fast, reaching or exceeding 60%, it has hovered around 20% or less in EJMT (Table 4). The insurance sector of EJMT, in line with most developing countries, is at the aggregate level characterized by inadequate capital, low investment returns, high operating costs, lax control over brokers, high levels of receivables, fraud and unduly high claims by some insured but otherwise low claims and settlements for the majority of customers, and protracted disputes and long delays in settlement. These problems reinforce the mutual mistrust that exists between insurance companies and consumers and they are caused by information problems that afflict the insurance industry. In Jordan, Morocco, and Tunisia, there are some private companies that operate with reasonable efficiency and acceptable solvency margins. However, in Jordan, the sector is highly fragmented into many small and inefficient companies. In Tunisia, there are some state-owned or statecontrolled companies that operate with large negative amounts of capital. In Morocco, the insurance market is affected by the past losses and outstanding claims of several companies that failed in the 1980s, but are currently under liquidation. All four countries experience long delays in settlement of claims that force them to maintain very high levels of reserves that in turn increase their operating costs. In Egypt and Morocco, insurance reserves are unusually high. The main type of insurance is mandatory motor insurance, which is subject to low regulated premiums. Motor insurance produces large technical losses, which force insurance companies to scrutinize claims and delay their settlement. In many cases, large court awards have encouraged claimants to go to court rather than accept reasonable settlements by insurance companies. Over time, a climate of mistrust has evolved between insurance companies and their clients. This has had adverse effects not only on motor insurance, but also on the development of other types of personal insurance, including household and life insurance. As in most developing countries, large industrial and commercial risks are reinsured by international reinsurance companies. This has protected the
57
Institutional Investors, Contractual Savings and Capital Market Development
Table 4.
Insurance Premiums and Assets.
Insurance Premiums (% of GDP)
Insurance Assets (% of GDP)
Share of Life Premiums (% of Total Insurance Premiums)
1993
1997
1993
1997
1993
1997
Egypt Jordan Morocco Tunisia
0.7 1.8 2.4 1.8
0.7 2.0 2.6 1.6
3.5 3.0 10.4 3.6
3.7 3.2 11.6 3.5
22.7 NA 21.2 6.3
22.4 NA 23.4 7.6
Argentina Brazil Chile
1.5 1.2 3.3
1.8 2.1 3.5
2.0 NA 10.1
2.0 2.6 13.7
16.9 10.2 57.9
31.6 18.4 65.9
Czech Republic Hungary Poland
2.5 2.2 2.0
2.9 2.3 2.7
6.0 3.2 NA
7.4 4.2 0.8
25.6 23.8 40.1
26.6 33.0 28.1
Japan Malaysia Singapore Sri Lanka
12.6 3.7 3.7 1.0
11.9 4.4 5.1 1.2
40.4 10.3 11.6 1.1
43.1 12.3 16.1 2.5
78.6 48.4 62.8 25.0
79.4 49.9 74.5 32.0
France Germany Italy Portugal Spain
7.6 6.7 3.2 4.4 4.4
9.3 6.5 4.2 5.4 5.4
35.9 26.3 14.3 6.6 18.4
55.9 33.8 15.2 10.8 22.4
60.3 39.5 30.2 31.6 32.9
68.5 41.7 45.3 47.0 47.0
8.6 12.2 9.3 11.7 8.9
9.4 17.3 11.9 11.2 8.5
49.1 72.3 63.9 76.7 37.9
68.6 78.7 80.8 96.4 42.3
51.5 79.9 59.6 64.3 41.7
56.2 79.9 69.8 70.1 45.4
The Netherlands South Africa Switzerland United Kingdom United States
Source: Own calculations based on data from Sigma and national sources.
domestic market from excessive retention of risk, but it has also caused the domestic market to be unable to develop the capacity to price and retain risk locally. EJMT have avoided the large losses suffered in Argentina and Brazil, where a high local retention ratio along with a state-owned monopoly reinsurer has been established. In contrast to Latin America and Eastern Europe, and more recently East Asia, the insurance industry in EJMT
58
WAFIK GRAIS AND DIMITRI VITTAS
continues to be relatively closed to foreign companies. Resistance to greater opening comes from both state-owned and private companies. The development of life insurance has been affected by the existence of a social security system that promises relatively high pensions to covered workers. Due to young demographic structure in EJMT, these promises are still credible. Cultural factors and religious interpretations may also explain the underdevelopment of life insurance. Life insurance corresponds to 2.30% of GDP in Malaysia and accounts for nearly 50% of insurance premiums. In Egypt and Tunisia, life insurance corresponds to 0.16% and 0.12% of GDP, respectively. The repressive regulatory environment, lack of competition and product innovation, and the prevailing mistrust of insurance companies have been of greater significance. In Malaysia, life insurance receives a boost from the success of the national provident fund since retiring workers may purchase annuity products. The rapid growth of life insurance in Chile has been attributed to the pension reform program that was implemented in the early 1980s. The Chilean life insurance sector also benefited from the compulsory purchase of term life and disability insurance that protects workers and their families from the financial consequences of serious accidents and death before reaching retirement age. A similar growth of life business is anticipated in Argentina. The need to provide for old age is also a factor in the large increase in the relative importance of life insurance in France, Italy, Portugal, and Spain. In these countries the aging of the population and the financial pressures facing the social security systems has undermined the credibility of overgenerous social security pensions. 5.2.3. Mutual Funds Mutual funds have expanded very fast in the 1990s in most advanced OECD countries as well as in some developing countries. In the United States, the assets of mutual funds exceeded 50% of GDP in 1997 when they overtook the total assets of commercial banks. In France, mutual funds have assets equivalent to 39% of GDP, while in most other advanced OECD countries they range between 20% and 30%. There are significant differences in the composition of mutual funds across countries. Equity funds are predominant in Anglo-American countries, while bond funds are more prominent in continental European and most developing countries. Money market mutual funds have prospered in countries that have imposed restrictions on the remuneration of current accounts and short-term deposits. They are relatively large in the United States, France, Australia, Spain, Greece, and Morocco. Mutual funds have
Institutional Investors, Contractual Savings and Capital Market Development
59
been authorized in Tunisia, Morocco, and Egypt. They have grown particularly fast in Morocco where they reached 4% of GDP in 1997 but may now stand at close to, or higher than, 10% of GDP. Unlike pension funds and life insurance companies, which have long-term liabilities, mutual funds have a short-term orientation because they allow their customers to redeem their shares on demand. However, mutual funds invest predominantly in marketable securities and may collectively exert a strong impact on capital market development. 5.2.4. Equity Markets As already noted, equity markets experienced an explosive growth over the past 20 years. This growth has been more prevalent and persistent in AngloAmerican countries, but more recent in European countries. In contrast, equity markets in Asia, Latin America, Eastern Europe, Africa, and the Arab world enjoyed periods of high growth interspersed with significant setbacks. Some countries experienced large growth in market capitalization and value traded even between 1993 and 1997 (Table 5). Contractual savings institutions can play a very important role in the development of a country’s capital market. This role depends on the allocation of their assets, which varies considerably from country to country, reflecting both historical traditions and differences in regulation. Real assets and equities are heavily represented in the portfolios of the U.K. pension funds. This is generally attributed to the equity cult that U.K. fund managers have developed since the mid-1960s in response to the high rates of inflation experienced by the U.K. economy between 1960 and 1980. But this pattern is also explained by the absence of legally imposed minimum funding requirements and by the use of pension obligations that are quasi-indexed to inflation. A preference for equities is also prevalent in South Africa. Pension funds and life insurance companies in South Africa offer market linked products, the value of which fluctuates with market prices. Real assets represent a smaller proportion of pension fund assets in the United States and other Anglo-American countries. In continental Europe, pension funds (as well as life insurance companies) have traditionally placed the largest part of their assets in government, corporate and mortgage bonds, and in long-term loans. This pattern is being repeated by pension funds and insurance companies in Eastern Europe. This is partly the result of investment regulations and partially the result of a traditional emphasis on conservative investment policies. Although pension funds and insurance companies are subject to upper limits on their holdings of equities (as well as overseas securities), and although their managers are
60
WAFIK GRAIS AND DIMITRI VITTAS
Table 5.
Equity Markets (% of GDP).
Market Capitalization
Value Traded
Turnover Ratio
1993
1997
1993
1997
1993
1997
Egypt Jordan Morocco Tunisia
7.5 75.8 8.6 6.0
23.1 71.4 31.0 18.1
0.4 25.1 1.9 0.3
7.7 7.1 3.1 1.5
5.3 33.1 22.1 5.0
33.3 9.9 10.0 8.3
Argentina Brazil Chile
13.2 40.2 85.0
17.8 29.6 93.8
4.3 13.1 6.3
8.8 25.4 9.8
32.6 32.6 7.4
49.4 85.8 10.4
Czech Republic Hungary Poland
NA 1.8 2.0
27.8 22.8 7.1
NA 0.3 2.6
13.3 17.0 5.6
NA 16.7 130.0
47.8 74.6 78.9
Japan Malaysia Singapore Sri Lanka
63.1 253.9 158.0 19.0
63.2 202.2 133.3 13.2
22.3 239.4 142.0 3.7
29.8 146.8 66.5 2.1
35.3 94.3 89.9 19.5
47.2 72.6 49.9 15.9
France Germany Italy Portugal Spain
32.3 21.2 13.5 13.1 22.8
44.9 35.3 26.3 30.7 50.1
13.9 15.9 6.7 5.8 9.9
28.8 48.6 17.3 20.5 85.2
43.0 75.0 49.6 44.3 43.4
64.1 137.7 65.8 66.8 170.1
50.8 105.7 98.8 109.2 72.4
117.5 165.2 190.7 141.8 119.2
21.6 10.0 71.0 44.5 50.5
79.0 30.5 193.1 62.9 123.1
42.5 9.5 71.9 40.8 69.8
67.2 18.5 101.3 44.4 103.3
The Netherlands South Africa Switzerland United Kingdom United States
Note: Market capitalization is yearly average. Turnover ratio is value traded divided by market capitalization. Source: IFC Emerging Markets Database.
seeking either increases in these limits or their complete abolition, restricted investments are well below the specified limits. However, a gradual shift of asset allocation of Dutch and Swiss pension funds in favor of equity investments has taken place in recent years. The pattern in other countries is similar to that of continental European countries. Investment rules have generally favored bonds. In Singapore, the Central Provident Fund invests over 90% of its funds in specially issued, nonmarketable and floating rate government securities that earn a
Institutional Investors, Contractual Savings and Capital Market Development
61
modest positive real rate of return. In Malaysia, the funds of the national provident fund are placed in government bonds that offer a reasonably high real rate of return. Investments in equities have increased in recent years. In both countries, individual workers are allowed to withdraw funds for investment in housing and an increasingly broader range of approved securities.7 In Chile, investments in corporate equities for privately managed pension funds were less than 20% of total assets in the late 1980s, mainly because of the imposition of tight restrictions on investment portfolios. The pension funds invested heavily in government, mortgage and corporate bonds as well as bank deposits. The gradual relaxation of investment rules has allowed the Chilean pension funds to invest more in equities. The limit on equities was raised to 37% of assets in the 1990s and the restriction on investment in foreign securities was lifted. Currently, Chilean pension funds invest less than 20% in domestic equities and as much as 15% in foreign securities. In EJMT, investment allocations have been subject to direct government influence, with the notable exception of Jordan, where any such influence has been indirect. In Egypt, most of the vast resources of the social security system are handed over to the National Investment Bank, which invests in public projects. This is similar to what happens in Malaysia, but with the important difference that in Malaysia the rate of return credited to individual accounts has been positive in real terms. In Tunisia and Morocco, contractual savings institutions have been required to invest in low-yielding government paper, in low-interest housing loans, and even in building low-rent housing units. This has depressed investment income, although in recent years investment rules have been substantially relaxed and new investment flows are now placed in assets yielding market rates of return. In Jordan, the Social Security Corporation has enjoyed greater freedom from government interference, but investments have been constrained by the highly conservative policies of the Corporation and by the shortage of attractive investment opportunities. Increasingly, the Corporation has been forced to place a growing percentage of its new flows into bank deposits. A new policy may be adopted in the future that will emphasize a more professional approach in asset allocation and may encourage the contracting out of fund management to professional asset managers. In most countries around the world, investments in foreign assets have been constrained by regulations in the form of either foreign exchange controls or unnecessarily tight prudential controls. Following the removal of
62
WAFIK GRAIS AND DIMITRI VITTAS
exchange controls and the relaxation of investment rules, pension funds in several countries have built up substantial holdings of foreign equities and bonds. These holdings are well over 20% in Australia, New Zealand, and the United Kingdom and around 60% in Hong Kong. International diversification may increase portfolio returns, especially if pension funds become too big for the local markets, but a more general result of diversification is a reduction in investment risk, stemming from the less than perfect covariance of returns in different national markets. Contractual savings institutions in most countries tend to display a strong ‘‘home bias’’, which may be attributed to the nature of their liabilities, the absence of efficient hedging facilities, and their preference for investing in markets and securities with which they are more familiar. Nevertheless, international diversification has been growing fast. Most countries, especially those with undiversified economies dominated by a few industries and a few family groups, should allow some foreign investments once a new pension system is well established. Another major influence in equity market development is the trend of globalization and integration of local markets with world markets. Among the countries reviewed in this chapter, Switzerland, Singapore, and Malaysia stand out for the high level of market capitalization and value traded as well as for the high level of development of institutional investors. However, in none of these countries do pension funds and insurance companies invest heavily in equities or play a dominant part in equity trading. Their impact is indirect and relates to the attractiveness of the respective countries to multinational enterprises and foreign institutional investors. In fact, Swiss multinational companies dominate the Swiss equity market. International investor interest in these multinationals probably explains the very large and persistent growth of traded values. International integration, especially the regional integration into the single European market, is a major reason for the fast development in recent years of equity markets in the countries of Southern Europe. Another factor is the growing preference of households for investments in mutual funds and life insurance companies as a response to the weak presence of pension funds and the concerns about the long-term credibility of social security pensions. Chile and South Africa show high levels of market capitalization with low levels of traded values. This reflects the ‘‘buy and hold’’ policies of institutional investors in these countries, a feature that is also present in the United Kingdom and the Netherlands. But restrictions on international investors also explain the low volume of trading in Chile and South Africa.
Institutional Investors, Contractual Savings and Capital Market Development
63
Since the lifting of international sanctions and the advent of policies of economic liberalization, international interest and the volume of trading have increased rapidly in South Africa. Chile continues to apply restrictions on foreign capital inflows and this has likely affected the volume of equity trading on the Santiago stock exchange. Jordan and Egypt appear to be better linked with foreign investors than Morocco and Tunisia, although all four countries have a long way to go to achieve reasonable integration with world markets.
6. CONCLUSIONS AND POLICY IMPLICATIONS EJMT have reasonably large contractual savings sectors by the standards of developing countries. Their capital markets have also experienced significant growth in recent years in line with most other developing countries. However, EJMT lag behind in the development of the insurance industry and in the efficient utilization of the long-term financial resources they mobilize. The last aspect is reflected in relatively low levels of equity trading and in low investment returns. Both of these are associated with the investment policies adopted by their institutional investors. Although EJMT have relatively young populations, their social security systems are faced with growing pressures because of the offer of benefits that are unsustainable, given current contribution rates and investment returns. As contribution rates are already quite high, in some cases they are very high by international standards, the only realistic options are either to enhance investment performance or to lower benefits. EJMT need to re-examine their social security systems to establish a closer link between contributions and benefits. This will help in decreasing contribution evasion and maintaining financial balance. They also need to consider various means for raising the investment returns of their pension funds and other institutional investors. Contracting out fund management to professional asset managers, selected on the basis of objective, meritocratic and prudent criteria is one option that needs to be considered. Allowing a partial privatization of the pension system by allowing part of the benefits from approved pension plans to be contracted out would be another option. Transforming the social security system from defined benefit to defined contribution could be a third option. Given the aging population, some or all of these measures may be inevitable. At some point, EJMT may be forced to contemplate more fundamental systemic reforms of their pension systems.
64
WAFIK GRAIS AND DIMITRI VITTAS
Another approach that would enhance the efficiency of the overall financial system would be to encourage the entry of strategic investors into the local banking, insurance, and fund management markets. A greater openness may also be warranted in the various support services, such as legal, accounting, actuarial and management consulting, that contribute to greater efficiency in the operation of capital markets. As noted above, EJMT are lagging behind many reforming countries in Southern and Eastern Europe, in East Asia and in Latin America. A greater openness would be essential in areas such as insurance and fund management where modern financial technology is highly complex and where successful transfer from more advanced markets is indispensable for accelerating growth and enhancing efficiency.
NOTES 1. In this research, bank development is not measured by the total assets of banks or even by total loans or total deposits, but rather by the amount of credit extended to the private sector, expressed in relation to GDP. This measure captures the higher economic efficiency of lending to the private sector rather than the government and the public sector, either directly or through the central bank. 2. This research, undertaken by Alberto Musalem, has not been completed but preliminary findings underscore the link between contractual savings and equity markets. 3. Compulsory contributions to funded pensions schemes, private or public, may also be perceived as a form of payroll taxation, especially if net investment returns are low relative to market benchmarks. However, implicit payroll taxes of funded schemes are usually lower than the explicit payroll taxes of unfunded pension pillars. 4. The capital market effects of contractual savings and institutional investors are discussed in greater detail in Vittas (1998a,b; 1999). See also Walker and Lefort (1999) for an empirical study of the impact of pension funds on capital market development in Latin America. 5. A scaled-premium approach sets the contribution rate for a period of 5–10 years (the period of equilibrium) at a level that is adequate to cover future benefit payments after taking into account expected net investment income and demographic factors. The contribution rate is reset at the end of each equilibrium period. 6. Despite imposing an inordinately high contribution rate of 26%, the social insurance system of Egypt succeeded in reducing the share of accumulated balances to GDP from 38% in 1988 to 32% in 1993. The culprit for this fall was the negative real rate of return of nearly 12% for most of this period. At this rate, the real value of balances declines by 47% in 5 years and by 78% in 10 years. The real rate of return on social insurance balances improved in recent years. It is now slightly positive in real terms on new and reinvested funds, though the weighted average return on all balances is still slightly below the rate of inflation. 7. These allocations are not included in the reported statistics.
Institutional Investors, Contractual Savings and Capital Market Development
65
REFERENCES Demirguc-Kunt, A., & Maksimovic, V. (1998). Stock market development and firms’ financing choices. World Bank Economic Review, 10, 341–369. International Finance Corporation (IFC) (1987, 1998). International Monetary Fund, Washington, DC. Kaminsky, G., Lyons, R., Schmukler, S. (1999). Managers, investors, and crises: Mutual fund strategies in emerging markets. Mimeo, World Bank. Levine, R., & Zervos, S. (1998). Stock markets, banks, and economic growth. American Economic Review, 88, 537–558. Vittas, D. (1998a). The role of non-bank financial intermediaries. Policy Research Working Paper No. 1892. Washington, DC: World Bank. Vittas, D. (1998b). Institutional investors and securities markets: Which comes first? Policy Research Working Paper No. 2032. Washington, DC: World Bank. Vittas, D. (1999). Pension reform and financial markets. Central America Reform Project, Discussion Paper No. 7. Cambridge, MA: Harvard Institute for International Development. Walker, E., & Lefort, F. (1999). Pension reform and capital markets: Are there any (hard) links? Mimeo, World Bank.
This page intentionally left blank
66
FINANCIAL LIBERALIZATION, SAVINGS, INVESTMENT, AND GROWTH IN MENA COUNTRIES Lahcen Achy ABSTRACT Over the last two decades Middle East and North African (MENA) countries like much of the developing countries have experienced a wave of liberalization of their financial sectors. There have been expectations that financial liberalization would enhance economic growth by stimulating savings and investment. The purpose of this chapter is three-folds: (1) to review the literature on the rationale for financial repression; (2) to examine the theoretical and empirical literature on the links between financial liberalization, savings, and investment; and (3) to assess empirically the effect of financial reforms on economic performance in the specific case of MENA countries.
1. INTRODUCTION Over the last two decades Middle East and North African (MENA) countries like many of the developing countries have experienced a wave of liberalization of their financial sectors. For a long time, the primary task of Money and Finance in the Middle East: Missed Opportunities or Future Prospects? Research in Middle East Economics, Volume 6, 67–94 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1094-5334/doi:10.1016/S1094-5334(05)06004-8
67
68
LAHCEN ACHY
the financial system was to finance the government needs, public enterprises, and priority sectors through mandatory holding of treasury bills and bonds issued by development banks. Interest rates subsidies to priority sectors have been reduced or eliminated. The monetary authorities started to manage liquidity through a more active use of reserve requirements and a more market-based allocation of refinancing. Stock markets legislation has been updated, and their management was transferred to the association of brokerage houses. New banking laws have increased the autonomy of the central bank and introduced prudential regulation in line with international standards. Finally, measures to increase competition by opening banks’ capital to foreign participation have been implemented. There have been expectations that financial liberalization would enhance economic growth by stimulating saving and investment. The purpose of this chapter is three-fold: (1) to review the literature on the rationale for financial repression; (2) to examine the theoretical and empirical literature on the links between financial liberalization, savings, and investment; and (3) to assess empirically the effect of financial reforms on economic performance in the specific case of MENA countries. The empirical assessment attempts to capture the effects of both banking sector and stock market developments. For data availability reasons, our investigation focuses on a sample of five MENA countries (Egypt, Jordan, Morocco, Tunisia, and Turkey) over the period 1970–1998. The rest of the chapter is organized as follows. Section 2 examines why governments adopt financial repression policies. Section 3 reviews the theoretical literature and empirical findings illustrate how financial liberalization can affect savings, investment, and growth. Section 4 addresses the impact of financial liberalization on economic performance in the specific context of MENA countries, and finally section 5 concludes.
2. WHY DO GOVERNMENTS ADOPT FINANCIAL REPRESSION POLICIES? Whether the degree of financial development is crucial or not for economic growth is a very debatable issue among economists. Lucas (1988) asserts that economists tend to over-stress the role of financial factors in economic growth. In contrast, a large literature on economic development states that the extent of financial intermediation in an economy is crucial for its growth (Goldsmith, 1969; Shaw, 1973; MacKinnon, 1973; Bencivenga & Smith,
Financial Liberalization, Savings, Investment, and Growth in MENA Countries
69
1991; Levine, 1997). Those who hold the latter view argue that by reallocating capital to the highest-return projects and dealing with moral hazard, adverse selection and transaction costs problems, financial intermediaries represent an essential catalyst for economic growth. So, why do governments adopt financial repression policies? Financial repression is the result of a nominal interest rate ceiling that is below the prevailing rate of inflation and currency depreciation. Under financial repression regimes, the monetary authorities impose high-reserve requirements, bank-specific credit ceilings and selective credit allocation, mandatory holding of treasury bills and bonds issued by the government, and a non-competitive and segmented financial system. Theoretically, there are several arguments in favor of financial repression. First, it allows the authorities to have better control over the money supply. Second, the government is assumed to know better than the market what investments are more desirable from a social perspective. Third, the government can protect lenders against usury practices by moderating the free determination of interest rates. Finally, financial repression is usually accompanied by interest rates below market rates, which reduces the costs of servicing government debts. Giovannini and De Melo (1993) emphasize the fact that financial repression is a taxation device used to finance the government needs at artificially low interest rates. They provide empirical evidence on the interactions between financial controls and tax policies (budgetary impact of financial repression) by estimating the amount of revenue that governments in developing countries have obtained from financial repression. Their empirical evidence covers 24 countries over the period 1972–1987. The government revenue directly arising from financial repression can be measured as the difference between foreign and domestic interest rates times the stock of domestic government liabilities. The results show that revenue from financial repression as a share of gross domestic product (GDP) has ranged from 0% in Indonesia, where domestic government debt is held exclusively by the central bank, to 6% in Mexico and Zimbabwe. In Algeria it represents 4.3%, in Morocco 2.31%, in Turkey 2.2%, in Tunisia 1.5%, and in Jordan 0.6%. Giovannini and De Melo (1993) report the two components through which financial repression was implemented: The interest rate differentials and the changes in exchange rates. The sum of these two components is equivalent to a tax rate of financial repression. These components are 8.49 and 1.59% for Algeria, 5.74 and 10.33% for Morocco, 5.98 and 7.22% for Tunisia, 0.85 and 54.64% for Turkey, and 6.71 and 0.47% for Jordan, respectively. These results suggest
70
LAHCEN ACHY
that in a number of MENA countries financial liberalization will generate a substantial budgetary problem, unless it is accompanied by appropriate fiscal measures to substitute for the revenue losses. However, there are theoretical as well as empirical arguments which suggest that financial repression may exert adverse effects on economic growth. Under such circumstances, it is relevant to investigate the government’s motivations for adopting such policies. According to Roubini and Sala-iMartin (1995) the main reason is that the financial sector represents a potential source of ‘‘easy’’ resources for the public budget through the inflation tax. Financial development, which also means a reduction in the transaction costs of transforming non-liquid to liquid assets, reduces the need to hold money. Therefore, financial development decreases the inflation tax-base and the possibilities to collect seigniorage. Therefore, tax system reforms aiming at reducing the need for seigniorage revenues have been stressed as a priority in structural adjustment programs before implementing financial liberalization reforms. To the extent that financial liberalization increases the efficiency of allocating savings to productive investments, financial repression will also have negative real effects.
3. FINANCIAL LIBERALIZATION, SAVINGS, INVESTMENT, AND GROWTH The financial system affects capital accumulation either by increasing saving rates (level effect) or by improving the allocation of savings among potential investors (reallocation effect). The availability of funds may also affect economic growth by financing technologic innovation. 3.1. Theoretical Review In an extensive survey, Levine (1997) breaks the primary function of financial sectors into five basic functions: mobilization of savings; allocation of resources; hedging and pooling risks; easing the exchange of goods and services; monitoring managers and exerting corporate control. Informational asymmetries and transaction costs create incentives for the emergence of financial markets that supply financial instruments. Banks can offer liquid deposits to savers and undertake a mixture of liquid low-return investments to satisfy
Financial Liberalization, Savings, Investment, and Growth in MENA Countries
71
demand for deposit and illiquid high-return investments. In doing so, banks provide complete insurance, against liquidity risk, to savers while facilitating long-run investments. The ability of banks and security markets to pool risk across individual projects, industries, and regions is also crucial for economic growth. By facilitating risk diversification, financial markets induce a portfolio risk toward projects with high-expected returns. Dornbusch and Reynoso (1989) argue that strong assertions concerning the positive effects of financial liberalization on economic growth are not supported by the evidence. To support their claim, they reconsider the conventional theoretical propositions and empirical facts against financial repression. First, the relationship between real deposit rates and the savings rates: Although there is a strong belief that higher interest rates stimulate savings, the offsetting income and substitution effects of increased interest rates mean an ambiguous total effect on saving. In developing countries, the lack of appropriate data and their poor quality make it harder to provide any evidence. Second, the relationship between financial depth and growth is not strong, and varies substantially across countries. Furthermore, Levine (1991) derives a model where more liquid stock markets (markets where it is less expensive to trade equities) stimulate longrun investment projects. This is because investors can easily sell their stake if they need liquidity before the project matures. He concludes that high-liquid stock market attracts investment in long-run high-return projects that boost productivity growth. Also, Mauro (1995) starts from the documented fact that Japan and continental Europe have experienced high saving rates and fast growth even though their stock markets have been relatively underdeveloped, while the U.S.A. and U.K. have been characterized by low savings and slower growth even though their stock markets have been well developed. To accommodate these facts, he incorporates some results from the precautionary savings literature into an endogenous growth model in which the young and the old share the output produced by their ‘‘family business’’. This model suggests that the existence of a stock market where investors can pool their risks is expected to reduce precautionary savings and – in a closed economy – investment and growth. Mauro’s paper shows that if uncertainty is of a multiplicative nature and utility across states is Constant Relative Risk Aversion (CRRA), then the mechanism will decrease savings and tend to lower growth if and only if the elasticity of inter-temporal substitution is below 1. However, there is a theoretical debate on whether high stock market liquidity improves saving rates. In fact, higher returns and better risk
72
LAHCEN ACHY
sharing may reduce saving. Furthermore, since more liquidity makes it easier to sell shares, it may reduce the incentives of shareholders to monitor managers (Shleifer & Vishny, 1986). Also, weaker corporate governance limits the effective resource allocation and affects economic growth. Regarding the reallocation and productivity effects of financial deepening, Greenwood and Jovanovic (1990) develop a model where the extent of financial intermediation and economic growth are endogenously determined. In their model, financial intermediaries can invest more productively than individuals because of their better ability to identify investment opportunities. They conclude that financial intermediation promotes growth because it allows a higher rate of return to be earned on capital, and growth in turn provides the means to implement costly financial structures. Concerning the role that the stock market plays in allowing investors to diversify their portfolios to hedge against idiosyncratic risk, Levine (1991) constructs an endogenous growth model in which the stock markets emerge to help agents to cope with the liquidity risk and investigates how the stock market affects investment incentives in ways that change steady-state growth rates. He shows that the existence of a stock market affects positively economic growth and may stimulate higher returns and more productive projects rather than safe storage of wealth. His paper also shows that taxing financial market activities lowers per capita growth rates.
3.2. Empirical Review To date the empirical relationship between financial system development and savings, investment, and economic growth has been difficult to test. Two streams of research may be distinguished. The first investigates more generally the relationships between financial development indicators and economic performance. The second focuses more specifically on the impact of financial liberalization on economic performance. King and Levine (1993) examine whether higher levels of financial development are positively associated with economic development using data for over 80 countries from 1969 through 1989. Three indicators have been constructed in order to measure financial development: (1) the ratio of liquid liabilities to GDP; (2) the importance of deposit banks relative to the central bank in allocating credit; and (3) the share of credit issued to non-financial institutions (in total credit and in GDP). This array of indicators provides a much better picture of financial depth than using a single indicator. Using
Financial Liberalization, Savings, Investment, and Growth in MENA Countries
73
cross-country regressions and after controlling for a set of country and policy characteristics, they find that higher levels of financial development are positively associated with faster rates of economic growth and physical capital accumulation. They also point out that financial development is a good predictor for long-run economic growth over the next 10–30 years. Atje and Jovanovic (1993) present a cross-country study of stock markets and economic growth over the period 1980–1988. They find a significant correlation between average economic growth and stock market capitalization for 40 countries. Levine and Zervos (1998a) study whether stock markets and banks promote economic growth. Their chapter extends that of Atje and Jovanovic (1993) in different dimensions: (1) by enlarging the sample size and the time coverage; (2) by using different measures of stock market development; and (3) by controlling for other economic and political factors that may affect economic growth. Using data on 47 countries from 1976 through 1993, they find that measures of stock market liquidity are strongly related to growth, capital accumulation, and productivity. Surprisingly, stock market size does not seem to be robustly correlated with economic growth. Conversely, bank lending to the private sector has a strong effect on economic growth. However, as indicated by Rajan and Zingales (1998), one may be still skeptical about the causality direction between financial development and economic growth. Specifically, financial institutions may tend to lend more if they expect future economic growth and if the stock market capitalizes the present value of growth opportunities. In this case, financial development can be viewed as a leading indicator rather than a causal factor. To overcome these issues, they suggest investigating microeconomic data on firms’ external finance. Using data at the industry level over the period 1980–1990, their findings indicate that ex-ante development of financial markets facilitates the ex-post growth of sectors dependent on external finance. Harris (1997) re-examines the empirical relationship between stock markets and economic growth using more appropriate instruments for investment and applying more advanced econometric techniques. In contrast to Atje and Jovanovic (1993), he finds no hard evidence that the level of stock market activity helps to explain growth in per capita output. Splitting the sample leads to similar results for the sub-sample of less developed countries. In the sub-sample of developed countries, however, the level of stock market activity helps to explain per capita growth but by less than half the value predicted by Atje and Jovanovic for their whole sample. Bayoumi (1993) investigates the interaction between financial deregulation and household savings behavior using regional data for the U.K. in the
74
LAHCEN ACHY
1980s. He concludes that financial deregulation was responsible for lowering the equilibrium level of savings by roughly 2.25% per year and for making savings more dependent on changes in wealth, income, and interest rates. Henry (2000) examines the relationship between stock market liberalization (the decision to allow foreigners to purchase shares in the country’s stock market) and the growth of real private investment using an event study approach on a sample of developing countries. His paper covers 10 countries (Argentina, Brazil, Chile, Colombia, India, Korea, Malaysia, The Philippines, Thailand, and Venezuela). Standard models of international asset pricing predict that stock market liberalization may reduce the liberalizing country’s cost of equity capital. This fall in the cost of equity will transform some investment project that had a negative net present value (NPV) before liberalization into positive NPV after liberalization. Data indicate that on average these 10 countries experience large, temporary increase in the growth rate of real private investment on the heels of stock market liberalization. The relationship between private investment growth and stock market liberalization persists after controlling for world business cycle effects, contemporaneous economic reforms, and domestic fundamentals. However, Henry cannot conclude that stock market liberalization causes investment booms, as the possibility of reverse causality cannot be ruled out. The evidence found by Kim and Singal (2000), and Bekaert and Harvey (2000) supports the fact that stock market liberalization causes a one-time revaluation of emerging market stock prices and a fall in the cost of capital. Why a country’s cost of equity falls after financial liberalization? There are two components in the cost of capital: the equity premium and the riskfree rate. First, stock market liberalization is expected to increase net capital inflows and this ‘‘supply effect’’ could reduce the risk-free rate. Second, more risk sharing between foreign and domestic residents should reduce equity premium. Increased capital inflows may also increase stock market liquidity, and increased liquidity reduces the equity premium (Levine & Zervos, 1998b). But, financial liberalization does not always lead to a fall in the cost of equity capital. It could increase the risk-free rate. This depends on whether the liberalization of restrictions on inflows is accompanied by a liberalization of restrictions on outflows. It also depends on whether the autarky risk-free rate (before liberalization) was above or below the world rate. Ultimately, whether a country’s cost of capital rises or falls following stock market liberalization is an empirical question that must be considered on a case-by-case basis.
Financial Liberalization, Savings, Investment, and Growth in MENA Countries
75
4. EMPIRICAL INVESTIGATION In this section we investigate the relationship between financial intermediation and growth. As was already pointed out economic agents rely on financial intermediaries to reduce information and transactions costs, to manage liquidity risk, and allocate investments optimally. By doing so, financial intermediaries influence savings and investment decisions, and hence may affect overall economic growth. Our purpose is to assess the existence of this effect and test the theoretical predictions regarding the relationship between financial development and economic growth in the particular case of some selected MENA countries. 4.1. Data Sources and Indicators 4.1.1. How to Capture Financial Liberalization? Financial liberalization reforms involve usually several key phases that are implemented in several years. These reforms cover interest rates, reserves requirements, credit allocation, bank ownership, pro-competition measures, security markets, prudential regulation, and openness of capital account (Bandiera et al., 2000). This process of financial liberalization has generally proceeded in many stages and the relaxation of different controls has been rarely simultaneous. As a result, it is difficult to define a single date in which liberalization takes place. Moreover, official and practical timing of any particular relaxation often do not coincide. There have also been partial reversals of prior liberalization in some countries. One can try to infer the timing of liberalization from the statistical properties of interest rate data themselves or other financial indicators (liquid liabilities as a share of GDP, deposit money bank assets to total financial assets, and credit by deposit money banks to private sector as a share of total credit or GDP). Following Bandiera et al. (2000), I construct a financial liberalization index that captures the exogenous changes in eight items (interest rate regulation, reserves requirements, credit allocation, bank ownership, pro-competition measures, security markets, prudential regulation, and capital account liberalization) (Appendices B–C). For each item a dummy variable that takes on the value 1 for a more liberalized regime and 0 otherwise is created. Using the principal component analysis, a weighted average of the more important principal components is computed. This indicator represents the financial liberalization index that will be used in the regression
76
LAHCEN ACHY
analysis. To this end, International Monetary Fund (IMF) country reports have been explored in order to construct database records of such policy changes.
4.1.2. Financial Depth Measures Assuming that the size of the formal financial intermediary sector relative to economic activity reveals the extent and the quality of financial services, many researchers use liquid liabilities over GDP (LL/GDP) as a proxy for financial development. However, this indicator reflects the overall size of the financial sector and does not distinguish between the allocation of capital to the private sectors, and to various governmental and quasi-governmental agencies. In this respect, it may not inform reliably on the extent to which financial services such as risk management and information processing are provided. Data on this indicator are presented in Table 1. In order to assess more accurately the contribution of commercial banks in financial intermediation, the ratio of deposit money bank domestic assets to total financial assets is used, including central bank domestic assets. The rationale for this measure is that banks are more likely to identify profitable investment, monitor managers, and mobilize savings than central banks (Levine, 1997). The weakness of this indicator is that in some countries the government influences the structure of bank assets and liabilities through mandatory holdings and controls on credit allocation. Table 2 presents data on this indicator. A more accurate indicator of financial development is provided by the value of credits by financial intermediaries to the private sector (excluding credit to money banks) divided by total domestic credit. The last indicator is computed as the ratio of claims on the financial private sector to GDP. Tables 3 and 4 present, respectively, data on these two last indicators.
Table 1.
Egypt Jordan Morocco Tunisia Turkey
Liquid Liabilities to GDP.
1970–1979
1980–1985
1986–1991
1992–1997
40.0 58.6 34.5 34.7 19.4
74.6 80.3 41.0 40.1 19.3
80.5 115.5 45.7 47.5 22.0
77.4 106.1 61.7 44.7 24.0
Source: IMF, International Financial Statistics.
Financial Liberalization, Savings, Investment, and Growth in MENA Countries
Table 2.
Egypt Jordan Morocco Tunisia Turkey
77
Deposit Money Banks Assets to Total Financial Assets. 1970–1979
1980–1985
1986–1991
1992–1997
56.2 70.1 50.2 80.3 60.2
50.3 78.4 48.8 78.5 51.4
52.9 71.8 42.8 75.3 61.4
68.0 73.0 61.6 79.6 74.6
Source: IMF, International Financial Statistics.
Private Credit to Total Domestic Credit.
Table 3.
Egypt Jordan Morocco Tunisia Turkey
1970–1979
1980–1985
1986–1991
1992–1997
24.8 72.4 47.1 84.6 56.3
24.0 74.2 39.4 87.4 45.4
26.4 61.7 44.7 87.2 52.8
39.2 75.2 55.1 93.8 62.5
Source: IMF, International Financial Statistics.
Table 4.
Egypt Jordan Morocco Tunisia Turkey
Private Credit by Deposit Banks to GDP.
1970–1979
1980–1985
1986–1991
1992–1997
14.3 33.9 13.2 — —
22.3 47.6 16.3 — —
25.6 61.2 17.2 51.6 13.4
27.5 62.0 25.7 50.5 13.5
Source: IMF, International Financial Statistics.
4.2. Econometric Analysis The econometric analysis is based on three main specifications involving three dependent variables (private savings, private investment, and real GDP growth rate). In the first two specifications, the variables are entered in the logarithmic terms (except for the real interest rate, RR) that contains negative values. One advantage of the logarithmicic transformation is that the regression directly yields elasticity coefficients for the private saving
78
LAHCEN ACHY
rates and private investment rates, with respect to the explanatory variables, making it readily possible to assess the economic relevance of the relationship. The logarithmic transformation also accounts for potential non-linearities in the relationship between the dependent and the explanatory variables. It is very likely, for example, that the effects of financial depth on private investment are minimal at low levels of financial depth and increase as the country’s financial system develops. The equations are estimated using a panel data approach and allowing specifically each of the five countries to have its own intercept (fixed-effects estimation). An F-test rejects significantly the hypothesis of a common intercept and supports the alternative of individual country intercepts. To test the effects of financial liberalization on private savings, private investment, and growth, various indicators of the extent of financial reforms are used alternatively. First, the four classical indicators of financial depth (total liquid liabilities of financial intermediaries as a percentage of GDP (LLY), deposit money banks assets as a share of total assets (BANK), private credit by deposit money banks to GDP (PRIVY), and private credit by deposit money banks to total domestic credit (PRIV)) are used to proxy indirectly for the scope of financial liberalization. In a second step, the financial liberalization index (FINX) constructed on the basis of available information on the eight main dimensions of financial reforms is used (Appendices B–C). The results are presented in Table 5 for private saving, Table 6 for private investment, and Table 7 for real GDP growth rate. To avoid omitted variables bias, different policy-related variables such as trade openness, exchange rate overvaluation, inflation rate, and real interest rate are included in the regressions as well. Various tests are conducted to examine the ways in which financial depth affects private saving, private investment, and economic growth and also to check the robustness of the estimated results. 4.2.1. Financial Liberalization and Private Saving The estimation results on the impact of financial liberalization on private savings using each of the five measures of the extent of financial development are shown in Table 5. While there are theoretical reasons which suggest that each of the included variables contribute in explaining private savings, the predicted sign for some variables is ambiguous a priori (Appendix A). The coefficients on all five financial indicators are negative although only two of them are statistically significant at the 95% level of confidence (LLY and FINX). The results support the hypothesis that an increase in financial
Financial Liberalization, Savings, Investment, and Growth in MENA Countries
Table 5.
The Impact of Financial Liberalization on Private Savinga.
Explanatory Variables
LLY
Financial depth
0.37
Real interest rate Real GDP growth Budget surplus/GDP Dependency ratio External debt/GDP Adjusted R2 F-statistic Prob(F-statistic) Observations
79
( 2.12) 0.02 (1.41) 0.005 (1.75) 0.03 ( 1.44) 0.81 ( 1.72) 0.20 ( 0.84) 0.31 5.72 0.00 114
BANK
PRIVY
PRIV
FINX
0.13 ( 1.20) 0.02 (1.35) 0.01 (3.50) 0.04 ( 2.23) 1.26 ( 1.86) 0.35 ( 1.94) 0.28 5.19 0.00 114
0.12 ( 1.55) 0.02 (1.34) 0.008 (2.31) 0.03 ( 2.13) 1.18 ( 2.06) 0.31 ( 1.74) 0.29 5.31 0.00 114
0.13 ( 0.91) 0.02 (1.37) 0.01 (3.58) 0.04 ( 2.34) 1.26 ( 2.14) 0.37 ( 1.90) 0.29 5.29 0.00 114
0.20 ( 3.23) 0.003 (1.18) 0.01 (2.03) 0.02 ( 1.92) 1.89 ( 4.63) 0.18 ( 2.13) 0.35 7.26 0.00 114
Note: These estimations are obtained using country-fixed-effect estimation over the period 1970–1997. The dependent variable is the private saving rate taking in logarithm. All regression variables are in logarithm, except for real GDP growth, real interest rate and budget surplus/ GDP. The estimates are corrected for auto-correlation and heteroskedasticity (generalized least squares, GLS, estimation). The t-statistics are given in parentheses. LLY is the ratio of liquid liabilities over GDP. BANK is the deposit money banks assets as a share of total assets. PRIVY is private credit by deposit money banks to GDP. PRIV is private credit by deposit money banks to total domestic credit. FINX is a financial liberalization index constructed on the basis of available information on the eight main dimensions of financial reforms implemented in the countries under study. Source: Author’s estimates. a Data on private saving used are from Loayza et al. (1998). significant at 10%. significant at 5%.
depth leads to a lower level of private saving. The coefficients represent the elasticities of private saving with respect to the financial depth indicators, since the regression variables are in logarithm. The first column of the table suggests that a 1% increase in liquid liabilities as a share of GDP leads roughly to a 0.37% decline in private savings. The elasticities of private savings with respect to credit issued to the private sector (as a share of GDP or as a share of total domestic credit) are invariably equal to ( 0.13). A similar result is obtained when using deposit money bank assets to total financial assets as a measure of financial development. Finally, the elasticity of private savings with respect to the composite index of financial
80
Table 6.
LAHCEN ACHY
The Impact of Financial Liberalization on Private Investmenta.
Explanatory Variables
LLY
Financial depth
0.58
Real interest rate Real GDP growth (t 1) Trade openness Inflation rate Exchange rate overvaluation Budget surplus/GDP Adjusted R2 F-statistic Prob(F-statistic) Observations
( 2.62) 0.02 (3.20) 0.01 (1.86) 0.79 (5.97) 0.13 ( 1.82) 0.006 ( 2.56) 0.01 (1.35) 0.59 16.59 0.00 109
0
PRIVY
PRIV
FINX
0.03 (0.14) 0.01 (1.44) 0.01 (2.36) 0.54 (4.16) 0.11 ( 1.31) 0.003 ( 1.53) 0.00 (0.02) 0.52 13.72 0.00 109
0.16
0.19 ( 1.24) 0.006 ( 0.89) 0.02 (2.46) 0.77 (3.50) 0.10 ( 1.19) 0.001 ( 0.92) 0.001 ( 0.09) 0.53 13.95 0.00 109
0.18 ( 1.27) 0.008 (1.21) 0.02 (2.41) 0.50 (6.06) 0.06 ( 1.43) 0.002 ( 1.10) 0.004 ( 0.50) 0.53 14.27 0.00 109
( 1.94) 0.008 ( 1.32) 0.01 (2.21) 0.57 (6.18) 0.13 ( 1.68) 0.002 ( 1.61) 0.005 (0.62) 0.54 14.63 0.00 109
Source: Author’s estimates. For footnote refer to that of Table 5. a Data on private investment used are from Pfeffermann et al. (1999). significant at 10%. significant at 5%.
liberalization is in the middle range ( 0.20). The negative impact of financial development in general and financial liberalization more specifically on savings suggests that liberalization may have relaxed credit constraints on households. Jappelli and Pagano (1994) found the same result in a crosscountry study. They argue that easing liquidity constraints on households by liberalizing consumer and mortgage credits, coupled with financial deregulation, lead to a decline in saving and economic growth rates. As noted earlier, the real interest rate can have a positive or negative effect on savings, depending on the relative magnitudes of the substitution and income effects. The evidence is also less clear as shown by the coefficients reported in Table 5. The estimates are positive, which suggest that higher real interest rates stimulate the saving behavior of private agents, but they are not statistically significant. Theoretically, in a closed economy the effect of economic growth on private saving is positive, but in a small open economy it becomes ambiguous; faster growth may reduce the savings rate by stimulating the
Financial Liberalization, Savings, Investment, and Growth in MENA Countries
81
Table 7. The Impact of Financial Liberalization on Real GDP Growth. Explanatory Variables Financial development Real interest rate GDP ( 1) Private investment (t+1) Human capital Trade openness Inflation rate External debt/GDP Term of trade changes Exchange rate overvaluation Adjusted R2 F-statistic Prob(F-statistic) Observations
LLY 2.03 ( 0.93) 0.16 (2.79) 2.65 ( 3.77) 3.48 (2.75) 4.31 (2.48) 1.62 (0.95) 0.88 (1.16) 1.80 ( 1.47) 0.62 (0.38) 0.01 (1.23) 0.34 4.95 0.00 127
BANK 1.96 ( 0.97) 0.15 (2.69) 2.60 ( 3.73) 3.88 (3.08) 3.60 (2.54) 1.87 (1.13) 1.24 (1.89) 2.75 ( 2.35) 0.22 (0.14) 0.007 (0.71) 0.35 5.19 0.00 127
PRIVY 2.05 ( 1.78) 0.16 (2.80) 2.79 ( 3.97) 3.57 (2.86) 3.86 (2.78) 1.69 (1.04) 0.93 (1.41) 2.11 ( 1.98) 1.1 (0.65) 0.006 (0.62) 0.36 5.26 0.00 127
PRIV
FINX
0.91 ( 0.68) 0.15 (2.64) 2.57 ( 3.70) 3.86 (3.07) 3.11 (2.24) 1.93 (1.16) 1.21 (1.84) 2.46 ( 2.30) 0.72 (0.42) 0.005 (0.49) 0.34 4.83 0.00 127
0.35 (0.17) 0.15 (2.67) 2.65 ( 3.27) 3.65 (2.94) 3.16 (2.19) 2.19 (1.34) 1.31 (1.95) 2.18 ( 2.07) 0.35 (0.22) 0.01 (0.82) 0.33 4.76 0.00 127
Note: These estimations are obtained using country-fixed-effect estimation over the period 1970–1997. The dependent variable is real GDP growth rate. All explanatory variables are in logarithm, except for real interest rate. The estimates are corrected for auto-correlation and heteroskedasticity (generalized least squares, GLS, estimation). The t-statistics are given in parentheses. For expansion of LLY, BANK, PRIVY, PRIV, and FINX, refer to that of Table 5. Source: Author’s estimates. significant at 10%. significant at 5%.
consumption of the young (Jappelli & Pagano, 1994). According to our estimation, the coefficient of economic growth is positive, statistically significant and quite robust across the five measures of financial development. It reveals a positive relationship between economic growth and private savings. This result tends to corroborate previous finding in the literature. A potential endogeneity bias may be suspected if growth is itself endogenous and positively correlated with the error term of the private saving equation.
82
LAHCEN ACHY
To account for this potential bias, the contemporaneous growth rate of GDP was replaced by the growth rate in the previous period. This specification yields results similar to those reported in Table 5. As expected, a higher-dependency ratio (the ratio of (the population under 15 + the population over 60) to the total population) increases family needs and puts more burdens on workers, thereby reducing savings. This negative coefficient is consistent with the life-cycle hypothesis. That is, if a high proportion of the population is not of a working age, then the dissaving patterns of this segment of the population will outweigh the saving patterns of the working cohort. The magnitude of the effect, however, does not seem to be statistically robust and vary depending on the financial depth indicator. The budget deficit as a share of GDP may affect private savings if economic agents are Ricardian and use private savings to offset changes in government savings. The negative and significant coefficient of the government budget surplus, in four out of five regressions, is consistent with Ricardian-like behavior. Finally, the results indicate a positive but insignificant coefficient on external debt as a ratio of GDP. 4.2.2. Financial Liberalization and Private Investment Theoretically, financial liberalization is associated with better allocation of capital and more efficient transformation of savings into investment. By mitigating the liquidity constraints faced by entrepreneurs and reducing transactions costs of converting illiquid to liquid assets, a more liberalized financial system should stimulate capital accumulation and have positive effects on economic growth. To test the effects of financial development on private investment, the same methodology applied for private savings is adopted. The results are reported in Table 6. The results seem to be surprising with regard to the theoretical expectations. The coefficients on financial development indicators as well as on the financial liberalization index indicate a negative impact of financial depth on private investment in the five MENA countries investigated. This finding is inconsistent with the previous literature using cross-country regressions. Levine (1993), for example, using a cross section of 77 countries over the period 1960–1989 found a positive effect of financial development indicator on real per capita capital growth. One potential explanation of our result can be found in the link between financial development, and the development of mortgage and consumer credit markets. By distorting private credit allocation in favor of households at the extent of lending to firms, financial liberalization not only have reduced private savings but also tended to reduce available loans for the
Financial Liberalization, Savings, Investment, and Growth in MENA Countries
83
business sector. To assess the validity of this explanation one needs detailed data on the allocation of private credit between households and firms. A complete characterization of the allocation of private credit (by sector, by maturity) is extremely useful to investigate the extent to which financial liberalization resulted in an effective increase in the flow of funds attributed to private investment. For data availability reasons, this exercise is left for future research. The real interest rate coefficient is positive in three cases out of five, although only significant when liquid liabilities as a share of GDP is used as a measure of financial development. In the two other cases, the real interest rate has a negative but insignificant coefficient. The positive sign on the real interest rate contradicts the theoretical prediction according to which a higher cost of capital would discourage private investment. This result may reflect uncertainty concerning future inflation, since investment decisions are taken on an ex-ante basis, while the real interest rate used here is the ex-post measure of the borrowing cost. The coefficient of economic growth is positive, statistically significant and quite robust across the five measures of financial development. This result implies a positive relationship between economic growth and private investment. To ensure that this result is not biased by the potential positive correlation of economic growth with the error term of the private investment equation, the contemporaneous real GDP growth rate is replaced with the real GDP growth rate in the previous year. To account for the effect of government policies on private investment, four variables were included in the private investment equation: the extent of trade openness, the rate of inflation, the degree of overvaluation of the real exchange rate, and the extent of the government budget surplus as a share of GDP. The first variable has the expected positive sign, highly significant and statistically robust, implying that trade openness exerts a positive effect on private investment. The second and third variables have the predicted negative sign, although only marginally significant. This negative sign is consistent with the damaging effect of uncertainty in the macroeconomic environment on private investors’ decisions. The fourth variable exhibits insignificant coefficients, although one may expect government borrowing in order to finance budget deficit to be harmful for private investment. 4.2.3. Financial Liberalization and Economic Growth The theoretical underpinning of the relationship between financial development and economic growth has recently received considerable attention in
84
LAHCEN ACHY
both theoretical and empirical research. The channels through which financial development could affect economic growth have already been presented in our literature review. The purpose here is to assess the relative contribution of financial factors using the framework provided by the recent empirical growth literature. The estimated growth equation relates real GDP growth to a set of measures of financial depth, real interest rate, and a set of control variables consisting of private investment rate, human capital indicator, trade openness, inflation rate, external debt as a share of GDP, annual change of terms of trade, and real exchange rate overvaluation. The estimation strategy is the same as before. The results are presented in Table 7. The first striking feature of our results is the negative and insignificant coefficients on financial depth indicators (except for private credit as a share of GDP). This result contradicts those of King and Levine (1993), who found a positive association between financial system development and economic growth in a cross-country context. Our results provide evidence that once fundamental variables such as private investment and human capital (proxied by the secondary enrollment rate), and policy-related variables such as trade openness, inflation rate, and the burden of external debt are controlled for, financial depth indicators fail to explain the growth experience in the five MENA countries under investigation. These findings indicate the existence of some interactions among variables that cause some of them to lose significance when they are included simultaneously. One channel of interaction presented earlier is through the effect of financial depth on private investment. The second striking feature is the positive, significant, and robust coefficient of the real interest rate. The ex-post determination of the real interest rate and the uncertainty surrounding the prediction of inflation rate ex-ante may be an explanation of this finding. The remaining results presented in Table 7 show that the coefficient estimate on lagged GDP, which stands for the catch-up term, is negative and significant. The coefficients on private investment and on human capital are positive and significant which is consistent with our expectations and corroborate the previous literature. The coefficient estimate of external debt as a share of GDP is negative and statistically robust which indicates that the burden of external debt has contributed negatively to the observed slow growth in MENA countries. Debt obligations absorb an important fraction of resources that could be mobilized for investment purposes. The results on the other variables need to be interpreted with extreme cautious due to the potential heterogeneity in growth patterns among the five countries.
Financial Liberalization, Savings, Investment, and Growth in MENA Countries
85
5. CONCLUSION Since the mid-1980s, a gradual liberalization of the financial system has taken place in MENA countries. Interest rates subsidies to priority sectors have been reduced or eliminated. The monetary authorities started to manage liquidity through a more active use of reserve requirements and a more market-based allocation of refinancing. New banking laws increased the autonomy of the central bank and introduced prudential regulation in line with international standards. Finally, measures to increase competition by opening banks’ capital to foreign participation were designed. The expectation was that higher financial development would enhance economic growth through the better mobilization of savings and more efficient allocation of capital. The aim of this chapter was to assess the impact of financial development on private savings, on private investment, and on economic growth. While our results on the effects of financial development on private saving are consistent with the previous studies, our results on the relationship between financial development and private investment on the one hand, and financial development and economic growth on the other are rather disappointing with regard to our initial expectations. The coefficients on financial development indicators, as well as on financial liberalization index, indicate a negative impact of financial depth on private investment in the five MENA countries investigated. This finding is inconsistent with the previous literature using cross-country regressions. One potential explanation of our result can be found in the link between financial development, and the development of mortgage and consumer credit markets. By distorting private credit allocation in favor of households at the expense of lending to firms, financial liberalization may not only have reduced private savings, but tended also to reduce available loans for the business sector. To assess the validity of this explanation one needs detailed data on the allocation of private credit between households and firms. Regarding the relationship between financial depth and economic growth, our results provide evidence that once fundamental variables such as private investment and human capital (proxied by the secondary enrollment rate), and policy-related variables such as trade openness, inflation rate, and the burden of external debt are controlled for, financial depth indicators fail to explain the growth experience in the five MENA countries under investigation. This is particularly plausible if financial liberalization have led to further distortion of credit allocation in favor of consumption and at the expense of productive activities.
86
LAHCEN ACHY
In order to investigate these issues in more depth, it would be worthwhile to undertake detailed analysis at the country level. Such analysis can take into account the effect of institutional differences and other country-specific developments. It can also capture the effect of private credit distribution between households and firms, among sectors, and between small and large firms.
NOTES 1. Appendix D describes the list of variables used in the chapter to investigate the relationship between financial liberalization saving, investment, and growth. It also presents data sources and the time span covered by available data for each variable.
REFERENCES Atje, R., & Jovanovic, B. (1993). Stock markets and development. European Economic Review, 37, 632–640. Bandiera, et al. (2000). Does financial reform raise or reduce saving? The Review of Economics and Statistics, 82(2), 239–263. Bayoumi, T. (1993). Financial deregulation and household saving. Economic Journal, 103(421), 1432–1443. Bekaert, G., & Harvey, C. (2000). Foreign speculators and emerging equity markets. Journal of Finance, 55, 565–613. Bencivenga, V. R., & Smith, B. D. (1991). Financial intermediation and endogenous growth. Review of Economic Studies, 58(2), 195–209. Dornbusch, R., & Reynoso, A. (1989). Financial factors in economic development. American Economic Review, 79(2), 204–209. Giovannini, A., & De Melo, M. (1993). Government revenue from financial repression. American Economic Review, 83(4), 953–963. Goldsmith, R. (1969). Financial structure and development. New Haven, CT: Yale University Press. Greenwood, J., & Jovanovic, B. (1990). Financial development, growth and the distribution of income. Journal of Political Economy, 98(5), 1076–1107. Harris, R. D. F. (1997). Stock markets and development: A re-assessment. European Economic Review, 41, 139–146. Henry, P. B. (2000). Do stock market liberalization cause investment booms. Journal of Financial Economics, 58, 301–334. Jappelli, T., & Pagano, M. (1994). Saving, growth and liquidity constraints. Quarterly Journal of Economics, 109(1), 83–109. Kim, E. H., & Singal, V. (2000). Stock market openings: Experience of emerging economies. Journal of Business, 73, 25–66. King, R. G., & Levine, R. (1993). Finance and growth: Schumpeter might be right. Quarterly Journal of Economics, 108(3), 717–737. Levine, R. (1991). Stock markets, growth, and tax policy. Journal of Finance, 46(4), 1445–1465.
Financial Liberalization, Savings, Investment, and Growth in MENA Countries
87
Levine, R. (1997). Financial development and economic growth: Views and agenda. Journal of Economic Literature, 35, 688–726. Levine, R., & Zervos, S. (1998a). Stock markets, banks and economic growth. American Economic Review, 88(3), 537–558. Levine, R., & Zervos, S. (1998b). Capital control and stock market development. World Development, 26(7), 1169–1183. Loayza, N., Lopez, H., Schmidt-Hebbel, K., & Serven, L. (1998). The world saving data base. Washington, DC: The World Bank. Lucas, R. (1988). On the mechanics of economic development. Journal of Monetary Economics, 22(1), 3–42. Mackinnon, R. (1973). Money and capital in economic development. Washington, DC: Brooking Institution. Mauro, P. (1995). Stock markets and growth: A brief caveat on precautionary savings. Economic Letters, 47(1), 111–116. Pfeffermann, G., Kisunko, G.V., & Sumlinski, M.A. (1999). Trends in private investments in developing countries: Statistics for 1970–1997. IFC Discussion Paper 37, Washington, DC. Rajan, R. G., & Zingales, L. (1998). Financial dependence and growth. American Economic Review, 88(3), 559–586. Roubini, N., & Sala-i-Martin, X. (1995). A growth model of inflation, tax evasion and financial repression. Journal of Monetary Economics, 35(2), 275–301. Shaw, E. S. (1973). Financial deepening in economic development. New York: Oxford University Press. Shleifer, A., & Vishny, R. (1986). Large shareholders and corporate control. Journal of Political Economy, 94(3), 461–488.
APPENDIX A. THE EFFECTS OF FINANCIAL LIBERALIZATION COMPONENTS ON SAVINGS Financial Liberalization Component Interest rate liberalization
Direct Effect
Higher deposit interest rates (price effect)
Effect on Savings
Substitution and income effects, total effect on saving ambiguous Reduction of reserve More resources available Net effect depends on other policy instruments requirements for lending (quantity (monetary policy: open effect) may also lead market operations) to a price effect Reduction of directed credit Reallocation within the Ambiguous effect on corporate saving and to priority sectors business sector (highreduction of household return projects) and saving more lending to households
88
LAHCEN ACHY
APPENDIX A. (Continued) Financial Liberalization Component
Direct Effect
Bank ownership (more privatization)
May be associated with an increase in lending to households More risk taking in lending and reduction of bank spreads. Wider range of saving opportunities Offset or moderate risk taking promoted through competition. May also reduce upward pressure on deposit rates Wider and more flexible range of saving instruments Flows of foreign funds and increase in rates of returns as barriers to capital outflow are removed
Pro-competition policies
Prudential regulation
Development of securities markets International financial liberalization
Source: Bandiera et al. (2000).
Effect on Savings
Reduction of household saving Ambiguous effect on saving
Ambiguous effect on saving
Can increase saving, the effect may take time to be effective Ambiguous effect on saving because banks can also borrow from abroad to sustain lending to local firms and households
Financial Liberalization Component Interest rate liberalization
Reduction of reserve requirements Reduction of directed credit to priority sectors
Egypt
Jordan
Morocco
Tunisia
Most interest rates Interest rates on deposits were liberalized in liberalized in 1990 and 1987, 1994, and 1991 1996. But some 1996: Controls on lending deposit rates and deposit rates remain regulated completely eliminated (interest on sight except a ban on deposits must not remuneration of sight exceed a ceiling of deposits and interest rates 2% points) for small saving deposits Proxy by total reserves as a share of total deposit money bank assets 1991: Liberalization of interest rates (deposits and lending)
1994: Credit ceiling for private sector abolished 1995: Credit ceiling for public sector abolished
1981: Interest rate ceiling abolished 1983: Ceilings reintroduced 1988: Ceiling eliminated
89
1991: Mandatory From 1991 to 1993, holding of treasury mandatory holding were debt instruments reduced. Obligatory was relaxed holding of treasury 1994: Abolition of the papers reduced from obligation for banks 35% in 1986 to 10% by to subscribe and end 1996 hold treasury bills
Turkey
Financial Liberalization, Savings, Investment, and Growth in MENA Countries
APPENDIX B. FINANCIAL LIBERALIZATION COMPONENTS AND FINANCIAL LIBERALIZATION INDEX (1)
90
APPENDIX B. (Continued) Financial Liberalization Component
Bank ownership (more privatization)
Pro-competition policies
Egypt
1996: Privatization of public banks ownership in joint venture and private banks
Jordan
Morocco
1989: Barriers to entry reduced by abolishing ‘‘Moroccanization’’ decree
Turkey
1996: Obligatory sectoral lending ratios abolished 1986: Measures to enlarge the scope of activities for foreign banks and off-shore banks 1981: Barriers to entry lowered
LAHCEN ACHY
Since mid-1980s: Further competition by breaking down the compartmentalization of activities between development and commercial banks. 1993: The new banking law, full integration of development banks
Tunisia
APPENDIX C. FINANCIAL LIBERALIZATION COMPONENTS AND FINANCIAL LIBERALIZATION INDEX (2) Financial Liberalization Component Prudential regulation
Development of securities markets
Egypt
Jordan
Prudential regulation Prudential regulation applied applied since progressively since 1993 1991 (liquidity ratios, the capital adequacy ratio, the solvency criterion (Basle Accord) 1997: Banks have to publish their audited financial statements in accordance with international standards 1992: Measures to ensure stock market revival
Morocco
Tunisia
Turkey
1986: New banking Prudential regulation 1991: Adoption of law becomes introduced (to be met by prudential effective. It 1996) regulations provides 1993: External audits supervisory and and off-site prudential reporting measures requirements 1994: Amendments to the banking law These requirements to be met in 1999
1994: Stock market 1983: The capital 1993: Stock market market board is capitalization and capitalization and established to turnover increased turnover increased promote and dramatically dramatically monitor the 1997: Electronic quotation 1996: Electronic securities markets quotations 1986: Istanbul Stock Exchange becomes operative
APPENDIX C. (Continued) Financial Liberalization Component International financial liberalization
Egypt
1991: Liberalization of the exchange rate. All transactions in the balance of payments including the capital account were fully liberalized (except investment in real estate)
Source: Table constructed by the author.
Jordan
Morocco
Tunisia
Turkey
1993: Current account convertibility. Liberalization of inward foreign direct and portfolio investment, and external borrowing by residents 1996: Inter-bank foreign exchange (FX) market created 1996: A Moroccan bank issued equities in international capital markets and a private Moroccan enterprise issued corporate bonds in the European markets
1994: Inter-bank FX market created 1997: Forward FX market established 1993: Current account convertibility. Liberalization of inward foreign direct and portfolio investment, and external borrowing by residents 1994: Tunisian government issued long-term bonds on the Japanese capital market
1984: FX deregulation. Banks are allowed to keep foreign currency abroad 1985: New restrictions introduced 1988: FX liberalized 1989: Capital movements liberalized 1990: Exchange rate liberalized
Financial Liberalization, Savings, Investment, and Growth in MENA Countries
93
APPENDIX D. DATA SOURCES AND INDICATORS1 Variable Liquid liabilities to GDP (LLY)
Time Span 1970–1997
Deposit money bank 1970–1997 assets to total financial assets (BANK)
Private credit by deposit money banks to GDP (PRIVY) Private credit by deposit money banks to total domestic credit (PRIV) Reserves to total deposits money banks (RES) Real interest rates (RR)
1970–1997
1970–1997
Description LL use IFS line 551 or if not available line 351 which is equal to M2. LL is a typical measure for financial depth but does distinguish between the financial sectors or between the use of liabilities Deposit money bank domestic assets are from IFS lines (24 þ 25)
Total financial assets ¼ IFS lines (12 þ 22 þ 42) Credit issued to the private sector as opposed to credit issued to governments and public enterprises. IFS (32D..ZF + 42D..ZF) Total domestic credit is from IFS (32 þ 42)
1970–1997
Total reserves are from IFS lines (20 þ 40)
1970–1997
RR is the deposit interest rate (IFS line 60L..ZF) less the rate of inflation measured by GDP deflator (line 99BIPZF) or if not available CPI (line 64). When 60L..ZF is not available, RR is computed from the lending rate (60P..ZF) less the spread (from World saving database) less inflation Private saving as a share of gross national disposable income (from World saving database: Loayza et al., 1998) Global Development Finance & World Development Indicators (for gross domestic investment); Pfeffermann
Private saving rate (PSAVY)
1970–1997
Private investment rate (PINVY)
1970–1997
94
LAHCEN ACHY
APPENDIX D. (Continued) Variable
Time Span
Growth (Grow)
1970–1997
Exchange rate overvaluation (EXOV)
1970–1997
Trade openness (OPEN)
1970–1997
Inflation rate (P) Secondary enrollment (SECENR) Budget surplus as a share of GDP (BUDY) Terms of trade (TOT) Share of population over 60 years (OLD) Share of population under 15 years (YOUNG)
1970–1997 1970–1997
Description et al. ‘‘Trends in Private Investment in Developing Countries’’ (for public investment and private investment) Annual percentage growth rate of GDP at market prices based on constant local currency Computed from real effective exchange rate index given by Global Development Finance and World Development Indicators Total trade (imports þ exports) as a share of GDP. Global Development Finance and World Development Indicators Computed from CPI (IFS line 64)
1970–1997
1970–1997 1970–1997
1970–1997
Source: Author’s estimates. CPI: consumer price index; IFS: International Financial Statistics; LL: liquid liabilities.
DETERMINANTS OF PRIVATE SAVINGS IN THE MIDDLE EAST AND NORTH AFRICA Kıvılcım Metin O¨zcan and Yusuf Ziya O¨zcan ABSTRACT This chapter investigates the relationship between private savings and a broad range of macroeconomic aggregates in the Middle East and North Africa (MENA) over the period 1981–1994. Private savings are explained by the growth rate of income and strong inertia. Public savings crowd out private savings only partially. A financial depth measure suggests that countries with deeper financial systems will tend to have higher private savings. Private credit and real interest rates capture the severity of borrowing constraints and the degree of financial repression. Inflation captures the macroeconomic volatility and has a positive impact on savings.
1. INTRODUCTION Traditional growth and saving analysis focus on the effect of higher savings on long-run growth and the impact of an increase in domestic savings on investment. The neoclassical model suggests that during the transition between steady states, an increase in saving ratios generates a higher growth Money and Finance in the Middle East: Missed Opportunities or Future Prospects? Research in Middle East Economics, Volume 6, 95–113 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1094-5334/doi:10.1016/S1094-5334(05)06005-X
95
96
KIVILCIM METIN O¨ZCAN AND YUSUF ZIYA O¨ZCAN
only in the short run (Solow, 1956). In the long run, the equilibrium rate of growth can be explained by structural and demographic variables. Endogenous growth theories developed by Romer (1986) and Lucas Jr. (1988) suggest that permanent increases in growth rates are caused by increased capital accumulation and in turn result in higher savings. If there is limited international capital mobility, higher domestic savings result in higher investment and economic growth. However, this theory is only valid if investment and domestic savings are highly correlated. There is a vast amount of empirical literature examining the correlation between domestic savings and aggregate investment. The investigation of saving behavior is also significant in empirical studies, although the lack of detailed and reliable data was the main bottleneck for a fuller development in this line of work (see Schmidt-Hebbel, Serven, & Solimano (1996) for a comprehensive survey). Among several empirical studies is Edwards’ (1996) investigation of the process of determining the saving rates. In contrast to previous studies, Edwards makes a distinction between private and government savings in the world economy. He discusses why saving ratios in Latin America have traditionally been so low, and compares them with some of the most successful countries in the world. Edwards also incorporates some policy-related, demographic, structural and political variables in determining the saving ratios. Per capita income growth seems to be the most important determinant of private and public savings; public savings are lower in countries with higher political instability and only partially crowd out private savings. Dayal-Ghulati and Thimann (1997) investigate the empirical determinants of private savings for a sample of economies in Southeast Asia and Latin America over the period 1975–1995. They made use of panel estimations to explore the relationship between private saving rates, and policy and non-policy variables. Their findings indicate that fiscal policy, particularly social security arrangements, influences private savings; macroeconomic stability and financial deepening also appear to be important variables in determining the saving behavior in those two regions. Loayza, Schmidt-Hebbel, and Serven (2000a) investigate the factors behind the broad saving disparities using a large cross-country time-series data, constructed for the World Bank Saving Project (WBSP). Their paper contributes to the literature by using the largest data set on aggregate saving measures compiled to date. They used the panel-instrumental variable technique, which allows correction for endogeneity and heterogeneity throughout the internal instruments. Motivated by previous studies, this chapter aims to contribute to the existing empirical literature by examining the determinants of private
Determinants of Private Savings in the Middle East and North Africa
97
savings for a sample of economies in the Middle East and North Africa (MENA). A cross-country time-series data set constructed for the WBSP is employed over the period 1981–1994 (for the details of the data set, see Loayza, Lopez, Schmidt-Hebbel, & Serven, 1998b). The data reveal that the mean private saving rate for the 15 selected MENA countries is greater than the world saving rate (0.26 vs. 0.16, respectively). However, the mean saving rate for MENA countries drops to 0.24 when the United Arab Emirates is excluded. Furthermore, with the exclusion of three high saving countries (Kuwait, Bahrain and the United Arab Emirates), the mean saving rate becomes 0.22. The median value is 0.25. When only seven MENA countries (Egypt, Iran, Jordan, Morocco, Syria, Tunisia and Turkey) are considered, the mean saving rate drops to 0.17, which is almost equal to the world saving rate. The force behind the savings in MENA countries can be delineated, to a large extent, by the growth rate of income. Public savings seem to play quite an important role in determining the rate of private savings. This chapter tests if public savings crowd out private savings and also incorporates the role of demographic, policy and non-policy variables as potential determinants of private savings in the region. It applies panel estimation technique to investigate the relationship between the private saving rate and a broad range of variables. The organization of the chapter is as follows. Section 2 summarizes the trends in private savings in the MENA region. Section 3 briefly presents the determinants of private saving rates in the literature. Sections 4 and 5 define an empirical private savings model and report the econometric results, respectively. Section 6 compares MENA countries to less developed countries (LDC) and to a group of Organisation for Economic Co-operation and Development (OECD) countries. The last section (Section 7) presents some concluding remarks.
2. TRENDS IN THE PRIVATE SAVINGS IN MENA REGION The data reveal that the private saving rate, that is, saving/gross private disposable income (GPDI), has an overall mean rate of 0.26 for 15 MENA countries with a standard deviation of 0.135. The descriptive statistics of the private saving rate of the 15 MENA countries, covering the period 1981– 1994, are illustrated in Table 1. Syria has the lowest mean saving rate
KIVILCIM METIN O¨ZCAN AND YUSUF ZIYA O¨ZCAN
98
Table 1.
Descriptive Statistics of Private Saving Rates (S) of 15 MENA Countriesa.
Mean
SBHR SCYP SEGY SIRN SISR SJOR SKWT SMLT SMAR SOMN SSAU SSYR STUN STUR SARE
0.397 0.257 0.160 0.272 0.289 0.189 0.324 0.260 0.207 0.287 0.186 0.104 0.216 0.251 0.624
Maximum Minimum
0.520 0.291 0.237 0.386 0.357 0.329 0.609 0.294 0.252 0.463 0.446 0.231 0.262 0.304 0.791
0.256 0.208 0.000 0.218 0.212 0.088 0.118 0.190 0.171 0.016 0.117 0.062 0.170 0.135 0.466
Standard Deviation
0.092 0.024 0.060 0.042 0.044 0.070 0.125 0.030 0.027 0.141 0.187 0.107 0.023 0.051 0.115
Jarque- Probability Observations Bera Normality 15,847 1,174 3,785 5,614 0.779 0.796 0.375 2,375 1,128 1,292 1,411 1,223 0.059 2,633 1,165
0.452 0.555 0.150 0.060 0.677 0.671 0.828 0.304 0.568 0.524 0.498 0.542 0.970 0.267 0.558
15 15 15 15 15 15 15 14 15 15 15 12 15 15 11
Source: Author’s estimates. Bahrain (BHR), Cyprus (CYP), Egypt (EGY), Iran (IRN), Israel (ISR), Jordan (JOR), Kuwait (KWT), Malta (MLT), Morocco (MAR), Oman (OMN), Saudi Arabia (SAU), Syria (SYR), Tunisia (TUN), Turkey (TUR) and the United Arab Emirates (ARE).
a
(0.104), while the United Arab Emirates has the highest (0.624). Due to the considerable range in the mean saving rate, it is preferable to categorize the countries into three groups. The first group consists of countries that have a mean saving rate lower than 0.20 (Syria, Egypt, Saudi Arabia and Jordan). The second group of countries has a mean saving rate between 0.21 and 0.30 (Morocco, Tunisia, Turkey, Cyprus, Malta, Iran, Oman and Israel). The third group comprises of countries with a mean saving rate higher than 0.31 (Kuwait, Bahrain and the United Arab Emirates). Among countries of the first group, Syria has a very different picture compared to the other countries in the region. In addition to having the lowest mean saving rate (0.10), Syria also has a high standard deviation (0.107), which is an indicator of serious fluctuations in the saving rate. The falling saving rate after 1980 seems to recover in 1984. Annual private saving rates of the 15 MENA countries in the 1980–1994 period are given in Table 2. Starting in 1985, Syria experienced fast drops in its savings, which reached below zero in 1987. Although Syria’s saving rate increased to 0.07 in
Private Savings of 15 MENA Countries Over the Sample Period 1980–1994.
Low Saving Countries Year 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 Average
SSYR SEGY 0.23 0.15 0.18 0.14 0.23 0.20 0.16 0.04 0.03 0.02 0.07 0.06
0.10
0.22 0.13 0.21 0.21 0.24 0.17 0.13 0.00 0.17 0.14 0.13 0.20 0.21 0.13 0.12 0.16
SJOR 0.33 0.29 0.25 0.19 0.22 0.09 0.15 0.15 0.21 0.24 0.22 0.13 0.11 0.13 0.13 0.19
Medium Saving Countries SSAU 0.33 0.06 0.09 0.12 0.05 0.02 0.17 0.08 0.26 0.31 0.45 0.38 0.36 0.32 0.32 0.19
High Saving Countries
SMAR STUN STUR SCYP SMLT SIRN SOMN SISR SKWT SBHR SARE 0.19 0.18 0.18 0.21 0.20 0.25 0.25 0.23 0.23 0.21 0.22 0.20 0.18 0.17 0.19 0.21
0.20 0.19 0.19 0.26 0.23 0.21 0.17 0.22 0.22 0.24 0.23 0.24 0.23 0.20 0.22 0.22
0.14 0.20 0.19 0.18 0.24 0.27 0.28 0.28 0.30 0.28 0.27 0.27 0.28 0.30 0.29 0.25
0.26 0.29 0.25 0.24 0.26 0.24 0.27 0.29 0.28 0.27 0.26 0.21 0.21 0.28 0.25 0.26
0.26 0.29 0.27 0.22 0.23 0.19 0.23 0.29 0.28 0.27 0.26 0.27 0.29 0.28 0.26
0.39 0.27 0.23 0.27 0.27 0.24 0.27 0.32 0.26 0.25 0.28 0.23 0.29 0.22 0.30 0.27
0.45 0.46 0.41 0.39 0.41 0.33 0.34 0.35 0.16 0.23 0.33 0.19 0.19 0.05 0.02 0.29
0.35 0.35 0.30 0.35 0.36 0.31 0.26 0.21 0.24 0.26 0.28 0.30 0.28 0.25 0.25 0.29
0.12 0.36 0.19 0.26 0.34 0.17 0.21 0.37 0.27 0.30 0.38 0.44 0.61 0.41 0.41 0.32
0.50 0.48 0.46 0.48 0.45 0.43 0.33 0.26 0.45 0.35 0.43 0.27 0.28 0.28 0.52 0.40
0.79 0.74 0.72 0.70 0.69 0.68 0.51 0.53 0.47 0.51 0.53
0.62
Determinants of Private Savings in the Middle East and North Africa
Table 2.
Source: Author’s estimates.
99
100
KIVILCIM METIN O¨ZCAN AND YUSUF ZIYA O¨ZCAN
1990, it declined again in 1991 to reach a negative rate. The rate for Egypt (0.16) seemed to fluctuate to some extent, and fell below the overall mean between 1985 and 1987. Despite a brief recovery, in the following 2 years, the saving rate fell again in 1993. Saudi Arabia (0.186) has the highest standard deviation among saving rates, which is an indication of an unstable saving behavior during this period. The country experienced a downward trend between 1980 and 1984. Despite small fluctuations in the following years, savings had an upward trend until 1990. After that period, savings decreased gradually to 0.32 in 1994. Jordan (0.189) is a country with considerable fluctuations in its savings. It was at its lowest rate (0.09) in 1985 and recovered gradually, reaching 0.24 in 1989. Savings in Jordan started to fall after 1985. Among the countries of the second group, Morocco’s saving rate (0.207) was the lowest and tended to decrease gradually after 1990. Its rate experienced an upward trend after 1993. The saving rate of Tunisia (0.216) was quite stable over the period. Despite some small fluctuations between 1982 and 1987, it tended to stay within the range 0.20–0.25. Turkey (0.251) had the median saving rate for the period. The relatively stable rate in Turkey reached a peak in 1988. It first followed a downward trend, then an upward one between 1988 and 1993. Savings decreased to some extent after 1993. Cyprus (0.258) is one of the most stable countries in terms of saving rates. Throughout the period, the rate of savings stayed between 0.20 and 0.30. Malta (0.261) had a similar saving behavior to Cyprus. Savings dropped below the 0.20 only once in 1985, after which it remained quite stable until the end of the period. Iran (0.272) experienced a drastic drop in its savings between 1980 and 1985. Following a recovery in 1985, savings remained less volatile and tended to increase after 1993. Oman (0.289) presents a unique case of a consistently falling saving rate. If the upward trends in 1988 and 1994 are disregarded, the country had a falling rate during the whole period. The relatively higher mean saving rate is obviously due to the higher initial savings between 1980 and 1984. Starting in 1993, the country had nearly zero savings. Israel (0.289) is also among the countries that exhibited a stable saving behavior during this period. The country experienced a serious downward trend between 1984 and 1988, but then an upward trend seemed to stabilize the saving rate around 0.25 after 1988. Among the third group of countries, Kuwait (0.324) followed, in general, an upward trend in the period. Savings were around 0.10 in 1980 and increased to 0.40 by 1994. Two serious troughs were observed in 1982 and 1985. Bahrain (0.397) was the country with the second highest initial saving rate (0.50) in 1980. Savings followed a gradual downward trend until 1987,
Determinants of Private Savings in the Middle East and North Africa
101
reaching a level as low as 0.25. This trend reversed in 1987 and then the rate fluctuated to some extent until 1992. The country experienced a drastic increase in savings in 1994 reaching a rate of 0.52. The United Arab Emirates (0.62) is the country with the highest saving ratio in the region. This is due to the highest initial saving ratio (almost 0.80) that fell to 0.53 in 1991 as the result of a straight downward trend until 1986. The following trends are obvious from these descriptions. First, countries with high saving ratios in the period had similarly high saving rates before the period. It seems that their higher saving ratios are due, to some extent, to spillover effects. Second, there are vast differences among the countries in terms of their mean saving ratios. This heterogeneity may have serious implications for the analysis of the determinants of savings in the region. Third, no country had a perfectly consistent trend (increasing, decreasing or stable) throughout the period. Although it is quite difficult to interpret these trends, either for each country or the entire region, one can attempt to pinpoint certain tendencies for both. All of these countries experienced either domestic or international conflicts during this period, or they shared borders with those countries that had domestic or international political conflicts. The Egypt–Israel War, Iraq’s attack on Kuwait, the Iran–Iraq War, the PKK problem in Turkey and Iraq, and the fundamentalist religious movements in North African countries are good examples of such conflicts. Politically, MENA region is one of the most unsettled regions of the world.
3. A REVIEW OF THE DETERMINANTS OF PRIVATE SAVING RATES The formidable task of understanding the determinants of private savings and its relation to growth has received equal attention in both industrial and developing countries. The previous empirical literature on the determinants of private saving and growth rates covers both industrial and developing countries (Aghevli, Boughton, Montiel, Villanueva, & Woglom, 1990; Edwards, 1996; Muradoglu & Taskin, 1996; Bailliu & Reisen, 1998; Masson, Bayoumi, & Samiei, 1998; Loayza, 2000a,b). For industrial-country samples, one can refer to Haque, Peseran, and Sharma (1999), and for developing countries to Corbo and Schmidt-Hebbel (1991), and Dayal-Ghulati and Thimann (1997). In the following sections, we discuss some of the variables that constitute the determinants of private savings.
102
KIVILCIM METIN O¨ZCAN AND YUSUF ZIYA O¨ZCAN
3.1. Fiscal Policy Variables The relationship between government and private savings has been a focal issue in fiscal policy literature. Fiscal policy affects saving behavior through revenue and expenditure. The most widely accepted view is based on the Ricardian paradigm: ‘‘For a given path of government spending, a deficit financed cut in current taxes leads to higher future taxes that have the same present value as the initial cut’’. Thus, the concept of ‘‘Ricardian Equivalence’’ can be defined as follows: A rise in the budget deficit due to a tax cut would not affect the government saving rate since private saving would rise by an equivalent amount. However, empirical results reveal that government saving partially crowds out private saving. Therefore, raising government savings helps to increase national savings (Dayal-Ghulati & Thimann, 1997).
3.2. Financial Market Developments It is difficult to find a good proxy to represent developments in financial market. The most available data is a set of narrow and broad monetary aggregates. Almost all studies have used the degree of monetization of the economy measured by the broad money (M2) in relation to gross domestic product (GDP) to measure financial market developments. It has been shown that financial market developments have a positive net effect on saving (Edwards, 1996; Dayal-Ghulati & Thimann, 1997; Loayza et al., 2000a). The effect of increases in real interest rates (RR) on savings has been the subject of a large number of empirical country studies. The existing evidence suggests that there is only a weak-interest elasticity of aggregate domestic savings for both developed and developing countries (Boskin, 1978; Giovannini, 1983). McKinnon (1991) reports that aggregate savings, defined as the percentage of gross national product (GNP) accounts, do not respond strongly to higher RR.
3.3. Demographic Variables According to the life-cycle model of Modigliani (1970), economic agents will have negative savings when they are young and have very low income, positive savings during their productive years and negative savings when they grow old and retire. Therefore, aggregate private savings will be affected by the age distribution of the population. When the share of the
Determinants of Private Savings in the Middle East and North Africa
103
working population increases relative to that of retired persons, savings are also likely to increase (Lahiri, 1989; Edwards, 1996; Higgins & Williamson, 1996; Dayal-Ghulati & Thimann, 1997; Loayza et al., 2000a). Demographic variables are long-run variables and they explain long-term trends in saving ratios. However, they provide no information about the short-term changes in saving behavior. 3.4. Income Level and Economic Growth In the life-cycle hypothesis framework, income growth is considered to have an important positive effect on private savings since savings and growth are highly correlated in the long run. Growth increases savings since it increases the income of the young people relative to that of the old. On the other hand, the permanent income hypothesis suggests a negative relationship between growth and saving since forward-looking agents believe that their permanent income will increase in the future and so they will dissave against future receipts. 3.5. Other Determinants of Savings For an open economy model, the terms of trade is a critical variable, particularly for the oil exporting countries (Ostry & Reinhart, 1992; DayalGhulati & Thimann, 1997; Loayza et al., 2000a). Open economy models should also include domestic interest rates that are related to foreign interest rates. However, in general obtaining the interest rates from the available sources proves to be a difficult task.
4. AN EMPIRICAL MODEL AND THE DATA 4.1. Empirical Private Savings Model Based on the discussion in Section 2, the empirical private savings equation can be specified as S tk ¼ b0 Sðt
1Þk
þ b1 GS tk þ b2 DLPtk þ b3 YDtk þ b4 ODtk
þ b5 U tk þ b6 T tk þ b7 C tk þ b8 M tk þ b9 RRtk þ b9 LY tk þ b10 DLY tk þ b 11 WI tk þ etk
ð1Þ
where tk denotes country k in time period t and L denotes the natural logarithm of the variables of interest. D is a difference operator and DL defines
104
KIVILCIM METIN O¨ZCAN AND YUSUF ZIYA O¨ZCAN
the growth rate of the particular variable. S is private savings as a percentage of GPDI and S(t 1)k is the one period lag of Stk. GS is the public sector’s saving as a percentage of GPDI and can be considered as one of the fiscal policy-related variables. M is the money supply plus quasi money (M2) as a percentage of GNP. P is the GDP deflator, and therefore, DLP is the rate of inflation. DLP can be treated as a macroeconomic stability variable. YD, OD, U, LY and DLY represent life-cycle variables, including the young dependency ratio (YD), the old dependency ratio (OD), the urbanization ratio (U), the log of real per capita GPDI (LY), and its growth rate (DLY ). To capture the characteristics of the financial sector, credit to the private sector (end of period) as a percentage of GPDI, M2/GNP and RR are used. To account for the effects of war years on the respective economies, a war dummy was created. It takes the value of one for the Iran–Iraq War period (September 1979 to July 1988) and for the Gulf War period (August 1990 to July 1992) and zero otherwise. War dummies are included only for those countries involved in conflicts. WI is the war dummy and represents the Iran–Iraq and Gulf War years, e is the error term with mean zero and constant variance s2 :
4.2. World Saving Database The data set is obtained from the saving database recently constructed by the World Bank, which is described in detail in Loayza, Lopez, SchmidtHebbel, and Serven (1998a). The database includes 150 countries and covers the period between 1960 and 1995. However, not all variables are available for every country and for every year. The effective time coverage of the data files varies substantially across modules and, within a given module, across countries and variables. In this study, the period covers from 1981 to 1994 for the selected MENA countries. In many analyses, the measurement of the variables stands as a formidable task. There is a considerable dispute about the definitions of private and public savings, which is the main source of the measurement problem. The definition of the public saving, for instance, is particularly important for the oil exporting economies included in this study, where oil wealth is held by the public sector, as well as for economies with a large public sector such as Turkey. This study adopts the same definition used in the World Saving Database (WSD), which deserves careful attention. Following Loayza et al. (1998b), the basic income measure is taken as gross national disposable income (GNDI), which is equal to GNP plus all net unrequited transfers from abroad. Gross national saving is defined as GNDI minus
Determinants of Private Savings in the Middle East and North Africa
105
consumption expenditure, both measured at current prices. Gross private saving is then computed as the difference between gross national savings and public sector savings, which is consolidated central government savings in terms of units of domestic currency. For the public sector, four alternative saving measures are available in the WSD. In this study, public sector saving is defined as central government saving. Although all demographic, financial deepening and fiscal policy variables are available in the WSD, it is unfortunate that the spread between average deposit and lending rates, deposit rates, treasury bill rate or money market rates are only available for a few MENA countries. However, discount rates, which are available for the majority of the countries in the data set, are used in the analysis. Similarly, terms of trade are only available for the period of 1982–1993, which shortens the sample period significantly.1
5. ESTIMATION RESULTS This section reports estimation results of the private savings equations for a 12-country data set (Cyprus, Egypt, Iran, Israel, Jordan, Malta, Morocco, Oman, Saudi Arabia, Syria, Tunisia and Turkey).2 The estimation was performed on panel data, which are available for a 14-year period from 1981 to 1994 (Syria 1981–1991 and Malta 1981–1993). Following Loayza et al., (2000a) a dynamic specification is used in order to allow for inertia, which is very likely to be present in the annual data. Inertia in saving rates can arise from the lagged effects of the explanatory variables on saving. Considering the lagged-dependent variable as a possible explanatory variable allows us to use dynamic specification and to discriminate between short- and long-run effects on saving. We employ the Generalized Least Squares (GLS) method using crosssection weights in which the constant terms are allowed to vary across the countries. This allows the study to capture the underlying country-specific factors. The total panel (unbalanced) observations are 164 and we estimated Eq. (1) with the covariance matrix adjusted to allow for heteroskedasticity within the sample using the White Adjustment Procedure.
5.1. Persistence The results in Table 3 confirm the persistence in the saving equation. The coefficient of the lagged-dependent variable (lagged private saving rate) is
KIVILCIM METIN O¨ZCAN AND YUSUF ZIYA O¨ZCAN
106
Table 3.
Determinants of Private Savings of 12 MENA Countries.
Variables St 1 M C YD OD U T GS DLP LY DLY Fixed effects CYP-C EGY-C ISR-C JOR-C MLT-C MAR-C OMN-C SAU-C TUN-C TUR-C SYR-C IRN-C R2 S.E. of regression F-statistics Durbin–Watson statistics
Coefficient 0.459 0.107 0.214 0.349 1.123 0.147 0.037 0.225 0.037 0.073 0.114
t-Statistics 7.870 2.771 5.361 2.411 1.619 2.868 3.779 7.754 3.137 3.086 2.604
Probability 0.0000 0.0063 0.0000 0.0172 0.1075 0.0048 0.0002 0.0000 0.0021 0.0024 0.0102
0.539 0.608 0.603 0.489 0.510 0.684 0.534 0.645 0.421 1.027 0.862 0.971 0.963 0.044 374.922 1.888
Note: Estimation method is GLS with cross-section weights. Sample period is 1981–1994. Included observations are 14 and number of cross-sections used is 10. Total unbalanced panel observations are 135. White Heteroskedasticity-Consistent Standard Errors (S.E.) and Covariance are calculated and cross-sections without valid observations are dropped. Source: Author’s estimates.
positive (0.46) and significant for the fixed effect model, which indicates a large degree of persistence. The study also shows that the long-run effects of the other private saving determinants are 1.85 times as large as their respective short-run effects. The result concerning persistence is consistent with the findings of previous research conducted by Loayza et al., (2000a).
Determinants of Private Savings in the Middle East and North Africa
107
5.2. Income Level and the Growth Rate The results show that the real (log) per capita income level has a positive (0.07) and statistically significant impact on the private saving rate, indicating that higher-income countries tend to save a higher fraction of their GDP. In addition, the real per capita growth rate was found to be statistically significant and positive (0.11), implying that a 1% increase in the growth rate would yield 0.11% higher savings in the short run. A 10% improvement in the terms of trade increases the private saving rate by 0.37% points in the short run. When the impact of saving on growth (through capital accumulation) is taken into account, it is possible to confirm the hypothesis that there is a circular effect, as faster growth leads to increased saving, which then leads to even higher growth (Edwards, 1996).
5.3. Demographic Variables The MENA region has the highest population growth rates among developing regions. The analysis reveals that, for MENA countries, all demographic variables (the young dependency ratio, the old-dependency ratio and urbanization) are statistically significant. However, only the coefficient of the urbanization ratio was negative, which is in line with the predictions of the life cycle and precautionary savings models. These findings partially confirm Deaton’s (1990) buffer stock approach to the private savings. The positive sign of the coefficients of the young- and the old-dependency ratio can be explained by child-rearing traditions. These strictly observed the traditions dictate that children live with their parents until they complete their education and get married. Raising a number of children is an expensive process and consumes much of the household’s savings. These savings can be considered as human capital investment made on the younger generation by their parents. In fact, when the young-dependency ratio increases, private saving ratio also increases. This tradition, coupled with the tradition of not allowing children to work for money until they finish their education, increases the burden on parents, and lessens their opportunity to save. In contrast to what the relevant literature presents, older people may start saving money after their children complete their education and get married. Their income may be lower as they grow older, but their expenses decrease as well after they have raised their children.
108
KIVILCIM METIN O¨ZCAN AND YUSUF ZIYA O¨ZCAN
5.4. Fiscal Policy Variables The impact of government savings on private investment is an issue that has received considerable attention, from a policy as well as a theoretical perspective. Our results indicate a negative and statistically significant coefficient ( 0.23) for the government savings to GPDI ratio, confirming that government savings will tend to crowd out private savings. Specifically, the private sector reduces its savings rate by 0.23% for each 1% increase in the public saving ratio within the same year when the policy change occurs. However, in the long run, the offset coefficient rises to 0.43. An important issue, especially regarding policy prescriptions, is whether or not this coefficient is statistically different from 1.0 (the case of complete crowding out). Testing this hypothesis yields the conclusion that the coefficient is indeed significantly different from 1, implying that crowding out is only partial. This indicates that the Ricardian equivalence does not hold strictly, and that the government can increase aggregate savings by increasing public savings (Edwards, 1996).
5.5. Financial Market Developments and Macrostability The financial variables included reflect the degree of development of the financial system, as well as the tightness of borrowing constraints. The results reveal that the M2/GNP ratio, used to reflect ‘‘financial depth’’, has a positive and statistically significant coefficient (0.107), as suggested by Loayza et al. (2000a). Private credit and RR also capture the severity of borrowing constraints and the degree of financial repression. We found negative and statistically significant coefficients for the private credit to GNDI ratio. If the flow of private credit rises by 1%, the private saving rate decreases by 0.21% on impact. This suggests that the relaxation of credit constraints leads to a decrease in the private saving rate (Japelli & Pagano, 1995; Loayza et al., 2000a). However, we could not find any statistically significant real interest rate effects on savings in the fixed-effect model, and therefore, it was dropped from model (1). In line with some studies in the literature, we approximated macroeconomic stability by using the inflation rate and found a positive (0.04) and statistically significant effect. However, one should be cautious while interpreting this result, as the indirect and positive effect of reduced inflation on savings (through growth) must also be taken into account (Loayza et al., 2000a).
Determinants of Private Savings in the Middle East and North Africa
109
The presence of war within the study period is included only for those countries involved in conflicts.3 The private saving equation was estimated both with a war dummy and without a war dummy. A war dummy has a negative ( 0.026) and statistically insignificant impact on the private saving rate.4
6. A COMPARISON BETWEEN MENA, LDC AND OECD COUNTRIES Having analyzed the determinants of private saving rates for MENA countries, we find it insightful to compare our estimation results on MENA countries with those of LDC and OECD countries as in Loayza et al. (2000a). The addition of the LDC and OECD countries in the analysis reveals an interesting fact in that the impact of the young dependency ratio turns out to be positive for the MENA group and negative for the LDC and OECD groups, respectively. However, the statistical significance of variables, varies across the groups (see Table 4). Looking at the estimation results, the persistence of saving is uniformly positive and statistically significant across the three groups, as is also the case for inflation, terms of trade and per capita income growth rate. The public savings to GPDI ratio also has a consistently negative and significant coefficient but is less than 1. The coefficient for the credit/GPDI ratio is negative and statistically significant, which is uniform across the samples. However, this uniformity disappears once we consider the M2/GNP ratio. Although its impact is found to be insignificant for LDC, the coefficient is significant for the MENA and OECD samples. The demographic variables also exhibit differences in significance across the samples, although all LDC and OECD samples have negative coefficients, which is in line with the life-cycle/precautionary saving hypotheses. However, for MENA, only the urbanization ratio has a negative impact on the private saving of the region. For the OECD group, all the three demographic variables are insignificant, unlike MENA and LDC, for which all are statistically significant. The three coefficients for demographic variables are small in the OECD sample, unlike LDC, where they are greater and statistically significant. This fact can be explained by the non-linearity of the effects of demographic variables, as well as the greater homogeneity across OECD populations in terms of urbanization and age structure
110
KIVILCIM METIN O¨ZCAN AND YUSUF ZIYA O¨ZCAN
Table 4. Determinants of Private Savings of MENA, LDC and OECD Countries. Variables St 1 M C YD OD U T GS DLP LY DLY
MENA 0.459 0.108 0.214 0.349 1.124 0.148 0.037 0.225 0.038 0.074 0.114
LDC 0.476 0.024 0.508 0.279 1.370 0.240 0.044 0.283 0.177 0.071 0.425
OECD 0.674 0.028 0.085 0.068 0.218 0.080 0.068 0.112 0.157 0.013 0.285
Source: Author’s estimates. Statistically significant coefficients.
(Loayza et al., 2000a). Although the coefficient of the level of real per capita income is consistently positive across samples, it has an insignificant coefficient for the OECD group. The effect of income and growth is higher in LDC and MENA than that in OECD. This finding supports the subsistence–consumption theory in that saving rates in low-income countries are more responsive to changes in the level and growth rate of income. The coefficient of the credit/GPDI ratio is also small in OECD countries, as the borrowing constraints in those countries are not strict. Therefore, a further increase in this ratio would not imply a significant relaxation in the borrowing availability in these countries.
7. CONCLUDING REMARKS This chapter examined the empirical determinants of private savings for a sample of selected MENA countries. The empirical results reveal that in the MENA region, private saving rates have strong inertia. The effects of a change in a given saving determinant are fully realized in the long run rather than in the short term (i.e. more than a year later). In the region, real per capita income has a positive impact on private saving rates, which indicates that ceteris paribus, more advanced countries, tend to save a higher percentage of their GDP. Real per capita growth rate is
Determinants of Private Savings in the Middle East and North Africa
111
also positively related to savings, which supports the hypothesis that there is a circular effect faster growth leads to increased saving and then to even higher growth. Moreover, while the negative impact of dependency ratios and urbanization rate lends support to the life-cycle hypothesis, they are statistically insignificant. The most important finding of this study is that government saving to GPDI ratio has a negative impact on the saving rate, confirming the claim that government savings will tend to crowd out private savings. For the MENA countries, increases in government savings will be offset by reductions in public savings. The findings further indicate that, although higher government savings crowd out private savings, they do it in a less than oneto-one manner and the Ricardian equivalence does not hold strictly. From a policy point of view, MENA financial depth and development, as measured by the M2 to GPDI ratio, suggests that countries with deeper financial systems will tend to have a higher private saving rate. Private credit and RR also capture the severity of borrowing constraints and the degree of financial repression. The precautionary motive for saving is supported by the findings that inflation captures the degree of macroeconomic volatility and has a positive impact on private saving in the area.
NOTES 1. Alternatively the data is extended to the years before 1982 by constructing the terms of trade simply as the nominal export/import ratio divided by the real export/ import ratio. However, only six MENA countries namely Egypt, Israel, Malta, Morocco, Tunisia and Turkey benefit from the data extension using facilities of the WSD modules. Therefore, we pursued our interest of analyzing the saving behavior of initially selected 15 MENA countries. 2. Bahrain, Kuwait and United Arab Emirates are dropped from the data set due to their high saving ratios (0.40, 0.32 and 0.62, respectively) compared to the remaining group of countries. 3. Since Kuwait is dropped from the data set, the war dummy only represents the Iran and Iraq War. 4. The equation estimated with war dummy is not reported here but it can be requested from the authors.
ACKNOWLEDGMENTS An earlier version of this article was presented at the Third Mediterranean Development Forum (MDF3): Voices for Change Partners for Prosperity,
112
KIVILCIM METIN O¨ZCAN AND YUSUF ZIYA O¨ZCAN
Financing Development Workshop, held in Cairo, March 2000, jointly sponsored by the World Bank Institute. The authors also benefited from the seminar presentations at Bilkent University and at the International Conference on Banking, Financial Markets and the Economies of the MENA, held in Byblos, Lebanon, May 2000, sponsored by the School of Business at Lebanese American University. The authors are grateful to the participants of the conferences, as well as Luis Serven, Klaus Schmidt-Hebbel, Jamal AlKibbi, Isac Diwan, Merih Celasun, Bilin Neyapti, and two anonymous referees who provided helpful criticism on an earlier draft. However, they are not responsible for any errors. We are also grateful to Go¨rkemli Demirel for excellent assistance. Financial assistance of the World Bank Institute is gratefully acknowledged.
REFERENCES Aghevli, B. B., Boughton, J. M., Montiel, P. J., Villanueva, D., & Woglom, G. (1990). The role of national saving in the world economy: Recent trends and prospects. International Monetary Fund, Occasional Paper No. 67, Washington, DC. Bailliu, J., & Reisen, H. (1998). Do funded pensions contribute to higher savings? A crosscountry analysis. Weltwirtschaftliches Archiv/Review of World Economics, 134, 692–711. Boskin, M. (1978). Taxation, savings and the rate of interest. Journal of Political Economy, 86, S3–27. Corbo, V., & Schmidt-Hebbel, K. (1991). Public policies and saving in developing countries. Journal of Development Economics, 36, 89–115. Dayal-Ghulati, A., & Thimann, C. (1997). Saving in Southeast Asia and Latin America compared: Searching for policy lessons. IMF Working Paper, WP/97/110. Deaton, A. (1990). Savings in developing countries: Theory and review. Proceedings of the First Annual World Bank Conference on Development Economics, Washington, DC, 61–96. Edwards, S. (1996). Why are Latin America’s savings rates so low? An international comparative analysis. Journal of Development Economics, 51, 5–44. Giovannini, A. (1983). The interest elasticity of savings in developing countries: The existing evidence. World Development, 11, 601–607. Haque, N. U., Peseran, M. H., & Sharma, S. (1999). Neglected heterogeneity and dynamics in cross-country savings regressions. IMF Working Paper, International Monetary Fund, Washington, DC, May. Higgins, M., & Williamson, J. (1996). Asian demography and foreign capital dependence. NBER Working Paper No. 5560, Cambridge, MA: National Bureau of Economic Research. Japelli, T., & Pagano, M. (1995). Saving, growth, and liquidity constraint. Quarterly Journal of Economics, 109, 83–109. Lahiri, A. (1989). Dynamics of Asian saving. Staff Papers, International Monetary Fund, Paper No. 36, 228–261. Loayza, N., Lopez, H., Schmidt-Hebbel, K., & Serven, L. (1998b). The world saving database. Word Bank Manuscript. Washington, DC: The World Bank.
Determinants of Private Savings in the Middle East and North Africa
113
Loayza, N., Lopez, H., Schmidt-Hebbel, K., & Serven, L. (1998b). Saving in the world: Facts, trends, correlations. Manuscript presented at the World Bank Conference on Saving in the World: Puzzles and Policies. Washington, DC: The World Bank. Loayza, N., Schmidt-Hebbel, K., & Serven, L. (2000a). What drives private savings across the world. Review of Economics and Statistics, 82, 165–181. Loayza, N., Schmidt-Hebbel, K., & Serven, L. (2000b). Saving in developing countries: An overview. World Bank Economic Review, 14, 393–414. Lucas, R. E., Jr. (1988). On the mechanics of economic development. Journal of Monetary Economics, 22, 3–42. Masson, P. R., Bayoumi, T., & Samiei, H. (1998). International evidence on the determinants of private saving. World Bank Economic Review, 3, 483–501. McKinnon, R. (1991). The order of economic liberalization: Financial control in the transition to a market economy. Baltimore, MD: Johns Hopkins University Press. Modigliani, F. (1970). The life cycle hypothesis of saving and intercountry differences in the saving ratio. In: W. Eltis, M. Scott & J. Wolfe (Eds), Induction, growth and trade: Essays in Honour of Sir Roy Harrod. London: Oxford University Press. Muradoglu, G., & Taskin, F. (1996). Differences in household savings behaviour: Evidence from industrial and developing countries. The Developing Economies, 34, 138–153. Ostry, J., & Reinhart, C. (1992). Private saving and terms-of trade shocks: Evidence from developing countries. International Monetary Fund Staff Papers Paper No. 39, 495–517. Romer, P. (1986). Increasing returns and long run growth. Journal of Political Economy, 94, 1002–1037. Schmidt-Hebbel, K., Serven, L., & Solimano, A. (1996). Savings and investment: Paradigms, puzzles, policies. World Bank Research Observer, 11, 87–117. Solow, R. (1956). A contribution to the theory of economic growth. Quarterly Journal of Economics, 70, 65–94.
This page intentionally left blank
114
PART II: COMMERCIAL BANKING
115
This page intentionally left blank
116
PROFITABILITY AND RISK MANAGEMENT IN BANKING: A COMPARATIVE ANALYSIS OF EGYPT AND LEBANON Sam Hakim and Simon Neaime ABSTRACT We investigate the profitability and risk management in banking in two prominent countries in Middle East and North Africa (MENA), Egypt and Lebanon, where banks operate under market-oriented economic regimes. The study covers the 1990s which witnessed banking sector reforms towards a more efficient financial system. Noting the differences in the structure of the banking system and the monetary changes in Egypt and Lebanon, we investigate the impact of liquidity, credit, and capital on bank profitability in each country’s banking sector. Based on our findings, we draw conclusions on the strength of risk management practices and enforcement of banking regulations.
1. INTRODUCTION Securities markets in the Middle East and North African (MENA) region have traditionally been less important in channeling financial funds. A fairly Money and Finance in the Middle East: Missed Opportunities or Future Prospects? Research in Middle East Economics, Volume 6, 117–131 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1094-5334/doi:10.1016/S1094-5334(05)06006-1
117
118
SAM HAKIM AND SIMON NEAIME
developed commercial banking system has taken the lead in attracting and channeling funds. The banking system has also played a prominent role in debt finance and financial intermediation. A sound banking system with good performance indicators necessitates sound risk management and regulatory frameworks. Assessing the overall performance of banks requires looking at both efficiency measures and risk-taking behavior. However, with the increasing trend towards financial integration in the MENA region, risk and banking management practices are still not up to par. In many instances risk is not accounted for, and banks suffer from operational and other inefficiencies. Considerable attention in the finance literature has been given lately to the concept of bank efficiency and the relationship that exists between economies of scale and scope, and the profitability of commercial banks. Benston et al. (1982) was among the first to have tested for economies of scope in U.S. banks. He also tried to test complementarities between different outputs of bank services. Mester (1996) argues that because bank efficiency models fail to account for risk-taking behavior, their results may be misleading. To overcome this disadvantage, his analysis incorporates the risk profiles of banks by including proxies for asset quality in the profit function. A similar finding is reinforced by Berger and Mester (1997). Empirical evidence in the U.S. (Federal Reserve Bulletin, 1999; 2000) suggests that cost savings does not necessarily justify increasing bank scale through controlling entry or encouraging mergers. In addition, small banks mergers should be encouraged over larger banks on the basis that operating costs are kept at a lower level. This holds whether simple accounting ratios are compared pre- and post-merger, holding industry effect constant, or in more sophisticated econometric analysis, using frontier cost functions. More recent evidence is available in Jagtiani and Khanthavit (1993), and Kwast (1995). Against this background, we investigate the performance and risk in two prominent countries in MENA, Egypt and Lebanon, where banks operate under market-oriented economic regimes. The study covers the 1990s, which witnessed banking sector reforms towards a more efficient financial system. Noting the differences in the structure of the banking system and the monetary changes in Egypt and Lebanon, we investigate the impact of liquidity, credit, and capital on profitability in each country’s banking sector. In both Egypt and Lebanon banks can be divided into three categories: (1) commercial banks, (2) business and investment banks, and (3) specialized banks for real estate, industry, and agriculture. Based on our findings, we draw
Profitability and Risk Management in Banking
119
conclusions on the strength of risk management practices, corporate governance, and enforcement of banking regulations. In addition in Egypt, Non-Bank Financial Intermediaries (NBFIs) comprise a mixed bag of institutions, ranging from leasing, factoring, and venture capital companies to various types of contractual savings and institutional investors (pension funds, insurance companies, and mutual funds). The common characteristic of these institutions is that they mobilize savings and facilitate the financing of different activities, but they do not accept deposits from the public. To remain consistent across countries, we focus in this chapter only on commercial banks in the case of Lebanon and Egypt.
2. EGYPT BANKING SYSTEM The Egyptian pound in recent years has been informally pegged against the US$, and has been technically over valued. One of the key issues the Egyptian government decided to tackle recently is the value of the pound and the question of a steadfast exchange rate mechanism. In 1999, Egypt’s foreign exchange reserves fell but remained at a comfortable level. Reserves stood at US$17.6 billion by August 1999, down from US$20 billion at the beginning of that year. Nevertheless, this gave Egypt sufficient reserves to cover nearly 12 months worth of imports. The fall in reserves though highlighted a lack of flexibility in dealing with market demands at the Central Bank of Egypt (CBE). More recently, foreign reserves have continued their downward trend. By June 2000, the Central Bank’s balance of foreign currency reserves had fallen to US$15.1 billion. In January 2001, after a series of liquidity problems and run on the US$, the Egyptian pound experienced high volatility on the exchange rate. In response to the lingering liquidity crisis, which has an adverse effect on the foreign exchange market, the government reiterated its belief in the effectiveness of the regulated market forces, and of all rights guaranteed by the Foreign Exchange Law. As a result, the Central Bank implemented a number of measures to restore stability of the market and sustain the confidence of market participants. The main measures consist of devaluing the Egyptian pound to LE 3.85 per US$, allowing banks and foreign exchange dealers to set their buying and selling rates for the US$ (bank notes and transfers) within a band of 71%. All other currency rates shall be determined according to the cross rates internationally. This measure was followed by the lowering of the discount rate in April 2001, from 11.5% to
120
SAM HAKIM AND SIMON NEAIME
Table 1.
Macroeconomic Growth Rates (Lebanon).
Year 1991/1992 1992/1993 1993/1994 1994/1995 1995/1996 1996/1997 1997/1998 1998/1999 1999/2000
GDP
Unemployment
1.9 2.5 3.9 4.7 5 5.3 5.7 6.1 6.5
9.2 10 9.8 9.6 9.2 8.8 8.3 7.9 7.4
Source: IMF and ESCWA.
11% a continuation of the reduction of the discount rate, which had been 12.25% 3 years earlier. Meanwhile, inflation rate continued to decline thus proving the success of the monetary policies adopted by the government, which took suitable and effective measures for activating banking operations and making domestic liquidity available in the banking system. In addition, real growth rate continued unabated, currently at 6.5% from a trough of 1.9% during the recession in 1991–1992. In parallel fashion, as Table 1 reveals, the unemployment rate pursued a steady decline to a low of 7.4%. Amid this changing environment, Egyptian banks pursued a policy of effective participation in pumping the necessary finance into the various economic sectors while attempting to diversify their activities, to include non-traditional services, in order to be able to apply universal banking. This strategy moved in line with the state’s tendency towards bolstering the capital market through increasing its contributions to investment projects on one hand, and revolving their portfolio in order to achieve the stock market’s equilibrium on the other. Moreover, the main Egyptian banks continued their gradual implementation of the state’s plan for privatizing the public stake in banks and joint ventures. Many banks also pursued efforts by applying modern technologies, such as computerization in order to upgrade the quality of their services. While the payback from these efforts is still long term, the immediate impact has not produced the positive effect it was expected. For example, loan and deposit growth for commercial banks have slowed to 12% and 8%, respectively (see Table 2). The trend is not particular to commercial banks alone but is pervasive covering also the specialized and investment banks, suggesting the need for a macroeconomic solution.
121
Profitability and Risk Management in Banking
Table 2.
Growth Rates in Egyptian Banking Activities (%).
Commercial Banks Loans –7 23 20 37 22 17 12 18 12
Specialized Banks
Investment Banks
Deposits
Loans
Deposits
Loans
Deposits
21 19 7 10 10 15 7 8 8
1 –2 14 24 18 19 20 19 7
20 19 38 30 30 12 30 24 9
5 5 10 22 18 24 16 20 11
–8 4 9 34 18 12 14 23 20
Source: Central Bank of Egypt.
3. LEBANON’S BANKS: A BACKGROUND Prior to 1975, the Lebanese government passed many legislative decrees, which enhanced the banking industry’s domestic and international stature. These included, the Banking Secrecy Law of 1956, the Free Convertibility Law of 1952, and the introduction of Joint Accounts in 1961. These attracted Arab funds and made Lebanon the main financial center of the Arab region. Despite 17 years of civil turmoil, the Lebanese banking sector was able to provide basic banking services without major disruptions. Still, the banking system experienced many difficulties and obstacles, which hindered its activities. For example, the 1988–1992 depreciation of the Lebanese pound (from LP 3 per dollar in 1988 the Lebanese Pound depreciated to LP 2300 per dollar in 1992), the dollarization of deposits and the subsequent deterioration of the banks’ capital base, the decrease in performance of human capital and expertise, the withdrawal of foreign banks from the domestic market, and the 1989–1991 crisis of ‘‘Problem Banks.’’ This crisis led to 19 bank mergers and the trend is expected to continue in the future (see Table 3). During the period 1992–1998, growth in the banking sector activities exceeded growth in gross domestic product (GDP). Total banks deposits amounted to twice the level of GDP (the 1999 estimates put it at US$ 15 billion). This is a clear reflection of the severe disturbances affecting the country’s economic situation. The most serious are: (1) increasing government debt financed at high interest rates and (2) better performing banks’ balance sheets and higher profitability.
122
SAM HAKIM AND SIMON NEAIME
Table 3.
Lebanese Mergers and Acquisitions 1993–1999.
Year 1993 1994 1995 1997
1998
1999 (under consideration)
Merged Bank Globe Bank Capital Trust First Phoenician Security Bank of Lebanon Geagaa Bank of Lebanon and Pakistan CCMO BLC Tohme Nasr Lebano-Africain Adcom Bank Orient Credit Universal Bank Foreign Trade Transorient Unibank Al Moghtareb Bank Litex Intercontinental Bank
Merging Bank Societe General Credit Libanais BLC Societe General Inaash Bank Audi Bank Byblos Bank Fransabank Audi Bank Audi Bank Audi Bank Fransabank Al-Madina Bank of Beirut UBL
Orbil
Source: Lebanese Banker’s Association. BLC: Beirut pour le Commerce; CCMO: Credit Commercial pour le Moyen Orient; UBL: United Bank of Lebanon.
During this period, Lebanese banks devoted substantial efforts to enhance their capabilities and operations to regain their prominent role in the region. Efforts were devoted to enlarge the capital base, increase resources and activities, create new retail products, invest in new technologies, recruit qualified personnel and train already existing ones. Those efforts were productive and private funds witnessed a 13-fold increase; deposits grew at a rate of 282% and loans increased by 302% during the period 1992–1998 (see Table 4). Lebanese banks also issued Global Depository Receipts (GDR) and Eurobonds on the international capital markets. These issues amounted to US$ 1.3 billion in June of 1998. In addition, banks introduced new savings products and financing schemes. These range from joint lending schemes, personal loans (housing, car, consumer, etc.), leasing, to investment packages as well as various saving schemes. Other specialized financing schemes were also introduced. First, IFC loans amounting to US$ 270 million of
123
Profitability and Risk Management in Banking
Table 4.
Main Features of the Lebanese Banking Sector 1992–1998 (Billions of LL).
Deposits Foreign deposits (US$ million) Total deposits Deposits/total deposits (%) Loans Loans/deposits (%) Private funds Total assets
1992
1993
1994
1995
1996
1997
1998
3697 4555
4744 6462
7844 7593
8994 9330
13,351 11,169
13,936 16,132
15,897 20,031
12,068 15,801 30.60 30
20,350 23,885 30,685 38,570 46,113 38.50 37.70 43.50 36.10 34.50
4804 6217 8169 10,695 13,290 16,022 19,300 39.80 39.30 40.10 44.80 43.30 41.50 41.90 264 444 676 1146 1943 2990 3620 14,634 18,809 24,285 29,055 37,183 45,633 55,031
Growth of the Banking Sector 1992–1998 (%) 1992/ 1993/ 1994/ 1993 1994 1995 Deposits Foreign deposits Total deposits Loans Private funds Total assets
28.3 41.9 20.8 29.4 68.2 28.5
65.3 17.5 5.9 31.4 52.3 29.1
14.7 22.9 18.7 30.9 69.5 19.6
1995/ 1996 48.4 19.7 33.8 24.3 69.5 28.0
1996/ 1997 4.4 44.4 22.6 20.6 53.9 22.7
1997/ 1998 14.1 24.2 19.5 20.5 21.1 20.6
1992/ 1998 330 340 282 302 1273 276
Source: Lebanese Banker’s Association.
which US$ 100 million as mortgage loans and the balance to finance productive sectors. Second, the European Investment Bank (EIB) provided euro 30 million to finance the construction of hotels outside the capital (see Table 5). Also, some banks provided fiduciary loans under the Inter-Arab Trade program. Third, the government has decided to disburse some of its dues of around LP 1130 billion to the private sector. This money, it is hoped, will stimulate productivity and create jobs. And, through subsidizing interest rates (5%) on industrial loans in LP the government envisages to encourage investments in these sectors. A Credit Insurance Agency – that would increase the customer base of creditors for small- and medium-sized companies – is currently being formulated. The payment system of banks was also enhanced through first the establishment of a clearing house for commercial chapters (in addition to the clearing of checks in LP and foreign currencies). Second, the Central Bank support for establishing ATM networks. Third, the establishment of an up-to-date information system that facilitates the
124
Table 5.
SAM HAKIM AND SIMON NEAIME
Foreign Issuing of Lebanese Banks (Millions of Dollars, as of 6/30/1998).
Eurobonds Certificates of deposits Global depository receipts Subordinated loans Total Euro-loans IFC EIB
210 675 262 150 1297 270 35
Source: Central Bank of Lebanon.
Table 6.
Structure of the Lebanese Banking Sector.
Number of operating bank Number of branches Representative offices Financial institutions Number of employees Number of ATMs
1992
1998
73 512 11 7 14,075 0
77 685 11 21 14,922a 205
Source: Lebanese Banker’s Association. a 1997.
collection of the government revenues (collection of Treasury funds as well as telephone, electricity, and car maintenance bills – see Table 6). Representing 9% of the GDP, the banking system continues to hold a considerable position in the economy, financing on average 45% of aggregate demand and 50% of imports while deposits amount to 90% of gross domestic savings. During the recent recession dates, the banking industry in Lebanon has been among the few resilient factors. Banks are conservatively managed, with loans representing a mere 35% of assets as the industry is strictly controlled by the Central Bank, which imposes strict regulatory ratios for risk management and control. For example, foreign exchange positions are forbidden, political loans do not exist, credits are strictly monitored and classified, related party transactions are restricted and subject to timely and accurate disclosure. Since 1998, all banks in Lebanon have abided by the International Accounting Standards. While such environment allowed major players to expand their market shares, the banking industry has experienced a relative decline since 1999 mainly as a result of a
Profitability and Risk Management in Banking
125
contraction in private and public demand, and a shrinking of net interest margins. Amidst this change, the pace of mergers and acquisitions has increased.
4. EMPIRICAL INVESTIGATION AND METHODOLOGY Our data consist of annual observations from 43 Lebanese and 62 Egyptian banks between 1993 and 1999. For Lebanon, the data was obtained from Bilan Banques, which tracks the financial statistics of Lebanese financial services institutions, as reported to the Lebanese Bankers Association. In the case of Egypt, the data is provided from the CBE. As our data contains information on cross-sectional units observed over time, a panel data estimation technique is adopted. This allows us to perform statistical analysis and apply inference techniques in either the time series or the cross-section dimension. The model takes the form yit ¼ ai þ bit xit þ uit
(1)
where i ¼ 2 cross sections (Lebanon and Egypt) and t ¼ 7 (1993–1999); yit is the return on equity (ROE) and represents bank profitability; and xit is a vector of three independent variables that were constructed to measure liquidity, credit, and capital. They have been selected on the basis of their potential relevancy to this model, and because of their importance in depicting a particular bank’s real financial position. Some of the independent variables will vary over time and cross sections, whereas others will only vary across sections. While the error terms are serially correlated for k41; they are independent of the regressors; that is E½ui;tþk;k X i;t ¼ 0: The residual covariance matrix for this set of equations is given by: O ¼ Eðuu0 Þ ¼ s2 I N I T
(2)
The intercept ai varies across banks to capture the specific effects for each country. White heteroscedasticity-consistent standard errors and covariances are computed. Greene (2000) details the estimation methodology. In what follows we discuss the three explanatory variables of our model. In order to measure liquidity we constructed a variable to measure the possibility that depositors will withdraw deposits in excess of deposits. Depositors may have good incentives to withdraw their deposits while sufficient funds remain in the bank, if they feel it is encountering financial difficulties. Moreover, a bank’s deteriorating liquidity position is an
126
SAM HAKIM AND SIMON NEAIME
important indicator of potential bankruptcy, and whether it is solvent or not is another issue. The liquidity variable employed in our model takes into account these factors, and shows a bank liquid assets as a percentage of its issued liabilities, it is defined as the total interest bearing liabilities over the total liquid assets. These consist of cash with banks, plus securities, which are the short-term Treasury-Bills. The liquidity variable is defined by X1 ¼
savings þ checking þ current þ fixed deposits þ banks cash with banks þ securities
(3)
The higher this ratio, the higher will be the liquidity risk, and the less the coverage of its liabilities, other things being equal. Savings accounts are defined as deposits that pay interest and have maturities of 1 week or longer. Withdrawals are allowed through only the saving book. Checking accounts are deposits that have no maturity, and withdrawals are done through checks. Current accounts are similar to saving accounts but they are subject to an income tax. Fixed deposits are deposits blocked for a period of time that pay a higher interest rate. Banks and correspondents are placement accounts that a bank opens with its correspondent bank. They are of two types: (1) Call Accounts, with 2 days notice before withdrawal, and (2) Blocked Accounts with longer maturities. Credit factor (CF): It represents the likelihood that borrowers will not repay their loans as promised. Assessing and pricing credit risk, and determining the appropriate amount of collateral is the concern of the bank’s management. Loan management is an art and not a science, and failing to apply appropriate loan policies may either be related to lack of expertise and training programs (i.e. poor management), or to fraudulent activities, such as the concentration of loans to friends, relatives, or associates. The ratio will consider the quantity of loans, and shows how much the bank has formed provisions to withstand default loans: X2 ¼
advances þ bills þ total contra accounts total assets
provisions
(4)
Advances are loans given by an individual bank to both the private and public sectors. Provisions are liability accounts formed as reserves for potential or actual losses emanating from bad or substandard loans. The contra accounts are financial services that generate income and create credit risk (e.g. letters of credits and lines of credits). Bills represent discounted securities purchased by banks. When provisions increase, the bank will be in a better position to withstand default on loans, and therefore has a better credit policy. The way the loan to asset ratio is calculated shows some
Profitability and Risk Management in Banking
127
deficiencies, since credit risk is in reality more a qualitative then quantitative measure. Capital adequacy (CA): A bank becomes insolvent, when it has a negative net worth, defined as the difference between the market value of its assets and liabilities. Capital serves many functions. First, any new entry to the banking system must pay its price of least capital requirements. Second, capital provides a bank with the required funds to service any new desired asset expansion and growth purposes. In the case of Lebanon, after the devaluation of the domestic currency, which started in 1988, most Lebanese banks became undercapitalized. Ratios of capital adequacy have shown high levels of inadequate capital. Whether they are true indicators of a bank’s capital position is questionable. Efforts, however, have been devoted to fix the levels of adequate capitalization to serve its most important functions mainly that of being a cushion of own funds formed to absorb losses incurred by a bank. Capital adequacy is measured by: X3 ¼
prior profits þ reserves þ capital assets þ contra accounts
(5)
Prior profits are carried forward from banking operations, which management intends to reinvest in the bank. Reserves constitute a separate account in which the profit has been reinvested. Capital is the initial price of entry that a bank has to pay to enter to the banking system. Other things equal, the higher the ratio (i.e. the equity of a bank), the higher is the ability of a bank to absorb external shocks or unforeseen losses.
5. EMPIRICAL RESULTS For profitability analysis, it is important to look at capital adequacy in two dimensions (across banks and over time) and distinguish explicitly between Egypt and Lebanon. This would allow us to separate the effects particular to each country. In particular, because bank capitalization was effected by variables specific to Lebanon, and beyond a bank’s own control, we examine the impact of capabilities on profitability over time and across countries. The other variables (liquidity and credit) will be limited to vary over time. These represent bank-specific variables and should reflect the direct activities a bank is undertaking. Our results are divided between fixed (variability over time), and non-fixed effects (over time and across countries). The results are provided in Table 7.
128
SAM HAKIM AND SIMON NEAIME
Table 7.
Cross Section and Times Series Analysis.
Dependent Variable: ROE Time Period: 1993–1999 – 2 cross sections yit ¼ ai þ bit xit þ uit Variable Liquidity Credit Capital adequacy – EGY Capital adequacy – LEB Fixed effects Intercept – EGY Intercept – LEB Adjusted R-squared Durbin–Watson stat
Coefficient
Standard Error
0.13 3.99 42.34 6.58
0.07 1.63 23.45 2.30
t-Statistic 1.68 2.45 1.81 2.86
0.035 0.733 73% 1.16
Source: Authors’ estimates. White heteroscedasticity-consistent standard errors and covariance. ,significance at 5% and 10%, respectively.
The intercept represents the basic return on bank equity for each country during this time period, in the absence of any regressors. The bank-specific effects include liquidity, credit, and capital adequacy. Turning to the analysis of the country-specific variables, we notice that the coefficient on the capital adequacy ratio is strong and significant. The sign is negative for both Egypt and Lebanon indicating that higher adequacy has an adverse effect on ROE. Generally, a higher capital adequacy ratio may lead to a higher total profit in absolute terms but not necessarily a higher (percentage) ROE. In fact, evidence reveals that a highly capitalized institution represented by a high capital to asset ratio, will have a tougher time producing lofty return to its shareholders. This is because capital represents a sunk cost. While it serves as a cushion against unforeseen events, it also bears a cost and ultimately holds back the institution from delivering the high percentage returns other under-funded banks may be able to produce. A similar argument is available by Kimball (1998). In terms of absolute levels, the coefficient on the capital adequacy variable is stronger for Lebanon and Egypt. This is expected, since Lebanese banks have suffered a larger attrition to their capital base making their profit returns relatively more significant than in Egypt. This is evidenced when we compare directly the ROE in Egypt vs. Lebanon as we do in Table 8. This table shows a significant slowing in profitability for Lebanese banks, a factor attributed primarily to a regression of Lebanese interest rates, which
129
Profitability and Risk Management in Banking
Table 8.
Egypt Lebanon
ROE Comparison (%) for Commercial Banks.
1993
1994
1995
1996
1997
1998
1999
4.0 55.3
16.2 58.1
7.1 75.0
4.1 74.4
7.7 56.4
8.1 22.6
6.9 11.1
Source: Bilanbanque, 1992–1999 and Central Bank of Egypt.
peaked during the period 1995–1996, when they reached a high of 38%. At that time, the high rates were a direct result of a classic crowding-out effect. Since, Lebanese interest rates have been on a steady decline (currently hovering around 11%), forcing Lebanese banks to accept more normal profits. Turning to the fixed effects, our results suggest that the credit variable is a good predictor of profitability across all banks. The sign of the credit variable is positive and significant suggesting that bank profitability is positive and a direct function of lending activities. A high ratio of loans to assets (net of any provisions for doubtful accounts) indicates additional risk taking on the part of a bank and therefore would translate into a higher ROE, to the extent that a higher calculated risk leads to a higher return. These results apply to banking in general irrespective to one country. Finally, turning to the liquidity variable, our results reveal that it is insignificant across all banks and seems to have no discernable impact on profitability.
6. CONCLUSIONS AND POLICY IMPLICATIONS In banking theory, the profit concept assumes that banks are price takers in input markets and can set output prices. The assumption is reasonable under the oligopolistic market structures of the banking systems in Egypt and Lebanon and the deregulated service-fees structure. The profit concept also assumes that all banks use the same technology. This simplifying assumption may seem a strong one since foreign banks are likely to use superior technology compared to local banks. Following the currency devaluation in the late 1980s and early 1990s, Lebanese banks suffered a major blow to their capital base. A similar argument applies to bank efficiency because the Lebanese strife forced many banks to delay investment in automation and information technology. As a result, the period prior to 1991 (when the Hariri government took office) is radically different from the period we analyze here and therefore is another
130
SAM HAKIM AND SIMON NEAIME
justification for ignoring it within this study. In a parallel fashion, in 2000 the Egyptian economy suffered a liquidity crunch, which ultimately led to the devaluation of the Egyptian pound in January 2001. This period is also not included in our analysis. As a result, our analysis should produce more meaningful comparison between Lebanon and Egypt. As strong similarities exist between the two countries’ financial services industries, we introduced a cross section and time series analysis by combining Egyptian and Lebanese banks into one single data set. The technique allows us to maximize statistical inference in a case where time series data were not available over a long period and had long frequency (annual data). In conducting panel data estimation, we allow some coefficients to vary only across time, while others were allowed to vary across countries and over time. In order to assess the overall bank performance in each economy, we specify and estimate a fixed effects model allowing only the intercept to vary over time and by country. Using data for empirical investigation from the financial statements of banks over the period 1993–1999, we specified and estimated a fixed effects model of bank returns with varying intercepts and coefficients. Our results show that ROE in banking is a direct and an increasing function of the bank’s lending activities, irrespective of the country. Regarding countryspecific effects, we find a strong link between capital adequacy and commercial bank return, with high capitalization acting as a hindrance to return. The negative impact is attributed to the fact that capital is a sunk cost, with large banks realizing high profits in absolute but not in percentage terms. The effect is stronger for Lebanon than Egypt, because Lebanese banks are relatively less capitalized than their Egyptian counterparts, a factor largely due to the severe attrition to capital Lebanese banks have experienced in the wake of the steep currency devaluation. Meanwhile, bank liquidity which represents a risk and solvency indicator seems irrelevant to bank profitability. In total the model explains about 74% of the variability of return on bank equity over time and across countries. These results should be used as a precursor to a more elaborate model which distinguishes between bank categories in each country carefully taking into account the type of activities of each. Our findings are expected to help policymakers in the MENA region set better performance targets, and enable bank managers allocate capital more efficiently across their business units. Our results will be useful to identify how commercial banks can better employ their current capital and evaluate their future performance.
Profitability and Risk Management in Banking
131
ACKNOWLEDGMENTS Financial Support from the University Research board of the American University of Beirut is greatly acknowledged. The authors are grateful to two anonymous referees for valuable comments and suggestions on an earlier draft.
REFERENCES Benston, G., et al. (1982). Scale economies in banking. Journal of Money, Credit and Banking, 14(2), 435–454. Berger, A., & Mester, L. (1997). Inside the black box: What explains differences in the efficiencies of financial institutions? Journal of Banking and Finance, 21, 895–947. Federal Reserve Bulletin. (1999). Profits and balance sheet developments at U.S. Commercial Banks in 1998, 85(6), 367–369. Federal Reserve Bulletin. (2000). Profits and balance sheet developments at U.S. Commercial Banks in 1999, 86(6), 364–367. Greene, W. (2000). Econometric analysis. New Jersey: Prentice-Hall. Jagtiani, J., & Khanthavit, A. (1993). Scale and scope economies at large banks: Including off balance sheet products and regulatory effects (1984–1991). Journal of Banking and Finance, 20, 243–256. Kimball, R. (1998). Economic profit and performance measurement in banking. New England Economic Review, 35. Kwast, M. (1995). Panel discussion of US banking consolidation and efficiency: Myths and reality. Journal of Banking and Finance, 17, 76–82. Mester, L. (1996). A study of bank efficiency taking into account risk preferences. Journal of Banking and Finance, 20, 1025–1045.
This page intentionally left blank
132
AN ANALYSIS OF MERGERS AND ACQUISITIONS IN THE TURKISH BANKING SECTOR Ays-e Mumcu and E. U¨nal Zenginobuz ABSTRACT This chapter explores various aspects of mergers and acquisitions in the banking industry within a simple model that allows explicit comparison of sector performance before and after the mergers and acquisitions. The industry structure we look at involves a few dominant banks and a competitive fringe, which we take as the structure most likely to resemble the Turkish banking industry in the aftermath of the ongoing restructuring process. Using a reasonable set of parameters to simulate the model, we perform comparative statics exercises regarding the impact of mergers among domestic as well as with foreign banks on equilibrium outcomes.
1. INTRODUCTION The Turkish economy has been hit by two very severe crises recently, namely those of November 2000 and February 2001. These crises caught the Turkish banking sector in the middle of a sweeping restructuring process, which
Money and Finance in the Middle East: Missed Opportunities or Future Prospects? Research in Middle East Economics, Volume 6, 133–162 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1094-5334/doi:10.1016/S1094-5334(05)06007-3
133
134
AYS - E MUMCU AND E. U¨NAL ZENGINOBUZ
was one of the critical components of the comprehensive disinflation program Turkey adopted at the beginning of 2000. The disinflation program involved tight fiscal and monetary policies, large-scale structural reforms, and a pre-determined exchange rate policy to serve as a nominal anchor in reducing inflation from its chronically high levels. In regard to the banking sector, the disinflation program of 2000 foresaw a revamping of the legal and regulatory framework for banking supervision in accordance with European Union and world standards, correcting the weaknesses in the private banking system, and finally the restructuring and privatizing of the state banks. Toward this end, the powers of the independent Banking Regulatory and Supervisory Agency (BRSA), which was established as part of a new banking law in June 1999, were further strengthened through a series of amendments approved in the Turkish Parliament in December 1999. As the single regulatory and supervisory agency to oversee the sector, the BRSA has completely independent jurisdiction over the entry and exit of banks and over changes to the regulatory framework. Prior to the disinflation program, the Turkish banking industry had faced distorted incentives due to the chronically high and erratic inflation levels, which went together with increasing government deficits during the previous decade. Excessive and persistent public sector borrowing requirements led to very high real returns to government issued securities, alluring some private banks to accumulate asset portfolios that were far from sound. Together with the slackening of entry requirements to the sector and the overall weakness of the regulatory framework, this environment contributed to the fragmentation of the banking sector into small banks. A significant number of small banks carrying weak asset portfolios became insolvent over time, and had to be transferred to the Savings Deposit Insurance Fund (SDIF).1 The first banking crisis in December 2000, as well as the one that followed the financial crisis in February 2001, can be seen as the culmination of persistent distortions in the Turkish banking sector. The new and tougher regulatory framework, together with the new macroeconomic environment that was to arise from the disinflation program of 2000, was expected to lead to significant changes in the Turkish banking sector. This included the increased likelihood of significant mergers and consolidation in the sector. Such a process was also expected to increase the presence of foreign banks in Turkey through acquisitions. Both of these developments have gained increased momentum after the already ailing Turkish banking sector was further hit by the November 2000 and the February 2001 crises.
An Analysis of Mergers and Acquisitions in the Turkish Banking Sector
135
Understanding the various impacts of mergers and acquisitions requires the modeling of the banking industry structure and behavior of banks. The objective of this chapter is to explore various aspects of mergers and acquisitions within a simple, albeit rigorous, model that allows for explicit comparison of sector performance before and after mergers and acquisitions. The model studies the strategic interaction among commercial banks in an imperfectly competitive banking industry, and allows us to consider the determinants of feasibility and desirability of mergers and acquisitions among domestic as well as foreign banks. Given the aims of the ongoing fundamental reform process in the Turkish banking sector, we abstract from many of the features that are currently relevant and concentrate on a sector structure that is expected to arise after the proposed restructuring is successfully completed. This amounts to looking at a banking industry in a stabilized economy, in which all public banks are privatized and banking regulations are in place and fully enforced. The main features of the stylized model we develop in this chapter can be summarized as follows: Adopting an industrial organization approach, we consider an imperfectly competitive banking industry. Banks are assumed to act as independent entities that react strategically to their environment where deposits received, loan amounts serviced, and net foreign position are viewed as strategic variables. The industry structure we look at involves a few dominant banks with a (competitive) fringe, which we take as the structure most likely to resemble the Turkish banking industry in the aftermath of the ongoing restructuring process. For a given as well as a changing number of firms, we first study the features of equilibrium outcomes, such as the size of banking industry, interest rates (deposit, loan, and interbank rates), etc. After determining the set of parameters that fairly represents the Turkish banking sector, we perform various comparative statics exercises regarding the impact of mergers among domestic as well as with foreign banks. Other comparative statics exercises that we carry out include the impact of deposit insurance, reserve requirement ratios, the country’s international credit rating, and the global economic conditions. The plan of the chapter is as follows: Section 2 provides a quick glance at the Turkish banking industry structure and summarizes the developments that have rendered it vulnerable to crises of November 2000 and February 2001. In Section 3 we present the model we study. Section 4 presents the behavior of the dominant group and the competitive fringe. Section 5 presents the equilibrium. Section 6 presents the empirical application. Discussion and conclusions follow in Section 7.
AYS - E MUMCU AND E. U¨NAL ZENGINOBUZ
136
2. A QUICK GLANCE AT THE TURKISH BANKING INDUSTRY STRUCTURE2 At the end of September 2001, there were 68 banks operating in Turkey. As of that date, the total assets in the Turkish banking sector amounted US$110.7 billion. Deposits stood at US$68.8 billion for the whole industry, while the total amount of loans extended by the industry equaled US$28.6 billion. Further data on the Turkish banking sector as of September 2001 is presented in Table 1. Of the 68 banks, 52 were commercial (deposit money) banks while the rest were investment and development banks. Of the 52 commercial banks operating in Turkey, 24 were privately owned domestic firms, 16 were Table 1.
Turkish Banking Sector (September 30, 2001).
Total Total Share in Share in Number Total Share in of Banks Assets (US$ Sector Loans (US$ Sector Sector Deposits (US$ million) (%) million) (%) million) (%) Total
68
110,732
100.0
28,642
100.0
68,796
100.0
Commercial banks State owned Privately owned – Five largest banks – Other Foreign owned SDIF managed
52
105,402
95.2
26,338
92.0
68,796
100.0
3 24
30,101 59,403
27.2 53.6
5,322 19,240
18.6 67.2
21,430 38,946
31.2 56.6
40,005
36.1
15,127
52.8
26,568
38.6
16
19,398 2,933
17.5 2.6
4,112 681
14.4 2.4
12,379 978
18.0 1.4
9
12,965
11.7
1,096
3.8
7,441
10.8
16
5,330
4.8
2,304
8.0
0
0.0
3 10
4,347 881
3.9 0.79
1,868 400
6.5 1.4
3
102
0.09
36
0.1
Development and investment banks State owned Privately owned Foreign owned
Source: The Banks Association of Turkey (2000), BRSA (2001).
An Analysis of Mergers and Acquisitions in the Turkish Banking Sector
137
privately owned foreign banks, while 9 were banks under the management of SDIF. As of September 2001, the total assets of the 24 private domestic banks constituted 53.6% of the total assets of the Turkish banking sector. The share of state banks in total assets stood at 27.2%, while foreign banks’ share of total assets was 2.6%. A cursory look at the evolution of the Turkish banking sector from early 1980s until the November 2000 crisis reveals that the sector has expanded significantly during that period. The number of banks increased from 43 in 1980 to 66 in 1990 and to 79 by the end of 2000. Total employment in the banking sector also increased considerably during the 1980–2000 period: Starting from 125,000 in 1980, it went up to 154,000 in 1990, and reached 170,000 in 2000. However, since the end of 2000 the number of employees has decreased by almost 15% to 147,453 in the first 9 months of 2001, indicating the extent of the shock received by the industry.3 Table 2 displays the changes in various financial indicators for the Turkish banking sector for the 1980–2000 period. Total assets of the Turkish banking sector increased from US$20.8 billion (28.6% of gross national product, GNP) in 1980, to US$58.2 billion (38.2% of GNP) in 1990, to US$155 billion (76.9% of GNP) in 2000. The share of state banks in banking sector total assets decreased from 4.1% in 1980 to 34.9% in 2000. Table 2. Turkish Banking Sector: Financial Indicators. In US$ Million (Unless Otherwise Stated) Total assets Total credits Securities portfolio Total deposits Savings deposits –TL –FX Non-deposit funding –Foreign banks Networth+profits Total assets/GNP (%) Total credits/GNP (%) Securities portfolio/GNP (%) Savings deposits/GNP (%) State bank assets/total sector assets (%) Off-balance sheet operations/total assets (%)
1980
1990
1994
1999
2000
20,785 11,168 1,339 10,188 4,288 4,288 — 1,289 — 1,147 28.6 15.4 1.8 5.9 44.1 —
58,171 27,342 5,997 32,564 19,343 11,914 7,429 11,760 3,460 5,903 38.2 17.9 3.9 12.7 44.6 —
52,552 20,559 5,955 33,191 24,190 8,612 15,578 9,019 2,675 4,409 40.3 15.8 4.6 18.5 39.6 49.5
133,533 40,206 22,955 89,361 58,807 24,701 34,106 22,934 12,073 7,840 71.7 21.6 12.3 31.6 34.9 103.5
155,237 50,931 17,848 101,884 64,352 26,628 37,724 29,435 16,284 11,367 76.9 25.2 8.8 31.9 34.2 100.8
Source: The Banks Association of Turkey (2000), BRSA (2001).
138
AYS - E MUMCU AND E. U¨NAL ZENGINOBUZ
Further restructuring in the first 9 months of 2001 reduced this figure down to 27.4%. The number of foreign banks increased from 4 in 1980 to 18 in 2000. The foreign banks’ share in total banking sector assets increased from 3.7% in 1992 to 5.4% in 2000, but this figure stood at 2.6% at the end of September 2001. The foreign presence in the Turkish banking sector is very low compared to similar emerging market economies. Tables 1 and 2 also reveal that the share of loans in total assets declined from 47% in 1990 to 33% in 2000, and finally to 28.5% at the end of September 2001. The ratio of loans to deposits also decreased from 84% in 1990 to 51% in 2000, and to 42.8% as of September 2001. The credit to GNP ratio stood at a low level of 25% at the end of 2000. There are a number of reasons why the Turkish banking sector has continually moved away from its traditional financial intermediation role since the beginning of 1990s. The last decade was one of persistently high and volatile inflation, as well as highly erratic growth rates for the Turkish economy. Liberalization of the current account in 1989 led to external capital inflows, but these were mostly very short term in nature. External capital inflows have proven to be very sensitive to macroeconomic instabilities putting stress on the Turkish economy, and economic fragility, in turn, has further contributed to the volatility of economic variables. This was accompanied by currency substitution, as confidence in the Turkish lira (TL) continuously eroded in the presence of high inflation. The maturity of bank funding sources has shortened significantly and at the same time the share of foreign currency liabilities in total liabilities has increased considerably. Another factor that contributed to the increase in the foreign exchange open position of the banking sector was the attractiveness of borrowing abroad to purchase government securities that offered very high real interest rates. All throughout the 1990s persistent and increasing public sector borrowing requirements were financed through various government securities sold in domestic financial markets. As for the state banks, their financial health was severely compromised, as they were increasingly being used to finance various government expenditures. This, together with inefficient management due to appointments based on political affiliation rather than merit, has resulted in the accumulation of excessive amounts of the so-called ‘‘duty losses’’, which led the illiquid state banks to borrow from the markets at very high rates and short maturities. In turn, this has contributed to practice among private banks’ of borrowing abroad to lend at very high domestic interest rates. For all practical purposes, there were no effective internal risk assessment and management systems that were in use, while liquidity, interest rate, and
An Analysis of Mergers and Acquisitions in the Turkish Banking Sector
139
foreign exchange risks in the Turkish banking sector continually increased throughout the 1990s. In addition, the banking system was not subject to effective supervision and regulation by an independent public agency. The increasingly precarious Turkish banking sector was put through an early test in the wake of the 1994 financial crisis, which was only contained when interest rates were allowed to rise very sharply and the TL depreciated by more than 50%. Banks with short positions in foreign exchange and a mismatch of maturities incurred large losses. In addition, a significant run on deposits occurred, which caused several banks to have severe liquidity problems. Further deposit withdrawals ensued, when three small banks had to be put on the liquidation block. The run on the banks could only be contained when the government introduced a 100% guarantee for savings deposits and declared that it would provide liquidity to the banks in need. These developments revealed that the Turkish banking sector was quite segmented in that it contained a small group of rather efficient and profitable banks, and quite a few small banks at the margin that were prone to falling insolvent in the face of financial downturns. The Turkish banking sector did recover from the 1994 crisis and managed to grow after 1995. However, a number of key issues, such as risk management, and prudent regulation and supervision of the banking sector, were not addressed in the aftermath of the 1994 crisis. At the beginning of 2000, Turkey began to address the need for structural reform through a new economic program; however, before policies could be fully implemented and take effect in the banking area, other aspects of the program had already started to seriously compromise the financial health of the banking sector once again. The disinflation program initially led to a sharp decline in the interest rates, but the government’s financing needed to continue. The pre-announced exchange rate path and the real appreciation of the TL allowed banks once more to borrow abroad at low rates to invest in longer-term fixed rate Treasury Bills. Soaring maturity mismatches and foreign currency open positions were again the end result. Delays in structural reforms in several key areas, together with the sharp increase in the current account deficit, led to a loss of confidence in the disinflation program by the fall of 2000. An outflow of foreign funds ensued, leading to sharp increases in Treasury Bill rates, which in turn led to serious financing difficulties for several banks that had concomitant maturity mismatches and foreign exchange open positions. During the resulting crisis in November 2000 several private banks had to be put under the management of SDIF. Despite efforts to calm the markets, the November 2000 crisis was never really contained, leading to an increasing loss of credibility for the
140
AYS - E MUMCU AND E. U¨NAL ZENGINOBUZ
disinflation program. Finally the exchange rate peg, the vital component of the disinflation program, had to be abandoned in February 2001. In the aftermath of the crisis, a number of banks became insolvent and had to be taken under the umbrella of SDIF. After the February 2001 crisis, a new economic program, named ‘‘Transition to a Strong Economy’’, was adopted. This program called for the completion of the structural transformation of the banking sector and is still in effect. In addition to efforts to further improve banking regulation and supervision, state banks and banks under the umbrella of SDIF are currently undergoing drastic financial restructuring. By 2001, the Turkish banking industry has already begun consolidating. As data for Tables 1 and 2 indicate acquisitions by foreign banks and several mergers among private domestic banks has already occurred. Further mergers and acquisitions are expected to follow in the near future.4 Table 3 presents data for 20 of the privately owned commercial banks in Turkey as of the end of September 2001. The first group, consisting of banks with assets in the US$5–10 billion range, is considered to have market power and identified as a ‘‘dominant’’ group of banks. The total assets of the five banks in this group made up about 64% of the assets of all the privately owned commercial banks, domestic as well as foreign.5 Most of the rest of the banks listed in Table 3 are considerably smaller in size. This group can be considered as the competition ‘‘fringe’’. The total assets of the 20 banks listed in Table 3 represent about 95% of all assets of privately owned commercial banks, and the percentage of deposits and loans held by the banks are even higher.6
3. A STYLIZED MODEL OF THE BANKING INDUSTRY A bank can be modeled as a firm that uses labor and physical capital as inputs to produce different financial services for depositors and borrowers.7 Banking activity is thus viewed as the production of deposit and loan services. We model the banking industry as an imperfectly competitive environment, in which banks act as independent entities. As an approximation of the structure of the Turkish banking industry we will assume that there are a few dominant banks together with a competitive fringe. The economy is assumed to be open so that the banking industry can borrow and lend freely in the world financial markets.
141
An Analysis of Mergers and Acquisitions in the Turkish Banking Sector
Specifically, we assume there to be a total of N domestic banks operating in an oligopolistic industry. There are n banks in the dominant a m banks in the competitive fringe, with, n+m ¼ N. Each bank is characterized by a cost function ci ¼ ðDi ; Li ; F i Þ Table 3.
Private Commercial Banks Ranked by Asset Size (September 2001)a. Total Share in Share in Sector Deposits Sector (%) (%) (US$ Million)
Total Assets (US$ Million)
Share in Sector (%)
Total Loans (US$ Million)
US$5–10 billion I˙s- Bankası Akbank Yapı Kredi Garanti Pamukbank Group totalb
9,491 9,451 8,638 6,780 5,644 40,005
8.6 8.5 7.8 6.1 5.1 36.1 (64.2)
3,247 2,838 3,425 2,161 3,456 15,127
11.3 9.9 12.0 7.5 12.1 52.8 (75.9)
7,084 5,866 5,839 3,830 3,949 26,568
10.3 8.5 8.5 5.6 5.7 38.6 (66.5)
US$1–5 billion Osmanlı Koc- bank Finansbank Su¨merbank Dıs-bank I˙mar Bankası Toprakbank TEB Group totalb
3,536 3,127 2,326 1,476 1,271 1,129 1,123 1,056 15,043
3.2 2.8 2.1 1.3 1.1 1.0 1.0 1.0 13.6 (24.1)
507 988 610 14 239 399 203 267 3,227
1.8 3.5 2.1 0.0 0.8 1.4 0.7 0.9 11.3 (16.2)
1,485 2,086 1,151 1,322 633 958 1,015 712 9,362
2.2 3.0 1.7 1.9 0.9 1.4 1.5 1.0 13.6 (23.4)
US$0–1 billion S- ekerbank Denizbank Alternatifbank Tekstil Bank Oyak Bank Anadolubank Tekfenbank Group totalb
933 933 776 595 355 336 163 4,092
0.8 0.8 0.7 0.5 0.3 0.3 0.1 3.7 (6.6)
0.8 0.6 0.6 0.2 0.4 0.2 0.1 3.0 (4.3)
713 617 496 397 259 256 144 2,881
1.0 0.9 0.7 0.6 0.4 0.4 0.2 4.2 (7.2)
221 185 183 69 104 63 28 854
Source: The Banks Association of Turkey (2000), BRSA (2001). a As of September 30, 2001, there were 24 privately owned domestic banks and 16 foreign banks. Of these, the largest 20 in terms of total asset size are reported in this table. b The figures in parentheses indicate the shares within all privately owned commercial banks, including the foreign owned.
AYS - E MUMCU AND E. U¨NAL ZENGINOBUZ
142
where Di are the deposits collected and Li the loans extended by bank i, and Fi represents bank i’s net position in the world financial market. The balance sheet of a typical bank will therefore look as follows: Assets
Liabilities
Loans (L) Reserves (R)
Deposits (D) Net foreign position (F)
Note that the reserves for a bank i is given by Ri ¼ Di
Li þ F i
Cash reserves, Ci, and the net position on the interbank market, Mi, will make up bank i’s reserves Ri. The cash reserves Ci bear no interest, and therefore will be set at the minimum level required by the Central Bank. Letting a be the reserve requirement ratio on domestic deposits and b the reserve requirement ratio on foreign credits set by the Central Bank, we will have C i ¼ aDi þ bF i
for all i
The banks collect deposits from households and borrow from international financial markets, and channel these funds to firms to finance their investment needs. The third actor in the real sector is the government. The government finances its deficit G either by issuing securities DB or by issuing high-powered money DM0. Assuming that households do not hold cash, the high-powered money is only used to finance banks’ compulsory cash reserves held in their accounts at the Central Bank. That is:
M0 ¼
N X i¼1
Ci ¼ a
N X i¼1
Di þ b
N X
Fi
i¼1
We will assume that banks are differentiated in terms of their perceived risk in the domestic financial market as well as in the foreign financial markets. Each bank i faces an upward sloping inverse supply function of savings rD i ðSÞ; where P S ¼ B þ D; with B being the outstanding government securities and D ¼ N i¼1 Di : Banks’ different perceived financial strengths are assumed D to be summarized by parameters li X1 such that rD i ðSÞ ¼ li r ðSÞ; D where r ðSÞ is the upward sloping inverse supply function of savings
An Analysis of Mergers and Acquisitions in the Turkish Banking Sector
143
faced by the least risky bank (i.e. the bank i* with li ¼ 1). Each bank i is assumed to face aPdownward sloping inverse demand function of loans rL ðLÞ; where L ¼ N i¼1 Li : The cost of borrowing from the international financial market for bank i is rFi ¼ mi rF ; where mi X1 is bank i’s perceived strength in the international financial market and rF is the risk free foreign interest rate. Finally, banks can borrow and lend in the interbank market at rate r.
4. THE BEHAVIOR OF THE DOMINANT GROUP AND THE COMPETITIVE FRINGE The choice variables of the banks are deposits Di, loans Li, and net foreign position Fi. Taking into account the operational costs, the profit of bank i will be given by pi ðDi ; Li ; F i Þ ¼ rL ðLÞLi þ rM i
li rD ðSÞDi
mi rF F i
ci ðDi ; Li ; F i Þ
(1)
where Mi, the net position of the bank i in the interbank market, is M i ¼ ð1
aÞDi þ ð1
bÞF i
Li
(2)
Substituting Eq. (2) in Eq. (1) yields pi ðDi ; Li ; F i Þ ¼ ½rL ðLÞ
rLi þ ½rð1
þ ½rð1
bÞ
aÞ
mi rF F i
li rD ðSÞDi ci ðDi ; Li ; F i Þ
ð3Þ
Thus, the banks’ profits are equal to the sum of the intermediation margins on loans on deposits, and on the net foreign position, net of operational costs . The banks in the competitive fringe are price takers in all three markets. That is, in addition to the risk free foreign rate rF, they take the loan rate rL and the deposit rate rD determined by the dominant group as given as well. The first order conditions for the banks in the competitive fringe are qpfi ¼ rð1 qDi
l i rD
aÞ
qpfi ¼ rL qLi
r
qci ¼0 qDi
qci ¼0 qLi
(4)
(5)
144
AYS - E MUMCU AND E. U¨NAL ZENGINOBUZ
and qpfi ¼ rð1 qF i
bÞ
m i rF
qci ¼0 qF i
(6)
f L D F f L D F F Let Lfi ðrL ; rD i ; ri ; rÞ; Di ðr ; ri ; ri ; rÞ; and F i ðr ; ri ; ri ; rÞ be the loan supply, deposit demand, and net foreign position of the banks in the competitive fringe, respectively, found by solving Eqs. (4)–(6) simultaneously. Define F Lf ðrL ; rD i ; ri ; rÞ
m X
F Lfi ðrL ; rD i ; ri ; rÞ
(7)
m X
F Dfi ðrL ; rD i ; ri ; rÞ
(8)
m X
F F fi ðrL ; rD i ; ri ; rÞ
(9)
i¼1
F Df ðrL ; rD i ; ri ; rÞ
i¼1
and F F f ðrL ; rD i ; ri ; rÞ
i¼1
as the total loan supply, the total deposit demand, and the total net foreign position of the fringe, respectively. The residual demand for loans that the dominant group will face will be Lr ðrL Þ ¼ LðrL Þ Lf ; where LðrL Þ is the total demand for loans (inverse of rL ðLÞ). Similarly, the residual supply of deposits function will be Sr ðrD ; l; mÞ ¼ SðrD ; l; mÞ Df B; where S r ðrD ; l; mÞ is the total supply of savings, with l ¼ ðl1 ; . . . ; lN Þ and m ¼ ðm1 ; . . . ; mN Þ: The banks in the dominant group are assumed to engage in quantity (Cournot) competition in determining their supply of loans and demand for deposits. Given the behavior of the competitive fringe, the first order conditions summarizing the dominant group banks’ maximizing behavior are qpdi qSr ðrD ; l; mÞ qci r D ¼ rð1 aÞ li Di ¼0 (10) þ S ðr ; l; mÞ qDi qDi qDi qpdi qLr ðrL Þ þ Lr ðrL Þ ¼ Li qLi qLi
r
qci ¼0 qLi
(11)
and qpdi ¼ rð1 qF i
bÞ
mi r F
qci ¼0 qF i
(12)
An Analysis of Mergers and Acquisitions in the Turkish Banking Sector
145
d L D F d L D F F Let Ldi ðrL ; rD i ; ri ; rÞ; Di ðr ; ri ; ri ; rÞ; F i ðr ; ri ; ri ; rÞ be the loan supply, deposit demand, and net foreign position of the banks in the dominant group, respectively, found by solving Eqs. (10)–(12) simultaneously. Define F Ld ðrL ; rD i ; ri ; rÞ
n X
F Ldi ðrL ; rD i ; ri ; rÞ
(13)
n X
F Ddi ðrL ; rD i ; ri ; rÞ
(14)
n X
F F di ðrL ; rD i ; ri ; rÞ
(15)
i¼1
F Dd ðrL ; rD i ; ri ; rÞ
i¼1
and F F d ðrL ; rD i ; ri ; rÞ
i¼1
as the total loan supply, the total deposit demand, and the total net foreign position of the dominant group, respectively.
5. EQUILIBRIUM Summing condition (2) over all banks will give the market clearing condition in the interbank market as: ð1
aÞ½Dd ðÞ þ Df ðÞ þ ð1
bÞ½F d ðÞ þ F f ðÞ ¼ Ld ðÞ þ Lf ðÞ
(16)
The market clearing conditions in the loan market and the deposit market will be given by LðrL Þ ¼ Ld ðÞ þ Lf ðÞ
(17)
SðrD ; l; mÞ ¼ B þ Dd ðÞ þ Df ðÞ
(18)
and
respectively. Note that all of the banks are assumed to be price takers in foreign credit markets. Solving (16)–(18) simultaneously will give the noncooperative equilibrium in the banking industry for a given N, li ; mi ; a; b; and rF. Some of the comparative statics exercises that can be carried out thus involve examining the impact of the changes in N, n, and m (mergers), a and b (reserve requirement ratios), li (deposit insurance), mi (country’s international credit rating or acquisitions by foreign banks), and rF (global economic conditions).
146
AYS - E MUMCU AND E. U¨NAL ZENGINOBUZ
In order to simulate the model we must assume particular functional forms for the equation. For the inverse demand function for loans and the inverse supply function of deposits we use: rL ðLÞ ¼ AL
BL L
(19)
rD ðSÞ ¼ AD þ BD S
(20)
and
respectively. As for the cost functions (K), we assume that the banks in the dominant group have a cost advantage over those in the fringe.8 We assume that the cost function of banks in the dominant group is linear in deposits and loans, but quadratic in funds borrowed from abroad. The functional specification adopted for the cost function of the banks in the dominant group is: sf (21) cd ðDi ; Li ; F i Þ ¼ K d þ sD Di þ sL Li þ F 2i 2 The banks in the fringe are assumed to have decreasing returns to scale in production. Their cost function is assumed to have the following form: r r r cf ðDi ; Li ; F i Þ ¼ K f þ D D2i þ L L2i þ F F 2i (22) 2 2 2 The effect of the changes in the parameters of the model will be assessed mainly according to their impact on welfare. We use the standard welfare measure of total surplus (the sum of depositors’ and creditees’ surpluses and total profits of the banks) in quantifying welfare. Mergers can take place either sequentially, which we will call sequential mergers, whereby one bank is acquired at a time, or through block mergers, whereby a number of banks are acquired together. The extent of consolidation will be much greater under block acquisitions than under sequential acquisitions. In the sequential case, the market power and profits of the remaining banks are increased at each stage, making further acquisitions more costly for the acquiring bank.
6. EMPIRICAL APPLICATION: MERGERS AND ACQUISITIONS Wherever possible we use data from the Turkish banking industry. When data are not available on a particular parameter, we concentrate on reasonable range of values and consider the robustness of results to changes in
An Analysis of Mergers and Acquisitions in the Turkish Banking Sector
Table 4.
147
Base Model Parameter Valuesa.
n0 ¼ 5
AD ¼ 0:01
K d ¼ 0:2
K f ¼ 0:03
ld ¼ 1
rF ¼ 0:03
m0 ¼ 15 a ¼ 0:11 b ¼ 0:11
BD ¼ 0:01 AL ¼ 0:9 BL ¼ 0:01
sd ¼ 0:01 sL ¼ 0:02 sF ¼ 0:06
rd ¼ 0:04 rL ¼ 0:05 rF ¼ 0:15
lf ¼ 1 m ¼ 1:05 mf ¼ 1:1
B ¼ 9:7
a
d
The TL values are in quadrillion for the relevant parameters (AD, AL, Kd, Kf, B).
the parameter values. We took the group of privately owned commercial banks listed in Table 3 as representing the completely privatized banking industry structure in our model. Thus, the base line values for n and m are 5 and 15, respectively. The base case for parameters used in most of empirical applications is summarized in Table 4. We use data from 1999, the most recent data before the November 2000 and February 2001 crises. In 1999, the reserve requirement was 6%, thus a was set at 6%, while b was set to 3%. In arriving at values for Kd and Kf we use the overhead costs to assets ratio of 6% reported by Demirgu¨c- -Kunt and Levine (1999). Given the blanket deposit insurance in Turkey that every bank enjoys without discrimination, the perceived financial strengths (from the viewpoint of depositors) of all banks in the base case are taken as equal i.e. we take ld ¼ lf ¼ 1: An average London Inter Banking Offering Rate (LIBOR) rate for 1999 (expressed in terms of TL) is estimated as the value for the risk free cost of borrowing in international markets. The values for md and mf are arrived at by considering the difference in borrowing costs of representative banks from the dominant and the fringe groups.9
6.1. Mergers among Fringe Banks 6.1.1. Will a Fringe Bank Acquire Another Fringe Bank? The first question we consider is whether a fringe bank will have an incentive to acquire another fringe bank. We first look at the case where the merged entity is still a fringe bank. The other case we look at is where the merged entity becomes a dominant bank. Table 5 exhibits, for the base case as well as with higher fixed costs for the fringe, the change in the profits of the fringe banks as the size of the fringe varies. When a fringe bank acquires another fringe bank and the merged entity still remains a fringe bank, the post-merger profits of the fringe banks are higher, as observed in Table 5. However, a merger will actually take
AYS - E MUMCU AND E. U¨NAL ZENGINOBUZ
148
Table 5.
Mergers among Fringe Banks (N ¼ 5)a.
m
pf (K f ¼ 0:03)
pf (K f ¼ 0:06)
Dpf
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
135.57 112.70 96.85 85.11 75.97 68.58 62.43 57.19 52.63 48.62 45.05 41.82 38.90 36.22 33.75
105.57 82.70 66.85 55.11 45.97 38.58 32.43 27.19 22.63 18.62 15.05 11.82 8.90 6.22 3.75
22.85 15.85 11.74 9.14 7.39 6.15 5.24 4.56 4.01 3.57 3.23 2.92 2.68 2.47
a
Profit figures(pf Þ are in TL trillion.
place if and only if both the acquiring and the acquired firms have incentives to merge. The maximum amount that a fringe bank will bid to acquire another fringe bank is the difference between its pre- and post-merger profits. A targeted fringe bank will accept an offer to merge if the bid it receives exceeds its pre-merger profits. Therefore, a merger will go through if and only if the maximum amount that the acquiring firm will be willing to bid exceeds the minimum amount the firm to be acquired is willing to accept. We observe from Table 5 that in the base case there is no profitable acquisition of a fringe bank by another fringe bank, regardless of whether the acquisitions are carried out sequentially or in block. The increase in the profits of banks when the number of fringe banks is reduced by one from 15 to 14 is less than the pre-merger profit of a single bank. If we assume that fixed costs of the fringe banks are higher, block acquisitions become profitable. In the high-fixed cost case presented in Table 5, 10 banks could profitably merge as a block. When there are 15 fringe banks, the profit of each fringe bank is TL 3.75 trillion. With 10 fringe banks merged, there will be six banks altogether in the fringe. The minimum total post-merger profit that will be sufficient to compensate each merging bank is TL 37.5 trillion. The profit of the merged bank will be TL 38.58 trillion in this case, which exceeds TL 37.5 trillion. We observe that sequential mergers are not profitable even in this case of high fixed costs.10
An Analysis of Mergers and Acquisitions in the Turkish Banking Sector
Fringe Banks Merging into a Dominant Banka.
Table 6.
n¼5
m
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 a
149
n¼6
pd
pf
pd
pf
719.78 607.96 522.82 455.34 400.23 354.16 314.94 281.03 251.36 225.11 201.67 180.60 161.52 144.14 128.22
135.57 112.70 96.85 85.11 75.97 68.58 62.43 57.19 52.63 48.62 45.05 41.82 38.90 36.22 33.75
518.22 434.94 371.22 320.49 278.89 243.98 214.15 188.27 165.53 145.35 127.29 110.99 96.18 82.67 70.26
109.01 91.41 79.06 69.80 62.52 56.57 51.58 47.29 43.54 40.22 37.23 34.53 32.07 29.80 27.70
Profit figures(pd ;pf Þ are in TL trillion.
As mentioned above, a merger among fringe banks may lead to their acquiring enough resources to become one of the dominant banks. The comparative statics exercise that is involved here requires checking whether there exist incentives for the fringe banks to become dominant banks. Table 6 shows that the profits of a fringe bank will increase when it is converted to a dominant firm, as expected. If the increase in profits exceeds the cost of acquiring the requisite technology to operate as a dominant bank, a single fringe bank will indeed choose to become a dominant bank.11 For example, the profit of a fringe bank in an industry with m ¼ 15 and n ¼ 5 is TL 33.7 trillion, while the profit of a dominant bank in an industry with m ¼ 14 and n ¼ 6 is TL 82.6 trillion. Therefore, if the cost of acquiring the requisite technology to operate as a dominant bank is less than TL 48.9 trillion, a fringe bank will choose to become a dominant bank. Also, we observe in Table 6 that two fringe banks could profitably merge to become a sixth dominant bank in an industry structure with m ¼ 13 and n ¼ 6. The profit of a dominant bank in the industry structure with m ¼ 13 and n ¼ 6 is TL 96.1 trillion. This exceeds TL 67.4 trillion, the total pre-merger profits of the two fringe firms. Note that the profit advantage of becoming a dominant bank decreases in this case, making mergers among fringe firms less likely. We observe in Table 6 that it would
AYS - E MUMCU AND E. U¨NAL ZENGINOBUZ
150
not be profitable for three fringe banks to merge and then convert into a dominant bank.
6.1.2. Should Fringe Banks Merge? We now turn to evaluating the social welfare desirability of mergers to see whether they conflict with the incentives of banks to merge. As can be seen from Fig. 1, for the base case of parameters we consider, merging firms in the fringe may in fact lead to a decrease in total surplus as long as the size of the dominant group remains the same (the solid line in Fig. 1). Recall from above that there were no incentives on the part of fringe firms to merge in this case either. On the other hand, if the fixed costs of the fringe banks were higher, it would improve welfare if the number of banks in the fringe was reduced through consolidations among themselves. In Fig. 1 the dashed curve displays such a case, where the total surplus maximizing number of firms in the fringe (when n ¼ 5) is 12, indicating the desirability of mergers with respect 18 n = 5, Kf = 0.03
17.7
n = 5, Kf = 0.06
17.4
Total Surplus
17.1 16.8 16.5 16.2 15.9 15.6 15.3 15 1
2
3
4
Fig. 1.
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20 m
Impact of changing m keeping n constant.
An Analysis of Mergers and Acquisitions in the Turkish Banking Sector
151
to the base case where m is 15. Although we would expect total surplus to increase with more competition, increasing m might not lead to higher welfare because of the fixed costs of operation. If fixed costs are sufficiently small, then increasing the number of banks in the fringe will always be welfare increasing. This is due to decreasing returns to scale in production for the fringe. Recall from above that, though sequential mergers were not profitable, block mergers did turn out to be profitable in this case. Note, however, that the reduction in the size of the fringe (to a total of six banks) leads to a smaller fringe than the optimal size identified in Fig. 1. Fig. 2 displays the impact on total surplus of changing the number of banks both the dominant and the fringe groups. We observe that merging a number of fringe banks and turning them into a dominant bank will not necessarily increase welfare. For example, combining two fringe banks (thus reducing m to 13) and turning them into one dominant bank (thus making n ¼ 6) will reduce welfare. This is despite the fact that competition among the dominant group has increased with the addition of one firm, and the improved variable cost when the fringe firms are merged and become a dominant bank. The high fixed cost of operating a dominant firm may 17.6
17.4
Total Surplus
17.2
17
16.8
n=3 n=4
16.6
n=5 n=6
16.4
16.2 1 2
3 4
Fig. 2.
5 6
7 8
9 10 11 12 13 14 15 16 17 18 19 20 m
Impact of changing both m and n.
AYS - E MUMCU AND E. U¨NAL ZENGINOBUZ
152 17.6
Total Surplus
17.4
17.2
17
16.8
n=4 n=5 n=6
16.6 1 2 3 4 5 6 7 8
Fig. 3.
9 10 11 12 13 14 15 16 17 18 19 20 m
Impact of changing both n and m when Kd is low.
exceed gains from increased competition and improved production efficiency. However, if cost structure is different e.g. when fixed costs are lower, the welfare result displayed in Fig. 2 may be reversed. Fig. 3 displays a case where it is desirable for the banks in the fringe to merge and become a dominant bank. Note also from Fig. 3 that whether it is desirable to merge the fringe banks or not will depend on how many dominant firms there are. If there are four banks in the dominant group it will not increase welfare to reduce the number of fringe banks. If, however, n is larger, then mergers in the fringe become desirable.
6.2. Mergers among Dominant Banks 6.2.1. Will a Dominant Bank Acquire Another Dominant Bank? Table 7 presents the change in the profits of the dominant banks as the size of the dominant group varies in the base case. Whether sequential or block
An Analysis of Mergers and Acquisitions in the Turkish Banking Sector
Table 7.
153
Mergers among Dominant Banks (m ¼ 15)a.
n
pd (K d ¼ 0:2)
pd (K d ¼ 0:25)
Dpd
1 2 3 4 5
1593.22 696.66 374.25 218.12 70.26
1543.22 646.66 324.25 168.12 78.22
896.5 322.41 156.13 89.89
a
Profit figures (pd Þ are in TL trillion.
mergers among dominant banks will be profitable depends very much on their fixed costs of operation. In the base case where fixed costs are relatively low, the increase in profits when competition is lessened with a sequential merger will not exceed the pre-merger profits of a targeted dominant bank. Block mergers will, however, be feasible: Four dominant banks can profitably merge at once, leading to a two-bank dominant group. On the other hand, when fixed costs are high, profits in the pre-merger stage are low, making mergers more likely. In the case presented in Table 7 (with K d ¼ 0:25), both sequential and block mergers are feasible. 6.2.2. Should Dominant Banks Merge? Whether dominant banks should merge or not depends on the fixed costs in the dominant group as well as on the size of the fringe. We can use Fig. 2 to consider the impact of changing the size of the dominant group on total surplus. Though there is no incentive for the banks themselves to merge in this case, it turns out that shrinking the size of the dominant group to n ¼ 4 would increase total surplus when m ¼ 15: Further reducing the size of the dominant group to n ¼ 3 would reduce the total surplus, however. Reducing the dominant group by one bank saves on the fixed costs at the expense of reduced competition; the competition effect takes over when there is further reduction in the number of banks and the negative impact on total surplus exceeds the savings from fixed costs. If fixed costs in the dominant group were higher, further reducing the size of the dominant group would have been warranted. One other factor that allows welfare improving reduction in the size of the dominant group is the presence of a sizeable fringe with 15 banks. If
AYS - E MUMCU AND E. U¨NAL ZENGINOBUZ
154
the competitive fringe were smaller, say, with m ¼ 6; increasing the size of the dominant group would have increased monotonically the total surplus.
6.3. Mergers Involving Both Dominant and Fringe Banks 6.3.1. Will a Dominant Bank Acquire a Fringe Bank? Consider now the case where fringe banks are bought out by a dominant bank in a sequential manner (so m decreases while n remains constant). Table 8 shows that, in the base case, there turns out to be no incentive for such acquisitions of fringe banks by a dominant bank. For block mergers to be profitable, one dominant bank will have to acquire 13 fringe banks at once, which indicates that this kind of mergers will be difficult to realize. When the fixed costs of the fringe banks are higher, then both sequential as well as block acquisitions of fringe banks by a dominant bank are feasible. For example, when K f ¼ 0:05; it will be desirable for a dominant bank to acquire fringe banks sequentially until the size of the fringe falls to 12. It would also be profitable to acquire the whole fringe as a block (see Table 8). Table 8.
Dominant Firms Acquiring Fringe Firmsa. n¼5
m
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 a
pd
pf (K f ¼ 0:03)
pf (K f ¼ 0:05)
Dpd
719.78 607.96 522.82 455.34 400.23 354.16 314.94 281.03 251.36 225.11 201.67 180.60 161.52 144.14 128.22
135.57 112.70 96.85 85.11 75.97 68.58 62.43 57.19 52.63 48.62 45.05 41.82 38.90 36.22 33.75
115.57 92.70 76.85 65.11 55.97 48.58 42.43 37.19 32.63 28.62 25.05 21.82 18.90 16.22 13.75
111.81 85.14 67.48 55.11 46.06 39.22 33.90 26.67 26.25 23.43 21.07 19.08 17.38 15.91
Profit figures (pd ;pf ) are in TL trillion.
An Analysis of Mergers and Acquisitions in the Turkish Banking Sector
155
6.3.2. Should a Dominant Bank Acquire a Fringe Bank? Welfare analysis in this case is similar to evaluating the desirability of mergers among fringe banks while keeping the size of the dominant group intact. Fig. 1 demonstrates, for the base case, that the acquisition of fringe banks by dominant banks may lead to a decrease in total surplus (the solid line in Fig. 1). On the other hand, if the fixed costs in the fringe banks are higher, then the acquisition of fringe banks by dominant banks will be improve welfare (the dashed curve in Fig. 1).
6.4. Changes in Policy Variables 6.4.1. Reserve Requirement Ratio Fig. 4 displays the impact of changing the reserve requirement ratio from 6% to 4%. This policy change does not have any impact on the desirability of mergers in the fringe. Decreasing a has the following impact on the equilibrium outcomes in this example: r decreases, rL increases, rD decreases, the profits of both the dominant and the fringe banks decrease, and total 18 17.8
Total Surplus
17.6 17.4 17.2 17 16.8 n = 5, alpha = 0.06
16.6
n = 5, alpha = 0.04
16.4 1 2 3 4 5 6 7 8
9 10 11 12 13 14 15 16 17 18 19 20 m
Fig. 4. Impact of changing a.
AYS - E MUMCU AND E. U¨NAL ZENGINOBUZ
156
welfare increases. A decrease in a amounts to cost reduction which favors the depositors and creditees.
6.4.2. Deposit Insurance Fig. 5 displays the impact of changes in the perceived financial strength of the banks. In the base case we consider, the dominant and the fringe banks are assumed to have the same financial strength from the viewpoint of the depositors. This is justified in the presence of blanket deposit insurance extended to depositors of all banks without discrimination, which has been the case in Turkey since 1994. Any change in the deposit insurance system that will lead to differences in the perceived financial strength of banks will be expected to change equilibrium outcomes. In the example exhibited in Fig. 5, an increase in the perceived riskiness of the fringe banks does not have an impact on the desirability of mergers among the fringe firms. Increasing lf has the following impact on the equilibrium outcomes in this example: r increases, rD decreases, rL increases, the profits of the dominant banks increase while those of the fringe banks decrease, and total 17500 17400
n = 5, lambdaf = 1 n = 5, lambdaf = 1.05
17300
Total Surplus
17200 17100 17000 16900 16800 16700 16600 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 m
Fig. 5.
Impact of changing lf :
An Analysis of Mergers and Acquisitions in the Turkish Banking Sector
157
surplus decreases. Though the change in total surplus is rather small in the case exhibited in Fig. 5, the increase in the market share of dominant banks in deposits may be considerable, leading to driving the fringe banks out of the sector.
6.5. Acquisition by Foreign Banks Consider a scenario in which foreign banks establish a considerable presence in the Turkish banking industry through acquisitions. If the dominant group is now dominated by foreign banks, one implication will be the ease of access to international financial markets. We can capture this change by considering a reduction in md : Fig. 6 displays the impact of such a change on total welfare. In the example exhibited in Fig. 6, a decrease in the perceived riskiness in the international markets of the dominant banks does not have an impact on the desirability of mergers among the fringe. Decreasing md has the following impact on the equilibrium outcomes in this example: r decreases, rD decreases, rL decreases, the market share of dominant banks in 18 n = 5, Mud = 1.05 n = 5, Mud = 1
17.7 17.4
Total Surplus
17.1 16.8 16.5 16.2 15.9 15.6 15.3 15 1 2 3 4
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 m
Fig. 6.
Impact of changing md :
AYS - E MUMCU AND E. U¨NAL ZENGINOBUZ
158
deposits decreases, the profits of the dominant banks increase and those of the fringe banks decrease, and total welfare increases.
6.6. Changes in Global Economic Conditions In our model the impact of the economic conditions external to the country is summarized by rF. In Fig. 7 we look at the impact of changes in this parameter. If the conditions in the world economy worsen so that rF increases, the equilibrium outcomes move in the following directions in the example exhibited in Fig. 7: r, rD, and rL all increase; the profits of both the dominant and fringe banks decrease; and total welfare also decreases. Note that mergers among fringe banks now become desirable.
7. CONCLUDING REMARKS We know from the industrial organization literature that a merger in an imperfectly competitive industry has two main effects. One is a decrease in 17.7 17.5 17.3 17.1 Total Surplus
16.9 16.7 16.5 16.3 16.1 15.9 15.7
n = 5, rF = 0.30
15.5
n = 5, rF = 0.38
15.3 1 2 3 4
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 m
Fig. 7.
Impact of changing rF.
An Analysis of Mergers and Acquisitions in the Turkish Banking Sector
159
competition due to the reduced number of players in the market. The only countervailing effect that can balance this negative impact is the possible efficiency gains that may be realized at the end of the merger. This will more likely be the case in industries where production exhibits increasing returns to scale. In such a case the merger will allow firms to take advantage of economies of scale, which results in a more efficient overall production in the industry. Whether or not the increase in welfare due to efficiency gains will outweigh the adverse effect of reduced competition depends on the specifics of the industry. We modelled the Turkish banking system and simulated a series of scenarios to assess the likely impact of mergers and acquisitions. Our exploration clearly shows the importance of empirically quantifying the cost structures and other related parameters for accurate assessment of the pros and cons of mergers in the banking industry. Depending on parameter values, especially the fixed costs of operating a bank, we identified cases where mergers among fringe banks as well as among the dominant banks turn out to be desirable from the welfare point of view. Note that the welfare measure we used to evaluate changes in banking industry structure as well changes in policy variables was the standard total surplus measure that gave equal weight to consumers’ (depositors’ and creditees’) surpluses and total industry profits. One can certainly envision more general social welfare functions that give higher weight to depositors’ and creditees’ welfare, and our analyses can easily be repeated for more general social welfare functions.12 Whether there will indeed be incentives to merge or not is a question related to, but different from the desirability of mergers. In cases where there is incentive to merge with adverse welfare impacts, the competition authority may want to step in to block the proposed merger. We identified such cases where dominant banks will have the incentive to make bids to acquire fringe banks that will be acceptable to the targeted banks. We also showed that whether the mergers proceed sequentially i.e. one firm is acquired first and then another, or a number of firms are acquired in block makes a difference for the desirability of mergers. As far as incentives to merge are concerned, the extent of consolidation will be greater under block acquisitions than under sequential acquisitions. In the sequential case, the market power and profits of the remaining banks are increased at each stage, making further acquisitions more costly for the acquiring bank. When mergers involving only the fringe banks are considered, we found that as the fixed costs of the fringe banks increase, block acquisitions be-
160
AYS - E MUMCU AND E. U¨NAL ZENGINOBUZ
come profitable. High fixed costs in this case also made decreasing the number of fringe banks preferable from the welfare point of view. The extent of consolidation that would come about could, however, be more than the optimal size identified. In the case of a number of fringe banks merging and becoming a dominant bank, total surplus will decrease in certain cases. This is despite competition among the dominant group and an improvement in the variable cost when the fringe banks become larger. The high fixed cost of operating a large bank may exceed gains from increased competition and improved production efficiency. Whether sequential or block mergers among dominant banks will be profitable depends very much on their fixed costs of operation. As the fixed costs of operating a large bank rise, block mergers as well as sequential mergers may become feasible. Whether dominant banks should merge or not depends on the fixed costs in the dominant group as well as on the size of the fringe. When the fixed costs of the fringe banks are high, then both sequential as well as block acquisitions of fringe banks by a dominant bank become feasible. Welfare analysis in this case demonstrates that the acquisition of fringe banks by dominant banks may increase or decrease total surplus depending on the fixed costs of the fringe banks. Changes in the reserve requirement ratio, the extent of deposit insurance, and improved perception of riskiness in international financial markets affect the equilibrium outcomes obtained. On the other hand, incentives to merge do not seem to be significantly affected by changes in these variables. A dramatic effect of removing the blanket deposit insurance may be the forcing of the fringe banks out of the sector, which will reduce total surplus. We also demonstrated how worsening conditions in the world economy may render consolidation of the fringe desirable from the welfare point of view. A number of lessons can be drawn from our analysis regarding mergers in the banking sector in Turkey. As for whether mergers will occur or not, the nature of the merged entity is of critical importance. If two fringe banks merge to become a dominant bank, this indicates not only an improvement in the cost structure but also enhanced market power. Thus even in cases where there is no incentive for fringe firms to merge to remain a fringe firm, they will merge if through merging they can exercise market power. The welfare consequences of such mergers have to be carefully analyzed, paying attention to the relative size of the fringe and the dominant group and the cost structure in the banking industry before and after mergers. Consolidations among both the fringe and the dominant banks may turn out to be excessive, and thus their extent has to be carefully assessed by competition
An Analysis of Mergers and Acquisitions in the Turkish Banking Sector
161
authorities. Though block mergers, where a group of banks are acquired by another bank, are not necessarily welfare reducing, this type of merger may end up stifling competition too much. Sequential mergers, where only one bank can be acquired at a time, are more difficult; thus, allowing only this type of mergers will be a safety check against excessive mergers. Finally it is noted that mergers among dominant as well as fringe banks will be impeded by a prisoners’ dilemma type interaction among each other. Consolidation in the sector will benefit all banks through reduced competition. But those that are not actually involved in mergers will not bear merging costs. For example, Table 1 reveals that if 10 fringe banks merge to reduce the size of the fringe from 15 to 6, those that do not take part in the merger will enjoy a profit increase of TL 34.83 trillion. Compared to its premerger profit, each of the 10 merging fringe bank will be able to derive much less benefit from the merger, since the total profits of the merged entity will not be more than the profit of fringe firms that remained out of the merger. Thus, each bank favors mergers in the sector but does not want itself to be a part of mergers. Unless mergers lead to dramatic improvement in the cost structure and lead to significant increase in market share, which in the case of fringe banks can only happen if the merged entity becomes one of the dominant banks, the prisoners’ dilemma identified will render consolidation in the sector highly unlikely.
NOTES 1. SDIF was established in 1983. The Turkish Central Bank had managed SDIF from its inception until August 30, 2000, at which time it was transferred to BRSA. 2. The Banks Association of Turkey and BRSA compiles the information and data presented in this section from various periodic reports. The web sites (in English) of the Banks Association of Turkey and BRSA can be reached at http:// www.tbb.org.tr/english and http://www.bddk.org.tr/english/mainpage/index_eng.htm, respectively. 3. Information on the developments in the Turkish banking sector over the 1980– 2000 periods is compiled from various BRSA reports, in particular from BRSA (2001). 4. An earlier version of this chapter predicted a Turkish banking sector with about 20 private commercial banks (see Mumcu-Serdar, Zenginobuz, & Orhan, 2001). Table 1 reveals that the number of private commercial banks had already fallen to 24 by the end of September 2001. 5. After September 2001, Garanti, the fourth largest private bank in Table 3, merged with Osmanlı, the sixth largest private bank according to Table 3, thereby becoming the largest private commercial bank in Turkey. With this merger, the share of the first five largest private commercial banks in total sector assets reached 69.8%.
162
AYS - E MUMCU AND E. U¨NAL ZENGINOBUZ
6. In these calculations, foreign owned commercial banks have been included among the privately owned commercial banks. The banks under SDIF management have not been included in this category. 7. For various models of banking industry structure and bank behavior, see Freixas and Rochet (1996). 8. Evidence of economies of scale in the banking industry is mixed. Various studies show that only very small banks have the potential to achieve scale economies and that the average cost curve becomes quickly more or less flat for the larger firms (see Dermine, 1999). 9. We thank Cevdet Akc- ay for providing us this information on the perceived riskiness of the privately owned Turkish commercial banks in international financial markets. 10. As fixed costs do not affect the marginal decisions of banks, Dpf is the same for all levels of fixed costs. However, fixed costs affect the level of profits and thus the feasibility of mergers. 11. The assumption here is that there are no other extra-economic barriers that would prevent a fringe bank from acquiring the requisite technology to become a dominant bank. 12. If there is concern regarding the stability of the banking system, giving less weight to profits may lead, in general, to more frequent exit and entry in the sector than desirable. In turn, this may compromise systemic stability of the sector. In addition to being the standard assumption in the industrial organization literature, total surplus reflects a neutral stance regarding this point.
ACKNOWLEDGMENTS We gratefully acknowledge the contribution of Mehmet Aras Orhan for an earlier version of this chapter. We also thank two anonymous referees for helpful comments. Go¨ze Aslankara provided assistance with data for the current version. The chapter draws from a project, titled ‘‘The Restructuring of the Turkish Banking Industry and Its Impacts on Competition’’, funded by Bogˇazic- i University Research Fund, Project No. 00C103.
REFERENCES BRSA. (2001). Towards a sound Turkish banking sector, May 15. ohttp://www.bddk.org.tr/ english/mainpage/index_eng.htm4 (December 30, 2001). Demirgu¨c- -Kunt, A., & Levine, R. (1999). Bank-based and market-based financial systems: Cross country comparisons. Policy Research Working Paper No. 2143, Washington, DC: The World Bank. Dermine, J. (1999). The economics of bank mergers in the European Union, a review of the public policy issues. INSEAD Working Paper No. 99/35/FIN.
PART III: MONETARY, FINANCIAL AND EXCHANGE RATE POLICIES
163
This page intentionally left blank
164
HYSTERESIS IN CURRENCY SUBSTITUTION: THE MIDDLE EAST AND NORTH AFRICA Wassim N. Shahin and Fadi G. Freiha ABSTRACT After reviewing the theoretical and empirical literature on currency substitution, a model is used in this chapter to empirically examine the state of dollarization in Middle East and North African countries, using Lebanon and Egypt as case studies. For Lebanon, despite the decline in inflationary expectations, the expectations of currency depreciation, and an increase in real interest rate differentials between domestic and foreign currencies, dollarization did not decline by the anticipated amount. For Egypt, unlike many Latin American Countries, currency substitution was successfully reversed for a period when the government managed to peg the value of the Egyptian pound to the dollar.
1. INTRODUCTION Dollarization, which has become a ‘‘generic’’ term for currency substitution, implies substituting one stable currency, usually the US$, for an unstable domestic currency. This phenomenon has been largely in existence since the Money and Finance in the Middle East: Missed Opportunities or Future Prospects? Research in Middle East Economics, Volume 6, 165–181 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1094-5334/doi:10.1016/S1094-5334(05)06008-5
165
166
WASSIM N. SHAHIN AND FADI G. FREIHA
1970s in several Latin American countries, namely Mexico, Uruguay, Peru, Argentina, Bolivia, and Venezuela. Dollarization also dominated countries like Egypt and the Yemen Arab Republic in the late 1970s and early 1980s, and has been high in Lebanon since 1985. Currency substitution also became common in transition economies in Eastern Europe in the early 1990s after measures were introduced to liberalize economic activity and allow the trading of foreign exchange and the presence of bank accounts in foreign currency.1 In all of these regions, dollarization has existed in different degrees as dollars performed the function of a store of value and even replaced the domestic currency as a medium of exchange and a unit of account in some economies. The literature addressing dollarization can be divided into two broad parts. Early studies examined the causes and implications of currency substitution to understand the salient features of this phenomenon. The second group of studies examined the state of currency substitution in the aftermath of the reversal of the factors that caused it. The main issue was to determine whether or not dollarization is a reversible phenomenon. That is, does dedollarization take effect through a self-adjustment mechanism given political stability and a reversal in inflationary expectations and expectations of currency depreciation, or is there hysteresis in currency substitution and asymmetry between the rate of dollarization and the factors that cause it? The purpose of this chapter is to shed light on the irreversibility of the dollarization phenomenon by examining the state of currency substitution in the Middle East and North Africa (MENA), using Lebanon and Egypt as case studies. These two countries, characterized by high rates of dollarization in the past, have witnessed major political and economic stability in the last few years. For Lebanon, we argue that despite a decline in inflationary expectations, the expectations of currency depreciation, and an increase in real interest rate differentials between domestic and foreign currencies, dollarization did not decline by the anticipated amount. We diverge from the literature on currency substitution that attributes such results to the hysteresis or irreversibility of dollarization. In the case of Egypt, we find that unlike many Latin American countries, currency substitution was successfully reversed for a period when the government managed to peg the value of the Egyptian pound to the dollar. However, when the government began to devalue in the year 2000, dollarization reoccurred. This chapter is divided into five parts. Section 2 reviews the literature on the causes and implications of dollarization by highlighting the results of several studies examining these issues. Section 3 analyzes the hysteresis or
Hysteresis in Currency Substitution: The Middle East and North Africa
167
irreversibility of dollarization. In Section 4, we conduct empirical analysis on currency substitution in Lebanon and Egypt to examine its reversibility given several years of political stability. Section 5 includes concluding remarks.
2. CAUSES AND IMPLICATIONS OF DOLLARIZATION Dollarization is usually demand induced by the banks and the non-bank public in countries where the institutional environment is conducive for currency substitution. Countries experiencing dollarization are generally characterized by one or more of the following few restrictions, if any: Restrictions on foreign exchange holdings or trading, the availability of all type of foreign accounts in the banking system encouraging the transactionary, as well as the store of wealth motive for holding money, or low transactions costs in switching between currencies and little or no control on capital flows. In addition, these countries lack well-developed financial markets and instruments in domestic currency causing the public to move to foreign currency accounts whenever the return on domestic ones declines. The causes of dollarization are both political and economic. Political reasons take the form of political and institutional instability, such as internal strife, regional conflicts, rigged elections, and political assassinations. Economic reasons attribute currency substitution to high and varying inflation rates limiting the purchasing power of domestic currency, high-inflationary expectations in the absence of reliable economic stabilization plans, expectations of domestic currency devaluation to reduce current account deficits and the value of public debt, high real interest rate differential between domestic and foreign currencies, and high real return differential due to further expectations of depreciation of the domestic currency. The early literature on dollarization attempted to test several of the above-mentioned causes. The research was mostly conducted on Latin American Countries after the dollarization of this region in the late 1970s and early 1980s. Some of the first empirical studies on the causes of the subject (Ortiz, 1983; Canto, 1985; Ramirez-Rojas, 1985; Marquez, 1987; ElErian, 1988) used structural money demand models to examine the determinants of dollarization. The models tested currency substitution, largely defined as the ratio of foreign deposits in the domestic banking system to total deposits, as a function of inflation rate differentials between countries,
168
WASSIM N. SHAHIN AND FADI G. FREIHA
interest rate spread on different currencies, and exchange rate expectations, along with some variables measuring lags and stock adjustment terms. Some studies defined dollarization differently to include cross-border foreign deposits of domestic residents, whereas others included domestic currency in the definition of domestic money. Common to all of the studies was the underestimation of dollarization due to the exclusion of the foreign currency in circulation variable as a result of difficulties encountered in its measurement. Statistical analysis gave large support to the significance of exchange rate expectations in causing dollarization in Mexico (Ortiz, 1983; RamirezRojas, 1985), Uruguay and Argentina (Ramirez-Rojas, 1985), the Dominican Republic (Canto, 1985), Venezuela (Marquez, 1987), and Egypt and the Yemen Arab Republic (El-Erian, 1988). The impact of interest rate differentials on the degree of currency substitution was mixed. Whereas the variable was of less significance in some of the studies analyzed (Ortiz, 1983; El-Erian, 1988), it was shown to be significant in others (Ramirez-Rojas, 1985), especially for Uruguay, where the interest rate differential was used as a proxy for the expected change in the exchange rate. The effect of inflation rate differentials on dollarization was also shown to be significant, especially for Argentina and Mexico, where this variable was used as the proxy for the expected change in the exchange rate (Ramirez-Rojas, 1985). The literature addressing the implications of currency substitution on economic activity stated and discussed the negative as well as the few positive implications. The negative impacts of dollarization are numerous warranting a close and serious examination of this phenomenon to recommend de-dollarization measures or to determine a policy of co-existence with currency substitution. First, highly dollarized countries lose their ability to define and control the monetary aggregates for two main reasons: (i) the inexistence of a proper and accurate measure of foreign currency in circulation, especially when this currency is used as a medium of exchange and (ii) the common and erratic flight of domestic currency and local currency deposits to cross-border ones in different denominations whenever there are signs of political instability and expectations of exchange rate depreciation. Second, such an environment impacts the ability of the monetary authority to plan and effectively implement monetary policy. Third, dollarization causes short-term fluctuations in currency prices leading to a wide range of volatility in the value of the exchange rate and in the expectations of depreciation. This issue was tested in Girton and Roper (1981), and Isaac (1989). The first study found that currency substitution produces instability in the sense that shifts in the expected rate of exchange rate change produce
Hysteresis in Currency Substitution: The Middle East and North Africa
169
larger movements in the exchange rate. This instability would be eliminated when the behavior of money issuers is endogenized. In the second study, Isaac examined the extent of exchange rate overshooting or undershooting to determine that the impact depends on the degree of currency substitution. Whereas this phenomenon may cause the exchange rate to overshoot in response to commodity demand shocks, it may actually moderate overshooting in response to money demand shocks. Fourth, the short-term volatility of the exchange rate lowers the long-run value of exchanging currencies. Fifth, lowering the long-run value of the domestic currency causes an increase in the long-term level of the inflation rate as a result of the imported component of inflation in open-economies running current account deficits. Sixth, the increase in inflation at a time when wages are expressed in domestic currency reduces the real return on labor. Seventh, this decline in real wages lowers the real return from taxation at a time of an increase in government expenditure due to inflation. This causes an increase in budget deficits and affects the planning, implementation, and effectiveness of fiscal policy. Eighth, the high dollarization rate in the banking system weakens the Central Bank’s ability to act as a lender of last resort in cases of banking crisis. Finally, dollarization largely affects government revenue from seigniorage profits. This issue was tested in Melvin (1988), who used reduced-form equations for Bolivia and Mexico to test the impact of dollarization on seigniorage profits in these countries. He found a systematic significant relationship between the two variables implying a reduction in profits. The positive impacts of dollarization also took several forms. First, allowing bank accounts in foreign currencies prevents capital flight providing needed funds for domestic investments and increasing the public’s portfolio choices. Ortiz (1983) tested the stability of the money demand function in Mexico following the 1976 devaluation. The results suggest that the instability problem associated with the currency substitution hypothesis has not been empirically meaningful in Mexico. He concludes that as long as dollars held in Mexico are close substitutes for ones abroad, short-run speculative capital flows should be reduced. Second, the dollarization of economies, especially ones characterized by high budget deficits and government debt, may force the government to tighten its expenditure due to the shortage of hard currency it holds relative to the size of spending. Finally, in cases where dollarization is symmetrical (two countries holding each others’ currencies), currency substitution serves to equate the real rates of return on both currencies, limit inflation, and harmonize monetary and fiscal policies.
170
WASSIM N. SHAHIN AND FADI G. FREIHA
3. DOLLARIZATION AND HYSTERESIS It was initially believed by many that decreases in inflationary expectations and the expectations of exchange rate depreciation coupled with some degree of political stability would lead to a decline in currency substitution. However, as the economic and political situations improved in several Latin American countries, it was noticed that dollarization remained high. Therefore, several studies were developed to examine the reasons behind the irreversibility of dollarization. This section will first highlight some of the models that have tested the reasons behind the irreversibility of currency substitution in Latin America. Then we will state some of the results derived from the experience of some Eastern European countries showing, based on casual empiricism, the decline in currency substitution with the reversal of factors that cause it. In a study on the hysteresis or irreversibility of dollarization in four Latin American countries, Bolivia, Mexico, Peru, and Uruguay, Guidotti and Rodriguez (1992) examined the relationship between dollarization and inflation rate differentials. The results derived based on unit root tests suggest for all four countries the non-stationarity of the dollarization time-series and the stationarity of the inflation rate differentials. Therefore, whereas shocks to the level of dollarization have permanent effects, shocks to the inflation differential are of a transitory nature. Using co-integration tests, the authors show that these two variables are also not co-integrated. They conclude that dollarization cannot be fully described by a function of the current interest rate or inflation rate differential as is usually implied by traditional currency substitution models. Therefore, a more complex dynamic relationship may exist between the two variables. To try and interpret hysteresis, the authors use dynamic analysis to define a band for the inflation differential within which there is no incentive to switch between currencies. They conclude that above the upper level of the band, the economy becomes more dollarized, whereas below the lower value, dedollarization occurs. The irreversibility of currency substitution in Bolivia was also captured in a study by Clements and Schwartz (1993). Using a standard currency substitution model, the authors show that even though economic variables are statistically significant, they have a fairly low explanatory power compared to variables that measure inertia in the dollarization process, such as the stock adjustment variable and the time-trend. The authors conclude that monetary and fiscal measures may be insufficient in reversing the dollarization process.
Hysteresis in Currency Substitution: The Middle East and North Africa
171
The experience of some selected transition economies in Eastern Europe seems to differ completely from the case of countries in Latin America. Viewing the decline of dollarization in these countries, Sahay and Vegh (1995) use a conceptual framework and raw data on the level of dollarization and its determinants to conclude that the reversal of currency substitution is primarily attributed to a higher real return on domestic currency assets as a result of lower inflation rates and more market-determined interest rates. The conclusions of this study are to be interpreted differently than ones conducted on Latin America since they are based on a conceptual framework spanning a maximum of a 4-year period beginning with the financial market liberalization of these countries in 1990. The experience of many Latin American countries differs in the sense that these countries witnessed long periods of dollarization starting in the 1970s in highly unstable economic and political environments where time-inconsistent and time-variant institutional changes occurred regularly, depriving these policies of their credibility. In such a setting, once the public has found means to economize on its domestic money, a reversal becomes difficult to achieve. These institutional factors have influenced the public’s expectations about confiscation risk affecting its decision of where and how to hold deposits in foreign currency (Savastano, 1996).
4. HYSTERESIS IN THE MENA REGION: LEBANON AND EGYPT This section starts by presenting the currency substitution model adopted in the testing methodology, the sample size, and the data used. Then we discuss the conditions of Lebanon and Egypt, and report the results on the reversibility of the currency substitution phenomenon. The construction of the model is based on the assumption that demand for foreign currency is a function of differentials between foreign and domestic interest rates and exchange rate expectations. The resulting framework consistent with the empirical literature on dollarization, especially Guidotti and Rodriguez (1992), Clements and Schwartz (1993), and Mueller (1994), is represented by Doll1t ¼ a1 þ a2 I þ a3 E þ t b
(1)
where Doll1 ¼ lnðð1=CS Þ21Þ represents the degree of dollarization in period t with b being a possible acceleration rate of CS (currency substitution). The variable will be estimated using values of b equal to 1. I is the interest rate differential between domestic deposit rate and foreign currency deposit
172
WASSIM N. SHAHIN AND FADI G. FREIHA
rate. E is the expected change in the exchange rate computed based on a static expectations variable proxied by the depreciation rate observed during the current month. is the error term. Some models have included expected inflation and inflation rate differentials. Monthly figures on inflation rates are not available in Lebanon and Egypt, and existing yearly inflation figures are unofficial. Therefore, we believe, as is the case in many studies, that interest rate differentials and exchange rates capture the movements in the inflation variables. The sample period of the study extends from January 1984 to September 2001 for Lebanon, and from January 1984 to June 2001 for Egypt (all monthly). Data sources are from the International Financial Statistics, which are based on official Central Bank reports of both countries. For both countries, the standard definition of dollarization is used, which is the ratio of foreign currency deposits to total deposits. The proxy for foreign currency deposit rate is the 3 month London Inter Banking Offering Rate (LIBOR) for Egypt for the period, and the 3 month LIBOR with adjustment for Lebanon until February 1995 when the Central Bank started reporting figures on foreign currency deposit rates, which were adopted until the end of the sample period. Domestic interest rates are average deposit rates for Lebanon and the discount rates for Egypt, which are approximately equal to the average between deposit and lending rates.
4.1. A Lebanese Application Lebanon experienced a major internal strife since 1975. However, the domestic currency started depreciating in 1984 leading to dollarization (Fig. 1). Lebanon provides a very suitable environment for currency substitution, as it is characterized by the availability of all type of bank accounts in foreign currency with no restrictions on the holdings or trading of foreign money and no control on capital flows. The causes of dollarization in Lebanon have been examined by Osseiran (1987). The standard causes discussed in Section 2 were found to hold. The institutional motive is captured by the degree of the economy’s openness. The deterioration of the quality of domestic money and the expected depreciation of the domestic currency were found to be significant factors. In early Fall of 1992, the Lebanese currency reached a historic low against the US$ standing at $1 ¼ 2420 LL as compared to $1 ¼ 5.84 LL in the first month of 1984. With the advent of the new government of Mr. Rafic ElHariri in October 1992, the currency started to regain its value, ending the
Hysteresis in Currency Substitution: The Middle East and North Africa
173
19
84 19 .01 84 19 .12 85 19 .12 86 19 .11 87 19 .10 88 19 .09 89 19 .08 90 19 .07 91 19 .06 92 19 .05 93 19 .04 94 19 .03 95 19 .02 96 19 .01 96 19 .12 97 . 19 11 98 19 .1 99 20 .09 00 20 .08 01 .0 7
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%
Fig. 1.
Dollarization chart of Lebanon. Source: Central Bank of Lebanon.
year at $1 ¼ 1838 LL. Given the halt in hostilities, and with the high expectations placed on the fast reconstruction of the country, the public was expected to restore its confidence in the domestic money as inflation rates began to decline and the Lebanese currency began to appreciate. The trend of currency substitution was also expected to reverse itself. A study by Mueller (1994) based on data going until 1993 examined the limited reversibility of the dollarization process. The author found that despite the significance of the expected depreciation of the domestic currency and of interest rates when cross-border deposits are included in the model, a ratchet effect measuring hysteresis is more significant. The conclusion that was reached pointed to an asymmetry between domestic and foreign currencies. We use a model including recent data going until 2001 (as opposed to Mueller’s with 1 year of observations after the beginning of the period of political stability in 1992) to test whether after 10 years of political stability, dollarization has become a reversible phenomenon in Lebanon. Our aim is to shed more light on the concept of hysteresis (if any) in the country by examining the symmetry between currency substitution and the factors that cause it. We tend to believe that even though the country witnessed an appreciation in the domestic currency, the exchange rate and the interest rates never reached their market-determined levels. This view is evidenced by some casual empiricism showing that interest rates that are determined through direct manipulations of the Treasury-Bills rates were raised regularly to peak levels at times of political disturbances, such as October 1995 when discussions about the presidential elections were taking place.2 The authority, in an attempt to prevent the flight away from domestic currency,
174
WASSIM N. SHAHIN AND FADI G. FREIHA
was regularly altering the Treasury-Bills rates to unprecedented and sometimes unexpected levels. This event would cause banks to raise their deposit rates on domestic currency to attract more deposits. Banks were able to conduct asset management in this fashion because they were required to hold at least 40% of their Lebanese pound deposits in Treasury-Bills (a rule that was relaxed in April 1997). As a result, many of them were surviving on the spread between the Treasury-Bills rate and the deposit rate.3 The artificial and forced increase in interest rates made many believe that the values of interest, and especially exchange rates had not yet reached their market-determined short- as well as long-term levels. Therefore, currency substitution did not decline to its expected levels possibly for this reason rather than because it is largely an irreversible phenomenon, as is the case in Latin America. In order to achieve our stated purpose, we use the standard definition of the proxy for currency substitution by defining dollarization as the ratio of foreign currency deposits in the banking system to total deposits. To test the significance of our specification, we start, following Guidotti and Rodriguez (1992), by testing for the presence of a unit root in the timeseries process of dollarization, the interest rate differential, and the rate of expected depreciation of the domestic currency.4 The version of the augmented Dickey–Fuller (ADF) test used allowed for a time-varying drift. The results are summarized in Table 1. The findings show that there is evidence at the 1% significance level that the dollarization proxy and the interest rate differentials have unit roots. Therefore, shocks to their levels have permanent effects, as their time-series process is non-stationary. The proxy for the rate of expected depreciation of the domestic currency appears to be stationary. Shocks to this variable seem to be transitory.
Table 1.
ADF Unit Root Test.
Dickey–Fuller t-Statistic
Level First difference
Mackinnon Critical Values
Dollarization
Expected Depreciation
Interest Rates Differential
1%
–3.154 –5.949
–5.563
–3.483 –5.962
–4.005
Source: Author’s estimates.
Hysteresis in Currency Substitution: The Middle East and North Africa
Table 2.
Johansen Co-integration Test.
Series: Doll1, E, and I Eigenvalue
0.116 0.067 0.044
175
Lags Interval: 1–4
Likelihood Ratio (LR)
5% Critical Value
1% Critical Value
Hypothesized Number of CE(s)
49.474 23.853 9.398
34.91 19.96 9.24
41.07 24.6 12.97
None* At most 1** At most 2**
Source: Author’s estimates. CE(s) is the co-integrating equation (s). LR test indicates three co-integrating equations at 5% significance level. *, **Denote rejection of the hypothesis at 1% and 5% level of significance, respectively.
It is noted that the dollarization proxy and the interest rate differentials (I) are shown to be integrated of the first order, Doll1(1), I(1) whereas the expected depreciation (E) is integrated at level E(0) (It is important to note that this was expected since the variable E is defined as a percentage change which is similar to the first difference of the two other variables.). Given a group of non-stationary series we are interested in determining whether the series are co-integrated to identify the long-run equilibrium relationship. To determine the co-integration of the three variables, we conduct a Johansen co-integration test. The results reached are summarized in Table 2. The results indicate that we can reject the null hypothesis of no co-integration among the variables at the 1% and 5% significance levels. These findings suggest that dollarization in Lebanon can be described by a function of the current interest rate differentials and expected depreciation. The co-integration between currency substitution and factors that cause it show that the variables obey a stable long-run relationship, which in a sense mimics the concept of long-run equilibrium. The post-1987 period (Fig. 1) seems to point to a trend similar to the pre1987 period. Therefore, we tested the period from 1987 onwards separately to determine whether structural adjustment of data is necessary. The results also show co-integration between the dollarization proxy and the independent variables in line with our findings for the whole period.
4.2. An Egyptian Application Egypt’s economic situation was volatile during the last two decades. In 1986, as oil prices dropped sharply causing a regional recession, Egypt directed its
176
WASSIM N. SHAHIN AND FADI G. FREIHA
focus on internal policy and borrowed heavily leading to macroeconomic imbalances. In the early 1990s, Egypt undertook a macroeconomic stabilization effort along with a long-term structural reform, such as institutional strengthening, privatization, trade liberalization, legal reforms concerning investment and commercial activities, and a restructuring of the financial sector. As indicated by Subramanian (1997), the most important element of the stabilization process was the reduction of the government deficit achieved by both a decline in government expenditures (subsidies, transfer payments, salaries, etc.) and a rise in government revenues (due to the appreciation of the Egyptian pound, rise in oil receipts, introduction of new taxes, etc.). According to Mongardini (1998), Egypt encountered a series of exchange rate depreciations, which led to a loss of confidence in the Egyptian pound. During the stabilization process, the Egyptian Government applied contractionary fiscal policies and set interest rates at high levels leading to a high inflow of capital and to the appreciation of the domestic currency; however, this appreciation, induced a loss of competitiveness in the tradable goods, and increased the current account deficit. This dilemma between keeping a strong domestic currency and a current account deficit should take into consideration whether the exchange rate is at market equilibrium level or overvalued. Mongardini (1998) analyzed the exchange rate structure in Egypt and segmented it into three periods: (a) Prior to 1987, the exchange rate market was composed of three components: The Central Bank exchange rate which was applied on most imports, exports of oil, and rice (it traded around LE 0.7). The commercial bank exchange rate, which was applied on tourism, exports that do not go through the Central Bank, and workers remittances (it traded around LE 1.36). It is important to note that both the above rates are subject to heavy interventions. The non-bank free market rate (unofficial), mainly used by the private sector for exchange. (b) Between 1987 and 1991, a new bank foreign exchange market was introduced with a rate set around the non-bank free market rate (around LE 2.16), which would later replace the commercial bank rate. It should be noted that the Central Bank rate was devalued to LE 2 per dollar and the new bank foreign exchange rate dropped to LE 3 to the dollar.
Hysteresis in Currency Substitution: The Middle East and North Africa
177
70% 60% 50% 40% 30% 20% 10%
19 84 19 85 19 86 19 86 19 87 19 88 19 89 19 89 19 90 19 91 19 92 19 92 19 93 19 94 19 95 19 95 19 96 19 97 19 98 19 98 19 99 20 00 20 01
0%
Fig. 2. Dollarization chart of Egypt. Source: Central Bank of Egypt.
(c) After 1991, the previously explained foreign exchange rates were replaced by two rates, the primary and the secondary rates which would later combine into a single rate. By examining the dollarization chart of Egypt (Fig. 2), we can clearly see the path of currency substitution from 1984 to 2001. We observe three trends: An upward trend from 1984 to 1991, indicating a rise in the dollarization rate, a downward trend from 1991 to 2000 indicating a process of de-dollarization that accompanied the stabilization program of the Egyptian Government, and the beginning of a third trend reflecting the reoccurrence of dollarization after the government started the devaluation process. It should be noted that the early stages of de-dollarization saw stability in both the exchange rate and the interest rate differentials (differential between the domestic interest rate and the LIBOR). Similar to our approach in analyzing data on Lebanon, we tested for the presence of a unit root in time-series process of dollarization, the interest rate differential, and the exchange rate expectations in order to determine the stationarity of each series.5 The null hypothesis of a unit root is rejected against the one-sided alternative if the t-statistic is less than (lies to the left of) the critical value. In Table 3, when the test was conducted on the first level of each variable, the test fails to reject the null hypothesis of a unit root in the two series (Doll1 and I) at the reported significance level, indicating that shocks have
178
WASSIM N. SHAHIN AND FADI G. FREIHA
Table 3.
ADF Unit Root Test.
Dickey–Fuller t-Statistic
Level First difference
Mackinnon Critical Values
Dollarization
Expected Depreciation
Interest Rates Differential
1%
–2.057 –4.474
–7.395
–1.737 –4.901
–4.007
Source: Author’s estimates.
Table 4.
Johansen Co-integration Test.
Series Doll1, E, and I Eigenvalue
0.237 0.069 0.026
Lags Interval: 1–4
Likelihood Ratio (LR)
5% Critical Value
1% Critical Value
Hypothesized Number of CE(s)
73.439 19.738 5.389
34.91 19.96 9.24
41.07 24.6 12.97
None At most 1 At most 2
Source: Author’s estimates. CE(s) is the co-integrating equation (s). LR test indicates one co-integrating equation at 5% level of significance. Denotes rejection of the hypothesis at 5% significance level.
permanent effects as the series are non-stationary. On the other hand, expected depreciation (E) is integrated at level E(0) (this was expected since the variable (E) is defined as a percentage change which is similar to the first difference of the other two variables). Moreover, when the test was conducted for the first difference of dollarization and interest rate differentials, these two variables became stationary at the 1% confidence level indicating that at the first difference level, shocks on dollarization and interest rate seem to become transitory (Doll1 and I are integrated of order one). To determine the co-integration of the three variables, we conducted a Johansen co-integration test summarized in Table 4. Sometimes, two variables will follow a random walk, but a combination of those variables will be stationary. In our model, although the rate of dollarization and exchange rate both behave as random walks, we would expect these two variables to move together over the long run, so that a linear combination of the two should be stationary.
Hysteresis in Currency Substitution: The Middle East and North Africa
179
The trace statistic rejects the hypothesis of no co-integration at the 5% level and the long-run test indicates one co-integrating equation. The above data show co-integration between I and Doll1. Since mid-1991, E has been stable. This stability, accompanied by an absence of permanent interest rate differential effect on Doll1, could have had a positive effect on the rate of dollarization (being one of the reasons of de-dollarization). Unlike many reforming countries, Egypt successfully managed to reverse its currency substitution rate. Empirical results show that this was possibly due to the fact that interest rate shocks did not have a permanent effect on dollarization and that exchange rate stability policies had a permanent positive impact. Both results lead to de-dollarization.
5. CONCLUDING REMARKS The literature on currency substitution points to the presence of hysteresis in the dollarization process. As the variables causing currency substitution, such as an increase in expected inflation, an increase in the expected rate of depreciation in the domestic currency, and a decline in the real return on this currency worsen, dollarization occurs. However, as these determinants improve, currency substitution does not automatically decline possibly implying the irreversibility of the dollarization phenomenon. In this chapter, preliminary results on Lebanon after 10 years of stability show that dollarization and factors that cause it seem to be integrated based on Johansen test. Therefore, we tend to believe that unlike the case of many Latin American countries, dollarization in Lebanon could be reversible given that the actual values of interest rates and exchange rates become market determined. We conclude that currency substitution was not reversed in Lebanon largely because the authority has been trying to protect the value of domestic currency through forced stabilization measures. These attempts caused currency substitution to remain relatively high as the public perceived actual levels of interest and exchange rates to be non-market determined. Policies aimed at allowing market-determined interest and exchange rates may decrease currency substitution even if the current value of the domestic currency unit declines. The public may perceive this new level as one closer to the currency’s long-run equilibrium, a fact that may reduce its expectations of currency depreciation. In the case of Egypt, the stabilization policy and the enormous fiscal adjustment pursued from 1991 to 2000 has managed to establish a stable
180
WASSIM N. SHAHIN AND FADI G. FREIHA
exchange rate, avoid a collapse in output, strengthen the financial system, maintain a sound balance of payments, and improve Egypt’s external viability making it less vulnerable to sudden shifts in investor sentiment. Unlike many Latin American and Eastern European countries, Egypt successfully managed to reverse its currency substitution rate for a period when it successfully managed to peg its currency to the US$. However, the latest devaluation measures that have started in year 2000 caused a reoccurrence of the dollarization phenomenon allowing us to conclude that the reversibility phenomenon in Egypt was temporary.
NOTES 1. A very thorough review of dollarization in Latin America is found in Savastano (1996). For an excellent survey of the recent experience of Eastern European countries with dollarization, refer to Sahay and Vegh (1995). 2. There were several instances between 1993 and 1997 when the Treasury-Bills rate would be increased to facilitate marketing government debt or to encourage the public to hold domestic currency in an attempt to preserve its value. 3. Until 1995, Banks in Lebanon were required to hold 60% of their deposits in Lebanese currency in Treasury-Bills, 10% in required reserves and 3% in required reserves at 6% interest rate. In 1995, the figure declined from 60% to 40% until it was relaxed in April 1997. It is to mention that banks were not required to hold reserves against dollar deposits until October 2001 when the new rule stipulates a 15% required reserve ratio against foreign currency deposits. 4. For the interest rate differential, we use the difference between LIBOR and the deposit rate on domestic deposits with maturities up to 3 months until February 1995 when the Central Bank started reporting dollar deposit rates. For the remaining years, we use the reported interest rate figures on dollar deposits in Lebanon instead of LIBOR. 5. Three unit root tests were conducted. In order to eliminate serial correlation, we specified four lagged first difference terms. In Table 3 the ADF tests for the following regressions: DDoll1t ¼ m þ gDoll1t 1 þ aDDollt 1 þ bDDollt 2 þ wDDollt 3 þ dDDollt DE t ¼ m þ gE t 1 þ aDE t 1 þ bDE t 2 þ wDDollt 3 þ dDE t 4 DI t ¼ m þ gI t 1 þ aDI t 1 þ bDI t 2 þ wDDollt 3 þ dDI t 4
4
REFERENCES Canto, V. (1985). Monetary policy, dollarization and parallel market exchange rates: The case of the Dominican Republic. Journal of International Money and Finance, 4, 507–521. Clements, B., & Schwartz, G. (1993). Currency substitution: The recent experience of Bolivia. World Development, 21(11), 1883–1893.
Hysteresis in Currency Substitution: The Middle East and North Africa
181
El-Erian, M. (1988). Currency substitution in Egypt and the Yemen Arab Republic. International Monetary Fund Staff Papers, Paper No. 35(1), 85–103. Girton, L., & Roper, D. (1981). Theory and implications of currency substitution. Journal of Money, Credit and Banking, 13(1), 12–30. Guidotti, P., Rodriguez, C. (1992). Dollarization in Latin America: Gresham’s Law in Reserve. International Monetary Fund Staff Papers, Paper No. 39(3), 519–544 International Monetary Fund, International Financial Statistics, Various Issues. Isaac, A. (1989). Exchange rate volatility and currency substitution. Journal of International Money and Finance, 8, 277–294. Marquez, J. (1987). Money demand in open economies: A currency substitution model for Venezuela. Journal of International Money and Finance, 6, 167–178. Melvin, M. (1988). The dollarization of Latin America as a market-enforced monetary reform: Evidence and implications. Economic Development and Cultural Change, 36, 543–558. Mongardini, J. (1998). Estimating Egypt’s equilibrium real exchange rate. International Monetary Fund Working Paper, WP/98/5. Mueller, J. (1994). Dollarization in Lebanon. International Monetary Fund Working Paper, WP/94/129. Ortiz, G. (1983). Currency substitution in Mexico: The dollarization problem. Journal of Money, Credit, and Banking, 15(2), 174–185. Osseiran, F. (1987). Currency substitution in Lebanon. Banque du Liban. Bulletin Trimestriel, 35, 4–11. Ramirez-Rojas, C. L. (1985). Currency substitution in Argentina, Mexico, and Uruguay, International Monetary Fund Staff Papers, Paper No. 32(4), 629–667. Sahay, R., & Vegh, C. (1995). Dollarization in transition economies: Evidence and policy implications. International Monetary Fund Working Paper, WP/95/96. Savastano, M. (1996). Dollarization in Latin America: Recent evidence and some policy Issues. International Monetary Fund Working Paper, WP/96/4. Subramanian, A. (1997). The Egyptian stabilization experience: An analytical retrospective. International Monetary Fund Working Paper, WP/97/105.
This page intentionally left blank
182
CHOOSING AN APPROPRIATE PALESTINIAN MONETARY REGIME Moyara Ruehsen ABSTRACT This chapter examines several viable monetary regimes including the introduction of a Palestinian currency operating under a managed float, a Palestinian currency operating under a currency board, a monetary union with Jordan, the status quo that permits the Jordanian dinar, Israeli shekel, and U.S. dollar as legal tender, and finally, dollarization coupled with the introduction of Palestinian coins. Each of these options is compared on the basis of whether or not it enhances macroeconomic stability, provides the benefits of seignorage, deters inflation, stimulates investment, and encourages fiscal and monetary discipline.
1. INTRODUCTION Every national entity has symbols of sovereignty, such as a national flag, and the issuance of official postage stamps, license plates, passports, and in most cases, a national currency. The Palestinian Authority, like any other emerging independent state, also aspires to issue its own national currency. Money and Finance in the Middle East: Missed Opportunities or Future Prospects? Research in Middle East Economics, Volume 6, 183–199 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1094-5334/doi:10.1016/S1094-5334(05)06009-7
183
184
MOYARA RUEHSEN
The Palestinian National Authority (PNA) stated its intention to introduce a new currency, the Palestinian Pound, as early as 1995. The announcement was met with both concern and surprise. Most of the concern was expressed by the Palestinian Authority’s neighbor, Jordan, as there were between $500 and $800 million of Jordanian dinars in circulation at the time in the West Bank and Gaza, amounting to one third of all Jordanian currency in circulation.1 There was also a great deal of surprise at the announcement given the small size of the Palestinian economy, and its vulnerability to external shocks, which an independent financial regime would exacerbate. In response to the Palestinian Authority’s decision to issue a new currency, many economists immediately advocated the adoption of a currency board. Soon, it became almost a foregone conclusion.2 A currency board more or less takes the place of a Central Bank.3 It requires every unit of domestic currency to be backed by a corresponding quantity of hard currency, such as the U.S. dollar (US$), held in reserve. This allows for the free convertibility of the domestic currency on demand at the parity rate, without the dangers of destabilizing depreciation.4 The benefits of such an arrangement are the built-in adjustment mechanisms and the automatic enforcement of fiscal and monetary discipline. The rationale behind the proposal was that a new Palestinian currency would be extremely vulnerable to internal and external shocks, but a currency board would at least protect the Palestinian pound from speculative attacks and force fiscal and monetary discipline on the Palestinian governing authorities even in times of crisis. This chapter analyzes some of the monetary options available to the Palestinians. Section 2 examines and evaluates some of the proposals that have been put forward so far, including an independent currency with a managed float, an independent currency managed by a currency board, monetary union with Jordan, the status quo, and finally dollarization coupled with Palestinian coinage. Section 3 offers a further critique of the currency board option since that option has received the most attention and favor by both Palestinians and other countries in the region. Section 4 concludes with recommendations for a monetary regime that encourages financial stability and foreign investment, and minimizes the risks of a serious monetary crisis.
2. PROPOSED OPTIONS There has been heated discussion over the shape of such a future monetary regime since the Palestinian Authority’s announcement in 1995 of its plan to
Choosing an Appropriate Palestinian Monetary Regime
185
introduce a Palestinian pound to replace the three currencies which are currently legal tender (the US$, the Israeli new shekel, and the Jordanian dinar). Ideally, a state’s monetary regime will provide a healthy financial environment that is conducive to economic growth. It follows that the ideal choice of a monetary regime is one which maximizes growth-conducive conditions. Some of these conditions include a stable and predictable macroeconomic environment with low inflation and a realistic exchange rate, a strong and resilient banking system, high savings and investment rates, and strong governmental institutions committed to fiscal and monetary discipline.5 The monetary arrangement which has generated the most support thus far is a new, independent currency managed by a currency board. However, other options have also been discussed, including a managed float, a currency union with Jordan, and continuation of the status quo. Finally, one option that has not been discussed in the literature, but which is worthy of more serious consideration, is dollarization. Each of these options is considered in turn, with emphasis on the arrangement’s ability to enhance prospects for economic growth by encouraging macroeconomic stability, providing the benefits of seignorage, deterring inflation, stimulating investment, and encouraging fiscal and monetary discipline.
2.1. An Independent Currency with a Managed Float For a nation-state seeking to declare its independence, an independent currency provides a powerful symbol of sovereignty, and offers many potential advantages such as independent monetary policy and seignorage. But this symbol of sovereignty carries many risks. First, a new currency is fragile by nature, as investors, both domestic and foreign, will be wary of holding assets denominated in the new currency until it has a proven track record.6 Second, newly independent states lack well-established financial institutions capable of withstanding speculative attacks. Third, introducing a new currency is likely to create new transaction costs. Finally, states declaring independence are likely to have enemies, who may wish to destabilize the new currency with a few carefully chosen actions or policy initiatives.7 How do these benefits and costs stack up in the case of the PNA? The symbolic advantage of having a national currency is a powerful one, but only if the new national currency is stable and inspires confidence. A weak currency, which continues to depreciate against the currencies of the country’s main trading partners, has the opposite effect of instilling pride, and can become an embarrassment. It also discourages investment in the domestic economy.
186
MOYARA RUEHSEN
Likewise, the benefits of having an independent monetary policy can also be problematic. On one hand, having one’s own currency allows a central bank to manipulate the supply of money in order to adjust for the optimal inflation/unemployment trade-off, particularly in the event of a crisis. Interest rate adjustments can cushion the blow of a shock and reduce the magnitude of a recession. On the other hand, independent monetary policy means a central bank has the freedom to print money in order to finance a deficit. For a poor, struggling country, this option can be extremely tempting. The introduction of a new and freely floating currency is likely to erode fiscal and monetary discipline. The only irrefutable benefit to having an independent currency is seignorage. However, in the Palestinian case, this benefit is difficult to estimate, and amounts only to a fraction of the foreign aid received on an annual basis. Estimates of seignorage range from $15 to $20 million per year (Naquib, 1998, p. 23; UNCTAD, 2001, p. 29).8 In sum, apart from seignorage, the benefits of having an independent currency are tenuous at best. The costs of an independent currency, in the Palestinian case, are likely to be quite high, given current conditions. First, years of ongoing conflict have left the local economy in shambles. Personal savings are quite low, and foreign private investment is minimal. This kind of environment does not provide a foundation for the introduction of a ‘‘strong’’ and stable currency or a stable macroeconomic environment. Continued depreciation or successive devaluations will only lead to more inflation and a worsening balance of payments. Second, the Palestinian economy has probably experienced more shocks, and is likely to experience many more in the foreseeable future, than any other emerging market economy outside of sub-Saharan Africa. Border closures with Israel, worker permit restrictions, and internal closures within the West Bank and Gaza, inhibit the movement of people and goods and result in significant hardship. With more than 100,000 Palestinian workers commuting into Israel before September 2000, the income foregone of one day of border closure alone amounted to as much as $1.5 million per day.9 Third, given the asymmetric trade relationship between Israel and the Occupied Territories, introducing a new currency, which is not interchangeable with the shekel, can present numerous difficulties.10 None of these factors bodes well for the introduction of a strong, independent currency or macroeconomic stability.
2.2. An Independent Currency Managed by a Currency Board A currency board would appear, at first glance, to address many of the concerns discussed above with the introduction of new currency. However, a
Choosing an Appropriate Palestinian Monetary Regime
187
currency board introduces new challenges. First, a currency board automatically enforces fiscal and monetary discipline on a country. By backing up every unit of currency in circulation with an equivalent amount of foreign reserves, the central monetary authorities cannot print money to finance a deficit. But at the same time, they cannot use monetary policy to address the inflation/unemployment tradeoff, and therefore relinquish the main advantage of having an independent currency in the first place. Second, reserves must be large enough to match a given definition of the money supply, which the Palestinians might be able to do better than most emerging market economies, but still it will not be easy. Third, the currency board mechanism allows for automatic adjustments via the interest rate. (When demand for the domestic currency falls and the anchor currency rises, the domestic money supply shrinks and interest rates rise in response, causing demand for the domestic currency to rise again.) This can be extremely harmful, particularly for economies with inflexible labor markets, and initial high rates of unemployment. A currency board is only advantageous in instances of hyperinflation. There is strong evidence that currency boards are an ideal, albeit temporary, solution for countries experiencing extremely high rates of inflation. Currency boards brought inflation down from triple-digit to single-digit figures in the cases of Argentina, Bulgaria, Estonia, and Lithuania (Kopke, 1999; Hanke, 2000; Ghosh, Gulde, & Wolf; 2000). However, there is little evidence to suggest that currency boards had any effect (positive or negative) on inflation rates when there was relatively little inflation before the currency board went into effect.11 Palestinian inflation rates have been relatively stable since Israel brought down its hyperinflation nearly two decades ago. As a result, it would seem that the Palestinians have little to gain in the way of inflation deterrence with the adoption of a currency board. Also, the attractive benefit of seignorage is minimized with a currency board, in part because the currency board does not have the authority to print money. As stated above, the benefits of seignorage with a independent floating currency will likely add up to $15–20 million per year. With a currency board, the benefits will be much smaller, so seignorage by itself does not provide enough of a reason to adopt a currency board. Finally, a currency board could potentially encourage investment by providing a critical level of macroeconomic stability. This assumes, of course, that international financial markets develop confidence in the Palestinian Monetary Authority’s (PMA’s) commitment to maintain the currency board in the medium to long run. If financial markets believe this commitment is serious, then in theory they have no reason to launch a
188
MOYARA RUEHSEN
speculative attack against the currency. However, as recent experience has indicated, a currency board does not stave off speculation, nor does a currency board insulate a country from external macroeconomic shocks. The best examples of a currency board’s inability to protect a country’s financial system from speculative attack are Hong Kong and Argentina. Hong Kong has had a currency board since 1983, with the Hong Kong dollar (HK$) fixed to the US$ at a rate of HK$7.8/US$. At the height of the Asian financial crisis, most pegged currencies throughout Asia were forced to fend off massive speculative attacks or give up and retreat to a float. Most analysts, however, did not include the HK$ in their bleak predictions, because in theory it made no sense to attack a currency which had an equivalent amount of foreign exchange reserves backing up every unit of domestic currency in circulation. However, the inconceivable happened, and on October 23, 1997, a run on the HK$ forced interbank interest rates into triple digits (Kasa, 1999). More attacks followed in January, June, and August of 1998. Needless to say, this had a devastating impact on Hong Kong’s property market and gross domestic product (GDP) growth for the next several quarters. A similar thing happened to Argentina in 1995 in the aftermath of the Mexican peso crisis, and again in the late 1990s in the aftermath of the Brazilian real crisis. Like Hong Kong, Argentina had a currency board in place, with each Argentine peso in circulation backed up with US$1 in reserve. But again, this insurance was not enough to scare off speculation. It has been suggested that currency traders will attack a currency board if they suspect that a government may soon abandon the currency board (Kasa, 1999). One factor which may influence such suspicions is the real appreciation of the exchange rate. This appears to have been the case in Argentina, where the Argentine peso appreciated steadily over time against the currencies of its main trading partners within the Mercosur customs union. Pressure from trade union groups and exporters led currency speculators to wonder about the imminent devaluation of the peso. This begs the question then that if the Palestinians choose to adopt a strict currency board, how soon will they be under pressure to abandon the board (and thus invite potentially destabilizing speculative attacks) when the Palestinian currency starts to appreciate vis-a`-vis other trading partners like Israel and Jordan? 2.3. Monetary Union with Jordan A monetary union offers many advantages for a group of countries meeting the criteria of an ‘‘optimum currency area’’ (Mundell, 1961). It can provide
Choosing an Appropriate Palestinian Monetary Regime
189
a large degree of macroeconomic stability for the smaller member countries, and thereby provide an environment conducive to domestic and foreign investment. Whether countries qualify as an optimal currency area depends on the degree of labor and capital mobility between them, with the mobility of labor being a critical factor.12 If countries meet these conditions, then a currency union offers many benefits, such as reduced transaction costs, and in theory, a more efficient allocation of capital and labor resources within the union. At least one author has come out in favor of monetary union with Jordan (Naquib, 1998; UNCTAD, 1996a,b), making the argument that Jordan and Palestine share more of the characteristics of an optimum currency area than Israel and Palestine. Naquib also describes the Jordanian capital market as ‘‘organically tied with Palestinian capital (p. 27).’’13 There is much truth to this, and as illustrated in Fig. 1, a significant portion of lending and cash transactions in the PNA are conducted in Jordanian dinars. However, what is not clear is the degree to which the two economies are tied together with respect to trade. Most of the Palestinian Authority’s trade (exports and imports) is currently conducted with Israel, nor does there appear to be much, if any, labor mobility between the Jordanian and Palestinian economies. That is not to say that this may not change in the future if present hostilities lead to a greater degree of separation between the Palestinian and Israeli economies. The question of fiscal and monetary discipline depends, of course, on the arrangement the Palestinian and Jordanian authorities negotiate. In a true currency union, monetary policy is a collective or joint responsibility. In this case, however, it is highly unlikely that the Jordanian monetary authorities would relinquish any authority over monetary policy. Rather, the Palestinians would be at the complete mercy of Jordanian monetary policy, regardless of whether it is appropriate for Palestinian needs. As will be discussed later, the Palestinian economy is subject to numerous shocks, and hence, has unique fiscal and monetary needs. A currency union will only succeed in the face of asymmetric shocks if there is a very high degree of capital and labor mobility (Mundell, 1961), which does not appear to be the case here. Finally, on the issue of seignorage, there is a precedent, even in instances where the larger economy has sole control over monetary policy, of sharing the rewards of seignorage. One example is the Rand Currency Zone, in which the tiny countries of Lesotho and Swaziland use the South African rand as their official currency. South African monetary authorities exercise full control over monetary policy, but they reimburse Lesotho and Swaziland for the rand seignorage generated within their borders (Hamed, 2000, p. 17).
190
MOYARA RUEHSEN Bank Deposits∗
*Bank deposits as of April 1999. Source: PMA 1999a. Lending∗∗
**Outstanding loans as of April 1999. Source: PMA 1999a. Cashed Checks∗∗∗
*** Source: PMA 1998, 1999b Cash Transactions∗∗∗∗ Israeli Shekel
Jordanian Dinar
US Dollar
**** Source: Author estimates based on sample figures in Hamed (2000) p.3.
Fig. 1. Composition of Bank Deposits, Loans, and Cash Transactions by currency. Source: Palestine Monetary Authority (1998, 1999a, b) and Hamed (2000, p. 3).
2.4. Status Quo At present, the Jordanian dinar, the new Israeli shekel, and the US$ are all legal tender within the Palestinian territories. Palestinian banks hold accounts and issue loans in all three currencies, and the private sector conducts business in all three currencies. The reasons for this arrangement are historical in nature. Prior to 1967, the Jordanian dinar was the official currency for the West Bank and Gaza. But after 1967, the Palestinian territories were
Choosing an Appropriate Palestinian Monetary Regime
191
forced to embrace a currency union with Israel. For the next two decades, the Palestinian economy suffered from the transmission of several macroeconomic shocks, principally the triple-digit hyperinflation Israel experienced in the 1980s.14 As a consequence, many Palestinian households switched to the Jordanian dinar as a safe haven for their personal savings. Over time, the US$ also became increasingly used as a safe haven. However, the true picture of currency composition is a complicated one. As can be seen in Fig. 1, most bank deposits reflect the historical explanation above, with 27% of deposits held in dinars, and 59% held in dollars. Also, the composition of lending is similar, with 36% of loans issued in dinars, and 62% in dollars. However, many cash transactions are still conducted in Israeli shekels, perhaps because so many Palestinian laborers are paid in new Israeli shekels. Despite the widespread reliance on the US$ by Palestinian banks, the status quo is frequently labeled a ‘‘dual currency area’’ referring to the use of both the Jordanian dinar and the new Israeli shekel. As such, it is subjected to twice as many macroeconomic shocks, or costs typically associated with exchange rate fluctuations. Monetary shocks originating in Jordan are passed on through the financial account because of the high degree of capital mobility between Jordan and the PNA, whereas monetary shocks originating in Israel are passed on through the current account, because of the high degree of goods and services transactions between Israel and the PNA (Naquib, 1998, p. 22). This suggests a lack of macroeconomic stability, and a high degree of vulnerability to inflationary shocks from its two neighbors. The climate for domestic and foreign investment is not necessarily compromised by this situation given the increasing use of the US$ in bank lending and deposits. However, it has been pointed out that commercial banks operating in the Palestinian territories are finding it increasingly difficult ‘‘to perform the function of transforming debt maturities, because of currency mismatching of assets and liabilities,’’ which is a disincentive to the extension of long-term loans (UNCTAD, 2001, p. 28). To sum up, the status quo is not an ideal arrangement. It suffices in the present climate of uncertainty, but it is not a viable long-term solution for the Palestinian economy.
2.5. Dollarization and Palestinian Coins Dollarization offers the Palestinian economy a monetary regime that is not too far removed from the status quo.15 It offers the ultimate in
192
MOYARA RUEHSEN
macroeconomic stability, by providing ‘‘a monetary policy as good as that of the United States’’ (Schuler, 1998b, p. 3). It would, in theory, result in lower real interest rates than a currency board, and would cushion the Palestinian economy from inflationary shocks from its neighbors. It would also impose fiscal and monetary discipline by default, as government authorities would not be able to print money to finance budget deficits. All of these factors would enhance the investment climate in the Palestinian territories, which would contribute positively to economic growth. It has been estimated that dollarization contributes an additional 1% of GDP growth per year for the average emerging market economy (Schuler, 1998b, p. 6). The only disadvantage is the loss of potential seignorage. While dollarization offers so many advantages, there is reluctance among many to consider this option, because it does not offer the symbolic benefit of having one’s own currency. However, this problem is easily addressed. If the Palestinians were to dollarize, they would need to solve the issue of small change, and they could issue Palestinian coins for this purpose. Other countries, which have instituted dollarization have also done this. Panama, for instance, uses the US$, but issues its own coins, while East Timor adopted the dollar in 2000, but decided to import plane loads of American coins to use as change. Not only was this expensive, but it was also highly impractical. Timorese coinage for every day transactions would have been cheaper, and would have provided the East Timorese with a symbol of national pride. The other advantage to issuing Palestinian coins is the seignorage benefit. However, this benefit is expected to be minimal compared with the $15–20 million expected with a fully independent Palestinian currency. Finally, the choice of a key currency (euro vs. dollar) warrants some discussion. Many newly independent countries have chosen dollarization or ‘‘euroization,’’ such as East Timor, Kosovo and Montenegro. The optimal choice for these countries depends on respective economic links to the dollar and euro zones. In the case of the PNA, much of the external aid currently allocated is in the form of US$. Also, many of the countries with which it has close financial relations (e.g. Jordan and the Arab states of the Gulf), peg their currencies to the US$. This last point is an important one because one potential danger of dollarization is the loss of trade competitiveness, which can occur if the currencies of the country’s trade partners depreciate significantly vis-a`-vis the US$. In conclusion, of the five possible currency arrangements discussed here, dollarization offers the most net benefits when judged on the criteria of macroeconomic stability, seignorage, inflation, investment climate, and
193
Choosing an Appropriate Palestinian Monetary Regime
Table 1.
Comparison of Different Palestinian Monetary Regimes.
Type of Monetary Regime
Enhances Macroeconomic Stability
Provides Benefits of Seignorage
Palestinian currency with managed float Palestinian currency with currency board Monetary union with Jordan Status quo Dollarization +issuance of Palestinian coins
Low
Yes
Medium
Deters Inflation
Stimulates Investment
Encourages Fiscal and Monetary Discipline
No
No
No
Yes
Possibly
Minimal
Yes
Medium
Possibly
Possibly
Minimal
Possibly
Medium–High High
No Minimal
Possibly Yes
Minimal Yes
Yes Yes
Source: Author’s estimates.
fiscal and monetary discipline (see Table 1). Issuing Palestinian coins alongside the dollar also addresses the important need of having a symbol of national pride.
3. A FURTHER CRITIQUE OF THE CURRENCY BOARD OPTION Of all the monetary regime options discussed above, the one which has received the most attention and support thus far is the currency board.16 As many analysts have operated with the assumption that the PNA is determined to introduce its own currency regardless of the risks, they have concluded that a currency board is probably the best option. However, a currency board offers none of the benefits of having one’s own currency, nor does it provide a guarantee against speculative attack. It is also far less superior to the dollarization and coinage option discussed above. Why does the support for a currency board continue to persist? One explanation is the respect for the Palestinian desire to have a symbol of national pride in the form of its own currency. This desire cannot be disputed, but issuing Palestinian coins provides almost as much symbolic pride
194
MOYARA RUEHSEN
as the issuance of Palestinian bills. A second explanation for the strong support of the currency board option is the patronizing and insulting assumption that most emerging market economies are too fiscally and monetarily irresponsible to manage their own currency well, and therefore must have their hands tied with a currency board. Finally, another explanation is the erroneous assumption held by many that a currency board will automatically provide a climate of macroeconomic stability. With regard to this last point, currency boards have come under increasing criticism in recent years precisely because they have not lived up to what they initially promised. In theory, speculators will not attack a currency board, because they know the currency board has enough foreign exchange in reserve to back up every unit of domestic currency. However, this assumes that speculators believe that the government will maintain the currency board and not abandon it. If currency markets suspect that a currency board will be abandoned, they will undoubtedly launch an attack by selling off the domestic currency. As explained earlier (see Note 4), when this happens interest rates will rise, precipitating a potentially serious recession at home. This happened to Argentina in 1995 after the Mexican peso crisis and again in 1999 after the Brazilian real crisis. Real per capita GDP growth fell to 4% in 1995 from nearly 5% the year before, and fell from nearly 7% in 1997 to 4% in 1999. The same thing happened to Hong Kong in 1997– 1998 during the Asian crisis. Real per capita GDP growth fell from just over 5% in 1997 to 7.5%. Estonia and Lithuania provide two more examples of countries with currency boards suffering a deeper and prolonged recession. After the Russian ruble crisis in 1998, Estonia’s real per capita GDP growth rate fell from a high in 1997 of just over 11%, to 0% by 1999. Lithuania’s growth rate fell from 7% in 1997 to 3% in 1999. Latvia, by contrast, which did not have a currency board in place, saw its growth rate only fall from 7% to 3% during the same period. It is also worth considering whether the Palestinian economy meets the basic preconditions which must be in place for a currency board to succeed. One such condition is a well-developed, well-supervised, and regulated financial system, encompassing not just the banking sector, but also the bond market and equity markets (Wagner, 2000, p. 193). Then there is the question of whether or not the PMA can come up with an adequate level of foreign exchange reserves. (The PMA is in a better position to do this than most other emerging market monetary authorities.) Finally, a currency board needs broad political support from the populace (Ghosh, Gulde, & Wolf, 2000, p. 295). If the Palestinian population finds that lending rates are generally higher under the currency board than they were under the ‘‘dual’’
Choosing an Appropriate Palestinian Monetary Regime
195
currency arrangement, any initial support for such an arrangement may quickly erode. Is a currency board really appropriate for the Palestinian economy? Currency boards have proved advantageous, albeit in the short term, in situations of hyperinflation. In the cases of Argentina, Bulgaria, Estonia, and Lithuania, the introduction of currency boards brought down triple-digit inflation to more manageable levels.17 However, hyperinflation has never been a problem for the Palestinian economy. Inflation has been in the single digits since 1995. Even if it were to become a problem in Palestine, the introduction of the currency board alone, as one author points out, did not solve the problem of hyperinflation in most of those countries. In Argentina’s case, for example, many economic fundamentals were improved at the same time the currency board was put in place, such as an improved tax administration, public sector reform, privatization, and trade liberalization (Naquib, 1998, p. 5). In the end, having a currency board in place while other trading partners were devaluing their currencies led to a serious overvaluation of the Argentine peso. The pressures of the recession that resulted soon became unmanageable.
4. CONCLUDING REMARKS AND RECOMMENDATIONS High-level staff members of the PMA, including the former Governor, do not appear to be in a blind rush to establish a currency board. They are well aware of the difficulties of implementing such a regime and appear open to the consideration of other options.18 As of mid-2001, the worsening conflict with Israel has changed many of the assumptions underlying the original currency board decision in 1999. If anything, the current conflict and its devastating impact on the Palestinian economy point to an increasing need for the Palestinians to obtain easier access to external markets and to develop those markets. Such markets are likely to be outside of the immediate region, so there is little rationale for the Palestinians to maintain a currency union with any neighbor. Furthermore, the currency board option does not provide the most important benefits of having one’s own currency. It also does not protect the currency from speculative attack, nor does it cushion the impact of external shocks. An alternative worthy of careful consideration is the Palestinians following other newly independent provinces and states, including Kosovo and
196
MOYARA RUEHSEN
East Timor, and dollarizing. The choice of the US$ for paper money would provide the Palestinian economy with improved macroeconomic stability. In combination with dollarization, the issuance of Palestinian coins for change (following the example of Panama) could serve as a symbol of national pride. The loss of potential seignorage by not having a new, independent currency would be outweighed by the gains from increased investment resulting from reduced financial risks and overall macroeconomic stability.
NOTES 1. United Press International (1996), Martin (1997), and Zak (1999). A study conducted by the Housing Bank for Trade and Finance (HBTF) subsequently suggested that the introduction of a Palestinian currency would not ‘‘affect the purchasing power or the potency of the dinar,’’ although future competition between the two currencies would likely occur (The Star, 2000). The Israelis, in contrast, expressed little concern. And by 1999, an Israeli government-sponsored research team came out in support of a new Palestinian currency, albeit one operating under a currency board arrangement (Voice of Palestine, 1999b). 2. See Hamed (2000, p. 2). ‘‘It is reasonable to assume that a Palestinian currency will be issued under a currency board arrangement, at least in the first few years;’’ United Press International (1998), PMA (1998, p. 89). By 1998, the design of the notes was already decided, with al-Aqsa Mosque on one side and a portrait of Yassir Arafat on the other (United Press International, 1998). At the end of 1999, the expectation was that the new currency would be issued within 1–2 years (MENA, 1999). By 1999, the PMA was already approaching the World Bank and International Monetary Fund (IMF) for assistance with the introduction and implementation of a new currency (Economic Intelligence Unit, 2000, p. 37). 3. For good, descriptive definitions of a currency board and how it functions (see Williamson, 1995; Montagnon, 1998; Schuler, 1998a; Gulde, 1999; Kopke, 1999; Ghosh, Gulde, & Wolf, 2000; Hanke, 2000; Wagner, 2000). 4. If too much of the anchor currency is demanded in exchange for the domestic currency, the money supply will shrink and domestic interest rates will rise, thereby attracting increased investment in the domestic currency once again. 5. For a more complete and excellent discussion of criteria and policy instruments designed to promote economic growth and to what extent the Palestinian economy measures up (see Abed, 1999). 6. Hisham Awartani, an economist at the Center for Palestine Research and Studies in Nablus, suggests that most Palestinians, if given the choice between a new Palestinian currency and the dollar, dinar or shekel, will opt for the dollar, dinar and shekel (Lynfield, 2000). 7. When Eritrea introduced its own currency in 1997, Ethiopia responded by not allowing Eritrean traders to use Ethiopian banks, and rerouting Ethiopian exports through the port of Djibouti in place of the Eritrean ports of Asmara and Massawa.
Choosing an Appropriate Palestinian Monetary Regime
197
8. One estimate ranges as high as $600 million (Voice of Palestine, 1999a). However, this figure is unreasonably inflated. 9. This estimate is calculated from figures in Shaban (1999, pp. 51–52). Amount is in 1995 US$. 10. Again, the Eritrean–Ethiopian case is instructive here. Before Eritrea introduced its own currency, and before hostilities broke out, 66% or Eritrea’s exports went to Ethiopia, but only 10% of Ethiopia’s exports went to Eritrea. There is similar asymmetry in the Palestinian–Israeli case, with as much as 85% of Palestinian exports going to Israel, but only 8% of total Israeli exports going to the Palestinian Authority. When Ethiopia shut out Ethiopian traders after the introduction of the Eritrean currency (nacfa), the Eritrean economy suffered considerably. 11. Hanke (2000, p. 57) illustrates the case of Bosnia–Herzogovina. In Bosnia’s case inflation was relatively insignificant, compared to other case studies, before and after the currency board was introduced. 12. One study which looked at numerous Middle Eastern countries found that the region as a whole did not meet the criteria of an optimal currency area, expect perhaps the Gulf countries (Sadeh, 1997). 13. See also The Star (2000), for a brief discussion of Palestinian and Jordanian economic ties. 14. See Arnon and Spivak (1996) for a further discussion of the transmission of shocks between the Israeli, Palestinian and Jordanian economies from 1967 until 1994. 15. Dollarization can also refer to the adoption of other key currencies, such as the yen or euro. 16. A conference was convened in Cairo on November 13–14, 1999 specifically to discuss the implications of a new Palestinian currency, steps toward implementation and the currency board option. The atmosphere at the conference was one of general support for the idea, with the outstanding question being when, not if, such a monetary regime should be implemented. 17. See Ghosh, Gulde, and Wolf (2000) for comparisons of all four cases, and Gulde (1999) for a more detailed discussion of Bulgaria’s experience. 18. Author interview and discussions with the former Governor of the PMA, Dr. Fouad Hamdi Beseiso, and senior staff at the PMA, Ramallah, November 1999.
REFERENCES Abed, G. T. (1999). Beyond Oslo: A viable future for the Palestinian economy. Research in Middle East Economics, 3, 29–58. Arnon, A., & Spivak, A. (1996). Monetary integration between the Israeli, Jordanian and Palestinian economies. Weltwirtschaftliches Archiv, 132(2), 259–277. Economic Intelligence Unit (2000). Occupied territories. Israel Country Report. Ghosh, A. R., Gulde, A., & Wolf, H. C. (2000). Currency boards: More than a quick fix? Economic Policy: A European Forum, 31, 269–321. Gulde, A. (1999). The role of the currency board in Bulgaria’s stabilization. Finance and Development, September, 36–39.
198
MOYARA RUEHSEN
Hamed, O. (2000). Monetary policy in the absence of a national currency and under currency board in West Bank and Gaza Strip. Discussion Paper, September, http://www.palecon.org/masdir/osamaright.html Hanke, S. H. (2000). The disregard for currency board realities. Cato Journal, 20(1), 49–59. Kasa, K. (1999). Why attack a currency board? FRBSF Economic Letter, November, 99(36), 26. Kopke, R. W. (1999). Currency boards: Once and future monetary regimes? New England Economic Review, May/June, 21–37. Lynfield, B. (2000). Launching a Palestinian currency is no small change. Christian Science Monitor, 92(88), 6. Martin, J. (1997). In search of a currency. World Business, January–February, 12. MENA News Agency (1999). Palestinian official discloses plans for future national currency. BBC Summary of World Broadcasts, 20 November. Montagnon, P. (1998). The price of the prize. Financial Times, 12 February. Mundell, R. A. (1961). A Theory of Optimum Currency Areas. American Economic Review, 51(4), 657–665. Naquib, F. M. (1998). The economics of currency boards: The case of the Palestinian economy in the West Bank and Gaza Strip. Cairo, Egypt. Paper presented at the 6th Annual Meeting of the ERF, October. Palestine Monetary Authority (1998). Third Annual Report. Ramallah: PMA Research and Monetary Policies Department. Palestine Monetary Authority (1999a). Statistical Bulletin, 11th Issue. Ramallah: PMA Research and Monetary Policies Department, June. Palestine Monetary Authority (1999b). Fourth Annual Report. Ramallah: PMA Research and Monetary Policies Department. Sadeh, T. (1997). The economic desirability of Middle-Eastern monetary cooperation. World Economy, 20(6), 809–827. Schuler, K. (1998a). Introduction to currency boards. http://users.erols.com/kurrency.html Schuler, K. (1998b). So you want to dollarize. http://users.erols.com/kurrency/soyoudlr.html Shaban, R. A. (1999). The Harsh Reality of Closure. In: I. Diwan & R. A. Shaban (Eds), Development Under Adversity: The Palestinian Economy in Transition (pp. 45–65). Washington, DC: World Bank. The Star (2000). Is a Palestinian currency a threat to the Jordanian dinar. Lexis-Nexis, June 8. UNCTAD (1996a). Private Investment in the Palestinian Territory: Recent Trends and Immediate Prospects. Geneva: United Nations Conference on Trade and Investment. UNCTAD/ECDC/SEU/13. UNCTAD (1996b). Prospects for sustained development of the Palestinian economy: Strategies and policies for reconstruction and development. Geneva: United Nations Conference on Trade and Development. UNCTAD (2001). The Palestinian economy: Achievements of the interim period and tasks for the future. Geneva: United Nations Conference on Trade and Development. United Press International (1996). Palestinian currency launch moves Jordan. Lexis-Nexis, April 1. United Press International (1998). Palestinians plan own currency in May. Lexis-Nexis, November 30. Voice of Palestine (1999a). Palestinian economy losing 600m dollars a year due to lack of currency. BBC Summary of World Broadcasts, May 18.
Choosing an Appropriate Palestinian Monetary Regime
199
Voice of Palestine (1999b). Israeli working group decides to back Palestinian currency. October 19, BBC Summary of World Broadcasts. Wagner, H. (2000). Which exchange rate regimes in an era of high capital mobility? North American Journal of Economics and Finance, 11, 191–203. Williamson, J. (1995). What Role for Currency Boards? Washington, DC: Institute for International Economics. Zak, M. (1999). Rabin would have acted differently. The Jerusalem Post, 8A, October 22.
This page intentionally left blank
200
DOLLARIZATION AND ITS LONG-RUN DETERMINANTS IN TURKEY Irfan Civcir ABSTRACT This chapter explains dollarization process in Turkey by an extended portfolio model where dollarization is determined by the relative rates of return of domestic and foreign currencies denominated assets, expected change in the exchange rate, exchange rate risk, and credibility of current economic policies. The econometrics results are in line with the intuitive predictions of the model. We have found that interest rate differential and the expected exchange rates are the dominant variables in determining dollarization. This chapter also provides evidence of inertia in the process of dollarization in Turkey.
1. INTRODUCTION During the last two decades, the Turkish economy experienced periods of high economic growth as well as severe economic crises. One remarkable characteristic of the Turkish economy has been high and persistent inflation. Starting in 1980 the economy embarked upon several economic reforms and Money and Finance in the Middle East: Missed Opportunities or Future Prospects? Research in Middle East Economics, Volume 6, 201–232 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1094-5334/doi:10.1016/S1094-5334(05)06010-3
201
202
IRFAN CIVCIR
financial liberalization. With the aim of integrating domestic financial markets with the rest of the world, capital account restrictions were eliminated and domestic financial transactions were allowed to be conducted in both domestic and foreign currencies. After this arrangement, the share of assets and liabilities in the banking system denominated in foreign currency increased very rapidly sometimes followed by partial reversals. The currency substitution literature has tried to explain why some economies use a foreign currency instead of the domestic currency in their everyday transactions and bank deposits.1 Basic analytical framework has been the consumer’s portfolio selection model where dollarization is determined by the relative rates of returns of domestic and foreign currency denominated assets. In some cases the rates of return differentials help to explain dollarization trends. However, in other cases the same model is proved to be less successful in explaining swings in dollarization ratios. In Turkey, real rates of return on domestic currency assets have increased significantly relative to foreign currency assets, but the dollarization ratio remained persistently high. This suggests that other variables should be incorporated into these models. Early works on currency substitution started in the late 1970s as an alternative explanation for the excessive variability experienced by the floating exchange rates of the major currencies after the abandonment of the Bretton Woods system. Another branch of literature – developed by Ortiz (1983), Canto (1985), Ramirez-Rojas (1985), Fasano-Filho (1987), Marquez (1987), Canto and Nickelsburg (1987), El-Erian (1988), Melvin (1988), Calvo and Vegh (1992), and Savastano (1996) – has focused on the determinants and characteristics of the currency substitution in developing countries. The nature of the currency substitution in these countries is different from the one analyzed by the literature on developed nations. In particular, in the less developed countries (LDCs) the process tends to be asymmetrical because domestic residents usually hold foreign money in their portfolios, but there is no foreign demand for domestic currencies. Currency substitution in developing countries is usually one of the ultimate consequences of high and variable inflation. High inflation, in turn, is the result of fiscal imbalances and leads to dollarization and eventually to currency substitution. Institutional factors play a crucial role in the process of currency substitution. When the restriction on holding of foreign currency is removed, the dollarization process usually begins with the foreign money replacing the domestic money as a store of value. As high inflation continues, some prices – particularly of real estates, cars, and other ‘‘big ticket’’ items – start to be quoted in foreign currency. After that, some transactions begin to be performed in foreign currency, especially those
Dollarization and its Long-run Determinants in Turkey
203
involving large transfer of funds. However, domestic money seems to retain its functions as a unit of account and medium of exchange in almost all nondurable goods (see Calvo & Vegh, 1992). The literature on currency substitution developed in many directions; one of which is the hysteresis literature. Work by Dornbusch and Reynoso (1989), Guidotti and Rodriguez (1992), Kamin and Ericsson (1993), and Clements and Schwartz (1992) shows that several Latin-American countries have hysteresis in the dollarization ratios. Hysteresis refers to the fact that the dollarization ratio has remained high in the face of declining inflation rates and increased real rates of returns on domestic assets.2 This phenomenon cannot be explained by the traditional currency substitution models which, following Calvo and Rodriguez (1977), assume that foreign assets (non-interest bearing) compete with domestic money in providing liquidity services. In these models, the ratio of foreign-to-domestic currency denominated assets depends positively on the nominal interest rate (when the domestic interest rates increase liquidity provided by domestic money decreases and liquidity provided by foreign money increases (see Sahay & Vegh, 1996). Therefore, a fall in inflation (i.e. a fall in domestic interest rates) should lower the demand for foreign currency relative to domestic currency. To explain hysteresis in these models, it is necessary to assume that there are some costs associated with the process of currency substitution which makes it irreversible. Dornbusch and Reynoso (1989), and Dornbusch, Sturzenegger, and Wolf (1990) argue that the process of financial adaptation is costly (due to sunk cost and learning by doing), which makes it irreversible. In a related explanation, Guidotti and Rodriguez (1992) assume that it is costly to switch between currencies and show that there is a ‘‘band’’ for the inflation differential above which dollarization continues to occur, even if inflation falls. A similar result is derived by Sturzenegger (1997), who emphasizes the public good aspect of currency substitution by assuming that it is costly to make transactions in foreign currency, but only to the extent that others have not yet begun to use it. Uribe (1997) also emphasizes the relation between private cost of transaction and the aggregate degree of dollarization. The literature cited above considers currency substitution, but not dollarization, which is the observable measure of currency substitution. With this observed data, the dollarization ratio should respond only to differences in real rates of returns between domestic and foreign currency denominated assets (see Sahay & Vegh, 1996). A fall in the domestic rate of inflation rate should not affect the dollarization ratio unless it affects the real rates of returns. Hence, de-dollarization would result only from a higher real return on domestic currency denominated assets. Conversely, dollarization would
204
IRFAN CIVCIR
be the consequence of very low real rates of return on domestic currency denominated assets together with the institutional freedom to invest in foreign currency denominated assets at home. In several developing countries, dollarization is associated with low real rates of return on domestic currency denominated assets (see Dornbusch & Reynoso, 1989; Savastano, 1996). In some countries, like Turkey, real rates of return on the domestic currency denominated financial assets were substantially higher than that of foreign currency assets, and dollarization stayed high. In such cases, real rates of return criterion do not seem to explain the behavior of the households in the dollarized country. Therefore, there must be other factors that affect the decision-making process by the public. There is an asymmetric substitution process between domestic and foreign currencies; that is, in a dollarized economy, the demand for foreign currency rises when the local currency depreciates, but falls by a lesser extent when the local currency appreciates (Civcir, 2003a). Yotopoulos (1997) explains currency substitution as an asymmetric reputation between hard and soft currencies. Hard currencies have a good reputation of stability, but soft currencies, in contrast, are expected to depreciate in a free currency market since they lack the reputation of being stable. Under these circumstances, and also when international financial transactions are present, there is asymmetric demand for domestic residents to hold dollars as a store of value – a demand not offset by Americans’ holding of domestic currency as an asset. This asymmetry tends to increase the price of dollar in the home country. This encourages dollarization with domestic residents fleeing from domestic currency in exchange for the dollar, which causes further deprecations. Expectations of depreciation feed on themselves to become self-fulfilling prophecies, in fixed, floating, or pegged exchange rate systems. With the asymmetric reputation still operating, fixed exchange rates have had to fend off the tide of domestic currency denominated asset holders in the country who want to hedge their wealth against depreciation of currency by buying dollars. This was done by offsetting this precautionary demand through increasing the supply of speculative short-term capital that the banking system borrowed in the international market. There is a causal relationship between dollarization and the inflow of hot money, in which the latter is lured by high interest rates in an attempt to thwart the former. Asymmetric substitution behavior of the households can be attributed to cost considerations. Once the fixed cost of an investment in new money management techniques are born, the new product or strategy remains in place and is not discarded even though interest rates, inflation, or depreciation rates decline.
Dollarization and its Long-run Determinants in Turkey
205
In high chronicle inflation countries, the asymmetric behavior of individuals to change currencies (ratchet effect) can be attributed to a costly process of developing, learning, and applying new strategies to beat inflation or get the highest return on their investment. Such strategies are commonly called financial innovations: inter alia, rapid switching between demand deposits and saving deposits in domestic currency, the evolution of high yielding or indexed money substitutes, the efficient use of overdrafts, the application of portfolio optimization methods, and most notably, the flight into foreign currency assets. Over time, an increasing proportion of the public resorts to these forms of financial innovations. There are a few incentives for households and enterprises to switch back to domestic currency after the end of instability or when a higher rate of return on the domestic currency is offered. This causes a more prolonged ratchet effect on the relative demand for foreign and domestic currencies. The credibility of the authorities’ stabilization efforts may shorten or extend the duration of the ratchet effect as well as influence its strength. Only an expectation of a significant decline in inflation or a considerable appreciation of currency, or a substantial increase in the real rates of return on domestic currency denominated assets can overcome the sunk cost of inflation beating strategies and provide enough incentives for households to eventually revert to traditional domestic money balances. In this chapter, we present a model in which an individual demands monetary assets denominated in both currencies taking into account the return that they offer and the risk associated with them. In particular, we include the exchange rate risk associated with the foreign currency deposits (FCDs). For the case of Turkey (1986–1999) we found significant evidence that the dollarization ratio has a cointegrated relationship with the interest rate differential and changes in the exchange rate, a measure of the exchange rate risk and credibility. This chapter is organized as follows. In Section 2, we present the macroeconomic developments and a brief description of the dollarization in Turkey. In Section 3, we present the model and the data. In Section 4, we present the results of cointegration tests for the dollarization ratios for the Turkish case. Conclusions and final thoughts appear in the final section.
2. HISTORICAL PERSPECTIVE During the last three decades, the Turkish economy has experienced periods of high economic growth as well as severe economic crises. During these
206
IRFAN CIVCIR
fluctuations, both internal economic conditions and external economic events played an important role. The current government, which came to power in early 1980, prepared a plan for a fundamental reorientation of Turkey’s economic policy with a medium-term adjustment program. The stabilization and adjustment program aimed not only at stabilizing the economy, reducing government budget deficit, inflation, and the current account deficit, but also at changing the development strategy that Turkey had followed for decades. This new strategy involved moving toward outward orientation and giving an increasing role to market forces. In the 1980s, several reforms were carried out, including financial sector reforms and liberalization of the capital account. The major aim of financial reforms was to integrate the domestic financial system with international markets. Over the course of 1980s, and again in the 1990s, the range of alternative domestic assets was enlarged as new financial instruments emerged (mutual fund shares, corporate finance bills, Treasury Bills and bonds of various maturities, asset-backed securities), and new financial markets were developed. Also, deposit interest rates were liberalized. The emergence of new domestic financial instruments and markets may have controlled the process of dollarization to the extent that these new domestic assets could compete with foreign assets in ensuring the liquidity and maintaining the value of financial wealth. In the period 1980–1983, the reforms aimed at eliminating multiple exchange rate practices, providing the commercial banks with more inclination in managing their foreign exchange positions, and allowing the exporters to hold a portion of their earnings in the form of foreign exchange deposits with commercial banks. A major step toward liberalizing the foreign exchange regime was taken in December 1983. Commercial banks were allowed to engage in foreign exchange operations and transactions in proportion to their foreign exchange liabilities. Finally, banks were allowed to open FCD accounts to residents, and restrictions on foreign travel and investment from abroad were greatly loosened and simplified. The liberalization of the capital account in August 1989 lowered the transaction costs associated with acquiring foreign balances, thereby enabling portfolio changes to be more responsive to the changes in relative returns on assets. Indeed, the package introduced in August 1989 (and amended in 1990) included substantial liberalization of regulations regarding transactions in securities, permission of residents to purchase foreign exchange from banks or other authorized agencies, and the rules regarding the repatriation of cash proceeds from non-residents sales of property.
207
Dollarization and its Long-run Determinants in Turkey
Foreign residents were allowed to purchase or sell any type of Turkish securities – registered on stock exchange, issued upon the permission of the Capital Market Board or issued by public institutions (through intermediary institutions including banks) and to transfer the gains from sales of these securities abroad (through banks and other authorized financial institutions). Moreover, residents were allowed to invest abroad in cash up to US$5 million, or its equivalent in other currencies, through banks and special financial institutions (the export of capital exceeding US$5 million limit was subject to approval). Residents in Turkey were allowed to secure foreign credits abroad in cash or in kind, provided that they used banks or special financial institutions as intermediaries. The widening of the possibilities of obtaining foreign balances may have increased the potential for currency substitution (OECD, 1994).
2.1. Dollarization After allowing commercial banks to engage in foreign exchange operations, and more importantly, to open foreign exchange deposit accounts to residents in December 1983, Turkish banks’ operations in foreign currency have grown substantially and foreign exchange deposits have become a major component of the broad money (see Fig. 1).
0.6 Foreign Currency Deposits (FCD) / (M2 + FCD) FCD / (Total TL Deposits + FCD) 0.5
0.4
0.3
0.2
0.1 1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
Source: IMF, international financial statistics
Fig. 1.
Dollarization ratios.
1996
1997
1998
1999
2000
208
IRFAN CIVCIR
Exchange Rate WPI Inflation Differencial 2.0
1.5
1.0
0.5
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
Source: IMF, international financial statistics
Fig. 2.
WPI inflation differential and exchange rate (annual % change).
Until late 1988, when there were high and volatile domestic–foreign inflation differentials and a depreciation of the nominal exchange rate that broadly reflected movements in inflation, the increase in FCDs could easily be explained by residents’ desire to hedge against inflation (see Fig. 2). However, after the full liberalization of the financial system in 1989, dollarization has continued besides the fact that real rates of return on FCDs were substantially less than Turkish lira (TL) denominated assets.3 This is mainly due to exchange rate risk, political uncertainty, and macroeconomic vulnerability. Fig. 3 illustrates real rates of returns on 3-month FCDs, 3-month time deposits, and Treasury Bills. It is clear from the figure that throughout most of the 1990s real rates of return on TL deposits were higher than that of FCDs. The 10-year average of real rates of return on TL deposits was about 20% while that rate was about 3% for FCDs. The highest rate of return from FCDs is obtained in 1994 (about 73%) when the policy-makers attempted to manipulate the auction rates of Treasury Bills in 1993/1994. As can be seen from Fig. 4, during the period 1986–1988 the share of FCDs in the total deposits increased from about 15% to 27%, while the share of time deposits in the total deposits decreased from about 66% to 42%. The development of sight deposits has shown more variation, but generally their share in the total deposits remained the same in this period. In 1988, a high return on TL deposits decreased the share of FCDs and
Dollarization and its Long-run Determinants in Turkey
209
1.5 T-Bills 3-Month Foreign Currency Deposits 3-Month Time Deposits
1.0
0.5
0.0
-0.5 1986
1987 1988 1989
1990 1991
1992 1993 1994 1995 1996 1997 1998 1999 2000
Source: IMF, international financial statistics
Fig. 3.
0.6
Annual real rates of returns.
Sight Deposits / Total Deposits Time Deposits / Total Deposits Foreign Currency Deposits / Total Deposits
0.5 0.4 0.3 0.2 0.1 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
Source: IMF, international financial statistics
Fig. 4.
Share of deposits.
increased the share of the time deposits. As can be seen from Fig. 4, the share of sight deposits has a declining trend throughout the period. It is important to notice that the share of time deposits and the FCDs follow an asymmetric pattern. This can be interpreted as economic agents moving from time deposits to FCD or vice versa. The process of currency substitution was slowed down for a significant period of time, beginning in 1989. Expectations of a real appreciation due to
210
IRFAN CIVCIR
the improved current account outlook increased demand for TL deposits. Moreover, liberalization of the capital account in August 1989 led to substantial capital inflows that contributed to the appreciation of the real exchange rate (RER). The RER appreciated by 26% during 1989 and inflation fell to around 60% per annum, from 75% in 1988. During 1990, the authorities pursued a tight monetary policy, which, coupled with an ample crop, led to a further reduction in inflation and further RER appreciation. Reflecting these developments, the dollarization ratio declined until August 1990. However, the decline in the dollarization ratio ended in 1991. The downward trend in inflation and nominal exchange rate depreciation was reversed by the increase in oil prices following the outbreak of the Gulf crisis. In addition, as a consequence of the expansionary fiscal policy followed in the period preceding the October 1991 election and large real wage increases, inflation accelerated reaching 71% by the end of 1991. The Gulf crisis in 1991 had increased uncertainties which resulted in an increase in the share of FCDs in the total deposits. This can be evaluated as a precautionary demand for FCDs. However, after 1991 the increases in the FCDs deposits cannot solely be attributed to this factor since the rate of return on TL denominated assets became higher than the rate of return on FCDs. It is important to note that FCDs to total deposit ratio might be a misleading indicator of dollarization, since financial reform alternative instruments have been introduced, such as Treasury Bills and repos, which is the Central Bank sale of government securities with the commitment to buy the securities back at a given price. Repos have been widely used by households as a highly rewarded substitute for time deposits, and by banks to finance offbalance sheet portfolios of Treasury Securities. Another point in this regard is that the maturity of the financial assets has shifted from long run to short run due to lack of stabilization. Consequently, in the context of continued expansionary policies and with the inflationary expectations firmly established given the outcome of labor agreements, FCDs increased steadily and continuously to reach over 50% of Broad money including FCDs (M2Y) at the end of 1993. To control its own balance sheet the Central Bank announced a monetary program which targeted key balance sheet components such as total domestic assets and total domestic liabilities from 1990 to 1992. Combined with a lax fiscal policy, this resulted in high real interest rates and appreciation of RER. During 1992 and 1993, the public sector deficit continued to rise. With no help from fiscal policy, the Central Bank announced a monetary program for 1993. Turkey was in a high interest rate-repressed exchange rate trap (Agenor, McDermott, & Ucer, 1997). Given the limits in
Dollarization and its Long-run Determinants in Turkey
211
domestic borrowing and with the idea of creating a downward pressure on interest rates, the government used foreign borrowing to finance both public and current account deficits. However, in the third-quarter of 1993, the deficits of Turkey looked unsustainable. Foreign creditors warned that they would not finance the deficits as before. It was apparent that crisis was shaping, but the government kept its intention of taking the necessary measures after the municipal elections to be held in March 1994. From November 1993 onward there was a rush to foreign currency. In January 1994, Turkey’s credit rating was lowered to below investment grade paper by two international rating agencies, which resulted in a run on the TL and a further shift into foreign currencies. To calm the panic and keep the exchange rate within certain limits, the Central Bank intervened in the foreign exchange market by selling its reserves and raised the official US$ exchange rate by 13.6% in a day in mid-January 1994. That did not stop the demand for foreign exchange and the Central Bank continued selling its reserves. From November 1993 to the end of March 1994, the sale of foreign exchange reserves amounted to US$7 billion, leaving the reserves at about US$3.3 billion. In April 1994, the increase in the price of foreign currency was about 60%. A number of Treasury Bill auctions were canceled in order to manipulate interest rates. This resulted in a sharp erosion of the financial market confidence. On 5 April 1994, a stabilization program was announced the week after the local elections. In spite of this announcement and a stand by agreement signed with International Monetary Fund (IMF), turbulence in the financial markets continued. Eventually, the government was unable to borrow in the domestic market. Borrowing resumed at the end of May 1994, when 3-month Treasury Bills auctions were conducted, resulting in a 50% return over 3 months. This permitted the debt financing of the public sector borrowing requirement (PSBR) and the repayment of the Central Bank advances. In 1994 the Lira depreciated significantly in real terms due to the overshooting of the exchange rate in the first half of the year. From mid-1994 onward, the exchange rate began to be used as a nominal anchor in the fight against inflation. Initially this policy was perceived to be a temporary one that would last until the end of 1994. Yet the nominal anchor policy continued in 1995 and the RER declined considerably. The period up to September 1995 saw large short-term capital inflows and a widespread shift back into TL instruments by residents on the back of financial market sentiments and still high returns offered on domestic securities, which were sustained by the strong pick-up in the real economy. This resulted in a
212
IRFAN CIVCIR
decline in the dollarization ratios. Monetary and exchange rate developments during 1995 were driven by recovery from the 1994 financial crisis, as well as difficulties in pursuing stabilization efforts. A disappointing inflation performance in 1995 and early 1996 may be largely traced to excessive growth of the money supply in the first half of 1995 and rapid growth of credit to the government after September 1995. During this first period, foreign reserves recovered rapidly reflecting currency substitution back into the TL. However, the general outlook changed again following the September announcement of early elections for December 24, 1995. Financial market sentiments shifted radically from September 1995 onward, leading up to the elections. Rapid money growth nevertheless continued, although its sources shifted. In the face of higher pre-election spending and political uncertainty, the Treasury experienced an increasing difficulty in financing the PSBR and started to use short-run advances from the Central Bank. As expected, Treasury Bill maturities shortened considerably, interest rates rose sharply, and the exchange rate also weakened significantly, reflecting a renewed currency substitution and a reversal of capital flows. From November 1995 onward, the Central Bank began quoting forward foreign exchange rates to stabilize markets, and to provide direction with respect to near term future rates. Sales of foreign exchange, including forward sales, resulted in a drop of some US$4.3 billion in reserves in November and December. Returns on TL instruments (measured in real terms) nonetheless remained high during the last quarter, despite the depreciation/inflation shock and diminished the pressure from both Treasury financing operations and open market operations by the Central Bank (OECD, 1996). The exchange rate, after experiencing a decline of 12% in December, depreciated at a somewhat steadier pace during the first quarter of 1996, and by early July 1996 rates on Treasury Bills had fallen by 120%. The foreign reserves had been partially restored, but currency substitution remained high reflecting the state of financial market confidence. In 1996, the setting of monetary policy was formulated in line with the 1996 budget. The policy framework set a target growth in net domestic assets which required its growth to be significantly below the inflation target. In addition, the Central Bank monitored growth in reserve money and helped to sustain stability in foreign exchange markets. Thus, policy had shifted away from active use of the exchange rate to reduce inflation. At the beginning of the year, the Central Bank announced this objective to the public. Except for two sub-periods, the May–July period of increasing political uncertainty and the November–December period during which the Treasury altered its manner of borrowing, it was observed that the changes
Dollarization and its Long-run Determinants in Turkey
213
in the monthly average foreign exchange basket was close to the monthly inflation rates. At the end of the year, the annual increase in the monthly foreign exchange basket was 77.3%, whereas the nominal depreciation of the TL against the US$ and the German mark was 84.1% and 70.8%, respectively. The annual increase in the wholesale price index, on the other hand, was 84.9% at the end of the year. Inflation remained steady at around 75% in the first half of 1997, but then accelerated to more than 100% by the beginning of 1998. A 3-year stabilization program was launched in early 1998 based on a more active role for monetary policy in the disinflation strategy, commitment to a tight fiscal policy, and abolition of the backward indexation of public sector wages. At the same time the economy proved vulnerable to the emerging market crises. The turmoil in Asia did not have an immediate impact on the economy. However, its effect was felt in the second half of 1998. The Russian crises in the second half of 1998 had an immediate impact on the Turkish economy. The second half of the 1998 was particularly difficult as reflected by the massive outflow of capital, rising real interest rates, and declining economic activity. As a consequence of high real interest rates on TL deposits, the FCDs decreased slightly. However, the high inflationary environment led to an increase in the FCDs in 1999 again.
3. THE EMPIRICAL MODEL Most of the empirical studies aimed at identifying the determinants of currency substitution and dollarization are based on a simple money demand function and incorporate expected exchange rates or inflation and/or interest rate differentials as the main determinants of dollarization. The main assumptions of these models are that the demand for foreign currency by residents is driven by the difference between the real rates of returns on domestic and foreign currencies. These, in turn, depend on the interest rates on domestic currency, foreign currency, expected exchange rates, and expected inflation rates. The econometric analysis in this chapter is based on a simple structural model of type used by El-Erian (1988), Ramirez-Rojas (1985), Rojas-Suarez (1992), and Clements and Schwartz (1992). However, we extend the model by adding real interest rate differentials on domestic and FCDs. Furthermore, a variable measuring the credibility of macroeconomic policies is added to the model. The model used in this chapter can be summarized as
214
IRFAN CIVCIR
follows: CS ¼ CSðse ; ðrd e
rf Þ; sr ; cr; DÞ
(1) d
where CS is the dollarization ratio, s the expected exchange rate, r the real rate of return on domestic currency, rf the real rate of return on foreign currency, sr the exchange rate risk (exchange rate misalignment), cr the credibility, D may be deterministic variables, seasonal dummies, impact dummy variables, or intercept. In this model we experimented with two different measures of dollarization. The first measure uses the FCDs of the residents within the domestic banking system as a percentage of broad money including FCDs (M2Y). The second measure uses the FCDs of residents within the domestic banking system as a percentage of total deposits of the residents within the domestic banking system. As in most of the studies on dollarization, the data only measures foreign currency in the domestic banking system; that is, it excludes foreign currency held abroad and in circulation. Therefore, the dollarization ratio underestimates its actual extent. However, as put forward by Clements and Schwartz (1992) as long as the institutional and legal environment stays the same, the explanation of the cases of the currency substitution based on the narrow definition will be valid in a strict sense. In this chapter, we investigate determinants of dollarization for the period of 1986:1 to 1999:12. FCDs were allowed in 1984; therefore, the legal environment did not change significantly for the period under investigation, and our results will not be affected. However, since the capital account was fully liberalized in 1989, the impact of this factor on the econometric model will be assessed by the Chow test for structural stability of coefficient estimates. The real interest rate differentials are measured by the difference between domestic currency and foreign currency interest rates in the domestic banking system. We also evaluated interest rate differentials between domestic assets and foreign interest rates to check whether capital inflows have any effect on the currency substitution ratio. Previous studies on the dollarization in Turkey have not included these variables in their models (see Akcay, Alper, & Karasulu, 1997). In a high inflation economy, even if the interest rate wedge between domestic and foreign currencies is in favour of domestic currency, the demand for foreign currency may not decline significantly because of the lack of credibility of the policies. For the period investigated in this chapter, there was a number of unsuccessful stabilization efforts by the authorities. In response to these developments economic agents demanded higher real interest rates on public debt instruments, shortened
Dollarization and its Long-run Determinants in Turkey
215
maturity of the debt issued, and increased their holding of FCDs. Based on these considerations, in the empirical analysis, credibility is proxied by average maturity of new domestic non-indexed public debt issues. The expected sign of its coefficient is negative. This means that when the credibility of economic policies increase, dollarization decreases (see Ozatay, 2000). Econometric studies on currency substitution have used different proxies for expected depreciation of the exchange rate. Ortiz (1983) used the difference between the official and RER for expected exchange rate depreciation. Cuddington (1983) used the ratio of the difference between forward and spot exchange rate to spot rate as a proxy for expected depreciation of the exchange rate. Ramirez-Rojas (1985) used the same measure for Mexico, the current differential between inflation rates in Argentina and U.S. for Argentina, and the differential between the domestic interest paid on the deposit denominated domestic currency and the domestic interest paid in FCDs for Uruguay. Clements and Schwartz (1992) used inflation rate differential between U.S. and Bolivia. Akcay et al. (1997) used estimated exchange rate volatility based on the estimation of purchasing power parity (PPP) for Turkey. Boero and Tullio (1996), in their German money demand function, used interest rate differentials between Germany and U.S., and percentage deviation of the mark and dollar rate from PPP to measure the currency substitution. In this chapter the expected depreciation is proxied by the actual monthly changes in the RER. Recently, there has been a renewed effort to understand the empirical determinants of currency crises. Most of the empirical studies have shown that the RER deviation from trend, or other forms of calculating RER misalignment, is an important variable in predicting crises. Kaminsky, Lizondo, and Reinhart (1998) find the RER to be the most reliable indicator in predicting future currency crises (in terms of their noise-to-signal ratio and longest lead time).4 This result can be interpreted as a sign that relative price is a key summary variable of several underlying fundamentals. Goldfajn and Valdes (1996) analyzed a large set of RER appreciation derived from an initial sample of monthly RER for 93 countries from 1960 to 1994 to evaluate whether RER misalignment leads to nominal devaluation (or future crises). They found that the probability of eliminating misalignment without nominal devaluation for different degrees of misalignment is extremely low. That means real overvaluation is usually corrected through nominal devaluation rather than inflation differentials. Goldfajn and Valdes (1998) provided empirical evidence that market participants do consider overvaluation in their formation of expectation.
216
IRFAN CIVCIR
The discussion above shows that RER misalignment can be taken as a devaluation risk. Therefore, in this chapter we will use RER misalignment based on PPP as a proxy for exchange rate risk. In constructing RER the wholesale price index is used. Misalignment is calculated as the deviation of the actual series from predicted series based on a regression of the RER on constant and trend (see Civcir, 2003b,c). An increase in the RER reflects an appreciation of the domestic currency. All the series are monthly and seasonally unadjusted, and the estimation sample extends from 1986:1 to 1999:12. Some evidence in the literature suggests that the length of the period is more important than the frequency of the data. We used monthly data for two reasons; some of the monetary variables are available as of 1986:1, and in a high inflation economy, economic agents tend to make their decisions quite frequently in small intervals.
4. ESTIMATION RESULTS This section presents unit root test for the variables of interest. Then Johansen’s (1988; 1991) maximum likelihood procedure is applied to test for cointegration among the variables of the model.
4.1. Integration Before modeling dollarization presence of univariate, unit roots were tested to determine the order of integration of the variables with and without trend term. Table 1 reports Augmented Dickey–Fuller (ADF) (1981) and Dickey– Fuller f-test statistics. The lag lengths for the ADF tests are determined by the akaike information criteria (AIC) and likelihood ratio (LR) statistics. When the serial correlation problem detected further, lag(s) added to eliminate the serial correlation. Tests are carried out with constant and trend, with constant, and without constant term included in the ADF regression. In this way the unit root allows for the alternative hypothesis trend stationarity and non-zero intercept on the series. The trend stationarity implies that the deviation of the series from a linear function of time follows a stationary process. Test results show that all the variables contain unit roots. That is, all the variables are non-stationary in levels, but stationary after first differencing.5 Dickey–Fuller f test statistics shows that all of the variables are difference stationary.
217
Dollarization and its Long-run Determinants in Turkey
Variables
Table 1.
Unit Root Test Results.
K
Levels A
log(FCD) – log(M2Y) log(FCD) – log(TDEP) (rd rf) pWPI pCPI se sr Cr 1% critical value* 5% critical value
12 12 9 11 11 10 3 12
2.496 2.222 3.381 2.908 2.941 3.537* 2.486 1.782 4.026 3.443
B 1.707 1.716 2.262 2.927* 2.992* 3.438* 2.453 1.005 3.478 2.882
F3
F1
4.402 5.200 5.600 3.360 3.849 6.663 3.395 2.353 8.730 6.490
4.679 5.372 5.639 5.420 6.636 11.592 6.633 1.748 6.700 4.710
C 1.316 1.214 1.915 0.088 0.387 1.242 2.453* 0.569 2.580 1.942
Notes: (1) k is the number of lagged dependent variables in the ADF regression. (2) Columns A, B, and C give the t-statistics from ADF regression including constant and trend, constant and without constant, respectively. (3) The critical values are from MacKinnon (1991). Source: Author’s estimates. *,** Denote rejection at 5% and 1% critical values. CPI: Consumer price index, WPI: Whole sale price index.
4.2. Cointegration There are number of alternative ways to conduct cointegration tests (for detailed survey see Maddala & Kim, 1998). Earlier tests for cointegration are based on Engle and Granger (1987). However, Banerjee, Hendry, and Mizon (1996) have noted that the small sample property of the Engle– Granger method is poor. Additionally, if the regressors in the model are endogenous and/or errors exhibit serial correlation, then the asymptotic distribution of the coefficients will depend on the nuisance parameters. Researchers have demonstrated that, in testing an equilibrium relationship, econometric methods are robust to simultaneity bias and potential endogeneity can make a significant difference to the outcome (Cheung & Lai, 1993). The cointegration tests in this chapter are conducted by using the reduced rank procedure developed by Johansen (1988), and Johansen and Juselius (1990). This method should produce asymptotically optimal estimates because it incorporates a parametric correction for serial correlation (which comes from underlying vector autoregressive, VAR) and the system nature of the estimator means that the estimates are robust to simultaneity bias. Alternatively, Phillips and Hansen’s (1990) methods are used; however, these estimators are not informative about the number of cointegration
218
IRFAN CIVCIR
relationships in a given vector. The Johansen method, based on maximum likelihood optimization, provides more robust results, especially when there are more than two variables involved (Gonzalo, 1994). Optimality of this method has been shown by Phillips (1991) in terms of symmetry, unbiasedness, and efficiency properties. Furthermore, it does not suffer from problems associated with normalization (see Johansen, 1995), and it is robust to departure from normality (Cheung & Lai, 1993; Johansen, 1995). The Johansen method detects the number of cointegrating vectors in non-stationary time series and allows for hypothesis testing regarding the elements of cointegrating vectors and loading matrix. Basically, the cointegration analysis is performed in the following unrestricted VAR framework: Dxt ¼
k X
Gi Dxt
i
þ Fxt
1
þ CD þ t
(2)
i¼1
where xt is the vector of non-stationary (in levels) variables, the matrix F has reduced rank equal to r and can be decomposed as F ¼ ab0 ; where a and b are pxr full-rank matrices and contain adjustment coefficients and the cointegrating vectors, respectively. D is a deterministic variable which may include constant term, the linear trend, seasonal dummies, and impulse dummies. Finally, the error term follows a normal process. In order to test for the number of cointegration relationships amongst the variables, Johansen (1988) and Johansen and Juselius (1990) provide two different tests to determine the number of cointegrating vectors, namely trace and maximum eigenvalue tests. In the trace test, the null hypothesis is that there are at most r cointegrating vectors and it is tested against a general alternative. In the maximum eigenvalue test, the null hypothesis of r cointegrating vector is tested against r þ 1 cointegrating vectors. Once we determine r, the number of relationships, we can do hypothesis testing on both loadings and cointegrating vectors. Restrictions can be imposed on the coefficients to test alternative theory-based hypotheses on the long-run value of variables. One problem with the Johansen and Juselius procedure is that it is not able to identify exactly the parameters in a and b matrices. Only if there is just one cointegrating vector found can we truly make concrete conclusions about any unique long-run relationship between the variables, otherwise we cannot. In Johansen’s cointegration analysis the deterministic part of the system is also important. Johansen (1992b) describes a procedure for selecting
Dollarization and its Long-run Determinants in Turkey
219
between a model with an unrestricted constant and one with restricted constant, or no linear trend. The basic idea is to estimate the restricted and unrestricted version of the model and accept the model with the fewest cointegrating vectors. If both models accept the same number of cointegrating vectors, then the restricted model should be used. Similarly, when there is an equivalent number of vectors in each model, one can use LR test statistics to test whether the restriction can be rejected (if we reject the null hypothesis of restricted constant, then constant term should enter to vector error-correction model (VECM) as unrestricted). However, even if the restriction of the constant to the cointegrating space is accepted by the LR test, the validity of the restriction should be judged carefully, since the restriction of the constant term to cointegrating space implies that the average first difference of the each of the endogenous variables is zero. This might be a strong assumption for some variables. Doornik, Hendry, and Nielsen (1998) statistically analyzed overspecified trend in the cointegration space and suggested that adopting a model that includes a trend in the cointegration space has low cost even when data-generating process (DGP) does not display trend. They found that including an unrestricted trend was problematic (for further details see Doornik et al., 1998). Therefore, we have restricted the trend in the cointegration space. Concerning the treatment of impulse indicator variables, Hendry and Doornik (1994) strongly recommended that these be entered unrestricted if they are used to establish an estimate of the innovation variance. They certainly advise against their restriction to the cointegrating space. Following these suggestions, we have included dummy variables unrestrictedly into the cointegration space.
4.3. Cointegration Results The Johansen procedure is used to determine the rank r and to identify a long-run dollarization relationship amongst the cointegrating vectors. The number of lags used in the VAR is based on the evidence provided by both the LR test6 and AIC; however, in the case of serial correlation a sufficient number of lags are introduced in order to eliminate the serial correlation of the residuals. The cointegration tests amongst cs, se, ðrd rf Þ; sr and cr include 10 lags in the VAR, a set of monthly centered seasonal dummy variables (see Johansen, 1995), a constant term and further, the estimates of VAR also include two impulse dummy variables: D8889 is included to capture the interest rate intervention during the period 1988:10
220
IRFAN CIVCIR
Table 2.
Full System Diagnostics.
Statistics
Values [P values]
Vector portmanteau 12 lags VAR 1–7 F(112, 284) Vector normality w2 ð8Þ Vector heteroskedasticity F(840, 112)
91.016 1.0321 [0.4116] 155.92 [0.0000] 0.18844 [1.0000]
Source: Author’s estimates. Indicates 1% significance level.
to 1989:3, and D94 is included to capture currency crises in 1994. We have restricted the trend and credibility variable in the cointegration space. The diagnostics in the form of vector statistics (and single equation statistics, which we do not report here) reported in Table 2 indicates that our VAR model is satisfactorily and a close approximation to actual DGP, apart from some non-normality of residuals. Table 3 reports the estimates of the Johansen procedure and standard statistics. In determining the number of cointegrating vectors we used the degrees of freedom adjusted version of the maximum eigenvalue and trace statistics, since in the existence of small samples with too many variables or lags the Johansen procedure tends to overestimate the number of cointegrating vectors (see Cheung & Lai, 1993; Gonzalo & Pitarakis, 1994). These test statistics strongly reject the null hypothesis of no cointegration in favour of one cointegration relationship. Table 3 also reports standardized eigenvectors, b0 ; and adjustment coefficients, a: The first row of b0 is the estimated cointegration vector (see Fig. 5), and can be written as: logðFCD=M2Y Þ ¼ 0:004 trend ðstandard errorsÞ
ð0:0017Þ
þ 0:005s
0:112ðrd ð0:0717Þ
r
ð0:0034Þ
0:001cr ð0:0003Þ
rf Þ þ 0:344s ð0:0642Þ
ð3Þ
The result is in line with the intuitive predictions of the dollarization model, and shows a negative relationship between the dollarization ratio and the real interest rate differential between 3-month deposit and 3-month FCD and credibility variable, and shows positive relationships between the dollarization ratio and the expected exchange rate and exchange rate risk. There is also a positive linear trend. From these results it can be seen that the interest differential and the expected exchange rates are the most dominant variables in determining dollarization. Although the credibility variable is highly significant, in terms of its magnitude it is relatively very small.
221
Dollarization and its Long-run Determinants in Turkey
Cointegration Tests and Related Statistics.
Table 3.
Contegration tests Eigenvalues
Hypothesis
L-max
%95 c.v.
L-trace
%95 c.v.
r¼0 ro1 ro2 ro3
47.94** 12.99 10.19 4.478
31.5 25.5 19 12.3
75.6** 27.66 14.67 4.478
63 42.4 25.3 12.3
0.333868 0.104255 0.082733 0.0372385
Standardised eigenvetors ðb0 Þ Log(FCD)-log(M2Y)
(rd rf)
1.000 2.916 1.033 1347.600
0.112 1.000 1.603 960.660
se
sr
0.344 0.849 1.000 1040.200
cr
0.005 0.007 0.130 1.000
Trend
0.001 0.001 0.001 3.699
0.004 0.008 0.029 9.990
Standardised adjustment coefficients ðaÞ 0.00697
0.005116
0.006422
0.000001
10.28700 10.26300 4.12960
1.000200 1.075800 1.075900
0.429960 0.455350 0.387210
0.000407 0.000517 0.000143
Log(FCD)-log(M2Y) (rd rf ) se sr
Weak exogeneity test statistics (Ho: weakly exogeneous)
LR-test, w2 ð1Þ
(rd rf)
se
sr
28.856 [0.0000]**
26.834 [0.0000]**
9.883 [0.0017]**
Joint weak exogeneity tests LR-test, w2 ð3Þ
48.701 [0.0000]** Multivariate unit root tests (Ho: variable stationary)
log(FCD)-log(M2Y) LR-test, w2 ð5Þ
50.728 [0.0000]**
(rd rf)
Se
sr
Cr
45.804 [0.0000]**
25.652 [0.0001]**
60.424 [0.0000]**
57.328 [0.0000]**
Notes: 1) The estimation period is 1986:1-1999:12. VAR includes 10 lags on each variables, a constant term, centred seasonal monthly dummy variables, D8889 dummy and D94 dummy variables. MDAY and Trend variables are restricted to the cointegration space. 2) The l-max and l-trace are maximum eigenvalue and trace test statistics, adjusted for degrees of freedom. The critical values are taken from Osterwald-Lenum (1992). 3) The multivariate stationarity, weak exogeneity and significance tests statistics are evaluated by assuming a single cointegration vector. 4) The values in [U] are p-values. The * and ** indicate rejection of likelihood ratio tests at 5% and 1% significancee levels, respectively
222
IRFAN CIVCIR
0
ecm 1
-5
-10
-15 1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
Source: Author’s estimates
Fig. 5.
Cointegration relationship.
The semi-elasticity of exchange rate variable (34.44%) implies that in the long run a reduction in the monthly rate of the exchange rate depreciation by a percentage point would decrease the ratio of FCDs to broad money by about 34%. Similarly, the equation indicates that a decrease in the spread of dollar and domestic currency deposits by 1% point would in the long run decrease the share of foreign money by about 11% points. The significance of the positive linear trend in the regression can be taken as evidence of dollarization hysteresis in Turkey. The Turkish government started to implement a disinflation program at the beginning of 2000 and the FCD to M2Y ratio decreased. The main element of the stabilization program is a pre-fixed exchange rate; therefore, depreciation of the exchange rate is also reduced significantly (0.009 per month). These developments suggest that there is a chance of reducing dollarization in Turkey. However, the interest rates on TL deposits and Treasury Bills are already positive in real terms for a long period. Therefore, a further increase in the real rate of return on domestic currency denominated assets may not significantly lower the degree of dollarization. The equilibrium correction coefficient is significant and small in magnitude. It indicates that economic agents do not immediately and fully adjust their holdings to variations in the relative rates of return on foreign currency balances. The adjustment coefficient, 0.007, indicates that only about 1% of adjustment between actual and desired balances occurs in the current month. Consequently, actual holdings in the previous month have a strong effect on the behavior of the current months. The significance and small magnitude of this coefficient is consistent with inertia in the process of dollarization in Turkey and pose difficulty in reversing the dollarization process.
Dollarization and its Long-run Determinants in Turkey
223
The inclusion of a variable measuring the interest rate differentials and acting as a proxy for the expected exchange rate in the same model might be considered odd, as interest rate differentials themselves could be considered as a proxy for expected depreciation. However, in this study the actual depreciation of the exchange rate is used as an expected depreciation. When we excluded the actual depreciation from the regression, no cointegration relationship was found and the interest rate differential becomes insignificant. Therefore, we retained the actual depreciation of the exchange rate and interest rate differentials in our model. In Turkey, the RER appreciated significantly after the full liberalization of capital accounts and after a number of stabilization efforts failed due to the lack of fiscal adjustment measures. Under these circumstances, economic agents have demanded substantial risk premiums for holding financial assets denominated in the domestic currency and shifted their portfolio in favour of foreign currency. To examine the impact of this factor on dollarization, we included trend deviation of RER as a proxy for exchange rate risk in our model. The results show that our measure of the risk variable is significant and has correct sign. Since the real rates of return on domestic assets have been positive in most of the period under investigation and the dollarization ratio had an upward trend, we also investigated dollarization hysteresis by including a ratchet variable which was the highest rates of deprecations in each month. However, this variable was not significant. To check the robustness of our results we further investigated the dollarization issue in Turkey by replacing the dollarization ratio with the share of FCDs in total deposits (TDEP). The results are very similar to the previous results (see Table A.1 in the appendix). We can test various hypotheses on the parameters of a matrix. The first interesting aspect is represented by the possibility of identifying long-run weak exogeneity of the variable(s) with respect to the parameters of equilibrium relationships. If the cointegration vector does not have any influence on a particular variable, then all the weights will be equal to zero. In this case, the variable is said to be long-run weakly exogenous for the long-run parameters and can be considered as driving the dynamics of the system as a whole. Table 3 also presents a weak exogeneity test of a given variable for the cointegrating vector. If a given row in a is equal to zero, disequilibrium in the cointegrating vector does not feed back directly onto the corresponding variable. The test results show that weak exogeneity is rejected for all variables. The weak exogeneity results justify a system approach to analyzing cointegration relationship and guides us in answering the question of
224
IRFAN CIVCIR
whether we have to model the dollarization in a single equation or in a system context. Finally, Table 3 also reports multivariate stationarity of the given variable. The tests are based on the assumption that there is only one cointegrating vector. Here, the null hypothesis is the stationarity of the variable. Furthermore, since the test is multivariate and includes a larger set of information, these statistics may have a stronger power against the ones based on the univariate test (see Johansen, 1995). We reject the null hypothesis of stationarity of all the variables.
4.4. Constancy Test on the Long-Run Equilibrium Parameter constancy is an additional and crucial issue to ensure a well-specified equation. The potential for parameter stability increases significantly during and after financial crises, and the factors affecting the dollarization ratio may change. In this section, we report the graphical instability test7 in Fig. 6. The first graph (a) shows one-step residuals and the
0.1
Res 1 Step
0 -0.1 1992 10
1993 5%
1994
1995
1996
1997
1998
1999
2000
1995
1996
1997
1998
1999
2000
1995
1996
1997
1998
1999
2000
1up CHOWs
7.5 5 2.5 1992 3
1993 5%
1994
Ndn CHOWs
2 1
1992
1993
1994
Source: Author’s estimates
Fig. 6.
Constancy tests on the long-run dollarization relationship.
Dollarization and its Long-run Determinants in Turkey
225
dollarization equation standard errors, second graph (b) shows sequentially estimated one step ahead Chow statistics. The third graph (c) shows breakpoint Chow test. In the first graph, residuals lie inside the 72 standard error bands indicating that parameter constancy is not violated. In the break-point Chow test for the sequence of 1991:10 to 1999:12 none is statistically significant at the 5% level indicating that constancy of the parameters cannot be rejected for the whole sequences of forecasts. In the one-step Chow test, only three points are above the 5% significance level, but these are numerically small and occur only in the 1994 crisis period. Constancy of the parameters indicates that, in general, the dollarization process in the long run remained unchanged over the sample period.
5. SUMMARY AND CONCLUSIONS Turkey experienced periods of macroeconomic instability and several stabilization efforts failed due to the structural weaknesses of the economy. Fundamental reforms, which are essential for moving toward a well-functioning market economy, have not been implemented or have been delayed for long period. Therefore, chronic high fiscal deficit and inflation remained in place for a prolonged period and the credibility of TL diminished significantly. Under these circumstances private economic agents responded by increasing their holding of FCDs in the domestic banking system to hedge against inflation and demanded high real interest rates on government’s debt instruments. In this chapter we tried to explain the dollarization process in Turkey by using an extended portfolio model. In this literature, dollarization is explained by a simple portfolio model where the dollarization ratio is a function of rates of return differential. In some cases the rate of return differentials help to explain dollarization trends (see Balino et al., 1999). However, in other cases the same model is proved to be less successful in explaining swings in dollarization ratios. In Turkey real rates of returns on domestic currency assets have increased significantly relative to foreign currency assets, but the dollarization ratio remained persistently high. This suggests that other variables should be incorporated into these models. In this study we extended the simple portfolio model where dollarization is determined by the relative rates of returns of domestic currency and foreign currency denominated assets, expected change in the exchange rate, exchange rate risk, and credibility of current economic policies.
226
IRFAN CIVCIR
The econometric results are in line with the intuitive predictions of the dollarization model, where there is a negative long-run relationship between the dollarization ratio, real interest rate differential between 3-month deposit and 3-month FCD, and the credibility variable. There are also positive long-run relationships between the dollarization ratio, and the expected exchange rate and exchange rate risk. We have found that interest differentials and the expected exchange rates are the most dominant variables in determining dollarization. The significance of a positive linear trend in the regression can be taken as an evidence of dollarization hysteresis in Turkey. Furthermore, the equilibrium correction coefficient is significant and small in magnitude. It indicates that economic agents do not immediately and fully adjust their actual holding to desired balances. Consequently, actual holdings in the previous month have a strong effect on the behavior of the current months. This can be taken as evidence of inertia in the process of dollarization in Turkey and pose difficulty in reversing the dollarization process. The inflationary expectations are highly sensitive to the exchange rate as devaluation has a high degree of pass-through to domestic prices in Turkey. Therefore, the policy implication of this study is that a credible disinflation program with an exchange rate anchor may reduce the dollarization ratio. However, a timely, smooth transition from a pegged exchange rate system to a more flexible one is difficult. Authorities resist abandoning the fixed exchange rate due to fear of damaging their policy credibility, but the longer the peg is retained, the more vulnerability builds up in the financial system and in the external account. In the end, a country exits from the peg as a result of crises, which increases dollarization. In Turkey the financial system is also highly dollarized (i.e. deposit dollarization as well as liability dollarization). There is a substantial capital inflow in the form of short-term borrowing, and devaluation will worsen the balance sheet of domestic banks. Devaluation in a context of weak banking system and large foreign exchange exposure in the private sector can damage the financial system and disrupt real activity. This implies that in highly dollarized economies devaluation as a policy option may be very costly. Dollarization also reflects the absence of macroeconomic stability and the existence of distortions in the financial markets. Under these circumstances, dollarization may complicate stabilization and cause extra volatility. Macroeconomic stability and fiscal disciplines are necessary for de-dollarizing the economy. Other direct measures should be avoided. Direct measures through forced conversion to the domestic currency have severe adverse affects. This will substantially diminish the government’s credibility and
Dollarization and its Long-run Determinants in Turkey
227
increase the confiscation risk perceived by the domestic residents. This will also increase the demanded interest rates later, as occurred in several LatinAmerican countries (see Savastano, 1990; Berg & Borensztein, 2000). The recent experience in Argentina shows that exchange rate stabilization brings stability in the short run at the price of instability in the long run (see Eichengreen, 2002). After the hyperinflation experience in 1980s, Argentina’s dollar-based ‘‘currency board’’ system increased the credibility of the price stability in the short run. However, this also exposed the economy to fluctuations in the relative value of major currencies. Starting in the mid1990s, Argentina’s competitiveness worsened relative to other countries which resulted in an increasing current account deficit and mounting external debt. External debt also reflected substantial deficit of central and provincial governments. Domestic residents and foreign investors saw an unsustainability of the current account deficit and external debt, and increased their expectation of devaluation sometimes in the near future. The convertibility law allowed them to shift their domestic currency into dollars and then to take the dollars out of country. The result was a loss of dollar reserves at the central bank and devaluation domestic currency in January 2002. This will diminish governments credibility in the future and increase the inertia in the dollarization.
NOTES 1. The terms ‘‘currency substitution and dollarization’’ have been used to depict the same phenomenon. However, the term ‘‘dollarization’’ indicates that a foreign currency serves as a unit of account or as a store of value, and not necessarily as a medium of exchange. In high inflation countries, foreign currency is first used as a store of value or unit of account and only at the later used as a medium of exchange. That is, currency substitution is normally the last stage of the dollarization process. More importantly, in practice, data on foreign currency circulating in the economy usually does not exist, dollarization ratios are used as an indicator of currency substitution (see Calvo & Vegh, 1992). 2. Hysteresis is also called ratchet effect which is used in the money demand literature. 3. There are many data problems in measuring the actual amount of currency substitution. The ideal measurement would include foreign banknotes circulating as medium of exchange and store of value in the economy, as well as checking accounts and short-term deposits denominated in foreign currency in the domestic banking system and abroad. 4. Kamisky et al. (1998) also show that interest rate differentials are not useful in predicting crises. Interest differentials do not adequately reflect expected depreciation, possibly due to the fact that changes in the interest differentials may reflect
228
IRFAN CIVCIR
short-run monetary policies that increase domestic interest rates or changes in the risk premium. These factors may produce enough noise that prevents extraction of reliable expected depreciation measures. 5. Following the suggestion of Dickey and Fuller (1981) the unit root tests are first performed for two roots, and if two roots are rejected then single unit root will be tested for. Results showed that none of the variables seems to show evidence of two unit roots, these results are not reported here but available from the author. Phillips and Perron (1988) unit root tests are also performed and the results are very similar to the ADF tests. Therefore, they are not reported here but available from the author upon request. 6. The LR ¼ trace (rest) (r) – trace (unrest) (r) w2 ðp rÞ; where p is the number of variables and r is the number of cointegrating vectors. 7. All the tests presented here employ the null hypothesis of parameter constancy.
ACKNOWLEDGMENTS The author thanks Sheetal K. Chand, Ragnar Nymoen, Ahmet Akıncı, and an anonymous referee for very useful discussion and comments. Author gratefully acknowledges financial support from Norwegian Research Council, and Faculty and Staff at the Department of Economics, Oslo University. The usual disclaimer applies.
REFERENCES Agenor, P. R., McDermott, C. J., & Ucer, E. M. (1997). Fiscal imbalances, capital inflows, and the real exchange rate: The case of Turkey. International Monetary Fund Working Paper WP/97/1. Akcay, O. C., Alper, C. A., & Karasulu, M. (1997). Currency substitution and exchange rate instability: The Turkish case. European Economic Review, 41, 827–835. Balino, T., Bennett, A., & Borensztein, E. (1999). Monetary policy in dollarized Economics. Occasional Paper 171. Washington, DC: International Monetary Fund. Banerjee, A., Hendry, D. F., & Mizon, G. E. (1996). The econometric analysis of economic policy. Oxford Bulletin of Economics and Statistics, 58(4), 573–600. Berg, A., & Borensztein, E. R. (2000). The pros and cons of full dollarization. International Monetary Fund Working Paper WP/00/50. Boero, G., & Tullio, G. (1996). Currency substitution and the demand for Deutsche Marks before and after the fall of Berlin wall. In: P. Mizen & E. J. Pentecost (Eds), The macroeconomics of international currencies. Cheltenham: Edward Elgar. Calvo, G. A., & Rodriguez, C. A. (1977). A model of exchange rate determination under currency substitution and rational expectations. Journal of Political Economy, 85, 617–625. Calvo, G. A., & Vegh, C. A. (1992). Currency substitution in developing countries: An introduction. International Monetary Fund Working Paper 92/40.
Dollarization and its Long-run Determinants in Turkey
229
Canto, V. A. (1985). Monetary policy, ‘‘dollarization,’’ and parallel market exchange rates: The case of the Dominican Republic. Journal of International Money and Finance, 4, 507–521. Canto, V. A., & Nickelsburg, G. (1987). Currency substitution: Theory and evidence from Latin America. Boston: Kluwer Academic Publishers. Cheung, Y.-W., & Lai, K. S. (1993). Finite-sample sizes of Johansen’s likelihood ratio tests for cointegration. Oxford Bulletin of Economics and Statistics, 55(3), 313–328. Civcir, I. (2003a). Broad money demand and currency substitution in Turkey. Ekonomicky Casopis/Journal of Economics, 7/51, 823–841. Civcir, I. (2003b). Before the fall was the Turkish lira overvalued. Eastern European Economics, 41(2), 69–90. Civcir, I. (2003c). The monetary model of the exchange rate under high inflation: The case of Turkish lira–US dollar. Finance a Uver, 53(3–4), 113–129. Clements, B., & Schwartz, G. (1992). Currency substitution: The recent experience of Bolivia. International Monetary Fund Working Paper WP/92/65. Cuddington, J. (1983). Currency substitution, capital mobility, and money demand. Journal of International Money and Finance, 2, 111–133. Dickey, D., & Fuller, W. (1981). Likelihood ratio tests for autoregressive time series with unit root. Econometrica, 49, 1057–1072. Doornik, J. A., Hendry, D. F., & Nielsen, B. (1998). Inference in cointegrating models: UK M1 revisited. Journal of Economic Surveys, 12(5), 533–572. Dornbusch, R., & Reynoso, A. (1989). Financial factors in economic development. NBER Working Paper No. 2889. Dornbusch, R., Sturzenegger, F., & Wolf, H. (1990). Extreme inflation: Dynamics and stabilization. Brooking Papers on Economic Activity, 2, 1–84. Engle, R. F., & Granger, C. W. J. (1987). Cointegration and error correction: Representation, estimation and testing. Econometrica, 55(2), 251–276. El-Erian, M. (1988). Currency substitution in Egypt and the Yemen Arab Republic. International Monetary Fund Staff Papers, Paper No. 35, 85–103. Eichengreen, B. (2002). Crisis prevention and management: Any new lessons from Argentina and Turkey. Background paper for the World Bank’s Global Development Finance 2002. Fasano-Filho, U. (1987). Currency substitution and liberalisation. Aldeshot: The Case of Argentina Gower Publishing Co. Goldfajn, I., & Valdes, R. O. (1996). The aftermath of appreciations. NBER Working Paper No. 5650. Goldfajn, I., & Valdes, R. O. (1998). Current account sustainability: Are currency crises predictable? European Economic Review, 42, 873–885. Gonzalo, J. (1994). Five alternative methods of estimating long-run equilibrium relationships. Journal of Econometrics, 60, 203–233. Gonzalo, J., & Pitarakis, J. Y. (1994). Comovements in large systems. Universite Catholique de Louvain CORE Discussion Paper, 9465. Guidotti, P., & Rodriguez, C. A. (1992). The dollarization in Latin America: Gresham law in reverse. International Monetary Fund Staff Papers, Paper No. 39(3), 518–544. Hendry, D. F., & Doornik, J. A. (1994). Modelling linear dynamic econometric systems. Scottish Journal of Political Economy, 41, 1–33. Johansen, S. (1988). Statistical analysis of cointegrating vectors. Journal of Economic Dynamics and Control, 12, 231–254.
230
IRFAN CIVCIR
Johansen, S. (1991). Estimation and hypothesis testing of cointegration vectors in Gaussian vector autoregressive models. Econometrica, 59(6), 1551–1580. Johansen, S. (1992b). Testing weak exogeneity and the order of cointegration in UK money demand data. Journal of Policy Modelling, 14(3), 313–334. Johansen, S. (1995). Likelihood-based inference in cointegrated vector autoregressive models. Oxford: Oxford University Press. Johansen, S., & Juselius, K. (1990). Maximum likelihood estimation and inference on cointegration – with applications to the demand for money. Oxford Bulletin of Economics and Statistics, 52, 169–210. Kamin, S., & Ericsson, N. (1993). Dollarization in Argentina. Board of Governors of the federal reserve system. International Finance Discussion Paper No. 460. Kaminsky, G. Lizondo, S., & Reinhart, C. M. (1998). Leading indicators of currency crises. International Monetary Fund Staff Papers, Paper No. 45, 1–48. MacKinnon, J. G. (1991). Critical values for cointegration tests. In: R. F. Engle & C. W. J. Granger (Eds), Long-run economic relationships: Readings in cointegration (pp. 267–276). Oxford: Oxford University Press. Maddala, G. S., & In-Moo Kim (1998). Unit root cointegration and structural change. Cambridge: Cambridge University Press. Marquez, J. (1987). Money demand in open economies: A currency substitution model for Venezuela. Journal of International Money and Finance, 6, 167–178. Melvin, M. (1988). The dollarization of Latin America as a market-enforced monetary reform: Evidence and implications. Economic Development and Cultural Change, 36, 543–558. OECD (1994). Economic surveys: Turkey. Paris. OECD (1996). Economic surveys: Turkey. Paris. Ortiz, G. (1983). Currency substitution in Mexico: The dollarization problem. Journal of Money, Credit and Banking, 15, 174–185. Osterwald-Lenum, M. (1992). A note with quantiles of the asymptotic distribution of the maximum likelihood cointegration rank test statistics. Oxford Bulletin of Economics and Statistics, 54, 461–472. Ozatay, F. (2000). A quarterly macro econometric model for a highly inflationary and indebted country: Turkey. Economic Modelling, 17/1, 1–11. Phillips, P. C. B. (1991). Optimal inference in cointegrated systems. Econometrica, 59(2), 283–306. Phillips, P. C. B., & Perron, P. (1988). Testing for a unit root in time series regression. Biometrika, 75, 335–346. Phillips, P. C. B., & Hansen, B. E. (1990). Statistical inference in instrumental variables regression with I(1) processes. Review of Economic Studies, 57, 99–125. Ramirez-Rojas, C. L. (1985). Currency substitution in Argentina, Mexico and Uruguay. International Monetary Fund Staff Papers, Paper No. 32, 627–667. Rojas-Suarez, L. (1992). Currency substitution and inflation in Peru. Revista de Ana´lisis Econo´mico, 7(1), 153–176. Sahay, R., & Vegh, C. A. (1996). Dollarization in transition economies: Evidence and policy implications. In: P. Mizen & E. J. Pentacost (Eds), Macroeconomics of international currencies, theory, policy and evidence. Brookfield: Edward Elgar. Savastano, M. A. (1990). Speculative attacks and currency substitution under managed exchange rate regimes in developing countries. Los Angeles: University of California. Savastano, M. A. (1996). Dollarization in Latin America: Recent evidence and some policy issues. International Monetary Fund Working Paper WP 96/04.
231
Dollarization and its Long-run Determinants in Turkey
Sturzenegger, F. (1997). Understanding the welfare implications of currency substitution. Journal of Economic Dynamics and Control, 21, 391–416. Uribe, M. (1997). Hysteresis in a simple model of currency substitution. Journal of Monetary Economics, 40, 185–202. Yotopoulos, P. A. (1997). Financial crises and the benefit of mildly repressed exchange rates. Working Paper Series in Economics and Finance No. 202, Stanford University.
APPENDIX Table A.1.
Cointegration Tests and Related Statistics. Cointegration tests
Eigenvalues 0.326315 0.110898 0.078765 0.037025
Hypothesis
L-max
%95 c.v.
r¼0 ro1 ro2 ro3
46.61** 13.87 9.68 4.45
31.50 25.50 19.00 12.30
L-trace 74.61** 28.00 14.13 4.45
%95 c.v. 63.00 42.40 25.30 12.30
Standardized eigenvectors ðb0 Þ log(FCD/ TDEP ) 1.000 4.407 0.744 630.750
(rd rf)
se
0.119 1.000 1.557 464.340
0.356 0.989 1.000 503.930
sr
0.006 0.033 0.096 1.000
cr
0.001 0.002 0.000 1.802
Standardized adjustment coefficients ðaÞ log(FCD/ TDEP) (rd rf) se sr
0.004138
0.004374
0.007076
0.000002
10.253000 10.262000 4.271200
0.423050 0.460210 0.469980
0.656990 0.700770 0.657620
0.000995 0.001228 0.000237
Weak exogeneity test statistics (Ho: weakly exogenous) (rd rf) LR test, w2 ð1Þ 30.030 [0.0000]**
se 28.007 [0.0000]**
sr 11.337 [0.0008]**
Joint weak exogeneity tests LR test, w2 ð3Þ 45.737 [0.0000]**
Trend
0.003 0.013 0.021 5.595
232
IRFAN CIVCIR
Table A.1.
(Continued )
Multivariate unit root tests (Ho: variable stationary) Log(FCD/TDEP) LR test, w2 ð5Þ 40.243 [0.0000]**
(rd rf) 40.682 [0.0000]**
se
sr
cr
19.235 [0.0017]**
57.365 [0.0000]**
53.003 [0.0000]**
Significance tests (t-statistics) (rd rf) Coefficient Standard error
0.1191 0.0657
se
sr
cr
Trend
0.3563 0.0670
0.0057 0.0026
0.0005 0.0002
0.0029 0.0014
Notes: (1) The estimation period is 1986:1 to 1999:12. VAR includes 10 lags on each variables, a constant term, centered seasonal monthly dummy variables, D8889 dummy, and D94 dummy variables. MDAY and trend variables are restricted to the cointegration space. (2) The L-max and L-trace are maximum eigenvalue and trace test statistics, adjusted for degrees of freedom. The critical values are taken from Osterwald-Lenum (1992). (3) The multivariate stationarity, weak exogeneity and significance tests statistics are evaluated by assuming a single cointegration vector. (4) The * and ** Indicate rejection of likelihood ratio tests at 5% and 1% significance levels, respectively.
FISCAL MONETARY AND EXCHANGE RATE POLICIES IN A SMALL OPEN ECONOMY Simon Neaime ABSTRACT After investigating debt sustainability in Lebanon, this chapter examines the conduct of monetary policy during the last decade. The empirical section looks at the long-run relationship between the nominal exchange rate and the inflation rate by employing Johansen co-integration technique. The resulting coefficient estimates suggest that while the Lebanese monetary authority has succeeded in containing inflationary pressures by adopting a monetary policy rule targeting the nominal exchange rate, it should have adopted a real exchange rate targeting policy to dampen the effects of its current policy on interest rates, public debt and budget deficits, and the growth in gross domestic product.
1. INTRODUCTION AND RELATED LITERATURE Inflation-prone economies, both in less-developed countries (LDC) and the European Monetary System (EMS), have often used a nominal exchange rate anchor as a means of achieving ‘‘inflation convergence’’. By pegging Money and Finance in the Middle East: Missed Opportunities or Future Prospects? Research in Middle East Economics, Volume 6, 233–249 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1094-5334/doi:10.1016/S1094-5334(05)06011-5
233
234
SIMON NEAIME
their currencies to a relatively low-inflation currency, such as the euro or the U.S. dollar (US$), and relying on high interest rate policies to defend their exchange rates, they have attempted to contain and roll back inflationary pressures. Even though this policy has helped them reduce inflation substantially relative to low-inflation countries, it has also generated persistent real exchange rate appreciations, losses in international competitiveness, large trade deficits, large budget deficits and the accumulation of sizeable debts, and in some instances serious currency devaluation crises. Some examples in the 1990s are Lebanon, Finland, Mexico, Britain, Italy, and Spain within the exchange rate mechanism (ERM) of the EMS.1 Lebanon has maintained its fixed nominal anchor at about Lebanese pound (LP) 1500 per US$ since the mid-1990s. A rigid nominal exchange rate regime combined with strict anti-inflationary policies has led to cumulative real exchange rate overvaluations, slow growth in gross domestic product (GDP) and an unsustainable fiscal policy. In addition, the collapse of the Lebanese currency has yet to materialize if some flexibility in the exchange rate is not allowed. It has frequently been suggested (Dornbusch, 1989) that countries with higher than average inflation rates could have escaped the problems posed by pegged nominal exchange rates, if they had pursued a fixed real exchange rate policy. Had these countries adjusted the nominal exchange rate according to the difference between foreign and domestic inflation so as to maintain a constant but competitive real exchange rate, they would have avoided real exchange rate overvaluations and high interest rates, and they would only have experienced moderate inflation.2 Thus, an alternative policy to pegging the nominal exchange rate is a real exchange rate peg. Several countries with high rates of inflation have historically adopted policies that peg the real exchange rate to avoid losses in competitiveness, including Argentina, Brazil, and Chile (Williamson, 1981). Also, in the literature, a number of authors have considered a similar exchange rate policy (Rodriguez, 1981; Dornbusch, 1982; Fuhrer & Moore, 1992; Edwards, 1996). In particular, the extended target zone proposal (see Williamson & Miller, 1987) envisages rules for the conduct of monetary policy to stabilize real exchange rates between the cooperating countries. According to this proposal, countries should determine a consistent set of real exchange rate targets, chosen to avoid prolonged real exchange rate misalignments and ensure medium- to longer-run current account equilibrium. However, previous studies have incorporated real exchange rate rules within a partial equilibrium framework. This is very limited because the authors are unable to capture the effects of the real exchange rate on all sectors of the economy.
Fiscal Monetary and Exchange Rate Policies in a Small Open Economy
235
This chapter examines empirically the success and implications of a monetary policy rule where the central authority chose to target the nominal exchange rate to contain inflationary pressures, in a small open economy, for the period 1993–1998. The announced objective of the policy was price stability. The Lebanese government set the reaction function for the monetary authority to devote all the attention to that specific objective, which was judged to be the most urgent for economic performance, after a period of extreme exchange rate and inflation volatility. Models that attempt to evaluate the success or failure of a given monetary policy differentiate between two types of policies: monetary policy rules that provide a regular supply of credit to an expanding economy and discretionary monetary policy rules that involve varying the money supply to counter shocks that may have caused the economy, to deviate from its intended objective. In general, when the monetary authority promotes price stability at the expense of the growth in real income it will react negatively to inflationary pressures. If the depletion of international reserves is a primary concern, the money supply is likely to be tightened. However, if the primary objective of the monetary authority is to finance government spending, money growth is expected to vary positively with price inflation and government spending. The rest of the chapter comprises three sections. Section 2 highlights the sustainability of fiscal policy by looking at debt and debt-related ratios. Section 3 evaluates the success in the conduct of monetary policy and outlines the method of investigation employed and summarizes the empirical results. Section 4 sets out the main conclusions and some policy implications.
2. THE SUSTAINABILITY OF FISCAL POLICY UNDER A PEGGED EXCHANGE RATE REGIME In its effort to restore confidence in the Lebanese economy and control the inflationary and exchange rate pressures of the early 1990s, the Lebanese monetary authority pursued a strict exchange rate regime pegging the LP to the US$. To attain this objective, interest rates in Lebanon were kept relatively high over the period under consideration. The stated objective for this high interest rate policy was twofold. On the one hand, the policy was intended to fight inflation, and at the same time prevent the private sector from substituting local currency for the US$. On the other hand, this policy was intended to attract foreign capital in the form of investments in Treasury Bills, the proceeds of which were used to finance the reconstruction of
236
SIMON NEAIME
the devastated infrastructure. The obvious outcome of this restrictive monetary policy was the accumulation of a sizeable public debt and recurrent budget deficits. We therefore turn next to an evaluation of the sustainability of fiscal policy in Lebanon using debt ratio analysis. Debt and debt-service indicators are widely used in the economic literature to assess an economy’s fiscal situation and the exposure to debt-related risks of liquidity and solvency. The standard indicators fall into two broad categories: flow and stock indicators. The flow indicators are based variables, such as GDP or exports. From an intertemporal perspective, these variables represent the resources that are available to meet debt obligations. Flow indicators may thus be useful in assessing solvency problems, since a solvency problem implies that an economy may never be able to service its debt out of its own resources. Stock indicators instead tend to reflect liquidity problems. There are three flow indicator ratios: debt/GDP, debt service/exports, and debt service/GDP. They relate debt and its servicing to resources that are available to meet these obligations, namely, exports and GDP. These indicators are useful for evaluating both solvency and liquidity risks, but have some limitations. The debt/GDP ratio indicates the amount of resources in a given economy that can be generated to repay the debt. A rising debt to GDP ratio signals that the rate of growth of debt exceeds the growth rate of the economy, and if this continues then the country will have difficulty in meeting its debt obligations in the future. Debt service measures3 include interest payments on all debt and amortization payments on long-term debt only. The assumption is that short-term debt is normally rolled over. A more comprehensive measure of debt service should include all amortizations. Fig. 1(a) shows that while in late 1970s, the debt to GDP ratio was 0.4, by the end of 2002 the ratio of Lebanese public debt had reached 1.8. This is a clear indication of a fast growing debt at a time when GDP growth was stagnating at around 1%. This led to Lebanon becoming one of the most indebted Mediterranean countries, as well as encompass to other highly indebted emerging countries, such as Argentina, Brazil, and Mexico. A huge debt of course entails a heavy debt service burden, and by the need of 2002, the debt service to GDP ratio stood at about 18% (see Fig. 1(b)). Even more alarming is the fact that in 2002, debt servicing had become about four times the size of Lebanese exports. In other words, every year the Lebanese economy had to generate hard currency of about US$3 billion to service its debt from sources other than the proceeds from exports, usually the main source of hard currency for an economy. This discrepancy put more strain
237
Fiscal Monetary and Exchange Rate Policies in a Small Open Economy Debt Service/GDP in (%)
Total Public Debt/GDP in (%) 2.0
24
1.6
20 16
1.2
12
0.8
8
0.4
4
0.0
(a)
60 65 70 75 80 85 90 95 100 Debt Service/Exports in (%)
4
0
(b)
60
65
70
75
80
85
90
95 100
Foreign Debt/Exports in (%) 20 16
3
12 2 8 1 0
4 60 65 70 75 80 85 90 95 100
(c) Foreign Debt/GDP in (%) 90 80 70 60 50 40 30 20 10 0
(e)
0
(d) 5
60
65
70
75
80
85
90
95
100
International Reserves/Short-Term Debt in (%)
4 3 2 1 0 60 65 70 75 80 85 90 95 100
(f )
60
65
70
75
80
85
90
95 100
Fig. 1. Public debt-related ratios (%): (a) Total public debt/GDP; (b) debt service/ GDP; (c) debt service/exports; (d) foreign debt/exports; (e) foreign debt/GDP; (f) international reserves/short-term debt. Source: Banque Du Liban (BDL), Lebanese Ministry of Finance and IMF.
on the balance of payments, the amount of foreign currency reserves available at the central bank, and subsequently on the local currency. The other broader flow indicators are the external debt/exports, and external debt/GDP ratios.4 These indicate the repayment capacity of a
238
SIMON NEAIME
country, and are used for evaluating solvency risk. A higher debt to exports ratio indicates a larger amount of resources needed to service obligations. This in turn implies increased balance of payments vulnerability and a larger repudiation risk. Foreign debt is 16 times the size of Lebanese exports. This is expected to put more pressure on the external balance (see Fig. 1(d)). After being below the 30% level in the 1980s and early 1990s, foreign debt has more than tripled since 1997 and stood at 80% of GDP by the end of 2002 (see Fig. 1(e)). The most useful stock indicator is the international reserves/short-term debt ratio. It relates the size of international reserves of the monetary authority to the amount of debt coming due within a year, and is an important indicator of liquidity risk. This indicator shows whether the economy has enough foreign exchange reserves to cover the amount of debt that is coming due in the short term.5 Fig. 1(f) indicates again that this ratio has been steadily declining since the late 1980s. This is mainly due to the fast growth in short-term debt, coupled with a steady decline in foreign currency reserves at the central bank. On average the central bank can at most cover 3–4% of the maturing short-term debt. This is signaling an increasing liquidity risk faced by the Lebanese government. The ratio analysis portrayed above presents a rather bleak and alarming picture regarding the Lebanese fiscal sustainability. While these indicators no doubt reflect current fiscal realities, the ratio analysis to fiscal sustainability used above and in the literature suffers from many shortcomings. The debt and debt-service ratios should be used in both static (point in time) and dynamic (intertemporal) context. Although these indicators can provide useful information about the ability to pay, no one indicator provides information on all the dimensions of a payment problem. Critical debt levels are likely to vary over time. So these ratios have to be accompanied by a more comprehensive economic evaluation. Therefore, it is important to turn to the next, more rigorous empirical analysis of monetary sustainability in Lebanon by studying the econometric time-series properties of the monetary variables of interest.
3. THE CONDUCT OF MONETARY POLICY: AN EMPIRICAL ESTIMATION The stated objective of the Lebanese monetary authority was twofold. While price stability was the main objective, targeting the nominal exchange rate was also being pursued. Whether this policy was successful in achieving its
Fiscal Monetary and Exchange Rate Policies in a Small Open Economy
239
stated objective is an empirical question, which we hope to answer by studying the time-series properties of the rate of inflation and the exchange rate. To test the static relationship between inflation and nominal exchange rate movements for the period 1993 1998, the following regression equation is used: Pt
Pt ¼ b0 þ b1 Det þ t
(1)
Pt
is the differential between the Lebanese (Pt) and the U.S.A. where Pt (Pt ) consumer price index (CPI) expressed in logarithm, and Det is the change in the logarithm of the exchange rate. b0 and b1 are constant parameters and t is a stationary stochastic process. Average monthly data used are for the period 1993 1998.6 Table 1 includes the results of the impact of the exchange rate on the inflation differential with no time lags. The estimation procedure uses the Newey–West estimation technique, and adjusts for heteroskedasticity and autocorrelation in the error term. The results indicate a strong positive correlation between the rate of change of the exchange rate and the change in the CPI differential for the period (1993 1998) a period of relative exchange rate stability. Specifically, a 1% appreciation in the pound (i.e. few LP needed to buy one US$), caused the Lebanese CPI to fall by 4.57%, assuming that the U.S. CPI remained unchanged. As the impact of exchange rate changes on inflation may also manifest itself with a time lag, lagged changes in the exchange rate were introduced as follows: Pt
Pt ¼ b0 þ b1 Det þ b2 Det
1
þ b3 Det
2
þ t
(2)
Eq. (2) tests whether the inflation differential responds to an appreciation of the LP with a lag. The results, summarized in Table 2, show that the apTable 1.
Regression of the Inflation Rate Differential on the Exchange Rate.
Dependant Variable LCPID Intercept Det R2 DW
Coefficient
t-statistics
1.4 11.9 0.26 0.3
110 3.5
Note: Newey–West HAC standard errors and covariance (lag truncation ¼ 3). Time period 1993–1998. Source: Author’s estimates.
240
SIMON NEAIME
Table 2.
Lagged Response of Inflation to Exchange Rates.
Dependant Variable LCPID
Coefficient
t-statistics
1.4 6.27 7.37 –0.49 0.32
66.9 2.75 3.11 –0.86
Intercept Det Det-1 Det-2 R2
Note: Newey–West HAC standard errors and covariance (lag truncation ¼ 3). Time period (1993–1998). Source: Author’s estimates.
preciation of the pound significantly impacted the inflation spread for up to 1 month. The effect seems to die out after 2 months. However, an R2 of about 30% suggests that other factors affected the rate of inflation as well. Liquidity injection via budgetary deficits was probably the most important factor. This deficit, measured as a percentage of GDP, declined only a little over the period as whole. 3.1. Unit Root Tests for Stationarity The existence of a long-run relationship between the exchange rate and inflation can also be tested. For this purpose the Johansen (1988) co-integration test was used, after establishing the non-stationarity of the series by applying the Augmented Dickey–Fuller (ADF) Unit Root Test.7 It is common for time-series data to demonstrate signs of non-stationarity; typically both the mean and variance of macro-economic variables trend upwards over time. In any case tests for non-stationarity are carried out as a preliminary step to explore the possibility of a significant long-run relationship between the variables concerned (i.e. co-integration tests).8 The following regressions were estimated: DPt ¼ b1 þ b2 t þ b3 Pt
1þ
2 X
di DPt
i
þ t
i¼1
Det ¼ a1 þ a2 t þ a3 et
1
þ
2 X
li Det
i
þ Zt
ð3Þ
i¼1
where D is the first-difference operator; bi ; di ; ai ; and li are constant parameters; t is a time trend; Zt and t are stationary stochastic processes.
Fiscal Monetary and Exchange Rate Policies in a Small Open Economy
241
To determine the order of integration of the two series model, Eq. (3) is modified to include second differences on lagged first and two lags of second differences. That is D2 Pt ¼ l1 þ l2 t þ l3 DPt
1
þ
2 X
mi D2 Pt
i
þ 1t
i
þ Z1t
i¼1
D2 Pt ¼ f1 þ f2 t þ f3 Det
1
þ
2 X
bi D2 et
(4)
i¼1
where D2 Pt ¼ DPt DPt-1 and D2 et ¼ Det Det-1 ; li ; mi ; fi ; bi are constant parameters; t is a time trend; Z1t and 1t are stationary stochastic processes. Two lagged-difference terms are included so that the error terms Zt and t in both the equations are serially independent. The null hypothesis is that di ¼ 0; that is, a unit root exists in Pt or et and the two series are non-stationary. To test for stationarity, the ADF test is applied to Eqs. (3) and (4), and the results are summarized in Table 3. Eq. (3) investigates the presence of Table 3.
ADF Unit Root Tests. Variable
t-statistic
ADF ADF (1) ADF (2) C Trend
et–1
Pt–1
1.75 0.45 0.62 –1.79 2.45
–2.92 1.61 –0.45 2.93 2.83 First Differences
t-statistic
ADF ADF (1) ADF (2) C Trend
Det–1
DPt–1
–4.85 –2.13 1.99 –4.1 2.99
–5.51 1.68 0.63 1.29 –0.34
Note: Sample period 1993–1998. ADF is the t-statistic for testing the null hypothesis of nonstationarity. The numbers in parenthesis are lag lengths. Source: Author’s estimates. McKinnon’s critical value at the 5% level of significance is –3.47. Indicates rejection of the null hypothesis at the 5% level of significance.
242
SIMON NEAIME
unit roots in both the Lebanese CPI and the LP/US$ exchange rate. The tstatistics on et–1 and Pt–1 are 1.75 and –2.92, respectively. The null hypothesis is di ¼ 0 (or unit root). Since both computed statistics are smaller in absolute terms than McKinnon’s 5% critical value (–3.47), the null hypothesis that di ¼ 0 cannot be rejected. Thus one can conclude that, the CPI and exchange rate series exhibit unit roots, which is another way of saying that the two series are non-stationary. Estimations of Eq. (4) indicate that the first-differenced series De and De are stationary. In absolute terms the ADF statistics of 4.85 and 5.51 exceed the critical value of 3.47, so that the hypothesis di ¼ 0 can now be rejected. That is, the first-differenced CPI and exchange rate data do not exhibit a unit root (i.e. the two series are stationary). The ADF tests indicate that each variable achieves stationarity only if converted to first differences. Thus, each variable is integrated of the first order Ið1Þ:9 Since Det and DPt are stationary, they are an I(0) stochastic processes, which means et and Pt series are I(1) time series; essentially they are random walks (i.e. non-stationary stochastic processes). 3.2. Co-integration Tests Testing for the existence of co-integration among economic variables has been widely used in the empirical literature. The existence would imply that two macro-economic series would never drift too far apart. The Johansen test is used to examine the existence of a long-term relationship between the Lebanese CPI and the exchange rate at the 5% and 1% levels of significance, respectively (Table 4). It is applied using alternative lag lengths in the vector autoregression (VAR). Table 4.
Johansen Co-integration Likelihood Ratio Test. t-statistics
Hypothesis Null
Alternative
r¼0 r p1
rX1 r¼2
40.055 2.44
Critical Values 5%
1%
15.41 3.76
20.04 6.65
Note: Sample period 1993 1998; maximum lag 2 years in VAR. Johansen Co-integration Likelihood Ratio Test is based on the trace of the stochastic matrix. The r represents the number of co-integrating vectors. Source: Author’s estimates. Indicates rejection of the null hypothesis of non-cointegration at the 1% level of significance.
Fiscal Monetary and Exchange Rate Policies in a Small Open Economy
243
The co-integration test is conducted for the period 1993 1998, and the existence of a long-term relationship between inflation and exchange rate movements could not be rejected (Table 4). Thus, there exists a long-run relationship between the two variables, such that a steady appreciation of the exchange rate has contributed to price stability.
3.3. Short-Term Dynamics of Inflation and the Exchange Rate While a significant long-term relationship between inflation and exchange rate movements is shown to exist, in the short-run this may or may not be the case. The two series are thus also examined to see how the exchange rate and inflation interact in the short-run. For this purpose, a vector errorcorrection model (VECM) is employed to tie the short-run behavior of inflation and exchange rates, respectively, to their long-run values. The VECM was first introduced by Sargan (1984) and later popularized by Engle and Granger (1987). Their two-step method is used to estimate the following equation: DPt ¼ a0 þ a1 EC t
1
þ a2 Det
1
þ a3 DPt
1
þ t
(5)
where D denotes the first differences, and t is the error term with the usual properties. Regression (5) relates the change in Pt to its first lag and that of et. In this equation Det-1 ; and DPt-1 ; capture the short-run disturbances in Pt whereas the error-correction term ECt–1 captures the adjustment toward the long-run equilibrium. If a1 is statistically significant, it tells us what proportion of the disequilibrium in Pt in one period is corrected in the next period. Eq. (5) assumes that movements in the exchange rate were determined exogenously. The estimates for this equation are presented in Eq. (6): DPt ¼ 0:001 0:49EC t ð0:38Þ
1
þ 1:1Det
ð 4:99Þ
ð1:91Þ
1
þ 0:28DPt
1
(6)
ð2:44Þ
R2 ¼ 0:27 The results suggest that short-run changes in the exchange rate of the LP have a significant impact on the rate of inflation, with a time lag of 1 month. This could be explained by the fact that imported goods are denominated in dollars. In periods of currency depreciation retailers change prices instantaneously to make windfall profits on already existing stocks, thereby shifting the additional price increases to consumers. Conversely, during periods of currency appreciation retailers adjust prices slowly and hence the effect on the price level is felt with a time lag and not with equal force, as in the
244
SIMON NEAIME
case of depreciation. Moreover, higher-inflation rates in a given period are likely to last well into the next period. Also, as indicated by the coefficient of the error-correction term, about 0.49 of the discrepancy between the actual and the long-run, or equilibrium, values of inflation is eliminated or corrected each month. To examine the possibility of a bi-directional relationship between the two variables, the error-correction model is re-estimated after switching the place of DPt and Det : Det ¼ f0 þ f1 EC t
1
þ f2 Det
1
þ f3 DPt
1
(7)
þ Zt
The results are reported in Eq. (8), and suggest that inflation has no statistically significant impact on the pound. That there is only a unidirectional relationship, Det ¼
0:001 þ 0:0007EC t
ð 5:36Þ
1
þ 0:46Det
ð0:10Þ
ð11:6Þ
1
0:013DPt
1
(8)
ð 1:63Þ
R2 ¼ 0:71 namely from the pound to inflation and not vice versa. As the Lebanese economy relies heavily on imports (i.e. 80% of goods consumed domestically are imported), it is not surprising that stabilizing the exchange rate contributes significantly to a lowering of the inflation rate. If exchange rate stability was maintained, this would be an important factor in helping maintain price stability. In both equations, the impact of the last period exchange rate on its current value was highly significant. This can be explained by the fact that during the period 1993 1998 there was a consistent gradual appreciation of the pound (2–3 pounds every month). Thus, other things being equal, a given appreciation in one period gave rise to expectations that this pattern would persist into the following period.
4. CONCLUSION AND POLICY IMPLICATIONS By pegging its local currency to the US$, and relying on a high interest rate policy to defend its exchange rates, the Lebanese central bank has been successful in containing and rolling back inflationary pressures. Even though, this policy has helped in reducing current inflation rates substantially relative to the early 1990s, it has also generated persistent real exchange appreciations, losses in international competitiveness, large trade
Fiscal Monetary and Exchange Rate Policies in a Small Open Economy
245
deficits, large budget deficits and the accumulation of sizeable debts. This policy has put the Lebanese economy on a non-sustainable fiscal path. Recent experience with fixed nominal exchange rates, both from the EMS and some LDCs, shows that fixed nominal exchange rates have generated substantial real exchange rate overvaluations, and thus have adversely affected the behavior of real interest rates, foreign debts, and GDP growth rates in high-inflation countries. This chapter has found that in the context of a small open economy, targeting the nominal exchange rate might be a successful way of containing inflationary pressure but also may lead to increases in the domestic real interest rate and public debt, and to lower levels of output and net exports. The Lebanese monetary authority has been successful in controlling the inflationary pressures of the early 1990s through a monetary policy rule of nominal exchange rate targeting. By targeting the nominal exchange rate the inflation rate has been below the 3% rate for the last 3 years. However, the costs of this policy are now felt at all levels of the economy. The real interest rate has remained high at around 12%, which has translated into a crowding out of the private sector’s investment, a huge trade account deficit and a very low rate of growth of GDP. Total foreign liabilities currently stand at US$35 billion, following a deficit of 56% in 2000. Debt servicing accounts for 43% of total expenditures and almost 88% of expected revenues. For a GDP of about US$17 billion the debt to GDP ratio is 200% which is the highest ratio in the Middle East region. An alternative policy for the Lebanese authority should have been perhaps to adjust the nominal exchange rate according to the difference between foreign and domestic inflation, so as to maintain a constant but competitive real exchange rate. This would have avoided real exchange rate overvaluations and the high interest rates used to maintain the currency peg, and the end result would have been to only have moderate inflation instead of near zero rates of inflation. However, one disadvantage of this type of real exchange rate targeting policy is that the monetary authorities do not know what the sustainable level of the real exchange rate is. Unless the authorities are able to choose the correct level at the time of the pegging, further adjustment will be necessary to avoid real exchange overvaluations. Second, the adoption of a policy that fixes the real exchange rate would lead to a ‘‘crawling peg’’ exchange rate regime and represent a fundamental change in the present European system of fixed but adjustable exchange rates. The solution to the above problem has been resolved through the adoption of the euro currency since 1999.
246
SIMON NEAIME
NOTES 1. See, for example, Dornbusch et al. (1995), Sze´kely (1996), Bruno (1995), and Edwards (1996), among others. Analysts agree that the appreciation of the real exchange rate and the current account were the main causes behind the EMS currency realignments and the devaluation of the markka and the peso. 2. This suggestion is directed more towards inflation-prone EMS countries (such as Italy, Greece, Portugal, and Spain) and also LDC countries with high-inflation rates and currency pegs. 3. Debt service is commonly computed on a cash basis instead of an accrual basis. If a country is in arrears on its debt payments, the debt service paid undercounts the true obligation. Thus, a better measure is debt service due instead of debt service paid. 4. Indicators using total debt have many limitations: The size of debt to GDP or exports is likely to be influenced by the stage of development of a country. Debtlevel-based indicators are expected to be higher in the early stages of development. Thus, using a debt-level-based indicator outside an inter-temporal or dynamic context can be misleading. Aggregate debt-level indicators do not provide any information on debt structure in terms of maturity, borrower (public or private) creditor, currency, or interest rate composition. All these aspects of debt structure have important implications in terms of determining vulnerability to external shocks. Thus, measures using total debt stock, instead of the characteristics of debt, ignore the fact that some of these debts are more vulnerable to an external shock than others. 5. This ratio also has limitation: (1) It may not be very useful for open economies that have a relatively large amount of short-term trade credits, because trade credits are less likely to be withdrawn during a crisis; and (2) it does not provide any information on the quality of international reserves. If international reserves are illiquid, then they cannot be used to meet immediate external obligations. 6. Data for 1990–1992 were not used because a structural change occurred in 1992 and high exchange rate of volatility during the same year. Regressions for the period 1991–1998 as a whole and for 1991–1992 did not show any significant statistical relationship between the exchange rate and inflation. 7. Co-integration tests involve two steps. First, each time series is examined to determine its order of integration. This involves Unit Root Tests based on the work of Fuller (1976), Dickey and Fuller (1979; 1981), and Phillips and Perron (1988). These tests involve the calculation of t-statistics for r ¼ 1 in ordinary least square (OLS) regressions of the form: X t ¼ rX t21 þ ut (1), where Xt is the variable of interest and ut is independent and identically distributed (iid), Nð0; s2 Þ: The t-statistic does not have the standard t-distribution; critical values under the null that r ¼ 1 are found in Fuller (Table 8.5.2). If ut violates the iid assumption, modifications to Eq. (1) must be implemented in conducting the test. The ADF (p) test supplements Eq. (1) with p-lagged changes in the dependent variable as additional regressors. Alternatively, the Phillips Perron test involves the estimation of Eq. (1), coupled with a non-parametric correction of the t-statistic for general forms of autocorrelation in the errors. Eq. (1) is often expressed in an alternative form as DX t ¼ ðr 1Þ X t 1 þ ut ¼ dX t 1 þ ut ; where d ¼ ðr 1Þ and D is the first-difference operator. This equation is equivalent to Eq. (1), however, now the null hypothesis is that d ¼ 0: This was the procedure followed in this chapter.
Fiscal Monetary and Exchange Rate Policies in a Small Open Economy
247
8. In regressions involving time-series data, the time variable t is often included as one of the regressors to avoid problems of spurious correlation. Data involving economic time series often tend to move in the same direction because of a trend that is common to all of them. To avoid such a spurious association, Pt and et are regressed on Pt–1, et–1 and a time trend t. For a more detailed discussion see Granger and Newbold (1974). 9. The level and first-differenced CPI, and exchange rate data plots show that the first-differenced series do not show any trend, while the series level do. These plots are available from author.
ACKNOWLEDGMENTS Financial support from the University Research Board of the American University of Beirut is greatly acknowledged. The author is grateful to two anonymous referees for valuable comments and suggestions on an earlier draft.
REFERENCES Bruno, M. (1995). Comment on Rudiger Dornbusch, Ilan Goldfajn, and Rodrigo O.Valde´s: Currency crises and collapses. Brookings Papers on Economic Activity, II, 278–285. Dickey, D., & Fuller, W. (1979). Distribution of the estimators for autoregressive time series with a unit root. Journal of the American Statistical Society, 74, 661–692. Dickey, D., & Fuller, W. (1981). Likelihood ratio statistics for autoregressive time series with a unit root. Econometrica, 49, 1057 1072. Dornbusch, R. (1982). PPP exchange-rate rules and macroeconomic stability. Journal of Political Economy, 90, 158–165. Dornbusch, R. (1989). Europe 1992: Macroeconomic implications. Brookings Papers on Economic Activity, II, 341–362. Edwards, S. (1996). Exchange-rate anchors, credibility, and interia: A tale of two crises, Chile and Mexico. American Economic Review Papers and Proceedings, 86, 176–180. Engle, R., & Granger, C. (1987). Co-integration and error correction: Representation, estimation and testing. Econometrica, 55, 251–276. Fuhrer, J., & Moore, G. (1992). Monetary policy rules and the indicator properties of asset prices. Journal of Monetary Economics, XXIX, 303–336. Fuller, W. (1976). Introduction to statistical time series. New York: John Wiley and Sons. Granger, C., & Newbold, P. (1974). Spurious regressions in econometrics. Journal of Econometrics, 2, 111–120. Johansen, D. (1988). Statistical analysis of co-integration vectors. Journal of Economic Dynamics and Control, 12, 231–254. Phillips, P., & Perron, P. (1988). Testing for a unit root in time series regression. Biometrika, 76, 335–346.
248
SIMON NEAIME
Rodriguez, C. (1981). Managed float: An evaluation of alternative rules in the presence of speculative capital flows. American Economic Review, 71, 256–260. Sargan, D. (1984). Wages and prices in the United Kingdom: A study in econometric methodology. In: D. F. Hendry, et al. (Eds), Quantitative economics and econometric analysis. Oxford: Basil Blackwell. Sze´kely, I. (1996). Discussion in Jeffrey Sachs, Aaron Tornell, and Andre´s Valasco: The collapse of the Mexican peso: What have we learned. Economic Policy, XXII, 58–60. Williamson, J. (1981). Exchange rate rules: The theory, performance and prospects of the crawling peg. New York: St Martin’s. Williamson, J., & Miller, M. (1987). Targets and indicators: A blueprint for the international coordination of economic policy. Washington: Institute for International Economics.
APPENDIX The dynamics of the Lebanese exchange rate, Lebanese monthly rates of inflation and U.S. monthly rates of inflation are shown in Figs. A1–A3.
3000
2500
2000
1500
1000
500 91
Fig. A1.
92
93
94
95
96
97
98
Dynamics of the Lebanese exchange rate (LL/US$): 1991–1998. Source: Banque Du Liban (BDL).
Fiscal Monetary and Exchange Rate Policies in a Small Open Economy
249
20
15
10
5
0
-5
-10 91
Fig. A2.
92
93
94
95
96
97
98
Lebanese monthly rates of inflation (%): 1991–1998. Source: Banque Du Liban (BDL), Lebanese Ministry of Finance.
15
10
5
0
-5
-10 91
Fig. A3.
92
93
94
95
96
97
98
U.S. monthly rates of inflation (%): 1991–1998 in. Source: IMF, International Financial Statistics.
This page intentionally left blank
250