:f ' ,: '~ ¾ *A" A 'U; i * t1 i Ii 0 iL~~~~~~~~~ i~~~S oil 4 1 d /J?S01X0J9Zew>01 @ f ~~~~~~~~~~~' gS ' '' si': F *- At''$t' 0 2 +i b @ o" 2 i i; Public Sector Deficits and Macroeconomic Performance Public Sector Def icits and Macroeconomic Performance Edited by William Easterly Carlos Alfredo Rodriguez Klaus Schmidt-Hebbel Published for the World Bank Oxford University Press Oxford University Press OXFORD NEW YORK TORONTO DELHI BOMBAY CALCUTTA MADRAS KARACHI KUALA LUMPUR SINGAPORE HONG KONG TOKYO NAIROBI DAR ES SALAAM CAPE TOWN MELBOURNE AUCKLAND and assodated companies in BERLIN IBADAN ©3 1994 The International Bank for Reconstruction and Development I THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. Published by Oxford University Press, Inc. 200 Madison Avenue, New York, N.Y. 10016 Oxford is a registered trademark of Oxford University Press. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior per- mission of Oxford University Press. Manufactured in the United States of America FLrst printing November 1994 The findings, interpretations, and condusions expressed in this study are entirely those of the authors and should not be attributed in any manner to the World Bank, tc its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. Library of Congress Cataloging-in-Publication Data Easterly, William Russell. Public sector deficits and macroeconomic performance I William Easterly, Carlos Alfredo Rodriguez, Klaus Schmidt-Hebbel. p. cm. "Published for the World Bank." Includes bibliographical references and index. iSBN 0-19-520988-5 1. Budget deficits-Developing countries. 2. Fiscal policy- Developing countries. I. Rodriguez, Carlos A. (Carlos Alfredo) II. Schmidt-Hebbel, Klaus. HI. International Bank for Reconstruction and Development. IV Title. HJ2216.E37 1994 339.5'23'091724-dc2O 94-11506 CIP Contents Foreword ix Acknowledgments xi Definitions and Data Notes xii Contributors xiii Overview William Easterly, Carls AIfredo Rodrguez, and Klaus Scizmidt-Hebbel Part I: Synthesis and Methodology 2 Part It: Country Case Studies 6 Conclusion 10 Note 11 PART 1. SYNTHESIS AND METHODOLOGY 13 1 Fiscal Adjustment and Macroeconomic Performance: A Synthesis 15 William Easterly and Klaus Schmidt-Hebbel Measurement of Public Deficits 18 Causes of and Remedies for Deficits 25 Deficits, Inflation, Real Interest Rates, and Financial Repression 38 Private Sector Response to Public Deficits 53 Public Deficits, Trade Deficits, and Real Exchange Rates 60 Condusions and Policy Implications 63 Appendix 1.1. Measurement of Deficits and Evaluation of Public Sector Solvency 67 Appendix 1.2. Sustainable Deficits 69 Appendix 1.3. Decomposition of Seigniorage 70 Notes 70 References 73 2 The External Effects of Public Sector Deficits 79 Carlos Aifredo Rodriguez Defict Financng and the Trade Balance 80 The Short-Run Process of Determining the Real Exchange Rate 84 Short- and Long-Run Interrelations between the Assets Markets, the Trade Surplus, and the Real Exchange Rate 87 v vi Contents Conclusions 95 Note 96 References 96 PART I1. COUNTRY CASE S-ruDIEs 99 3 Argentina: Fiscal Disequiibria Leading to Hyperinflation 101 Carlos Alfredo Rodriguez Recent Developments: 1-lyperinflation and Political Crisis 101 Background 102 Measurement of the Public Sector Deficit 107 Inflationary Financng of the Deficit 116 Ilflation, Revenue from Money Creation, and the Structure of Financial Markets 122 Demand for Money and the Limits on the Inflation Tax 126 The Process of Determining Public Debt and the Interest Rate 135 The External Effects of Public Sector Deficits 149 General Summary and Conclusions 257 Appendix 157 Notes 164 References 165 4 Chile: Fiscal Adjustment and Successful Performance 167 Jorge Marshall and Klaus Schmidt-Hebbel Fiscal Policy, Decomposition of Nonfinancial Deficits, Quasi-Fiscal Deficits, and Sustainability 174 Deficit Financing, Interest Rates, and Inflation 180 Crowding-Out and Crowding-In of Private Consumption and Private Investment by the Public Sector 191 Relative External Prices and the Trade Balance 202 Conclusions and Policy Implications 207 Appendix 4.1. Revenue and Expenditure Functions 211 Appendix 4.2. Derivation of the Asset Portfolio Model 213 Appendix 4.3. Derivation of the Trade Surplus Equation 225 Appendix 4.4. Data Sources, by Section 216 Notes 217 References 221 5 Colombia: Avoiding Crises trough Fiscal Policy 225 Willitam Easterly Historical Background 225 Fiscal Deficits, Real Interest Rates, and Inflation 237 The Real Exchange Rate and the Fiscal Deficit 254 Conclusion 262 Appendix 5.1. Balance Sheet Identities for Money and Credit Market Equilibria 263 Appendix 5:2. Definitions of Variables, Sources, and Data 263 Appendix 5.3. Data for Regressions for Consumption, Investment, and Money 265 Contents vii Appendix 5.4. Data for Regressions for the Real Exchange Rate and Resource Balance 266 Appendix 5.5. National Accounts, Colombia 1950 Prices 267 Notes 270 References 271 6 Cote d'Ivoire: Fiscal Policy with Fixed Nominal Exchange Rates 273 Christophe Chamley and Hafez Ghanem External Shocks 274 Public Revenues and Expenditures 278 Impacts of Fiscal Policy 288 Conclusion 302 Appendix 6.1. The Size and Burden of the Public Sector 304 Appendix 6.2. Public Investment 305 Appendix 6.3. Costs of and Retums to Education 306 Notes 307 References 308 7 Ghana: Adjustment, Reform, and Growth 310 Roumeen Islam and Deborah Wetzd Economic Policy and Fiscal Deficits since the 1960s 310 Fiscal Deficits and Financial Markets 319 Fiscal Deficits, Private Consumption, and Private Investment 329 Fiscal Deficits and the External Sector 335 Conclusions 353 Appendix 7.1. The Model 354 Appendix 7.2. Dynamics 355 Appendix 7.3. Time-Series Data for Principal Variables, Ghana, 1961-90 357 Appendix 7.4. Central Government Revenue Lost as a Result of Smuggling, Ghana, 1960/61 through 1981/82 366 Appendix 7.5. Decomposition of the Changes in the Central Government Deficit According to Changes in Economic Policy Variables, Ghana, 1972/73 through 1988 367 Notes 369 References 370 8 Morocco: Reconciling Stabilization and Growth 374 Riccardo Faini The Budget Deficit: Evolution and Financing 375 Asset Demand, Seigniorage, and the Inflation Tax 383 Investment and Saving Decisions 389 Modeling the Impact of Fiscal Policy 396 Conclusions 403 Appendix. The Model 404 Notes 408 References 411 viii Contents 9 Pakistan: Fiscal Sustainability and Macroeconomic Policy 413 Nadeem U Haque and Pefer!. Montiet Overview 414 Fiscal Policy in Pakistan 418 Deficits and Inflation 427 The Effects of Fiscal Policy on Economic Behavior 437 Policy Simulations 445 Conclusions 451 Notes 455 References 456 10 Zimbabwe: Fiscal Disequilibria and Low Growth 458 Felipe Morandd and Klaus Schmidt-Hebbel Determination and Sastainability of Public Sector Deficits 463 Deficit Financing and Financial Markets 474 Crowding-Out of Private Consumption and Private Investment 485 External Accounts, Real Exchange Rates, and the Public Deficit 492 Growth Prospects 497 Conclusions and Policy Implications 500 Notes 504 References 507 AFTERWORD 511 The Political Economy of Fiscal Deficit Reduction 513 Vito Tanzi Macroeconomic Imbalances and Government Response 513 Macroeconomic Policy, Credibility, and Biases in Official Pronouncemnents 515 The Control Issue 51B Conclusion 522 Notes 522 References 524 APPENDIX AND INDEXES 525 Statistical Appendix 527 William Easterly and Klaus Schmidt-Hebbel Author Index 551 Subject Index 553 Foreword In setting out the lessons of development experience, the authors of World Development Report 1991 surely did not hesitate before writing that a "prudent fiscal policy is the foundation of a stable macroecon- omy" (p. 110). The consequences of excessive fiscal deficits are well known: inflation, the crowding-out of private investment, debt prob- lems, and balance of payments problems. So why should the World Bank have spent resources on a research project examining the conse- quences of fiscal deficits? A sufficient reason is that adjectives such as "prudent" and "excessive" are inadequate guides to action. It is generally agreed that budget deficits which, since 1986, have exceeded 8 percent of gross national product (wNP) bear much of the blame for the current economic disaster in the former Soviet Union. But Pakistan's fiscal deficits have exceeded 8 percent of GNP each year since 1985. If a deficit of 8 percent a year is excessive, why did Pakistan experience steady growth along with single-digit inflation in the 1980s? Are the inevitable consequences merely being delayed? Or are some countries able to sustain large deficits for some time without ill consequences? If so, what determines the size and duration of sustainable deficits? These are among the questions studied in this volume. Another reason for this study is that the conventional wisdom should always be suspect. The results reported here do contain at least one major surprise: the link between inflation and budget def cits is very weak, at least for the countries examined in this study. The connection between budget deficits and inflation is evidently compli- cated: inflation of triple and more digits is always associated with large deficits, but large deficits do not necessarily lead to high inflation. This book shows World Bank research at its best. The country studies draw on detailed knowledge and data from a wide range of countries; they are well constructed, well informed, and valuable in their own right and are the main sources from which conclusions are drawn. The conclusions are presented concisely and convincingly by the editors in the introductory chapters. ix x The busy policymaker should certainly read the introductory chap- ters. Th"ie serious researcher will benefit from working at and reflect- ing on the country studies as well. Stanley Fischer Professor of Economics Massachusetts Institute of Technology luly 1994 Acknowledgments The research presented in this volume is the outcome of the World Bank research project on Macroeconomics of Public Sector Deficits. The World Bank Research Committee provided valuable comments at the initiation of the project. We thank the members of the project's advisory committee, Robert Armstrong, Edgardo Barandiarhn, Mohsin Khan, Danny Leipziger, Johannes Linn, and Costas Mi- chalopoulos, for their dedication and their contributions to the proj- ect's preparation and execution. We are also indebted to Mario Blejer, Vittorio Corbo, and Stanley Fischer for continuing comments and support. Maria Cristina Almero provided valuable research assis- tance. We are also indebted to Nancy Levine for her excellent editorial support. To all of them and to the participants in the World Bank Conference on the Macroeconomics of the Public Sector Deficit, held June 20-21, 1991, in Washington, D.C., go our thanks. xi Definitions and Data Notes Statistics are typically for calendar years (Argentina, Chile, Colombia, C6te d'lvoire, and Morocco) or for fiscal years, denoted by a slash (Pakistan and Zimbabwe). Data for Ghana are typicaUy based on fiscal years through 1982 and on calendar years thereafter. Variables are defined in each chapter. A dot over a variable repre- sents a time derivative (change per unit of time). A bar denotes an average level of a variable. An asterisk on a variable indicates that the variable is measured in foreign currency. A hat (^) denotes a propor- tional rate of change. Blanks in tables reporting regression results denote variables omitted from the regressions. A billion is a thousand million. xii Contributors Christophe Chanmley Professor, Department of Economics, Boston University William Easterly Principal economist, Macroeconomics and Growth Division, Policy Research Department, The World Bank Riccardo Faini Professor, Department of Economics, University of Brescia Hafez Ghanem Senior economist, World Development Report, The World Bank Nadeem Ul Haque Deputy Division Chief, Research Department, International Monetary Fund Roumeen Islam Economist, Middle East and North Africa Regional Office, Country Department i, The World Bank Jorge Marshall Vice President, Central Bank of Chile Peter J. Montiel Division Chief, Macroeconomics and Grwth Division, Policy Research Department, The World Bank Felipe Morand6 Director, ILADES(Georgetown Univer- sity Economics Program Carlos Alfredo Rodriguez Director, Centro de Estudios Macro- econ6micos de la Argentina Klaus Schhmidt-Hebbel Senior economist, Macaoeconomics and Growth Division, Policy Research Department, The World Bank Vito Tanzi Director, Fiscal Affairs Department, International Monetary Fund Deborah Wetzel Economist, Office of Development Eco- nomics and the Chief Economist, The World Bank xiii Overview William Easterly, Carlos Alfredo Rodrfguez, and Klaus Schmidt-Hebbel The following are the main findings and conclusions of this volume. 1. The correlation of fiscal deficits with any one indicator of macro- economic imbalance (such as inflation, real exchange rates, or market-determined real interest rates) is close to zero. This is explained by the great variety of ways in which governments finance their deficits. 2. Deficits, however, are unambiguously bad for growth. Reliance on taxation of financial assets to finance deficits-through surges of monetary financing or issuing of domestic debt at controlled nominal interest rates-is bad for private investment and for growth. Reduction of deficits through conventional tax increases is no more contractionary of short-run demand than taxing financial assets and is far preferable for long-nm growth. 3. Large fiscal deficits are largely explained by conscious policy choices and not by external shocks or by feedback from domestic economic conditions. 4. We reject the fiscal approach to inflation, whereby inflation is thought to be determined largely by the monetary financing requirements of the government. The long-run association between monetary financing and inflation is indisputable. How- ever, the amount of revenue from monetary financing is small, and some countries are even beyond the level of inflation at which inflation tax revenue is maximized. Thus it is doubtful that revenue motivations are behind inflation. Large temporary accelerations of monetary financing do not consistently result in higher inflation. 5. The conventional wisdom that public investment is good for private investment is contradicted by the evidence in half of the case studies, where public investment has a negative and statis- tically significant effect on private investment. The negative association in some cases is explained by the likelihood that public investment is replacing rather than complementing pri- vate investment. Concentration of public investment on infra- structure and on privatization of other state enterprises would 1 2 Overuiew ensure a complementary relationship between the public and private sectors. 6. Increasing public saving (reducing the deficit) is the only policy measure known to be effective in raising national saving. In only a few cases is there evidence that lower deficits (higher public saving) increase private consumption and thus lower private saving, and even in these cases the effect is far less than one for one. 7. Fiscal deficts lead to current account deficits and overvalued currencies. Conversely, fscal stabilization is a prerequisite for extemal adjustment and real depreciation of the currency. The decade of the 1980s was the decade of fiscal adjustment; so far, the 1990s seem to call for more of the same. The severe erosion of fiscal discpline in the 1970s and the first half of the 1980s has proved difficult to reverse. This widespread fiscal crisis coincided with the worst decade for growth and macroeconomic instability in the post- World War [I era. Attaining macroeconomic balance has become the foremost priority of industrial and developing economies alike; often, it is the measure of government success. Despite the consensus on the need to reduce public sector deficits, developing countries offer a bewildering range of experience with deficits. From country to country, deficits may lead to high and vari- able inflation, to debt crises, or to low inflation with crowding-out of investment and growth, while in some countries moderately high deficits seem not to generate macroeconomic imbalances at all. What explains these differing outcomes? This volume presents the findings of a World Bank research project that sought to answer this question through a careful examination of ten country case studies.' The case studies and the synthesis chapter use a common framework to assess the macroeconomic consequences of public deficits in a small but representative sample of developing countries. Part I: Synthesis and Methodology Chapter 1 synthesizes and extends the results of the case studies. It finds that sinple correlations between fiscal deficts and individual indicators of macroeconomic imbalance-inflation and interest rates- are weak to nonexistent. A simple explanation follows from the theo- retical framework of the study: since each of the many ways of financ- ing a high fiscal deficit carries its own macroeconomic outcome, the deficit will not be correlated with any one type of outcome. The chapter however, finds a strong negative correlation between deficits Overuiew 3 and growth. No matter how deficits are financed, the consequences are bad news for the long-run health of the economy. While most of this book treats fiscal deficits as a policy lever chosen by the government, the synthesis chapter examines the extent to which deficits are driven by external shocks (involving, for example, conmrodityr prices and foreign interest rates) or by domestic macro- economic variables (such as inflation, domestic interest rates, and real exchange rates). Only in Colombia do foreign shocks explain more than 50 percent of the (modest) variation in deficits. External shocks explain very little of the variation in countries with more dramatic changes in deficits, such as Chile, Ghana, and Zimbabwe. The Olivera-Tanzi effect, whereby inflation lowers tax revenue, shows up in only two cases, Ghana and Colombia. A real exchange rate depre- ciation was found to increase the deficit in Chile and Ghana, lower it in Mexico, and leave it roughly unchanged in Colombia, Thailand, and Zimbabwe. Exogenous policy changes account for most of the variation of deficits in significant episodes of fiscal improvement or deterioration in the case studies. Although inflation and deficits show no simple correlation, the synthesis confirms other studies that find a long-run association between morney creation (as a ratio to gross domestic product--GDP) and inflation. Figure 1.1, in chapter 1, shows that the relationship follows the typical "Laffer curve"; revenues from money creation first rise and then fall as the inflation rate rises. Revenue from money creation is often identified with the "inflation tax" on money hold- ings (which is slightly inaccurate because revenue from money cre- ation also includes the real growth in money demand). Maximum "revenue" is between 70 and 160 percent inflation. As the maximum ",revenue" is approached, the tradeoff between additional monetary financing and inflation becomes more unfavorable: an additional per- centage point of GDP of monetary financing induces only 5 additional percentage points of inflation in Thailand but 97 additional percent- age points of inflation in Argentina. The case studies lind much higher revenue-maximizing inflation rates, compared with our cross- sectional estimate, in the high-inflation countries and much lower rates in the low-inflation countries. The synthesis chapter argues that these findings result from the false assumption that money demand has a constant inflation elastidty. Once a variable inflation elasticty is taken into account, the true revenue-maximizing inflation rate is dose to the cross-sectional estimate of between 70 and 160 percent inflation. With all the sound and fury about money creation, it is easy to forget how small it is as a source of financing. In a sample of fifty-one countries over nearly twenty years, three-quarters of the annual observations of inflation tax revenue are less than 2 percent of cGDP. In 4 Overniew many countries excise tax revenues on a single product account for as much revenue as the inflation tax does. Considering the steep costs associated with inflation, it is hard to believe the fiscal theories of inflation that ascribe inflation to revenue motivations. The inflation tax tail is just too small to be wagging the inflation dog. The real significance of money creation is its potential for generat- ing large temporary revenues in times of crisis. Even in those cases, temporary bursts of inflation tax revenue did not always result in accelerations of inflation. The most consistent link between money creation and inflation is the long-run association described above. (Those countries resorting to large temporary use of inflation did have lower growth, however.) Huge one-time taxes on financal assets also lowered the potential for future revenue from money cre- ation. In the most spectacular example, the Ghana study found that an episode of demonetization of the largest-denomination note in 1982 (along with an earlier deposit expropriation) permanently low- ered revenue from money creation by 1 to 2 percentage points of GDP. High fiscal deficits are significantly associated with highly negative real interest rates. Fmancial repression was a popular means of "financing" fiscal deficits; in our study Ghana, Mexico, and Zin- babwe maintained nominal interest rate controls that resulted in neg- ative real rates. (Mexico libealized interest rates in 1988.) The control of interest rates was a costly way of mobilizng a small amount of 'revenue" (about 1 percent of GDP) through the implict tax on deposits. The ratio of private credit to cDp in financially repressed economies is ordy about a third that in unrepressed economies. The paucity of private credit in high-deficit, financially repressed economies and the increase in interest rates in high-deficit, unre- pressed economies both have the effect of driving out private invest- ment. The conventional wisdom that deficits crowd out investment is reaffirmed by these studies. A debt-financed defict is particularly damaging because rising public debt service crowds out even more investment over time. In Colombia, for example, a debt-financed expansion of 09 percentage point of GDP a year in the deficit over three years would lead private investment to fall 0.5 percentage point of GDP in the first year and 09 percentage point in the last. Deficits under financial repression had even stronger instantaneous impacts on private investment than those under decontrolled markets. The conventional wisdom fares less wel in the analysis of public investment. The studies of Chile, Colombia, Ghana, and Mexico found negative effects of public investment on private investment; the Argentina study found zero effect. These results are consistent with the theoretical ambiguity of the relationship between public and pri- Overiew 5 vate capital. If public capital substitutes for private capital (a public steel mill, for example), then public investment drives down the rate of return to private capital and lowers private investment. These results do not support the argument that fiscal austerity implemented through cuts in public investment has hurt private investment. Various theories have raised the hope that private saving would rise when public deficits rose, attenuating the potential crowding-out of private investment. The Ricardian equivalence theory holds that the private sector saves in anticipation of future tax obligations; an alternative theory is that fiscal deficits directly aowd out private consumption through widespread rationing of goods and controls on financial markets. Only three of the case studies-Argentina, Morocco, and Zimbabwe-showed evidence of increased private sav- ing in response to deficits, and the savings were much less than offsetting. The evidenice in these cases suggests that direct crowding- out of consumption caused increased saving. If increased private saving only slightly offsets increases in deficits, it follows that reducing deficits wfll be effective in raising national saving. Since other research finds little effect of policy variables (such as interest rates or tax incentives) on saving, reducing public expendi- ture is the only policy measure known to be effective in increasing national saving. Chapter 1 strongly confirms the prediction of the theoretical frame- work that fiscal deficits appreciate real exchange rates, as well as the corollary that real devaluation and fiscal adjustment go together. The synthesis also finds a close association between fiscal and exter- nal balances in the case studies (figure 1-2 in chapter 1). The remark- able robustness of these relationships across the case studies is the strongest evidence yet in the literature for the "fiscal approach to the balance of payments"-the idea that external deficits are primar- ily a result of fiscal deficts. The study also found important evi- dence for the effects of the composition of public spending on the real exchange rate, an effect often predicted and seldom documented. Chapter 2 sets out the framework for interpreting these results and for implementing similar approaches in other countries. The synthesis poirnts out the key role of economic growth in deter- mining the macroeconomic consequences of public deficits. In the 1980s fast-growing economies like India, Pakistan, and Thailand were able to get away with relatively high deficits because of the accom- panying rapid growth of domestic money demand and other means of financing. Countries that had slow or negative growth in the 1980s, such as Argentina, Brazil, and Mexico, found that an exceptionally large dose of fiscal adjustment was required to restore macro- economic stability. 6 Overview Part II: Country Case Studies The authors of the individual case study chapters have described the particular macroeconomic features of each country and the theoretical tools and econometric tests that were developed to deal with some of those special features. Foremost among the issues discussed are financial repression (Ghana and Zimbabwe), exchange rate controls and black markets for foreign exchange (Ghana), commodity booms and the accommodating fiscal reactions (Colombia, Cbte d'lvoire, and Morocco), financial deepening that allowed for monetary financng of deficits (Palistan), and the financing of central bank losses-the quasi-fiscal deficit (Argentina and Chile). There are large differences in the basic macroeconomic back- grounds of the countries descrbed in the case studies. On the one hand, Argentina had a long history of triple-digit annual inflation and finally experienced hyperinflation in 1989 and 1990. Chile, on the other hand, with a past history of high inflation, was able after the mid-1 973s Li i.plement serious structural reforms that removed high inflation from the macroeconomic picture Whereas in Argentina the inflation rate is the overwhelming indicator of macroeconomic imbal- ance, in Chile discussion centers around export growth, social expen- diture, and real interest rates. The other high-inflation country in the sample of case studies is Ghana. There, however, inflation never reached the triple-digit leveL Unlike Argentina, Ghana instituted a wide range of price and exchange rate controls and rationed credit and foreign exchange. During 1972-83 the government ran large fiscal deficits that had to be financed through monetary expansion because of the small size of the local financial market and the reduced availability of extemal credit. The controls on interest rates generated negative real rates, inducing significant disintermediation arnd further reduc-ing the base for the inflation tax. A black market for foreign exchange developed because of the negative real interest rates and foreign exchange controls. All these factors worsened the allocation of resources and reduced the chances for implementing structural adjustment In Argentina the very high inflation rates and the frequent reliance of authorities on large devaluations induced the public to shift toward holding foreign exchange (mainly U.S dollars). Because of the lack of a significant degree of indexation in the financal system, the natural inflation hedge was foreign exchange, creating the phenomenon known as "dolarization." Economies with indexed financial mar- kets, such as Brazil or Chile, have not experienced doa1arization because the public uses indexed deposits to hedge against inflation. To the extent that dollarization reduces the demand for high-powered money, it magnifies the inflationary impacts of a given degree of Overuiew 7 monetary deficit financing. The worst case is illustrated by Argen- tina's experience: dollarization became widespread, the real mone- tary base shrank to less than 2 percent of GDP, and, as the deficit continued, hypeinflation resulted. Paldstan is completely different from the high-inflation dollarized economies. Although the average deficit for the public sector was 7 percent of GDP for the period 1973-88, inflation has not exceeded 10 percent since 1976. The financing of deficits was evenly divided between the Central Bank, new issues of domestic debt, and external debt. Low inflation was the result of strong growth in money demand during the past two decades as new activities were incorporated into the monetary economy. (A virtually identical experience has been observed in other countries of the region, induding India.) In addi- tion to issuance of base money, the deficits were also financed by imposing on banks lquidity requirements that required the holding of government debt The remaining fiscal gap has been dosed by the steady flow of foreign aid. From the case study one might think that Pakistan is different because it was able to resort to monetary financing of deficts without inducing high inflation. However, the process of financal deepening cannot continue forever, and the inflationary consequences of defict financng wil eventualy be felt The high fiscal deficts may not have aeated inflation so far, but they did contribute to a weak external sector. Sustained current account deficits require the continuous inflow of new capital for financing in the context of an economy highly dependent on imports of intermediates. Zimbabwe represents a combination of the problems faced in Ghana and Pakistan. Although no extraordiary growth in money demand has been observed, the high fiscal deficits have been financed through new issues of domestic debt placed in the private sector. Effective foreign exchange rationing prevents Zimbabweans from converting their portfolios into foreign exchange and prevents imports from booming in response to domestic price controls and the high level of aggregate demand induced by the deficts. The result is a high level of private saving combined with compulsory publid debt placements at low, often negative, real interest rates. Ztbabwe is another country that has apparently reached the big spender's para- dise of being able to run fiscal deficits without inflation. The other face of the big spender's paradise is that Zimbabwe's domestic debt rose from 54 percent of cDP in 1980 to 86 percent in 1988, while external debt reached 42 percent of GDF in 1985. At those debt le'els, the interest on the debt becomes the most significant item in government spending and starts a dynamic process of its own: high interest requires the issuance of more debt to pay for it, which in turn produces a higher interest rate. Two alternatives are open for a country 8 Overiew with levels of indebtedness such as those of Zimbabwe in the mid-1970s. (a) melt the debt through devaluations, as Argentina often did, rsking hyperinflation, or (b) generate a level of fiscal deficit that will permit the interest on the debt to be genuinely serviced. According to the authors of chapter 10, levels close to the sustainable fiscal deflcit-but still inconsistent with recovery of investment and growth- were attained in Zimbabwe after the partial fiscal adjustment of 1987188. Commodity booms appear to help generate public expenditures that are not easily reduced when the boom is reversed. This has been the case in two of the countries studied: Colombia and Morocco. The rise in phosphate prices in the mid-1970s allowed Morocco to increase government spending. As prices feIl in the late 1970s, the govem- ment did not reduce spending but resorted to external borrowing. Because this foreign borrowing took place at a time of rising interest rates and deteriorating terms of trade, a foreign exchange crisis soon occurred, in 1983& At no point during this difficult period did authori- ties resort to monetary financing of the deficit; thus inflation never became an issue in Morocco. In Colombia fiscal deficits increased sharply after the end of the 1975-78 boom in coffee prices, during which the government had helped spend the extra available funds. The micrease in government spending had two effects: first, it created permanent spending needs that were not easily reversed when financing was no longer available; and, second, because foreign exchange was used to finance expendi- tures on domestic goods, the real exchange rate significantly appreci- ated, damaging the prospects of the nontraditional export sector. Co- lombia learned from the mismanagement of the 1975-78 coffee boom and did not repeat the mistake of increasing spending during the next boom, in 1986; instead, it properly sterilized the extra foreign exchange proceeds by issuing public debt. Colombia has had conservative and stable macroeconomic polices and has learned from the few errors it comrnitted in the past. Fiscal deficits have been very low by regional standards. (The highest national government defict since 1960 has been 5 percent of GDP, in 1982.) The result has been moderate inflation, high growth, and a satisfactorv external position. The Colombia case study precisely quantifies the theoretical Iela- tionship between fiscal deficits and inflation. The detafled analysis of money demand presented in the case study shows the limited possi- bilities for the use of the inflation tax. According to the calculations presented, at 22 percent annual inflation, the revenue from money creation is about 2 percent of GDP. Increasing revenue to 2.65 percent of GDP requires inflation to jump to 100 percent a year. It is clear that the extra 80 points in inflation do not justify an extra revenue collec- Overview 9 tion of barely more than half a point of GDP. Colombians must be aware of this fact because for the past fifteen years inflation has remained in the 15 to 25 percent range. As in Colombia, C6te d'Ivoire's macroeconomic shocks were mostly external in nature: the boom in cocoa and coffee prices in the znid-1970s allowed for an increase in government spending that was reversed only slowly after prices came back to normal levels. Because of the rules of the CFA zone, the country could not rely on domestic financing of the deficits in any significant amount and instead resorted to external debt. The extemal financing of deficits contrib- uted to the appreciation of the real exchange rate. The slow fiscal adjustment allowed extemal debt to grow from 30 percent of GDP in the early 1980s to 100 percent of GDP at the end of the decade. As debt grew and spending was not cut, the government experienced a pay- ments cisis that forced it to make drastic and disorganized spending cuts. We can speculate that as the fiscal adjustment progresses and foreign borrowing ends, the real exchange rate should rise above historical levels in order to generate the resource balance required to service the now-higher external debt. Argentina is unique among the case studies; the magnitude of its disequilibria makes it a textbook example of the effects of incorrect government policies. In Argentina monetary financing of deficits immediately creates inflation. Fiscal deficits increase domestic absorp- tion and induce trade deficits that also allow for a real appreciation in the exchange rate. Finally, as the deficits are financed with domestic debt, they induce real interest rates much higher than the productvity of capital and thus reduce the ability of the government to repay. Every year between 1960 and 1988, Argentina ran primary fiscal deficits. The monetary financing of those deficits has implied triple- digit inflation levels since 1975 (except in 1980 and 1987, when infla- tion was only around 80 percent, and 1989, when it reached 4,927 percent). External debt was also used to finance deficts but was cut as a source of financing after the regional debt crisis of 1982. Since then, external debt has grown because of unpaid miterest, reaching about $60 bfllion in 1990. Domestic debt has been issued in large quantities but has frequently been melted by the huge devaluations to which Argentines became accustomed in the 1970s. The result of the govern- ment's melting its own debt was that the market asked for higher real interest rates to continue holding the remaining lower real levels of debt. In consequence, after each devaluation meltdown, interest rates rose and the remaining debt continued rising at rates higher than before; in only a short time the debt reached still higher levels, and the government tried a new devaluation. The Argentine experience is in sharp contrast to that of Colombia, which kept inflation low by maintaining steady and sound fiscal poli- la Overview des and so grew at satisfactory rates. Argentina presented a picture of steady fiscal imbalance with very high inflation and no growth in income per capita during the twenty-year period 1969-89. Compari- son of the experiences of these two countries best exemplifies the high costs of fiscal imbalance and unpredictable intervention on the part of economic authorities. Chile is the only country studied that has successfully achieved significant structural adjustment during the past two decades. Tax reform, social security reform, and rationalization of the public sector were the main fiscal shocks after the mid-1970s. The economy was opened to international trade to a degree not yet experienced by any other Latin American country. After the 1983 balance of payments crisis, exports did diversify, and they grew more than imports. Chile was able to continue regularly serving its foreign debt and to save substantial amounts of reserves through the copper stabilization fund and thanks to an unusually high world price for copper during 1989-91. In recent years (1988-90) Chile's nonfinancial public sector has had surpluses with which it has canceled its outstanding debt with the central bank. From this perspective, Chile is completely different from the other countries studied, where the main problem was how to finance the deficits rather than how to invest the supluses. Chile's basic econonmc structure is no different than tat of neigh- boring countries. As in Colombia and Argentina, the Chilean data confirm all the relationships hypothesized at the beginning of the study: monetary financing creates inflation, debt financirng raises interest rates, and external financing induces real appreciation and trade deficits. The basic difference lies in the fact that Chile relied on effective adjustment rather than just devising strategies for financing the disequllibria. The reforms implemented before 1982 were an important stabilizing factor in the rapid recovery of the Chilean econ- omy during the 1980s. The three basic secrets of Chile's success wuith fiscal policy management were its stable policy environment, its con- sistent policies, and an increased emphasis on human capital and physical infrastructure in public sector investment. Conclusion Financing deficits has been shown to be a significant source of infla- tion, unsustainable indebtedness, and other types of macroeconomic instability. Price and interest rate controls and foreign exchange rationing often result from a futile attempt to hide the unavoidable costs of deficit spending; they eventually make both macroeconomic instability and growth performance worse. Low and stable fiscal defi- cits are a necessity for the favorable long-run prospects of a country, Overview 11 as well as for avoiding the short-ran macroeconomic ills of high infla- tion, high real interest rates, and real overvaluation of the currency. Note 1. The ten case studies were Argentina, Chile, Colombia, Cote d'Ivoire, Ghana, Mexico, Morocco, Pakistan, Thailand, and Zimbabwe. Because of space constraints, only eight countries are discussed in this volume. (Work- ing papers on Mexico and Thailand are available from the World Bank, Policy Research Department, Macroeconomics and Growth Division, Washington, D.C.) The Statistical Appendix presents complete data for all ten countries and also brings together hard-to-find information, collected in the course of the project, on consolidated nonfinancial public sector deficits, real interest rates, inflation rates, and inflation tax revenues in fifty-nine other countries. Part I Synthe sis and Methodology Fiscal Adjustment and Macroeconomic Performance: A Synthesis William Easterly and Klaus Schmidt-Hebbel Fiscal deficits were at the forefront of macroeconomic adjustment in the 1980s, in both developing and industrial countries. They were blamed in large part for the assortment of ills that beset developing countries during the decade: overindebtedness, leading to the debt crisis that began in 1982; high inflation; and poor investment and growth perfor- mance. In the 1990s fiscal deficits still occupy the center stage in the massive reform programs initiated in Eastern Europe and the former U.S.S.R. and by many developing countries on all continents. Many issues are raised by the successes and failures of fiscal adjust- ment. Not the least of these is how to define and measure fiscal adjustment. What are the most meaningful measures of public sector deficits? How should one assess fiscal stance, public sector solvency, and sustainability of deficits? While the analytical literature tends toward a definitional and methodological agreement on this issue (see Blejer and Cheasty 1991 for a comprehensive survey), empirical applications still differ widely. Once measurement issues are settled-and before analyzing the consequences of deficits-some frequently posed questions are: How important are the macroeconomic causes of fiscal deficits? What role do domestic and foreign shocks play in relation to changes in fiscal policy in the evolution of deficits? What are the most effective policy instruments for fiscal adjustment? Regarding the macroeconomic impact of deficits, a recurring ques- tion is whether larger public deficits are always associated with higher inflation. Sargent and Wallace's (1985) "monetarist arithme- tic" answered this question affirmnatively. But the relationship is blurred because governments finance deficits by borrowing, as well as by printing money. The relationship is further muddied by other factors, such as unstable money demand, inflationary exchange rate depreciations, widespread indexation practices, and sticky expecta- tions (see Kiguel and Liviatan 1988; Dombusch and Fischer 1991). 15 16 Fiscal Adjustment and Macroeconomic Perfonnance: A Synthesis Interest rates are another ambiguous factor. Do deficits raise domestic real interest rates when governments rely heavily on domestic debt financing, or is this relationship also blurred by such factors as financial repression (Easterly 1989; Giovannini and de Melo 1993) and the high degree of substitutability between public debt and other assets held by the private sector? Looking beyond domestic financial markets, a central issue of fiscal stabilization involves how private consumption and investment react to deficits. Will consumers reduce their spending when taxes are raised and increase it when taxes are lowered? Or wll they offset only changes in government consumption-without reacting to changes in government tax or debt financing-as argued by Barro (1974)? This issue is stfll not empirically settled for industrial countries (see Hay- ashi 1985; Bernheim 1987; Leiderman and Blejer 1988; and Seater 1993 for surveys of empirical studies on Barro's Ricardian equivalence proposition of one-to-one crowding-out of private consumption by public consumption). There is, however, growing evidence for devel- oping countries against the Ricardian hypothesis (Haque and Montiel 1989; Corbo and Schmiidt-Hebbel 1991). In regard to government investment, does a higher level of public capital spending crowd in or crowd out private investment? Theory predicts that this will depend on the degree of substitutability or complementarity of private and public capital (see Easterly, Rodriguez, and Schnidt-Hebbel 1989), and the lmited available evi- dence for developing countries confirms this ambiguity (see Blejer and Khan 1984; Khan and Reinhart 1990). Public deficts could also have indirect effects on private consump- tion and investment if real interest rates rise in response to higher domestic debt financing. Although theory predicts that real interest rates will h-ave an ambiguous effect on private consumption, private investment should decline unambiguously with higher interest rates. A growing body of evidence for developing countries supports the notion that private consumption is insensitive to real interest rates (Giovanniini 1983, 1985; Corbo and Schrnidt-Hebbel 1991; Schmidt- Hebbel, Webb, and Corsetti 1992). Surprisingly, many studies also show little response of private investment to interest rates in develop- ing countries (see the surveys by Rama 1993 and Serven and Sol- imano 1993). Finally, how do fiscal imbalances feed into external deficits? One should expect a strong link between fiscal and current account deficits in financially open economies when either consumers are not Ricar- dian or the national versus imported composition of public and pri- vate sector spending differs. The role that fiscal imbalances played in the overborrowing by developing countries that led to the 1982 debt crisis is widely recognized (see Dornbusch 1985; Sachs 1989). But William Easterly and Klaus Schmidt-Hebbel 17 more systematic evidence linking public deficits with external deficits and real exchange rate appreciations is still lacking. The underlying theoretical framework of this study is simple. The consequences of deficits depend on how they are financed. As a first approximation, each major type of financing, if used excessively, brings about a macroeconomic imbalance. Money creation to finance the deficit often leads to inflation. Domestic borrowing leads to a credit squeeze-through higher interest rates or, when interest rates are fixed, through credit allocation and ever more stringent financial repression-and the crowding-out of private investment and con- sumption. External borrowing leads to a current account deficit and real exchange rate appreciation and sometimes to a balance of pay- ments crisis (if foreign reserves are run down) or an external debt crisis (if debt is too high). In its analysis of the effects of deficits, the method applied here focuses first on the monetary and financial market implications of deficits. Next, the direct and indirect effects of public spending, taxa- tion, and deficits on private consumption and investment are addressed. Finaly, the impacts of public deficits on external disequil- ibria and the real exchange rate are identified. Each step in this study applies a common framework-for deficit measurement, sustainability, macroeconorric sensitivity, monetary and financial markets, private consumption and investment, the trade deficit, and the real exchange rate-to a set of ten case studies. To put the case study results into broader perspective, selected issues are addressed with the use of a fifty-nine-country sample that includes both developing and industrial countries. The fully spe- cified models based on behavioral relationships can be found in Easterly, Rodriguez, and Schmidt-Hebbel (1989); Marshall and Schmidt-Hebbel (1989); Fischer and Easterly (1990); and Rodriguez, chapter 2 in this volume. This chapter summarizes the empirical evidence-based on econometric estimations and policy simulations for each country-and derives the relevant policy implications.' Drawing on a representative set of case studies makes possible inferences on the unsettled issues mentioned above that are more reliable than those based on pooled cross-country studies or individ- ual case studies. The selection criteria for the ten cases-Argentina, Chile, Col- ombia, C8te d'lvoire, Ghana, Morocco, Mexico, Pakistan, Thailand and Zimbabwe-stressed both the diversity of fiscal and macro- economic regimes and experiences and the sample's ability to repre- sent the developing world at large. As wil be dlear frcom the discus- sion below, the ten countries indLude countries that underwent fiscal adjustment and those that did not; high- and low-deficit countries; large and small economies; low- and high-inflation countries; coun- 18 FiscalAdjustment and Macroeonomic Performance: A Synthesis tries with and without developed financial markets; and countries with and without access to foreign financing. This chapter begins with a discussion of alternative measures of fiscal deficits, deficit sustainability, and the interaction between defi- cits and the macroeconomy. The second section examines the causal- ity from the macroeconomy to the deficits; the contribution of exter- nal and domestic shocks in comparison with that of shifts in fiscal policy in changing the deficit; and the most effective policy, instru- ments for fiscal adjustment. The three subsequent sections address the macroeconomic conse- quences of deficits. First the focus is on the relationship between the domestic financing of deficts, inflation, and real interest rates; then the relationship of deficits to private consumption and investnent is analyzcd; and finally the spillover effects of deficits on external imbal- ances are examined. The chapter closes with the main conclusions and policy implications of the analysis. Measurement of Public Deficits How the public deficit is measured has an important bearing on an accurate analysis of the maaoeconomic implications of deficits.2 Appendix 1.1 discusses briefly two dimensions of deficit measure- ment: public sector composition and economic relevance. To ilustrate how misguided a too narrow measure of public sector defict can be, box 1.1 compares consolidated nonfinancial public sector (cNn's) and quasi-fiscal deficits in Argentina and Chile. Box 1.2 Ilustrates the differences between alternative measures of Morocco's central gov- errunment deficit. With regard to deficit measurement, the approach taken by this chapter and by the research project at large is to combine the widest possible sector coverage (subject to data availability) and the most meaningful definitional choice, driven by the issues addressedL3 Sustainable Public Dficits Sustainable public deficit measures are denved by looking at the below-the-line financing constraints of the deficits, following the accounting approach to public sector solvency (see appendix 1.1). Our purpose is to compare actual deficits and sustainable deficit levels. Estimrates of sustainable prmary deficits are derived by hold- ing constant the ratios of public liabilities to output for feasible values for the macroeconomic variables that determine market demands for public liabilities, as discussed in appendix 1.2. Figure 1.1 comoares sustainable and actual primary surplus levels for six relevant fiscal experiences during the 1980s.4 Sustainable pri- Wiliamn Easterly and Klaus Schmidt-Hebbd 19 Box 1.1. Quasi-Fiscal Deficits in Argentin and Chile Quasi-fiscal deficits of the central bank amounted to a cumulative 55 percent of cDP during 1982-5 in Argentina and to a cumuative 41 percent of cap during the same period in Chile. In Argentina quasi-fiscal deficits were roughly as large as conventional deficits during the period (box figure 1.1); the sum of both exceeded, on average. 25 percent of GDP per year! In Chile quasi-fiscal deficits exceeded, on average, 10 percent of cDP per year, more than double the conventional deficits. Both cases illustrate how misleading nonfinancial public sector deficits can be. For emple, whfle conventional deficits were falling during 1984 in Argentina, the fiscal stance of the overall public sector, induding the central bank, deteriorated greatly. In Chile con- ventional deficits underestimate both the 1981-85 fiscal crsis and the subse- quent fiscal adjustment. Box figure 1.1. Conventional and Quasi-Fiscal Deficts, Argentina and Chile, 1979-89 Argentina Chile Percent Percent 25 - - - -- - -- 25 - _ _ _ _ _ 20 * - - - 20 15 ---- 15 ___ 10 -- -- ----- 10A 0 - - --------------- 5 -- -5 19791980 1985 1989 19791980 1985 1989 - Conventional publc sector deficit -Quasi defidt mnary surplus levels diverge widely, not only because of the different levels of public liability stocks and macroeconomic variables in each country but also because the calculations were made for different public sector coverages. The levels range from 1.4 percent of aDP for the total (financial and nonfinancial) public sector in Chile to -2.8 20 Fiscal Adjustment and Macroecanomic Perfomnmce: A Synthesis Box 1.2. Alternative Deficit Measures for Morocco The decline in Morocco's cash-basis deficit gives only a partial picture of the significant fiscal improvement achieved during 1983-88 (see box figure 12). The country was able to lower its accruals-basis deficit at an even more rapid pace by reducing accumulation of arrears and starling to repay them in 1986. At the same time, the nominal deficit fell faster than the economically more meaningful operational deficit as a consequence of the dedine of the inflation component of domestic interest payments resulting from lower inflation. Box figure 12. Deficit Measures, Morocco, L983-88 Percentage of GDr- 14 10 - 6 4 1983 1984 1985 1986 1987 1988 - Nominal deficit (accuals basis) Operational deficit (acmrals basis) +- Nominal deficit (cash basis) - Arrears percent of GDP (a sustainable primary deficit) for the central govern- ment in Ghana. How do the calculated sustainable deficts compare with actual levels during the 1980s? In Chile the massive public sector adjustment during the 1980s (comprising both the nonfinancial deficit and the central bank's quasi-fiscal losses) pushed primary surpluses in 1988 and 1989 well beyond the upper bound of sustainable levels. Col- ombia reached sustainable primary surplus levels in 1987-89 after significantly strengthening its fiscal stance. Morocco also pursued strong fiscal adjustment polices, achieving a primary surplus level in 1988 even higher than that required to reduce public debt as a share of Figure 1.L Actual and Sustainable Public Sector Primary Surplus in Six Countries, 1980s Chile (consolidated total public sector primary surplus) Percentage of cGP 10.0 5.0 - -5.0 -10.0- -15.0 - -20.0-- tI - 1982 1983 1984 1985 1986 1987 1988 1989 Colombia (consolidated nonfinancal public sector primary surplus) 1.3 0 -40 1983 1984 1985 1986 1987 1988 1989 Ghana (central govenment primary surplus) 0 - -2.8 -4.0 -8.0- 1984 1985 1986 1987 1988 1989 1990 (Figureconinues on theUouwingpage.) 21 Figure 1.1 (continued) Morocco (central government primary surplus) Percentage of cop 4.0 - 0 -22 -4.0 -8.0~ 1983 1984 1985 1986 1987 1988 1989 Palistan (general govermnent primary surplus) 2-0 - 0 I -1.7 _ -4.0 -8.0 1983/84 1984/85 1985/86 1986/87 1987/88 Zimbabwe (general govermment primary surplus) 6.0 - _ __ 0 -1.7 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ -4.0 -8.0 -10.0- 1983/84 1984/85 1985/86 1986/87 1987188 1987/88 Actual prinmary surplus Sustainable primary surplus (midpoint) ---- --- Upper and lower bounds of primary surplus Source: Chapters 4,5,7,8,9. and 10 inthisvolume 22 William Easterly and Klaus Schmidt-Hebbel 23 output, although subsequently its fiscal stance deteriorated some- what- Pakistan's fiscal deterioration raised its primary deficits beyond the 1.7 percent of CDP level consistent with stable debt-to-output ratios. Fialy, Zimbabwe's modest fiscal adjustment in 1987-89 reduced its primary deficit to within the broad range of values con- sistent with sustainable levels, but stfll distant from an upper-bound level of sustainable prmary surplus consistent with an adverse mac- roeconomnic scenario. Although these calculations are based on simple assumptions, they provide useful benchmarks for evaluating fiscal stance from a longer- run perspective. The next section generalzes these results by focus- ing on the fiscal performance of the ten country cases and distin- guishing between the consolidated nonfinancial and quasi-fiscal deficits. Correlations of Deficits with Other Macroeconomic Variables To obtain an overview of the relationship between fiscal deficits and relevant macroeconomic variables, we collected data on a large sam- ple of countries, induding members of the Organization for Economic Cooperation and Development (oEcD). (For data and sources for the sample, see the Statistical Appendix in this volume.) The usefulness of fiscal deficts as an indicator of overall economic performance is reflected by our calculation of simple correlations between public sec- tor balances (as a percentage of GDP) and other major macroeonornic variables, as shown in table 1.1. There are good reasons not to expect very strong correlations. Fis- cal deficits are measured in different ways across countries, introduc- ing some measurement error into the sample. In addition, the theo- retical relationship between deficits and other macroeconomic variables depends crucially on the means of financing them. (More generaly, simple correlations may fail to be significant because of the omission of other variables.) Despite these caveats, we find a sigrdfi- cant statistical relationship between the public sector balance and many, although not all, macroeconomic peformance variables. Per capita growth is significantly and positively related to fiscal sur- pluses.5 There is also an interesting negative and significant correla- tion between per capita growth and the variance of fiscal balances. Low and stable fiscal deficts are associated with high growth. FLscal balances are positively related to real interest rates, contray to the usual prediction that deficits lead to high interest rates and surpluses to low rates. Since there are a large number of negative real interest rates in the sample, this finding is probably explained by an association between financial repression and fiscal deficits. Fiscal bal- ances are negatively related to money creation (seigniorage), confirm- 24 Fisca Adjustment and Macroeconomic Performance: A Synthesis Table 1.1. Cross-Section Correlations of Consolidated Public Sector Balance as a Percentage of GDP with Other Variables Variable Corrdution t-satistic GDP growth 0.19 1.46 GoP per capita growth 0.37 3.02** Real interest rate 0.31 2.34* Money creation (percentage of GDP) -0.33 -2.40** Inflation -0.16 -1.21 Investment (percentage of CDP) 0.24 1.70* Total consumption (percentage of Gm) -0.48 -3.97** Private consumption (percentage of GDP) -0.38 -3.00** Real exchange rate -0.I5 -1.10 Current account (percentage of GDP) 0.54 4.76** Black-market premium -0.35 -2-65** Variance of deficits with per capita CGP growth -0.36 -288** Significant at 5 percent level (one-tailed). * Signfficant at 1 percent level (one-tailed). Note: Public surplus is positive; deficit is negative. PubLic balances of OECD countries are eneral govemment. Sample size varies between fifty and fifty-nine. Period of averages is longest period for which data are available for each pair of concepts for eadh country. For the real exchange rate, an appeciation is an increase. Source: OECD, OECD Economic Outlook; World Bank data. For real exchange rate, Dollar 1990, uing purchasing power parity (Prr) comparisons from Summers and Heston 1988. ing that countries which run high deficits do so in part through greter reliance on seigniorage. (But de Haan and Zelhorst 1990 find that the correlation holds only for high-inflation countries.) However, inflation rates show little correlation with fiscal balances, perhaps reflecting in part the nonlinear relationship between money creation and inflation discussed later in this chapter. Fiscal balances are positively, although weakly, related to total gross domestic investment, offering at least superficial support to the notion that deficits crowd out investment. The fiscal balance is nega- tively related to both total and private consumption. This finding is superfically consistent with the hypotheses that taxes crowd out pri- vate consumption and that public and private consumption are com- plementary. It is inconsistent with the Ricardian notion that govern- ment spending lowers consumption. Fiscal balances are correlated with external current account bal- ances across countries. This lends support to the "fiscal approach to the balance of payments," which says that fiscal imbalances are the main sources of extemal imbalances.6 A suggestive association is found between fiscal balances and black-market premiums, indicat- ing that countries with high deficits are more likely to control the foreign exchange market tightly and that deficits drive up the pre- mium created by such controls. Real exchange rates, however, show William Eisterly and Klaus Schmidt-Hebbel 25 little association with fiscal deficits. This may reflect the extent to which trade intervention differs across countries. There are two messages to be carried away from this set of gener- ally strong (but occasionally weak) associations between fiscal bal- ances and macroeconomic performance. The first is that despite prob- lems of comparability across countries, the fiscal balance is a useful indicator of macroeconomic health. The seconid is that in order to trace the relationships between deficits and specific macroeconomic variables, such as inflation and real interest rates, deeper analysis is needed, with carcful attention to the bidirectional causality among the variables and the underlyingbehaviaral relationships. The follow- ing sections address these issues. Causes of and Remedies for Defidts Liis section focuses on the determinants of deficits and the compo- nents of successful programs for fiscal stabilization. We start by assessing the contnrbution of foreign and domestic macroeconomic shocks in relation to that of shifts in fiscal policy in the evolution of public deficits. We ten identify the main policies used in successful adiustment efforts. Senitivity of D seficit to Foegn Variables Foreign shocks are a source of fiscal instability in many developing countries. Fluctuating export prices and foreign interest rates mean that commodity exporters and highly indebted countries face an inherent instability which often hinders fiscal adjustment efforts.7 Changes in export prices affect the public sector directly, via the profits of the exporting state-owned company or marketing board, or indirectly, through taxes on profits or on exports. (See G0b and associates 1988 for a detailed study of the macroeconomic and fiscal consequences of commodity price booms in oil-exporting countries.) The quantitative impact of the export pnce shock on government accounts depends on the tax and property structure, the amount exported, and the magnitude of the price shock. In countries that face high export price volatility and that export through a large state- owned enterprise (as in Chile and Mexico) or a marketing board (as in Cote d'Jvoire and Ghana), fiscal accounts are sensitive to terms of trade shocks. If the foreign trade structure is diversified, the private sector is the main exporter, and export taxes are low or absent (as in Pakistan and Zimbabwe), public sector accounts do not suffer signifi- cantly from export price volatility. Import prices affect pubLic expendi- ture in some countries. In Morocco, for instance, the decline in im- ported food prices was the main cause of the substantial decline in 26 Fiscal Adjus:ment and Macroeconomic Performance: A Synthesis subsidies to the private sector in the mid-1980s. Changes in foreign interest rates affect highly indebted countries with a high share of variable-interest debt, such as Argentina and Colombia. In addition to measuring the impact of foreign shocks on public accounts, it is illuminating to assess the contribution of shocks to overall public sector deficits. Both dimensions are presented for six countries in table 1.2. The first column shows the average absolute change in public deficits as a result of different foreign shocks over the relevant sample periods. For instance, foreign shocks contrib- uted, on average, to a variation in the public sector deficit amounting to 2.3 percent of ODP in Chile and 0.3 percent in Zimbabwe. Chile and Thailand are highly sensitive to changes in export prices or terms of trade, while in Colombia and Morocco the average contibution of export price shocks to deficits is only about 1 percent of cDP. In Zimbabwe the influence of terms of trade shocks on government revenue is negligible. In comparison with terms of trade shocks, fluc- tuations in interest rates have much lower effects on public deficts; they contribute, at most, OA percent of CDP to the variation of public sector deficts in our sample. The average relative contrnbution of foreign shocks in the second column measures the degree of correlation between foreignrshock- induced deficits and the overall public deficit. In Chile, Colombia, and Thailand adverse foreign shocks increase deficits, with shares varying between 12 and 50 percent of the total fluctuation of deficits. In Ghana the tiny foreign interest shocks are uncorrelated with defi- cits. In Morocco and Zimbabwe, however, foreign shocks have the opposite sign from the changes in overall deficits, indicating that domestic macroeconomic shocks and fiscal policy changes more than compensate for the influence of adverse foreign shocks. 'Even moderate shocks could explain much variation in deficits. In Colombia, for example, moderate shocks have a huge influence-as much as 50 percent-on the variabflity of deficits. Because Colombia did not require such substantial fiscal adjustment during the relevant sample period (1989), foreign shocks had a more significant role in the evolution of its deficit. In Chile, which experienced the greatest foreign shocks, the relative contnbution of these shocks to deficit variability has been a low 12 percent. During 1973-88 Chile embarked on massive fiscal adjustment programs that overshadowed the influ- ence of foreign shocks. Optimal responses to shocks depend on whether the shocks are temporary or permanent. Purely transitory shocks should be (dis)saved and hence reflected by public deficits, whereas permanent shocks should induce corresponding changes in expenditure or reve- nue without affecting deficits. In the case of public sectors that own large commodity-exporting comparnies (as ir. Chile, Mexico, and William Easterly and Klaus Sclamidt-Hebbel 27 Table 1.2. Contribution of Foreign Shocks to Public Deficits Average relative contribution of Averageabsolute foreign Shocks to variation of public variation of deficts attributable publicdeficits to foign shtocks (peruentage of inrieRion Country and shock (percentage of GDP) of deficits) Chile, 197348 Foreign shocks 2.3 12 Changes in copper pnce 2.7 15 Changes in foreign interest rates 0.4 -3 Colombia, 1984-89 Foreign shocks 1.0 50 Changes in coffee fund 1.2 59 Changes in surplus of ofl company 0.9 -9 Ghana, 2972(73 to 2988 Changes in foreign interest rates 0.1 0 Monxro. 1971-88 Changes in contributions of phosphate company 0;8 -17 Thailand, 1970-88 Changes in terms of trade 2.2 41 Zimbabwe, 1980181 to 19888!9 Changes in foreign interest rates 0.3 -3 Note: The first column computes the annual average absolute variation of the deficit caused by the corresponding changes in foreign variables. (The exception is Chile; figures are based on period averages for 1973-75, 1975-81, 1981-86, and 1986-8a.) If more than one foreign variable is considered, the sum of the average absolute varia- dons for the individual variables differs from the average absolute varation of the cornbined shocks because of the opposite signs of individual variations. The second column reflects the average relative contnbution of foreign shocks to the variation cf pubLic deficits, defined as: dv,(sign dd Il Z ldd i-D 11 i-t where di is the change in the deficit in period i, dvi is the change in the deficit caused by variable v in period i, t is the initial period, and n + 1 is the total number of periods. Source: Authors' clulaltions and information from country case studies Listed in the references to this chapter, Morocco) or tat collect large revenues from private exporters (Cote d'Ivoire and Ghana), price-stabilization or revenue-stabilization funds, such as those implemented in Chile and Venezuela, or hedg- ing through risk-sharing contracts, are efficient mechanisms for iso- lating the budget from temporary export price shocks. 28 Fiscal Adjustment and Macroeconomic Perfonnance: A Synthesis Sensitivity of Deficits to Domestic Macroeconomic Variables A second group of variables that affect deficits and are also outside the direct control of fiscal policymakers consists of domestic macro- economic variables. In the following discussion, we concentrate on four variables that often have strong effects on public budgets: infla- tion, the real interest rate, the real exchange rate, and output. INFLATION. Inflation affects budget deficits through various chan- nels.8 Anticipated inflation raises nominal interest payments to domestic debt holders. Inflation also affects the primary deficit (the Olivera-Tanzi effect).9 Collection lags for taxes that are not fully indexed (for example, nominally fixed excise taxes) lead to dedining real revenue when inflation increases. Inflation also tends to lead to public demoralization and hence to lower tax compliance. If, how- ever, income brackets are nonindexed, higher inflation leads to bracket creep and hence to higher direct taxation. Real publc current expenditure declines with inflation when public wages or transfers are -tot indexed. Whereas in many countries the net effect of inflation is to increase primary deficits, the budget structure could conceivably reverse this effect. Table 1.3 summarizes the effects of inflation on public deficits in seven countries and identifies the channels through which they oper- ate. Results from estimated tax revenue functions allow us to classify countries according to the net influence of inflation on tax revenue. Inflation lowers aggregate tax revenue in Colombia and both direct and indirect taxes in Ghana. The only positive effect of inflation on taxes is found for direct tax revenue in Zimbabwe, where nonindexa- tion of income brackets leads to bracket creep. Short collection lags, indexation of tax revenue, and indexation of income brackets could be behind the nonsignificant effects of inflation on tax revenue in Chile (direct and indirect taxes), Morocco (total taxes), Pakistan (direct, indirect, and trade taxes), and Zimbabwe (indirect and total taxes). Table 1.3 provides some partial evidence concerning the effects of inflation on expenditure categories. Transfers to the private sector in Chile decline with inflation, presumably as a result of incomplete indexation. No evidence of a significant effect of inflation on aggre- gate public expenditure could be found in Morocco. In most countries the net influence of inflation is to raise nominal public sector defidts, as a result of the dominating effect of rising prices on interest payments and tax revenues. A good example is Thailand where, according to econometric results, an increase in inflation of 10 percentage points raises the CNH'S deficit by 0.9 per- centage points of GDP."( Williatm Easterly and Klaus Schmnidt-Hebbel 29 Table 1.3. Effect of Inflation on Public Deficits Negative 2ero Positifve Effect on tari revenue Colombia: total taxes Chile: direct taxes, Zimbabwe: direct taxes (1972-87) indirect taxes (1970171 to 1988/89) Ghana: direct taxes, (1973-89) indirect taxes Morocco: total taxes (1970/71 to 1988) Pakistan: direct taxes, indirect taxes, trade taxes (1972/73 to 1987188) Zimbabwe: indirect taxes, total taxes (1970/71 to 1988189) Effect on public expenditure Chile: transfers Morocco: public (1973-89) expenditure Effect on the public deficit Thailand (1971-88) Note: The effect of inflation on deficits via nominal interest payments on the debt is excluded as a separate channel of transmission. a. Because of nonindexation of income brackets. Sounre: Country case studies listed in the references to this chapter. REAL iiETREsr RArE. Real interest payments (and hence both the nominal and the operational deficit) obviously increase one-to-one with the real interest rate. Inflation shocks that are unexpected (or, even if expected, are not reflected in higher nominal interest rates because of controls on interest) reduce real interest rates and hence the operational deficit. For instance, in Ghana the one-period rise in inflation from 30 percent in 1982 to 115 percent in 1983 increased the nominal cNFPS deficit only slightly but reduced the operational deficit significantly as a result of the drop in real interest rates to negative levels. Since the mnid-1970s financial liberalization with partial or complete deregulation of interest rates has increased the sensitivity of deficits to the real interest rate. After early and radical financial liberaliza- tions in Chile (1974-75) and Argentina (1977), the 1980s saw partial or complete liberalizations in Mexico, Morocco, and Zimbabwe. Whereas the massive rise in real interest rates in Chie during the 1970s did not impinge on the deficit because of the virtual absence of domestic interest-bearing debt, the increasing domestic debt stocks of the 1980s, in conjunction with moderately high interest rates, added 30 Fiscal Adjustment and Macroeconomic Performnce: A Synthtesis to the burden of the central bank, which holds most of the domestic debt in the public sector. In Morocco partial liberalization of interest rates since 1984 has significantly increased the cost of domnestic debt to the treasury. It is estimated that a future increase of rates on gov- eminent debt to competitive market levels could add 2 percentage points of GDP to the deficit. Less access to foreign financing after 1982 forced countries to com- bine deficit reduction with increased reliance on domestic financing. A case in point is Pakistan. Its domestic nonbank borrowing increased from 4.8 percent of CDP in fiscal 1980181 to 7.4 percent in 1987188, and its domestic interest payments rose by 1.5 percentage points of GDP as a result of both higher domestic debt stocks and higher interest rates. REAL EXCHANGE RATE. A real exchange rate (RER) depredation raises public expenditure (measured in local currency units) by increasing foreign interest payments and the cost of tradable capital and intermediate goods acquired by the public sector. Public sector revenue is boosted by a real depredation that raises surpluses of firms producing tradable goods, as well as direct and indirect taxation on production or sales of tradable goods. The net effect of the RER on the deficit (in real terms or as a share of GDr-) hence depends on the relative weights of traded and nontraded items in public expenditure and revenue. Table 1.4 summarizes the effects of the RER on tax revenue, profits of state-owned enterprises (soEs), transfers, and consolidated defi- cits. In Colombia total tax revenue was reduced by real devaluation- presumably because a moTe devalued Riw reduces quantitative import restrictions or because of a highly elastic import demand. The op- posite is true for Ghana and Zimbabwe, where various revenue categories (direct and total taxes in Ghana, direct and indirect taxes in Zimbabwe) are increased by devaluation, presumably because tradable-goods activities (sales and production) are taxed more heavily than nontradable-goods activities. Because the remaining tax categories are shown to be insensitive to the RER, aggregate tax reve- nue rises with an RER devaluation in Ghana and Zimbabwe. Real devaluations have positive effects on public budgets in coun- tries in which a significant share of soEs consists of companies that produce tradable goods. This is especially true when the big com- modity exporters are public enterprises, as in Chile, Colombia, Mex- ico, and Morocco. Devaluations also boost net revenues from profits of agricultural marketing boards; this is clearly the case in Cote d'Ivoire. A computation of the net effect of the RER on the CNFPS deficit combines the previously mentioned effects on public revenue with William Easterly and Klaus Schmidt-Iikbbd 31 Table 1A. Effect of a Real Exchange Rate (RER) Devaluation on Public Deficits Negative_Positiue Effect on tax revenue Colombia: total taxes (1972-87) Ghana: direct taxes, total taxes (1970171 to 1988) Zimbabwe: direct taxes, indirect taxes (1970/71 to 1988189) Effect on profits or transfers from state-owned enterprses (SOEs) Chile: surplus of soEs and copper taxes Colombia: surplus of coffee fund and state oil company Cote d'lvoire: revenue from cocoa and coffee marketing boards Mexico: surplus of SOE Morocco: contributions of state phosphate company Net effect on the CNFPS deficit Increases defict Close to zero Lers defrcit Chile Colombia Mexico Ghana Thailand Zimbawe Source: Country case studies listed in the references to this chapter. the large and positive effect of the RER on foreign interest payments and with any effects on public expenditure. In some countries the interest effect dominates whatever positive effect the Rem has on the pmary deficit. The opposite is true in Mexico, where the share of oil- related federal revenue in CDP (7.9 percent in 1989) is more than twice as large as interest payments on dollar-denon-inated debt (3A percent in 1989). In Colombia, Thailand, and Zimbabwe a real devaluation has little or no influence on the overall deficit. ouTPur. Transitory output shocks affect nonfinancial public defi- cits because of changing tax bases and transfer payments to the pri- vate sector. This anticyclical behavior of public deficits motivated tra- ditional Keynesian prescriptions of using the budget as an automatic stabilizer to counteract "autonomous" demand shocks. In countries with nonindependent central banks or countries under extreme financial crises, the anticydical behavior of the nonfinancial deficit is reinforced by the anticyclical quasi-fiscal operations of the financial public sector. Cases in point are Argentina and Chile during the financial crisis and recession of the early 1980s (see box 1.1). Trend growth is sometimes seen as a cure for public deficits; if growth is high enough, it is argued, tax bases expand and countries 32 Fisal Adjustment and Macroeconomic Performance: A Synthesis Box 1.3. How Sensitive Are Deficits to Macroeconomic Shocks in Zimbabwe? Box table 1.3 shows to what extent Zimbabwe's cNFPS deficit is affected by domestic and foreign macroeconomic shocks. C(he estimates are based on 1987188 and 1988189 crFzs budgets.) The domestic real interest rate has a significant effect on the deficit as a result of Zimbabwe's high domestic public debt a 1 percentage point increase in the real interest rate raises the deficit by 0.4 percentage point of GOP, as the ratio of domestic debt to Gop stands at 40 percent. It is interesting that inflation has a lower positive effect on the deficit than does the real interest rate. The reason is that the effect on the deficit via higher nominal interest payments is in part compensated by the positive effect of bracket creep on revenue from income taxes. A devaluation contributes to a slightly lower deficit in Zimbabwe: the higher foreign interest bill is more than compensated by increased tax revenues from -import taxes and direct taxes on tradables-producing sectors. Growth seems to have a strong effect on deficits1 but its magnitude is overestimated because the calcu- lation considers the influence of cDp only on tax revenue, not on public expenditure. Box table 1.3. Macroeconomic Shocics and the CNNPS Deficit Change in CNFPSdeffdt Change in macroeonomik determinants (percentage points of GDP) Inaease in domestic inflation (1 percentage point) 0.31 Increase in domestic real interest rate (1 percentage point) 0.40 Devaluation of real exchange rate (1 percent) -0.06 Growth of real cDP (1 percent) -0.37 Increase in foreign interest rate tl percentage point) 0.25 Sourcc Chapter 10 in this volume, table 10.5. can grow out of deficits. This view is flawed for two reasons. First, it neglects the fact that not only tax bases but also successful pressures for higher public expenditure rise with output levels. Second, growth will not materialize if public deficits are high, inflation and real inter- est rates are high, and private investment is therefore depressed. Box 1.3 ilustrates the preceding analysis by showing how sensitive Zimbabwe's CNFPS deficit is to changes in the four domestic maao- economic variables and the foreign interest rate. William Easterly and Klaus Schmidt-Hebbel 33 Fiscal Policies In tiis section we compare the influence of fiscal policy variables and that of foreign and domestic macroeconomic variables in the evolu- tion of public deficits. Using time-series results for the decomposition of public sector deficits according to the three groups of deficit deter- minants (based on the deficit decomposition methodology of Mar- shall and Schnidt-Hebbel 1989), we compute the contnbution of each group to changes in public defidts. Figure 1.2 presents the average relative contribution of the tree groups of variables to changes in cNns deficits and the pattern of deficts in Chile, Ghana, and Zim- babwe."' The evolution of the public deficit in these three countries reflects the influence both of temporary (or cyclical) shocks and, par- ticularly in the cases of Chile and Ghana, of structural po'iy shifts that brought about lower trend deficits. Chile's fiscal experience in 1973489 reveals four distinct periods: mas- sive fiscal stabilization (1973-76); consolidation of pubic sector re- trenciment (1977-80); crisis and deficit explosion (1981-4); and again, significant fiscal stablization (1985-89). Fiscal policyaakers are the main actors behind this experience, which achieved CNRFS surpluses dose to 5 percent of GDP. On average, the relative contribution of fiscal policy variables to changes (and therefore to trend reduction) in the deficit was 142 percent. Changes in fiscal policy variables thus compen- sated for the strongly negative contribution of domestic macroeconomic variables and the slightly negative contnbution of external variables. Ghana is a case of gradual, but also highly successful, fiscal adjust- ment. There, too, the contribution of fiscal policy variables to the turnaround was massive, explaining 91 percent of the change in the deficit. Improvements in domestic macroeconomic variables helped to a smail extent, contnbuting 11 percent to the fluctuations and structural correction of the central government defict in Ghana. The substantial deterioration in Zimbabwe's cNFPs budget after 1980 was partly reversed when a limited fiscal stabilization began in fiscal 1987188. Zimbabwean policymakers compensated for the influ- ence of variables beyond their control: fiscal policy variables explain 110 percent of the variation of public deficits, neutralizing the nega- tive contribution of foreign interest shocks to the deficit. A central conclusion emerges from the cases of Chile, Ghana, and Zimbabwe: fiscal policy variables dominate absolutely these countries' experiences of fiscal adjustment or deterioration. External and domes- tic macroeconomic shocks play a minor, and often even negative, role in the cyclical variation and the structural changes in public sector budgets. Active fiscal policies are both the main culprits in fiscal crises and effective instruments in bringing about fiscal stabilization and adjustment. 34 Fiscal Adjustment and Macroecontomic Performance: A Synthesis Figure 1.2 Nominal CNFPS Defict and Deficit Decomposition by Main Determinants, Chile, Ghana, and Zimbabwe, 197.1-89 Defiit (percentage of GDP) 25 - no 20 91 Zimbaw le * -2 10 V 0 -10 4 1971 1975 1980 1985 1989 a--l Ghanaa = External variables Chile Domestic macroeconomic vanables ' Zimbabweb Fiscal policy variables a. Central governmentdefict; fiscal yeardata forl971/72 to 1981/8/ b. Fiscal year. Source- Authozs calculations based on country data in chapters4, 7, and 10 in this volume. Table 1.5 identifies the contribution of specific policies in ten rele- vant country experiences, one for eadh country in the sample. Three countries (Chile, Mexico, and Thailand) achieved strong and rapid fiscal adjustment. Four (Colombia, Ghana, Morocco, and Zimbabwe) followed a more gradualist approach of fiscal retrenchment. Pakistan experienced moderate deterioration, and Argentina and CBte d'Ivoire experienced massive fiscal deterioration. Loss of control over public consumption (particularly wage levels and employment levels) is a major cause of a looser fiscal stance. One dramatic example is Argentina, where as a result of the large increase in current expenditure, investment fell by almost 5 percentage points of CDI' during the period of extreme fiscal deterioration, 1977 .82. The other is Zimbabwe, which, during a period of fiscal retrenchment William Easterly and Klaus Schmidt-Hebbel 35 (1986187 to 1988/89), was not able to avoid further increases in its public wage bill by 4.0 percentage points of CDP. Cote d'Ivoire's fiscal deterioration, too, was partly attributable to rising current expendi- ture. Conversely, the examples of strong austerity policies in Chile (197375), Ghana (1975176 to 1988), Mexico (1986-89), and Thailand (1985-88) illustrate the important role played by reductions in cur- rent expenditure and, in particular, by cuts in wages and public employment. Cutting transfers and subsidies is often an effective way to contrib- ute to both fiscal stabilization and market deregulation. In Ghana and Zimbabwe lower transfers and subsidies contnbuted greatly to deficit reduction, by 5.4 and 5.0 percentage points of CGDP, respectively. On the revenue side, tax reforms are at the heart of efforts to address structural deficits. In Chile reforms of direct taxes and the introduc- tion of the value added tax (vcr) raised revenues by a staggering 10.5 percentage points of cisp, while Zimbabwe's 1988 tax reforms yielded a significant 4.2 percentage points of CDP. 1-igher tax revenue also helped Colombia, Mexico, and Thailand to reduce their deficits. Rationalization of public enterprises and reforms of agricultural marketing boards constitute the fourth element of successful stabiliz- ation in our ten-country sample. Higher operating surpluses of soES contributed significantly to improving structural defidts-in Chile by a dramatic 8.4 percentage points of GDP and in Colombia and Ghana by smaller amounts. Conversely, the drastic deterioration in Cote d'Ivoire was caused by the decline in revenue from the cocoa and coffee revenue stabilization funds as a result of continuing producer price supports during a period of dedining world prices. An encouraging finding from our sample is that successful fiscal retrenchment does not have to rely on lower public investment. In the most dramatic fiscal tumaround (Chile, 1973-75) public capital formation was not reduced. In the three countries in which public investment fell during fiscal adjustment-Colombia, Ghana, and Mexico-the reduction was moderate. Only one case of fiscal retrenchment (Thailand) relied heavily on cutting public investment. Conversely, the two largest declines in public investment occurred in Cote d'Ivoire and Argentina durirg periods in which public deficits exploded. In the case of Argentina public capital expenditure contin- ued its systematic decline beyond 1982, reaching a thirty-year low of 6.1 percent of GWD in 1987. We conclude that successful (that is, sustainable) nonfinancial pub- lic sector adjustment typically requires simultaneous action on four fronts: reducing an overblown government bureaucracy; cutting transfers and subsidies to the private sector (except for efficient and targeted social programs); enacting tax legislation for increased, Table 1.5. Contribution of Policies to Fiscal Adjustment or Deterioration in Ten Countries (percentage points of COP) Deficit level at start Country experiemr and end of period Change i;i deficit and contribuftio of fiscal policy chaanges Argentina: 4.7; 15.1 Change in deficit +10.4 1977-82 Higher current expenditure +15.1 deterioration Lower capital expenditure -4.7 Chile; 20.6; 2.1 Change in deficit -18.5 1973-75 Lower public employment -4.3 adjustment Higher revenue from tax reform -10.5 Higher soE operating surplus -8.4 Colombia: 6.3; 2.2 Change in deficit -4.1 1984-89 Lower public wages/salaries -1.2 adjustment Lower fixed investment -2.1 Higher tax revenue -2.1 Higher SOE operating surplus -1.9 COte d'lvoire: 1,7, 14.4 Change In deficit +12.7 1984-89 Higher current expenditure +3,6 deterioration Lower capital expenditure -5.1 Lower tax revenue +2.9 Lower revenue commodity fund +12,7 Ghana: 15,1; -0.4 Change in deficit -15.5 1975/76-88 Lower wage bill -1.3 adjustment Lower expenditure on goodslservices -1.6 Lower transfers/subsidies -5.4 Lower public Investment -1,8 Mexico: 14.9; 5.1 Change in deficit -9.8 1986-89 Lower current expenditure -2.5 adjustment Lower other expenditure -4.6 Lower public investment -0.7 Higher direct tax revenue -3.0 Higher VAT revenue -0.9 Morocco: 12.1; 4.1 Change in deficit -8.0 1983-88 Lower expenditure on goods/services -2.9 adjustment Lower transfers/subsidies -1.7 Lower capital expenditure -3.3 New petroleum levy -3.4 Pakistan: 4.8; 8.3 Change in deficit +3.5 1980/81 to 1986187 Higher noninterest current expenditure +2.9 deterioration Lower direct tax revenue +0.8 Lower indirect tax revenue +1.9 Thailand: 8.6; -0,2 Change In deficit -8.8 1986-88 Lower public wages/salaries -1.4 adjustment Lower public investment -3.5 Higher revenue -2.2 Zimbabwe: 14.4; 10.0 Change in deficit -4.4 1986187 to 1988/89 Higher public wages/salaries +4.0 adjustment Lower transfers/subsidies -5.0 1988 direct tax reform -2,4 1988 custom duty reform -1.8 Note: SOE, state-owned enterprise; vAT, value added tax. Data refer to the central government for Ghana and Morocco and to general government for Pakistan. In all other cases the data refer to the consolidated nonfinancial public sector deficit. Soutrce: Country case studies listed in the references to this chapter. 38 Fil Adjustment and Macroeconomic Perfornamce: A Synthesis broadly based direct and indirect taxation; and reforming or privatiz- ing public enterprises and commodity marketing boards. Efficient public investment, particuarly in social or physical infrastructure, should not only be exempted from fiscal cuts but should be expanded to encourage economic growth. Deficits, Inflation, Real Interest Rates, and Financial Repression As shown by the cross-country evidence in figure 1.3, the relation- ships between deficits and inflation and between deficits and interest rates are far from simple. At low to medium rates of inflation, there is no relationship across countries between long-term inflation rates (1980-88) and public deficits. However, the countries with the highest inflation rates-Argentina and Mexico during the 1980s-had signifi- cantly higher deficits than counties with lower rates. Similarly, domestic real interest rates show no correlation with public deficits across countries except in the case of high-deficit, high-interest-rate Argentina. The lack of correlation across countries between deficits and infla- tion and between deficits and interest rates is primarily attributable to the different ways in which countries finance their public deficits. To account for the effects of these differences, a more detailed under- standing is needed of the links between domestic deficit financing and inflation and interest rates (or financial repression). Thiis section first considers the relationship between seigniorage, inflation, and money demand and compares steady-state seigniorage levels with one-shot seigniorage episodes. It then reviews the empiri- cal evidence on the relationships between specific sources of deficit financing and inflation, real interest rates, and financial repression. Seignorage Laffer Curve and Misspecified Money Demand Any notion that fiscal deficts and inflation display a simple relation- ship fails for two reasons. The first is that countries make different choices about seigniorage to finance their deficits, partly because they differ in the extent to which other means of finance are available. The second reason is that money creation and inflation are nonlinearly related. The scattergram shown in figure 1.4 suggests a conventional "Laffer-curve" relationship between the inflation rate and revenue from seigniorage, with revenue falling off at some point because of the elastic response of money demand. The exact maximum of the curve is sensitive to the inclusion of the extreme points: with Argen- tina the maximum is at 160 percent inflation, while without Argentina it is only at 68 percent. (A similar point is made by Fischer and William Easterly and Klaus Schmidt-H ebbel 39 Figure 1.3. Fiscal Deficits, Real Interest Rates, and Inflation Rates in Case Study Countries, 1980-88 Averages Consolidated total public sector deficit (percentage of CDP) A ---------- 20.01 ; , ___________ __--_ _ _---- :-----*A-- ---- -5~~ ~ a Z-imbabwe A ---- --Momroan ----------- Cotrdlitin:s-A--- ---------!------- ---- --et ChileA--______\r_ 7-5- __ IL ----- .rlkistan Thailand *---- --- ---- A-. 52 ----.Colombia Cases of\ Realinterst financaal - Inflationrate rate (percent) rA A (Percent) 10 5 0 -5 -10 -15 10203040506070 '240 Note- The consolidated total public sector defidts are for the CNFPS in all ten countries, but quasi-fiscl deficits are also included for Argentina, Chile, and Mexico. ource: For the defict series, the country studies (see References); for the inflation rates and nominal interest rates (used to compute real interest rates), an, Internationl Financial Statistics, various years. Easterly 1990, who also note that the growth rate affects whether deficits are inflationary.) Econometric estimation of a quadratic equation statistically con- firms the Laffer curve, as shown by the following cross-country relationship: TfY = 0.01 + 0.043r - 0.137r2 (4.9) (4.1) (-2.31) R2 = 0.44 where TY is average seigniorage revenue as a ratio to GDP in 1970-89 and r is average inflation in 1970-89; t-statistics are in parentheses.'2 These cross-section results differ significantly from calculations of revenue-maximizing inflation from individual time-eries results for the case studies.'3 A regularity is that countries with high-inflation have very high seigniorage-maximizing inflaticn rates (in Argentina the rate is 966 percent, in Chile, 792 percent, and in Ghana, 125 percent); countries with moderate inflation have more moderate max- imizing rates (Colombia's is 80 percent); and countries with low infla- 40 Fiscal Adjustment and Macroeconomic Performance: A Synthesis Figure L4. Inflation and Seigniorage Seigniorage (percentage of rDP) * ~~~~~~Including Argentina 4- 3 Exdudmg Argentina 0 o 20 40 60 80 100 120 140 160 180 200 Inflation, rate (percent) Source: Authors' regression and Statistical Appendix, tables AS3 and AA. tion have low maximizing inflaGon rates-Thailand's is only 4 percent! One hypothesis to explain these huge differences across counties is misspecification of money demand. Conventional estimates of the seigniorage-maximizing inflation rate typically make use of a Cagan money demand, which implies a constant semielasticity of money demand with respect to inflation or interest rates (Cagan 1956). East- erly, Mauro, and Schmidt-Hebbel (forthcoming) show that the elastic- ity of substitution in transactions between money and bonds deter- mines how the inflation semielasticity of money demand changes as inflation rises. Allowing for a variable semielasticity, the authors report estimates of seigniorage-maxinizing inflation-varying between 266 and 303 percent per year-for a panel sample of eleven high-inflation countries. Their results are consistent with a semi- elasticity that increases with inflation (that is, higher inflation hastens the flight away from money and toward financial assets that provide protection from inflation).14 Steady-State Seigniorage versus One-Shot Seigniorage Episodes Given the attention devoted to seigniorage in the literature, it is easy to forget how small it is as a source of revenue. Table 1.6 shows the William Easterly and Klaus Schmidt-Hebbel 41 Table 1.6. Average Seigniorage in OECD and Developing Countries Average seigniorage, Highest 1970-88 excise Prxduct subject to Country (percentageof GOP) tax, 1985 highest excise tar OECD countries Austria 0.9 iO Wine Belgium 0.5 1.1 Mineral oil Canada 0.4 0.7 Gasoline Denrmark 0.4 1.1 Cigarettes Finland 0.6 1.3 Fuel France 0.6 0.4 Insurance Germany, Fed. Rep. 0.7 1.3 Mineral oil Greece 2.8 2.2 Fuel Italy 2.2 1.7 Mineral oil Japan 1.0 0.6 Liquor Netherlands 0.6 0.8 Petroleum Norway 0.6 1.5 Vehide transfer Spain 2.3 1.3 Petroleum Sweden 0.6 1.0 Petroleum United States 0.4 0.3 Motor vehicle fuels Average 1.0 1.1 n.a. Developing countries Argentina 4.2 2.5 Fuel Bangladesh 1.0 - - Bolvia 2.9 - _ Brazil 2.3 0.2 Electricity Burkina Faso 1.1 0.7 Beverages Chile 3.7 - Colombia 21 0.6 Gasoline Cbte d'Ivoire 1.3 1.1 Petroleum Dominican Republic 1.6 1.8 Petroleum Ecuador 1.8 0.3 Beer Ghana 3.1 - - Honduras 0.8 0.5 Beer India 1.5 0.7 Textiles and Jute Indonesia 1.4 0.9 Tobacco Jamaica 1.9 - Jordan 5.0 Kenya 1.1 - Korea 1.6 0.8 Liquor Malawi 2.0 - Malaysia 1.3 0.7 Petrol Mexico 3.1 1.4 Gasoline Morocco 1.7 1.2 Tobacco Nigeria 1.1 - Pakistan 2.0 - Paraguay 1.9 0.9 Fuel (Table ontinues on the fUowing page.) 42 Fiscal Adjustment anid Macroeconomic Perfonnance: A Synthesis Table 1.6 (continued) Average segnio rage, Highest 2970-88 excise Product subject to Country (percentage of GDP) tax, 1985 highiest excise tax Peru 3.6 4.1 Gasoline Philippines 1.0 - - Sri Lanka 1.3 - - Thailand 1.0 1.5 Petroleum products Trinidad and Tobago 0.9 - - Turkey 3A - Venezuela 1.5 0.5 Liquor Zaire 4A 0.3 Tobacco Zambia 2.0 1.9 Petroleum Zimbabwe 1.1 - Average 2.1 1.1 n.a. - Not available. n.a. Not applicable. Source: For excise taxes, IMp 1986; for seigniorage, Statistical Appendix, table A.2. average seigniorage for a sample of industrial and developing coun- tries for which data are available. Seigniorage is calculated as the ratio to real GDP of the yearly sum of deflated monthly changes in the money base. The generally small amount of seigniorage for the ten case studies is typical of the overall pattern of seigniorage among all countries. The maximum amount of average seigniorage revenue over an extended time is less than 5 percent of GDP. Seigniorage is mainly a phenomenon of developing countries-among industrial countries, only Greece, Italy, and Spain had seigniorage above 1 per- cent of CGP. Average seigniorage is more than twice that level in developing countries.'5 Seigniorage revenue is of the same order of magnitude as revenue from individual excise taxes (see table 1.6). Why, then, are macro- economists so preoccupied with taxes on money, as against taxes on beer, jute, or cigarettes? Perhaps one reason is that seigniorage can be a large source of temporary revenue during times of crisis. The time-series averages for seigniorage conceal tremendous year-to-year fluctuations. Figure 1.5 shows a frequency distribution of the individual yearly observa- tions for the same sample of countries as in table 1.6. Although nearly half the sample is concentrated in observations of less than 1 percent of CDP, there is a significant number of observations of high- seigniorage revenue, reaching as high as 13 percent of CDP. The aver- age time-series coefficient of variation in the sample is 90 percent.16 This suggests that a fruitful approach to seigniorage would be to study the episodes of high seigniorage to see how they happened and William Easterly and Klaus Schmidt-Hebbel 43 Figure 1.5. Frequency Distribution of Annual Seigniorage Observations, Fifty-one Countries, 1970-88 Frequency Cumulative frequency (percent) (percent) 40 .10 90 80 30 70 -60 20 50 40 -30 '10 20 10 4-3 3-2 2-1 1-0 0-1 1-2 2-3 3-4 4-5 5-6 77-8 8-9 9-10 10-11 11-1212-13 L- LNegative -J Seigniorage (percentage of GDP) O Frequency Cumulative frequency Sou rce: Statistical Appendix, tables A.3 and A4. what their consequences were. A number of the case studies in this project include such episodes; bursts of seigniorage appear in Argen- tina in 1975 and 1983, in Chile in 1971, in Ghana in 1978 and 1983, and in Mexico in 1982. A different type of one-shot seigniorage took place in Ghana in 1979, when the government captured 2.5 percent of CDP through a currency conversion and partial expropriation of deposits, and in 1982, when the government again expropriated private wealth through the demonetization of the largest-denomination note. These episodes brought a short-term gain to public finances at considerable long-mn cost-the Ghana case study estimates tat seigniorage was permanently lowered by 1 to 2 percent of GDP because of the fall in money demand after the 1979 currency expropriation. Table 1.7 looks at episodes of high (conventional) seigniorage in the broader sample. Of a total of 1,143 observations of forty-nine coun- tries for various years, we identified eighteen instances in which seig- niorage was more than 4 percentage points above the average seigniorage-to-GDP ratio in the particular country.'7 We see that Table 1.7. Episodes of High Seigniozage Spikes Comnmuetnls of spike Iiinlatiou (percentt, December Seigniorage (percentage of CGDP) (perceitage of GDP) to Deceimber} Sedgniorage Chiange hilGovt (rci) Clawuge iv: spike rearl moey AbWcv Growlt (percent) inflatiorr Yr (o eAitiorg base + average Growlit Average Inflation otvr Aivrage Year of from Averae average inflation fn year growtvt int year of previous infain/ion Coutntry spike average) seigniorage mlontey base tax of spike mMe spike year rae Argentina 1975 9.0 4.2 -4.0 6.5 -0.5 2.2 336.1 296.2 105.4 Bolivia 1982 7.5 2.9 -0.8 7.9 -4.4 2.7 296.6 271.4 54.5 1983 5.3 - -3.3 8.6 -6.5 - 327.8 31.3 - 1984 7.0 - -3.9 20.8 1.0 - 2,176.2 1,848.4 - Chile 1971 6.7 3.8 10.9 -3.0 9.1 1.5 19.4 -15.6 91.3 Denmark 1985 4.2 0.4 4.3 -0.2 4.3 2.4 3.6 -2.0 7.7 Dominican Rep. 1986 4.9 1.6 5.3 -0.6 3.0 5.7 6.5 -21.8 13.1 1988 5.7 - 2.2 3.0 1.3 - 57.5 32.5 - Ghana 1978 4.7 3.1 -0.6 3.8 9.8 1.7 108.5 -2.4 38.6 Jamaica 1984 5.4 1.9 4.7 1.0 -1.4 -Q.2 31.2 14.5 17.0 Mexico- 1982 7,9 3.1 1.4 3.7 -0.6 5.0 98.9 70.2 29.8 Peru 1985 6.7 3.6 2.7 2.5 2.2 2.2 158.3 46.8 82.1 1988 6.0 - -7,3 14.8 -8.0 - 1,722.1 1,607.5 - Trinidad/Tobago 1982 4.2 0.9 4.3 0.0 -4.7 1.6 10.8 -0.8 10.5 Zaire 1976 4.1 4.4 0.5 2.2 -5.5 3.1 78.8 42.1 47.1 1982 4.9 - 4.4 -0.6 -0.4 - 41.0 -12.0 - 1987 5.7 - -0.1 4.1 0.6 - 106.5 68.2 Zambia 1986 6.0 2.0 5.4 1.1 0.2 1.8 34.6 -23.7 18.9 Average for "spike" episodes or countries 8.8 2.7 1.5 4.2 0.0 2.5 311.9 236.2 43.0 Average for thirty-nine "nonspike" countries in sample - 1.4 - - - 4.5 - - 11.2 -Not available. Nole: Spikes are defined as seigniorage more than 4 percentage points of cGD above the average seigniorage-tO-cr ratio for the country. Components do not sum to the "spike" because of the covariance term. Sortrce: Authors' calculations based on data in the Statistical Appendix and In IMF, International Fitnacial Statistics, various issues. 46 Fiscal Adjustment and Macroeconomic Performance: A Syn thesis "spikes" of high seigniorage are indeed short-lived. The only epi- sode that lasted more than one year was the Bolivian hyperinflation of 1982-84. The episodes are associated with developing countries; among OECD countries only Denmark indulged in a seigniorage spike. One might have thought that these bursts of seigniorage revenue would be associated with accelerations of inflation, but this hypoth- esis is, surprisingly, not confirmed by the data. Of the sixteen epi- sodes (treating Bolivia, 1982-84, as a single episode), only nine showed rising inflation-roughly the same proportion as in the broader sample. There is no evidence for the supposition that infla- tion shows a lagged response, as the following year's inflation also shows no tendency to accelerate. Some of the episodes of rising infla- tion are quite spectacular. Bolivia in 1982-84 and Peru in 1988 experi- enced classic hyperinflation in which real money demand fell but inflation soared to four digits. Large accelerations of inflation also took place in Argentina in 1975, Mexico in 1982, and Zaire in 1987. A decomposition of the seigniorage spikes into components associ- ated with the real change in the money base and the inflation tax helps explain the cases in which inflation did not accelerate (see appendix 1.3). Table 1.7 shows that the real change in the money base explains most or all of the above-average seigniorage in seven of the sixteen cases. In six of these cases, inflation declined. The lack of a dose association between acceleration of inflation and bursts of seig- niorage comes about because in nearly half the cases seigrniorage was driven mainly by real money balances. An understanding of this phenomenon would require more careful examination of individual cases, but apparently there was scope for temporary increases in seigniorage revenues through actions such as raising reserve require- ments or through exploitation of exogenous increases in demand for money. Price controls were used in Chile in 1971 to generate a "real" change in money demand, but inflation exploded in the following yearm8 Of course, the dassic inflationary method worked just as well as a method for generating bursts of seigniorage. One-time inflation taxes that were more than 8 percentage points of GDP above average were recorded in the hyperinflation in Bolivia and Peru, while less spec- tacular increases were registered in Argentina, Ghana, and Zaire. The money base fell in all these cases, limiting the potential for further inflation taxes. The CDP growth rates during the seigniorage spikes were not so high as to make money creation a recommended method of raising revenue. In twelve of the sixteen episodes growth during the episode was below the average growth rate for that country, and eight cases actually registered negative growth of gross-not per capita-output. We must remain skeptical about whether growth was poor because of William Easterly and Klaus Schmidt-Hebbel 47 the unusually high seigniorage or countries resorted to seigniorage because economic recession dried up other revenue sources. But it is interesting that the countries with spikes have a lower average growth than other countries in the sample, perhaps reflecting a ten- dency toward higher average seigniorage and inflation, as well. In condusion, seigniorage may be more important as a source of temporary increases in revenue than as a steady-state phenomenon. But the link between these temporary seigniorage surges and infla- tion is weak- A surprising number of episodes of high seigniorage are attributable to increases in real money balances instead of to acceler- ated inflation, illustrating the scope for achieving temporary revenue increases through various actions by the monetary authorities other than printing money. But the poor performance of countries that resort to such measures does nothing to further the case for using 'bursts of seigniorage as an instrument of public finance. Fiscal Deficits, Seigniorage, and Inflation Average long-term (1965-89) monetary financing or seigniorage is 2.3 percent of GDP in the ten sample countries (table 1.8)-close to the average of 2-1 percent for all developing countries, but twice the level of 1.0 percent for the OECD countries (see table 1.6). Seigniiorage arnd inflation show an association across the ten sample countries. Thi short-run relationship between money financing and inflation is typ- ically blurred by factors such as indexation practices, slowly changing expectations, slow portfolio substitution, and inflationary exchange rate depredations. In the long run, however, the tradeoff between inflation and money creation becomes increasingly unfavorable, explaining why seigniorage is generally used ordy as a last resort. Table 1.8 reports the amount of additional inflation required to achieve another per- centage point of GDP in long-mn seigriorage revenue. The figures are derived from estimated Cagan-type, constant-inflation or interest- elastic, money demands for the case countries. The tradeoff is still favorable in countries with low inflation (in -nailand the additional inflation is just 5 percentage points), worsens in moderate-inflation countries (15 to 20 percentage points in Colombia and Ghana), and becomes untenable in countries with high inflation (97 percentage points in Argentina), where moneyholders replace most of their local currency holdings with foreign currency and interest-bearing assets. These results are remarkably similar to the simulation results in table 1.9, which are based on more comprehensive portfolio substitu- tion and deficit-financing models (and, in the case of three countries, general equilibrium macroeconomic frameworks). The long-term effects on price levels of transitory deficits financed by money cre- 48 FiscalAdjustment and Macroeconomic Performance: A Synthesis Table 1.8. Seigniorage, Inflation, and Margnal Inflation Revenue, 1965-9 Perrcentage incres in inflation required to Seigniorage collkt additional (pecnage Intlation seigniorage of? Country of GDP)a (P4rrt)b percantage point of GDP Case study countries Argentina 4.2 1153 97 Chile 3.7 56.6 23 Colombia 2.1 17.7 15 C6te d'Ivoire 1.3 7.6 Ghana 3.1 31.6 20 Meidco 3.1 28.9 - Morocco 1.7 6.1 8-26 Pakistan 2.0 8.0 Thailand 1.0 5.7 5 Zimbabwe 1.1 7.7 10 Average 2.3 28.5 n.a. Othercountries Average, 35 developing countries 2.1 - n.a. Average, 15 industrial countries 1.0 n.a. - Not available. n.a. Not applicable. a. Seigrioage is defined as the nominal change in the money base each month divided by the consumeT price index (crn) for that month. The typical method of caliclating the ratio of the nonina chage in the y base over theentire year to the annual nominal crDp can seriously overstate seigniorage in high-inflation countries. Although interest paid on reserves should also be subtracted to get a true estimate of seigniorage, the data are generally lacking, and, in any case, few developing countries pay interest on reserves. Where interest is paid, it appears that it is quantitatively unimportant. An imnportant exception is Argentina, where the combination of high inflation and interest paid on reserves makes this adjustment important. We use the Argentine seigniorage series used by Rodriguez in chapter 3 of this volume. Periods covered are generaly 1965-89 but vary depending on the availability of data. b. Average annual rates of change in the cm between 1964 and 1988. Source: For annual data, Statistical Appendix, table A.2. For average annual rate of change, IMF, International Financial Statistics, various years. For Argentina, Colombia, Ghana, and Morocco, data in the last column are from the country case studies listed in the references to this chapter; for Chile, Thailand, and Zimbabwe, these data are calculated from seigniorage and inflation rates in the first two columns and from long- run money demand inflation semielastidties in the country studies. ation, taling into account feedback effects on inflation from asset substitution (and endogenous output response in the cases of Pakistan and Colombia), are similar to those obtained using the sim- ple Cagan form. For the four reported countries, the additional infla- William Easterly and Klaus Schmidt-Hebbel 49 Table 1.9. Simulation Results for Long-Term Effects of Fiscal Deficits on Inflation and Real Interest Rates (percent) Effect ofa I percentage paint increase in the deficit-to-COP ratio On the ral interest On the pnce le;d, rate, with domestic Country uith moneyflnandng debtfinancing Chile 14 0.1 Colombia 14 3.0 Morocco - 0.2 Pakistan 18 1.1 Zimbabwe 10 2.7 -Not available. Note: This table presents the long-term effects of a transitory (one-year) increase in the public deficit, financed by issuing either domestic non-interest-bearing monetary liabilities or domestic interest-paying debt. Both the short-term effects and the cross- effects (of money financing on the real interest rate and of debt financing on inflation) are of less interest and vary from country to country because of differences in model structures. The results for Chile and Zimbabwe are based on portfolio models com- birted with the public sector budget equation; those for Colombia, Morocco, and Pakistan are based on maoeconomic-ptfolio general equilibrium specifications. Sou rce Ccuntry case studies listed in the references to this chapter tion required to collect 1 additional percentage point of GDP through seigniorage ranges from 10 percent for Zimbabwe to 18 percent for Pakistan. Considering the unfavorable tradeoff in most cases and the widespread consensus on the undesirability of inflation, it is difficult to believe that revenue motivations alone explain chronic high infla- lion. (See Blejer and Liviatan 1987 and Kiguel and Liviatan 1988 for similar conclusions.) Fiscal Deficits and Interest Rates or Financial Repression There are two ways in which fiscal deficits can affect domestic real interest rates and financial markets. First, if interest rates are not controlled, a high fiscal defidt financed through domestic borrowing would be expected to result in high real interest rates. Second, if interest rates can be and are controlled, the implict tax on financial assets could be a hidden source of revenue for the government. By liberalizing interest rates, financial reform has shifted deficit financing from implicit financial repression revenue to explicit debt issuing in many developing countries. Argentina, Chile, Colombia, Morocco, Pakistan, and Thailand introduced financial reforns in the 1970s, and their real interest rates reached positive levels in the 1980s (table 1.10). Ghana, Mexico, and Zimbabwe, however, mnaintained varying degrees of domestic interest controls during most of the 1980s 50 Fiscal Adjiustment and Macroeconomic Performance: A Synthesis Table 1.10. Real Interest Rates under Financial Reform or Financial Repression in the 1980s Tax revniue on deposits Real intterest rates on «depisis affributable to financiat ,ercno repression, 1980-88 Country 1970-79 1980-88 (pffrentageofCDP.)' Argentina -17.2 4.8 nra. Chile -15.9 8.1 n.a Colombia -6.3 0.7 na. Ghana -18.8 -18.3 0.5 Mexico -4.6 -8.4 1.6 Morocco -3.1 1.8 n.a. Pakistan -3.4 2.1 na. Thailand -0.5 6.5 n.a. Zinbabwe -3.7 -4.3 0'Se n.a. Not applicable. a. Average annual real interest rates on time deposits at banking system, calcuated using the consumerprice index. b. Average annual revenue calculated as the difference between domestic real inter- est rates and the average real interest rate in OECD countries (0.9 percent) multiplied by deposits outstanding as a percentage of mDr. c. 1980-87. Sourr Country case studies listed in the references to this chapter; Statistical Appendix, tables A.A and A-5- (Mexico ltberalized its rates in 1988), as reflected by negative average real interestrates. The implicit tax from financial repression is normally computed as the product of the interest tax (typically, the difference between the foreign and domestic interest rates) and the outstanding stock of the relevant public liability or the time deposits in the financial system. In the latter case the financial repression revenue collected by the finan- cial system is often transferred back to the public sector via com- pulsory placements of government debt in financial institutions or through unpaid reserves held at the central bank. Average annual revenue for the three countries from financial repression of deposit interest rates during 1980-88 ranged from 0.5 percent of GDP for Ghana to 1.6 percent of GDP jor Mexico (see table 1.10). Holding down nominal interest rates under high inflation was a quick and easy way of obtaining revenue to compensate for the loss of external financing after 1982. Table A.5 of the Statistical Appendix presents estimates from other studies of revenue from financial repression. Although calculations differ widely because of different methodologies, there is a consistent finding that Ghana, Mexico, and Zimbabwe reaped significant amounts of revenue from controls on domestic interest rates during the 1980s.19 William Easterly and Klaus Schmidt-Hebbet 51 Revenue from financial repression in these three countries is com- parable to the 1.0 to 2.1 percent of GDP of average seigniorage col- lected in OECD and developing countries (see table 1-6). Although the implicit tax from financial repression is less visible than seigniorage and its inflation tax component, its deleterious effects on financial intermediation-and hence on the quantity and quality of private investment-is probably as strong as that of inflation. In fact, control- ling interest rates was a costly strategy for private credit and invest- ment, which remained depressed throughout the 1980s.2A In reaction, partial decontrol of interest rates in Ghana and Zimbabwe and com- plete interest liberalization in Mexico since the mid-1980s have reduced or abolished taxes on financial intermediation. Figure 16 shows the evolution of domestic private credit in the case studies. There are large differences in domestic private credit stocks between countries with deregulated financial markets-where private credit reaches an average 30 percent of GDP-and those with stringent financial controls, where the corresponding average ratio hovers around 10 percent. Mexco's experience well illustrates the effects of financial repression. Financial controls intensified after 1981 as infla- tion soared, and the ratio of private credit to cDP dropped below already low levels. Following financial liberalization in 1988, the ratio doubled within two years. In Ghana private credit was at a dismally low level, reflecting years of financial repression, induding two epi- sodes of outright expropriation of financial assets. By contrast, coun- tries that abstained from repressive interest rate controls, such as Chile and Thailand, had very high levels of private credit, which may partly explain their superior investment and growth performance in the late 1980s. The alternative to financial repression is government borrowing at market interest rates. Table 1.9, above, reports simulation results (based on the portfolio and general equilibrium frameworks referred to earlier) for the long-term effects on the real interest rate of a transi- tory 1 percentage point increase in the deficit (in relation to GDP), financed by floating domestic debt, in five country cases. The effects vary widely in the five countries, reflecting differences in the willing- ness of asset holders to shift from alternative forms of saving. Flat demands for domestic debt imply that real interest rates increased by a modest 0.1 to 0.2 percentage point in Chile and Morocco. Low asset substitutability between domestic debt and alternative private sector asset holdings in Colombia, Pakistan, and (after interest decontrol) Zimbabwe-partly because of high domestic debt levels-explains increases in real interest rates that range from 1.1 to 2.7 percentage points. The mplication for those ffiree countries is that when domes- tic borrowing is high and costly-which could lead to a domestic debt spiral such as the one in Argentina described in box 1.4-there is no 52 Fiscal Adjustment and Macroeconomic Performance: A Synthesis Figure 1.6. Private Credit under Financial Liberalization and Repression in Nine Countries, 1980-90 Without interest rate controls Percentage of GDP 70 --- / \ 60 -/ _ _ Chile 50 -G 50 / ~~~~~~Thailand_ fd Cote d'lvoire _---___--^ l 40 7, 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 WNith interest rate controls Percentage of CDP 30 Ps 20 ZMorocco 10 0 ~ r - I " - 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 SOzn-ce: !F, !tternatWinal Financal Stntistics, various issues. William Easterly and XJaus Schmidt-Hebbel 53 Box 1.4. Argentina's Unusual Financial Behavior The massive decline in Argentina's ratio of private credit to GDP, shown in figure 1.6, reflects an unusual kind of financial behavior. The government, far from controlling interest rates, oscillated between paying high interest rates and "melting down" domestic liabilities through surprise devaluations and other methods (including a forced conversion of time deposits into govemn- ment bonds of questionable value in 1990). This tactic was necessary because the high interest rates themselves fueled the accumulation of more debt, in a classic example of a debt spiral, or Ponzi game. Borrowing was increased in order to pay the interest on the debt, which implied yet higher interest and borrowing in the next period, and so on. Although the government managed to keep persuading the public to buy domestic debt, increasingly high inter- est rates were required. The Argentina case study in this book chronides the rise in nominal interest rates at the outset of successive economic plans, each of which opened with a devaluation. Initia Nominal interest deualuation ratepermonth Stabilizon plan and date (percent) (percent) Austral (June 1985) 40 7 Primavera (August 1988) 24 10 Bunge-Born I (July 1989) 200 17 Bunge-Bom 1 (December 1989) 54 60 Erinan January 1990) 220 100 choice except to continue financial repression (as Zimbabwe has done up to now) or to pursue the more desirable course of fiscal adjustment (as in Morocco or Colombia throughout the 1980s and to the present). These results for domestic debt financing and real interest rates (or financial repression) and those for seigniorage and inflation indicate strong correlations in both cases in developing countries. The increas- ingly unfavorable tradeoffs between these financing sources and tle rates of return on goverm-nent liabilities-leading in the extreme cases to hyperinflation, debt repudiation, or the virtual disappearance of domestic capital markets-imply that there is no alternative to fiscal adjustment for ensuring monetary and financal stability. Private Sector Response to Public Deficits The macroeconomic effects of deficits are determined to a large extent by the direct response of private spending-consumption and investment-to changes in the deficit and in its composition. The way 54 Fiscat Adjustnent and Macroeconomnic Performance: A Synthesis in which governments adjusted their fiscal imbalances during the 1980s (frequently by cutting public investment) was often costly for private sector investment. In the ten sample countries private invest- ment declined sharply from an average of 13 percent of CDP in 1981 to 9 percent in 1986. MeanwhIle, consumption, both public and private, was relatively insulated. Not even the sharp increases in public con- sumption of the 1970s-expansions that had much to do with the subsequent fiscal crises were moderated during the adjustments of the 1980s. To provide some insight into how the private sector responds to fiscal policies, we first identify the channels of transmission between fiscal policies and private spending and then assess their empirical relevance. The empirical inferences are derived from econometric estimations (based on annual time series for each country) for both private consumption and investment, consistent with optimizing behavior under liquidity constraints. Here we summarize orly the qualitative response to variables related to fiscal policy; the chapters on individual countries provide the estimated coeffidents of both fiscal and nonfiscal variables. Private Consumption and Fiscal Policies Fiscal policies affect private consumption and saving through two major channels: disposable income and rate of return (real interest mte). An increase in the deficit brought about by a cut in current taxes boosts private consumption by imcreasing disposable income, accord- ing to the standard Keynesian hypothesis that consumers increase spending when their current income rises. The permanent income hypothesis asserts that only a permanent tax cut significantly affects consumer spending; thus, if the tax cut is temporary, the consump- tion effect wifi be minimal. Both hypotheses are denied by the Ricardian equivalence hypoth- esis, which caims that consumers react the same whether govern- ment spending is financed through debt or taxes because consumers foresee that a tax cut today, paid for by a deficit and borrowing, will lead to a tax increase in the future. In anticipation of that future tax increase, consumers save rather than spend the income from the tax cut So a tax cut that simply substitutes debt finance for tax finance of unchanged government spending would leave consumer spending unchanged-and would lower it as a share of (now higher) disposable income. However, if government consumption is increased, private consumption should decline one-to-one with each dollar of higher permanent government spending. The argument, first skepticaly postulated by Ricardo and affirmed in the recent literature by Barro William Easterly and Klaus Schmidt-Hebbel 55 (1974), rests on-in addition to many secondary assumptions-two main and rather stringent assumptions (Seater 1993): that consumers are concerned with their own future welfare and that of their descen- dants, and that consumers can shift consumption over time by bor- rowing or lending whenever they wish. There is another reason-unrelated to the Ricardian hypothesis- why a tax cut could cause private saving to rise or a government consumption increase could cause private consumption to fall. If there are strict government controls on domestic credit and external capital flows, with government having the first claim on credit, an increase in the deficit (a fall in government saving) reduces the credit available to the private sector, forcng private saving to rise or con- sumption to fall. This effect, which may be difficult to distinguish from the Ricardian hypothesis, may be termed the direct-crowding- out hypothesis. The real interest rate determines how consumers schedule their const 2nption over time, assuming that they have access to credit. The effect of the interest rate on today's consumption levels is ambiguous because of the offsetting substitution, income, and wealth effects. An increase in interest rates causes consumers to substitute consumption tomorrow for consumption today, but it also induces consumers to feel richer and thus to spend more both today and tomorrow-unless this wealth stems significantly from future income streams dis- counted by the (higher) interest rate. Credit controls or borrowing constraints would block the effect of the real interest rate on consumption. Table 111 summarizes the qualitative effects of the abore- mentioned fiscal policy variables on private consumption.2l (For brev- ity, the discussion exdludes any reference to other consumptior determinants included in the estimations, such as the terms of trade, foreign saving, or money.) For most of the countries, both current (or temporary) and long-nm (or permanent) disposable income levels are fournd to be important determinants of private consumption-and often by magnitudes halfway between those implied by the Keyne- sian hypothesis and those implied by the permanent income hypothesis. Does public saving (or the deficit) affect private consumption directly, as implied by the Ricardian hypothesis and the direct- crowding-out hypothesis? Fcr most countries it does not: permanent public saving is not significant in Chile, Mexico, or Pakistan; current public saving or deficits do not affect consumption in Colombia, Coe d'Ivoire, Ghana, or Pakistan. In three cases, however, changes in public saving (or surplus) cause consumption to move in the same direction, which is consistent with either hypothesis. Private con- sumption rose with permanent public surpluses in Argentina and Table 1.11. Qualitative Effects of Fiscal-Policy-Related Variables on Private Consumption Disposable flcorne Pfblic savinjg Public5u'rphlus Counrry Ciirrent Penlmanent Crrenit Pentanelet CtrreDit Pennraseret Real ititerest rate Argentina, 1915-84, 1961-84 (+) - - (+)a Chile, 1960-88 (+) (+) - (0) - - (0) Colr'nbia, 1971-86 (+) (+) (0) - - _ CMte d'lvoire, 1972-87 (+) - - - (0) - Ghana, 1969170 to 1988 t+) (+) - - (0) - (0) Mexico, 1981.1-89.4 (+) (0) (0) - (- () Morocco, 1972-88 - (+) - - - (+) (0) Pakistan, 1963--87 - (+) - (0) (0) - Thailand, 1971-87 (+) - - -- +) Zimbabwe, 1965-88 +) (±) - (+) -- (0) - Not available. Note: The positive and negative signs correspond to statistically significant coefficients; (0) denotes a coefficient not significantly different from zero. Specifications and estimation techniques vary by country. The dependent variable (private consumption) enters in levels for Argentina, Ghana, and Pakistan, in log levels for Morocco and Thailand, in both levels and log levels for Colombia, in the ratto to national income for C6te d'lvoire, and in the ratio to private disposable income for Chile, Mexico, and Zimbabwe. a. The specification does not permit a clear distinction between current and permanent forms. Soarce: Country case studies listed in the references to this chapter. William Eiasterly and Klaus Schmidt-Hebbel 57 Morocco and with permanent public saving in Zimbabwe. Although the coefficients were significant and positive, they were much lower than those for permanent disposable income, implying-contrary to the Ricardian hypothesis-that tax cuts would affect consumption. It also implies that public saving would still have a positive net effect on total saving. These three cases could have supported the Ricardian explanation only if the countries had freely operating financial markets so that consumers could shift their consumption over time in anticipation of future tax increases. In fact, however, Argentina did not liberalize its financial markets until 1977 (late in the sample period), while Morocco and Zimbabwe had institutional arrangements that gave the public sector preferential access to domestic credit. These facts sug- gest that direct crowding-out of private consumption by public defi- cits is the more likely explanation. Similar results for a different sam- ple of developing countries were found by Corbo and Schmidt- Hebbel (1991). The ten case studies provide little evidence that real interest rates have a positive effect on private saving-a result consistent with simi- lar findings for other developing country samples (Giovarnini 1983, 1985; Corbo and Schmidt-Hebbel 1991; Schmidt-Hebbel, Webb, and Corsetti 1992). The real interest rate showed significant effects in three countries. Rising real interest rates depressed private consump- tion in Mexico (signaling the dominance of the intertemporal substi- tution effect) but increased consumption in Colombia and Thailand. The absence of significant results in five other cases suggests either that the substitution, income, and wealth effects tend to cancel each other out or that borrowing constraints prevent consumers from responding to interest rate swings by shifting consumption over time. Borrowing constraints are also behind Haque and Montiel's (1989) rejection of Ricardian equivalence for a set of developing countries. Private Investment and Fiscal Policies Fiscal policies affect private investment through three major chan- nels: public investment, public deficits, and the user cost of capital. Public capital could be a close substitute for private capital, driving down the rate of return on private investment. Public investment in steel plants is an obvious example. However, governments also invest in activities that do not attract private investment but that raise the rate of return of other private projects, such as infrastructure projects. Thus the higher the complementarity of public and private capital, the more likely it is that public investment wil have a net positive effect on private investment. Table 1.12. Qualitative Effects of Fiscal-Policy-Related Variables on Private Investment Purblic capital Ptblic sector Cost of capital Counatry Stock Flow Deficit Colisumption Revenue User cost Real (JJterest rate Argentina, 1915-84 (0) (-) (M) - Chile, 1961-88 - (-)l(0) - - - (-)/(O) - Colombia, 1925-88 (-) - - (0) C6te d'lvoire, 1972-87 - - (-) Ghana, 1967-88 - () - - - (0) Mexico, 1970-89 - (-)I() - - - (-) Morocco, 1972-88 (+) - - (-) Pakistan, 1972173 to 1987188 (+) - - - (-) Thailand, 1971-87. - (+) H) - - Zimbabwe, 1965-88 - (+) - - -(-)- - Not available. Note: The positive and negative signs correspond to statistically significant coefficients; (0) denotes a coefficient not signIfic2ntly different from zero. Specifications and estimation techniques vary by country. The dependent variable is private investment for all countries except Cote d'lvoire and Pakistan; it enters in levels for Argentina, in log levels for Thailand, in the ratio to GDP for Chile, Ghana, Mexico, and Zimbabwe, in the log ratio to GOP for Morocco, and in the level, log level, or ratio to GDP for Colombia. In the case of Pakistan the dependent variable Is the ratio of private capital stock to GDr. Because of data limitations for COte d'lvoire, Ihe dependent variable is the ratio of domestic Investment to national Income. Souirce: Country case studies listed in the references to this chapter. Wr71am Easterly and Ktaus Schmidt-Hebbel 59 If there is repression of domestic interest rates and the public sector is given preferential access to domestic credit, the public deficit could crowd out private investment. When interest rates are not regulated, deficit financing through domestic borrowing tends to raise real inter- est rates, diminishing the profitability of investment by raising the user cost of capital. (The user cost of capital is determined by the real interest rate, the price of investment goods, and investment incentives.) Table 1.2 summarizes the qualitative effects of fiscal policy vari- ables on investment (For brevity, the discussion excludes any refer- ence to other investment determinants induded in the estimations, such as the marginal product value of capital, foreign saving, firm profits, or banking credit to firms.) Consistent with the theoretical ambiguity of the relationship between public capital and private investment, the case studies found sharply different results. For Pakistan each percentage point increase in the ratio of public capital stock to output results in a 2.1 percentage point increase in the ratio of private capital stock to output. A similar relationship is found for Zimbabwe; a higher public capital stock also raises private invest- ment, but the effect is smaller than in Pakistan. By contrast, an increase in public capital stock in Chile and Colombia tends to lower private investment. Some of the country studies used public investment rather than public capital stock, again finding opposite effects in different coun- tries. For Ghana and Mexico increasing public investment reduces private investment (although the effect was weak for Mexico), while for Thailand private investment rises with public investment. For Argentina no significant relation was found. The Morocco study found that public investment contributes to growth, from whlich it can plausibly be inferred that private investment rises with public capital formation. Thus, only three countries provide direct evidence for tlhe wide- spread presumption that public sector investnent is good for private investment. These findings confirm previous studies for developing countries (Blejer and Khan 1984; Khan and Reinhart 1990), with ambiguous results regarding the effect of public on private invest- ment. By way of comparison, Aschauer (1989) finds that higher public capital strongly increases private investment in the United States. It seems reasonable to infer, then, that for countries with a negative relationship (Chile, Colombia, Ghana, and Mexico) or no relationship at all (Argentina), public investment is concentrated in activities which substitute directly for private investment. Public deficits have a negative effect on private investment in COte d'¶voire, where the effect is weak, and in Thailand, where it is strong. For Argentina the study decomposed the deficit into its three main 60 Fiscal Adjusfnent and Macroeconomic Perfonnance: A Synthesis components and found that public investment does not affect private capital formation but that public consumption and public revenue do, in directions consistent with the crowding-out hypothesis. The infer- ence, then, is that deficits tend to crowd out private investment through domestic financial markets in Argentina, C6te d'Ivoire, and Thailand. Although many studies have found that private investment is insensitive to real interest rates, the results for the sample countries show a surprisingly strong negative effect in five of them, with only two (Colombia and Ghana) showing no relationship. The effect of interest rates on private investment is strongest in Morocco and Pakistan, moderately strong in Zimbabwe, and weakest in Chile and Mexico. Public Deficits, Erad e Deficits, and Real Exchange Rates For the 1980s real ecchange rates are dosely correlated with the behavior of fiscal deficits in many developing countries. The major fiscal adjustment in COte d'lvoire in 1982-85 was accompanied by real depreciation; subsequent fiscal backsliding occurred in 1985-88 together with real appreciation. The large reduction in the fiscal defi- cit in Colombia in 1983-88 was accompanied by real depredation. In the same way Chile's real depredation of 1984-88 was contem- poraneous with a fall in the deficit. Ghana's reform program after 1982 included both a deficit reduction and a real depreciation of the official exchange rate (as well as a depreciation of the real black- market exchange rate). Morocco experienced both a deficit reduction and a real depreciation in 1982-85; in Thailand a deficit reduction and a real depreciation occurred together in 1985-88. These associations support the finding of Edwards (1989) that nominal devaluations last as real devaluations only if accompanied by fiscal adjustment. In order to provide more systematic evidence on the linkages between the fiscal defict, the trade deficit, and the real exchange rate, the project case studies tested behavioral relations for the two latter variables on the basis of Rodriguez's two-sector dependent-economy model with optimal capital accumulation (see chapter 2 in this vol- ume). This framework permits the derivation of a two-step relation- ship between the fiscal deficit and the real exchange rate: the fiscal deficit (among other determinants of investment and saving behav- ior) affects the extemal deficit, which then determines the real exchange rate consistent with the dlearing of the domestic goods market. The empirical evidence summarized in the following sections is based on econometric estimations {on annual time-series data for each country) for both the trade surplus and the real exchange rate. William Easterly and Klaus Schmidt-Hebel 61 Only the qualitative response to fiscal variables is reported here; the chapters on individual countries provide quantitative results on both fiscal and nonfiscal variables. Table 1.13 summarizes the sensitivity of the trade surplus to three fiscal variables: the deficit, public consumption, and public invest- ment. For eight countries-Argentina, Chile, Colombia, Cote d'Ivoire, Ghana, Mexico, Thailand, and Zimbabwe-there is signifi- cant evidence that rising external surpluses are correlated with higher public surpluses. A similar relationship-reducing the fiscal deficit by reducing public investment improves the trade balance-was found for Pakistan on the basis of a comprehensive macroecononic model. That fiscal adjustment is a major determinant of external adjustment is also implied by the hypothesis that fiscaI policy is an effective instrument for raising national saving, as the substantial evidence presented in the preceding section shows. Table 1.14 summarizes the sensitivity of the aggregate real exchange Table L13. Qualitative Effects of Fiscal-Policy-Related Variables on the Trade Surplus Public surplus Public expeniture Country Total Primary Operational Consumption Investmet Argentina, 1963- - (+) - - - Chile, 19608 - - (+) Colombia, 1970388 - (+) Cbte d'Ivoire, 1971-81 - (0) Cbte d'lvoire, 1979-89 - (+) - - Ghana, 1970488 - _ _ Mexico, 1970-89 - (+) Morocco, 1974- - - - (-) - Pakistan, 1983184 to 1987188 (-) Thailand, 1972-89 ( - - Zirnbabwe, 19658 - - ( - Not available. Note: The positive and negative signs correspond to statistically significant coeffi- cients;(O) denotes a coefficient not significantly different from zero. Specifications and estimation techniques vary by country. The dependent variable (current account or trade balance) enters as a ratio to GDP for Argentina, Chile, Colombia, C6te d'lvoire, Mexico, and Thailand; in levels for Ghana, Morocco, and Pakistan; and as a log ratio to GDP for Zimbabwe. The coefficient for Ghana is for aggregate private expenditure. The effects for Morocco and Pakistan are not the coefficients for one structural equation but represent the general equilibrium effect of a change in the exogenous variable on the current account surplus (in Morocco) or the trade surplus (in Pakistan). For Morocco the sign reflects the deterioration in the current account as a result of a foreign-financed increase in government consumption. For Pakistan the sign reflects the improvement in the trade surplus based on the impact of deficit reduction through lower public investment. Source: CountTy case studies listed in the references to this chapter. 62 Fiscal Adjustment and Macroeconontic Perjormance: A Synithesis Table 1.14. Qualitative Effects of the Trade Surplus and Fiscal- Policy-Related Variables on the Real Exchange Rate Country Tradic surpus Publicexpenditure Public deficit Argentina, 1964-87 C-) (+) Chile, 196D-88 -) (-) Colombia, 1967-8 () (-) CBe d'Ivoire, 1972-87 (+) Cote d'Ivoire, 1972-89 C-) (0) Ghana, 1970-88 - (1) Mexico, 1970-89 (-) () Morocco, 1974-88 (-) (+) Pakistan, 1983184 to 1987188 (H) Thailand, 197249 C-) Zimbabwe, 1965-88 _(- (+) - Not available. Note: The increase in the real exchange rate equals appreciation. The positive and negative signs correspond to statistically significant coefficients; (0) denotes a coeffi- cient not significantly different from zero. Specifications and estimation techniques vary by country. The dependent variable (real exchange rate) enters as levels for C6te d'Ivoire, Ghana, and Thailand; as levels distinguished between the relative export price and the relative import price for Chile, Mexico. and Zimbabwe; as natural log- arithms of the import price for Argentina; and as natural logarithms of the real exchange rate for Colombia. The effects for Morocco, Pakistan, and Thailand are not the coefficients for one structural equation but represent the general equilibrium effect of a change in the exogenous variable on the corresponding endogenous variable. For Morocco the reported effects combine the simnulation results of a domestic-debt- financed increase in public expenditure and a foreign-financed increase in public expenditure. For Pakistan the effect of an appreciation of the real exchange rate is brought about by a 10 percent reduction of the public defidt through lower public investment, which causes domestic prices to rise with a fixed nomiinal exchange rate. For Thailand the reported effect sumunarizes the simulation results of domestically financed deficits, which cause a trade deficit and a real exchange rate depreciation. Sourc Country case studies listed in the references to this chaptec rate to the trade surplus and to fiscal variables. For eiglht countries (Argentina, Clile, Colombia, Cote d'Ivoire, Mexico, Morocco, Thai- land, and Zimbabwe) higher trade surpluses lead to depredation of the real exchange rate. For Ghana a higher public defict leads directly to appreciation of the real official exchange rate, taking into account the existence of a black market in foreign exchange. The only contrary result was for Pakistan, where deficit reduction through reduced public investrnent leads to appreciation of the real exdhange rate because of the depressing effect of lower public investment on domestic output. These findings, together with those on the positive relation between trade deficits and fiscal deficits, strongly support the central hypothesis of this section: a lower fiscal deficit leads to a lower trade deficit, which in turn leads to a real exchange rate depreciation. William Easterly and Klaus Schmidt-Hebbet 63 The studies also examined Rodriguez's hypothesis (see chapter 2 in this volume) that, for a given trade deficit, an increase in public spending affects the real exchange rate. This effect occurs because an increase in public spending for a given trade deficit implies a corre- sponding decline in private spending. if the public sector has a higher propensity than the private sector to spend on imports rather than on domestic goods, a shift toward more public and less private spending implies an increased demand for imports and a corresponding depre- ciation of the real exchange rate. Tests of this hypothesis show split results for the sample countries. Higher govemment spending leads to an appreciation of the real exchange rate for Argentina, Cote d'Ivoire, Morocco, and Zimbabwe and to a depreciation for Chile, Colombia, and Mexico. These empirical results support the notion that the real exchange rate is sensitive to both policy Lnd extemal variables, including, prominently, the fiscal deficit. The strong contnbution of fiscal adjust- ment to external adjustment and to a corresponding depredation of the real exchange rate, as found in the ten-country sample, is reflected in figure 1.7. The figure confirms that the dominant macro- economic policy trend of the 1980s in these countries was fiscal and external adjustment. However, this average trend of steady improve- ment from 1982 to 1988 was not confined to the sample countries. Other developing countries showed similaz, although less pro- nounced, reductions of the public deficit, and industral countries also cut their deficits in half during the same period. A major conse- quence of fiscal adjustment was sharp reductions in current account deficits, supported by massive real exchange rate depredations. Condlusions and Policy Implications This chapter has summarized empirical evidence on various contro- versial issues that occupy the center stage in the discussion of the macroeconomics of fiscal adjustment or deterioration. The evidence is drawn mostly from a sample of ten case studies that are highly repre- sentative of the structural diversity of developing countries. This fea- ture strengthens the conclusions reached and shows them to be rele- vant for the developing world at large. The use of adequate measures of actual and sustainable public defi- cits was addressed first. Wide public sector coverage (including pub- lic enterprises and the central bank) and exclusion of economically irrelevant categories such as the inflation component of domestic interest payments permnit the derivation of more meaningful deficit measures. These measures can be compared with estimates for sus- tainable deficit levels to evaluate the need for fiscal adjustment. Public budgets are very sensitive to foreign and domestic macro- economic shocks in the short run. But the empirical findings show 64 Fiscal Adjustment and Macroeconomic Performance: A Synthesis Figure 1.7. Fiscal and External Balances and Real Exchange Rate: 1980-88 Averages for Ten Developing Countries Fiscal and extemal balances (percentages of GDP) Real exchange rate a - - - - - --120 -2 110 -4 L¶I!10 -6 - 90 -8- Ba -10 -70 1980 1981 1982 1983 1984 1985 1986 1987 1988 F-1 Fiscal balance Current account balance * Real exchange rate Notc An appreciation of the real exchange rate is shown as an increase. Sources: Chapters 4,5,7,8,9, and 10 and Statistical Appendix in this volume; , Inteenttio,ial Financial Stistics, various issues. that shocks explain only a minor part of the medium-term variation of public deficits. The major factor explaining change is fiscal policy. Policymakers are to be blamed for fiscal cises and praised for fiscal improvements-luck is a very minor determiant of fiscal stance. The first set of issues concerning the macroeconomic consequences of deficts that this chapter addresses relates to the linkages between public deficits, inflation, and real interest rates. Although cross- country correlations between deficits and inflation and between defi- cits and real interest rates were found to be weak at best, the sample countries offer strong evidence that, in the medium term, money financing leads to higher inflation and debt financing leads to higher real interest rates or increased financal repression. As deficit financ- ing mounts, the terms become increasingly unfavorable to the extrac- tion of these unconventional taxes from the private sector. The evidence soundly refutes the Barro-Ricardian proposition that consumers react the same to conventional taxes, unconventional taxes (inflation or financial repression), and debt financing. The William Easterly and Klaus Scizmidt-Hebbel 65 notion that private saving can be mobilized through higher real inter- est rates (brought about by increased debt financing or financial liber- alization) was also rejected. Both findings are in line with the recent empirical evidence on private saving behavior in developing countries. Higher interest rates, however, have a negative effect on pnvate investment. This finding is consistent with investment theory, but it contradicts some of the empirical evidence showing that investment is insensitive to interest rates in developing countries. Public invest- ment was found to crowd out private investment in some counbties and, in others, to crowd it in. This result confirms previous studies showing that the net effect of public investment on private invest- ment depends on the composition of the former-on whether it is a complement to or a substitute for private inr 'estment. Finally, strong and systematic evidence was also found to support the hypothesis that fiscal deficts spiR over into current account defi- cits, leading, in turn, to a real exchange rate appreciation- The main policy implications derived from these findings can be summarized as follows. - Fisical adjustment. Estimations for six countries suggest that sus- tainable piimary surpluses vary between 2 and -2 percent of GcP (equivalent to nominal deficits of between 2 and 6 percent of GDP). These figures, which depend on the combination of macroeconomic conditions and outstanding public liabilities, are a far cry from actual deficits in countries such as C6te dlvoire and Pakistan. Successful fiscal adjustment can be seen as proceeding in two stages. First, deficits are reduced to sustainable levels, consistent with stable debt-output ratios and normalized financial markets, as, for instance, in Morocco. A second phase of deeper fiscal adjustment supporting a strong private sector response-as in Chile or Morocco- involves reaching nominal public surpluses or only slight deficits, thus allowing the public sector to reduce its indebtedness in relation to the domestic private or external sectors. The ten-country sample suggests that sustainable adjustment typ- ically requires action on four fronts: reducing an overblown govern- ment bureaucracy; cutting transfers and subsidies (other than effi- cient, targeted social programs); enacting tax legislation for increased, broadly based direct and indirect taxation; and reforming or privatiz- ing public enterprises and commodity marketing boards. * Fiscal deficits and inflation. To the extent that deficits are financed by money creation, the relationship between fiscal deficits and inflation is indisputable. However, the tradeoff between additional inflation and revenue is increasingly unfavorable to the latter, as documented by the Laffer-curve behavior of money demands. 66 Fiscal Adjustment and Macroeconomic Perfomance A Synthesis Governments desperately lacking other income sources and with short planning horizons often start bursts of accelerated money print- ing and inflation that yield, for a brief period, abnormally high infla- tion tax revenue. This revenue falls as soon as people reduce tieir money holdings in response to higher inflation; that reaction leads to growing macroeconomiic instability and relative price variability. Since the inflation tax (as well as financial repression, discussed below) is a tax, there is no reason to expect conventional fiscal adjust- ment to be any more contractionary than adjustment through infla- tion (or financial repression). * Fiscal deficits, real interest rates, and financial repression. If domestic financial markets are not repressed but external financing is not easily available, higher deficits financed by domestic debt raise domestic real interest rates. When domestic financial markets are integrated with world capital markets, higher domestic public borrowing leads to external capital flows and higher foreign debt, without much affecting domestic real interest rates. The story is different when the government represses financial markets through controls on domes- tic interest rates, compulsory public debt placements, and controls on external capital flows. If the nominal interest rate is fixed, higher fiscal deficits lead to repressed (even negative) real interest rates, implying high taxes on financial intermediation. But the poor eco- noniic performance that follows from strong financial repression- depressed private credit and the attendant collapse of private investment-hardly recommends this unconventional form of taxation. * Budget deficts and prrivate consumption. Rejection of the notion that consumers are indifferent between taxes or debt finance carries the policy implication that increasing public saving-or reducing public deficits-is the most effective contribution fiscal policy can make to raising national saving. However, increasing real interest rates through domestic debt financing or financial liberalization will not raise private saving. * Budget structure, deficits, and private investment Real interest rates and private sector credit do significantly affect private investment. So whether there is financial repression or not, increasing public deficts reduces private investment. The composition of public spending mat- ters as well, since more public investment depresses private invest- ment in some cases-typically, when large public enterprises compete with private fimns and enjoy preferential access to domestic financial resources. The implication is that the prospects for higher private investment and growth are improved by three policy measures: restructuring and privatizating public firms and marketing boards; concentrating public investment on public and social infrastructure; and deregulating domestic financial markets-including removal of William Easterly and Klaus Schzmdt-Hebbel 67 credit ceilings, compulsory credit allocation, preferential access of the government to credit, and interest controls. * Fiscal deficits, trade deficits, and real exchange rates. The evidence of the strong relationship between public and external deficits comple- ments the policy implication derived from the finding that private saving does not offset changes in public saving: fiscal adjustment is very effective in boosting national saving and therefore in increasing the trade surplus as well. Exchange rates are driven by fimdamentals and not the other way around. This should serve as an antidote to the mistaken notion of many policymakers that nominal devaluation alone can restore macroeconomic balance. As Khan and Lizondo (1987) have hypothesized, real exchange rates are also affected by whether the government spends more on tradables than on nontrad- ables. Policymakers should pay attention to the composition of gov- emient spending when deciding on an accommodatirg exchange rate policy. * Fiscal deficits and growth. High deficts are strongly correlated with low growth. Inflation raises uncertainty and distorts relative prices, hurting private investment and resource allocation. The conventional notion that public investment is good for private investment and growth received niixed support. Countries that were forced to shift from extemal to internal financing of deficits-often because of a debt crisis induced by fiscal mnismanagement-showed particularly poor investment and growth performance in the 1980s. Growth itself makes deficits less harmful: countries such as Pakistan and Thailand could sustain larger deficts because of strong growth, while eco- nomic collapse worsened the macroeconomic effects of deficits in Argentina, C6te d'Ivoire, and Mexico. The virtuous circle between growth and good fiscal management is one of the strongest argu- ments for a policy of low and stable fiscal deficts. Appendix 1.1. Measurement of Deficits and Evaluation of Public Sector Solvency Altemative measures for public sector composition stretch from the central government to the consolidated nonfinancial public sector (which consolidates the central govermment with local government, social security, and nonfinancial public enterprises) and to the consol- idated total pubic sector (adding to the first consolidation the central bank and, possibly, the public commercial banks). Although deficit measures based on the widest public sector coverage are the most accurate measure of fiscal stance and public sector resource transfers, they are often not readily available and are not free of controversy. (See the references in the text on the problems of measuring quasi- fiscal deficits.) 68 Fiscal Adjustment and Macroeconomic Performance: A Synthesis An alternative to cash-based deficits is measurement of deficits on an accruals (or payment-order) basis, which reflects income and spending actions measured at the time they take place, even if they do not immediately involve cash flows. Accumulation of arrears on interest, wage, or goods expenditure would cause accruals-basis defi- cits to be larger than cash-basis deficits. Box 1.2, above, Illustrates the differences between alternative deficit measures for Morocco. A popular alternative to the nominal cash deficit is the operational deficit, obtained bv deducting from the nominal cash deficit the part of nominal interes payments on public debt attributable to inflation, which merely compensates the debt holder for the erosion of the real value of public debt caused by inflation and which, under asset- equilibrium conditions, is reinvested by the debt holder. This correc- tion is particularly important in high-infiation, high-domestic-debt countries. The primary deficit deducts net interest payments from the nominal cash deficit. Use of accruals-based deficits permits a move away from conven- tional deficits or intratemporal budget constraints and toward deficit measures consistent with public sector net worth or public inter- temporal budget constraints. While the latter would constitute the economically most meaningful measures of fiscal stance and public sector solvency, they are, unfortunately, not observable. Three approaches have been adopted in dealing with this issue. The first is the accounting approach to public sector solvency developed by Bui- ter (1983, 1985, 1990) and van Wijnbergen (1989), which derives esti- mates of sustainable deficit levels as those that can be financed with- out raising debt levels (in relation to GDP) under feasible rates of growth, real interest, and inflation. This methodology has been applied to many countries (for instance, by van Wijnbergen and others 1992 to Turkey and by de Melo 1990 to Morocco), and it is used in deriving the sustainable deficit measures in the sample countries, as reported in the section on sustainable public deficits. A less stringent requirement than the constancy of debt-output ratios imposed by the accounting approach is to test for the no-Ponzi- game condition on public debt followed by the neodassical solvency approach. This methodology, developed by Hamilton and Flavin (1988), Grilli (1989), Wilcox (1989), and Buiter and Patel (1990), checks for public sector solvency by comparing the rate of growth of the public debt (in relation to GDP) with the real interest rate. If the debt ratio systematically grows faster than the real interest rate, the public sector is considered insolvent. Among the applications of the meth- odology are the recent studies by Buiter and Patel (1990) for India and by Werner (1991) for Mexico. The third method diverges from the first two methods by focusing on prices instead of quantities, testing for discounts on public debt paper. William Easterly and Klaus Schmidt-Hebbel 69 Appendix 1.2. Sustainable Deficits This appendix derives the equation for the sustainable primary deficit calculations reported in the first section of this chapter. It is based on the accounting approach to public sector solvency developed by Bui- ter (1983, 1985, 1990) and van Wijnbergen (1989). We start with the budget constraint of the consolidated total public sector, which consists of the nonfinancial and financial subsectors, the latter including the central bank. The budget constraint equates the above-the-line total nominal deficit (the sum of the primary deficit and total net interest payments) to below-the-line financing sources (the change in monetary and nornmonetary public debt holdings). (11) P+ i pD = + D + ED* py py py Py Py Py where PD is the consolidated total public sector primary deficit, P is the GDP deflator, y is real GDP, i is the domestic nominal interest rate, D is the stock of domestic public debt, E is the nominal exchange rate (domestic currency units per unit of foreign currency), it is the foreign nominal interest rate, D* is the foreign public debt stock (in current- price foreign currency units), and M is the base money stock. All vari- ables are in current-price domestic currency units unless otherwise noted. Simple manipulation of equation 1.1 permits derivation of the ratio of the primary deficit to cDL' as: (1.2) pd =m+ ( n)m +a(n-r)d +a*+(n-r* -)d* where the lower-case variables pd, d, d*, and m are defined as the ratios of PD, D, E, DI, and M, respectively, to GDP at current prices; sr is the domestic rate of inflation; n is real GDP growth; r is the domestic real interest rate, r* is the foreign real interest rate; and e is the rate of real exchange rate depreciation. Equation 1.2 shows that the primary deficit of the consolidated public sector, as a share of GDP, is constrained to not exceed the sum of six financing sources: revenue from the sum of the inflation tax on the monetary base and growth-induced increase in money demand; the excess of domestic growth over the relevant real interest cost of domestic and foreign debt; and increasing demands for monetary and nonmonetary debt. Primary deficits are sustainable if they do not entail ever-increasing shares of debt and money to income. In the absence of explicit demands for public liabilities, the accounting approach to public solvency defines sustainability in the more restric- tive sense of constant ratios of debt to output and of money to output, consistent with steady-state (constant) inflation and interest rates. Therefore the country applications summarized in the first section of the chapter calculate sustainable primary deficits as determined by 70 Fiscal Adjustment and Macroeconomic Perfonnantce: A Synthesis equation 1.2, after imposing the steady-state condition of constz liability-to-income ratios. In most cases the calculations assume that 1988-90 liability-to-output ratios are the relevant steady-state values. Country applications differ in public sector coverage (central, general, nonfinancial, or total public sector), and equation 1.2 is modified accordingly. Appendix 1.3. Decomposition of Seigniorage The decomposition is based on the following equation for seig- niorage, T: (1.3) T,= Pt Mt_l + Mt-Mt- where Pt is the price level at time t, M, is the real money supply at time t, and r, is the inflation rate at time t. The steady-state value of seigniorage is given by: t.l.4) 1 +_M + g Il+i1l+g i+ where a bar denotes an average level of a variable and it is assumed that real money grows in proportion to output, with output growth given as g and the trend value of real money as M. The first term gives the inflation tax component of seigniorage, while the second gives the seigniorage that accrues from an increase in real money balances. The deviation of seigniorage from the average can then be given as follows: .13) Tt-T= (.1 + Wr 1 + -)(Mtl- + g + (Mt Mt-, - g M) A YT it 1 T+ 7-r) (Mt l1 +g) The first term is the above-average seigniorage attributable to the above-average inflation tax rate. The second term gives the above- average revenue attnbutable to the above-average real money base. The third term gives the real change in the money base minus the amount that would take place as money grows with output. The last t.erm is the covariance of inflation and money. Notes This chapter draws on aII the other chapters in the volume. The authors benefited from discussions with and comments by Jorge Baldrich, Nancy William Easterly and Klaus Schmidt-Hebbel 71 Birdsall, Mario Blejer, Vittoria Corbo, Giancarlo Corsetti, Shantayanan Devarajan, Ricardo Ffrench-Davis, Nicolas Eyzaguirre, Stanley Fischer, Michael Gavin, Ravi Kanbur, Johannes Linn, Carlos Rodriguez, Vito Tanzi, and Martin Werner, as well as the participants at the World Bank Conference on the Macroeconomics of Public Sector Deficits (Washington, D.C.), the Tenth Latin American Meeting of the Econometric Society (Punta del Este), and seminars at Columbia University, the Ministry of Finance of China (Beij- ing), the Ministry of Finance of Costa Rica (San Jose), CEMA-Universidad de San Andres (Buenos Aires), and the Central Bank of Chile (Santiago). They are also grateful for the comments and assistance of Paolo Mauro, for research assistance by Maria Cristina Alnero-Siochi, Piyabha Kongsamut, and Raimundo Soto, and for interaction with the other chapter authors. 1. The chapter presents in more detail the p oject's findings and implica- tions discussed in Easterly and Schmidt-Hebbel (1993a, 1993b). 2. The most complete study to date on the rmeasurement of fiscal deficits is Blejer and Cheasty (1991). Other references on alternative deficit measures include Tanzi (1985); Eisner (1986); Blejer and Chu (1988); ICotlikoff (1988); Fischer and Easterly (1990); and Buiter (1990). MF (1986) and United Nations (1968) discuss cash and accrual deficits in more detail. Robinson and Steia (1988) and Teijeiro (1989) survey issues concerning quasi-fiscal deficits. 3. The next section presents measures of sustainable prmary deficits for sector coverages that range-because of varying degrees of data availability- from the central govemment to the total consolidated public sector. The questions addressed in subsequent sections require the use of cash-based operational (or nominal) deficit measures for the widest available public sector coverage. The discussion of correlations of deficits with other economic variables presents separate 1978-89 data for nonminal consolidated nonfinan- cial and central bank quasi-fiscal deficits, while the fifty-nine-country correla- tions rely on nominal consolidated ptblic sector or general government bal- ances (depending on data availability), as reported in table A.1 of the Statisti- cal Appendix in this volume. The section on causes and remedies is based on measures of the consolidated nonfinancial public sector deficit, and the subsequent section is based on consolidated total (nonfinancial plus quasi- fiscal) public sector deficits-the nominal or operational measures are indif- ferent here. Finally, the last two sections use operational consolidated nonfi- nancial public deficits because of the lack of long time-series for quasi-fiscal deficits. 4. In two cases (Chile and Zimbabwe) upper and lower bounds, consistent with poasible deviations of the relevant macroeconomic variables from base- case levels, are added to the midpoint estimates. Here, as well as in tihe other four cases, the relevant macroeconomic variables used are those that deter- mine the primary deficit: the rates of output growth, inflation, domestic and foreign real interest, and real exchange rate devaluation; see equation 1.2 in appendix 1.2. 5. An interesting short-run counterpart to this result is the suggestion of Giavazzi and Pagano (1990) and Blanchard (1990) that fiscal austerity can be expansionary. 6. See, for example, Balassa (1988); Reisen and van Trotsenburg (1988); Bartoli (1989); Sachs (1989); and Rodriguez (chapter 2 in this volume). Note, 72 Fiscal Adjustment and Macroeconomic Performance: A Synthesis however, that the link breaks down if the Ricardian hypothesis of offsetting private saving holds (Frenkel and Razin 1987; Leiderman and Blejer 1988). 7. In broad terms, countries face four types of foreign shocks: changes in the price and interest conditions of their foreign trade and their credit flows, and changes in quantity constraints affecting their foreign trade and their credit flows. While quantity constraints are rather uncommon in foreign trade (abstracting from countries affected by global embargoes or partial trade restrictions on certain goods), massive changes in borrowing constraints are a stylized fact in credit markets. The aftermath of the 1982 debt crisis implied, in fact, a massive change in regime in the form of foreign resource constraints suddenly faced by most developing debtor economies. Although borrowing constraints constitute a strong foreign shock affecting below-the-line financ- ing sources, we focus only on changes in foreign terms of trade and interest rates, which impinge directly on above-the-line deficits. S. The channels mentioned here add bracket creep and transfer effects to the five-item list of Dornbusch, Sturzenegger, and Wolf (1990). 9. This is also called the Keynes-Olivera-Tanzi effect; see Olivera (1967) and Tanzi (1977). Sometimes the Keynes-Olivera-Tanzi effect is used more restric- tively to denote the tax erosion effect mentioned below. 10. The calculation is based on a reduced-form equation estimated for the CNFPS deficit in Thailand and on 1988 data for the deficit, inflation, and GDP. 1i. The average relative contribution of each group of deficit, leterminants is calculated on the basis of the equation presented in the note to table 1.2; hence the equation is now used separately for extemal, domestic macro- economic, and fiscal policy variables. However, in order to present the rela- tive contribution of each group of variables, di is defined here as the explained change in the deficit, not the actual change as in table 1.2 Hence the average relative contribution of external variables to actual deficits in the second column of table 1.2 differs from the average relative contribution of external variables to explained deficits in figure 12. The average absolute deviations between actual and explained deficit changes in percentage points of GDP are 0.9 for Chile (1974-88), 2.0 for Ghana (1972173-1988), and 1.4 for Zimbabwe (1981182-1988189). 12. Seigniorage is defined here as the ratio to GDP of the change in high- powered money during the year, averaged over 197049. The number of observations is forty-nine countries, including Argentina; the quadratic term is significant whether or not Argentina is included. See tables A.3 and A.4 of the Statistical Appendix for time-series data on seigniorage and inflation rates for forty-nine QEECO members and developing countries. 13. Barro (1990) also suggests that the maximum of the Laffer curve is at inflation rates around 100 percent. Edwards and Tabeflini (1990) present sug- gestive evidence for seigniorage Laffer curves in a number of developing countries. 14. Dornbusch, Sturzenegger, and Wolf (1990) describe the progressive substitution of interest-bearing assets for money in high-inflation episodes. 15. Similar magnitudes were found in the study by Fischer (1982). 16. The coefficient of variation is calculated over 1970-89 for a reduced sample of twenty-six countries with data over that period (to standardize the number of observations, which affects the variance). The coefficient of varia- tion is the standard d-viation divided by the mean. William Easterly and Klaus Schmidt-Hebbed 73 17. There were actually twenty-one such observations, but three were found to involve changes in measurement of the money base and were discarded. 18. This is not a general pattern, however; of the nine cases in which seigniorage is explained largely by the change in the real money base, four had rising inflation the following year, two had essentially unchanged infla- tion, and three had falling inflation the next year. Dombusch, Sturzenegger, and Wolf (1990) note the rise in real money balances in the early stages of hyperinflation. Our story is a different one: exogenous or policy-induced rises in real money balances allow large temporary seigniorage without hyperinflation. 19. Estimates by Giovannini and de Melo (1993) find much higher revenue from financial repression for Mexico, Morocco, Pakistan, and Zimbabwe than do the other studies because they calculate the tax rate as the ex-post differ- ence between domestic and foreign interest rates, including devaluation. These countries were experiencing steady real devaluations in the early 1980s, which tends to raise the estimate of the tax rate when this method is used. 20. Chamley and Honohan (1990), Easterly (1989), and Giovannini and de Melo (1993) estimate financial repression revenue and discuss the costs of financial repression. Chapter 5 presents evidence from cross-section regres- sions that financial repression has a negative effect on long-run growth. Dombusch and Reynoso (1989), however, argue that financial repression is costly only under very high inflation. 21. The consumption specification of the case studies and table 1.11 include (a) disposable income (gross income, including domestic debt interest pay- ments, less tax revenue) and (b) puhlic saving (tax and nontax revenue less current government expenditure, including domestic debt interest payments) or the public surplus (total government revenue less total governnent expen- diture), as consumption determinants. This follows Corbo and Schmidt- Hebbel (1991) in distinguishing between 'he Keynesian and permanent- income hypotheses (according to which only disposable income matters) and the Ricardian and direct-crowding-out hypotheses (according to which only gross income net of government consumption matters, so that disposable income and public saving should have a high and positive common coeffi- cient in the above-mentioned specification)- References Country Case Studies Alberro-Semerena, Jose Alberto. 1991. "The Macroeconomics of the Public Sector Deficit in Mexico during the 1980s." World Bank, Policy Research Department, Washington, D.C. Chanfley, Christophe, and Hafez Ghanem. 1991. "Fiscal Policy with Fixed Nominal Exchange Rates: Cote d'lvoire.' World Bank Policy Research Working Paper 658. Revised as chapter 6 in this volume. Easterly, WilliamL 1991. "The Macroeconomics of the Public Sector Deficit: The Case of Colombia." World Bank Policy Research Working Paper 626. Revised as chapter 5 in this volume. 74 Fiscal Adjustment and Macroeconomic Performance: A Synthesis Faini, Riccardo. 1991. "The Macroeconomics of the Public Sector Deficit: The Case of Morocco." World Bank Policy Research Working Paper 631. Revised as chapter 8 in this volume. Haque, Nadeem Ul, and Peter J. Montiel. 1991. "The Macroeconomics of Public Sector Deficits: The Case of Pakistan." World Bank Policy Research Working Paper 673. Revised as chapter 9 in this volume. Islam, Roumeen, and Deborah Wetzel. 1991. "The Macroeconomics of Public Sector Deficits: The Case of Ghana." World Bank Policy Research Working Paper 672. Revised as chapter 7 in th-is volume. Marshall, Jorge, and KIaus Schmidt-Hebbel. 1991. "Macroeconomics of Pub- lic Sector Deficits: The Chile Case Study." World Bank Policy Research Working Paper 6%. Revised as chapter 4 in this volume. Morandd, Felipe, and Klaus Schrnidt-Hebbel. 1991. "Macroeconomics of Public Sector Deficits: The Case of Zimbabwe." World Bank Policy Research Working Paper 688. Revised as chapter 10 in this volume. Ramangkura, Virabongse, and Bhanupongse Nidhiprabha. 1991. "The Mac- roeconomics of the Public Sector Deficit: The Case of Thailand." World Bank Policy Research Working Paper 633. World Bank, Policy Research Department, Washington, D.C. Rodriguez, Carlos A. 1991. "The Macroeconomics of the Public Sector Defi- cit: The Case of Argentina." World Bank Policy Research Working Paper 632. Revised as chapter 3 in this volume. Genera References Aschauer, David. i989. "Is Public Expenditure Productive?" Jounal of Mone- tary Economics 23 (March): 177-200. Balassa, Bela. 1988. "Public Finance and Economic Development." Policy Research Working Paper 31. World Bank, Office of the Vice President, Development Economics, Washington, D.C. Barro, Robert F. 1974. "Are Government Bonds Net Wealth?" Journal of Politi- cal Economy 81 (December): 1095-1117. -~. 1990. Macroeconomics. New York: Wiley. Barro, Robert F., and Xavier Sala-i-Martin. 1990. "World Real Interest Rates." In Stanley Fischer, ed., NBER Macroeconomics Annual. Cambridge, Mass.: Massachusetts Institute of Technology Press. Bartoli, G. 1989. "Fiscal Expansion and External Current Account Imbal- ances." In Mario I. Blejer and Ke-young Chu, eds. Fiscal Policy, Stabilization and Growtuh in Developing Countries. LIternational Monetary Fund: Washing- ton, D.C. Bemheim, B. Douglas. 1987. "Ricardian Equivalence: An Evaluation of The- ory and Evidence." In NBER Macroeconomics Annual 1987. Cambridge, Mass.: Massachusetts Institute of Technology Press. Blanchard, Olivier. 1990. "Comment on 'Can Severe Ficl Contractions be Expansionary?"' In Starney Fischer, ed., NBER Macroeconomics Annual. Cambridge, Mass.: Massachusetts institute of Technology Press. William Easterly and Klaus Schimidt-Hebbel 75 Blejer, Mario 1., and Adrienne Cheasty. 1991. aThe Measurement of Fiscal Deficits: Analytical and Methodological Issues." foumal of Economic Litera- ture 29(4) (December): 1644-78. Blejer, Mario I., and Ke-young Chu, eds. 1988. Measurement of Fiscal Impact: Methodological issues. IMF Occasional Paper 59. Washington, D.C.: Interna- tional Monetary Fund. Blejer, Mario I., and Mohsin S. Khan. 1984. "Government Policy and Private Investment in Developing Countries." International Monetary Fund Staff Papers 31 (2): 379403. Blejer, Mario I., and Nissa-n Liviatan. 1987. "Fighting Hyperinflation: Stabi- lization Strategies in Argentina and Israel, 198546." Interational Monetary Fund Staff Papers 34 (September): 409-38. Buiter, Wfllem H. 1983. ""Measurement of the Public Sector Deficit and Its Implications for Policy Evaluation and Design." Interational Monetary Fund Staff Papers 30 (June): 306-49. -. 1985. "A Guide to Public Sector Debt and Deficits." Economic Policy (November): 13-79. -. 1990. "The Arithmetic of Solvency." In Willem H. Buiter, Principles of Budgetary and Financial Policy. Cambridge, Mass.: Massachusetts Institute of Technology Press. Buiter, Willem H., and U. Patel. 1990. "Debt, Deficits and Inflation: An Application to the Public Finances of India." NBER Working Paper 3287. National Bureau of Economic Research, Cambridge, Mass. Cagan, Phillip. 1956. "The Monetary Dynamics of Hyperinflation." In Milton Friedman, ed., Studies in the Quantity Theory of Money. Chicago, in.: Univer- sity of Chicago Press. Chamley, Christophe, and Patrick Honohan. 1990. "Taxation of Financial Intermediation." Policy Research Working Paper 421. World Bank, Coun- try Economics Department, Washington, D.C. Corbo, Vittorio, and Klaus Schmidt-Hebbel. 1991. "Public Polices and Sav- ing in Developing Countries." Jounal of Deveopment Economics 36(1): 89- 116. de Haan, J., and D. Zelhorst. 1990. "The Impact of Government Deficits on Money Growth in Developing Countries." Journal of lInternational Money and Finance 9: 455-69. de Melo, Martha. 1990. "Fiscal Aspects of External Debt: A Case Study of Morocco." World Bank, Policy Research Department, Washington, D.C. Dollar, David. 1990. "Outward-Oriented Developing Economies Really Do Grow More Rapidly: Evidence from 95 LDcs, 1976485." World Bank, Coun- try Economics Department, Washington, D.C. Dornbusch, Rudiger. 1985. "Overborrowing: Three Case Studies." In Gor- don W. Smith and John T. Cuddington, eds., International Debt and the Developing Countries. A World Bank Symposium. Washington, D.C. Dornbusch, Rudiger, and Stanley Fischer. 1991. "Moderate Inflation." Policy Research Working Paper 807. World Bank, Office of the Vice President, Developmental Economics, Washington, D.C. Dornbusch, Rudiger, and Alejandro Reynoso. 1989. "Financial Factors in Econonmic Development." NBER Working Paper 2S89. National Bureau of Economic Research, Cambridge, Mass. 76 Fiscal Adjustment and Macroeconomic Performance: A Synthesis Dombusch, Rudiger, F. Sturzenegger, and H. Wolf. 1990. "Extreme Inflation: Dynamics and Stabilization." Brookings Papers on EconomicActivity 2: 1-64. Easterly, William. 1989. "FLscal Adjustment and Deficit Fmancing during the Debt Crisis." In Ishrat Husain and Ishac Diwan, eds. Dealing with the Debt Crisis. A World Bank Symposium. Washington, D.C.: World Bank. Easterly, Villiam, and Kaus Schmnidt-Hebbel. 1993a. "Fiscal Adjustment and Macroeconomic Performance." Outreach 10 (May). World Ban&, Policy Research Department, World Bank. Easterly, William, and Klaus Schmidt-Hebbel. 1993b. "Fiscal Deficits and Macroeconomic Performance in Developing Countries." World Bank Research Observer 8 (2) (July): 211-37. Easterly, William, Paulo Mauro, and Klaus Schmidt-Hebbel. Forthcoming. "Money Demand and Seignorage-Maximizing Inflation." Jounal of Money, Credit, and Banking. Easterly, Wllliam, Carlos A. Rodriguez, and Klaus Schmidt-Hebbel. 1989. "Research Proposal: The Macroeconomics of the Public Sector Deficit." World Bank, Country Economics Department, Washington, D.C. Easterly, William, Robert King, Ross Levine, and Sergio Rebelo. 1992. How Do National Policies Affect Long-Run Growth? A Resarch Agenda. World Bank Discussion Paper 16t Washington, D.C. Edwards, Sebastian. 1989. Real Exchange Rates, Devaluation and Adjustment. Cambridge, Mass.: Massachusetts Institute of Technology Press. Edwards. Sebastian, and Guido Tabelln 1990. "Explaining Fiscal Policies and Inflation in Developing Countries." NsER Working Paper 3493. National Bureau of Economic Research, Cambridge, Mass. Eisner, Robert. 1986. How Real Is the Federml Deficit? New York: Free Press. Fscher, Stanley. 1982. "Seigniorage and the Case for a National Money." Joumal of Political Economy 90 (2) (April): 295-313. Reprinted in Stanley Fscher, Indexing, Inflatior. and Ecanonoic Policy (Cambridge, Mass.: Massa- chusetts Institute of Technology Press, 1987). -. 1991. "Growth, Macroeconomics and DevelopmenL"Irn- Stanley Fischer, ed., NBER Macroeconomics Annual. Cambridge, Mass.: Massa- chusetts Institute of Technology Press. Fischer Stanley, and William Easterly. 1990. "The Economics of the Govem- ment Budget Constraint." World Bank Research Observer5 (2): 127-42. Frerniel, Jacob A., and Assaf Razin. 1987. Fiscd Policies and the World Economy. Cambridge, Mass.: Massachusetts Institute of Technology Press. GeTl, Alan, and associates. 1988. Oil Windfals: Blessing or Curse? New York: Oxford University Press. Giavazzi, Francesco, and Marco Pagano. 1990. "Can Severe Fiscal Contrac- tions Be Expansionary?" In Stanley Fischer, ed., NBER Macroeconomics Annual. Cambridge, Mass.: Massachusetts Institute of Technology Press. Giovannini, Alberto. 1983. "The Interest-Elasticity of Savings in Developing Countries." World Development 11(7): 601-07. -. 1985. "Saving and the Real Interest Rate in LDCS." Journal of Develop- ment Economics 18 (August): 197-217. William Easterly and Klaus Sclhmidt-Hebbel 77 Giovannini, Alberto, and Martha de Melo. 1993. "Govemment Revenue from Financial Repression." American Economic Review 83(4): 953-63. Grilli, Vittorio. 1989. "Seigniorage in Europe." In Marcello De Cecco and Alberto Giovannini, eds., A Europeatz Central Bank? Perspectives on Monetary Unification after Ten Years of tie EMS. Cambridge, U.K.: Cambridge Univer- sity Press. Hamilton, J., and Marjorie Flavin. 1988. "On the Limitations of Govemment Borrowing: A Framework for Empirical Testing." American Economic Review 76 (September): 808-19. Haque, Nadeem Ul, and PeterJ. Montiel. 1989. "Consumption in Developing Countries: Tests for Liquidity Constraints and Finite Horizons." Review of Economics and Statistics 71(3): 408-15. Hayashi, Fuio. 1985. "Tests for Liquidity Constraints: A Critical Survey." NsER Working Paper 1720. National Bureau of Economic Research, Cam- bridge, Mass. 'F (International Monetary Fund). 1986. A Manual on Governtment Finance Statistics. Washington, D.C. - Various years. International Financil Statistics. Washington, D.C. Ihan, Mohsin, and Samuel Lizondo. 1987. "Devaluation, Fiscal Deficits and the Exchange Rate." World Bank Economic Review 1(2): 357-74. Khan. M. S., and C. M. Reinhart. 1990. "Private Investment and Economic Growth in Developing Countries.'t Worfd Development 18 (1): 19-27. Kiguel, Miguei, and Nissan Uviatan. 1988. 'Inflationary Rigidities and Orthodox Stabilization Policies: Lessons from Latin America." World Bank Economic Reiew 2 (3): 273-98A IKotlikoff, Larry. 1988. "The Deficit Is Not a Well Defined Measure of Fiscal Policy." Scence 241 (August): 79i-95. Leiderman, Leonardo, and Mario I. Blejer. 1988. "Modeling and Testing Ricardian Equivalence: A Survey." Intemational Monetary Fund Staff Papers 35 (March): 1-35. Marshall, Jorge, and Klaus Schmidt-Hebbel. 1989. "Economic and Policy Determinants of Public Sector Deficits." Policy Research Working Paper 321. World Bank, Country Economics Department, Washington, D.C. oEco (Organization for Economic Cooperation and Development). Various issues. OECD Economic Outlook. Paris. Olivera, J 1967. "Money, Prices and Inflation Lags: A Note on the Dynamics of Inflation." Banca Nazionale dd Lworo Quarterly Review 20: 258-67. Poterba, James. 1988. "Are Consuiners Forward-Looking? Evidence from Fiscal Experiments." American Economic Reiew 48 (2) (May): 413-81. Rama, Martin. 1993. "Empirical Investment Equations for Developing Coun- tries." In Luis Servcn and Andrds Solimano, eds., Striving for Grwwth after Adjustment: The Role of Capital Formation. Washigton, D.C.: World Bank. Reisen, Helmut, and Axel van Trotsenburg. 1988. "Developing Country Debt: The Budgetary and Transfer Problem." Development Centre Study. Organization for Economic Cooperation and Development, Paris. Robinson, David J., and Peter Stella. 1988. "Amalgamating Central Bankand Fiscal Deficits." In Mario I. Blejer and Ke-young Chu, eds., Measurement of 78 Fiscl Adjustment and Macroeconornic Performance: A Synthesis Fiscal Impact: Methodological Issues. 1Mw Occasional Paper 59. Washington, D.C.: International Monetary Fund. Sachs, Jeffrey, ed. 1989. Developing Country Debt and the World Economy. Chi- cago, D1.: University of Chicago Press. Sargent, Thomas J., and Neil Wallace. 1985. "Some Unpleasant Monetarist Arithmetic." Fedeml Reserve Bank of Minneapolis Quarterly Review 9 (Wmter): 15-31. Schmidt-Hebbel, Klaus, Steven B. Webb, and Giancarlo Corsetti. 1992 "Household Saving in Developing Countries: First Cross-Country Evi- dence." World Bank Economic Review 6(3): 529-47. Seater, JohnJ. 1993. "Ricardian Equivalence." Journal of Economic literature21 (1): 142-90. Servin, Luis, and Solimano, Andris. 1993. "Private Investment and Macro- economic Adjustment: A Survey." In Luis Servmn and Andres SolinUano, eds., Striving for Growth after Adjustment: The Role af Capital Fonat ion. Washington, D.C.: World Bankl Summers, Robert, and Alan Heston. 1988. "A New Set of International Com- parisons of Real Product and Price Levels: Estimates for 130 Countries." Review of Incorne and Wealth 34: 1-25. Tanzi, Vito. 1977. "Inflation, Lags in Collection, and the Real Value of Tax Revenue." International Monetary Fund Staff Papers 24: 154-67. - . 1985. "Fiscal Management and External DebtProblems."' In Hassanali Mehran, ed., Extcrnal Debt Management. Washington D.C.: International Monetary Fund. Tanzi, Vito, Mario L Blejer, and Mario 0. Teijeiro. 1987. "Inflation and the Measurement of Fiscal Deficits." Inteational Monet Fund Staff Papas 34 (December): 711-38. Teieiro, Mario 0. 1989. "Central Bank Losses: Origins, Conceptual Issues, and Measurement Problems." Policy Research Working Paper 293. World Bank, Country Economics Department, Washington, D.C United Nations. 1968. A System of National Accounts. New York: United Nations. van Wiinbergen, Sweder. 1989. "Extemal Debt, Inflation and the Public Sec- tor. Towards Fiscal Policy for Sustainable Growth." World Bank Economic Review 3 (3): 297-320. van Wijnbergen, Sweder, P 'a Anand, Ajay Chhibber, and Roberto Rocha. 1992. External Debt, Fiscal e-olicy, and Sustainable Growth in Turkey. Baltimore, Md.: Johns Hopkins University Press. Werner, Martin. 1991. "Is Mexico Solvent? Testing the Sustainabffity of the Government's Fiscal Policy." Presented at the Tenth Latin American Meet- ing of the Econometric Society, August, Punta del Este, Uruguay. Wilcox, David. 1989. "The Sustainability of the Government Deficits: Impli- cations of the Present Value Borrowing Constraint." Jourmal of Money, Credit and Banking 21 (August): 291-306. World Bank. 1988. World Development Report 2988. New York: Oxford Univer- sity Press. 2 The External Effects of Public Sector Deficits Carlos Alfredo Rodnrguez This chapter analyzes the effects that public sector deficits and the means of financing deficits have on a specific set of macroeconomic variables related to the external sector. These variables include the real exchange rate, the trade balance, the current account, and the level of external indebtedness. The defidt of the public sector, as measured by public sector bor- rowing requiremernts, is the result of the difference between govern- ment spending and government tax revenues. Therefore, in descrb- ing the effects of a given deficit, it is imperative to separate the effects of the financing of the defict from those derived from the given levels of government spending or taxation. In order to do this we have to design a conceptual experiment. In our case we shall assume that there is available a neutral tax-for example, a value added tax or a consumption tax sudh that changes in the level of the tax do not affect the relative structure of demand for goods or assets. The deficit is generated by reducing this neutral tax and increasing the level of debt financing accordingly, whether the financing be external or internal. From this perspective, what we will be analyzing are the effects of tax versus debt financing in the context of an open economy. In the case of internal debt financng, the government may resort to issuing interest-bearing debt (bonds) or non-interest-bearing debt (money). The issue of tax versus debt finandng has received a lot of attention in the literature in reference to the well-known Ricardian equivalence prnposition. The Ricardian proposition maintains that a tax reduction financed with debt will have no real effects on the economy if the public discounts the future taxes to service the debt and :naeases savings by the exact amount of the tax reduction. However, the empirical validity of the Ricardian equivalence is incondusive. (For a general survey on issues related to Lte Ricardian equivalence, see Leiderman and Blejer 1988). In the context of an open economy, the real exchange rate is a crucial relative price for the alocation of resources in the extemal sector. This relative price certainly will be affected by the composition of government spending and may also be affected by the means of 79 80 The Extenial Effects of Public Sector Deficits financing such spending. In the first section we discuss the Ricardian equivalence proposition in relation to the external effects of tax versus debt financng. In the next two sections we assume that the equiva- lence proposition does not hold and descibe the resulting short-run and dynamic effects of government spending on variables of the external sector. Deficit Financing and the Trade Balance We are concerned with the short-run effects of deficit financing on the levels of the real exchange rate, the trade and current accounts, the levels of domestic and foreign indebtedness, and finally, the inflation rate to the extent that the defict is financed with money creation. The following variables are defined: (221) Y = gross domestic product (GDP). Fs= net financing from private sector to government: taxes plus acquisition of domestic paper (debt or currency minus interest col- lected on domestic debt). (222) Fpg = T + dC/dt + dD/dt-i- D where T = taxes, C = money, and D = intemal government debt. F,p, = net financing from the foreign to the government sector. (2.3) F,p = E * dDjfIdt-it E - Dp where Dp = external private debt and E = exchange rate. An asterisk (*) means that the variable is measured in foreign currency. (2.4) G = government spending on goods. Fes = net financing from the foreign to the government sector. (2.5) F = E- d(D)Idf-it* -E - Dg where D* = external government debt. Private Sector Budget Constraint (2.6) Gp = Y + Fp - Fps = private spending on goods. Government Budget Constraint (2.7) Gg = FM + Feg = government spending on goods. Total Svending on Goods (Z8) GT = Gp + Gg = Y + F, + Fe. - CaTros Alfr(do Rodn'guez 81 Starting from equation 2.8, we can derive a set of propositions that will be the basis for the subsequent analysis. * Proposition 1. Total spending on goods can exceed total output only if it is externally financed (follows from equation 2.8). * Proposition 2. For a given composition of total spending of goods between tradable and nontradable, the real exchange rate depends on the difference between total spending and total out- put of goods-in other words, on the trade balance deficit that is equal to the amount of external financing (to be proved later). * Proposition 3. Govermnent financing strategies will affect the real exchange rate ordy if they affect the trade balance (follows from proposition 2). * Proposition 4. Government financing strategies wil affect the trade balance only if the Ricardian equivalence proposition does not hold. If this is the case, a tax reduction finnced through increased debt (intemal or exteral) will result in some increase in private spending. As a consequence the tradn sur, lus wil deteriorate and the real exchange rate should faL. We would therefore observe that a fiscal deficit generates a real appreciation. Proposition 4 is the starting point for our analysis. The relevant question is whether the goverrunment financing strategies can affect the level of private spending-in other words, the issue of crowding- out, in this case referring also to external borrowing. In order to discuss the effects of deficit financing on the real c ,:change rate, we have to define a neutral experiment through which the defiit icrease does not affect the composition of total spending (which would be an obvious way to affect the real exchange rate). The experi- ment will be a tax reduction coupled with an equivalent inarease in government indebtedness (internal or external). In this way, we assume that a deficit is generated without a corresponding increase in the rate of government spending. There are three ways to finance such a deficit increase domestic debt, increase external debt, or increase the rate of money creation. We shall discuss each method separately. Tax Reduction Financed by External Government Borrowing Consider a Siti ation in which the government switches from tax financing to external financing. If the private sector reacts by investing the tax savings in foreign assets, there will be no effect on total spend- ing or on the trade surplus. The real exchange rate wil not be affected because government borrowing will be unable to affect the trade bal- ance. In terms of equation 2.8, the increase in F; is matched exactly by a decrease in Fi, so that their sum remains unchanged. This conclu- 82 The External Effects of Public Sector Defiits sion follows from a straight generalization of the Ricardian equivalence theory for foreign borrowing. The issue was analyzed in the context of an optimal model by Frenkel and Razin (1986), Auernheimer (1987), and Leiderman and Blejer (1988) and receives some empirical confirma- tion from the Argentine experience during 1978-81. During 1978-SI the Argentine government acquired a substantial external debt that was matched to a great extent by outflows of pri- vate capital. These outflows took place later, when it was already perceived that the government's borrowing and exchange rate poli- des were doomed to failure. There was a transitional period, how- ever, when the government debt was building up, and during this time the trade deficit deteriorated substantially-although part of the deterioration may have been attributable to the trade liberalization that took place, coupled with the policy of the quasi-fixed exchange rate. It is therefore not clear whether the private capital outflow observed was a private compensation for the increased government debt or a simple speculative movement induced by expectations of a large devaluation. This issue will be addressed in chapter 3. As mentioned in Leiderman and Blejer (1988) there is a wide variety of reasons why the Ricardian equivalence proposition may not hold, even in an open economy. For example, the authors mention the existence of borrowing constraints, distortionary taxation, uncer- tainty about the imposition of the required future taxes, and differ- ences in planning horizons for the private and public sectors. We miight add to this list the risk-induced differentials in rates of interest at home and abroad and differences in the spending propensities of taxpayers and bondholders. Given all these arguments, we feel that the degree of substi- tutability of private and government external borrowing is an empiri- cal issue in need of clarification and that a better understanding can be obtained from the specific experiences related in the case studies. Tar Reduction Financed by Internal Borrowing When the government deficit is financed with internal debt, a substi- tutability similar to that of private and government external borrow- ing wil result. If Ricardian equirvalence holds, the lower taxes will be used by the private sector to acquire the increased internal issue of debt, and total private spending will not be increased. However, port- folio composition may indirecdy affect the mix of spending between consumption and investment goods. If the private sector purchases the internal debt with increased foreign. debt, external financing wfll increase and the trade balance and the real exchange rate will be affected. In this case the Ricardian proposition would not hold because private spending has increased Carlos Alfredo Rodrfguez 83 by the exact amount of the tax reduction. Here again, the issue should be subject to empirical verification: is government borrowing inter- mediated externally by the private sector or not? Tax Reduction Financed through Inflation Tax A tax reduction financed through inflation tax is the most obvious example of neutrality, since it amounts to the substitution of one tax by another. Thus no direct effect on the rate of private spending should be expected. However, a differential tax has been imposed on a single financial asset-money-and this may have short- and long- run effects on the desired rates of acquisition of other assets, in par- ticular external assets. The higher inflation rate may stimulate larger desired holdings of external assets by the private sector. In the short run this implies larger capital outflows and, through the reduced rate of private spending, a larger trade surplus (and a higher real exchange rate). In the long run, as foreign private assets grow larger, the interest income will increase. This means that the trade surplus must be lower, since the interest eamed must be spent on foreign goods. Therefore, the long-run effect should be to lower the real exchange rate. This analysis suggests that the nonneutrality of the deficit in the case of the inflation tax is attributable to the use of a nonneutral tax on one domestic asset-money-and not to the validity or lack of validity of the Ricardian equivalence proposition- Gneral Conclusions A deficit financed with debt, whether domestic or foreign, will affect the trade surplus only if the reduced taxes affect the rate of private spending. If the private sector uses the reduced taxes to acquire the new issues of internal debt (when the deficit is internally financed) or to acquire foreign assets (when the deficit is extemally financed), there will be no effects on the rate of private spending, and therefore there will be no relation between the deficit and the trade balance or the real exchange rate. In this case the Ricardian equivalence proposi- tion will be valid, and the choice of tax or debt financing will be neutral. This result also holds true in the case of an open economy. Inflationary financing of the deficit wil affect the extemal sector through the portfolio-induced effects on desired private holdings of foreign assets. We expect a higher inflation rate to have opposite effects on the trade balance in the short ran and in the long run. In the short run higher inflation should improve the trade balance; in the long run higher inflation should prove detrimental to the trade balance- 84 The External Effects of Pubtic Sector Deficits In the next two sections we describe in detail the relationship between the real exchange rate and the levels and financing methods of government spending, assuming that Ricardian equivalence does not hold-for example, that government deficits do have an impact on trade deficits and therefore on the real exchange rate. The analysis will focus on the short-run impact and the dynamic response of endo- genous variables. The purpose will be to derive a set of basic struc- tural relationships that can be used for empirical estimation. The Short-Run Process of Determining the Real Exchange Rate Consider an economy producing two types of goods, tradable (7) and nontradable (N), with prices PT and PN. We define the real exchange rate (e) as the relative price of tradable versus nontradable goods: e = PT/PN. (The analysis in this section draws from and extends the results presented in Rodriguez 1982.) Nominal spending on goods by the private sector is denoted by Gp and government spending on goods is denoted by GC. Total spending on goods (absorption) is the sum of private and government spending: (2.9) G = Gp + Gr Nominal GDnP is denoted by Y, and the difference between GDP and nom-tinal absorption is the trade surplus (TS): (2.10) TS = Y-G. On the demand side, assume that the private sector spends a frac- dion b(e) of total private spending on nontradable goods: (2.11) Gp, = b(e) - Gp. In the same way, the government spends a fraction b8 on nontrad- able goods: (2.12) GSn = bg Gg Total nominal spending on nontradable goods is therefore: (2.13) Gn Gr + Gg8 = b(e) G,, + b8g Gg. The ratio of government spending to GDP is defined as the policy parameter (2.14) g = G8gY. On the supply side, the nominal value of output of nontradable goods is represented as proportional to nominal GDP: (2.15) Y = a(e) -Y. Equilibrium in the market for nontradable goods requires: Carlos Atfredo Rodriguez 85 (2.16) Gn =Y- Substituting (2.13) and (2-15) into (2.16), we obtain: (2.17) h(e) - Gp + bi - Gs = a(e) - Y. Substituting CGp = G - GC and G. = g - Y, we can express (2.17) as: (2.18) b(e)(G-g-Y)+b8 g-Y=a(e)*Y. Collecting terms, we can express this equation as the condition for the excess demand for nontradable goods (EDNT) and make it equal to zero: (2.19) EDNT =b(e) * G-[a(e) + g - (b(e)-b8)- Y = 0. Finally, defining (2.20) es= 1-(GIY) as the ratio of the trade surplus to cDP and substituting into (2.19), we obtain: (2.21) EDNT-b(e) - (1 - ts) -a(e) + g- bc) - bg] =0 -E(e, ts, g, bs). Wairasian stability requires that dElde > 0. The other derivatives are: dE/dts C O and dE/db8 <0 where dEldg Z 0 depending on b(e) z bg. Given these derivatives, we can solve explicitly for the real exchange rate (the market-clearing relative price) as a function of the other determinants: (222) e = F(ts, g, bg) where the signs under the variables indicate the expected signs of the partial derivatives. According to the previous equillrium condition, the real exchange rate should depreciate as the trade surplus inaeases. The reason is simple: a larger trade surplus means a reduction in spending in rela- tion to income. Part of the reduction in spending faIls on nontradable goods, so their price must fall (the real exchange rate rises). An increase in government spending on nontradable goods, bsf should raise their price, so the real exchange rate must fall. An increase in overall government spending for a given trade surplus implies that the government share in total spending has increased so that it has displaced private spending. In this case the demand for nontradable 86 The Extemal Effects of Public Sector Deficits goods will rise or fall depending on who has a larger propensity to spend on this type of goods; this accounts for the ambiguity in the sign of the partial derivative with respect to g. In this analysis we assumed the constancy of the terms of trade and therefore used an aggregate of tradable goods. A more general analy- sis would account for at least the existence of export and import- competing sectors. In that case the real exchange rate would measure the relative price of some aggregate of both tradable-goods prices. The equilibrium value of the real exchange rate in this context should also depend on the relative price of both tradable goods-in other words, the terms of trade-as well as trade taxes and subsidies that create a differential between the internal and external terms of trade. The interrelation between commercial policy instruments andK the equilibrium level of the real exchange rate has been addressed by Dornbusch (1974), Harberger (1988), Rodriguez (1989), and Sjaastad (1980), among others. Assume that there are two tradable goods-exportables and importables-with domestic prices determined by the following arbi- trage conditions: (2.23) Px = P*x- (1-Tx) Pm = Pt*m (1 + Tm) where the variables with asterisks refer to the (constant) foreign cur- rency prices and Tx and Tm are ad valorem trade taxes. There are now two relative prioes in this economy that we may denominate as the export and the import real exchange rates: (2.24) ex = Px/Pn em = Pm/Pn. Since there are now three goods in the economy, the shares of expenditure and output of nontradable goods should now depend on both relative prices: (2.25) a = a(ex, em) b = b(er, em). Substituiting (2.25) into (2.21), it is dlear that the equilibrium condi- tion in the market for nontradable goods (2.22) is now changed to: (2.26) ex = ex(em, ts, g, bg). The internal terms of trade are defined as: (2.27) TT = er/em = (P'x/P*m) (1 - Tx)J(l + Tm) = T* (1 -Tx)I(1 + Tm). Carlos Alfredo Rodriguez 87 This expression allows us to replace em in (2.26) by its equivalent in terms of ex,, TTh, and trade taxes, so that we end up with the follow- ing reduced-form equation: (2.28) ex = F(T, Tx, Tm, ts, g& bg). Since em is a function of ex, 7T, and trade taxes, we can also repre- sent the market equilibrium of nontradable goods by the equivalent condition (2.29) em = G(TT, Tx, Tm, ts, g, bs). Finally, assuming that we still want to refer to a single concept of the real exchange rate, we can define it as an average of the two real exchange rates: (2.30) e= z*ex + (1-z) * em =Z F(-) + (1 - z) G() = H(rr, Tx, Tm, ts, g, b1).. As shown in Rodriguez (1989), the average real exchange rate will still be positively correlated with the trade surplus, but the relation with the terms of trade will vary depending on the weights used to construct it. Equation 2.30 or one of the variations allowed by (2.29) or (2.28) is the basis for the empirical studies on the process of determination of the real exchange rate to be conducted in the country case studies. It follows from this section that government actions affect the real exchange rate at three different levels: total expenditures, composi- dion of expenditures, and extemal financing of the deficit (which is captured by the effects of financing on the trade surplus). As discussed previously, the contribution of the government to the trade surplus is directly measured by the government's ability to obtain foreign financing of its deficit, this financing is adjusted by whichever compensating capital flows are generated from the private sector. However, we still have to discover the process of determining the trade surplus of the private sector in relation not only to government-determined parameters bat also to the desired rate of accumulation of domestic and foreign assets in the private sector. This is done in the next section. Short- and Long-Run Interrelations between the Assets Markets, the Trade Surplus, and the Real Exchange Rate In the previous section we derived the relationship between the real exchange rate, the terms of trade, trade taxes, the trade surplus, and the level and composition of govemment spending. We also saw that 88 Thze External Effects of Public Sector Deficits the trade surplus depends on foreign financing (or lending) from the private and public sectors. While the capital flows of the public sector can be considered a policy variable related to the deficit-financing strategy, private capital flows still have to be explained, since they are an endogenous variable (except in the limiting case in which there is no capital mobility). In this section we extend the general equilibrium model discussed above to incorporate the assets markets and to determine the equilib- rium level of the trade surplus. (For further discussion on the interac- tion between the trade balance, the real exchange rate, and the assets markets, see Dornbusch 1973, Calvo 1981, Frenkel and Rodriguez 1982, and Rodriguez 1982.) The interrelation between the assets markets, the trade balance, and the real exchange rate becomes evident when the effects of the imposition of the inflation tax are analyzed. As mentioned in the first section, the effects on the external sector and on the real exchange rate of deficit financing through the inflation tax will not be neutral. The reason is that the inflation tax, as a nonneutral tax on a pardcular domestic asset (money) sets the incentive for a portfolio shift in favor of foreign assets. Before going into the formal derivation of all the general equilbrium relationships, we provide an intuitive explana- tion of the most basic interrelation, using the example of the inflation tax. The Inflation Tax and the Assets Markets Consider an economy that produces and consumes both tradable and nontradable goods. Individuals hold domestic money and interest- bearing foreign assets. The differential rate of return between both types of assets is the foreign interest rate plus the expected rate of devaluation. In long-run equilibrium the expected rate of devaluation is equal to the rate of inflation. An increased rate of monetary expan- sion generates the expectation of higher devaluation and inflation, and a process of substitution of foreign assets for domestic money begins. For analytical simplicity we assume that there is a freely float- ing exchange rate. (A fixed exchange rate would be inconsistent with the use of the inflation tax.) In the process of ruraing down cash balances and acquiring foreign assets, the nominal exchange rate is expected to rise because both the stock of money and foreign exchange are fixed at a given moment in time. The rise in the nominal exchange rate (the price of tradable goods) also induces, by substitution, some increase in the price of nontradable goods, although in a smaller magnitude, as we shall show later. Carlos Alfredo Rodriguez 89 The short-mn adjustment is therefore obtained through rises in prices and the exchange rate that reduce the real value of the total asset holdings of the private sector. The reduction in real asset hold- ings reduces the demand for nontradable goads and thus allows for the increase in the real exchange rate and the improvement in the trade surplus. The improvement in the trade surplus starts a process of accumula- tion of foreign assets. As holdings of foreign assets accumulate, the pressure on the exchange rate is reduced, and the real exchange rate begins to fall back to its original level. However, since foreign assets are larger than before, the service account shows a larger surplus. As a consequence, in the new long-run equilibrium the trade balance must show a larger deficit since the current account must be bal- anced. In condusion, the imposition of the inflation tax raises the real exchange rate during a transitional period and lowers it in the new long-run equilibrium. This discussion suggests that the real stocks of assets and the infla- tion rate should be added as variables in the equation to determine the trade balance because they are linked to the desired rate of accu- mulation of foreign assets. We now proceed to a formal demonstra- tion of these points in the context of a model that also incorporates domestic issues of public debt. A Dynamic General Equilibrium Model of Determination of the Real Exchange Rate The model developed here describes the dynamic effects on the exter- nal accounts and the real exchange rate of changes in the inflation tax, the foreign interest rate, and the stock of internal public debt. The context of the model is an economy in which individuals hold three types of assets: domestic money (M), a domestic bond denominated in foreign exchange issued by the government (b), and a foreign asset (D). The three assets are assumed to be imperfect substitutes, and the relative demands for the assets depend on the differential rates of return offered. Since we shall be analyzing the effects of the inflation tax, derived from the continuous issuance of money, we have assumed that the government bond is indexed. If it were fixed in nominal terms, as money grew the relative amount of this bond would approach zero. The alternative would be for the govenmuent to issue money and nominal bonds in order to keep constant the ratio between them. The assumption that bonds are already indexed to the price level or to some of its components such as the exchange rate simplifies the analysis without loss of relevance. 90 The External Effects of Public Sector Deficits The economy produces and consumes both tradable and nontrad- able goods. The excess supply of tradable goods is the trade surplus. The trade surplus plus the interest earnings on foreign asset holdings (the current account) equals the change in the stock of these assets. Demands for goods depend on the two nominal prices (Pt and Pn) and nominal expenditure on goods (G). Demands are assumed to be homogeneous of degree zero in all nominal variables. The variable E represents the nominal exchange rate, which is assumed to equal the nominal price of tradable goods (E = Pt). For simplicity we assume that the revenues of the inflation tax are redistributed neutrally to the public and that the interest an the internal public debt is also financed through a neutral tax. Supplies of both goods depend on the relative price (e = PtlPn = E/Pn) and on factor endowments, which we assume are fixed (we abstract here from growth considerations). Given those assumptions, the supply and demand functions take the following form: (2.31) Cm = Cn(E, Pn, G) = Cn(e, G/E) + + Ct = CI(E, Pn, G) = Ct (e, GIE) - + (2.32) Qn = Qn(e) Qt = Qt(e) -+ We define GDP, measured in terms of tradable goods, as: (2.33) y(e) = Qt(e) + Qn(e)le = GDP. The derivative of y(e) with respect to e is defined as: (2-34) y'le) = [QT'(e) + (lie) Qn'(e)] - Qn(e)le2 =-Qn(e)le2 c 0 since the term in brackets is identically equal to zero by the envelope theorem. The trade surplus, measured in foreign exchange, equals the differ- ence between cDlP and expenditure: (2.35) TS = y(e) - GIE. We define ts = TS/y(e) as the ratio of the trade surplus to CDP. Substituting ts into the demand for Cn, we can express it as: (2.36) Cn = CG[e, (1 - ts) - y(ef) = Cn(e, ts). If the trade surplus were to be zero, the demand for Cn would unambiguously depend positively on e (this follows from the Slutsky Carlos Alfredo Rodriguez 91 expansion of the price effect on the demand for Cn). If ts c 0, how- ever, an income effect operating in the wrong direction appears. We assume that the substitution effect dominates, so that the demand for nontradable goods depends negatively on its relative price. We there- fore assume the following signs for the partial derivatives of Cn: (2.37) d(Cn)ld(ts) c 0 d(Cn)/d(e) > O. Equilibrium in the market for nontradable goods requires that the relative price, e, adjust to equal supply and demand: (2.38) Qn(e) = Cn(e, ts). It is clear from equation 238 that an increase in the trade surplus is associated with a lower level of expenditure and therefore with a higher real exchange rate. As expenditure faUs, demand for nontrad- able goods falls, at 4d its relative price is reduced: (2.39) e = e(ts) c' > 0. Equation 2.39 detemines the real exchange rate that equflibrates the market for nontradable goods as a function of the proportional excess of expenditure over GNP (ts). However, ts is also an endogenous vari- able, and we will now turn our attention to determining it. Since the trade surplus is associated directly withl the desired rate of accumulation of foreign assets, we must tur to the description of the assets markets in order to determine the equilibrium level of the trade surplus. The level of nominal assets, A, is defined as: (2.40) A=M+E b+E-D=E (m+b +D) We assume that there is a desired long-nm level of real assets (a*) and that people adjust their expenditures in order to reach it gradu- ally. The desired level of real assets could be defined as a proportion of income or in terms of either commodity. To simplfy the analysis, it is convenient to define the desired level of real assets as constant in terms of foreign exchange: (2.41) A* =2 - E. The level of nominal expenditures on goods equals the sum of nominal income 1Y = E y(e)], foreign interest earned (r* * E * D), a fraction of the excess of actual asset holdings over the desired long- run level, and the amount of foreign transfer needed to finance local government spending, fes: (2-42) G = Y + E r* - D + z (A-A*) +f 92 Thie External Effects of Public Sector Deficits The trade surplus therefore equals (Y - G)IE, using (2.42) and (2.40): (2.43) TS =(at-m-b-l})- r - -fes Equation 2.43 describes the determination of the equilibrium trade surplus. It is directly related to the desired rate of accumulation of assets and to the interest earned on foreign assets. A fiscal deficit financed abroad should be subtracted from (2.43). What we have determined, then, is the structural form for the desired rate of private foreign savings. Equation 2.43 still has endogenous variables in the explanation of the trade surplus to the extent that m can change at any instant through jumps in the exchange rate. In order to determine the equi- librium value of m, we have to describe the conditions of the portfolio balance equilibrium. - The rate of return for holding domestic money is -I, where I is the expected inflation rate. The rate of return on the domestic indexed bond is d + i-r, where d is the expected rate of devaluation and i is the doIlar rate paid by the bond. Finally, the rate of return for holding the foreign asset is r* + d - 1. Since there are three assets, there should be two portfolio-preference functions that we assume depend on the difference between the rates of return of the two assets involved: (2.44) mID =L(+d + L' < O (2a45) bID = H(i -r) H' > 0. The stock of the domestic indexed bond is a variable subject to government control. It is clear that the government cannot resort to bond financing as a permanent source of revenue in the absence of growth. We therefore consider b a policy parameter that takes a fixed value and proceed to analyze the effects of changes in its level. For the moment, we assume that the expected rate of devaluation is a constant parameter. Substituting (2.44) into (2.43), we obtain: (2.46) TS=z*[a*-b-(1+L)D]-r*-D. According to (2.46), the trade surplus depends on the stocks of domestic and foreign assets held (which are constant at a point in time), the foreign interest rate, and inflationary expectations. We can now normalize TS by y(e) to obtain the variable ts: (2.47) ts z-.2e* - b -(1 + L) - D - r' DfzJly(e) L =L(r* + d). Note that in equation 2.47 the real exchange rate enters into the determination of the ratio of the trade surplus to C;DP not because it affects the trade surplus but because real GCmP depends on it. Caros Alfnido Rod rguez 93 Short-run equilibrium is achieved when the home goods market is in equilibrium (2.39) and when ts equals the desired rate of assets accumulation (2.47). Around the steady-state equilibrium, assets equal the desired level, so that a* = b + (1 + L) * D. We now proceed to evaluate the short-run response of the trade surplus to changes in the different parameters when the changes take place in the vicinity of the steady-state equilib- rium. These changes are obtained from differentiation of the short- run equilibrium conditions (2.39) and (2.47). After differentiation, the changes in the ratio of the trade surplus to GDP become: d(ts)ld(D)sr = -z- [(1 + L) + r*]I[. (1 - 1)1 c 0 d(ts)Id(r*)sr = -(z L' D + D)4Iy * (1 - J)] S 0. d(ts)Id(a*)sr = zly a(1 - J)] > 0 d(ts)ld(4,>)sr = -I[y -(1 -J)] < 0 where J-=e'-r* D-y'Iy cO. By (2.39), the real exchange rate depends (positively) only on the trade surplus; the effect of other parameters such as trade taxes (fl) and the level or composition of govenmuent spending have already been analyzed in the previous section and are assumed constant in this section. The partial derivatives in (2.48) therefore also give the sign of the short-run response of the real exchange rate to changes in the different parameters or state variable (D). In particular, it follows that an instantaneous depreciation of the real exchange rate takes place whenever the inflation tax or desired assets are raised, while an appreciation follows from increases in the stocks of domestic or foreign assets held by the private sector (b or D)). Algebraically, these short-run derivatives are: d(e)ld(D)sr = -e' [z(1 + L) + r*]I[y- (1-J)] c 0 d(e)ld(b)sr = -z - e'[y (1 - J)] < 0 d(e)ld(d)sr = -z L'e' D[y- (1 - J)] > 0 d(e)lId(r*)sr = -D e'(z- L' + 1)fy * (1 - J)] Q 0 d(e)ld(a*)sr = z e'I[y(1 - J)] > 0 d(e)ld(f)sr = -e'/y- (1 - 1)] < . In order to dose the model, we have to describe the process of fornation of the expected rate of devaluation. The model described 94 The External Effects of Public Sector Deicits here is similar in reduced form to that of Calvo and Rodrfguez (1977). In that study we closed the model using rational expectations and further showed that a quasi-rational nule of assuming that d equals the rate of monetary expansion yields identical qualitative results. For simplicity, we assume that expectations of devaluation are equal to the constant rate of monetary expansion, mu: d = mu = (1/A{) (dM/dt).1 At any instant, ts and e are jointly determined by the values of the state variable D and the parameters r*, d = mu, and b. Characterizing the dynamic behavior of foreign assets -requires specification of the current account, CA, which equals the trade sur- plus plus foreign interest earnings: (2.50) CA-d(D)Idt = z [a*-b-(1 + L) D. The differential equation 2.50 describing the trajectory of foreign assets is stable. The stock of foreign assets converges asymptotically to the desired long-run level (Dss): (2.51) Dss = (a -b)(1 + L). According to (2.51), the long-run stock of foreign assets depends on a*, the stock of the indexed government bond (b), the foreign interest rate (r*), and the inflation tax rate (d = mu). Algebraically, these changes are: d(Dss)dQa*) = 1/(1 + L) > 0 d(Dss)Id(b) = -11(1 + L) < 0 d(Dss)Id(mu) = -(a* - b) - L'1(1 + L)2 > 0 d(Dss)td(r*) = -(at - b) -LI1(1 + L)2 > 0. We can now compute the long-mn effects on the real exchange rate of changes in the different parameters. The difference between the short-run effects presented in (2.49) and the long-run effects is that we must take account of the adjustment of D to its long-run value Dss. For example, the long-run change in e in response to a change in mu is computed as: (2.53) d(e)ld(mu)ss = d(e)ld(mu)sr + d(e)Id(D)sr d(D)Id(mu)ss. Equations 2.54 summarize the long-run effects of parameter changes on e: d(e)ld(mu)ss = e' r* D - L'/[( + L) y (1 - 1J)] c 0 (2 ^ .54)*ss=e r/l L) *y* (I1-)]cO d(e)ld(b)ss = e' -r*[(l + L) - y (1 - 1)] > 0 d(e)ld(r)ss =e'.D [r -L' -(1 + L)]I[(J + L) - y - (1 -J)] < 0. Carlos Alfredo Rodriguez 95 It is interesting to compare the difference between the short- and long-run responses of the real exchange rate to changes in the differ- ent parameters. The table below presents the qualitative effects on the real exchange rate of changes in four variables: mu a* b ri Short run + + - Longrun - - + + The most striking feature of the table is that in all cases the direction of the short-run impact of a parameter change on the real exchange rate is the opposite of the direction of the long-run change (except for r*, which has an ambiguous short-run effect). An increase in the inflation tax rate, depreciates e in the short run and appreciates it in the long run. The same qualitative effects take place when the desired level of assets is inaeased. An increase in the stock of government debt appreciates e in the short run and depreciates it in the long run. The short-run impact of a higher foreign interest rate is ambiguous in the short run but unambiguously induces an exchange rate deprecia- tion in the long run. Condusions In this chapter we have developed a methodology for analyzing the effects of government spending and the ways that its financing is related to variables in the external sector. The level and composition of government spending affect the real exchange rate because of existing differences in propensities to spend between the govemment and private sectors. The fiscal deficit may or may not affect the external sector, depending on the empirical validity of the Ricardian equivalence proposition. If such equivalence does not hold, we expect government deficits to have direct effects on the economy's overall rate of spending and therefore also on the trade balance. Changes in the trade balance are bound to have both impact and dynamic effects on the real exchange rate. The impact effects are derived from the required expenditure switching, which is necessary to convalidate the new level of the trade balance that is compatible with the spending increase. The dynamic effects are the result of induced changes in the desired rate of accumulation of foreign assets. It follows from our dynamic analysis that it will not be possible to find a stable static relationship between the real exchange rate and the structural parameters without allowing for the fact that the level of foreign assets changes through time. It is therefore necessary to esti- mate a two-equation model: one equation relating the real exchange rate to the level of the trade balance and another determining the trade balance as a function of a set of parameters that includes the fiscal defidt and the stock of foreign assets held- Finally, the trade 96 The External Effects of Public Sector Deficits balance plus foreign interest eamed (the current account) will deter- mine the evolution over time of the stock of foreign pasets. Note 1. If d is not a constant, the derivation should proceed from the differentia- tion of the portfolio-balance relation (2.44): mu - = -(L'/L) (dtl#dt)e + (1ID)' (dD/dt). In the above expression, the hat (A) over a variable indicates the proportional rate of change. If there are rational expectations, the expected change in t should equal change (abstracting from uncertainty or random shocks). Other- wise it can also be assumed that the expected rate of devaluation is formned according to a process of adaptive expectations. In any event, this expression is the basis for the endogenous determination of the expected rte of devaluation. References Auemheimer, Leonardo. 1987. "On the Outcome of Inconsistent Programs under Exchange Rate and Monetary Rules, or 'Allowing Markets to Com- pensate for Government Mistakes.` " Journal of Monetary Economics 19 (March): 279-305. Calvo, Guillermo A. 1981. "Devaluation, Levels Versus Rates." Journal of International Economics 11(2): 165-72. Calvo, Guillermo A., and Carlos A. Rodriguez. 1977. "A Model of Exchange Rate Determination Under Currency Substitution and Rational Expecta- tions." Joournal of Political Econowmy 85: 617-25. Dornbusch, Rudiger. 1973. "Currency Depreciation. Hoarding and Relative Prices." Joumal of Political Economy 81 (August): 893-915. - . 1974. "Tariffs and Nontraded Goods." Journal of International Eco- nomics 4 (May): 177-85 Frenkel, Jacob A., and Assaf Razin. 1986. "Fiscal Polides in the World Econ- omy." Journal of Political Economy 94, pt.1 (June): 564-94. Frenkel, Jacob A., and Carlos A. Rodriguez. 1982. "Exchange Rate Dynamiucs and the Overshooting Hypothesis." International Monetary Fund Staff Paper 29 (March): 1-30. Harberger, A. C. 1988. "Trade Policy and the Real Exchange Rate." World Bank, Economic Development Institute, Washington, D.C. Leiderman, Leonardo, and Mario 1. Blejer. 1988. "Modeling and Testing Ricardian Equivalence: A Survey." International Monetary Fund Staff Paper 35 (March): 1-35. Rodriguez, Carlos A. 1978. "A Stylized Model of the Devaluation-Inflation Spiral." International Monetary Fund Staff Paper 25 (March): 76-89. - . 1982- "Gasto P6iblico, D6ficit y Tipo Real de Cambio: Un AnSlisis de sus Interrelaciones de Largo Plazo." Cuadernos de Economia (Chile) 19 (August): 203-16. Carlos Alfredo Rodrfguez 97 , 1989: "Macroeconomic Policies for Structural Adjustment." Policy Research Wor!cing Paper 247. World Bank, Country Economics Depart- ment, Washington, D.C. Sjaastad, Larry A. 1980. "Commercial Policy, True Tariffs and Relative Prices." In John Black and Brian Hindley, eds., Current Issues in Commercial Policy and Diplomacy. New York: St. Martin's Press. Part II Country Case Studies Argentina: Fiscal Disequilibria Leading to Iypation Carlos Alfredo Rodriguez This chapter was written during the first months of 1989, as hyper- inflation started to develop. The main developments during 1989-93 are not included in the analysis but are briefly described here. Recent Developments: Hyperinflation and Political Crisis On February 6, 1989, Argentina's Central Bank was forced to suspend the convertibility of the austral because of the intense demand for foreign exchange in the daily auctions. The black-market premium skyrocketed, and hyperinflation began to develop. In the presidential elections held in April 1989 the Peronist party defeated the incumbent Radical party; President-elect Menem was scheduled to assume the presidency in October. Quasi-fiscal imbalances and a rush against the austral combined to induce a price explosion. Inflation was 33 percent in April, 78 percent in May, and 114 percent in June. The outgoing administration lost control of the situation, and riots broke out. Three econonmcs ministers were put on trial over a period of three months. Finally President Alfonsin resigned early, transferring power to Menem on July 9, 1989. During that month inflation reached 197 percent. Contrary to expectations, Menem sided with conservative eco- nomic policies and named members of the Bunge and Bom group to the Ministry of Economics. Inflation initially subsided, but the money supply continued to be fueled by the significant. debt of the Central Bank. In December 1989 another run against currency occurred, the minister of economnics resigned, and hyperinflation reappeared, with inflation rates of 40 percent in December, 79 percent in January 1990, 61 percent in February, and 95 percent in March. In January 1990 the new economics minister, Ermant Gonzalez, launched the "BONEX plan," which virtually converted most govem- ment debt and bank deposits into a ten-year dollar-denominated bond called BOrNEX 89. Real M2 (non-interest-bearing money plus time deposits) was reduced to a mere 3.1 percent of cDIP, drastically down from 18 percent in 1986. The refinancing of the public debt made 101 102 Argentina: Fiscal Disequiibria Leading to Hyperinflation possible the elimination of the quasi-fiscal deficit and the adoption of measures to permanently stabilize the economy. The administration announced an ambitious plan of privatization and deregulation. In February 1991 there was a new run against the currency, which resulted in significant devaluation and.an upsurge in inflation. Gonzalez resigned and was replaced by Domingo Cavallo, the for- eign minister, who had long been preparing for the job of economics minister. In April Cavallo announced the convertibility plan that became the basis for the stabilization enjoyed in 1993, when this chapter was written. At the insistence of Cavallo, Congress passed a law that effectively converted the Central Bank into a currency board. The law set the exchange Tate at $1 to the peso, and the Central Bank was allowed to issue monetary base only for the purpose of purchas- ing dollars at the established rate. Since the passage of this law, impressive results in the areas of stabilization and structural adjustment have been achieved. External debt has been renegotiated under the Brady agreement; a significant portion of public enterprises has been privatized (including the tele- phone company, the national airline, tans, and the electricity, gas, and oil monopolies); and moves have been made toward a more open economy and toward deregulation in the social security system, the tax system, and labor markets. Real CDP grew by about 8 percent annually during 1991-92. Public finances have improved dramati- cally, and the country has been unning primary surpluses since 1991. The most significant shadow on the economic front is the relative overvaluation of the currency, wvhich is being sustained by the large capital inflows that have taken place in the past three years. These capital inflows in part represent the repatriation of the funds that had left during the 1980s. Dollar deposits at commercial banks grew by about $11 billion after the hyperinflation. In addition, the government used about $5 billion from the proceeds of privatization to finance local. spending. Finally, the low U.S. interest rates induced capital flows toward areas with higher rates, including Argentina. Background Argentina has had a sad economic history in recent decades, as illus- trated by the indicators shown in table 3.1. In 1989 real GiDP per capita was at about the level of the early 1960s (figure 3.1). This meant a quarter century without growth at a time when most of the rest of the world enjoyed a period of economic achievement. CDP per capita reached a historical high in 1974; Argentina matched this level in 1977 and 1979 but has never surpassed it. In 1980 GDP per capita began declining, which resulted in an accumulated decrease of 23.5 percent .in the ten-year per-od 197948. During the same period, the con- Carlos Alfredo Rodriguez 103 Table 3.1. Key Macroeconomic Indicators, Argentina, 1980-92 Gross GDP Real investmentas imports Inflation (millions of exchange skare of GDP (millions of Year (percent) austraks)b ratec (percent) U.S. dollars) 1980 87 1,129 100 25.3 10,541 1981 131 1,054 258 22.7 9,430 1982 209 1,002 517 21.8. 5,337 1983 433 1,032 381 20.9 4,504 1984 687 1,060 368 20.0 4,585 1985 385 1,014 373 17.6 3,814 1986 81 1,072 379 17.5 4,724 1987 174 1,095 421 19.6 5,818 1988 387 1,068 312 18.6 5,322 1989 4,923 1,018 556 15.5 4,200 1990 1,343 1,022 155 14.0 4,079 1991 84 1,114 169 14.6 8,275 1992 17 1,210 148 16.7 14,872 a. Annual rate, December to December. b. Real GoP at 1970 market prices. c. CPI ratio in relation to U.S. free-market exchange rate. Source: Central Bank of Argentina. Figure 3.L GDP Per Capita (1960 Prices), Argentina, 1960-90 1960 pesos (thousands) 75 - _ _ 70 ~ - ------ 65 -___ 60- 7- . so 5 - -------- -------- 1960 1965 1970 1975 1980 1985 1990 Source: Country case study. 104 Argentina: Fiscal Disequilibria Leading to Hyperinflation sumer price level increased by 3.36 million, equivalent to an annual compound rate of inflation of 349 percent. We do not intend to explain the reasons for Argentina's economic stagnation; instead we will concentrate on the influence of public sector behavior on that process. In particular, we will be concemed with the effects of government spending, taxation, and deficit financ- ing on the rest of the economy. We will not deal with the effects of government regulation because that topic deserves a separate volume- Govemment has played an important role in Argentine society. It taxes, spends, produces a wide variety of goods and services, regu- lates financial markets, supplies financial services, systemnatically resorts to incomes policies, and regulates foreign trade. Although the regulatory aspects of government action defy any possible quantita- tive measure, an impression of the size of government involvement in economic activity can be obtained by looking at the relative share of government spending in CDP. Figure 3.2 shows that government spending systematically tended to grow faster than GDP Unifl the last crsis of the Argentine economy, which began to develop in 1982. After that, spending began to fall, more as a consequence of limited resources than because of deliberate political action. The fall in the relative share of government spending, however, came too late to avoid a crisis that was already well in the making and that brought the country to a state of hyperinflation in 1989, when the inflation rate approached 5,000 percent and CDP fell by about 7 per- cent. The December-to-December values of the rate of inflation in the consumer price index (cpr) are shown in table 32 It is dear that inflation has been an ever-present phenomenon in recent decades and that it has been explosive. The hypeinflation of 1989 does not appear to be the end of the story. After the monthly inflation rate reached a peak of 195 percent in July 1989, inflation returned to the single-digit level. But in December 1989 a new hyperinflation started. It peaked at 95 percent in March 1990 and apparently ended in April 1990, when the monthly rate was "ornly" 11.5 percent. A quick characterization of government action in Argentina would be that the government has generally tended to increase spending and run deficits. Governments do not need to run fiscal surpluses all the time, especially when the economy is growing. But the Argentine economy has been stagnant over the past two decades, and the gov- ernment has run fiscal deficts every year since at least 1961 (figure 3.3). Indeed, the govenment has run a primary deficit (excluding interest payments) every year in our sample dating back to 1961. This means that in each of the past twenty-eight years, after paying for current and capital spending, the government has not had any resources left for servicing interest on its internal or external debt. Carlos Alfredo Rodrlsuez 105 Figure 32- Total Expenditures of the CNPPS, Argentina, 1961-87 Percentage of GDr 60 55 50 - - 40 35 30 .... -__.-- 1960 1965 1970 1975 1980 1985 1987 Notr. CNFTs, consolidated nonfinancial public sector. Sourcc- Country case study. Table 3.2- Annual Inflation Rates in the Consumer Price Index, Argenrtin, 1961-92 (percent) Year Rae Mar Re 1961 165 1977 160A 1962 30.4 1978 169.8 1963 23-9 1979 139.7 1964 18-2 1980 87.6 1965 38.2 1981 1313 1966 30.0 1982 209.7 1967 27.3 1983 433.7 1968 9.6 1984 688.0 1969 6.6 1985 385.4 1970 21.7 1986 81.9 1971 39.1 1987 174.8 1972 64.2 1988 387.7 1973 43.8 1989 4,923.7 1974 40.1 19-O 1,343.0 1975 335.1 1991 84.0 1976 347.6 1992 17.5 Sowre: National Statistics Institute (INDEC). 106 Argentina: Fiscl Disequilibrina Leading to Hyperinflafion Figure 3.3. Primary Deficit of the CNFPS, Argentina, 1961-88 Percentage of oDP 15.0 - _ 12.5 - 10.0 75 -t 4 ----9- 5.0 7-5- n.o - - -- --- 1960 1965 1970 1975 1980 1985 1988 Source- Country case study. (Dornbusch and de Pablo 1987 and Rodriguez 1989 have discussed the fiscal nature of Argentina's apparent inability to serve its mostly public foreign debt.) As a consequence the government has resorted systematically to issuing money and interest-bearing debt- The out- come of this deficit policy has been a systematic tendency of the economy to run high inflation and high real interest rates as a conse- quence of the pressure on the credit markets from the incremental government borrowing. For an economy that is not growing and that faces a positive real rate of interest, rnning a positive primary deficit implies an ever- growing stock of public debt in relation to GDP. Of couxse, real gov- ernment debt did not grow continuously because occasionally the existing stock of debt was melted down by bursts of inflation-fueled by large devaluations in the face of a foreign exchange crisis-that exceeded nominal interest rates. (For more on the relationship between defcits and devaluations, captured by increases in the level or in the rate of change of the exchange rate, see Calvo 1981a, Rodriguez 1978, and Femandez 1989.) The tendency to melt down the existing stock of debt by imple- menting unexpected devaluations was eventualy discounted by the market; in recent years the market has demanded and has obtained an increasing degree of indexation of the public debt by either the price level or the price of foreign exchange. This meant that the gov- Carlos Alfredo Rodriguez 107 emnment could no longer melt down the stock of real debt and had to face the critical problem of a growing real debt in the context of a persistent tendency to run primary deficits. The pnmay deficit has been falling since 1983 as a result of reduced spending and higher revenues. This trend, however, is not enough to reverse the increasing reluctance of the public to hold the intemal goverrunent debt or domestic currency. The government has increas- ingly had to resort to the use of forced investments by the banking system (depositas indipvnibles). The fall in demand for assets denomi- nated in domestic currency induces real interest rates that are incon- sistent with the real equilibrium of the economy- The shift out of domestic currency leads to an increasing degree of dollarization of the economy and a tendency for recurrent currency runs that result in frequent macroeconomic devaluations- The government was forced continuously to raise interest rates so that it could induce the public to keep holding the domestic currency and could roll over the public debt. The higher interest rates were paid back by issuing more debt and money, and the senrice of the debt became the major source of money creation. The system finally exploded when the country entered hyperinflation in May 1989. The hyperinflation, however, was not able to melt down the stock of interest-earning government debt because much of the debt reached maturity between one and seven days, and interest rates actually had a tendency to anticipate devalua- tions. In January 1990 the government replaced all interest-bearing obligations in the financial system-that is, government debt plus interest-earning deposits-with an issue of dollar-denominated gov- ernment paper that paid the London interbank offered rate (LBOR) and matured at ten years. This chapter is concerned with how public sector deficits and the methods of financing them affect inflation, interest rates, the real exchange rate, private savings and investments, and the external bal- ance. The next section takes up measurement issues. Measurement of the Public Sector Deficit In the past fifteen years many important changes have occurred in the distribution of revenues and expenditures among different levels of government in Argentina. These changes were a result of (a) the transfer of important expenditure items, such as elementary and sec- ondary education and local transport systems (including Buenos Aires' subway), from the central to the provincial and cty govern- ments; and (b) changes in the tax laws and the rules under which tax revenues are distributed among the central and the provincial gov- ernments (the federal copartidpation law). Provincial and local gov- 108 Argentina: Fiscl Dsequilibria Leading to Hyperinflation ernments have no well-defined budget constraints in Argentina, and the distribution of resources (which are collected mainly by the cen- tral government) is highly unpredictable. This makes it especially important in the Argentine case to work on the basis of consolidated nonfinancial public sector (CNFPs) data. Studies dealing with the behavior of the consolidated nonfinancial public sector in Argentina indlude Fundaci6n de Investigaciones Econ6micas Latinoamericanas- FREL (1987) and Schenone (1987). It is also important to account for the operations of the Central Bank as a source of substantial amounts of revenue (because of inflationary money creation) and no less substantial losses. The losses are a result of (a) the purchase of international reserves (induding those required to service the foreign debt); (b) the offering of swaps and other "exchange insurance" mechanisms, which have frequently been used to attract short-term foreign financng; and (c) losing operations of the Central Bank with the domestic financia system (including the bailout of failing financial institutions). In a later section we present estimates made by nEL of losses incurred by the Central Bank of Argentina, which should be considered components of public sector expenditures. Consolidated Nonfinancial Public Sector Table 3.3 presents official data on "above-the-line" CNFPS expendi- ture, revenue, and financing according to the "intemational meth- odology." (fIe international methodology computes only the operat- ing surpluses or deficts of public enterprises as components of public sector revenue and expenditure; it does not consider the current reve- nue and expenditure of these enterprises.) Figure 3.4 shows the evo- lution of the total expenditures and revenues of the cNRrS.. Note that revenues historicaly have sholi rr a growth trend similar to but below the growth of expenditures. Total revenues rose from a low of about 23 percent in 1%2-65 to a high of 41 percent in 1985. (The revenue measure does not incorporate current revenues from public enter- prises.) This high growth in fiscal revenues in the face of a stagnant economy should be enough to invalidate the commonly voiced claim that Argentina's basic problem is that the public does not pay taxes. The fact is that fiscal pressure is very high and has grown at a much faster rate than has GDP in the past twenty-five years. Even though fiscal pressure has been high, fiscal spending has also grown and has exceeded revenues. Expenditures of the CNFPS reached a level of 55 percent of GDP in 1983 (up from 30 percent in the 1960s). A serious attempt was then made to reduce government spending. Spending was reduced to 11 percent of GDP, but the decrease was wiped out by the increased quasi-fiscal spending by the Carlos Alfredo Rodrfguez 109 Figure 3A. Total Expenditures and Revenues of the CNFPS, Argentina, 1961-87 Percentage of GDr 60- 50- - --- ------- 25 - * * 1960 1965 1970 1975 1980 1985 1987 Expenditure ------- Revenue Source: Country case study. Central Bank in servicing its accumulated financial liabilities. The last column of table 3.3 represents the difference between total expendi- tures and resources. A deficit has to be financed through increases in short- and long-run debt (domestic or foreign), advances from sup- pliers, or money creation in the form of credit from the Central Bank and increases in short-term financial liabilities. Table 3A presents the "'below-the-line" items that describe the composition of the financing of the fiscal deficit. In every year from 1961 to 1988 the financial result has been negative; the government has always had to resort to printing money or issuing short-term financial liabilities in order to balance the budget. According to the official data there has been no financing of the cNFps from the Central Bank since 1986. Until mid-1985 Central Bank financing of the CNFPS was recorded in an account labeled "advances to the treasury." Since the stabilization attempt launched in June 1985, no more drawings have been made on that account, according to official statistics. Actu- allv, since 1985 the cnFPs has borrowed from the Central Bank by obtaining rediscounts for public banks and other public enterprises- (Text continues on pzge 174.) Table 3.3. Consolidated Nonfinancial Public Sector "Above-the-Line" Data, Argentina, 1961-88 (percentage of CDP) Current Cuitreh? Capital Capital Total Total Deficit or revenule expenditture revenu4e expenditure reveniea expeadilureb finrancing nteeds Total (1) (2) (3) (4) (5) (6) (6-5) 1961 23,17 17.41 0.55 10.68 28.67 33.03 4.36 1962 18.63 17.84 0.76 9.32 23.85 31.61 7.76 1963 18,47 17,93 0.68 8,22 23.37 30.37 7.00 1964 18.00 18.11 0.37 6.89 22.81 29.43 6.62 1965 18.86 16.88 0.35 6.41 22.31 26.39 4.08 1966 20.13 18.98 0.36 6.45 24.52 29.45 4.93 1967 23.35 18.26 0.48 7.65 27.78 29.86 2.08 1968 22.80 17.24 0.54 8.33 27.58 29.81 2.23 1969 22.53 16.53 0.45 8.21 26.67 28.42 1.75 1970 22.83 16.65 0.43 8.48 26.68 28.55 1.87 1971 20.51 17.04 0.34 8.20 24.72 29.12 4.40 1972 18.35 15,66 0,32 8.81 23.00 28.80 5.80 1973 19.01 19,11 0,11 7.50 25.07 32.55 7.48 1974 22.12 21.87 0.49 8.81 30.09 38.17 8.08 1975 15.87 22.16 0.15 8.94 27.10 42.18 15.03 1976 18.14 16.92 0.15 '13.06 27.78 39.47 11.69, 1977 22.93 14.78 0,40 13.26 28.85 33.56 4.71 1978 25.44 19.75 0.35 12.54 31.29 37.79 i.50 1979 24,44 21.10 0.25 10.51 31.84 38.35 6.51 1980 26.89 25.75 0.31 9.51 33.75 41.22 7.47 1981 24.39 29.11 0.25 9.65 32.95 46.21 13.26 1982 22.53 29.90 0.49 8.56 30.51 45.63 15.12 1983 23,08 29.03 0.23 9.68 40,24 55.39 15.15 1984 22.49 27.03 0.24 7.82 33.65 45.57 11.92 1985 26.96 26.90 0.25 7.06 40.92 46.94 6.02 1986 26.17 24.54 0.21 7,48 40.10 44.83 4.73 1987 23.33 23,23 0.25 6.13 34.59 40.13 5.54 1988C - - 7.39 -Not available, a. Includes carryover from previous fiscal years and contributions. b, Includes contributions. c. The data available for 1988 do not include the deficits of provinces and the city of Buenos Aires. For the remainder of the consolidated public sector the deficit Is 6.93 percent of cD. We estimate a deficit for the provinces and Buenos Aires equal to the 1987 deficit, 0.46 percent of cDP. Source: Ministry of Economics, Argentina. Table 3.4. Consolidated Nonfinancial Public Sector "Below-the-Line" Data, Argentina, 196148 (percentage of cDP) Deficit or Net f_ianctnig' uaEililies financhig Advaines iteeds Total Credit fromn TolIal Central Year (1) (2) Totai Domeslic foreign suppliers (1)-(2) Banik Change 1961 4.36 2,86 2.86 2.38 0.48 0 -1.50 0.19 1.32 1962 7.76 4.52 4.52 3.77 0.75 0 -3.24 0.71 2.53 1963 7.00 3.15 3.15 2.26 0.89 0 -3.85 1.37 2.49 1964 6.62 1.09 1.09 0,71 0.38 0 -5.53 0.49 5.04 1965 4.08 1.20 1.20 0,50 0.70 0 -2.88 0.83 2.04 1966 4.93 0.31 0.31 0.01 0.30 0 -4.62 0.28 4.35 1967 2.08 -0.17 -0.17 0.12 -0.29 0 -2.25 1.00 1.26 1968 2.23 0.42 0.42 0.47 -0.05 0 -1.81 0.71 1.10 1969 1.75 0.94 0,94 0.57 0.37 0 -0.81 0.72 0.10 1970 1.87 0.86 0.86 0.05 0.81 0 -1.01 3.38 -2.36 1971 4.40 1.92 1.92 0.73 1.19 0 -2.48 2.43 0.06 1972 5.80 1.93 1.93 0.89 1.04 0 -3.87 1.81 2.06 1973 7,48 0.91 0.91 0.79 0.12 0 -6,57 4.47 2.10 1974 8.08 1.87 1,87 1.00 0.87 0 -6.21 5.60 0.62 1975 15.08 1.88 1.88 1.77 0.11 0 -13,20 9.74 3.46 1976 11.69 4.14 4.14 3.03 1.11 0 -7.55 3.41 4.13 1977 4.71 2,83 2.83 1.54 1.29 0 -1.88 2.44 -0.56 1978 6.50 5.49 5.49 3.54 1.95 0 -1.01 0.86 0.16 1979 6.51 5.53 5.53 3.77 1.76 0 -0.98 -0.30 1.27 1980 7.47 2.99 3.41 1.53 1.88 -0.42 -4.48 3.59 0.89 1981 13.26 8,03 8.27 3.97 4.30 -0.24 -5.23 5.32 -0.09 1982 15,12 6,30 6.39 5.07 1.32 -0.09 -8.82 7.29 1.51 1983 15.15 -1.63 -1.34 -1.89 0.55 -0.29 -16.78 16.60 0.19 1984 11.92 -1,49 -1.48 -0.59 -0.89 -0.01 -13.41 6.19 7.22 1985 6.02 0.58 0.63 -0.30 0.93 -0.05 -5.44 2.33 3.11 1986 4.73 -0.02 0.04 -1.01 1.05 -0.06 -4.75 0 4.74 1987 5.54 4.17 4.22 1.03 3.19 -0.05 -1.37 0 1.37 1988 7.3- - - - Not available. a. All data are net. Source: Ministry of Economics, Argentina. 114 Argentina: Fiscal Disequilibria Leading to Hyperinflation and these loans have not been paid back. The CNFPS has also resorted to placing dollar-denominated treasury bills with the Central Bank in exchange for local currency, a procedure that has been labeled as external financing but that is actually equivalent to printing money, since these dollar-denominated treasury bills will very likely never be paid back. The Central Bank also has become the recipient of a large fraction of the service of the foreign debt. Quasi-Fiscal Expenditures In countries like Argentina central banks often suffer substantial losses-losses that can never be recovered-on loans to the private financial system or from the bailout of failing financial institutions. In 1977 Argentina's Central Bank started to collect interest on the fraction of reserve requirements that corresponded to nonremunerated bank deposits (demand deposits) and to pay interest on the reserves held on account of interest-bearing time deposits. The balance of these opera- tions is kept at the Monetary Regulation Account, which has proved to be a source of additional deficit because the interest paid has exceeded the interest collected. In 1985 the Monetary Regulation Account was modified to incorporate remunerated and nonremunerated reserve requirements. In addition, the Central Bank started to sterilize liquid funds by issuing a variety of short-term liabilities, indluding short-term certificates of deposit (cDs) and lump-sum mandatory deposits, that absorbed part of the liquidity of the commercial banks. Table 3.5 presents estimates by FEL of the quasi-fiscal expenditures of the Central Bank of Argentina for the period 1960-85. It is difficult to detexmine in advance when the quasi-fiscal deficit will result in additional money creation because much of the interest on the Cen- tral Bank's liabilities has beern paid by creating more such liabilities. This mechanism gave rise to a situation in which the Central Bank gradually absorbed a growing fraction of the lending capability of commercial banks. As of 1989, it is reported, more than 80 percent of the assets of the commercial banks were placed in liabilities of the Central Bank. This way of managing liabilities generated a situation in which the Central Bank, instead of being the "lender of last resort,"' became the "'borrower of first resort." The implications for monetary policy and the eventual development of hyperinflation will be discussed in the section dealing with monetary policy. Table 3.6 summarizes the data on the fiscal deficit. Another important source of quasi-fiscal deficit was the losses from swaps and other exchange rate insurance procedures. These opera- tions, which were concentrated in the 1982-85 period, led to the Central Bank's absorbing most of the outstanding extemal debt of the private sector. (The process is described in detail in Rodnrguez 1989.) Carlos Alfredo Rodriguez 715 Table 3.5. Quasi-Fiscal Expenditures of the Central Bank, Argentina, 1960-5 (percentage of CGP) Accumtulation of Net intenal ianal Yawr loans" 1EAb Swapsc resenrs Total 1960 -0.32 2.09 1.77 1961 1.82 -2.26 -0.44 1962 0.34 -3.08 -2.74 1963 -0.85 0.92 0.07 1964 -0.52 -0.81 -1.33 1965 0.11 0.28 0.39 1966 -0.39 -0.16 -0.55 1967 -1.24 2.66 1.42. 1968 -0.03 -0.14 -0.17 1969 0.94 -1.33 -0.39 1970 -0.54 0.48 -0.06 1971 0.66 -1.75 -1.09 1972 -0.51 0.69 0.18 1973 -1.83 0.77 -1.06 1974 -2.79 -1.14 -3.93 1975 -2.27 0.16 -1.30 -3 41 1976 -6.21 5.00 3.46 2.25 1977 -5.86 1.28 0.57 3.59 -0.42 1978 -2.36 2.72 1.98 2.34 1979 -0.82 0.86 2.80 2.83 1980 3.17 -0.61 -3.43 -0.87 1981 2.94 -1.38 3.99 -5.33 0.22 1982 7.57 4.98 16.63 -3.74 25.45 1983 -3.69 1.04 8.59 -1.06 4.87 1984 3.80 4.06 10.01 -0.30 17.56 1985 -1.78 1.56 4.63 2.99 7.00 Note: Blanks signify not applicable. a. Annual changes in loans to the financial system minus annual changes in reserve requirements. b. The Interest Equalization Account (Cuenta de Regulacidn Monetaria) of 1977 and the cost of remunerated reserve requiremnents and other remunerated liabilities intro- duced in its place in 1985. c. Losses for swaps and other "exchange insurance" mechanisms. Sounr: FIEL 1967. The PIEL figures for this period (see the third column from the right in table 3.5) overestimate the quasi-fiscal expenditure impact of these policies because the exchange losses they implied were not presently realized but were simply documented as public extemal debt. In addi- tion, these figures include interest accrued, although not actually paid, on that debt. To the extent that the external debt was not fully 116 Argentina: Fisal Disequilbia Leading to Hperinflation Table 3.6. Summary Results for Fiscal Deficit, Argentina, 196148 (percentage of GDP) Pnimnry Intcrest, defict Total CPS Quasi-fisal Total Year CPS of CPS deficit CPS deficit deficit 1961 0.67 3.69 4.36 -0.44 3.92 1962 0.83 6.93 7.76 -2.74 5.02 1963 0.88 6.12 7.00 0.07 7.07 1964 1.11 5.51 6.62 -1.33 5.29 1965 0.75 3.33 4.08 0.39 4.47 1966 0.75 4.18 4.93 -0.55 4.38 1967 0.59 1.49 2.08 1.42 3.50 1968 0.53 -1.70 2.23 -0.17 2.06 1969 0.39 1.36 1.75 -0.39 1.36 1970 0.41 1.46 1.87 -0.06 1.81 1971 0.48 3.92 4.40 -1.09 3.31 1972 0.57 5.23 5.80 0.18 5.98 1973 0.53 6-95 7.48 -1.06 6.42 1974 0.81 727 8.08 -3.93 4.15 1975 0.69 14.39 15.08 -3.41 1L67 1976 1.47 10-22 11.69 2.25 13.94 1977 1.14 3.57 4.71 -0.42 4.29 1978 1.92 4.58 6.50 2.34 8.84 1979 2.03 4.48 6.51 2.83 9.34 1980 1.85 5.62 7.47 -0.87 6.60 1981 3.88 9.38 13.26 0.22 13.48 1982 628 8.84 15.12 25.45 40.57 1983 3.50 11.65 15.15 4.87 20.02 1984 2.93 8.99 11.92 17.56 29.48 1985 2.93 3.09 6.02 7.40 13.42 1986 2.37 2.36 4.73 na. n.a. 1987 1.88 3.66 5.54 n.a. n. 1988 2.09 5.30 7.39 n.a. _n.a. n.a. Not applicable. Source: Tables 3.3 and 3.4. serviced, this new liability did not result in money creation. Some money creation took place, however, through the implementation of a variety of debt conversion mechanisms, including debt-equity swaps and onlending, which in effect implied the repurchase of exter- nal debt with newly printed money or short-term financial liabilies issued in local currency. Inflationary Financing of the Deficit As was mentioned above, we have doubts about the relevance of the official information presented in table 3.4 on the actual financing of Carlos Alfredo Rod dgsez 117 the deficit. In particular, that information shows no Central Bank financing of the CNFPS during the years 1986-87. But direct treasury borrowing from the Central Bank was in fact carried out in such a way as not to show openly in the accounts. For example, rediscounts were granted to public enterprises, and treasury paper was denominated in dollars to avoid direct borrowing in australes. Public construction programs were implemented by the Banco Hipotecario (the state mortgage bank) and were fully financed through Central Bank redis- counts that were never returned. Banco Hipotecario is alleged to have lost about $1 bilion in 1987-88 because of such operations, and all this was financed through money creation. Measuring the Inflation Tax Because of these questions, we resorted to measuring directly from the accounts of the monetary sector the fraction of the deficit that was financed by the printing of money. We measured the revenue from money creation as the absolute monthly change in M1 (non-interest- bearing money) divided by the exchange rate for the U.S. dollar in the free market. This provided a monthly series of U.S. dollar revenue from money creation. We then added up the series for each year to obtain annual revenue. Finally, we divided nominal GDP for each year by the average exchange rate in the free maiket to estimate dollar c.P. Dividing the revenue series by GDP, we obtained revenue from money creation as a percentage of CDP. - This series has a serious problem that makes it not comparable with our series for the CNFPS deficiL Since the Central Bank intervenes in the foreign exchange market by buying or selling foreign exchange, in many instances the changes in Ml are a consequence of increases in money demand that are satisfied through purchases of international reserves. Conversely, when the reserves are lost, we observe a signifi- cant fall in the revenue from money creation, as measured. Since reserve purchases or sales are not considered a public expenditure in the accounts of the cNFPs, we have resorted to subtracting those reserve changes from our series of revenue from inflation. With this correction the series of revenue from money creation obtained from the monetary data shows a clear correlation with the series of the deficit of the cNFPs obtained from the fiscal data. Both series are shown in figure 3.5. Almost systematically, the deficit of the consolidated public sector (the last column in table 3.6) exceeds the revenue from money creation. The difference between the two series can be taken as an approximation of the part of the total deficit that was financed by issuing interest-eartning debt. The generation of pub- lic debt should have been larger than this amount because the Central Bank had its own quasi-fiscal deficit that also had to be financed. We 118 Arsentina: Fiscal Discquilibia Leading to Hyperinflation Figure 3-5. Revenue from Money Creation and Total Deficit of the CNFPS, Argentina, 1961-87 Percentage of GWD 17.5 - 15.0 - --I*-' 12.5- _ < \^ _ 10.0 0 - 7.50 A I 12.5 - ---- - ----- ------___ _ Sj 1960 1965 1970 1975 1980 1985 1987 -----Deficit Revenue from money creation - So5urce Counry case study. do not believe tlhat the FlEL data in table 3.5 are comparable with our series because part of what FEL calls qulas:i-fiscal expenditure m'ay actually be indirect financing of the cNFps decidt through the mecha- nisn-ts discussed in the preceding section. The FI. data also consider the changes in reserves as part of the quasi-fiscal expenditures. With these caveats, dle data2 in figures 3.5 and 3.6 desmnbe the events quite closely in tens of the actual methods of deficit fi cing used in recent years. From 1964 throug 1975 the deficit of the CNFPS did not exceed significantly the revenue from money creation. Ii means that the iscal deficit was mainly fiarnced by creating money ather than by issaing deot. The stuacion qhanged drastlically in 1976, when debt finanng apparengtybecame a significant means of financ- ing the deficit. This ahange ooincided with the fall of ehe Perorest goverWiment and ead,te iation of the mRigury res.ne. en 1977 itere was a signisfcant financial mberalization that opened the domestic financial market to foreign investors. The pefiod 1977- 7d was mareed by foreign borrowing, and revenue frmomn money Teh ation fell weti bylow historical nevels. The banking dasis of ealy 1980 put an end to this stage of foreign fimancing of the deficit and cleared C-- riasAlfredo Rodfguez 119 Figure 3.6. Estimated Use of Debt Finaning, Argentina, 1961487 Percentage of CDP 10.0 - 75 5.0- -_ 0.0 - 1960 1965 1970 1975 1980 1985 1987 Souict Country case study. the way for the next stage of interal debt financing, which lasted until 1985. Begining in 1985, a serious effort was made to reduce the total defiat of the aN'. The use of the inflation tax, however, did not fall proportionately to the deficit of the CNFPS because of the mcreasing financial pressures of the service of the internal debt, which was held mostly in the Central Bank The Fisca Deficit and the Inflation Tar: Statistical Evidence In the preceding section we saw that the revenue from money cre- ation appeared to be closely associated with the deficit of the consoli- dated public sector. This should not surprise us because priting money is one of the only two ways of financing the deficit of the cNFPs-the other being issuance of new interest-earning debt. In this section we identify more dosely the nature of this ilationship, using regression analysis. Our methodology consists of considerng the revenue from money creation as an endogenous variable that is explained, among other things, by the need of the cNFPs to finance its deficit. We therefore consider the deficit of the cNes as the exogenous variable that gives nse to the need to resort to an inflation tax. Because of the possibility of financing the deficit with internal or external debt, we should not expect a stable one-to-one relation between the inflation tax and the 120 Argentina: Fiscal Diseuilibria Lading to Hyperinflation deficit. In fact, as the previous discussion indicates, debt financing was more widely used in some periods than in others. Nevertheless, one should consider that debt generation in excess of real growth wfll eventually have to be canceled or be financed through inflation, as the market for placing new debt eventually dries up. Since it was also obvious from the inspection of the data that, starting in 1976, more debt financing was used in the financing modality, we have indluded a dummy (D76 equals 0 through 1975 and equals 1 from 1976 on). Regression 3.1 reports tie results concering the relation between the infl. tion tax (rrAx) and the total deficit of the consolidated nonfinancial public sector (DEFT). (3.1) ox = 1.248 - 2.138D76 + 0.564DEFr. (1.82) (-2.85) (6.19) Adjusted R2 = 0.6 Durbin-Watson statistic = 2.31 Sample period = 1963-87 The regression was also run with correction for first-order auto- correlation of residuals, AR(1), but the AR(1) coefficient did not turn out to be significant. The coefficient of the DEFT variable is 0.56, mean- ing that 56 percent of the cNFPs fiscal deficit is financed through money creation. The coefficient of DEFT on rrAx is highy significant and justifies the presumption that deficit financing is a significant factor explaining the need for the inflation tax. The negative coeffi- cient on the D76 variable indicates that since 1976 there has been a tendency to use less inflationary financing for any amount of DEFT and that as a consequence there has been a shift toward more debt financing. We also tried to use the prmary deficit (PD) of the CNFPS, which indudes no interest service, as the explanatory variable for rrAx and found the surprising result that it works much better than the total deficit (DEFr). The results are reported in regression 3.2. (3.2) Ax = 0.936 - 121D76 + 0.70PD. (1.5) (-1.98) (7.47) MA(1) = -1.12 (f-value = -2-4) Adjusted R2 = 0.69 Durbin-Watson statistic = 1.59 Sample period = 1963-87 The results of regressions 3.1 and 3.2 suggest the possibility of a tendency to finance the primary defict with money and to roll over the interest expenditures in the form of new debt. To check for that possibiity, we inlduded thr interest expenditure of the cNm's as an additional explanatory variable, together with the primary deficit. If Clos AifredodRodRiguez 121 all deficits, independent of their sources, received the same treat- mertt, we should expect the coefficent on interest expenditure (IE) to be the same as the coefficient on the primary deficit. This does not tum out to be the case, as the results of regression 3.2 show. The primary deficit retains the same coefficient of 0Q7, while the interest expenditure turns out to be insignificant in explaining the inflation tax. We condcude that over the sample period authorities have used the inflation tax to finance primary expenditures, while interest expenditures that result from the existing stock of govenmment debt have tended to be refinanced by issuing more debt. (3.3) rrx = 0.929 - 1.26 276 + 0.697PD + 0.0271E. (1.4) (-1.4) (6.78) (0.08) MA(1) = -0.62 (t-value = -2.36) Adjusted R2 = 0.68 Durbin-Watson statistic = 1.59 Sample period = 1963-87 To determine the effects of the public sector defict on inflation, the natural next step is to ascertain the relation between the inflation rate and the revenue from money creation. Such linkage is provided by the demand for real cash balances, which has the inflation rate, taken as a measure of the opportunity cost of holding money, as a determin- ing variable. Precise estimates of money demand for Argentina wil be derived in the next section, but we can already obtain some prelim- fia-ry estimates by running a regression of the inflation rate on the series of revenue from money creation. The response of the inflation rate to the printing of money in order to finance the deficit need not be instntaneous, since there may be lags in the adjustment of prices to changes in the muney supply. We there- fore include inflation that lags one year as an explanatoxy variable for current inflation in addition to current revenue from money creation. The results are presented in regression 3A which indicates a dear association between the revenue from money creation and the result- ing inflation rate. Here we have assiamed a linear relation between rrAx and the inflation rate, a fact that may not be valid for high inflation rates because as inflation raises the base of the tax, real cash balances decrease, and actual revenue may in fact fall. In the next section we derive the precise nonlinear relationship using an esthmate of the demand for money. Our results here are therefore an approximation valid for inflation rates to the left of the maxiudm of the revenue curve. The results of regression 3. indicate that the long-run effect of raising 1 percentage point of C;DP from money creation is associated with an additional 97 percent inflation rate. Regressions 3.2 and 3A provide a structural framework for the relation between the public 122 Argentina: Fisal Disequrbiln Leding to Hyperinflation sector deficit and inflation (INF). An additional 1 percentage point of primary deficit is financed with 0.7 percent of revenue from money creation (the rest is financed with debt), and the effect of collecting this revenue from money creation is about 67.9 percent of additional inflation (97 x 027). (3.4) INF =-84.98 + 0.751 [NF (-1) + 29.34 rrAx. (-1.76) (5.89) (3.60) Adjusted R2 = 0.63 Durbin-Watson statistic = 1.898 Sanmple period = 1964-87 The implied long-nm relation is: [NE = -340 + 97.2PD. I nflation, Revenue from Money Creation, and tlhe Strtctre of Financial M01arkets As might be expected in an economy subject to frequent shocks, Argentina's financial markets are highly volatile and cannot be easily described by a simple set of instruments. The financial reform of 1977 freed interest rates and allowed banks to offer interest-bearing deposits (plazas fijos).' Before the 1977 reform, interest rates were set by the Central Bank, and credit was nomally rationed, as real interest rates tended to be negative. The financial wealth of Argentines can tentatively be divided into five main groups of assets: * Currenqr plus demand deposits (aggregate MI) * rime deposits (plazs fijos) denominated in local currency a Dollar-denominated bonds of the government (BoNEx) - Foreign financial assets, mostly denominated in U.S. currency * Government paper denominated in local currency, induding, at times, treasury bills and Central Bank cos This list roughly describes the alternatives open to the public since the reform of 1977. Unfortunately, we do not have reliable data describ- ing asset holdings prior to that reform. Loosely speakdng, however, we can say that the Argentine economy has experienced a sustained pro- cess of demonetization and dollarization in recent decades. The real amount of Ml (non-interest-bearing money) has systematically decreased since 1970, as shown in figure 3.7. (Mhe apparent real increase in M, during 1973-75 was actualy a-result of a price freeze that in mid-1975 brought about the inflationary explosion known as Rodrigazo.) The time path of real M2 (MI plus time deposits) has been quite dependent on the evolution of interest rates paid on time deposits (figure 3.8). CfrTosAltfredRodrIguez 123 Fipge 3.7. Real Value of M1, Argentina, 1960-90 Index (1960 =100) 175 .150~~~~~~~ 125 - -------- - 125 75 25---- 50- , , I 1960 1965 1970 1975 1980 1985 1990 Sotrce: Country case study. Figure 3.8. Real Value of MA, Argentina, 1960-90 Index (1960 =100) 2254 200 - 175 125 ..- 75 - 50 1960 1965 1970 1975 1980 1985 1990 Source. Country case study. 124 Argentina: Fiscal Disequilibria Leaditng to Hyperinflation The debt policy of the government has much to do with the perfor- mance of the financial portfolio of the private sector because the gov- emmnent has gradually become the economy's "borrower of first resort," and as a consequence most of the financial assets of the private sector (except for holdings in foreign exchange) are either directly or indirectly the result of loans to the public sector. Tables 3.7 and 3.8 provide estimates of the domestic interest-earning debt of the treasury and the Central Bank. Practically all of the Central Bank debt is directly held by commercial banks in the form of compulsory reserve requirements (depositos indiomibles) or, at times, as voluntary holdings of the Central Bank's cDs. The commercial banks, in turn, obtain their funds by raising interest-earning deposits from the pub- lic. What we therefore observe in practice is a system in which most of the public's deposits in commercial banks are lent to the Central Bank and are used to finance the fiscal deficit. Part of the deposits of the public may be lent to the prvate sector, but that amount has been gradualy displaced in favor of lending to the Central Bank (see table 3.10, below). It follows that most of the lending capability generated by the public's demand for M2 is absorbed in the form of domestic liabilities of the Centrl Bank. The pressure put on the financial markets by the government debt is best appreciated by evaluating this debt at the commercial exchange rate. Normally authorities try to stabilize the economy by fixng the exchange rate at the level given by the commercial ratea As the credibility of the stabilization plan decreases, interest rates rise, and the stock of government debt tends to rnse in terms of dollars. When the stock of debt, particularly the short-term debt of the Cen- tral Bank, gets out of line with the available reserves, pressures mount against the currency, and a devaluation eventually follows. Normally devaluations are successful in reducing the dollar value of the government debt denominated in australes but are not as success- ful in reducing the interest rates in dollar equivalent paid on the remaining stock. As a consequence, immediately after the devalua- tion the remaining stock of debt continues rising at rates far above the level consistent with a fixed exchange rate, and a new crisis starts to develop. Table 3.9 shows the evolution of the total domestic liabilities of the Central Bank and the total value of M2 (currency plus demand and tine deposits). The ratio between both concepts has oscillated between 59 and 86 percent, depending on how much of the available credit of the commercial banks is absorbed by the Central Bank. A new stabilization attempt, the Prirnavera plan, began in August 1988 with a stock of internal debt of $8.6 billion and ended with $18 billion in March 1989, eight months later. Since GDP stood at about $70 billion, it is dlear that a rate of debt accumulation on the part of the Curios Alfredo Rodrfguez 125 Table 3.7. Interest-Bearing Internal Debt of the Public Sector, in Australes, Argentina, December 1985-November 1989 (mill ions of current australes) Yearand Tresury GCetral montlt debt Bankdebt Total 1985 December 312 3,427 3,739 1986 January 315 3,383 3,698 February 316 3,829 4,145 March 307 4,293 4,600 April 313 4,079 4,392 May 302 4,782 5,084 June 300 5,039 5,339 July 30B 5,316 5,624 August 319 5,765 6,084 September 322 6,250 6,572 October 334 5,616 5,950 November 330 6,742 7,072 December 1,297 7,072 8,369 1987 January 1,666 7,505 9,171 February 1,798 8,035 9,833 March 2,409 8,399 10,808 April 2,724 9,226 11,950 May 2,898 9,139 12,037 June 3,311 9,796 13,107 July 3,630 10,523 14,153 August 5,277 11,853 17,130 September 5,743 11,859 17,602 October 8,040 13,016 21,056 November 8,086 14,146 22,232 December 9,117 16,755 25,872 1988 January . 11,103 17,488 28,591 February 13,152 20,160 33,312 Mardh 12,200 27,318 39,518 April 13,997 31,358 45,355 May 15,282 38,226 53,508 June 18,572 47,956 66,528 July 22,666 60,905 83,571 August 24,503 80,441 104,944 September ZL400 97,737 119,137 October 22,629 107,146 129,775 November 22,635 118,341 140,975 December 23,657 130,517 154,174 (Table confinues on thefolloirng page) 126 Argentina: Fiscal Disequfibria Leading to Hyperinflation Table 3.7 (continued) Yearand Tr,wy Cttrrc1 month debt Bankdebt Total 1989 January 28,434 152,794 181,228 February 45,542 170,508 216,050 March 76,457 202,606 279,063 April 143,719 307,604 451,323 May 207,980 707,948 915,928 June 500,913 1,558,682 2,059,595 July 877,854 2,363,255 3,241,109 August 912,544 3,216,690 4,129,234 September 989,721 3,767,581 4,757,302 October 995,725 4,155,349 5,151,074 November 1,432,058 3,743,756 5,175,814 Sore: World Bank, b rid Debt Tables, various years. public sector of 13 percent of GDP in only eight months was unsus- tainable, and a foreign exchange crisis was inevitable. The collapse of the Primavera plan in February set off a series of devaluations that melted down the debt to $5.7 billion by July 1989. The Bunge and Born plan was then launched by the newly elected authorities with a big devaluation and the announcement of a fixed exchange rate. In the next four months domestic debt rose to $7.9 billion (evaluated at the new exchange rate), mostly because of the interest service of the inherited debt. This time the market was aware of the final effects of rapid rates of debt accumulation and did not wait for debt to reach levels similar to those of the prior stabflization attempt A new crisis in December forced the abandonment of the Bunge and Born plan and the conversion, on January 1, 1990, of all time deposits (and the reserve requirements that backed them) into a ten-year BONEX. As a counterpart of the conversion of the liabilities of commercial banks into BoNEx, al government debt with the banks was also tuned into soNEx, as was most of the government paper in the hands of the public. (The exceptions were cash, demand deposits, time deposits up to about $300, and the existing stock of BoNEX from previous years.) Demand for Money and the Limits on the Inflation Tax Households have a pure transactions demand for real cash balances of local currency, which we can expect to be positively related to real income and negatively related to the cost of holding those balances (the cost being the expected rate of inflation). In this section we CarlosAlfredo Rodriguez 127 Table 3.8. Interest-Bearing Intemal Debt of the Public Sector, in U.S. Dollars, Argentina, December 1985-November 1989 (milions of U.S. doDars at the commnercial exchange rate) Yearand Treosury Gent nd month debt Bank debt Totl 1985 December 389 4,278 4,668 1986 January 393 4,223 4,616 February 395- 4,780 5,175 March 383 5,360 5,742 April 378 4,927 5,306 May 356 5,629 5,985 June 343 5,768 6,111 July 340 5,882 6,223 August 331 5,974 6,304 September 307 5,950 6,257 October 305 5,135 5,441 November 287 5,858 6,145 December 1,069 5,831 6,900 1987 January 1,289 5,805 7,094 Febmary 1,300 5,809 7,109 March 1,563 5,450 7,014 April 1,768 5,987 7,755 May 1,822 5,745 7,566 June 1,941 5,741 7,682 July 1,916 5,555 7,472 August 2,495 5,604 8B099 September 2,337 4,826 7,163 October 2,479 4,013 6,492 November 2,304 4,030 6,334 December 2,579 4,740 7,319 1988 January 2,8727 4,524 7,396 February 3,034 4,651 7,686 March 2,510 5,619 8,129 April 2,437 5,459 7,895 May 2,269 5,674 7,943 June 2,301 5,941 8,242 July 2,348 6,309 8,657 August 2,042 6,703 8,745 September 1,783 8,145 9,928 October 1,851 8,766 10,617 November 1,786 9,337 11,123 December 1,802 9,943 11,745 rkble ontamues on thefollowing page.) 228 Argentina: Fiscal Disequilibria Leading to Hyperinflation Table 3.8 (continued) Year and Treasury Cntral mantle debt Bank debt Toal 1989 January 2,081 11,181 13,262 February 3,143 11,766 14,909 March 4,999 13,248 18,247 April Z811 6,017 8,829 May 1,671 5,686 7,357 June 2,382 7,411 9,792 July 1,546 4,163 5,710 August 1,393 4,911 6,304 September 1,511 5,752 7,263 October 1,520 6,344 7,864 November 2,186 5,716 7,902 Source World Bank, World Debt Tables, various years. assume that the expected rate of inflation equals the actual rate in the current period. Knowledge of the demand for real cash balances is critically important for estimating the effect of deficit-induced money creation on inflation, as well as for estimating the limits of money creation as a means of financng. Given the level and volatility of the Argentine rates of inflation, as well as the unending series of radical changes in monetary policy, one would not expect to find stable demand-for-money parameters over extended periods of time. Rather, we should expect to see the revenue-maxnmizing rate of inflation, and the maximum revenue itself, changing over time. For this reason, we have chosen to esti- mate the demand for real cash balances through two different approaches. 1. The first approach makes use of monthly data for the perod January 1984-June 1988, a period that covers the time from the return to democracy to the onset of the Primavera stabilization plan, ending with the first hyperinflationary episode of 1989. The most important monetary event of this period was the Aus- tral stabilization plan, launched in June 1985. 2. The second approach uses long-term series of annual data (19688) and incorporates a time-dependent dummy variable to capture the possibility of structural change. We found that an additive dummy variable (D77) which takes the value of 1 for the period from 1977 on and 0 otherwise best captures the structural change that took place as a consequence of the financial liberaliz- ation of 1977. Carlos Alfredo Rodrfguez 129 Table 3.9. Domestic Liabilities of the Central Bank and Money Stock, Argentina, December 1985-March 1989 (millions of current australes; end-of-month data) Total stock Cetitral Batik of domestic Yearand doiWestICliabilities motney (M2i. Ralio, mont/s- (1) (2) (1)/(2 1985 December 8,581 10,033 0.86 1986 January 8,365 10,538 0.79 February 8,366 10,971 0.76 March 9,443 11,517 0.82 April 9,319 12,079 0.77 May 10,214 13,044 0.78 June 10,863 13,840 0.78 July 11,447 14,675 0.78 August 12,351 15,336 0.81 September 12,532 15,876 0.79 October 11,615 17,288 0.67 November 13,184 18,176 0.73 December 14,015 20,120 0.70 1987 January 14,867 21,126 0.70 February 14,978 21,964 0.68 March 15,593 23,011 0.68 April 16,619 24,097 0.69 May 17,286 25,765 0.67 June 18,082 27,679 0.65 July 19,717 29,898 0.66 August 21,095 31,530 0.67. September 21,568 34,314 0.63 October 24,881 38,811 0.64 November 26,139 42,602 0.61 December 30,274 48,075 0.63 1988 January 31,864 52,343 0.61 February 33,752 56,984 0.59 March 42,924 65,494 0.66 April 48,575 74,191 0.65 May 58,660 85,827 0.68 June 71,448 102,730 0.70 July 92,428 120,617 0.77 August 117,793 150,250 0.78 September 143,378 175,167 0.82 October 155,276 193,011 0.80 (Table continues on thefollowing page.) 130 Argentina: Fiscal Disequilibria Leading to Hyperinflation Table 3.9 (continaed) T*otal stock entral Bank of domestic Yearand donmsic liabilities money fhPt) Ratio, month (1) (2) (1)/(2) November 174,917 215,552 0.81 December 200,514 259,270 0.77 1989 January 221,780 287,405 0.77 February 243,199 315,141 0.77 March 277,949 368,864 0.75 Source: For M2 data, FnL 1987; for remunerated Central Pank liabilities, World Bank data; for nonremunerated monetary base, iMp, International Financial Statistics, various years. The vstimation on the basis of monthly data precludes the use of income senes to estimate a velocity function. As a consequence, we estimate a demand for real cash balances with the inflation rate as the only explanatory variable in our regressions. To avoid the simul- taneous determination problem (between current-period real cash balances and current-period inflation), we have used a two-stage least squares (TSLS) estimation procedure with inflation lagged up to three periods. TSLS is also used to estimate annual data. The following equations report the regression results for the annual data. Table 3.10 reports the results for the monthly data. (3.5) LV 1.9090 + 4.5484 INF + 0.67087D77 (49.49) (5.83) (9.89) Instruments: INF(-1), INF(-2), LV(-1), D77 MA(l) = 0.925 (t-value - 3.98) Adjusted R2 = 0.96 Durbin-Watson statistic = 1.85 F-statistic = 206 where LV = AO + A INF LV = ln(GDP/M1) V = annual velocity of circulation of M1 GDP = annual nominal GDP Ml = average of monthly holdings of nomial M1 iNF = equivalent monthly inflation rate for the year D77 = financial reform dummy (1 for 1977-88, 0 otherwise). Looking at figure 3.9, we can appreciate the significant changes that have occurred over time in the demand for real M1 as measured Carios Alfredo Rodriguez 131 Table 3.10. Regressions for Monthly Money Demand Variable (a) (b) (C) C 13.213 13.185 8.202 (195) (346) (5.4) L41(-l) - - 0.378 (3.29) [NP -3.01 -2.85 -1.763 (-6-04) (-10.2) (-4.29) MA(1) - 0.50 0.45 (3.4) (2.8) AR(1) 0.308 - - (2.56) Estimated long-run Al coefficient -3.01 -2.85 -2.83 Adjusted R2 0.61 0.67 0.81 Durbin-Watson 1.31 1.92 1.89 F-statistic 42 56 75 Note: Method of estimation, two-stage least squares; instruments, tNF(-1), INF-2), mw(-3), and (only in regression c) LMI(-1); sample period, February 1984-June 1988. Regression (a): LMI = C + AO LM1(-1) +41 INF, where [MI = In(MI/cpi). Regression (b): INF = cnIO1(-1J) -1. by its velocity of circulation. Dunrng 1960-74 velocity remained approximately stable in the range of 6-7. In 1975 velocity began an upward swing that showed no signs of stopping and that took it to a value of 33 in 1988. Unofficial data estimates put velocity around 50 in the second half of 1989, after the hyperinflation of June and July of tat year. The rise in velocity in 1975-76 may have been caused by high price instability during those two years; we have reason to believe, however, that a structural change took place in 1977, when interest rates were totally freed for the first time in decades and t1he public was able to invest in short-term time deposits at market- determined interest rates. This structural change was bound to reduce the real demand for M1, as alternative assets were now avail- able. Our empirical results confirm this assumption; the value for the structural change dummy is 0.67, meaning that there was a 95 percent [exp(0.67) - 1] upward shift in velocity as a result of the financial liberalization. The estimates from the monthly data yield long-run estimates of the semilog elasticity (coefficient Al corrected for the effect of lagged M1) that range between -2.83 and -3.01, implying a monthly revenue-maximizing inflation rate between 33 and 35 percent. The regression using annual data yields a comparable estimate for Al of -4.54, which is compatible with a maximum revenue rate of 21.9 132 Argentina: Fscal Disequilibria Leading to Hyperintflalion Figure 3.9. Annual Velocity of Circulation of MI, Argentina, 1960-90 Velocity 40 35 . ....... 30- 20- 15 ---- 10 ---- 1960 1965 1970 1975 1980 1985 1990 Sourc= Country case study. percent per month. All of these estimates of the maximum revenue inflation rate should be corrected by the monthly real growth rate, if any real growth is to be assumed. Our estimates of the semilog elastidity of demand for M., fall withidn the range of other empirica studies. For example, Kiguel and Neu meyer (1989), using monthly data for the period July 1982-March 1985, find values for Al in the range -2.4 to -3.8 Their estimates of Al for the fixed-exchange-rate period of the tabla cambiaria January 1979-January 1981), when there was a high degree of capital inbility, yields somewhat higher estimates, from -4.9 to -6.0. In order to estimate the possible range for the revenue from money aeation, we have deided to use the results from the regression based on annual data, since this procedure allows for a dtirect estimate of veloity. Computing the Maximum Revenue from Money Creation In this section we derive an estimate of the relation between mone- tary financing of the deficit and the resulting inflation rate. In particu- lar, we derive an estimate of the maximum revenue that can be obtained from the inflation tax before the economy enters into hyperinflation. Consider a demand for high-power money of the following form: (3.6) M- V(1e) =Y p - Q Carios Alfredo Rodriguez 133 where Q = CDP, p = price level, V = velocity, and M = stock of high- power money. By issuing high-power money, the government can finance part of its current expenditures, in the amount (dMldt)Ip. As a fraction of GDP, the revenue from money creation is: (3.7) IR = [-(1/V) (dVldt) + (I + g)JIV'(e). We shall be concerned here with the possibiity of sustainable deficit financing through the inflation tax. Leaving aside the transitory effects of the transition from one equilibrium to another, the sustain- able steady rate of deficit financing through the inflation tax is derived from (3.7) under the assumption that velocity remains con- stant at the level determined by the actual inflation rate and that expectations have been adjusted so that actual and expected inflation rates are identical. Under those circumstances, the steady-state sus- tainable revenue from inflation becomes: (3.8) IR = (I + g)IV(l). There is a maximum amount of revenue that can be raised with the inflation tax before generating hyperinflation. This amount corre- sponds to the solution of the maximum value for the equation: maxi- mum inflation tax = maxU + g)IV(l) Attempts to raise a higher reve- nue than this maximum will require ever-increasing rates of money creation and inflation Assume the following form for the velocity fumction: (3.9) log(V) = Vo + b - . In terms of the standard form for velocity shown in (3.9), the steady- state revenue from infiation (IR) is: (3.10) IR (I + g)IV() = (I + g) exp(-V0- bl). The function IR takes a maximum when (3.11) d(IR)(dl = exp(-V0 - bl) - (I + g) - b exp(-VO - H) = 0. The solution to this expression yields the maximum revenue inflation tax: (3.12) IMAX=(Ilb)-g which is the continuous time rate of inflation for the period over which velocity is defined. The corresponding maximum revenue is derived as (3.13) nRLAx = [exp(-VO-1 + b-g)]Ib. It is clear from this analysis that (3.10) is valid provided the expres- sion in parentheses is less than IRMAX. Any deficit in excess of uiMAX cannot be financed through the inflation tax because equation 3.10 134 ArgEntina: Fiscal Disequiibria Leading to Hypeinnfation will not have a solution-in other words, ever-increasing rates of monetary expansion and inflation will be necessary to finance the deficit, and the system will enter into hyperinflation. When annual data are used, the estinmate of money demand yields the following expression for the sustainable revenue from the infla- tion tax as a percentage of annual CDP: (3.14) IR = 12- (NF + g) exp(-2.5798 - 4.5484-NF) where n-f and g are the corresponding monthly inflation and growth zates and ER is the armual revenue from the inflation tax as a fraction of annual income. Assuming a real growth rate of 2 percent per year, g takes a value of 0.00165 per month. With this value forg, the rate of monthly inflation that maximizes RIYis equal to: (3.15) INF* = (1lb) - g = 0.2184 This maximum revenue rate is 21.8 percent per month, or 966 percent per year. The assocated maximum revenue from money creation is 7A percent of GDP, and velocity at this inflation rate takes the value of 35.5- To illustrate the workings of the maximum inflation tax, assume a COP of $70 bfilon (about the level of Argentina's GDP in 1989). With velocity at about 36, money demand is $1,945 nillon. At a monthly inflation rate of 21.8 percent, revenue from inflation is $424 million per month (0.218 x 1,945), or $5,088 million per year, which is equiva- lent to 7.3 percent of annual GDP. It should be noted that as the inflation rate approacies uqF*, the revenue finction becomes increasingly elastic with respect to the inflation rate. This means that small changes in revenue require large changes in inflation. As revenue reaches the maximum level, there is no increase in the inflation rate capable of generating any sustainable increase in the rate of revenue. Figure 3.10 -shows the relation between revenue and inflation derived from our estimated revenue function, assuming a growth rate of 2 percent per year. The figure shows little gain and much cost from raising inflation above 10 per- cent per month. At 10 percent per month the revenue is about 6 percent of annual Gop. Raising the maximum amount of extra reve- nue, 1.3 percent, requires that the inflation rate increase from 10 to 21.8 percent per month. Therefore, at the margin, the extra 1.3 per- centage points of GDP in additional revenue requires an increase in the annual inflation rate from 213 to 966 percent. Raising the first 6 percentage points of revenue requires only 213 percent inflation. In conclusion, our estimate for the money demand eqvition for Argentina yields an estimated maximum-revenue inflation tax of about 22 percent per month and an estimated maximum revenue of CaIrlos Alfredo Rodnrguez 135 Figure 3.10. Annual Revenue from Inflation and the Monthly Inflation Rate, Argentina Annual inflation revenue (percentage of GDUT 7.5 5.0- U~~~~~~~~~~~~~~~~~~~~~~~~~ t 0 10 20 30 40 50 60 Monthly inflation (percent) SowCountrycase sy 7.3 percent of GDP. The revenue estimate should be modified if all of the tax on Ml is not collected by the monetary authorities. As of December 1988 the monetary base was 52 billion australes and M1 was 53 billion australes, suggesting that our estimate of the revenue from inflation is approximately correct, since the M1 multiplier of the monetary base is about unity. The Process of Determining Public Debt and the Interest Rate There are two fnancial markets in Argentina, the formal market and the informal market. The informal market consists of the mercdo intermpresario (iterfirm financial market), in which firms borrow and lend among themselves through a set of institutions called "money- market desks." No record of the volume of these transactions exists, although newspapers do quote the interest rates at which transac- tions are settled. There is no accepted estimate of the size of this market. Most of the transactions are handled in one to seven days. The formal financial market is concentrated in the commercal banks and flnanceras. Both are subject to Central Bank regulation, under which deposits are guaranteed and reserve requirements are 136 Argentina: Fiscal Disquilibria Leading to Hyperinflation seL The foremost instrument on the borrowing side of the banks and financieras is the interest-earning time deposit. From time to time a variety of other istruments have been offered to the public, includ- ing a wide range of indexed deposits (indexed to the dollar, the cpi, or components of the price index). Time deposits are sold to the public and traditionally bear a maturity ranging from seven to thirty days. The average maturity for deposits has rarely exceeded fourteen days and in recent years has been dose to seven days. During 1989, follow- ing several episodes of long weekends coupled with forced bank holidays, it was not uncommon for 90 percent of the time deposits of the system to come due on the same day. Banks use their deposits to grant credit to the private sector or to acquire Central Bank assets. In 1989 about 80 percent of all assets of the commercial banks consisted of liabilities of the Central Bank. Reg- ulations on how those Central Bank liabilities are remunerated have changed over time, but on average the liabilities have paid an interest rate equal to the average cost to the commercial banks of raising the funds plus a spread that covers operating costs and allows for a profit. In some instances banks seem to have been interested in hav- ing their liquid funds absorbed by the Central Bank through increases in remunerated reserve requirements. In a country like Argentina which is subject to many financial uncertainties, it may actually be safer to lend to the entity that regulates the industry (the Central Barik), particularly since such lending is profitable. (For further infor- mation on the dynamics of the financial system that led to the hyper- inflation of 1989, see Almansi and Rodriguez 1989 and Fernandez 1989.) From this description we can conclude that for all practical pur- poses commercial banks raised deposits in order to lend them to the Central Bank. The result was a grossly distorted financial system in which, as of December 1989, about 2,500 bank branches with 140,000 employees administered $4,700 miion worth of deposits, or $33,600 per employee. After the meltdown of January 1990, total bank deposits sank to $1,384 million, but approximately the same number of employees were trying to retain their jobs, so that there was about $9,885 worth of deposits per bank employee! About half the bank deposits and half the employees were in state banks. The obviously needed restructuring of the Argentine financial system has not yet taken place, even though the system is grossly oversized. In such a system it is very difficult to ascertain how the interest rate on deposits is determined. One might think that the marginal 20 percent of private borrowers would generate a tendency for rates to approxnmate the productivity of investment, but this is not so. Most private creditors borrow on a daily basis not for investment purposes but to finance temporary financial disequilibria by overdrawing their Cauros Aifredo Rodriguez 137 checking accounts. The interest rate appears to be determined by the short-rmn liquidity available in the system. Short-Run Analysis We have found a significant relation between the stock of time deposits outstanding and the real interest rate that these deposits yield. The general rule for banks in determining interest rates is to set the rates at the level required to roll over all deposits (principal plus interest), unless the Central Bank intervenes by providing the required cash so the banks can reduce the outstanding level of deposits. In terms of the standard money-supply multipliers, the monetary base of the Central Bank (its liabilities) is equal to the sum of reserve requirements: (3.16) MB = a1Mh + a2D where a,, the average reserve requirement on M1, is very dose to unity since the reserve requirement on demand deposits has tended to equal unity. The coefficient a2 of reserve requirements on interest- earning deposts (D) is also dose to unity as a result of the Central Bank's policy of gradually absorbing most of the lending capability of commercial banks. The demand forM1 is assumed to take the form: (3.17) M= pL(pie) where pie is expiected inflation. The demand for real-interest-earning bank deposits depends on the expected real return to be earned: (3.18) Dip = d( - pie). The growth rate of the total monetary base, in turn, equals: (3.19) d(MB)Idt = (i + s) 'a2 D + DEE where s is the spread paid over the cost to the banks of raising the deposits and DEF is the financing needs of the nonfinancial public sector. Equations 3.16 through 3.18 are not enough to determine the four endogenous variables M1, D, p, and i. The fourth missing equation is the key to determining the interest rate. This equation depends on the structure of the operations of financial markets. If there is still enough of a link with the real sector through bank credit, we can assume that the real interest rate is determined bW the marginal pro- ductivity of capital. Given inflationary expectations, this condition determines 4, and equations 3.16-3.18 wfll determine the remaining variables M1, D, and p. 138 Argentina: Fisal Disquwlbria Leading to Hyperinflaion If there is perfect mobility of capital, the real interest rate will be determined by the external rate, and the nominal interest rate will therefore also be exogenously determined, given inflationary expecta- tions. In recent years (198249), however, the currency in general has not been convertible, and we do not observe any close link between the productivity of capital and the cost of bank credit. For practical purposes, most bank cedit goes to the Central Bank, which simply pays the cost of raising the credit. A missing element in this descip- tion is the Central Bank's method of determnining the level or the cost of its interest-earning debt. The growth rate of this debt depends on the interest rate and the shares of remunerated and nonremunerated liabilities in the total monetary base. We hypothesize that the Central Bank at times I-as aimed at controlling the ratio of remunerated to nonremunerated liabilities in the monetary base to ensure that its total liabilities grow at the desired rate. This means that at any point in time the total amount of interest-earning deposits in the system is fixed in nominal terms. Since the public can freely shift from D to M1, the constancy of D means that banks must offer whatever interest rate is needed to induce the public to roll over all of its deposits (prncipal plus interest) at a given point in time. The alternative is to try to keep the nominal interest rate constant by allowing depositors to shift freely between M1 and interest-earning deposits. Consider the tradeoff faced bv the Central Banlk when the demand for interest-earning deposits falls. The bank has the option of keeping the interest rate constant by allowing depositors to convert all their excess supply of D into M1. If that is done, however, the price level must jump because the real demand for M1 has not changed. There- fore, if interest-earning deposits fall and the Central Bank feeds the run by substituting a nonremunerated for a remunerated monetary base, the price level increases (or, in other terms, the currency is devalued in the black market). This behavior characterizes the periods df low interest rates and high black-market premaiums. The alternaive is to keep M1 and D constant and to allow the nominal interest rate to adjust so that the run is stopped by a higher return on deposits. The higher real interest rate reduces pressures in the parallel foreign exchange market, and the black-market premium initally falls. In this case there is no effect on M1. (We have assumed that M1 depends only on expected inflation; otherwise we would have to make some minor adjustments to the analysis that follows, but the thrust of the analysis would remain the same.) Since the nominal interest rate has increased, the growth rate of the monetary base is now higher (according to equation 3.19)- This behavior charac- terizes periods of high real interest rates and low black-market pre- miums for the currency. This situation is, however, unsustainable because the monetary base starts growing at rates inconsistent with Carlos Alfredo Rodriguez 139 pnce stability, and the black-market exchange rate starts to depreci- ate. Eventually the system breaks down as the Central Bank is forced to devalue the official exchange rate and to allow M1 to increase in order to reduce interest rates and the pressure of the quasi-fiscal deficit. The inherently unstable system implies wide oscillations in real interest rates and the real exchange rate as depositors try to anticipate sudden devaluations (the preferred instrument for melting down the quasi-fiscal deficits) by exchanging their deposits for cash with which to buy U.S. dollars in the black market If the government tries to stop the portfolio shift, it must resort to higher interest rates to force the rollover of the deposits. This mechanism reduces the short-run pres- sures on the currency but increases the rate of growth of the mone- tary base and therefore generates expectations of even larger devalua- tions. The run feeds on itself until authorities give in to pressures and devalue. Empirical Estimates The magnitude of the changes in the interest rate needed to accom- modate fluctuations in the demand for deposits is bound to depend on the interest elastcity of such demand. The more inelastic the demand is, the larger should be the inaease in the interest rate neces- sary to induce the rolover of the deposits in the face of an exogenous fall in demand. We have attempted to estimate this elasticty using data for the period following the financial reform of 1977, when a fundamental structural change allowed the free-market determina- tion of interest rates. The result was an unprecedented change in the ratio of M2 to M1 (figure 311). The increase in the ratio of interest- earning deposits to M1 began in 1977 and was completed by approxi- mately miid-1980. Since then, this ratio has oscillated in response to the relative returns on both assets. Regression (a), presented in the first column of table 311, shows the estimates of the real demand for interest-earing deposits using the ordinary least squares (OLs) estimation method and monthly data covering the periodJanuary 1978 through December 1988. Regression (b), shown in the second colun of table 3.11, estimates the same demand for the subperiod January 1984 through December 1988. We assume that the real demand for deposits depends on the nominal interest rate for the current month and on the expected inflation rate. The regressions shown use the current-period inflation rate as the measure of the expected inflation rate. No significant changes were found when actual inflation figures for the next period were used. The high value and significance of the coefficient for the lagged endo- genous variable strongly indicates a slow adjustment process. Regres- 240 Argentina: Fiscal Disquilibria Leading to Hyperinflation Figure 3.11. Ratio of Interest-Bearing Bank Deposits to MI, Argentina, 1961-90 Ratio 5 4- 2 -- - -----_ 2 - _:__ _- -- 1960 1965 1970 1975 1980 1985 1990 Source: Country case study. siaon (a) uses the data for longer periods and regression (b) uses the data for shorter periods. Regression (c), shown in the last column, is calculated for the longer period, using instrumental variables, to correct for the problem of simultaneous detemination of interest rates and inflation rates. In all cases the sign of the nominal interest rate and inflation rate is what was theoretically expected: the nominal interest rate is positive and the inflation rate is negative and has an absolute value equal to that of the interest rate. The results confirm the assumption that the demand for real deposits depends on the real interest rate paid on them. The coefficients are signifiant at the 1 percent confidence level and, when the oLs estimation method is used, do not differ in absolute value between the two periods, indicating that no significant structural change took place in the post-financial-reform years 1978-88. The estimates in regression (c), measuring the longer period with the use of instrumental variables, remain highly significant, but their absolute values for the interest elasticity are higher than in the ois estimate. In this case the semilog interest elasticty with respect to the monthly real interest rate is about 1.1 in the short run and 11i0 in the long run. Thus a 10 percent decrease in the demand for deposits would be compensated by an increase in the monthly interest rate of 9 percentage points in the short run and 0.9 percentage points in the CarlosAlfredo RoddrTguez 147 Table 3.11. Demand Estimates for Interest-Bearing Deposits Variable (a) (b) rC) C 0.981 0.758 0.991 (4.6) (2-1) (4.0) Al 0.929 0.946 0.928 (60) (35) (51) A2 O.t.'9 0.524 1.10 (6.0) (3.92) (3.59) A3 -0.765 -0.659 1.09 (-9.6) (-6.2) (-4.52) Adjusted R2 0.977 0.964 0.973 Durbin-Watson 1.72 1.87 1.83 F-statistic 1,876 530 1,601 Number of observations 132 60 132 Note: For regressions (a) and (b): method of estimation, ordinary least square; instru- ments, log(DIP) = C + Al - log(DIP)(-I) + AZ *NT + A3 - INF, where D = A2 -Ml (interest-bearing deposits from FiEL databank), P = consumer price index, and INT nonoinal interest on deposits, monthly basis (IMF. Interational Fnncial Stafistics, var- ious years, line 60L); sample period for regression (a), January 1978-December 1988; for regression (b), January 1984-December 1988. For regression (c): method of estima- tion, two-stage least squares; instruments, log(DP(-1), INFt(-), INF(-2), irr(-l), nc(-2); sample period, January 1978-December 1988. long run. Unfortunately, the instrumental variables technique did not yield satisfactory results for the period 1984-88, probably because the instruments used could not capture the sharp fluctuations experi- enced in both nominal interest rates and inflation rates. In the system described above, the real interest rate on time deposits of the institutionalized financial system depends on a deli- cate equilibrium between expectations of meltdowns and the govern- ment's need to refinance its debt with the financial system-a debt that is the counterpart of most of the interest-earning deposits at the commercial banks. If we assume that the govemment controls the rate of devaluation and can always outsmart the market by devaluing by more than was expected, we arrive at the conclusion that the government determines the ex post real rate of interest to a certain extent. The ex post real interest rate in the financial system is basically determined by the only significant creditor of the system: the Central Bank. This determIination is not made on a monthly basis, to be sure, but through the meltdowns that systematically take place. Why do depositors agree to remain in such an uncertain asset situa- tion? One reason is that they are systematically tempted with attrac- tive ex ante real interest rates. Devaluations happen inevitably, but they are usually instituted by a newly appointed economics minister-the previous minister having been just fired because of his 142 Arg1;nmina: Fiscal Disequiibria Leading to Hyperinflation now-obvious policy mistakes. Why people have believed in each new minister and have accepted the rollover of the remaining deposits in spite of just having been melted down remains unexplained, but in fact the credibility of new ministers is decreasing over time as the same pattern is repeated over and over again. Consider the initial nominal interest rates after the last full-fledged stabilization plans, all of which were based on a large devaluation that, it was promised, would be the last ever. The Austral plan (June 1985) devalued by 40 percent; immediately after the devaluation, the interest rate was about 7 percent per mont. The Priravera plan (August 1988) devalued by 24 percent; the subsequent interest rate was about 10 percent per month. The first part of the Bunge and Born plan devalued by 200 percent; the initial interest rate was 17 percent per month. The second part of the Bunge and Born plan (December 1989) devalued by 54 percent; the initial interest rate after the deval- uation was about 60 percent per month. Fmally, the Erman plan (January 1989) made mandatory the conversion of all time deposits into ten-year dollar-denominated government bonds that started trading at about 30 to 40 percent of par value, and the initial interest rate was about 100 percent. In the days prior to the announcement of the Erman plan, monthly interest rates reached levels of 600 percent per month for large depositors because of the expectations of a forth- coming meltdown. After the meltdown, which took place on January 2, 1990, nominal interest rates fell only to 100 percent per month-an indicator of the market's lack of confidence in the success of the new plan. It is dear that the institutionalized credit market in australes has no significant role in generating new credit. All of the poteiLtial credit tends to be absorbed by the govenment as a result of its need to finance deficits. Since 1985 or earlier, the prmary role of the market has been in refinancing the stock of public debt that has its counter- part in the stock of bank deposits. A change may have taken place after March 1990, when the government announced yet another effort at fiscal adjustment and a reform of the Central Bank's charter was undertaken, for the second time in less than a year. It is still too soon to judge the success of these efforts. An additional source of aedit for the government has been the iSSa- ance of different types of treasury debt and Central Bank instruments that are indexed either to prices or most commonly, to the U.S. dollar One such govemment paper is the uoNEx series of bonds, first issued in 1980. BoNEx bonds pay UIBOR rates and have a ten-year maturity. They usually trade below par, in spite of the fact that so far all BoNEx bonds have been regularly served. The internal rate of return on BoNEC bonds (if held to maturity) is usually taken as a measure of the marginal cost of borrowing in U.S. dollars for Argentina. Carlos Alfredo Rodriguez 143 BONEX bonds are in high demand by Argentines; they are widely held as collateral for loans in the informal financial market. Since BONEX bonds can be legaUy traded in secondary markets for either australes or U.S. dollars, they are the mechanism through which firms and individuals can buy or sell U.S. dollars legally. (Holding U.S. dollars is legal; trading them in the free market rate has usually been illegal.) Because of the wide acceptability of BONEX bonds, their internal rate of return (iRR) provides a good measure of the equilib- rium U.S. dollar interest rate in Argentina; other U.S. dollar or indexed rates tend to use the BONEX rate as the preferred reference rate. Instruments that are subject to devaluation risk or have dubious possibility of collection pay rates higher than the iRR on the uoNEx The BoNEx rate puts a floor under other U.S. dollar rates in the infor- mal system; no one in Argentina will lend at less than this rate. The ERR on the BONEx has shown significantly less variability than the ex post realized U.S. dollar rate on austral deposits (figure 3.12). As of January 1990 the stock of outstanding BONEX (series issued from 1980 to 1987) was stable at about $2,200 million, an amount larger than the U.S. dollar value of M1. The EoNEx rate sets the reference rate for an even wider informal market for U.S. dollar- indexed operations. The BONEx rate is not affected by exchange rate expectations (it is already set in U.S. dollars), and it is the last asset that the market expects to be melted down by default. Changes in the BoNEx rate can be assumed to be determined by changes in world U.S. dollar rates and in the creditworthiness of the government. The creditworthiness of the govenmuent is not something to be taken for granted. The government stopped serving its external debt with commercal banks in 1987. The debt traded at less than 20 per- cent of par in 1988 and fel to 12 percent in early 1990. The BONEX trades at parities above 70 percent because it has attained a sort of preferred status at the moment of servicing the debt. Other sup- posedly preferred instruments have already been subject to forced refinancing and melting down; an example is the LEDOL, a 90-day dollar-denominated bil issued by the Central Bank in August 1989. LEDOL was refinanced compulsorily with a ten-year treasury bond that immediately started trading at 30 percent of par. In January 1990, before the treasury bond was even issued, it was replaced by the new issue of BONEX 1989, which was used to purchase most of the out- standing austral-denominated government debt (and the plazo fijos) under the Erman I plan of January 2, 1990. The best measure of the creditworthiness of the govemrment is its ability to generate a fiscal surplus in relation to its level of indebted- ness. As has already been seen, the government has systematically run a primary deficit in recent years. As a result, the goverunent was eventually forced practically to default on its external debt, a process Figure 3.12. Ex Post Dollar Rate on Austral Deposits and BONEX Rate, Argentina, 1986-August 1989 Percent 50 40 - .................................................................... ... 30 ..........................- -.................................. 20 - . ............................. .............................. -10 -. .......................................................... .......................... .. ..... -0 -. .... .. . ;. -30 ....... . ... ..................... . . ..... ... 40 . 1986 1987 .1988 1989 Rate an austral deposits BONEX rate Sourc= Country case study. that began graduaUly in 1982. The service of BONEx actually competes with a1l other instruments of the internal debt. We therefore suggest the existence of a positive tradeoff between the intemal rate of return on the BoNK and the stock of internal debt of the government. In addition, we would expect that the BONEX rate is related to the oppor- tunity cost of exteral funds as measured by a risk-free rate such as prime or LIBOR. In defining the relative value of the stock of government debt, three deflators come to mind: the price level, the offical exchange rate, and the free-market exchange rate. We have found that the official exchange rate is the deflator that yields the best econometric results in the sense that the real stock of debt (measured in U.S. dollars at the official exchange rate) is the one measure best associated with the interest rate on BONEX. One reason may be that much of the stock of government debt was generated by the Central Bank as a result of attemnpts to maintain constant the nominal value of 'the offiial exchange rate. The U.S. dollar value of the stock of debt at the official exchange rate is therefore a measure of the pressures against the sustainability of that rate. The higher is the stock of government debt Carlos Alfredo Rodr(guez 145 evaluated at the official exchange rate, the less likely it is that the current set of policies can be maintained, and the risk premium of lending to the government increases. The regression results assessing the link between the BONEx rate and the stock of internal government debt are shown in equation 3.20. The regression estimates, using the OLS method, indicate a strong effect of the level of government debt on the BONEX rate. In the long run a 10 percent increase in the U.S. dollar value of the govern- ment debt results in an increase of 3.2 percentage points in the annual BONEX rate. (3.20) IRRBONE X -9.13 + 0.797iRRBONFX (-1) + 6.53 log (DGCOM) (-2.5) (11.7) (3.37) Adjusted R2 = 0.84 Durbin-Watson statistic = 1.66 Durwin h = 1.23 Sample period = April 1986-September 1989 where DGCOM is the dollar value of the stock of internal public debt evaluated at the commercial exchange rate and [RRBONEX iS the annual equivalent of the internal rate of return on B1ONEX. Correction for first- order autocorrelation did not yield a statistically significant AR(1). The Durwin h test also shows no significant autocorrelation. The external interest rate, measured by the monthly U.S. prime rate, was not significant in previous regressions and was not indluded here. A possible explanation for the lack of significance of the extenal interest rate might be that since 1982 Argentina has not had access to external credit and therefore a link between domestic and external interest rates should not be expected. The fact is, however, that Argentines do hold an estimated $30 billion or more in foreign assets (as against $2 billion-$3 billion for M1 in local currency). Since Argen- tines can freely shift the composition of their portfolio between for- eign assets and local paper, we should expect some association between the local and external interest rates. We have no doubt that such a relationship exists, but it is difficult to capture statistically for a short period, such as the one analyzed here, especially since the U.S. prime rate did not experience any significant variation in comparison with the sharp oscillations in the BONIEX rate that were induced by changes in the market's evaluation of the risk of lending to the gov- ernment (see figure 3.13) Interest Rates and Inflation in the Steady State In this section we turn to the long-run steady-state tradeoff between interest rates and government debt policy. In order to focus on the 146 Argentina: Fiscal Disquilzbria Leading to fHyperinflation Figure 3.13. Stock of Government Debt (Official Dollars) and BONEX Rate, Argentina, 1986-August 1989 Deviation from sample average 4- 3-. . ... .. ............................. ....................... - -. ................... .. ...... -1 - t ~~~~~~~................. .............. ..................... ..... .. ., -2 1986 1987 1988 1989 Governnent debt soNEx rate Sow-rc: Country case study. essential elements of the process-in particular the fact that interest is being paid on money by printing more money-we will assume that there is a 100 percent reserve requirement on both remunerated (D) and nonremunerated (M1) money. Under this assumption, we can write the monetary base, MB, as (3.21) MB = M1 + D. We also assume the following behavioral relationships between the demand for the two kinds of monies and the inflation rate (pi) and nominal interest rate: (3.22) Mi = P L(pz) L' < O. (3.23) D =PF(i-pi) F' > 0. The change over time of the monetary base is given by: (3.24) d(AB)Idt = i- D + def Carlos Alfredo Rodrfguez 147 where the variable def represents the nominal budget deficit and i D represents the Central Bank's remuneration of interest-beanrng deposits D. The real budget deficit, g. is given by: (3.25) g = def/P. Finally, we use the Fisher equation to define the real interest rate, R, as: (3.26) R = i-pi. In the steady state the following equality must hold: (3.27) [d(MB)IdtJ - (1fMB) = pi. Dividing (3.24) by MB and using (3.25), we can write: (3.28) [d(MB)Idtl (11MB) = i DIMS + g - P/MB. From equations 3.21, 3.22, and 3.23 we can express the real mone- tary base as: (3.29) MB/P = L(pi) + F(R). Hence, using equations 3.23, 3.26, 3.28, and 3.29, we can rewrite the steady-state equiibrium condition, (3.27), as follows: (3.30) L(pi) * pi = F(R)* R + g. In an economy In which the real sector, or the international capital market, determines the real interest rate, the equilibrium condition established by equation 3.30 determines the inflation rate. The equi- lbrium inflation rate is the one that delivers the inflation-tax revenue required to pay for the real budget deficit, g, and the real service of the Central Banks debt (the real quasi-fiscal deficit), F(R) R. There are, of course, real interest rates for which equation 3.30 has no solution. This corresponds to the situation in which the amount of resources required by the sum of the two deficits exceeds the maxi- mum stationary inflation tax. There will normally be many real interest ratec for which there are two solutions for equation 3.30. For these cases we assume that the monetary authority chooses the lowest-inflation-rate (or, the efficient) solution. This assumption is important because, as we will see later, it determines the sign of the equilibrium relationship between the real interest rate and the inflation rate. Differentiating equation 3.30, we otbserve that along the steady-state equilibrium relationship between r and pi we must have: (3.31) d(pi)ld(R) (F + R - F')1(L + pi- L'). As long as the economy stays on the efficient side of the inflation-tax Laffer curve, the right-hand side of equation 3.31 must be positive- 148 Argentina: Fiscal Disequilibria Leading to Hyperinflation that is, an inarease in the real interest rate corresponds to an increase in the equilibrium inflation rate. It is obvious from our model that in an economy in which the real sector does not determine the real interest rate, the financial sector alone cannot determine both the real interest rate and the inflation rate. The financial sector provides us with one equation, (3.30), but we have two unknowns, R and pi. The Argentine economy seems to be precisely such an economy. On the one hand, the supply of loan- able funds comes fron deposits, both remunerated and nonremune- rated, that people hold not as an altemative investment but simply for liquidity reasons. On the other hand, the demand for loanable funds is related merely to the short-term liquidity needs of business firms. Saving and investment in Argentina-the capital market-no longer function in local currency. As is descnibed in this chapter, the Argentine monetary authorities tried to use the apparent degree of freedom provided by the lack of connection between the financial market and the real sector during the 1982-89 period to manipulate the real interest rate and so control the composition of the Central Bank's liabilities. In particuar, the authorities tried to prevent the expansion of the Central Bank's non- remunerated liabilities by raising the real interest rate as much as required by the market to hold remunerated liabilities. TIis was done in the belief that it is only the expansion of the nonremunerated liabilities that causes prices to rise over time. The result of this mone- tary policy has been a sort of "unpleasant monetarist arithmetic." By raising the real interest rate the Argentine monetary authorities imcreased the required inflation-tax revenue, thus making necessary an increase in the equilibrium inflation rate. By choosing the composition of its liabilities between remunerated and nonremunerated debt, the Central Bank chooses a point in the tradeoff given by equation 3.31. This can be seen by dividing equation 3.23 by 3.22 and denoting by a the ratio of remunerated to non- remunerated Central Bank debt. The relationship between a, pi, and R is given by: (3.32) a = F(R)IL(pi). Assumiing that the equilibrium is at the efficient side of the Laffer revenue curve, equation 3.30 describes an upward-sloping relation- ship between R and pi. For a given or, equation 3.31 descnibes a downward-sloping relation between R and pi. The intersection of both schedules determines the unique steady-state values of R and pi. A higher a is associated with a rightward shift in the downward- sloping schedule (equation 3.31) and therefore with a higher R and pi in the new steady state. Carlos Alfredo Rodriguez 149 The nature of the tradeoff between remunerated debt and inflation is now dear. In the short run, increasing a (reducing liquidity by issuing interest-eaming debt) helps reduce pressures on inflation. In the long run, the rate of nominal monetary expansion must be higher in order to finance not only the previous deficit but also the real interest service on the larger remunerated debt. As a consequence, the inflation rate, as well as the real interest rate, must be higher, since the real stock of debt is higher and depositors in the banks have to be induced to hold the extra deposits with which to finance the extra government debt-whether these deposits be called remune- rated reserve requirements, treasury bills held by banks, compulsory bank investments, or something else. The External Effects of Public Sector Deficits The external effects of public sector deficts can be analyzed in a two- step process, taking, first, the effects of the fiscal deficit on the level of aggregate spending and therefore on the trade balance defict, and, second, the effects on the real exchange rate of the changes in aggre- gate spending, as measured by the trade balance. Additional side effects are caused by-portfolio shifts induced by changes in the rate of inflation that result in changes in the desired rate of accumulation of foreign assets and therefore in the trade balance. Finally, the rate of government spending may affect the real exchange rate if the govern- ment's propensity to consume nontradable goods is different from that of the private sector. The Theoretical Framework for the Determination of the Real Exchange Rate Consider an economy with three broad aggregates of goods: export- able goods, import-competing goods, and nontradable goods, with nominal prices Px, Pm, and Ph, respectively. The concept of the real exchange rate (RER) is intended to be a measure of some aggregate of the nominal prices of tradable goods (Px and Pm) in terms of nontrad- able goods (Ph). In general, however, this economy must have two relevant relative prices: PxlPh and PmIPh, which we shall term the export real exchange rate (RERx) and the import real exchange rate (RERM)- As relative prices, RERX and RErvm are endogenously determined and so cannot be considered policy variables. For a given (equi- proportional) change in the equilibrium values of the RERs, however, an accommodation can be made in the nominal exchange rate so that the RERS get to their new equilibrium values without need for varia- tion in the domestic prices of nontradable goods. In addition, a nomi- 150 Argentina: Fiscal Disequilzblia Leading to Hyperinflaon nal exchange rate policy that indexes this nominal variable to some aggregate price level may force the R.R measures to dep,;t from their equlibrium levels for long periods of time. It is important, therefore, to have available a structural model of determination of the equilb- rium values of the Ria so that nominal exchange rate policy does not force such a departure. (Earlier studies on real exchange rate deter- mination in Argentina indude Cavallo and Pefia 1984, Diaz-Alejandro 1981, and Rodriguez and Sjaastad 1979.) The model used here follows the one presented in Rodriguez (1989a). This model assumes that for internal balance to be achieved, there must be an equilibrium relationship between the three nominal prices and the rate of nominal spending. Such a relationship can be interpreted as the condition for equliium in the market for nontrad- able goods. In functional form, the equilibrium can be expressed as: (3.33) Dh(Ph, Pmn, Px) A - Sh(Ph, Pm, Px) Y = 0 where A is the nominal rate of absorption, Y is nominal income, Dk(-) is the fraction of total absorption of goods devoted to the purchase of nontradable goods, and Sh is the fraction of the value of the output of nontradable goods in total nominal CDP. In equation 3.33 we have assumed that supply of and demand for nontradable goods are homogeneous of degree one with respect to the levels of nominal output or absorption, respectively. Since equation 3.33 must be homogeneous of degree zero in aU nominal variables, we can deflate by Ph to obtain: (3.34) Dh[PmIPh, (PxIPm) (PmiPh)I - t -rs) = SJ[PmlPh, (PmlPx) -(Px/Pm)] or (3.35) PERM GI(PmIPX), ts] where ts = (Y - A)(Yis the trade balance surplus normalized by GDP. In logarithmic form, equation 3.35 can be expressed as: (3.36) log(RERM) = Co + w - log(PmIPx) + z - ts. Since log(RRm) = log(PmlIPh), we can interpret (3. 36) as the equation that determines the equilibriuma value of Ph given the exogenous values of Px, Pm, and ts. Both Px and Pm are determined by foreign prices and commercial policy, whereas ts is determined by macro- economic variables related to the equlibrium rate of foreign savings (to be analyzed later). From this perspective, we expect w to be posi- tive and between 0 and 1 because it is the elasticity of Ph with respect to an increase in the nominal price of exports when fs and Pm are held constant. (For detailed analyses of these relationships, see Dormbusch 1974, Harberger 1988, and Sjaastad 1980.) Carlos Alfredo Rodrfguez 151 Since Ramw = R (X CPxlPm), it follows that equation 3.36 can also be expressed as: (3.37) log(RERx) CO + (1 - w) [log(PxIPm)] + z ts. In general, we expect the parameter z to be positive because it repre- sents the effect on Ph of an inaease in absorption in relation to income; as some extra spending falls on nontradable goods, Ph rises, and thus RERX and RERM must fall. In terms of (3.34), as ts rises, absorption falls and so does Ph; it follows that z must be positive. It is usual to refer to the real exchange rate as the relative price of some average pnce of tradable goods. Assume that this average is formed in the following way: (3.38) log(PTA) = a * log(Px) + (1 - a) - log(Pm). The average real exchange rate would then be: (3.39) log(AvRER) = a -log(Px) + (1 - a) - log(Pm) - log(Ph). Since log(Ph) = -CO + w - log(Px) + (1- w) - log(Pm) - z-ts, we can substitute this expression into equation 339 to obtain: (3.40) lOg(AVRn) = +z - ts + (a -w) log(PxlPm). It follows from equation 3.40 that if the aggregation parameter, a, is chosen to be identical to the structural parameter, w, the AVRE will not depend on the terms of trade or commercial policy. In general, depending on the aggregation weights used, an average RER could depend on the terms of trade in any conceivable way. Assume now that the government also demands nontradable goods according to Gh - Ag, where Gh is the share of government expenditure that consists of nontradable goods and Ag is total spend- ing by the government. Market equilibrium is now given by: (3.41) DhAp +Gh-Ag=Sh where Ap is private sector absorption. Defining total absorption as A = Ap + Ag, it foliows that all of the prior analysis is still valid if Gh is identicaly equal to Dh-for example, if the demand of the government is identical to the demand of the private sector If Gh is greater than Dh, an inaease in govern- ment spending, for a constant total absorption, implies that the demand for nontradable goods will rise and that therefore the .ER must fall (there has been a shift in the composition of absorption toward the sector with the higher propensity to consume nontradable goods). Conversey, if Gh is less than Dk, an increase in government spending, for constant total absorption, wil mean a higher Rn 152 Argentina: Fiscal DiSequiliba Lending to Hyperinflation at equilibrium. As a consequence, the relationship between the Rnn and the rate of government spending is subject to empirical determination. The final expression to be tested empirically-one that incorporates the possibility of governent spending at a rate different from that of the private sector-is therefore: (3.42) log(RuM) = C0 + w log(Pm/Px) + z ts + e Ag where the sign of e is the same as the sign of the difference between the propensities of the govermment and the private sector to consume nontradable goods. If the government has a higher demand for non- tradable goods than the private sector, an increase in government spending for a given level of total demand implies a shift in demand toward nontradable goods and, therefore, a fall in the import real exchange rate. Unfortunately, monthly or quarterly series of government spend- ing are not available, and we were therefore restricted to the use of annual data for the period 1964-87. The results presented below show the estimation of the structual relationship for the import real exchange rate. This variable is constructed as the ratio of the imported component of the wholesale price index (wpi) to the consumer price index (cpi): uRERm = log(price of imports in wri/cpn) The explanatory variables are: * TSCDP, the ratio of the trade account balance to cO;P. (Since the trade account is denominated in U.S. dollars, the nominal GDP was converted into U.S. dollars using the official exchange rate for commercal transactions from the Fn.L data bank.) The expected sign of the effect of this variable on RERM is positive. * LPXM, the internal terms of trade-in logarithmic form, log(Px/Pnz), equal to the ratio of the agricultural component of the wPi to the imported component of the same pnce index. This variable incorporates the substitution effects attributable to the external ternts of trade and of taxes and subsidies on foreign trade. The expected sign of the effect of this variable on RERM = PmlPh is negative (a rise in Px increases Ph by a smaller propor- tion and thus reduces Pm(Ph). * Government spending, captured by the ratio of nominal govem- ment spending to nominal cDP. The expected sign of this vari- able is negative, under the reasonable assumption that the gov- ernment has a larger propensity to consume nontradable goods than does the private sector. Carios Alfredo Rodrfguez 153 (3.43) LREM = 5.16 + 0.O7TSGDP - 0.49LPXM - .O2TEGDP. (17.3) (2.5) (-2.8) (-2.9) Instruments: TSGDP(-1), TSCOP(-2), PD, PD(-1), LPXM, LPXM(-1), TEGDP, TEGDP(-1) AR(1) = 0.34(1.56) Adjusted R2 = 0.53 Durbin-Watson statistic = 1.63 Sample period = 1964-87 The empirical results show that all three variables have the expected signs and are highly significant in the determination of Argentina's import real exchange rate. The trade surplus coefficient equals 0.07, indicating that a 1 percentage point increase in the ratio of the trade surplus to GDP is associated with a 7 percent increase in the import real exchange rate. The coefficient of the internal terms of trade is approximately equal to 0.5. This means tat a 10 percent rise in the price of imports (or of exports) results in a 5 percent rise in the CI, which is our measure of the nontradable goods price index. Finally, the coefficient of the ratio of government spending to GDP is equal to -0.02, meaning that a i percent increase in this ratio is associated with a 2 percent decrease in the real exchange rate. This result imples that the government has a higher propensity to spend on nontradable goods than does the private sector- To correct for simultaneous determination bias, the regression was estimated with two-stage least squares. Instrumental variables were used for the ratio of the trade surplus to GDP (instruments were the current and lagged primary deficits of the public sector, lagged ratios of trade balance to awP, and current and lagged remaing explana- tory variables). All the coefficients are significantly different from 0 at the 2 percent confidence level or less. Figure 3.14 shows the actual relationship between the ratio of the trade balance to GDP and the import real exchange rate. (Both variables are normalized by their means.) Since the late 1970s Argentina's foreign debt has risen from a negli- gible level to almost 100 percent of Gro- (In 1990 the debt-induding the arrears accumulated since 1988-reached $66 billion.) Servicing this debt would require a trade surplus of about 10 percent of GDP just for the nominal interest (assuming an interest rate of 10 percent per year and also assuming that the goverrment, which is the main debtor, has a fiscal surplus large enough to purchase the trade surplus requied to pay its debt). According to our regression results, generating a ratio of trade balance surplus to GDIP (TSGDP) of 10 percent, as required to fully service the nominal interest on the Argentine foreign debt, would imply a real exchange rate 70 percent higher than if there were no 154 Argentina: Fisc Disequilzbria Leading to Hyperinflation Figure 3.14. Real Exchange Rate and Trade Balance, Argentina, 1964-87 Percent 3- 2 -- -- 1964 66 68 70 72 74 76 78 80 82 84 8687 ----- Real exchange rate (deviation from sample average) -m Tde balance (share of CDP) Source: Country case study. need to servrice external debt. Actually, however, the real exchange rate. in 1989 was at levels sinilar to those that prevailed in the early 19,70s, when the debt was nonexistent. This is in part a reflectxon of the fact that Argentina has not serviced its external debt since 1988. Other factors also have worked toward the maintenance of a rela- tively low real exchange rate. In particular, the increase in real gov- errnment spending, according to our results, has been biased toward nontradable goods and has therefore'tended to lower the equlilibrium reail exchange rate- Govenunent spending as a fraction of CDP inceased from about 27 percent in the early 1970s to 56 percent in 1983, falling back to 43 percent in 1987. According to the empinrcal results in equation 3.43, each additional percentage point of GDP irt goverrnment spendirtg is associated with a 2.1 percentage point fall int -he real ex ahange rate. The Trade Balance and shae Fisca) t As explained in chapter 2 dn tis vAolume, the trade defict may or may not be infIuenced by the fiscal deficit, depending on the validity of ere Carlos Alfredo Rodriguez 155 Ricardian equivalence hypothesis. Other variables that affect the trade deficit are debt service, debt levels, and inflation rates, which may affect the desired composition of portfolios and therefore the rate of capital flow. Another variable that theoretically belongs in the trade balance equation is the terms of trade (Laursen-Metzler) effect. an improve- ment in the terms of trade increases real income and may thus induce an increase in the desired stock of foreign assets. However, the rela- tionship is not clear and unambiguous, since a debtor country facing an increase in real income may decide it can support o lrger stock of foreign debt. The final word on the relationship between the terms of trade and the trade balance, therefore, wll be empirical. The deficit variable chosen consists of the deficit of the consolidated public sector before any interest service (primary defict), normalized by CDP. The terms of trade series are from the Economic Commission for Litin America and the Caribean (ECLAC). No series on the stock of fiL.Ugn assets held by Argentines are available. We tried two vari- ables as proxies: (a) the balance on the service account of the balance of payments and (b) a measure of foreign assets held, constructed from the accumulated sum of current account surpluses. None of the variables tested to capture the effects of the level of foreign indebtedness turned out to be significant in explaining the trade balance. One would expect that the effect of the level of foreign debt (or its service) on the trade balance surplus would have a posi- five sign, indicating that the economy makes some adjustment in order to service its debt. In the regression presented in equation 3.44, however, the debt service variable has the correct sign but is not significantly different from 0, as indicated by the t-value of only 0.83. In the regression presented in equation 3.45 we test the series gener- ated for the stock of foreign debt by accumulating the past current account deficits from 1960 onward (arbitrarily setting the initial level of debt equal to 0). This variable (lagged one period in order to repre- sent initial debt during the current period) also has the correct sign, but again the coefficient is not statistically significant. These results simply verify the obvious observation that the Argentine economy has not paid its external debt and therefore has not faced up to the need to adjust the level of the external trade surplus. It turns out that only two of the variables, inflation and the fiscal deficit, are significant in explaining the trade surplus. In both cases the coefficients of the variables have the theoretically expected signs and are significantly different from 0 at the 2 percent or less confi- dence level. Higher current inflation improves the trade balance sur- plus in accordance with what is theoretically expected from the port- folio model discussed earlier. (It is not possible to illustrate the opposite long-rn effect because of the insignificance of the coeffi- 156 Argentina: Fiscl Disequilibna Leading to Hyperinflation cient of the foreign assets held.) The primary deficit of the CNFPS iS shown to deteriorate the trade balance surplus, as expected in an economy in which the Ricardian equivalence proposition is not fully valid. The coefficient value of 0.32 indicates that one-third of the value of the primary fiscal defict results in a trade deficit. The trade deficit in turn recquires a lower real exchange rate (a real appreciation) for the home goods market to dear. We therefore find a negative relationship between the real exchange rate and the primary deficit of thLe CNFPS. The next three regressions use the following variables: TSGDP = ratio of trade surplus to GOp INF = December-to-December inflation rate of the CPI PD = ratio of primary deficit in the CNFPS to GDP FASGDP = ratio of foreign assets to CDP (constructed by accumulat- ing current account surpluses) StRGOP = ratio of the service account surplus to GDP MU = rate of expansion of Ml (instrumental variable for infla- tion). All regressions were performed using the two-stage least squares method in order to correct for the simultaneous determination of TSGDP and inflation (PD, FASGDP, and SERGDP were considered exog- enous variables). (3.44) TSGDP = 1.63 + 0.14sERGDP + 0.009iNr - 0.29PD. (2.3) (0-83) (2-5) (-2.2) lnstnnments: SERGDP, INF(-1), PD, PD(-1), MU, MU(-I) Adjusted R2 = 0.42 Durbin-Watson statistic = 1.60 Sample period = 196348 (3.45) TSGDP = 1.90 - 0.005FAsGDP (-1) + 0 .OluN - 0.33PD. (2.5) (-0.09) (3.2) (-2.2) Instruments: FASGDP (-1), I[F (-1), PD, PD (-1), MU, MU (-1) Adjusted Rz = 0.36 Durbin-Watson statistic = 1.65 Sample period = 196348 (3.46) TSGDP = 1.94 + 0.01 [NF - 0.32PD. (3.17) (4.5) (-2.7) Instruments: INF (-1), PD, PD (-1), MU, MU (-1) Adjusted R2 0 .39 Durbin-Watson statistic = 1.65 Sample period = 196348 Carlos Alfredo Rodriguez 157 General Summary and Conclusions Argentina has had a very troubled economy in recent decades, and the behavior of its public sector may have been instrumental in that pro- cess. Government spending grew systematically faster than GDP until the last crisis of the Argentine economy began to develop in 1982. Since then, spending has started to fall, more because of resource constraints than as a result of deliberate political action. The decrease in the rela- tive size of government spending came too late to prevent the financial crisis that brought the country into a state of hyperinflation in 1989. The government ran a primary deficit (not including any interest payments) every year from 1961 to 1989. As a consequence, it had to resort to issuing money and interest-bearing debt. This in turn led to a systematic tendency of the economy to experience high real interest rates and inflation. A permanent positive primary deficit, coupled with high real interest rates and a stagnant economy, would be expected to lead to an ever- growing stock of public debt in relation to GDP. Real government debt did not grow continuously, however, because occasionally the existing stock of debt was melted down by outbursts of inflation driven by the large devaluations accompanying a foreign exchange crsis. Revenues of the CNFPS have historically shown a growth trend similar to that of expenditures, although at systematically lower levels. This high growth in fiscal revenues in the face of a stagnant economy should be enough to invalidate the commonly voiced claim that Argentina's basic problem is that the private sector does not pay taxes. In fact, not only is fiscal pressure very high, but it has also grown at a much faster rate than that of GDP in the past twenty-five years. Even though fiscal pressure has been high, fiscal spending has also grown and has systematically exceeded revenues during the three decades covered by this study. The significant changes experienced by Argentina after 1990 were a result of a genuine desire on the part of authorities to promote struc- tural adjustment, an attitude partly induced by the debacle caused by hyperinflation. It is dear that after the adjustment that has taken place, it is very unlikely that the economy will revert to its earlier condition. The significant reduction in the size and role of the public sector is not likely to be reversed, either. Appendix This appendix complements the discussion of asset markets by assessing the impact of fiscal policy variables on private saving and investment. 158 Argentina: Fiscal Disequilibria Leading to Hyperinflation Private Consumption In this section we examine how Argentine private consumption rea. ts to an increase in the public sector financing needs as a result of either an increase in public expenditure or a decrease in public revenue. A critical issue in this context is whether domestic public debt can or cannot be considered net private wealth in Argentina. Assuming -itional economic behavior, the question becomes whether capital market imperfections are strong enough to produce net private wealth effects from changes in public financing strategies (debt versus taxes). Because of data limitations, what we present here is a simplified version of the framework proposed in the project. We estimate a private consumption function that depends on fiscal expenditures and revenues. In contrast to the usual practice, we do not include the real interest rate as an argument of the consumption function. Between the 1940s and the early 1970s Argentina experienced nearly permanent financial repression, rendering the government-imposed interest rate (charged in the formal financial market for rationed credit) meaningless from the point of view of resource allocation. This situation changed after the financial reform of 1977, which eliminated the former direct regulation of credit. However, as we argue in "The Process of Determining Public Debt and the Interest Rate," above, the interest rate in local currency, which was determined in a financial market working with a horizon of seven days or less, basically reflected the state of short-run liquidity rather than the intertemporal tradeoffs faced by economic agents in Argentina. Data in the regressions reported below, we used the annual database of the Fundaci6n Mediterranea for the years 1913-84. The original vari- ables taken from that database were: PBIPM = real CGNP at market prices CONSUMP = real private consumption GGN = nominal national goverrment expenditure IGN = nominal national goverrnment revenue PPBICF-= implicit prices of GOP at factor cost. Using the GNP implict prices (PPBicr) series as a deflator, we con- structed the series for real national govermnent expenditure (RGGN) and revenue (RIGN). CarlosAIfredo Rodriguez 159 Regressions To account for structural change, all the regressions were conducted both for the entire database sample, 1914-84, and for the 1960-84 period. The private consumption functions we estimated are: (3.47) CONSUMP(t) = 50.51 + 0.79PBIPM(t - 1) (4.21) (17.80) -61.O8RGC 4(t - 1) - 22.5IMCN(t - 1). (-2.79) (-0.66) R2 = 0.98 Adjusted Ri = 0.98 Durbin-Watson statistic = 1.36 Sample period = 1914-84 (3.48) CoNSUMP(t) 176.81 + 0.70PBeM(t - 1) (3.11) (9.88) 46.18RCCN(t -1) - 38.70EuGN(t - 1). (-1.62) (-0.83) R2 = 0.927 Adjusted R2 = 0.916 Durbin-Watson statistic = 1.98 Sample period = 1960-84 In both regressions only public expenditure has a coefficient signifi- cantly different from zero. Furthermore, the coefficient is negative, as we would expect either from a simple wealth effect or from combined wealth and substitution effects. Government revenue has no signifi- cant effect, consistert with the Ricardian equivalence proposition. Correcting for first-order autocorrelation does not change our results. Only public expenditure has - significant negative impact on private consumption. Given that both the consumption and the fiscal series are autocorrelated, we also tried the alternative procedure of normalizing the private consumption, public expenditure, and reve- nue data with GNP. The normalized variables are: NCON1913:1 1984:1 = CONSUMP(T)IPBIFM(7) NGGN 1913:11984:1 = 100*[GGN(CO/PPBICF(T)]IPBIPM(T) NICG 1913:1 1984:1 = 100*[IGN(2)/pIcF(7)1/PBIPM(7 In this case we have included the one-period lagged value of nor- malized private consumption and both past and current values of normalized government expenditure and revenue as explanatory variables. The ordinary least squares regression estimates for both sample periods are: 160 Argentina: Fiscal Disequilibria Leading to Hyperinflation (3.49) NCON(t) 0.49 + 0.46NCON(t - 1) (5.44) (4.5) + 003NccN(t) - 0.72NCCN(t- 1) (0.12) (-2.30) -0.10NIGN(t) + 0.08NIGN(t - 1). (-0.24) (0.21) R2 = 0.58 Adjusted R2 = 0.55 Durbin-Watson statistic = 1.85 Sample period = 1914-84 (350) NCON(t) = 0.36 + 0.53NCON(t - 1) (2.89) (3.82) + 041NC;N(t) - 0.86NcGN(t -1) (1-74) (-3.76) -0.25NiGN(t) + 0.59NICN(t - 1). (-0.88) (1.64) R2 = 0.673 AdjustedR2 = 0.587 Durbin-Watson statistic = 2.078 Sample period = 1960-84 Again, the lagged value of public expenditure has a negative impact on private consumption, while public revenue shows no statistically significant effect. On the basis of this evidence, it would be tempting to declare Argentina a Ricardian economy. That is not possible, however, because in Argentina changes in (conventional) tax revenues have nor- maUly been associated with changes in irflation-tax revenue. To illus- trate, we have regressed the revenue from money creation, computed as NFT(t) = [M(t) - M1(t - 1)]IPPcF(t), against the government deficit (DEF = RGGN - RIN), obtaiiing the following results: (3.51) INr(t) = 0.12 + 078DEF(t). (2.11) (9.55) AR(1) = 0.56 R2 = 0.866 Adjusted R2 = 0.864 Durbin-Watson statistic = 2.07 Sample period = 1914-84 As equation 3.52 shows, a substantial fraction of the government deficit, as conventionally measured, has been financed by money creation. Actually, only in the 1970s did it become common practice to finance the deficit in the capital markets. As equation 3.53 shows, Carios Aifredo Rodriguez 261 about 90 percent of the deficit seems to have been financed with money creation between 1914 and 1970. (3.52) INFr(t) = 0.03 + 0.90DEF(t). (0.79) (7.00) AR(1) = 0.53 R2 = 0.762 Adjusted R2- 0.758 Durbin-Watson statistic = 2.08 Sample penrod = 1914-70 Clearly, for most oi this century the traditional choice in Argentine public finances was between conventional taxes and the inflation tax, not between conventional taxes and debt, as a test of the Ricardian hypothesis would require. Equation 3.53 shows very neatly the shift to debt financing in the 1970s. As we have discussed elsewhere, the shift can be dated to 1977. Unfortunately, lack of private consumption data with a quarterly fre- quency prevents us from testing the Ricardian hypothesis for the only period (the past twelve years) for which a test would really make sense. (3.53) INnT(t) = 0.49 + 0.62DEF(t). (4.07) (5.55) ARC1) = 0.05 R2' = 0.707 AdjustedR2 = 0.682 Durbin-Watson statistic = 2.09 Sample period = 1970-84 Taking into account the importance of inflationary finance in Argentina, we have also estimated equations 3.47 and 3.48 by com- puting the revenue from money creation as part of the total tax reve- nues of the Argentine government. In this form we make sure that any change in "revenues," given public expenditure, must mean a change in public debt. In equations 3.54 and 3.55, the variable (TAX) is the sum of conventional public revenues (RIGN) and the revenue from money creation (INn). (3.54) CoNsuMP(t) = 76.55 + 0.69PBIPM(t - 1) (4.35) (15.54) -110.39RGGN(t -1) + 81.63TAX(t - 1). (-3.13) (2.26) AR(1) = 0.37 R2= 0.986 Adjusted R2 _ 0.986 Durbin-Watson statistic = 2.10 Sample period = 1915-84 162 Argentina: Fiscal Disequilibria Leading to Hyperinflation (3.55) coNsuMP(t) = 259.42 + 0.54PBtpm(t - 1) (2.74) (540) -116.O7RCCN(t - 1) + 109.70TAX(t - 1). (-2.08) (1.61) AR(.1) = 0.26 R2 =0.922 Adjusted RZ =0.911 Durbin-Watson statistic = 1.998 Sample period 1961484 We again observe the negative impact of public spending. The reve- nue variable, TAx, is significantly different from 0, but it is positive, which probably captures the expansionary effect of money creation rather than the contractionary effect that the inflation tax would pro- duce if the Ricardian hypothesis were not valid. Private Investment In this section we expect to find a negative impact of deficit financing on investment as a result of rising domestic interest rates or of stricter credit rationing (in the case of financial repression). DnA. We use the same database descnibed in the section on con- sumption behavior. The additional original variables taken from that database were: INVESTr aggregate investment INVSTGOB = government inivestment PUBCON = public consumption. From aggregate (uNvEsT) and government (INvsTcOB) investment we calculated private investment (PiNv) and normalized private invest- ment (NfNV) as follows: PINV = INVESr - INVSTGOB NPINV = PINV/PBIPM. REGRESSIONS. Following a procedure similar to that used with con- sumption, we regressed normalized private investment against the normalized fiscal variables. The results of regressing normalized pri- vate investment against current normalized public expenditure and revenue, for the 1914-84 sample, are: CaT/Os Alfredo Rodriguez 163 (3.56) NPINV(t) = 0.78NPlNV(t - 1) - 0.10NGGN(t) (10.49) (-0.73) + 0.49N1GN(t). (245) R2 = 0.67 Adjusted R2 = 0.66 Durbin-Watson statistic = 1.58 Sample period = 1914-84 As expected, there is a significant positive impact of public revenue on private investment. Given public expenditure, an increase in pub- lic revenue reduces the deficit and this has a positive impact on investment. Since we saw in the consumption module that changes in public revenue do not affect private consumption and, hence, pri- vate saving, it is likely that the positive impact on investment comes from a freer or perhaps merely a smoother working of the capital market. The introduction of lagged values of public revenue and expenditure does not change the conclusion in any fundamental way. Equation 3.57 shows the results of regressing normalized private investment against the current values of normalized public expendi- ture and revenue for the 1960-4 sample period. (3-57) NvrTV(t) = 0.70NPINv(t - 1) + 0.02NGG{N(t) + 0.48NIGN(t). (3.71) (0.11) (1.45) R2 = 0.238 Adjusted R2 = 0-169 Durbin-Watson statistic = 1.77 Sample period = 1960-84 We can see that the results are considerably weaker than for the larger sample. The coefficient of revenue still has the correct sign, and it has a substantially larger t-statistic value. It is not, however, significantly different from 0 at a 5 percent significance level. In both equations 3.56 and 3.57 the coefficient of public spending tums out to be insignificant. To see if aggregation made a difference, we regressed private investment against public consumption and govermment investment. Equations 3.58 and 3.59 show the results. Again, as we observed with normalized variables, public revenue has a significant positive impact on private investment. The difference comes from the expenditure side: public consumption has, for both sample periods, a significant negative impact on private investment. Contrary to what one would expect, if we were to assume that gov- ernment investment is bound to produce positive externalities on private economic activity (as it definitely did in the nineteenth cen- tury with the railroads), government investment has no significant effect on private investment- 164 Argentina: Fisal Disequilibria Leading to Hyperinflation (358) PNv(t) -14.87 + 0.24PnzM(t - 1) (-126) (5.28) -1.35PUBCON(t -1) + 39.07RIGN(t -1) (-2.72) (1.77) + 0.24INVSTGOB(t - 1). (0.52) AR(1) = 0.42 R2 = 0Q47 AdjustedR2 = 0.944 Durbin-Watson statistic = 1.85 Sample period 1915-84 (3.59) PINv(t) = 61.75 + 0.23PBrM(t - 1) - 1.99PUBCON(t - 1) (1.16) (3.29) (-3.15) + 45.85RIGN(t -31) + 0.49iNVSTGOB(t - 1). (1.37) (0.68) AR(1) = 0.20 IC2 = 0.789 Adjusted R2 = 0.745 Durbin-Watson statistic = 1.93 Sample period = 19614A Notes The author is grateful for very useful comments and suggestions by Aquiles Almansi, who provided invaluable help for the econometric analysis pre- sented in the appendix. 1. The financial liberalization was but one of the many policy irents of what came to be known as the Stabilization Plan of December 1978. The plan began to be put into effect after the military coup of March 1976; the financial opening of 1977 was followed by the tabla canbiaria (prefixed exchange rate path) and the trade reform of 1978. The plan was abandoned after March 1981 in the middle of a set of serious disadjustments, among them currency overvaluation, persistence of inflation, and the external debt problem. The lack of fiscal adjustment has been cited as the main reason for the failure of this stabilization attempt that aimed at making struchurl adjust- ment the centerpiece of the policies being followed. Literature covering developments during this period include Calvo (1981b, 1986); de Pablo (1983); Fernandez (1982); Rodriguez (1982a, 1982kb, 1983); and Sjaastad (1982). The sequential order of financial and trade liberalization has also been men- tioned as a factor contributing to the failure of the plan, an issue that is analyzed in Edwards (1984). Curlos Alfredo Rodriguez 165 References Almansi, Aquiles, and Carlos Rodriguez. 1989. "Reforma Monetaria y Finan- ciera en Hiperinflaci6n." Working Paper 67. Centro de Estudios Macro- econ6micos de Argentina (CEMA), Buenos Aires. Auernheimer, Leonardo. 1987. "On the Outcome of Inconsistent Pr,grams under Exchange Rate and Monetary Rules, or 'AlUlowing Markets to Com- pensate for Government Mistakes.'" Joumnal of Monetary Economics 19 (March): 279-305. Calvo, Guiflermo A. 1981a. "Devaluation, Levels versus Rates." Journal of International Economics 11: 165-73. - .1981b. "Reflecciones Teoricas Sobre el Problema de Estabilizaci6n en Argentina." Worldng Paper 29. Centro de Estudios Macroecon6micos de Argentina (CMA), Buenos Aires. - 1985. "Currency Substitution and the Real Exchange Rate: The Utility Maximization Approach." Journal of International Morney and Fimnace 4 (June): 175-88. - 1986. "Fractured laberalism: Argentina under Martirnez de Hoz." Economic Devlopmentand Cultural Change 34:511-33. Calvo, Guillermo A., and Carlos Rodriguez. 1977. "A Model of Exchange Rate Determination under Currency Substitution and Rational Expecta- tions."Journal of Political Economy 3: 617-25. Cavallo, Domingo, and Angel Pefia. 1983- "Defidt, Endeudamiento del Cobierno y Tasa de Inflacidn: Ar5entina 1940-1982, Estudios." Fundaci6n Mediterranea, C6rdoba, Argentina. 1984. "Gasto PNblico y Ti,o Real de Cambio." Instituto de Estudios Econ6micos sabre la Realidad Argentina y atinoamericana (IEERAL), C6r- doba, Argentina. de Pablo, Juan Carlos. 1983. "El Enfoque Monetario de la Balanza de Pagos en la Argentina: Analisis del Programa del 20 de Diciembre de 1978." El Timestre Econ6mkio (Mexico) 50 (April-June): 6450-69. Diaz-Alejandro, Carlos. 1981. "Tipo de Cambio y T&rinos de Intercambio en la Republica Argentina." Working Paper 22. Centro de Estudios Macro- econ6micos de Argentina (cEmA)v Buenos Aires. Dornbusch, Rudiger. 1974. "Tariffs and Nontraded Goods." Joumna of Inter- national Economics 4 (May): 177-83. Dombusch, Rudiger, and Juan Carlos de Pablo. 1987. "Argerttina: Debt and Macroeconomic Instability." NBER Worling Paper Series 2378. National Bureau of Economic Research, Cambridge, Mass. Edwards, Sebastian. 1984. "The Order of Liberalization of the External Sector in Developing Countries." Princeton Essays in International Finance 156. Princeton, N.J. Fernandez, Roque. 1982. "La Crisis Financiera Argentina." Working Paper 35. Centro de Estudios Maaoecon6niicos de Argentina (cMA), Buenos Aires. - . 1989. "Hiperinflacion, Repudio y Confiscacion: Los Lirites del Bman- cianinento hnflacionario." Working Paper 65. Centro de Estudios Macro- econ6n-icos de Argentina (cEMA), Buenos Aires. 166 Argentina: Fiscal Disequilbria lding to Hyperinflation RiEL (Fundad6n de Irvestigacones Econdmicas Latinoamericanas). 1987. El Gasto Nblicao en Argentina 1960-1985. Buenos Aires. Frerklel, Jacob A., and Assaf Razin. 1986. "FIscal Polces in the World Econ- omy." Journal of Political Economy 94, pt. 1 (June): 564 -94 Frenkel, Jacob A., and Carlos Rodriguez. 1982. "Exchange Rate Dynamics and the Overshooting Hypothesis." Interational Monetary Fund Papers 29 (March): 1-30. Harberger, A. C. 1958. "Trade Policy and the Real Exchange Rate." World Bank, Economic Development Institute, Washington, D.C. IMF (International Monetary Fund).- Various years. Intermationa Financial Sta- istis. Washington, D.C. Kiguel, Miguel A., and Pablo Andrds Neumeyer- 1989. "Inflation and Sei- gnorage in Argentina." Policy Research Working Paper 289. World Bank, Country Economics Department, Washington, D.C. Leiderman, Leonardo, and Mario I. Blejer. 1988. "Modeling and Testing Ricardian Equivalence, A Survey." International Monetary Fund Staff Papers 35 (March): 1-35. Rodriguez, Carlos A. 1978. "A Stylized Model of the Devaluation-Inflation Spiral." International Monetary Fund Staff Papers 25 (March): 76-8. 1982a. "The Argentine Stabiization Plan of December 20th." World Deuelopment 10 (September): 801-11- 198Tb. "Gasto Pdblico, Deficit y Tipo Real de Cambia: Un Analisis de sus Interrelaciones de Largo Plazo." Cuadernos de Economia (Chile) 19 (August): 2D3-16. .-, 1983. "Politicas de Estabilizaci6n en la Economia Argentina." Cuadernos de Economfa (Chile) 20 (April): 21-42. 1989. "'Managing Aigentina's External Debt: The Contribution of Debt Swaps." World Bank, Latin America and the Caibbean Regional Office, Washington, D.C. Rodriguez, Carlos A., and Larry A. Sjaastad. 1979. "El Atraso Cambiario en Argentina: Mito o Realidad." Working Paper 2. Centro de Estudios Macro- econ6micos de Argentina (cEmA), Buenos Aires. Schenone, Osvaldo H. 1987. "El Comportamiento del Sector Publico en Argentina; 1970-1985." Working Paper 60. Centro de Estudios Macro- econ6micos de Argentina (CEMA), Buenos Aires. Sjaastad, Larry A. 1980. "Commercial Policy,'True' Tariffs and Relative Prices." In John Black and Brian Hindley, eds., Current rssues in Commerca Policy and Diplomacy. New York: St. Martin's Press. -. 1982. "The Failure of Economic Liberalism in the Southern Cone." 1982 Bateman Lecture. University of Western Australia, Nedlands World Bank. Various years. World Debt Tables. Washington, D.C. 4 Chile: Fiscal Adjustment and. Successful Perrce Jorge Marshall and Klaus Schmidt-Hebbel After almost two decades characterized by extreme volatility in eco- nomic performance, Chile has been showing remarkable results since the mid-1980s. Gross domestic product (GDTp) has grown at an aver- age 6.7 percent since 1986, unemployment fell from 28 percent in the early 1980s to 5 percent in 1992, and inflation has been cut in half, to 13 percent. Gross domestic investment increased from 12 percent of GDP in 1984 to 21 percent in 1992, while the current account deficit shrank from 10.7 percent of GDP in 1984 to 1.5 percent in 1992. Exter- nal debt has been serviced regularly, and its total amount has been reduced through debt conversion. The reasons behind this success are to be found in the deepening of structural reforms and the adop- tion of coherent macroeconomic policies, among which fiscal policy played a key role. This chapter analyzes the macroeconomic determinants and effects of public sector deficits in Chile after 1974. The focus is on key issues related to the interaction between public deficits and macroeconomic variables and the channels through which deficits are transmitted to the economy. We begin with a brief historical overview of the background against which fiscal policy was implemented.1 The evolution of selected mac- roeconomic indicators during the past two decades is summarized in table 4.1. Table 4.2 reports the structure of the consoldated nonfinan- cial public sector (cNFPs) deficit and its financing during 1974-88.2 Table 4.3 presents an estimate of quasi-fiscal Central Bank losses dur- ing the 1980s. The first half of the 1970s was dominated by serious macroeconomiic imbalances inherited from the Allende government. The last year of the Allende administration, 1973, was characterized by a huge cNFrs deficit resulting from laige losses by public enterprises, low fiscal reve- nue, and overexpanded public sector employment, wages, and social security payments. The deficit, which exceeded 20 percent of GDP, was largely financed by Central Bank monetized loans. After the 1973 military takeover, Chile followed a strategy of market deregulation, privatization of public enterprises, and opening of the 167 168 Chile: Fiscal Adjustment and Successful Perfonnance Table 4.1. Selected Macroeconomic Indicators, Chile, 1970-92 (percent) CNFPS CEntnS Bank Cumgent deficit as qmasi-fiscal account share of losses as CDP deficit as Year GDP shareof GDP growth Inflation shareof GDP 1970-73 23.4 - 0.7 204.3 2.4 1974 5.5 - 1.0 369.2 0.5 1975 2.1 - -9 343.3 5.2 1976 -4.0 - 3.5 197.9 -L7 1977 -OA - 9.9 84.2 3.7 1978 -1.4 - 82 37.2 52 1979 -4.6 - 8.3 38.9 5.4 19 -5A - 7.8 31.2 7.0 1981 -OA - 5.5 95 14.4 1982 3.9 14.0 -14.1 20.7 9.2 1983 3.5 6.1 -0.7 23.1 5.4 1984 4.6 7-9 6.3 23.0 10J7 1985 2.9 13.1 2.4 26A 8.3 1986 2.0 -0.3 5.6 17.4 6.9 1987 0.2 0.8 5.7 21.5 4.3 1988 -3.6 -0.1 7.4 12-7 0.7 1989 -3.8 -02 10.0 21.4 3.1 1990 -0.5 - 2.1 273 2.1 19921 -A1 - 6.0 18.7 -0.5 1992 -0.8 - 10A 12.7 1.5 - Not available. Note: The consoUdated nonfinancial public sector (CNFPS) deficit is in nominal terms and comprises general governmert and nonfinancial public enterprise. Source. For Central Bank quasi-fiscal losses, authors' estimates. For other data, Cen- tral Bank of Chile and Budget Office, Ministry of Fnance, Chile. economy to foreign trade and capital flows. Along with these struc- tural reforms, the goverunent applied a strict stabilization program through monetary control, fiscal discipline, and general retrenchment by the public sector. Government spending was cut sharply by reduc- ing real wages, employment, and public investment. Tax rates were raised, and many firms that had been transferred to the public sector by the Allende government were returned to the private sector. The remaning public enterprises were forced to maximize profits and had to face hard budget constraints. The 1975 tax reform introduced value added taxation, significantly simplified the tax system, and indexed tax bases to inflation. As a result, the 1975 CNFPS deficit fell dras- tically, to 2.1 percent of GDP. From 1976 to 1981 the public sector showed sizable surpluses. Fiscal deficits having been stabilized, public finances were further favored by a recovery of growth, the new, more broadly based tax system, Table 4.2. Nonfinancial Public Sector Revenue, Expenditure, Surplus, and Financing, Chile, 1974-88 (percentage of GOr) Item 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 General governmenit Current revenues 30.3 34.9 37.4 38.6 33.2 32.5 32.9 32.1 29.9 27.7 28.7 28.6 28.2 28.4 28.7 Direct taxes 5.7 8.3 7.0 5.4 5.3 5.2 5.4 5.5 4.8 3.1 3.4 3.1 3.2 3.1 2.9 Indirect taxes 13.0 14.6 14.0 14.9 13.7 13.3 13.4 14.8 13.8 14.6 16.3 17.1 17.1 17.3 14.5 Copper taxes 1.7 1.3 2.6 2.6 2.8 1.9 1.9 0.2 1.0 1.9 1,3 0.5 0.6 1.1 4.7 Social security 3.1 3.4 3.4 3.7 3.7 5.3 5.6 4.7 3.3 2.8 2.8 2.4 2.5 2.2 1.9 Other income 6.8 7.3 10,3 12.1 7.7 6.7 6.6 6.9 7.2 9.3 5.0 5.5 4.9 4.7 4.7 Current expenditures 26.4 27.6 31.0 33.0 26.8 24.8 24,5 26,7 31.9 30.6 30.7 29,5 27.4 26.3 23.2 Purchases 4.3 4.4 3.7 4.9 4.7 3.0 3.1 2.9 3.3 3,2 3.4 3.2 3.0 3.3 2.7 Wages 10.0 9.3 8.9 11.0 10.0 9,2 9,1 8,8 10.3 8.9 8.5 7,8 7.4 6.8 6.2 Domestic interests 0.4 0.5 0.2 0.8 0,9 0.7 0.5 0.2 0.0 1.3 1.8 2.4 1.2 1.7 1.8 External interests 0.9 2.1 2.0 0.9 0.7 0.5 0.4 0.3 0.5 0,5 0.6 0.8 1.1 1.3 1.3 Total transfers 6.4 10.3 12.3 12.2 9.8 10.8 .10.9 14.1 17.3 16.3 16.2 15.0 14.3 13.0 11.2 Other expenditures 4.3 1.1 3.9 3.2 0.7 0.5 0.5 0.4 0.4 0.4 0.2 0.3 0.3 0.2 0.1 (Table canlinutes o01 thefollowinig page.) Table 4.2 (continted) Item 1974 1975 1976 2977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 Saving 3.9 7.3 6.4 5.6 6.5 7.7 8A4 5.5 -1.9 -2.9 -2.0 -0.9 0.9 2.1 5.5 Public investment 8.7 5.7 3.1 4,2 3.5 3.2 2.6 2.5 2.1 2.1 2.3 3.1 3.3 3.0 2.9 other expenditures 1.9 1.6 0.1 0.5 0.8 -0.7 0.3 0.1 -1.8 -1.9 -0.7 -0.4 -0.8 -0.6 -0.8 Surplus -6.6 0.0 3,2 0.9 2.1 5.1 5.5 2.9 -2.3 -3.1 -3.5 -3.6 -1.7 -0.3 3.4 Public eitterprises Current reveinues 32.8 35.9 31.0 26,6 24.7 27,0 25.8 20.8 24.2 29.5 29.8 35.3 35.4 34.7 37.8 Current expenditures 25.3 30.2 20.5 18.4 18.3 17.6 16.3 15.6 16,5 18.5 19.2 21.2 22.6 21,8 22.2 Saving 7.5 5.7 10,5 8.2 6.4 9,4 9.5 5.2 7.7 11.0 10.6 14.1 12.8 12.9 15.6 Taxes and transfers 4.8 5.4 6.9 5.2 4.4 8.3 7.5 5.6 7.1 8.6 8.2 9.6 9,1 9.7 12.6 8 Investment 3.9 3.5 3.0 2.7 3.2 1.9 2.6 2.6 2.6 2.6 3.7 4.0 4.7 4.0 3.3 Other income 2.3 1.1 0.3 -0.8 0.4 0,2 0.5 0,4 0.3 -0.1 0.3 0.3 0.5 0.9 0.6 Surplus 1.2 -2.1 0.9 -0.5 -0.8 -0.6 -0.1 -2.5 -1.6. -0.4 -1.1 0.8 -" 5 0.1 0.3 Nonfinancial puiblic sector Consolidated surplus -5.5 -2.1 4.0 0.4 1.3 4.6 5.4 0.4 -3.9 -3.5 -4.6 -2.9 -2.1 -0.2 3.7 Financing 5.5 2.1 -4.0 -0.4 -1.3 -4.6 -5.4 -0.4 3.9 3.5 4.6 2.9 2.1 0.2 -3.7 Internal 5.2 5.0 -0,9 -0.1 -4,0 -4.6 -5.3 -3.2 1.7 2.4 1.9 -1.2 -1.0 -2.2 -7.3 External 0.3 -2,9 -3.1 -0.3 2.7 0.0 -0.1 2.8 2.2 1.1 2.7 4.0 3.1 2.4 3.6 Souirce: Budget Office, Ministry of Finance, Chile. Jlage Marshall and Klaus Schmidt-Hebbed 171 Table 4.3. Central Bank Losses rom Quasi-Fiscal Operations, Chile, 198289 (percentage of GDP) Item 1982 1983 1984 2985 1986 1987 1988 1989 Loans to bankrupt financial institutions 8.6 0 0 0 -0.3 -0.2 - - Purchase of commercial barns' bad laans 0.6 2.4 -0.2 0 0.3 -0.1 -0.1 -0.2 Domestic debt rescheduling 4.8 2.9 1.5 4.6 0.9 0 - - Preferential exchange rate program 0 0.3 0.9 1.1 0.1 0.3 - - Exchange rate guarantee program 0 0.5 2.8 3.5 0.2 0 - - Foreign exchange capital losses 0 0 2.9 3.9 -1.5 0.8 - - Total quasi-fiscal losses 14.0 6.1 7.9 13.1 -0.3 0.8 -0.1 -0.2 - Not available. Note. This table presents annual losses consistent with the capitalized losses of table 4.5. For losses from purchase of bad loans and rescheduling of domestic debt, the estimates of net losses were obtained by multiplying gross losses (presented in tables 4 and 6 in Eyzaguirre and Latrailaga 1990) by the ratios of net loss to present value of disbursements calculated from table 4.5. These ratios are 0.193 for purchase of bad loans and 0.248 for domestic debt rescheduling. Source Eyzaguirre and Larrafiaga 1990; authors' calculations. and the profit-maximing behavior of public enterprises. The liberalization-cum-stabilization process seemed to be very successful. Eaflation, which had reached 1,000 percent in 1973, fell to 30 percent in 1980 and almost converged toward intemational levels in 1981-the stated objective of the fixed-exchange-rate policy pursued since mid-1979. GDP growth averaged 7.2 percent durng 1976-81. How- ever, p-ivate consumption and investment grew explosively, at rates that subsequently proved unsustainable. Although part of the private sector deficit was financed by public surpluses, the private sector increasingly relied on foreign saving; the current account deficit reached a record 14.4 poercent of GDP in 1982.3 At the erd of 1981 Chile was burdened by a huge foreign debt and an unstable domestic financial system-a result -of inconsistent exchange rate and wage policies, a government-supported euphoria that contributed to private overspending, and lax regulation of the financial system. Given this setting, a slight negative shock could 172 Chile: Fiscal Adjustment and Successful Performance have triggered a crisis. What happened was that a number of strong adverse foreign shocks hit the economy simultaneously. Declining terms of trade, higher foreign interest rates, and the end of voluntary foreign lending contributed to the massive 1982-83 financial crash and economic recession. In 1982 alone, real GDP declined by 14.1 percent, and inflation climbed to 20.7 percent. The recession lowered public sector revenues, particularly those from trade-related taxes. The huge unemployment rate forced the government to undertake a series of emergency support programs that raised goverrunent spending. In 1984 a structural adjustment program was initiated to address three pressing problems faced by the Chilean economy: excessive concentration of exports, low levels of national saving, and a precar- ious financial system. From the perspective of fiscal policy this period is characterized by a major fiscal adjustment program aimed at both stabilization and consolidation of the structural reforms begun in 1974. Six poicy reforms affected the nonfinancial publc sector. First, current expenditures were curtailed. Second, income taxes were reduced significantly, with the purpose of encouraging private! saving and investment. Third, revenues were falling as a result of the 1981 social security reform.4 Fourth, a nev impetus was given tu the priva- tization program initiated in 1974. Fifth, as terms of trade improved in 1988 and 1989, a new tax reduction-this time affecting value added rates-transferred additional resources to the private sector. Sixth, government expenditure was delinked from volatile public sector copper revenue by a copper price stabilization fund, established in 1985. The outcome of these reforms, boosted by a favorable evolution of foreign and domestic macroeconomic conditions, was a gradual hut massive imnrovement in fiscal stance. The CNFPS defict, which had reached 4.6 percent-of GDP in 1984, was almost eliminated by 1987, and a sizable cNw's surplus was recorded in 1988-89. Inportant public policy innovations during the 1980s also took place within the Central Bank. In response to the 1982-83 financial and exteral crises, the goverrunment decided to bail out the private sector by providing Central Bank loans at subsidized rates and extending exchange rate subsidies to debtors in foreign currency.5 The losses from these operations reached staggering levels: 14.0 per- cent of GDP in 1982 and an average annual 10.3 percent of GDP throughout 1982-85, the four-year period of financial bailouts and interest and exchange rate subsidies. The Central Bank financed most of its losses by issuing domestic debt, which increased by US$5.7 billion between 1982 and 1969. In addition, its foreign liabilities increased by US$4.0 billion during the same period The combination of fiscal reform in the nonfinancial public sector and the lender-of-last-resort emergency financing provided by the Central Bank contributed to a swift recovery of the economy and public Jorge Marshall and Klaus Schmidt-Hebbel 273 finances after 1985. Resource allocation and private spending also responded strongly to the massive real devaluations of 1982-85, which were supported by a nominal crawling-peg exchange rate policy, based on purchasing power parity, that was occasionally interrupted by cor- rective discrete devaluations. As a result of these policies, and witi the help of favorable terms of trade after 1988, the current account deficit fel to an average 1.9 percent in 198849. GDP growth resumed strongly after 1985, as a consequence first of capacity recovery and then of a vigorous expansion of capacity as private investment responded to the incentives provided by structural reforms and a stable macroeconomy. In 1988 and 1989 monetary and fiscal policies were relaxed (the latter through tax cuts), contnbuting to some overheating of the economy. By the end of 1989 annualized iniflation had reached almost 30 percent, and GDP growth hit double-digit levels. In 1990 the newly independent Central Bank pursued a strongly contractionary monetary policy, and the new democratically elected Aylwin government obtained parliamentary approval for a tax increase. Tax revenue was raised by about 3 percentage points of GDP from 1991 onward through a combination of a higher value added tax rate (18 percent, up from 16 percent), a higher profit tax rate (15 percent, up from 10 percent), an expanded tax base, and higher per- sonal income taxes. After monetary policy was eased in 1991, GDP growth-which was fairly low in 1990-inaeased vigorously, reach- ing a record 10.4 percent in 1992, while inflation fell to 12.7 percent. The cim's showed systematic supluses that averaged 0.9 percent in 1990-92. The current account showed an average deficit of 1.0 percent of GDP during the same three-year period, implying that the private sector ran an average deficit of 1.9 percent of cop, evenly financed by the public sector and the rest of the wo.;d. To analyze the interaction between fiscal deficits and macro- economic variables, this chapter focuses on four key issues. The impact of macroeconomic and external variables on public sector deficits. The focus is on how vulnerable public finances are to variables beyond the direct control of fiscal policymakers and how fiscal policy reacts to exogenous shocks. Our hypothesis is that even though the Chilean public sector faces an unstable external environment, active fiscal policies have been able to counteract both external and domestic shocks. * The effect of the deficit on inflation and interest rates, and the effects of alternative strategies of deficit financing. Fiscal policy was an important factor behind the relatively low inflation and interest rates during the 1984-90 period. A puzzle to be explained is why the huge quasi-fiscal deficit did not destabilize the economy. 174 Chile: Fiscal Adjustment and Successful Performance The effect of fiscal policies on private sector consumption and spending. This chapter focuses on the relative contributions of foreign saving and fiscal policy to the private consumption boom in the late 1970s and early 1980s and to the decline in private consumption during the adjustment period. Public policies also contributed to the strong expansion of private investment in 1984-92. Perhaps the most important issue regarding the relation between fiscal variables and private spending in the Chilean case is the explanation for the opposite trajectories of private con- sumption and investment since the mid-1980s. * The key role of the exchange rate in Chile's adjustment. The strategy of trade deficit correction and export diversification re- quired a sustained real exchange rate depreciation.6 Our main hypothesis is that fiscal adjustment has contributed to a higher trade surplus and hence to a more depreciated real exchange rate. We begin with a review of the evolution of fiscal policy in Chile during the 1974-88 period, emphasizing the structure and decom- position of cNFps deficits, the scope of quasi-fiscal operations, and the sustainability of public sector deficits. A three-asset portfolio model is then developed in order to estimate the impact of domestic deficit financing on interest and inflation rates and to assess the implications of alternative strategies of deficit financing. The direct and indirect effects of fiscal policy on pnvate consumption and investment are analyzed, and the impact of public sector deficits on real exchange rates and the current account is estimated. The final section summa- rizes the main findings and their policy implications. Fiscal Policy, Decomposition of Nonfinancial Deficits, Quasi-Piscal Deficts, and Sustainability Fiscal deficits are affected by fiscal polices and exogenous variables; the former are under the direct control of policymakers and the latter are not. An assessment of the relative contribution of these two types of variables to the evolution of fiscal deficits helps in understanding the sign and the net effect of fiscal policy actions. Since 1974 Chile has faced severe shocks, and fiscal policy has been systematically ori- ented so as to compensate for the effects of these shocks on the deficit. For example, when falling copper prices increased the CNFPS deficit by 3.1 percentage points of GDP in 1973-75, tax and expendi- ture policies were used to reduce the deficit. Similarly, lower tax revenue was allowed to increase the deficit during 1987-89, when the rise in copper prices reduced the deficit by almost 4 percentage points of GDP. Jorge Marshall and Klaus Schmidt-Hebbel 275 Decomposition of Norfinancial Public Sector Deficit The decomposition of the cNFps attempts to measure the contribution of exogenous and endogenous fiscal policy variables to the deficit, following closely the methodology presented elsewhere?7 Endo- genous variables are defined as those under the control of fiscal policymakers, such as tax rates, public investment, expenditure, and public sector wages. Exogenous determinants of the defict include domestic variables (such as output, the real exchange rate, and infla- tion) and exteral variables (for example, copper prices and foreign interest rates) that are beyond the direct influence of fiscal policy- makers. Appendix 4.1 presents estimated functions for tax revenue, public enterprise surplus, and transfers to the private sector and iden- tifies the role of fiscal policy and exogenous variables in their behav- ior. The decomposition of the CNFPS deficit makes use of these esti- mated relations. Table 4A summarizes the net impact of fiscal policy and of domestic macroeconomic and external variables on the nonfinancial public sector deficit for 1973-88. The data are shown for four subperiods that are relevant from the point of view of both the fiscal stance and the state of the overail economy: sharp stabilization (1974-75); reform, recovery and euphonra (1976-81); crisis and stabilization-cum-adjustment (1982- 86); and recovery under continuing reform (1987-88). The sharp decrease in the 1973-75 public sector deficit is largely explained by fiscal policy variables. The main policy shifts that were conducive to the massive fiscal adjustment were tax reforms (which raised indirect and direct effective tax rates), the decline in public employment, and the surge of the public enterprise surplus. Fiscal balance was restored two and a half years after the experience of huge public sector disequilibria at the end of the Ailende administration. The 1976-81 period was characterized by relatively stable sur- pluses. The data in table 4.4, however, suggest that this stability was the result of two opposing forces. On one side, fiscal policy variables-effective tax rates, public sector real wages, and the exter- nal public debt-in conjunction with a declining inflation rate and a backward indexation scheme, raised the deficit On the other side, large increases in income and the recovery of copper prices lowered the deficit. During 1982-86 the deficit increased as a result of fiscal policy actions, including reduced direct tax rates, a declning number of affiliates to the state-run social security system, and the rising public debt. These changes reflected, respectively, the tax reform of 1982, the socisl security reform of 1981, and increasing - debt-financing of deficts swollen by government intervention in the economy during tlhe crisis. 176 Chile: Fiscal Adjustment and Successful Perfonnance Table 4.4. Consolidated Nonfinancial Publ]ic Sector (CNFPS) Deficit Decomposition, Chile, 1973-88 (change in public sector defict, in percentage points of CGDP) Itemn 1974-75 1976-81 1982-86 1987-88 Fiscal policy variables -14.1 13.2 5.7 5.2 Indirect tax rate -5.4 5.5 -2.4 1.8 Direct tax rate -3.3 3.5 2.2 0.7 Copper tax rate -1.8 0.0 0.0 -0.8 Public employment -4.3 -2.1 -0.1 0.0 AffiLiates to social security -0.1 L8 1.1 2.4 Public investment -0.8 -1.8 0.8 0.0 Public sector real wage 0.2 3.3 -1.5 0.4 Foreign public debt 1.7 2.8 2.9 0.5 Domestic public debt 0.0 0.0 2.7 0.1 Real purchases of goods and servces -0.3 0.2 0.0 0.1 Domestic macroeconomic variables 4.7 -7.5 -2.8 -2.7 Real cD? growth 3.4 -10.8 -1.3 -2.9 Real exchange rate -0.7 0.3 -1.2 0.0 Inflation rate 2.1 4.6 -0.1 0.1 R-eal wage growth -0.1 -1.6 -0.4 -0.1 Domestic interest rate 0.0 0.0 0.2 0.2 External variables 2.9 -LS 0.8 -3.6 Copper price 3.1 -2.3 1.3 -3.9 Foreign interest rate -0.2 0.5 -0.5 0.3 Nondecomposed accounts Public enteprise surplus (after transfers) - -8.4 0.4 -2.1 -0.7 Other public sector income -4.2 0.6 2.0 0.1 Other public sector expenditure -3.5 -2.3 -0.9 -0.2 Explained change in deficit -22.6 2.3 2-7 -1.9 Nonexplained change in deficit 4.1 -4.7 -0.9 -3.9 Total change in deficit -18.5 -2.4 1.8 -5.8 Note: For instance, the NFPs deficit fell by 2.4 percentage points of Gcp between 1975 and 1981 (compare the corresponding levels in table 4.), as reflected by the last line of the second column. The change in the deficit explaied by our variables is an increase of 2.3 percentage points of aDP (third line fwom bottom), which is overexplained by the combined effect of fiscal policyvariables, causing a deficit rise of 13.2 percentage points of CDP fi(rst line). Source: Authors' calulations based on the methodology of Marshall and Sdhmidt- HebbeI 1989a, 1989b. The 1987-88 recovery of the nonfinancial public sector was associ- ated with favorable changes in domestic and external sector variables- in parlicular, the surge in real income and copper prices. However, fiscal policy variables such as a decrease in the value added tax rate and Jorge Marshall and Klaus Schmidt-Hebbel 77 the continuing reduction of affiliates to the social security system had a negative impact on public finances. To compare the role of fiscal polices with the influence of exog- enous variables in the evolution of the cnFPs deficit in Chile, we computed the average contribution to the explained change in the defict over the 1974-88 penrod of three factors: fiscal policy (142 per- cent), domestic macroeconomic variables (-41 percent), and external variables (-1 percent).5 These figures confirm the massive predomi- nance of fiscal policy changes in both the cyclical and trend evolution of the CNFPS deficit. Although external and domestic shocks beyond the control of fiscal policymakers are important in shaping deficts- as reflected in table 4.4-they have been counteracted by massive policy changes. Hence, even in an economy such as Chile's in which the budget is subject to strong shocks, policymakers are to blame for fiscal detenroration and are to be praised for successful stabilization. Quasi-Fiscal Operations and Domestic Debt An essential feature of the Chilean adjustment process after 1982 has been the quasi-fiscal operations of the Central Bank in support of a bankrupt financial system-in effect subsidizing financial insttutions and private debtors. During the early 1980s the Central Bank, at that time under the aegis of the Ministry of Finance, responded to an acute balance of payments and domestic financial crisis (caused in part by a lax banking law) with a program of massive bailouts, credit, and exchange rate subsidies. Tnese operations, however, were lim- ited to the 1982-85 emergency period. The main quasi-fiscal programs were: a Loans to bankrupt financial institutions. In 1981 the Central Bank provided emergency loans to eight financial institutions that faced insolvency. All eight institutions went bankrupt, and the outstanding Central Bank loans turned into losses. * Purchase of bad loans. The Central Bank purchased bad loans from private banks, which made future repurchase commitments. The loans were extended at subsidized interest rates. * Domestic debt rescheduling. The Central Bank rescheduled domes- tic debts and financed this operation through loans at negative spreads. s Persen. exchange mnte. After the massive 1982-83 devaluations, which raised the domestic-currency cost of foreign debt service, private debtors with foreign-currency-denominated liabilit;es were allowed to purchase, at a subsidized exchange rate, for'Agn exchange for servicing their debt. * Exchange rate guarantees. The Central Bank purchased foreign exchange with the commitment to sell it back after one year at the 178 Cile: Fiscal Adjustment and Successful Performance purchase exchange rate adjusted by domestic inflation. The mas- sive 1982-85 real exchange rate depreciation caused correspond- ing losses to the Central Bank. * Foreign exchange capital losses. The acquisition of private external debt pushed the Central Bank into a net foreign debtor position, which implied capital losses as a result of the large real devaluations. Table 4.5 reports estimated capitalized losses incurred by the Cen- tral Bank as a consequence of quasi-fiscal operations. Losses were estimated as the present value-as of December 1989-of Central Bank disbursements net of the value of assets acquired through quasi- fiscal operations. The total estimated loss is about US$9.0 billion, almost 40 percent of 1989 GDP. The equivalent cash flow is about US$540 million, taking into account an average 6 percent interest rate on Central Bank liabilities. The Central Bank financed most quasi-fiscal operations by issuing domestic debt. As a xesult, its outstanding domestic debt increased by US$5.7 billion during the 198249 period, while its external debt inaeased by US$4.0 billion. To compensate for the increase in Central Bank liabilities, during 1983-86 the general government transferred to the Central Bank approximately US$7.2 billion in low-interest trea- sury bonds. In this way the government recognized that the Central Bank was its financial agent during the financial crisis. The Central Bank still exhibits a significant cash-flow deficit (table 4.6) as a result of its holdings of low-yield treasury bonds, which pay a minimum real return of 2 percent per year, and its high-interest liabilities, which cost an average real 6 percent interest per year- The deficit is financed pre dominantly by the Bank's seigniorage revenue. Table 4.5. Capitalized Value of Central Bank Losses from Quasi- FLiscal Operations as of December 1989, Chile (milDions of U.S. dollars) Prsentvalueof Esimaed Quasi-fiscal opendion disburse, rfs value ofasses Net loss Credit to bankrupt financial institutions 1,930 0 1,930 Purchase of bad loans 3,114 2,513 601 Domestic debt rescheduling L570 1,180 390 Preferential exchange rate -3,320 0 3,320 Exchange rate guarantees 1,585 0 1,585 Foreign exchange capital losses 1,227 0 1,227 Total 12746 3,693 9,053 Soure: Eyzaguire and Larrafiaga 1990. Jorge Marshall and Klaus Schmidt-HebbrJ 179 Table 4.6 Estimated Central Bank Deficit, Chile, 1991 (percent) lItem Share of GOP Expenditure 2.95 Domestic liabilities 1.66 External liabilities 0.83 Dollar-denominated domestic liabilities 0.48 Revenue 1.05 Intemational reserves 0.58 Loans to the financial sector 0.47 Treasury bonds 0.43 Deficit 1.47 Souzr F.yzaguirre and Lrranhaga 1990. How Sutainable Is the Deficit? Whereas the nonfinancial pubLic sector showed an average annual surplus of 0.9 percent of GDP over 1990-92, the Central Bank of Chile reported in 1991 a deficit dose to 1.5 percent of CDP (table 4.6). A 0.6 percent total public sector deficit-approximaLeJy equivalent to a pri- mary deficit of 3.5 percent of CDP for the total pub-lic sector-seems manageable. However, public finances could be placed under stress by the large total public debt, which increased from US$7.9 billion in 1981 to more than US$22 billion in 1989. This figure indudes both domestic and external debt, held by the financial and nonfinancial public sectors. To Ilustrate the implications of public debt for public finances, we perform a simple test of deficit sustainability, based on a condition of a nonincreasing debt-to-Gp ratio imposed on the consolidated total public sector budget constraint.9 If the deficit is financed through issuing money, domestic debt, or foreign debt, financiang of the pri- mary deficit as a fraction of CDP (d) can be written as: (4.1) d=m+m(r+n)+b-b(r-n)+b*-b*r*-naL) where r is the inflation rate, n is the GDP growth rate, r is the real interest rate paid on domestic debt, r* is the real interest paid on foreign debt, and iLis the rate of real devaluation.'0 The outstanding stocks of base money, domestic public debt, and external public debt as shares of CDP are denoted by m, b, and b*, respectively. A dot over a var; ble denotes thLe time derivative. Table 4J7 reports simulation results for the primary public surplus levels required for (a) maintaing a constant debt-to-GDP ratio and (b) decreasing this ratio by 5 percentage points of GOP per year. Two macroeconomic scenarios-a base scenario and an adverse case-are 180 Chile: Fisagl Adjustment and Successfut PejormnWce Table 4.7. Consolidated Total Public Sector Primary Surpluses under Alternative Public Debt Paths (percent) Redl intenzst Real GOP Primnay surplus Item ate growth as share of GDP Canstant ratio of public debt to GDP Base scenario 6.0 5.0 -0.8 Adverse scenario 10.0 2.0 3.6 Decrease in 6atio of public debt to GDP by 5 percent peryear Base scenario 6.0 5.0 4.2 Adverse scenario 10.0 2.0 8.6 Naot Simulations assume 15 percent inflation, a constant real exchange rate, and domestic real interest rates equal to foreign real interest rates. considered for each altemative. Under the base scenario, with output growing at 5 percent and with a real interest rate of 6 percent, the public sector can maintain its debt ratio by running a primary deficit of 0.8 percent of GDP, equivalent to a 4.8 percent total deficit. Alter- natively, the public sector is able to reduce significantly its debt-to- GDP ratio-by 5 percentage points of GDP per year-when running a primary surplus of 4.2 percent of GDP. Under an adverse macro- economic scenario, with 2 percent growth and real interest rates at 10 percent, the public sector would have to generate a primary surplus of 3.6 percent of GDP to maintain a constant debt-to-GDP ratio and would need a primary surplus of 8.6 percent of GDP to reduce the debt-to-cDP ratio. The implication is that the current deficit levels of the total consoli- dated public sector (0.6 percent total, 3.5 percent primary) are sus- tamable under current macroeconomic conditions. Ratios of public debt to GDP could be maintained at their current levels even under more adverse conditions. Deficit Finartcng, Interest Rates, and Inflation This section analyzes the effect of the public deficit on inflation and interest rates in Chile under alternative financing strategies. The per- formance of the key rates and financial aggregates is depicted in figures 4.1 and 4.2. Interest and inflation rates have followed a rather similar pattern since the mid-1970s. Their relative stability-by Latin American standards-contrasts with the erratic behavior of money an" narticularly with the upward trend of domestic public debt. The latter aggregate, starbng at a stable and low level of about 10 percent of quarterly GDP before the 1982 crisis, rose sharply, to 80 percent of forge Marshall and Klaus Schmidt-Hebbd 181 Figure 4.1. Interest and Inflation Rates, Chile Real interest rate, 1975:.1-1988:4 Percentage per month 10 8 6 4- 2 0- -2- -4- -6 -7 -- ~I . .. . . . . . , . .. . .. ., . . . . . . . . . . . 19751 76:1 77:1 78:1 79:1 80:1 81:1 82:1 83:1 84:1 85:1 86:1 87:1 88:1 Year and quarter hIfladion rate, 19752-1988:4 Percentage per month 70- 60 50 40- 30 - 20- 10 _ 0- 1975:2 76:1 727:1 78:1 79,1 80:1 81:1 821 83:1 84:1 85:1. 86:1 87:1 88:1 Year and quarter Source: See appendix 4A. 182 Chile: Fiscal Adjustment and Successful Performance Figure 4-2- Money and Public Debt, Chile Money as a share of GDP, 1975:1-1988:4 Percent 30 28- 26- 24- 22 20 16- 14- 19-75:1 76:1 -77:.1 75:1 -79:1 80:1 81:.1 82:1 83:1 84-1 85:1 86:1 57:1 88:1 Year and quarter Public debt as a share of GDP, 1981:.4-1988:4 Percent 200- 180 160- 140- 120- 100 60- 40- 19S1:4 82. 83:1 84:1 85:1 86:1 87:1 88:1i YearTand quarter Source: See appendix 4A. Jorge Mrshall and Klaus Schmidt-Hebbel 183 GDP in the first quarter of 1983, and reached a peak of 190 percent of GDP in the first quarter of 1986, after wich it started to decine. As was true for Latin America at large, Chile overborrowed during 1978-81 and faced an adverse external environment in the early 1980s. This triggered a crisis of vast proportions, forcing major adjust- ments in macroeconomic policy in general and fiscal policy in particu- lar. What is unique to the Chilean experience is that the 1982 crisis and the decline in external financing hit a stabilized public sector that had shown significant surpluses before 1982. This favorable initial condition allowed the government (induding the Central Bank) to finance the deterioration in the public sector budget after the debt crisis without destabflizing the economy. However, both the 1981 pension reform and the quasi-fiscal operations of the Central Bank in 1982485 raised measured total public sector deficits. To a large extent these deficits were financed by increased domestic public debt, a liability demanded by the private social secuinty system put in place by the pension reform. The empirical nmodel presented in this section emphasizes these specific features of the post-1982 fiscal adjustment undertaken in Chile, introducing modifications to the framework pro- posed by Easterly (1989) on the basis of a model developed by Mujica (1990). Public Deficit Financing and Private Sector Portfolios A growing literature on fiscal deficits deals with the relation between fiscal deficits and economic growth, the sensitivity of macroeconomic variables to changes in the public sector budget, and the channels through which fiscal deficts impinge on the economy. This literature emphasizes the implications of an mcreased reliance an domestic financing of public deficits as a xesult of the sharp reduction in exter- nal financing in most highly indebted developing countries (see, for instance, Buiter 1988; Dombusch 1985a, 1985b, 1988; and van Wijn- bergen 1989). The point of departure of the model is the basic government financ- ing identity, which may be stated by rewriting equation 4.1 for the nominal total consolidated public sector deficit: (4.2) Gt + it41Dt + 4ti,etDtr= fIt + D, + 4b; where G is the prmary deficit, D is the stock of domestic public debt, Dt is the stock of foreign public debt, H is the stock of high-powered money, i is the domestic nominal interest rate, i* is the foreign nomi- nal interest rate, and e is the nominal exchange rate. The identity states that the total public deficit, comprising the priary deficit and interest payments, is financed by changes in foreign and domestic public liability holdings. However, since for most developing coun- 184 Chile: FisCaI Adjustment and Successful PerformWance tries after 1982 external borrowing has been quasi-exogenous, an inaease in the primary deficit must be financed basically by issuing bonds or by creating high-powered money. Therefore the model emphasizes the tradeoff associated with the choice between money and domestic debt financing for given levels of the primary deficit and foreign financing. The choice has implications for both real and nominal variables. This section focuses on inflation and interest rates. The model uses a standard portfolio framework for the private sector's holdings of money (MI), public debt, and foreign interest- bearing assets. Focusing on the demands for the first two assets and assuming short-run stock adjustment costs, the short-run equilibrium values of real balances and real public bonds holdings can be ex- pressed as follows (see appendix 4.2 for a formal derivation): (4.3) Mt - Pt = 'k + *lYt + a2zt + 03(4t+ 6,) + 04(Mt-n - Pt-1) + 45St + 06wt (4.4) dt-pt = 4o + $134 + 42t + 3(4+ 61) + 4(dtjd -lPt1) + 45dt + 46Wt where (in addition to the variables defined earlier) m is the nominal money stock, w is Teal financial wealth, y is real GDP, p is the domestic price level, d is the nominal public debt stock, S is a nominal supply shock (defined as the difference between the actual and expected rates of change of money), e is the nominal exchange rate, and a is the expected change in the exchange rate. All variables except i, i*, and a are expressed in natural logarithms. Subscripts t and t-1 denote time. The O, and 4j coefficients are nonlinear combinations of the structural coefficients, as shown in appendix 4.2. The money shock variable S captures the shock-absorber hypoth- esis proposed by Canr and Darby (1981), whereby unexpected mone- tary supply shocks cause unplanned transitory monetary hoarding. A similar term is included in the specification of the public debt market equilibrium condition: the actual rate of change in the stock of nomi- nal public debt (a) is hypothesized to affect the level of short-term public bond holdings. This variable may reflect the demand for public debt by the private social security corporations since 1981, which accommodated-at least in part-the financing of deficits originating in the social security reform and the quasi-fiscal operations of the Central Bank. This partial demand accommodation to the large sup- ply of public debt implies that the interest rate would not rise by as much as in the absence of demand accommodation. Equations 4.3 and 4.4 are quite general because they do not merely incorporate features of open and dosed economies but also allow for the possibility of distinguishing between the short- and long-run Jorge Marshall and Klaus Sdrnmidt-Hebbel 125 effects of shocks in the money and public bond markets by assuming partial short-term adjustment in both markets. Since the nominal stocks of money and domestic public debt are determined by the government budget equation (4.2) and real wealth is exogenous, equations 4.3 and 4.4 describe an adjustment mechanism for domes- tic prices and interest rates. Inverting them allows the derivation of the following reduced-form equations for the two latter variables (where the U2 and w- coefficients are defined as in appendix 4.2): (4.5) it = Q + %lyt + 02('t + 4) + 03(mt-I - Pt-1) + 04(d-1 - -i) + tI6m, + n6d + %d,1dt + 28S, + 09w, (4.6) Pt = 1r0 + 7r1 + TiYt2 + r + 5t) + 7r3(m,l - Pt-1) + 74(d,t--Pt-1) + 7r5Mt + 7r6d, + ir7c4.1 + .8S, + 7r9wt. Empirical Results This section presents the results obtained from the estimation of the short-run demands for real balances and real public debt, based on quarterly data for the Chilean economy."' We also show results for the reduced-form equations for the interest rate and the price level in order to assess the ability of the model to descnbe inflation and inter- est rate paths directly. The estinated structural asset demands correspond to equations 4.3 and 4.4, amended by indcuding exponential time trends, as re- flected by the first lines of tables 4.8 and 4.9 The results for short-run money demand in table 4.8 report coefficients with the expected signs, and almost all are significant at conventional levels. In particu- lar, the significance levels of the unexpected nominal monetary shock (Se) and the lagged dependent variable (m_1 - pF-i) suggest that real balances have been highly sensitive to money supply shocks and that the effect of any variable on holdings of real balances is spread over a significant period. Finally, the negative coefficient of the time variable is related to the strong financial innovations that took-place between the mid-1970s and the late 1980s as a consequence of financial liberal- ization. Taken as a whole, these results are consistent with the hypotheses of stock adjustment and unexpected monetary shocks implied by the model and suggest that conventional money demand models cannot suffciently explain tne behavior of real balances dur- ing this period in Chile. The results for the short-run demand for public debt in table 4.9 are consistent with some of the hypotheses advanced earlier. For exam- 186 Chile: Fiscal Adjushnent and Success Performance Table 4.8. Estimation Results for Money Demand, Chile, 1976:1-1988:4 Method Variable OLS AR NTLS 40 -1.582 -1.959 (1.3) (1.5) #1 0.415 0.486 (3.5) (3.9) -2 -3.101 -3.595 (4.1) (4.6) 43 -0.165 -0.102 (°09) (0.6) 04 0.7M 0.660 (15.2) (12.6) 45 0.148 0.151 (3.5) (4.1) r, -0.004 -0.004 (4.1) (4.0) ci 0.298 0.340 a1 1.390 1.430 1.394 - - az -10.410 -10.590 -10.414 af3 -0.550 -0.300 -0.555 Adjusted R2 0.972 0.979 0.972 Durbin's h 1.947 0.570 1.947 p 0.285 Note: The equation is: Mt - Pt =, + Oyt + j,2! + 43(it*+ a) + 44(mt,. - p,tJ) + 0<;, + rgnmE. The estimation methods were ordinary least squares (oLs), maximum likelihood estimation with first-order autoregressive residuals (ARI), and nonlinear least squares (Nra). Coefficients a, az2, and a3 are the long-run asset demand coefficients, and parameter a denotes the short-rnu asset adjustment parameter (see appendix 4.2). Numbers in parentheses are t-values, and p is the residual first-order correlation coeffi- cient. A blank denotes the omission of the specific variable from the regression. ple, slow stock adjustment is reflected by the significant coefficient of lagged public debt holdings. However, the actual rate of change in the stock of nominal public debt was not significant in preliminary regres- sions; hence table 4.9 reports the estimates of equation 4.4 without this variable. Output has a negative effect on debt demand-a result that reflects more the peculiarity of the sample period than any spe- cific feature of the Chilean economy. Another interesting resut refers to the relative size of the interest rate semielasticities of money and public bond holdings and the speed of adjustment in the money and lorge Marshall and Klaus Schmidt-Hebbel 187 Table 4.9. Estimation Results for the Demand for Public Debt, Chile, 1983:1-1988:4 Medhod Variable OLS ARI NLS *0 55.005 51.932 (3.8) (3.7) *1 -4.249 -3.956 (3.3) (3.2) 42 - 19.024 19.175 (2.6) (2.7) $3 -2816 -2.820 (2.1) (2.1) $4 0.305 0.281 (4.6) (4.2) 45 0.081 -0.082 (4.5) (4.4) pA .0.695 0.719 1is -6.115 -5.502 -6.117 sv ;27.380 26.668 27.381 I3 -4.052 -3.922 -4.054 Adjusted R2 0.872 0.816 0.872 Durbin's h 0.375 0.570 p 0.285 Note: The equation is: dt - Pt = Jbo + vlY + 42i, + 43(I;+ 6,) + 4'4(d,..r - pt-i) + %.d, + r2TnME. For methods and definitions, see table 4.8. A blank denotes the omission of the specific variable from the regression. public bond markets. The estimates confirm a larger impact on infla- tion when fiscal deficits are money financed than when they are debt financed. Table 4.10 reports the reduced-form equation results for the interest rate and price level equations (4.5 and 4.6), on the basis of the 1983.1- 1988.4 sample. All variables with an unambiguous effect on the inter- est rate and the price level present the expected signs. The ordy exception is lagged real public bond holdings, which has the wrong sign in both the interest rate and the price level equations but is not significant.12 In addition, for the variables with a priori ambiguous signs in the reduced-form equations, the signs of the estimated coeffi- cients coincide in all cases with the signs implied by the results of the structural equations. 188 Chile: Fiscal Adjustnent and Successful Performance Table 4.10. Estimation Results for Reduced-Form Equations for the Rates of Interest and Inflation, Ordinary Least Squares Method,', Chile, 1983:2-1988:4 Independert variable Initerest rate Price Jewl Constant -1.170 -8.810 (2.5) (1.3) y, . 0.084 0.170 (1.9) (0.3) ij*+ 8, 0.106 0.632 (4.8) (1.9) Mt-i ~ Pi-I - 0.055 -0.110 (2.8) (0.4) di-, - pl-I 0.00026 0.064 (0.03) (0.6) m, - -0.014 0514 (-1.9) (2.0) di 0.004 0.182 (0.5) (1.6) St 0.0039 -0.068 (0.5) (0.6) Adjusted R2 0.757 0.977 Durbin-Watson 1.72 150 Note: Figures in parentheses are t-values. Figure 4.3 shows the steady-state relationship between inflation rates and the inflation-tax revenue derived from the estimated money demand coefficients and assuming no growth in exogenous variables. At an inflation rate of 2 percent per month, the inflation tax is about 0.5 percent of cDp. Little tax gain is obtained from higher inflation. Raising the inflation rate from 2 to 20 percent per mont-the revenue-maximing rate-increases seigniorage revenue by only 1.5 percentage points of GCDP. Simulation Results The results reported in the first columns of tables 4.8 and 4.9 were used to simulate the impact on inflation and interest rates of a tempo- rary increase in the fiscal deficit under alternative strategies of domes- tic financing. To simulate the twelve-quarter effects of a transitory one-quarter public sector deficit, we assume that the economy is at an initial steady-state equilibrium with a mnoney-to-GOP ratio of 0.2 and a debt-to-cop ratio of 1.0.13 Table 4.11 reports the short- and long-term effects of money and debt-financed one-quarter fiscal deficits of 2 percent of cGD.. The Jorge Marshtall and Klaus Sc&midt-Hebbel 189 Figure 4.3. Seigniorage Revenue, Chile Percentage of GDI' 2.2 2.0- 1t6 - 1.6 1.2 - 1.0 0.8 - 0.6- 0.4 /- 0.2 0 1 4 7 10 13 16 19 22 25 28 31 34 37 40 Inflation rate (percent) Sourcec Auftrs' calculations. money-financed fiscal shock causes a fall in the interest rate of 0.39 percentage point in the first period and a long-run decline of about 0.27 percentage point. The effect on inflation is about 7 percent in both the short and the long run. Hence most of the consequences of a money-financed deficit are felt during the first period. In addition, since the increase in the price level never matches the increase in the stock of money required to finance the fiscal shock, real monetary holdings are affected. The dynamic behavior of the interest and infla- tion rates in response to this money-financed deficit is shown in figure 4.4. Table 4.11. Simulation Results of aTemporary Fiscal Deficit, Chile Item Short nun Long run Effects of a money-financed increase in the fiscal deficit Change in interest rate (percentage points) -0.39 -0.27 Change in the price level (100 d log p) 7.33 7.25 Effects of a public debt-financed increase in the fiscal diit Change in interest rate (percentage points) 0.09 0.05 Change in price level (100 d log p) 0.28 0.55 190 Chile: Fiscal Adjustment and Successfui Performance Figure 4.4. Simulation Effects of a Money-Financed Deficit on Interest and Inflation Rates, Chile Interest rate Percent per month 0.1 -0.3 -0.4 - -0.5 - 0 1 2 3 4 5 6 7 8 9 10 11 12 Quarter Inflation rate Pernent per month 7- 6 5 - 4 3- 2- 1- I 0- l_ - , - C * * * l -1-- I I I I I I I I III-I 0 1 2 3 4 5 6 7 8 9 10. 11 12 13 14 15 16 Quarter Jorge Marshall and Mauws Schmidt-Hebbel 191 A temporary deficit of 2 percent of GPD financed by public debt raises slightly both the interest rate and the price level (see figure 4.5 for the dynamic response). The increase in the price level required to maintain the portfolio equilibrium is much lower now than und.er the money-financed deficit because of the higher interest rate. To sumrnarize, these results suggest that the main impact of a change in the fiscal deficit occurs in the first quarter. The only excep- tion is the gradual rise in prices when the fiscal deficit is financed by domestic borrowing. The results also suggest that debt-financed defi- cits have a small effect on interest and inflation rates-an outcome that reflects the specific features of the estimation sample penrod. The net transfer payments to the private sector implied by the social secu- rity reform and the quasi-fiscal operations of the Central Bank during the 1980s, which were accommodated by the rising demand for public debt paper by the new private social security corporations, are the main reasons for the small impact of debt-financed fiscal deficits on interest rates and prices. Crowding-Out and Crowding-In of Private Consumption and Private Investm.ent by the Public Sector This section assesses the impact of the public sector on private sector spending in Chile. The focus is on the sensitivity of private consump- tion and investment to fiscal vanrables, in addition to the indirect effects of these variables through interest rates, inflation, and private disposable income. H-ow private saving and capital formation are affected by fiscal policies has significant implications for both short- run stabili:zation issues and long-run growth prospects. Figure 4.6 shows the evolution of the ratios of private consumption to private disposable income and private investment to GDP during the past three decades. It is not surprising that private consumption has tended to be countercydlical-in other words, consumption does not decline as strongly as does private disposable income during recessions such as those of 1975 and 1982-83. Two high-consumption episodes coincided with the fiscal-monetary expansion of the early 1970s and the foreign-financed private spending boom of ten years later. A distinct structural change took place after 1984: private con- sumption as a share of disposable income reached the lowest values in three decades and remained at that level throughout the 1985-88 recovery-an important counterpart to the significant current account adjustment that took place in the 1980s. The private investment rate showed wild swings during the 1960- 88 period in Chile. After reaching a histonrcal low of 5 percent of GOP during the early 1970s, private investment began a continuous recov- ery and eventually boomed in the early 1980s. Private capital forma-. 192 Chile: Fiscal Adjustment and Successful Perfonnance Figure 4.5. Simulation Effects of a Debt-Financed O-cficit on Interest And Inflation Rates, Chile Interest rate Percent per month 0.10- 0.09 0.08 0.06 - {_ _ 0.05- 0.04- 0.03- 0.02 - 0.01 - o- l I I l I l I I 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Quarter Inflation rate Percent per month 0.28- 024- 020- 0.16- 0.12- 0.08 0.04- 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Quarter Jorge Marshall and Klaus Sd:,midt-Hebbel 193 Figure 4.6. Private Consumption and Investment, Chile Ratio of private consumption to private disposable income, 1960-88 Ratio 1.08 1.04 1.00 0.96 0.92 0.88 0.84 1960 1965 1970 1975- 1980 1985 1988 Ratio of private investment to GDP, 1961-8 Ratio 0.15 - 0.14 - 0.13 - 0.12 0191 - l6 90 -17 90l6 l8 0.08 S 0.07- 0.06 0.05 0.04 - I ii ii 11 1 1 1 1 1 1 1 1961 1965 1970 1975 1980 71985 1988 Source: See appendix 4A. 194 Chile: Fiscal Adjustment and Successul Performance tion then took a dive that coincided with the 1982-83 recession and the subsequent policy uncertainty. Investment remained low in 1984 86 before picking up again in 1987 and 1988 and growing to remarka- bly high levels-about 20 percent-in the early 1990s. Private Consumption The direct effects of fiscal policies on consumption or saving operate through public saving (or the deficit) and its composition. If the strin- gent conditions required for Ricardian equivalence are satisfied, a permanent rise in public saving-if it comes about via lower public spending-is exactly offset by an increase in consumption. Because disposable income does not change, the reduction in private saving matches the increase in public saving. A rise in public saving does not affect consumption at all if it is a result of higher taxes. But as dispos- able income is reduced by the size of the tax, the reduction in private saving also matches the increase in public saving (the latter being equivalent in macroeconomic terms to issuing additional public debt). The opposite results are predicted by the Keynesian hypothesis (and the permanent income hypothesis without Ricardian equivalence): current (permanent) taxes matter for consumption, not current (per- manent) public spending levels.'4 The indirect effects of public (fiscal-monetary) policies on consump- tion operate through the impact of public deficits and their financing on major relative prices-real interest rates, inflation, and the real exchange rate-that affect private consumption. Real interest rates affect private consumption only if the substitution, income, and human wealth effects do not offset each other. But this seems not to be the case, to judge from the growing evidence showing that con- sumption is not sensitive to real interest rates)'5 Although the first- order effect of inflation is on the composition of the savings portfolio and not on saving or consumption flows, its second-order effects may reduce saving if inflation is associated with capital flight or raises precautionary saving as a result of higher uncertainty. An expected devaluation reduces the consumption-based real interest rate and leads to capital flight; the first effect could increase actual consump- tion, while capital flight raises the measured consumption-to-income ratio if it reduces measured income.16 Finally, substitution effects from fiscal policies could arise if fiscal spending on privately appro- priated services such as education, health, and nutrition, and direct transfers to households reduce the need for private consumption expenditure on these items, leading to possible declines in aggregate private consumption (see Easterly, Rodriguez, and Schmidt-Hebbel 1989)_ Jorge Marshall and Klaus Schmidt-Hebbel 195 Equation 4.7 specifies the ratio of private consumption to private disposable income, following the previous discussion and a frame- work developed by Corbo and Schmidt-Hebbel (1991):17 7 V -CIO AlDy + 02DY +3dM4 s + p4t + 5 DYpt DYp, pDYvt PCv CrYrR, + ,_i FSt where Cp is private consumption expenditure, DYp is current private disposable income, ozYpv is permanent private disposable income, PS' is permanent public saving, r, is the consumption-based real interest rate, vr is consumption inflation, P13m and P¢t are deflators for imported and domestic private consumption goods, respectively, crrR is the sum of public expenditure on privately appropriated services and direct transfers to consumers, H is base money, and FS is foreign saving. The coefficients -, 0l, and 2 are associated with the correspond- ing Keynesian, permanent income, and Ricardian hypotheses.18 Note, however, that (2 could be positive if the public sector has pref- erential access to bank credit, forcing pnrvate consumption (and investment) to adjust residualy. This is a form of institutional or direct crowding-out of private saving by public sector saving that is obviously unrelated to the Ricardian notion of unconstrained, forward-lookilg private consumers who internalize the govern- ment's intertemporal budget constraint. Expected signs of the coeffi- cients are: o, 01 0, 07, S 8> 0; 06 < 0; 03, 04. 65 0 0-19 Equation 4.7 was tested using annual data for the 1960-88 period, which is the sample period used most frequently in this section and the next. This period is characterized by significant structural breaks and changes in policy regime, which, because of their intensity, necessitate caution in the interpretation of the statistical results pre- sented below. Table 4.12 presents a selected number of empirical results for the consumption function. In general, the results for the static- expectations version dominate statistically those related to the partial perfect-foresight case.20 The average Chilean consumer appears to be partly KCeynesian, partly permanent-income forward-looking: under the static version marginal propensity to consume out of current pri- vate disposable income is approximately 0.70, and marginal propen- sity to consume out of permanent income is about 0.30. Under the partial perfect-foresight version the corresponding propensities are of a similar magnitude, dose to 0.50. It is interesting to note that con- sumers do not react to permanent public saving; that is, the consoli- dated public sector deficit (less public investment) has no effect on private consumption. The policy implication is clear: an increase in Table 4.12. Private Consumption Estimation Results, Chile, 1960-88 PDYP PSf PCM CPTR H FS Durbin- Equation C DYP DYP rc 7rc DY-, Y DYp p R2A Watson Static expectationts I OLS 0.71 0.28 0.18 0.05 -0.02 -0.14 0.12 0.29 0.24 0.79. 1.77 (10.4) (4.1) (0.8) (L7) (-0.5) (-3.0) (0.4) (3.2) (1.7) 2 OLS 0.74 0.25 0.36 -0.13 0.23 0.37 0.72 1.52 (9.8) (3.2) (1.6) (-5.2) (2.6) (2.5) 3 ML 0.66 0.26 0.29 -0.08 0.45 0.41 0.59 0.90 1.75 (10.0) (3.7) (1.2) (-2.2) (3.8) (2.8) (3.5) 4 OLS 0.72 0.27 -0.13 0.24 0.48 0.70 1.38 (9.4) (3.5) (-5.0) (2.6) (3.6) 5 ML 0.67 0.27 -0.09 0.44 0.42 0.57 0.89 1.50 (10.0) (3.9) (-2.4) (3.7) (2.8) (13.4) m Pnrtinl perfecdforesiglit 6 OLS 0.49 0.49 0.002 0.0004 0.08 -0.19 0.04 0.28 0.87 0.63 1.32 (2.6) (2.8) (0.005) (0.0008) (1.2) (-3.6) (0.09) (1.5) (4.0) 7 ML 0.38 0.50 0.38 -0.03 0.03 -0.11 0.24 0.52 0.68 0.67 0.89 1.39 (2.9) (3.6) (0.8) (-0.6) (0.4) (-2.1) (0.5) (3.0) (4.1) (4.1) 8 ML 0.42 0.48 0.14 -0.10 0.58 . 0.71 0.62 0.89 1.41 (3.4) (3.7) (0.4) (-2.4) (4.8) (4.6) (3.9) 9 ML 0.43 0.48 -0.10 0.57 0.73 0.60 0.90 1.43 (3.5) (3.80) (-2.6) (4.9) (4.9) (3.7) Note: The dependent variable Is the ratio of private consumption to private disposable income (CpDY). The estimation methods were ordinary least squares (oLS) and maximum likelihood (ML). Figures in parentheses are I-statistics; p is the first-order reslLai correlation coefficient. R2A is adjusted R'. A blank denotes the omission of the specific variable from the regression. Jorge Marshall and Klaus Schmidt-Hebbel 197 public sector saving has a strong effect on national savings, as it is offset only in part by a decline in private saving. Of all relevant prices, only the real exchange rate (given by the relative prices of the imported and domestic consumption goods com- ponents of aggregate consumption) affects private consumption. A 10 percent real devaluation, which would affect directly this relative price, would reduce aggregate consumption by approximately 1 per- centage point of private disposable income. But neither the real inter- est rate nor the inflation rate has any consistently significant separate effects on private consumption. This outcome corroborates earlier studies on the interest insensitivity of consumption in Chile (Arrau 1989; Schmidt-Hebbel 1987, 1988). In addition, there is no evidence of a substitution effect of public spending on privately appropriate ser- vices (such as education and health) or of transfers on aggregate consumption. Monetary holdings and foreign saving flows exert sig- nificant positive influences on private consumption. Whereas higher base money holdings tend to relax domestic borrowing constraints faced by consumers, higher foreign saving is associated with weaker foreign resource constraints. The effect of weaker foreign constraints on private consumption is very strong: from these results one can infer that between 42 and 71 percent of the dramatic current account correction that took place after 1981 was borne by private consumers. The policy implications of our results are dear. First, there is no evidence of a direct effect of the public deficit (taken separately from the influence of taxes through disposable income) on private con- sumption; consumers in Chile are neither Ricardian nor directly crowded-out by public spending. A transitory defict reduction financed by a transitory tax hike reduces private disposable income and hence private consumption and saving. If the tax hike is perma- nent, private consumption will take the brunt of the reduction in (current and permanent) private disposable income, without private saving being affected significantly. Second, domestic financing of public defiats has no indect effects on consumption through changes in the real interest and inflation rates. However, if inflation reduces real base money, private consump- tion will be affected. Foreign adjustments have strong effects on private consumption in Chile. The combination of lower foreign saving and a higher real exchange rate has a strong negative effect on private con- sumption expenditure, as evidenced during the post-1982 adjustment penod. Private Investmnent Following Easterly, 'Zodriguez, and Schmidt-Hebbel (1989) and empirical investment studies of Chile by Solimano (1992) and of Table 4,13. Private Investment Estimation Results, Chile, 1961-88 PK Ki - .K1 PRO H_1 FS Durun- Equaliont C D60s D7Os PUCK Y Y Y VUCK VY p R2A Watson Static expectationis I OLS 0.22 0.06 -0.10 0.12 -0.05 -0.11 0.17 -0.11 -0.07 -0.01 -0.01 0.62 1.29 (3.5) (0.8) (-1.5) (2.3) (-1.5) (-2.2) (2.0) (-1.4) (-0.5) (-1.8) (-0.1) 2 ML(AR1) 0.17 -0.03 0.05 0.04 -0.06 -0.07 0.26 -0.11 0.05 -0.002 -0.27 0.79 0.59 1.93 (2.7) (-0.5) (0.1) (1.1) (-1.8) (-16) (2.8) (-1.6) (0.4) (-0.) (-1.3) (5.9) 3 ML (AR1) 0.10 -0.08 0.05 0.01 -0.07 0.25 0.13 -0.26 0.81 0.51 1.72 (1.7) (-1.2) (1.6) (0.5) (-1.9) (2.5) (1.4) (-1.2) (6.2) Partial perfect foresigIt 4 OLS 0.21 -0.27 -0.09 0.04 0.04 -0.18 0.03 0.22 -0.11 -0,02 -0.15 0.71 1.19 (3.4) (-2.4) (-0.8) (0.8) (0.8) (-3.1) (0.3) (0.9) (-0.9) (-1.3) (-0.7) 5 ML (AR1) 0.18 -0.21 -0.15 0.16 -0.02 -0.12 0.16 -0.10 -0.06 -0.001 -0.18 0.63 0.67 1.81 (2.7) (-2.1) (-1.7) (3.1) (-0.2) (-2.0) (1.7) (-0.4) (-0-05) (-1.1) (-0-8) (3.2) 6 ML (AR1) 0.20 -0.09 -0.001 0.08 -0.12 0.17 0.17 -0.46 0.80 0.57 1.96 (2.5) (-1.0) (-0.002) (2.0) (-2.3) (1.8) (2.0) (-2.2) (5.9) Note: The dependent variable is the ratio of private investment to GDP (1p/. The estimation methods were ordinary least squares (oLS) and ravdmum likelihood (ML). Figures In parentheses are t-statistics; p is the first-order residual correlation coeffident; R2A is adjusted R2. A blank denotes the omission of the specific variable from the regiession. Jorge Marshall and Klauw Sclimidt-Hcbbd 199 Morocco by Schmidt-Hebbel and Miller (1992), we specify a behav- ioral function for private investment that will depend on profit, cost, and tax variables, as well as on liquidity constraints and risk determi- nants. To reduce the incidence of spurious correlation, we scale all nonstationary variables to GDP. Therefore we specify the following generic equation for the ratio of private investment to GDP: (4.8 8 -1PUCKI, y Pjiv PCEf 'M- PRO FC ye Y1 ye p11( y1 ' y, ' y ' y - (-)H (R) Q?) H- (+ H+ H+ -,_l -S, VUCK, ,VtY where lp is private fixed-capital investrment, Y is GDP, UCK is the user cost of capital, PucK is the estimated permanent UCK, PKp.l is the permanent estimate of private sector capital, PjP,,,IP. is the price ratio of imported and domestic private investment components, COT is corporate tax revenue, PCOT is the permanent estimate of COT, Kg is public sector capital stock, PRO is corporate profits, FC is banking credit flows to firms, H is base money, FS is foreign savin& vuca is the coefficient of variation of ucK, and WV is the coefficient of varia- tion of GDr. The expected signs of partial derivatives are denoted below each variable in the equation.21 The empirical results for the private investment functions are presented in tables 4.13 and 4.14, for both the static and partial perfect-foresight expectation alternatives for all relevant right-hand variables.22 In all the equations reported in table 4.13, the permanent user cost of capital (PUCK) has a positive effect, which is also significant in most cases. This seems to be a reflection of the extremely strong structural breaks that occurred during the 1960-88 period with regard to the functioning of the Chilean financial market, the determination and the levels of the real interest rate, and hence the dependence of investment on the cost of domestic financial capital. Two corrections were made to face this problem. First, Multiplicative dummies for PuciK were specified, separately for- the 1960s (D60s) and the early 1970s (D70s) (see table 4.13). Second, the original specification was tested for the 1976-88 period, which began after the 1974-75 financial interest liberalization (see table 4.14). However, the brevity of this more homogeneous period suggests the need for caution in interpret- ing the results of the table because of the small number of degrees of freedom. The coefficients of the period-specific dummies in table 4.13 (for 1961-88) tend to present the correct negative sign and are significant in some of the reported results. For example, regression 5 shows that Table 4.14. Private Investment Estimation Results, Chile, 1976-88 K I K I PRO H-_ FS Durbin- EqIOtiotn C PUCK F T 7 VUCK VY p RzA W4on Static expectations I OLS 0.35 0.05 -0.09 -0.14 0.02 0.51 -0.02 0.003 -0.26 0.88 3.69 (4.1) (0.7) (-1.8) (-2.2) (0.14) (1.7) (-0.1) (0.5) (-1.0) 2 OLS 0.36 -0.06 -0.17 0.26 0.56 0.82 2.95 (0,5) (-2.0) (-7.3) (1.1) (-2.3) o 3 ML (AR1) 0.37 -0.07 -0.18 0.36 -0.65 -0.51 0.95 2.19 cD (16.9) (-3.1) (-11.7) (1.9) (-3.3) (-1.9) Partial perfect foresight 4 OLS 0.52 0.16 -0.19 -0.07 -0.24 -0.32 -0.05 0.001 -0.27 0.86 2.91 (2.6) (1.3) (-1.0) (-0.5) (-1.2) (-0.5) (-0.2) (0.1) (-0.5) 5 ML (ARl) 0.45 0.10 -0.24 -0.49 -0,55 -0.37 0.95 2.09 (13.8) (3,2) (-11.1) (-2.5) (-2.9) (-; 1. 1) Nole: The dependent variable is the ratio of private investment to CDP ([IlY). The estimation methods were ordinary least squares (oLs) and maximum likelihood (ML). Pigures in parentheses are t-statistics; p is the first-order residual correlation coefficient; R2A Is adjusted R2. A blank denotes the omission of the specific variable from the regression. Jorge Marshall and Klaus Schmidt-Hebbel 201 while the coefficient of PUCK is still positive for the 1974-88 period (as reflected by its value of 0.16), its net influence is dose to zero for the early 1970s (0.16 for PucK minus 0.15 for D70s gives a net effect of 0.01) and is negative for the 1960s (0.15 for PucK minus 0.21 for D6Os gives a net effect of -0.06). However, for the 1976-88 period and the static-expectations alternative, the user cost of capital is negative and significant-although its coefficient is small-in the best results reported in table 4.14. The lagged ratio of capital to current output appears to be negative and highly significant in most reported results. This is not surprising: private investment is strongly procyclical. The consistently negative influence of public sector capital stock on private investment suggests that either public investment tends to substitute for private invest- ment or private investment is directly crowded out by public invest- ment because of the public sector's preferential access to domestic saving. Although direct crowding-out could have been present until the mid-1970s, the domestic financal lberalization rules it out for the post-1975 period. Whatever the reason, dearly there is no evidence for a complementarity of public and private capital that leads to crowding-in of private investment by public sector capital spending in Chile. However, to assess possible noise stemming from the indlu- sion of this variable, we also present results that omit it. The influence of liquidity constraint variables depends strongly on the period selected. This is very consistent with the change in the role of interest rates and hence of the user cost of capital during the 1960- 88 period: quantity constraints were of greater importance before domestic financial lberalization than afterward. Precisely the same is true of firm profits and foreign saving, which had a strong influence on investment during the 1960-88 period as a whole but disappeared from the scene during the 1976-88 period. Base money plays an ambiguous and unstable role: whereas in most subperiods and speci- fication alternatives it exerts a negative, nonsignificant influence, it presents a positive and significant, although minor, effect only in regression 3 (static expectations) for the 1976-88 period (see table 4.14). The coefficient of variation of the user cost has a negative, although usually insignificant, influence during the complete 1976-88 period, and its effect disappears during the second subperiod. Mudh more important is the variability of GDP, which affects private investment consistently and negatively during the entire period and particularly strongly in more recent years. From the empirical results we can draw a number of policy implica- tions. Financing of public deficits through debt would be felt increas- ingly by private investors in Chile because of its upward pressure on domestic real interest rates in unrepressed financial markets. The 202 Chile: Fiscal Adjustment and Successful Performance other side of financial liberalization is the weakening of both domestic and foreign liquidity constraints: firm profits and foreign saving were important investment deteminants before financial liberalization but vanished afterward. However, foreign adjustment has affected pri- vate investment negatively through the relative price of investment goods, which increased as a consequence of the real devaluations and so raised the user cost of capital. No evidence was found on public investment's aowding-in effects on private investment. The relevant variable-public capital stock- tends to reduce private capital formation, suggesting that public investment competes with private capital formation. Finally, investment is negatively affected by the variability of GDP. More stability in the external environment and in domestic macroeco- nomic policies influences private capital formation positively. Relative External Prices and the Eade Balance This section focuses on the impact of fiscal policies on the external sector, especially with regard to the trade defict and the real exchange rate. The framework applied here is derived in two steps. The first involves determination of the relative prices of exportables and importables from the equilibrium condition in the nontradable goods market; the second implies derivation of an equation for the trade balance. Figure 4.7 depicts the evolution of relative export and import prices and the ratio of the trade surplus to GDP during the past three decades in Chile. The price of exports relative to nontradables is strongly correlated with the international copper price, which is the main com- ponent of the country's external export prices and terms of trade. The price of imports relative to nontradables is a more useful measure of the real exchange rate relevant for decisions on spending and produc- tion (other than for copper). However, both measures of the real exchange rate tended to move together during 1975-88, a period of smaller terms of trade fluctuations than those experienced during 1960-75. The two periods of massive rzat exchange rate appreciations-the early 1970s and the early 1920s-coincde with the policy-induced spending frenzy of the Allende government and the foreign-financed private spending spree of Pinochet's plata dulce. Trade deficits, depicted in the lower panel of figure 4.7, reached their highest levels during those years-the most dramatic observation being the 10 per- cent of GDP trade deficit observed in 1981. The required current- account corrections after 1973 and 1981 led to drastic reversals of the trade deficits as a result of massive exchange rate depreciations and reductions in domestic spending. A distinct feature of the post-1984 JorgeMarsthal and Klaus Schmidt-Hebbel 203 Figure 4.7. Relative Export and Import Prices, Chile Export and import prices, 1960-88 Index (1977 1.0) 1.0 1.6 A 1.2 0.0 0.6 . OA 03- _,,,: .,,,,,,,, 1960 1965 1970 1975 1980 1985 1988 = 1XPt ratio of exportable to nontradable prices - = P/PN, ratO of importable to nontradable prices Note: Px/P., export prices; PMlPr4, import prices. Ratio of trade surplus to GDP, 1960-88 0.06 0.06 - 0.02 -0.02 -0.04 -0.06 -0.08 -0.10 -011I -- --r I I i 1960 1965 1970 1975 1980' 1985 1988 Source: See appendix 4.4. Table 4.15. Relative Export and Import Prices Estimation Results, Chile, 1960-88 TS ; DtVrbih- Eqitatioti c TT* im ~~~~ ~~~~~~~D75 p R2A Watsonl Relalivc export price 1 OLS -0.21 0.63 -0.04 2.48 3,64 0.51 0.86 1.29 (-1.2) (8.5) (-0.8) (4.0) (3.0) (4.5) 2 ML (ARI) 0.01 0.59 -0.05 2.22 2.50 0.38 0.52 0.77 1.76 (0.04) (6.9) (-0.9) (3.4) (1.7) (4.0) (3.0) 3 ML (AR1) -0.02 0.56 2.39 2.80 0.37 0.51 0.77 1.74 (-0.01) (7.7) (3.8) (2.0) (4.0) (3.0) Relative import price 4 OLS 1.20 -0.15 -0.09 2.10 -0.62 0.61 0.75 1.03 (6.7) (-2.0) (-1.9) (3.4) (-0.5) (5.4) 5 ML (AR1) 1.26 -0.14 -0.09 2.00 -1.09 0.47 0.62 0.72 1.65 (5,4) (-1.8) (-1.7) (3.3) (-0.8) (5.7) (3.9) 6 ML (R1) 1.09 -0.13 -0.09 2.04 0.47 0.60 0.72 1.62 (10.7) (-1.7) (-1.6) (3.5) (5.7) (3.6) 7 Mt (AR1) 1.08 -0.18 2.26 0.46 0.66 0.70 1.58 (9.8) (-2.5) (3.8) (5.5) (4.6) Note: The dependent variables are the relative export price (ex) and the relative import price (cei,). The estimation methods were ordinary least squares (OLs) and maximum likelihood (NL). FiSures In parentheses are t-statistics; p is the first-order residual correlation coefficient; R2A is adjusted R2, A blank denotes the omission of the specific variable from the regression. Jorge Marshall and Klaus Schmidt-Hebbel 205 external adjustment was the maintenance of massive trade surpluses, exceeding 6 percent of GDP in 1988. These surpluses are the counterpart of the significant decline in private consumption as a share of income (discussed in the preceding section) and a progressive correction in public sector deficits (discussed in the section on fiscal policy, above). Relative Prices of Exports and Im;ports Foliowing Easterly, Rodriguez, and Schmidt-Hebbel (1989) and Rodriguez, chapter 2 in this volume, the real exchange rate, defined as the relative price of tradable and nontradable goods, is derived from the continuous market-clearing condition for nontradable goods in the Salter-Swan-Corden-Dombusch tradition for the small open economy- Extension of this paradigm to a three-sector distinction between exportables, importables, and nontradables permits rewrit- ing of the market-learing condition as either one of the following functions for the relative pzices of exportables and importables923 (4.9) ex, aex T tMt,,-- Nt t t Mt'' - V (I Mt) ( .(4-10) t Pt t p e? !* Y. t t' \ {_) (C)() (C) (-) where ex and em are the relative prices of exportabIes and import- ables, respectively; Px, PM' and PN are the absolute pnrces of export- ables, importables, and nontradables; rr is foreign terms of trade; t4 is the ad valorem tariff rate for imports; TS is the current-price trade surplus; G is current-price (general) government spending (public consumption plus public investment); GN is current-price (general) government spending on nontradable goods; and Y is current-price cDP. This framework was applied to Chile using annual data for the 1960488 period. The results, reported in table 4.15 for both equations, tend to favor the relative export price interpretation as the relevant reduced-form equation for the equilibrium in the nontradable goods market. The external terms of trade are an important determinant of both export and import pnces, although their quantitative influence on export prices is much stronger. The average tariff rate has a nega- tive (although not significant) influence on both relative prices-a surprising result in the case of the import price, which should be positively affected by tariffs. The ratio of the trade surplus to output has an extremely high, consistent, and symmetric effect on both relative prices: an increase 206 Chile: Fiscal Adjustment and Successfid Performance in the trade surplus of 1 percent of CDP implies a 2.0 to 2.5 percent devaluation of the real exchange rate. Finally, aggregate government spending has a strongly positive impact on the relative prices of exports and a nonsignificant negative influence on the relative price of imports. In general, its influence could be of either sign.M' The Trade Balance The trade balance is both the difference between output and absorp- tion and the goods-markets counterpart of the accumulation of net foreign assets. In Rodriguez, chapter 2 in this volume, and Easterly, Rodriguez, and Schmidt-Hebbel (1989), the trade surplus is drectly related to the accumulation of net foreign assets (NFA). Private net foreign asset accumulation depends on the difference between desired and actual private NFA holdings, the former being replaced by its main determinants: the covered interest differential between domestic and foreign rates, domestic public debt, the terms of trade, and income. Publc NA accumuiation will reflect directy, with a negative sign, the (operational) public sector deficit for given stocks of domestic public debt and base money. Hence, under this foreign asset accumulation version the ecquation for the trade balance can be written as325 (4.11) TSt TS it i -(it (%E + i+te') NEAt1. Bt-1 00 -Y, Y ;1 1 +'(i,p-e;-iqe)' Y, ' YtI Y where TS is the cunrent-prce trade surplus, Yis current-price cDP, C is a constant, i is the average annual domestic nominal interest rate (using the average of active and passive rates), i is the average annual nomi- nal external interest rate paid on net foreign assets, El is the expected rate of nominal devaluation (defined below), B- is current-price domestic public sector debt (at the end of the preceding period), NEA-1 is current-price net foreign assets (total foreign debt less intemational reserves at the end of the preceding period), and OD, is the current- price operational public sector deficit. The expected signs of partial derivatives are denoted in parentheses below each variable. Another way of viewing the trade balance is to derive it from the macroeconomic equilibrium condition (see appendix 4.3 for a detailed derivation): (4.12) TS , (S C i-(+ Ef + itlt NFA1..1 Bt-1 ODgr YPI Yt Y 1 + (it*+ E( c + it*£3 :Yt t Y, Yt (+) pt GaCr p) rices. where YP IS permanent GDP at current pTices. Jorge Marshall and Klaus Schmidt-HebbeJ 207 Two differences arise between the asset-accurnulation version of the trade surplus (equation 4.11) and the macroeconomic equilibrium version (equation 4.12). First, the effect of domestic public debt, net foreign assets, and the interest rate differential are ambiguous in equation 4.12. Second, equation 4.12 includes the ratio of permanent to current output (an inverse measure of the business cycle). The subsequent empirical application will compare the relative relevance of these two approaches to the trade surplus for the Chilean economy. The empirical results are shown in table 4-16. Lines 1 and 3 present results for the complete specification. Of the right-hand determi- nants, the interest rate differential (IRD), total net foreign asset hold- ings, and the stoclk of domestic public debt holdings present positive, mostly significant signs for expectations under both versions. To cap- ture a possible break of the relationship between the trade surplus and the stock of net foreign asset holdings in 1981, the multiplicative dummy D81 for NFA was included for 1981. The positive signs of TwRD, NFA-1, and &B1 contradict the asset-accumulation version of the trade balance in equation 4.11 but are consistent with the ambiguous signs postulated by the alternative version of equation 4.12. The curTent exteral terms of trade have a consistently significant but low positive influence on the trade balance. The influence of the business cyde is not significant. A dummy for 197942 (D7982) sig- nals the particular regime of high access to foreign credit that dharac- terized those years. The trade deficit was 2 to 3 percentage points of GDP higher during the period, unhindered by either domestic or for- eign limits on foreign borrowing (see McNeIs and Schmidt-Hebbel 1993). Finally, the operational public sector deficit has a negative, although not significant, effect on the trade surplus under the static- expectations complete specification for the dependent variable (lne 1). To test for the influence of this variable when most other variables are omitted, we run the specification of line 2. Keeping in mind the variable exdusion bias, we conclude that an inaease in the public defict of I percentage point of GDP tends to reduce the Chilean trade surplus by a maximum 0.29 percentage point of GDP. Condusions and Policy Implications A number of important findings and lessons emerge from this study of the macroeconomic causes and consequences of budget sector defi- cts in Chile. The first, unavoidable observation concerns the wild gyrations of public sector deficits during the past two decades. In fact, each decade comprises a complete cycle of fiscal and macroeconomiic Table 4.16. Trade Surplus Estimation Results, Chile, 1960-88 NFA-1 B-i 002 YP Durbirt- Eqiwtfona C IRD 081 y y 7T* D7982 RzA Walsont Static expectnlions 1 OLS -0.04 0.01 -1.76 0,50 0.04 0.03 -0.12 -0.03 -0.03 0.68 1.84 (-0.3) (1.5) (-3.1) (1.9) (1.0) (2.0) (-1.0) (-1.0) (-2.1) 2 OLS 0.006 -0.29 -0.06 0.39 1.84 (1.0) (-4.1) (-2.6) Partial perfect foresightt 3 oLS -0.08 0.03 -1.61 0.50 0.07 0.04 0.03 -0.02 -0.02 0.73 1.97 (-0.5) (1t9) (-2.6) (2.4) (2.0) (3.3) (0.2) (-0.1) (-1.4) Note: The dependent variable is the ratio of the trade surplus to cDP (TSIV). The estimation method was ordinary least squares (om) and maximum likeli- hood (ML). Figures In parentheses are t-statistics. A blank denotes the omisslon of the specific variable from the regression; R2A Is adjusted R2. aJorge Marshall and Mlaus Schmidt-Hebbel 209 crisis, recovery, and consolidation. Starting with a cNFPs defict of 23.4 percent of GDP in 1970-73, which, in conjunction with adverse foreign stocks, forced a major stabilization effort tat was very suc- cessful on the fiscal side, the 1970s culminated in a 5.4 percent sur- plus in 1980. In the early 1980s the conjunction of negative external developments and misguided wage and exchange rate policies caused a major macroeconomic cnsis, which led to deterioration of the public finances of the financial and nonfinancial public sector. Quasi-fiscal rescue operations by the Central Bank directed at domes- tic debtors and the private financial system, combined with social emergency programs and a recession-induced fall of revenue in the nonfinancial public sector, caused large total public sector deficts in 1982-85. A strong condusion can be drawn regarding the relative roles of fiscal policy shifts and of shocks exogenous to the control of the fiscal authority: fiscal policy shifts were responsible for more than 100 per- cent of the cyclical variations and structural corrections of nonfinan- cial public sector deficits during 1974-88. Public sector retrencment and reform were major causes of changes in CNaws deficits: the tax reforms, public sector rationalzations, and privatizations of the mid-1970s corrected the initial deficits, while the 1981 social security reform and the tax reductions in the mid- to late 1980s contributed to the deficits. The large swings in copper prices were traditionally the most destabilizing factors outside the control of policymakers until the copper stabilization fund introduced a stronger separation between public spending and copper revenue- An implication of the high quasi-fiscal deficts was the huge buildup of total public debL Nonetheless, our calculations show that the solvency of the total public sector is not jeopardized by the current fiscal stance: a deficit of 0.6 percent of CDP-reflecting a 1.5 percent deficit of the Central Bank and a 0.9 percent surplus of the CNFPS-is perfectly sustainable under normal macroeconomic conditions. In the section on defict financing, we explored the implications of domestic financing of public deficits for the interest rate and the price level im Chile. The effects of alternative strategies of domestic financ- ing were discussed and contrasted in the context of a portfolio model with a parti adjustment structure for both the money and the public debt markets. The simulations based on the estimated portfolio model showed relatively conventional results: money-financed public deficits reduce interest rates and are strongly inflationary, while debt- financed deficits raise interest rates and are only weakly inflationary. Less conventional was the result that the moderate fluctuations of interest rates and price levels during the 1980s were attnbutable in large part to the rising demand for public debt by the new private pension funds. This rise permitted the accommodation of domestic 210 Chile: Fiscal Adjustment and Successful Performance financing of deficts that were largely caused by quasi-fiscal opera- tions and social security reform. Therefore, one should not generalize Chile's experience with fiscal deficits and their financing during the 1980s. In the section on crowding-out and crowding-in of private con- sumption and investment, we went a step further in analyzing the macroeconomic implications of public sector deficits. The focus was on the sensitivity of private consumption and investment to fiscal variables, in addition to the indirect effects of fiscal policies through interest rates, inflation, and private disposable income. Clear policy implications can be drawn from our empirical results on private con- sumption. There is no evidence of a direct effect of the public deficit on private consumption; consumers are neither Ricardian nor directly crowded-out by public spending in Chile. A transitory deficit reduc- tion financed by a transitory tax hike reduces private disposable income and hence private consumption and saving. If the tax hike is permanent, private consumption will take the brunt of the reduction in (current and permanent) private disposable income without signifi- cantly affecting private saving. Also, domestic deficit financing through changes in the real interest rate and the real inflation rate exercise no indirect effects on consumption. If, however, inflation were to reduce real base money, private consumption would be affected. Adjustment to foreign shocks has strong effects on private consumption in Chile. The combination of lower foreign saving and a higher real exchange rate has a large negative impact on private con- sumption expenditure. This was dearly observed during the post-1982 adjustment period, when private consumption fell to his- torical lows, while private investment reached historical highs. From the empirical results for private investment behavior, the fol- lowing implications can be drawn. Financing of the public deficit through issuance of domestic debt is increasingly felt by private investors in Chile, as a consequence of the upward pressure of this form of financing on domestic real interest rates. The other side of the coin is the weakened role of both domestic and foreign liquidity con- straints: whereas firm profits and foreign saving were important investment determinants before financial liberalization, they have not been sigrificant imvestment determinants since the mid-1970s. No evidence was found for public investment crowding-in effects. In fact, the relevant variable (public capital stock) tends to reduce private capital formation, reflecting public-private investment substitution. Foreign adjustment has affected private investment negatively through the relative price of investment goods, which has increased with the real devaluations. However, extemal adjustment coincded with a strengthening of the development strategy and stable policy rules, reducing significantly policy uncertainty and fluctuations in Jorse Marshall and Klaus Schmidt-Hebbel 211 output. This reduction of systemic risk and macroeconomic uncer- tainty (which negatively affect irreversible investment decisions) has had a major beneficial impact on private capital formation in Chile, as illustrated by our results and by the current record levels reached by this variable. External adjustment during the 1980s has implied a large real exchange rate depreciation and a 12 percentage point correction of the ratio of the trade surplus to CDP (from a deficit of 10 percent of GDP in 1981 to a surplus of 2 percent in 1992), which was required to ensure continuous debt servicing. Our results reflect the sensitivity of the trade surplus to the public sector deficit; however, its low parameter is also consistent with the fact that a significant share of the external adjustment was borne by private consumers. What lessons emerge for fiscal policy management from the Chil- ean experience? First, the Chilean case suggests that there are no quick-fix remedies for domestic and external imbalances. Adjusting fundamentals-among which public sector imbalances are para- mount-requires perseverance by policymakers in order to build up the credibility of private agents that is required for an adequate financial-markets and imvestment response. Second, balanced public accounts provide the best foundation for adjusting to external shocks. There is little doubt that the basic stablization-cum-structural-reform strategy implemented in Chile before the 1982 debt crisis (even taking into account some serious policy mistakes) was the major determi- nant of the rapid recovery and subsequent high-growth path fol- lowed by the Chilean economy during the 1980s and early 1990s. Finally, although this chapter has not focused in detail on public investment, we infer from other country experiences that public investment in infrastructure and human capital is a key to economic growth. The public sector should focus inaeasingly on high-return investment in human capital and physical infrastructure; this would be the government's main contribution toward ensuring that Chile continues on a high-growth path. Appendix 4.1. Revenue and Expenditure Functions Tax revenue functions for three tax categories, the public enterprise surplus, and transfers to the private sector were estimated before performing the cNFPs deficit decomposition in the section on fiscal policy. The estimation results for the revenue functions are reported in table 4.17. Income is the main exogenous (for the policymaker) determinant of direct and indirect tax revenue- Inflation shows a negative sign con- sistent with the Olivera-Tanzi effect in the case of indirect taxes but is not significant at the 10 percent level. Tax reforms (or endogenous Table 4.17. Estimation Results for Behavioral Public Sector Variables, Chile, 1973-89 lndepet'denl arpiab)es Exchalaie Inflation CopPer Durbin- Dependent varinble Colistat,t Iuconwre rate rate price Dl D2 R2A Watsorn l. Direct taxes, 506,3 3.36 298 -494 0.73 1.7 (0.93) (1.92) (2.2) (-3.0) 2. Indirect taxesb 364.9 10.65 -0.91 786 1,833 0.95 1.6 (C.';) (4.4) (-1.5) (2.1) (2.9) 3. Copper taxesc -84.9 -6.34 7.41 27.7 0.54 1.4 (-0.2) (-1.9) (1.6) (3.4) 4. Surplus of public firmsc -2,869 43.0 23.5 0.92 1.9 (-4.6) (10.3) (3.3) 5. Total transfers to private sector 1,121 11.2 -3.8 0.65 1.6 (0.5) (1.8) (-1.6) Note: The simulalion method was ordinary least squares. Figures in parentheses are I-statistics; R2A is adjusted R2. A blank denotes the omission of the specific variable from the regression. a. Income Is lagged by one peTiod; DI is 1 for 1975-81 and 0 for 1982-89. D2 is 0 for 1973-84 and I for 1985-89. b. Dl Is 1 for 1973-83 and O for 1984-89, D2 is 0 for 1975-83 and 1 for 1984-89. c. The sample period is 1975-89. d. Income is lagged by one period. Jorge Marshall and Klaus Schmidt-Hebbel 213 fiscal policies) are very significant determinants of revenue, as reflected by the period dummies that signal tax reforms. Revenue from special taxes on public sector copper mines is countercyclical and, not sur- prisingly, increases with the real exchange rate and the international copper price. The latter two exogenous deteminants also contribute to the state-owned enterprise surplus, which is dominated by tradable- goods-producing firms, including the copper industry. Finally, total transfers to the private sector (including social security- payments) appear to be procyclical and tend to fall with inflation. Appendix 4.2. Derivation of the Asset Portfolio Model This appendix presents the basic building blocks of the model dis- cussed in the section on deficit financing. The point of departure is the budget constraint of the public sector given by equation 4.2 in the text. In each period individuals allocate their wealth (which is fixed at any point in time) into three broad assets: domestic money, public sector bonds, and interest-bearing foreign assets. The long-run port- folio demands for these three assets are: (4.13) mld- pt = a0 + alyt + a2it + a3(it+ 6t) +.Wt (4.14) dt-d = P + ilYt + 02zt + P3(i+ at) + Wto (4.15) e, + mr= ro Y + ry ± r2it + r3(it*+ 6t) + W, where w is real financial wealth, y is domestic real output, p is the domestic price level, i is the domestic nominnal interest rate, it is the foreign nominal interest rate, m is the domestic nominal money sup- ply, m' is foreign nominal asset holdings, d is public domestic debt, e is the nominal exchange rate, and 6 is the expected rate of change of the exchange rate. All variables except i, i* and 6 are expressed in natural logarithms. The subscript t refers to time, and the superscript d denotes demand. The expected signs are denoted below each coeffi- cient. The expected signs of the coefficients are as follows: a,o ao a 0; 4O2, C3 < ; Plo,1 -0;02 > 0; 13 < 0; 70, T 0; T7<. ; 73 > 0 Wealth elasticities of long-term asset demands are assumed to be 1. The wealth adding-up constraint implies that only two of the three demand equations are independent, and we will therefore focus on the equilibrium conditions in the domestic money and public debt markets. Assuming a partial adjustment mechanism as a result of the existence of implicit adjustment costs, the change in the stock of real balances and public debt holdings can be written: (4.16) -(m, -,)=( - Pt) - (mt_. - Pa-i)) + OS1 (4.17) ( - Pt) = A[(dd - p,) - (dt, - pt-1)] + nid,. 214 Chile; FiscalAdjustment and Successful Perfonnance The expected signs of the coefficients are as follows: a > 0; A s 1; 0 > o; nf <1. Dots over variables denote time derivatives. In addition to the partial adjustment mechanism, equation 4.16 indludes a nominal money supply shock term (S), defined as the difference between actual and expected rates of change in nominal money balances. This term captures the shock-absorber hypothesis proposed by Carr and Darby (1981), which states that when the gov- ernment changes the rate at which it creates money, the result is a net unplanned increase in money holdings by individuals. As in the case of the money market, a partial adjustment mechanism is assumed in the market for public bonds. Equation 4.17 also indudes the actual rate of change in the stock of nominal public debt as part of the determination of the accumulation of real public bonds. This term stands for te inaease since 1981 in the demand for public debt by the private pension funds, which accommodated the financing of deficits implied by the social security reform and the quasi-fiscal operations of the Central Bank-in Chile. This partial demand accommoodation to the increase in the public debt implies that the interest rate did not rise by as much as if demand accommodation were absent. After substituting equations 4.13 and 4.14 into 4.16 and 4217 and solving for the current values of real balances and real public bonds, equations 4.3 and 4.4 are obtained. The short-rnm coefficients of asset demand equations (4.3-4.4) and the coefficents of the reduced-form equations for the nominal interest rate and the price level in 4.54.6 are detemined by the structural coefficients of equations 4.13, 4.14, 4.16, and 4.17 as follows: =5o- -~ o FoO * ~~~~~~~~~0- -D9Z2 = P 2 IL2 - .r3=a3 *3 -,03 -*- . +4WLa -4ljL o+ = t5 =n * +~~~~~~~~~6 Or '6 =IL ca.- a,4ao -o32ro $L02 aca1 - F. A,4a01 - a10%J -° Cira2 -- i-a2 ua% 3 - a3i3) 122 - 7-a IL02G'(2 A02 aa2 lorge Marshal and Maus Schmidt-Hebbel 215 1-o ou )A 03 -X3 P2D - aa2 I42 - uaC2 nP= - ua 4= 2-°a2~~~43 1W- u°2 -1 - $02 Q5 1rs = A62 Ga 02t 12- CFC2 - na2n - i) PD2 - ua2 /42 - Ua12 - - n- -unc%2 17= 77= n7c~~ -0 72 89= - p8a- 7. '4~ - aa Appendix 4.3. Derivation of the Trade Surplus Equation This appendbx focuses on the derivation of equation 4.12. The trade surplus in current prices, defined by the excess of exports over imports, is equal to the difference between current-price output and total absorption (A): (4.18) TS=P.A-P,, m= Y-A =Y + (rB1 + i*NFAg-.-T)-ODg- A where r is the domestic real interest rate paid on government bonds, T is total taxes net of transfers paid by the private sector, Ap is private absorption (private consumption and investment expenditure), A. is public sector absorption, and OD., is the operational public sector deficit. After substituting general functional forms for output and absorp- tion consistet with theory (as, for instance, in McNelis and Schmidt- Hebbel 1993) into equation 4.18, we obtain: (4.19) TS = Y(i* r,. . .)+ 1 + irB_l - Ti - ODS -AP(NFAP, B, NFAg, i*, r, TTh, YP) (+) (+) (O, +)QT,-X?,-) (+) (+) 216 Chile: Fisa Adjustment and Successful Performance Equation 4.12 is obtained by rewnrting equation 4.19 as a functional form for the ratio of the trade surplus to Y after scaling all relevant variables to current-price GDP. Appendix 4.4. Data Sources, by Section Dficit Financing, Interest Rates, and Inflation The definitions of and sources for the variables in this section are: i = domestic nominal interest rate. Monthly average of the effective interest rate paid on short-term deposits (30-89 days). Source: Banco Central de Chile (1990). i foreign interest rate. Monthly average uBOR for 180-day loans in U.S. dollars. Source: IMF, International Financial Statistics. 6 = expected devaluation rate of the nominal exchange rate. Sources: for 1975:1-1984:3, Le Fort and Ross (1985); for 1984:4- 196884, a uraiate autoregressive integrated moving-average pro- cess was estimated for the nominal exchange rate of the banking system on the basis of data from Banco Central de Chile (1990). M. = narrow money. The original series (sources: Vial and Man 1986; Banco Central de Chile 1990) was deseasonalized- Anticipated money holdings were vqtimated from an autoregressive process of orderfour- D = quarterly domestic public debt. Source: annual data from Lar- rafiaga (1989), corresponding to the debt of the Central Bank held by the private sector The quarterly interpolation was performed follow- ing the method descrbed inJadresic (1990). Y = quarterly real GOPR Sources: Haindl (1986); Banco Central de Chile (1990). Crofwing-Out and Crowding-In of Private Consumption and Private Investent by the Public Sector The definitions of and sources for the variables are: Cp= private consumption. Source: Banco Central de Chile (1990). Cg public consumption. Source: Banco Central de Chile (1990). Y = annual real GDP. Source: Banco Central de Chile (1990). Ip = private fixed-capital investment. Sources: Solimano (1992); Zucker (1988). UcK = user cost of capital. Source: Banco Centr al de Chile (1990). MPK = margia product of capital. Source: Banco Central de Chile (1990). forge Marsial and Klaus Schmidt-Hebbel 217 K aggregate real capital stock. This series was elaborated on the basis of Solimano (1992) and Zucker (1988) by splicing and assuming an initial capital-to-CDP ratio of 2.5. In addition, the ratio of public to private capital stock was assumed to be equal to the ratio between public and private investment. H = nominal base money stock. Source: Banco Central de Chile (1990). The remaining variables were obtained directly from Banco Central de Chile or were derived from the variables defined above according to the definitions provided in the text. Relative Extenal Prices and the Trade Balance The definitions of and sources for the variables are: PXIPN = relative price of exportable and nontradable goods. Sources: for exportable goods, Banco Central de Chile (1990); for the nontradable goods index, Sdhmidt-Hebbel, Castro, and Leng (1990). PMTPM = relative price of importable and nontradable goods. Sources: for the price of importable goods, Banco Central de Chile (1990); for the nontradable goods index, Scdmidt-Hebbel, Castro, and Leng (1990)- tM = nominal ad valorem tariff rate for imports. Sources: Ffrench- DaVrs (1984); Lagos and Aedo (1984). The remaining variables were obtained from Banco Central de Chile (1990). Notes The authors benefited from extensive discussions with Osvaldo Larrafiaga and Patricio Mujica. Bela Balassa, Edgardo BarandiarAn, Wldliam Easterly, James Hanson, Miguel Kiguel, and Juan Carlos Lerda gave us excellent com- ments on previous drafts. Francsco Berrasconi and Eduardo Saavedra pro- vided efficent research assistance. We thank all of them, and we retain responsibility for any remaining errors. 1. References on macroeconomic adjustment and reforis in Chile during the past two decades inldude Corbo (1985, 1990); Edwards and Edwards (1987); Ffrench-Davis and de Gregorio (1987); and Morande and Schmidt- Hebbel (1988). 2. This chapter uses three public sector deficit measures differentiated by public sector coverage: the consolidated nonfinancial public sector deficit (comprising general government and nonfinancial public enterprises); the financial deficit (comprsing quasi-fiscal losses of the Central Bank); and the total public sector deficit (comprising the first two ). In addition, it distin- guishes between the total (or nominal) deficit and the primary deficit, the difference between the two latter being explained by interest payments on domestic public debt. 3. For a discussion of the causality betwee domestic spending, foreign lending and the real exchange rate during this period see Morandd and Schmidt-Hebbel (1988), in particular chapter 5. 4. Since May 1981 Chile has gradually replaced the existing pay-as-you-go state pension scheme with a privately managed and fuly funded system. The transition deficit caused by this reform sarted at 3.2 percent of cm in 1982, reached a peak of 4.8 percent in 1991, and fell gradually thereafter (Arru 1991, 1992). Although this type of pension reformn increases measured public defcits (while private sector surpluses incease by the same amount), it is important to point out that such a reform, if financed by issuing publk debt, has no first-order effects on the macroeconomy- The reform provides the means of financing the transition deficit by creating a demand for public debt paper by the new private pension funds (see Arrau 1991 and Arrau and Schmidt-Hebbel 1993 for detailed discussions). 5. For a detaied cakulation of Central Bark quasi-fisal operations and losses, see Eyzaguirre and Lraiiaga (990). 6. The exoange rate definitions used houghout tis chapter are, for the nominal exchange rate, the prce of foreign currency in units of domestic currency and, for the real exchange rate, the relative price of tradable goods in units of nontradables. Hence a real devaluation means a higher real exchange rate. 7. See Marshall and Schlitidt-iebbel (1989a, 1989b) and a compact presen- tation in the appendix of Morandd and Sdhmidt-Hebbel (1991). . The average relative contaution of each group of deficit detemants to the explained variation of the deficit (see table 4.4) is computed as: - ~~~~~[~ 4drhsind,)] f id where di is the explained change in the deficit, dv; is the change in the deficit caused by variable category v (the total vaiations that resulted from fiscal policy, macroeconomic~ and exteni variables in table 4.4, with nondecom- posed accounts as part of fiscal policy variables), sign d, is the sign (positive or negative) of d, and the subperiods i aare the four considered in table 4.4. 9. See Buiter (198) and van Wjubergen (1989) for this notion of defit sustanabiity and its relation to public sector solvency. 10. The real domestic interest rate r is defined as the difference between the nominal domestic interest rate and domestic inflaton. The real foreign inter- est rate (*) is the difference between the nominal foreig interest rate and foreign inflation. The rate of real devaluation is defined as the rate of increase of the nominal exchange rate plus foreip inflation nmnus domestiinflation. 11. The sample period for the money equation is 1976:1-1988.4, while for the domestic debt equation it covers 1982:1-1988:4. 1.2. This finding can be explained in terms of the values of the estimated coefficients of the struchural equations. When the value of the coefficient related to the speed of adjustment in the public bond market (g) is high- which is the case according to our estimations-the model implies smal Jorge Marshal and Klaus Schmidt-Hebbel 219 values for the coefficents of lagged real public bonds in the reduced-form equations (see appendix 4.2). 13. Recall that flow variables, such as car, are quarterly figures. Therefore the corresponding annual ratios have to be divided by 4. 14. The Ricardian hypothesis has been widely rejected in empirical studies for industral countries. Most of these studies identify the existence of perva- siveborrowing constraints as the main cause for its rejection. A study for a set of developing countries by Haque and Montiel (1989) tests for two different causes, which could explain a deviation from Ricardian equivalence higher private than government discount rates (attnbutable to Blanchard-Yaari infinite-lived households that face a mortality probability) and liquidity con- straints. They find significnt evidence that the latter causes a deviaton from Ricardian equivalence without much support for such a role for the former. Borrowing constraints, proxied by current income or financial asset holdings, are also major determinants in the caoss-developing-country studies for pri- vate consumption by Rossi (1988) and for household saving by Schmidt- Hebbel, Webb, and Corsetti (1992). 15. Among the empirical studies showing savmg to be insensitive to the real interest rate in developing countries, see Corbo and Schmidt-Hebbel (1991), Giovannini (1985), and Schridt-Hebbel, Webb, and Corsetti (1992). For an alternate view, see Balassa (1989) and Fry (1988). 16- For the theory and the Latin American experience on the role of consumption-based real interest rates in intertemporal consumption alloca- tion, see Dornbusch (1983. 1985a, 1985b, 1988). and for the relation between savmg and capital flight see Drombusch (1989). 17. The scaling of nonstationary variables to private disposable income and of nonstationary variables to bGm in the private investment and trade surplus equations that foilow reduces the incidence of problems derived from nonsta- tionary time series. 18. Note that equation 47 allows for testing the separate influence of per- manent public saving (PS), in addition to the influence of permanent private disposable income (PrM). If PS is not significantbut rm is (that is, if Ricardian equivalence is rejected in favor of the permanent income hypothesis), taxes affect consmption, but government spending does noL 19. Permanent private disposable income and permanent public saving are consistent with the following definitions for their corresponding acrrent values: DYpt = GoP, - NFpp,, - T, + r,D,, and %S, = Tt- CG, - Npwct - rt Q,, where GDP is gross domestic product, N'p is net foreign payments made by the private sector, Sc is current public saving, Cc is public consumption, and xm,, is net foreign payments made by the publc sector Note that D now refers only to the domestic public debt. For the expected 'permanent" values of any variable (private disposable income and public saving in this section, and other variables in the investment section that foflows), we specify two alternatives. The first is to assume expectations of the permanent values that are consistent with partial perect foresight, defined as the simple aver- age of the contemporaneous variable and two periods into the future, for any variable x: Px, (x,-i +-+ + x,l.,,)I3, where Px defines the expected per- manent value of variable x. The second altemative is the simple static- 220 Cuile. Fiscal Adjustment and Successful Performnce expectations specification, which allocates a 100 pecent weight to the con- temporaneous value in the preceding average. Similar assumptions are made with respect to expected consumption inflation (and expected investment inflation in the section on private investment). A fist alternative takes actual inflation between today and tomorrow as the relevant proxy for rationally expected inflation. The second alternative is static, specifying the expected price change to be equal to the actual price change between yesterday and today. 20. This result mimics the results for the quarterly structural consumption function for Chile in Schmidt-Hebbel (1988), whidh are stronger for the back- ward- than for the forward-lookiing expectations spedfications. 21. The definition of permanent variables is the same as for consumption determinants, explained in footnote 19. The current real user cost of capital is defined as: Uc,r = (P,t,JP iR - P)I(1 + Pjl) +6] where Pe is the private investment deflator, P, is the GMo deflator, iF, is the nomnal interest rate on banking loans to firms, P,, is the expected rate of change of the private investment deflator, and a is the real rate of capital depredation. The private sector capital-output ratio (the inverse of the aver- age product of capital) stands for both the neodassical marginal product of capital (which is a linear transformation of the marginal product under a Cobb-Douglas techology) and the Keynesian potential-to-actual-output ratio. Note that private and public capital add to the total domestic capita stock: K, = K0, + K,. Expected investment inflation is based on an estimated autoregressive structure. All expected permanent variables are specified according to the two hypotheses mentioned above: the partial perfect- foresight alternative and the static version. Fmally, the two coefficients of variation are defined as fiveperiod moving variances for two periods back, the current period, and two periods forward. 22- The final results reported in the tables exclude three variables that appear in equation 4.8: the relative price of the two aggregate investment components, corporate taxes, and firn credit. The exclusion of the first vari- able from prelminary results was justified by its implausibly high coefficient, which affected many other parameters. The other two variables were not induded because of lack of data. 23. For the same reason discussed in the preceding section (to avoid spu- Tious correlation), we scale nonstationary variables (such as the trade sur- plus, government expenditure, and net foreign assets) to appropriate scale variables in equations 4.9-4.12. 2& It was not possible to include the nontradable component of public spending as an additional determinant because of lack of data. A significant fraction of the massive 1975 devaluation could not be explained by any of the preceding variables and hence was treated as an outlier. 25. Current-price cDP is used as the relevant scale variable. Hence the positive sign of the constant term C reflects the hypothesized positive effect of income on the trade balance, when multiplying equation 4.11 by the latter and abstracting from the presence of multiplicative terms involving income and all nonscaled right-hand variables. Jorge Marshall and Klaus Schmidt-Hebbel 221 References Arrau, Patricio. 1989. "Intertemporal Monetary Economics: Evidence from the Southern Core of Latin America." Ph.D. diss. University of Pennsylva- nia, Department of Economics, Philadelphia. . 1991. "La Reforma Previsional Chilefia y su Financiamiento durante la -Žransici6n." Colecci6n Estudios C1EPLAN (Chile) 32 (June): 5-44. . 1992 "El Nuevo R4imen Previsional Chileno." In Fundaci6n Friedrich Ebert de Colombia (Frcot), Regmenes Previsionales. BogotA . Arrau, Patricio, and lGaus Schmidt-Hebbel. 1993. "Macroeconomic and Intergenerational Welfare Effects of a Transition from PayAs-You-Go to - Fuly-Funded Pension Systems." World Bank, Policy Research Depart- ment, Washington, D.C. Balassa, Bela. 1989. "The Effects of Interest Rates on Saving in Developing Counties." Policy Research Working Paper 56. World Barik, Office of the Vice President, Development Economics, Washington, D.C Banco Central de Chile. 1990. Indicadores Econ6micos y Sociales 1960-1988. Santiago. Buiter, Willem H. 1988. "Some Thoughts on the Role of Fiscal Policy in Stabilization and Strucural Adjustment in Developing Countries." NERt Worling Paper 2603. National Bureau of Economic Research, Cambridge, Mass. Car, Jack, and Michael K Darby. 1981. "Me Role of Money Supply Shocks in the Short-Run Demand for Money-" Journa of Monetary Economics 8 (2): 183-99. Corbo, Vittorio. 1985. "Reforms and Macroeconomic Adjustments in Chile during 1974-1984." World Development 13 (18): 893-916. 1990. 'Public Finance, Trade and Development The Chilean Experi- ence." In Vito Tanzi, ed., Public Fnance and Economic Development. Detroit, Mich: Wayine State University Press. Corbo, Vittorio, and Jaime de Melo. 1989. "External Shocks and Policy Reforms in the Southern Cone: A Reassessment." In Guilermo A. Calvo, ed., DeLbt, Stabiliation and Development: Essays mi Memory of Carlos Diaz- Alejandro. Oxford, U.K.: Basil Blackwell. Corbo, Vittorio, and Klaus Sdhinidt-Hebbel. 1991. "Public Policies and Sav- ing in Developing Countries." Joumal of Developmet Economics 36 (1): 89- 116. Cortaz, Ren4, and Jorge Marshall. 1980. "Indice de Precios al Consumidor en Chile: 1970-1978." Colecain Estudios CIEPLAN (Chile) 4:159-201. Dornbusch, Rudiger. 1983. "Real Interest Rates, Home Goods, and Optimal External Borrowing." Journal of Political Economy 91 (1): 141-53. - . 1985a. "External Debt, Budget Deficits and Disequilibrium Exchange Rates." In Gordon W. Smith and John T. Cuddington, eds., Intemnational Debt and the Developing Countries. A World Bank Symposium. Washington, D.C -.1985b. "Overborrowing: Three Case Studies." In Gordon W. Smith and John T. Cuddington, eds., Inenatirnal Debt and the Developing Coun- tries. A World Bank Symposium. Washington, D.C. 222 Chile: Fiscal Adjustnent and Successul Performtance . 1988. "Mexico: Stabilization, Debt and Growth!" In Georges de M&il and Richard Portes, eds., Economic Policy. A European Forum. Cambridge, U.K.: Cambridge University Press. ---n. 1989. "Capital Flight Theory, Measurement and Policy Issues." Mas- sachusetts Institute of Technology, Department of Economics, Cambridge, Mass. Easterly, William. 1989. "Fiscal Adjustment and Deficit Fmancing during the Debt Crisis." In Ishrat Husain and Ishac Diwan, eds., Dealing with the Debt Crisis. A World Bank Symposium. Washington, D.C. Easterly, Wiliam, Carlos A. Rodriguez, and Klaus Schmidt-HebbeL 1989. "Research Proposal: The Macroecononmics of the Public Sector DeficiL" World Bank, Country Economics Department, Washington, D.C. Edwards, Sebastn, and Alejandra C. Edwards. 1987. Monetarism and libemal- izatiorn 7he Chilean Experiment. Cambridge, Mass.: Ballinger. Eyzaguirre, Nicols, and Osvaldo Larraiaga. 1990. "Macroeconomfa de los Operaciones Cuasifiscales en Chile." Economic Commission for Latin America and the Caribbean, Santiago. Ffiench-Davis, Ricardo. 1984. "Indice de Predos Externos." In Colicc6n Estudios CIEPLAN (ChiDe) 13: 87-106. Ffrench-Davis, Ricardo, and Jose de Gregorio 1987. "Origenes y Efectos del Endeudamiento Extemo en Chile." El Trimestre Econ6mice (Me,dco) 54 (January-March): 159-78. Fry, Maxwefl J 1988. Moan,e Interest and Banking in Economic Development. Baltimore, Md.: Johns Hopldns University. Giovannini, Alberto. 1985. "Saving and the Real Interest Rate in tars." Journal of Deelopment Econonics 18 (August): 197-217. Haindl, Erik. 1986. 'Trimestralizacid6n del KcB por Origen y Destino." Estudios de Econoina (Chile) 13 (1): 117-53. Haque, Nadeem LI, and Peter J. MontieL 1989. "Consumption in Developing Countries: Tests for Lquidity Constaints and Friite Horizons." Review of Economics and Sfatics 71 (3): 408-15. wmF (International Monetary Fund). Various years. International Financial Sta- tLirs. Washington, D.C Jadresic, Esteban. 1990. "Trimestralizaci6n sin Uso de Series Reladonadas." Notas Titicas CIEPLAN (Chile). Lagos, Luis Felipe, and Christiin Aedo 1984. "Protecci6n Eetiva en Chile 1974-1979." Documento de TraWo 94, Instituto de Economia, Pontifida Uni- versidad Cat6lica de Chile, Santiago. Larraiaga, Osvaldo. 1989. "El D6ficit del Sector Piblico y la Polftica FLscal en Chile." Economic Commission for Lation America and the Caribbean, Santiago. Le Fort, Guillermo, and Crisian Ross 1985. "La Devaluad6n Esperada, una Aproximaci6n Bayesiana: Chile 1964-1984." In Sene Inuestigac6n 72, Uni- versidad de Chile, Departamento de Economia, Santiago. Marshall, Jorge, and Klaus Schmidt-Hebbel. 1989a. "Economic and Policy Determinants of Public Sector Deficts." Policy Research Worldng Paper 321. World Bank, Country Economics Department, Washington, D.C Jorge Aarshall and Klus Schmidt-Hebbel 223 . 1989b. "Un Marco Analitico-Contable para la Evaluacidn de la Polftica Fiscal en America Latina." Sede PolUtra Fisal 1. Economiic Commission for Latin America and the Caribbean, Santiago. McNetis, Paul, and Klaus Schmidt-HebbeL 1993. "Fiancial Laberalization and Adjustment: The Cases of Chile and New Zealand." Journal of Intenta- tional Money and Finance 12: 249-77. Morands, Feipe, and Klaus Schmidt-HebbeL 199L "Macroeconomics of Public Sector Deficits: The Case of Zibabwe. Policy Research Working Paper 688. World Bank, Country Economics Department, Washington, D.C , eds 1988. Del Auge a la Crisis de 1982: Ensayos sabre Estabiliai6n Finan- dem y Endedamiento en Chile, LAnes-Georgetown and Instituto Inter- americano de Mercados de Capital, Santiago, Chile. Mujica, Patricio. 1990. "Financiamiento Dom4stico del Deicit Fiscal y sus Efectos Macoecon6micos: Un Modelo para Chile." Econoric Commission for Latin America and the Caribbean. Santiago- Rossi, Nicola. 1988. "Govermnent Spending, the Real Interest Rate, and the Behavior of Liquidity-Constrained Consumers in Developing Countries." International Monetary Fund Staff Papers 35 (1): 104-40. Scdnuidt-Hebbel, Klaus. 1987. "Terms of Trade and the Current Account under Uncertainty." Andlisis Econn6ico 2 (1): 67489. -- 1988. "Consumo e Inversidn en Chile (1974-1982): Una Interpretacid6n 'Real' del Boom." In Felipe MorandE F. and Klaus Schnxiidt-Hebbel, eds., Del Auge a la Crisis de 1982. Ensayos sabre Estabilacidn Financiea y Endeuda- mnuento en Chile, uA.Es-Georgetown and Instituto Interamericano de Me- cados de Capital, Santiago, Chile. Schidt-Hebbel, Klaus, and Tobias MiLmer 1992. "Private Investment under Maaoeconomic Adjustment in Morocco." In Ajay Chhibber, Mansoor Da- lanmi, and Nemat Shafik, eds., Reuiving Private Investment in Developing Countries. Amsterdam: North-Holland. Schmudt-Hebbel, Klaus, FrancLsca Castro, and Ivan Leng. 1990. "Una Base de Datos Tdmestrales para la Eoanomia Chilena." Sene Investigad6n 24. -nLDEs-Georgetown and Instituto Interamericano de Mercados de Capital, Santiago. Schmidt-Hebbel, Klaus, Steven B. Webb, and Giancarlo Corsetti. 1992. "Household Saving in Developing Countries: First Cross-Country Evi- dence." World Bank EmnomicReiew 6 (2): 529-47. ServEn, Lis. 1990. "A RsSM-x Model for Chile." Policy Research Working Paper 508. World Bank, Country Economics Department, Washington, D.C Solimano, AndrEs. 1992. "'How Private Investment Reacts to Changing Mac- roeconomic Conditions: The Case of Chile in the 1980s." In Ajay Chhlbberr Mansoor DailaIrd, and Nemat Shafik, eds., Reviving Private Investment in Developing Countries. Amsterdam: North-Holland. van Wijnbergen, Sweder. 1989. "External Debt, Inflation and the Public Sec- tor. Toward Fiscal Policy for Sustainable Growth." WorLd Bank Econonic Review 3 (3): 297-320. 224 Chile: Fiscl Adjustment and Successful Pejbnnance Vial, Joaquin, and BArbara Mafn. 1986. "Series Monetarias Chienas 1960- 1985. " Estudias de Economia 13 (1): 191-22. Zucker, A. 1988. "Comportamiento de la Inversi6n en Capital Fijo en Chile: 1974-1987." Thesis. Instituto de Economfa, Pontificia Universidad Cat6lica de Chile, Santiago. Colombia: Avoiding Crises through Fiscal Policy William Easterly This chapter analyzes the macroeconomic effects of public sector defi- cits in Colombia. The first section reviews the historical evolution of fiscal poliqr in Colombia, with emphasis on the adjustment program of 1985-89, and uses an econometrically estimated model of the money and credit markets to examine how fiscal deficits affect the inflation rate and the real interest rate. The subsequent section dis- cusses the relationship between the fiscal deficit and the real exchange rate, using a reduced-form model of traded and nontraded goods. Historical Background Colombia is justly celebrated in Latin America for its prudent macro- economic management, of which careful management of fiscal defi- cits is the cornerstone. Even its occasional departures from conserva- tve macroeconomic policy seem tame by Latin American standards. In this section we review the broad outlines of macroeconomic policy in Colombia and look in more detail into the adjustment that has taken place since 1985. Macroeconomic Management, 1960-89 From the mid-1960s through the early 1970s, macroeconomic policy in Colombia was mostly conservative, supportive of an export-oriented development strategy that was associated with high growth of both GvP and trade. (See table 5.1 for basic macroeconomic indicators; see Garcia Garcia 1989 for a more detailed review.) The main factor in the second half of the 1970s was the surge in coffee export revenues, which the authorities were partially successfu:l in sterilizing. In the early 1980s the end of the coffee boom coincded with a large increase in public investment, especially in the energy sector, which led to an incipient balance of payments and debt crisis. The crsis was largely avoided thanks to a strong adjustment effort that began in 1985 and has continued to the present. 225 Table 5.1. Basic Economic Indicators, Colombia Five-Year Averages, 196-89 (percent) hIdicalor 1960-65 1965-70 2970-75 1975-80 1980-85 1985-89 Real exchange ratea 99.7 112.8 129,9 123.3 118.5 161.4 Inflation rate 11.1 9.3 16.4 24.2 23.1 24.0 Fiscal balance as a share of GDPb -1,2 -0.5 -1.5 -0.1 -3.5 -1.8 Fiscal balance as a share of GDr- - - - -2.1 -5.8 -2.6 Fiscal balance as a share of GDPd -2.2 -1.8 -2.9 -1.5 -5.7 - Real interest rate On loans - 2.4 4.4 10.3 13.0 LI On deposits - - -2.2 -1.0 4.3 6.1 GDP growth 4.7 6.0 5.7 5.4 . 2.3 3.6 Current account balance as a share of GDP -2.3 -2.7 -2.8 0.6 -5.1 -1.1 Real private investment as a share of real GDP 11.9 10.2. 9.2 8.1 7.4 8.2 Public external debt as a share of GDPe - - 20.0 15.3 22.6 37.2 Public external debt as a share of exports of goods and servicese - - 124.1 79.3 133.0 18717 -Not available. a. Depreciation is up. b. IMF, Interntaionil Financial Slalislics Yearbook 1989 (national government only); data until 1987 only. c. World Bank data. d. Garcia Garcia 1989. e. World Bank, World Debt Tables; public external debt data until 1988 only. William Easterly 227 Figure 5.1. Central Government Balance, Colombia, 1960-87 Percentage of GDI' 2 1 - . - ' ,/ a ~~~~SURPLUS 0 -1 g /f; _ 0 = ;, .:~~~~.V. DEpr. - i -2 -_[;;,1,t0' -3 -4 -5 1960 62 64 66 68 70 72 74 76 78 80 82 84 86 87 Source: IMF, tntenational Finanidni Statistics Yearbook-, 1989. Figure 5.1 presents a historical perspective on fiscal management. The only series available for a sufficiently long period is the IMF'S International Financial Statistics (irs) series on the national government deficit. The -IFS data are not consistent with the fiscal data used else- where in this chapter but do show a similar pattern. There have been three episodes of loose fiscal policy at roughly ten-year intervals-in 1961-62, in 1972-73, and in 1981-84, of which the last was by far the most severe. Bach episode was followed by rapid fiscal adjustment that avoided a prolonged crisis. The impression of relative macroeconomic stability is confirned by the behavior of the inflation rate (figure 5.2). By Latin American stan- dards (although not by world standards) the rate is very stable, stay- ing within a band of about 15 to 35 percent since the early 1970s. We see roughly three distinct periods: (a) during most of the 1960s, infla- tion oscillated in response to periodic large adjustments of the exchange rate; (b) following the introduction of the crawling peg in 1967, the inflation rate stayed very stable for about five years; (c) after the episode of loose fiscal policy in 1972-73, inflation accelerated, remaining at more than 20 percent after the coffee boom led to some monetization of reserve inflows. After the end of the coffee boom, fiscal expansion replaced reserve inflows as a source of money cre- ation. Subsequent fiscal contraction was just sufficent to match the reduction in external credit availability during the debt crisis, so that the need for money creation continued. Figure 5.3 shows the evolution of money creation, roughly defined as the change in base money over GDP. We see that the reliance on seigniorage has been remarkably stable since the mid-1960s, aside 228 Colombia: Avoiding Crises through Fiscal Policy from the burst of money creation associated with the coffee boom in the late 1970s. The behavior of the real exchange rate mirrors changes in macro- economic policies, as shown in figure 5.4. The exchange rate was quite volatile in the early 1960s, but after the introduction of the Figure 5.2. Inflation Rate, Colombia, 1960-89 Percent per year 35 30- 25- 20- 15- 10 5 1960 62 64 66 .68 70 72 74 76 78 80 82 84 86 8889 Source: World Bank data. Figure 5.3. Seigniorage, Colombia, 1960-88 Percentage of GDP I 4 -1 0 1960 62 64 66 68 70 72 74 76 78 80 82 84 86 88 Source: World Bank data. William Easterly 229 Figure 5.4. Real Exchange Rate, Colombia, 196089 Percentage deviation from 1975 value 30 20- l0- 0 -10' -30 -40 -50 1960 62 64 66 68 70 72 74 76 78 80 82 84 86 88 89 Notcr Depreciation is up. Source World Bank data. crawling peg, it showed a steady depreciation (depreciation is defined as up). With the onset of the coffee boom in 1975, the deterio- ration was reversed, and the real exchange rate appreciated steadily untifl the early 1980s. During the adjustment effort that began in 1985 the real exchange rate once again deprecated substantially. We see some association between the behavior of the real exchange rate and the episodes of fiscal contraction and expansion. The greatest appre- ciation of the currency came during the periods of expansionary fiscal policy in the early 1960s and early 1980s. The episode of loose fiscal policy during the early 1970s did not seem to have had much effect. We will examine in a subsequent section the relationship between deficits and the real exchange rate. The real interest rate shows a more erratic path than inflation and the real exchange rate (figure 5.5).1 This reflects the controls on inter- est rates prior to 1974, after which financial hiberalization began by fits and starts. Interest rates were mostly market-determined during the 1980s, with occasional temporary controls such as those imposed in 1988. The fiscal expansion of the 1980s was associated with some rise in the real interest rate, although interest rates remained high even after the fiscal adjustment. A puzzle of macroeconomic behavior in Colombia is the long-term decline in the private investment ratio, as shown in figure 5.6. (The ratio is defined in real terms to avoid any relative price effect.) The fiscal adjustment of the first haif of the 1980s was associated with a decline in investment, while the adjustment program of 1985-88 was 230 Colombia: Avoiding Criss through Fisazl Policy associated with a small rebound of private investment. The fal in private investment in the late 1970s and early 1980s was associated with high real interest rates and a high relative price of capital goods. But the inarease in the interest rate was modest compared with the secular decline in investment. As the figure shows, public investment Figure 5.5. Real Deposit and Loan Rates, Colombia, 1970-88 Percent 20 A 15 - 5 0 -5- -10 1970 72 74 76 78 80 82 84 86 88 Real deposit rate ...----- Real loan rate Source World Bank data. Figre 5.6. Public and Private Investment, Cojombia, 1950-88 Percentage of GDP 18 716 14 12 10 8 6 4 2 o 1950 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 - Real private investment -ii-Real public investment Source World Bank data. William Easterly 231 was rising at the same time that private investment was falling, sug- gesting crowding-out in the long run. The effect of public capital on private investment will be examined statisticaly later in this chapter. Public Sector Accounts, 1975-85 The public sector deficit was under control until 19814. Deficits were small during the heyday of the coffee boom of 1975-78. The end of the boom brought increased deficits, which worsened in 198184 because of a surge in public capital spending. The genesis of the fiscal crisis of the early 1980s can be seen by separating out the effects of changes in coffee revenues and public capital spending-the two largest single influences on the deficit. As figure 5.7 shows, if public capital spending had remained constant at its 1975 value (with the coffee surplus still taking its actual values), no fiscal problem would have developed. On the conrary, the deficit would have remained relatively constant until 1982 and would then have steadily declined; by 1985 a large surplus would have been registered. Expanded public capital spending was the source of fiscal imbalances in the early 1980s. If the coffee surplus had remained the same as in 1975, the deficit would have been larger in the 1970s, implying that some fiscal problems were hidden by the good fortune of the coffee bonanza. The increase in deficits in the early 1980s was financed for the most part from domestic sources, although external borrowing also increased. As figure 5.8 shows, the expansionary fiscal policy did lead to a rapid growth of the publc debt ratio in the 1980s, which helps to explain the incipient external debt crisis of 1983-84. But the debt accu- mulation was from a relatively low base compared with most other Latin American countries. The debt ratio decreased during the coffee boom of 1975-79, so that the fiscal adjustment of 1985-89 was enough to avoid a full-blown debt cisis. The restrained use of exteral financ- ing of the public sector in earlier periods helps to explain Colombia's avoidance of the kind of debt crisis that bedeviled its neighbors. It helped also that Colombia's fiscal expansion during 1981-84 was comparatively modest. Unlike the situation in Mexico and Argentina, the current balance of the public sector always showed a surplus, amounting to more than 2 percent of GDP. Moreover, some of the expansion in public investment was in oil exploration and develop- ment, which paid off with a surge in oil exports after 1985. The Adjustment Program of 2985-89 In late 1984 the Colombian government began a major adjustment program which succeeded in eiminating the fiscal imbalance that had 232 Colombia: Avoiding Crises through Fiscal Policy Figure 5.7. Public Sector Deficit, with and without Coffee and Capital Spending Effects, Colombia, 1975-87 Percentage of GDP 10 Efli offhighs - 5 iDficmaxplts ~ EEE ofe inesd /7/>1°Phlslwrd~~~~~~caitl ening 0 Dapitalspending Ffixfect o1 vawe lue -i--Deficit wth coffee and capitalsupeningfxda 17au 810 1975 76 77 78 79 80 81 82 83 84. 85 86 87 - Tota ipublicndeficnt --.*--Deficit with capital spending fixed at 1975 value -efit with coAm ee and capital spending fixed at 1975 value Source World Bank data. Figue 5<%. Public Extemna Debt, Colombia, 197"88 Percentage of GDrP 80 r 60 _; 1970 71 72 73 74 75 76 777 78 79 80 81 82 83 84 85 86 87 88 Colombia IHighly indebted countries ------ Latin American countries Source: World Bank data. William Easterly 233 threatened to cause a macroeconomic crisis. It is useful to distinguish two main phases of adjustment. In the initial phase the overall deficit was sharply reduced, from 6.3 percent of GDP in 1984 to only 0.3 percent in 1986. In the second phase, 198749, the deficit increased again, to about 2 percent of CDP. The surge in coffee export revenues, a good deal of which accrued to the public sector, made 1986 an anomalous year. Domestic financing of the deficit in the first phase was sharply reduced, from nearly 5 percent of C;DP in 1984 to -2.7 percent in 1986. Domestic financing then inaeased again to between 1 and 2 percent of GDP. It is notable that net external financing continued to be avail- able during the first phase of the adjustment. Colombia did not reschedule its external debt as other Latin American borrowers did. Although commercial banks were reluctant to continue lending in the wake of the global debt crisis, commercial financing was arranged with the support of two World Bank adjustment loans. A large flow of extemal finance in 1986, which came at the same time as the surge in coffee revenues, was successfully sterilized through public sector repayments of debt to the Central Bank and repayment of public short-term debt. In the second phase, external borrowing was reduced by the authorities in favor of domestic borrowing and money ceation. This choice may help to explain the continuing high real interest rates noted earlier. Because of the importance of the fortuitous coffee boom of 1986 for adjustment, a question arises regarding the extent to which the fiscal improvement was associated with permanent policy changes. Table 5.2 addresses this issue by decomposing the deficit into permanent policy components (structural trends) and other components- The first part of the table separates out various fiscal components whose variation is a result of nonpolicy factors or temporary policy changes. These indude the overal balance of the National Coffee Fund and the transfers it made to the rest of the public sector, which are affected mainly by international coffee prices.2 Also included is the balance and transfers of ECOPEROL, the national oil company, which became a large source of finaning for the rest of the public sector by 1987, thanlks to the development of a major new oil field. The fiscal improvemernt was helped in the first phase of the adjustmnent by the increased profits of the National Coffee Fund, which was the beneficiary of the surge in international coffee prices in 1986, and in the second phase by ECOPETROL. There were also some intrinsically temporary develop- ments that influenced the national government's balance, the main one being a tax amnesty that increased revenues in 1987.3 Taking all these factors into account, we find that there was still a substantial improvement in the stmctural defidt. Less than 1 percent- Table 5.2. Structural Trends in Fiscal Policy during Adjustment, Colombia, 1984-89 (percentage of cUr) Clhantge, lItem 1984 1985 1986 1987 1988 1989 1984-89 Total consolidated public sector deficit 6.3 3.5 0.3 2.0 2.1 2.2 -4.0 National Coffee Fund Surplus 0.4 1.4 3.2 -0.6 0.0 0.1 -0.3 Transfers to public sector 0.0 0.0 0.1 0.3 0.0 0.0 0.0 ECOPFTROL Surplus 0.1 -1.2 -0.2 08 0.0 0.4 0.3 lTansfers to public sector 0.1 0.1 0.2 0.6 0.6 0.7 0.6 National goverment, temporary revenue Items'1 0.1 0,1 0,3 0.5 0.1 0.1 0,0 Structural deficit 7.0 3.9 3.8 3.5 2.9 3.6 -3.4 Fixed capital formation -8.8 -8.3 -6.7 -6.1 -6.4 -6.7 2.1 Structural current deficit -1.6 -4.4 -2.9 -2.6 -3.5 -3.1 -1.3 Wages and salaties -6.9 -6.3 -5.9 -5.7 -5.7 -5.8 1.2 Structural current deficit net of wages -8.7 -10.7 -8.8 -8.3 -9.2 -8.9 -0.2 Import surcharge 0.1 0.6 0.7 1.0 1.1 1.0 1.0 Structural current deficit net of wages and Import surcharge -8,6 -10.2 -8,1 -7.3 -8.1 -7.8 0.8 Interest payments -2.4 -2.9 -3.1 -4.0 -4.2 -3.7 -1.3 Structural current primary deficlt net of wages and Import taxes -11,0 -13.0 -11.3 -11.3 -12.3 -11.5 -0.5 Note: - indilcates surplus. a. Includes coffee tax (2,5 percent of coffee exports); backpayment of duties by EcorPERoL Decreto 399-1986; and special revenue from the tax amnesty (in 1987). Souirce: World Bank data. Wiliam Eastedy 235 age point of GDP of the improvement in the deficit during 1984-89 is explained by nonstructural factors, leaving a fiscal improvement of 3.4 percentage points of C;DP to be explained. Colombia benefited from good luck, but the main part of the adjustment was achieved through its own efforts. Another related question is to what extent the fiscal improvement was achieved at the expense of long-run growth. In the second part of table 5.2 the change in the "structurar' defict is explained by only three specific fiscal components: public investment, wages, and a surcharge on imports (although these factors were somewhat offset by the rse in interest payments). The reduction in public imvestment accounts for 2 percentage points of the reduction in the structural deficit. We examine the composition of this change below. The cut in spending on public wages, which accounts for 1 percent of the improvement in the fiscal deficit, largely reflects expedience, since it was achieved through a decline in real wage rates rather than a ratio- nal retrenchment of public employment. These two items alone fully explain the improvement in the structural deficit over tbe adjustment period. The last two items in table 5.2 axe roughly offsetting. The inaease in interest payments reflects the consequences of the previous buildup of debt, the shift toward paying market interest rates on domestic debt, and the effect of the real devaluation on external inter- est payments. The major source of increased revenue for keeping up with the increased interest burden was the increase in the tax on imports, through a surcharge of 8 percent of import value beginning in 1985. Although increased revenue was desirable, the means cho- sen were again driven mairly by expedience rather than by rational long-run policy. Higher growth in the future is likely to require increased openness, which will eventually make necessary a reduc- tion of tariff rates. (In fact, a trade liberalization was begun in 1990.) To evaluate the reduction in public investment, we need more information about the composition of the cuts. Figure 5.9 shows the evolution of public investment since 1970. As was noted earlier, much of the fiscal expansion that triggered the near-crisis was a result of public investments in electric power and in extractive activities (coal mining and oil production). There was also some expansion in socal sector investments. The sectors that had expanded earlier were pre- cisely those that were cut during the adjustment program, with the largest adjustment in the electric power sector. In hindsight, the latter cuts were questionable, since electric power shortages developed in the early 1990s (partly as a result of mismanagement in the power sector, however). The large reduction of investment in the minig sector mostly reflected the phasing-out of lumpy expenditures on the development 236 Colombia: Avoiding Crises through Fiscal Policy- Figure 5.9. Public Investment by Sector, Colombia, 1970-89 Percentage of GDP 7- 6 5 4 -1 , 970 - 1975 1980 1985 198 LIElectric power X ,s~~~~~~~~~~~~~~- O Sad services Agiriculture S Gura: World BanLk data- Of new oil and coal fields. Expenditure on kiTfastructure actuaUly expanded during the adjustment pr-ogram, largely because of spernd- ing in the telecornununications sector and the construction of a sub- way (of dubious economic value) in Medellin. Mdore questionable for long-run prospects wvas the large reduction in social investments dur- ing the adjustment program and the failure to increase investment in agriculture. The conclusions of table 5-2 are rather nmixed.-They suggest that some means chosen for fiscal adjustment were prejudicial to long-run growth. Improved growth performance wMl probably requiire thie -reversal of some of the fiscal adjustinent measures of 1985 89 and their replacemlent by less distortionary fiscal instnunents to maintain he defict within bounds. Nevertheless, the form of adjustent was less damaging to growth than in otherLatin American countries- William Easterly 237 Fiscal Deficits, Real Interest Rates, and Inflation To analyze the effect of the financing of fiscal deficits on interest rates and inflation, we use the model developed for this study. For conve- nience, the basic structure is presented here. Basic Framework The basic relationship is that of the fiscal deficit (DEE) to its means of financing: (5.1) DEF = s + LdS + tE +BN where L4. is government borrowing from the banking system, Lg is aedit from the Central Bank, Fg is foreign debt, E is the nominal exchange rate, and BNOLS is net other liabfities of the government. The evolution of government borrowing from the banking system and the Central Bank determines the equilibria in the market for credit and the money market, xespectively. These two markets are deared by the domestic interest rate and the price level. Thus it is the composition of the financing of the deficit that determines the interest rate and inflation rate in this model.4 The equiibrium in the credit market is given by the identity that assets equal liabilities for commercial banks (see appendix 5.1): (5.2) Ldg 5 (1-r)Pc[M1(1-C) + QMJ Ldp OTHD where r is the reserve requirement on both narrow money and quasi money, P, is the consumption deflator, M1 is real demand for narrow money, c is the ratio of currency to narrow money, QM is real demand for quasi money, Ldp is private demand for credit, and OTH is net other assets of the baning systeim (As a simplification, we assume one reserve ratio for all types of deposits.) Equation 5-2 merely says that deposits in the banking system, less the amount set aside for reserves, must be equal to private and public credit demand plus net other assets of the banking system. Private credit demand, Ldp, will evolve with private investment, p: (5.3) Ldp = (1 + r)(1 - 6)Ldp(-1) + P- Ipy where r is the inflation rate (in terms of the GDP deflator), a is the rate of depredation of physical capital, P is the GODP deflatorg 4p is private investment, and y is the ratio of private credit to the private capital stock. This expression is derived from the assumption that the ratio of private domestic debt to the nominal value of capital (the leverage ratio) stays constant over time. This implies that the inflation compo- nent of interest payments on the debt (r times the previous stock of debt) will always be rolled over. A constant share of gross investment 238 Colombia: Avoiding Cries through Fiscal Poliy - will be financed by borrowing, but there will be an adjustment for depreciation on past capital, as reflected in past debt. The equilbrium in the money market will be given by the condition that the supply of high-powered money, H, equals the demand: (5.4) HP -M [c+ r(l-c)] + rP QM where the components (cP Ml), [r(1 - c)P- M1], and (rP - QM) correspond to currency demand, reserves on demand deposits, and reserves on qrasi money, respectively. The supply of high-powered money, in turn, must be consistent with the outstanding credit to the government by the Central Bank: (55) Lc=H - NFA6 - OTHC where NFAcb is net foreign assets of the Central Bank and oTHc is net other assets of the Central Bank. Substituting equation 5.4 into equa- tion 5.5 gives the condition that the assets of the Central Bank must equal its liabilities (see appendix 5.1). Econometrically Estimated Behavioral Equations To complete the framework discussed above, we need to specify behavioral equations for money demand, quasi-money demand, pri- vate investment, and total output and to determine interest rate spreads. We also must estirate an equation for private consumption to determine the private saving flow that coiresponds to accumula- tion of money and quasi money. PORTwo.UO DEmANDS. For private portfolio demands, we suppose a three-asset system of money, quasi money, and foreign assets. It is enough to specify behavioral equations for the first two, with foreign assets determined as a residual from the balance sheet condition for the private sectot Although the holding of foreign assets is legally restricted iln Colombia, the existence of a thriving underground econ- omy and flows of remittances from abroad make for a high degree of de facto capital mobility. The existence of a parallel-market premium plies that mobility is not perfect, but the premium is seldom more than 5 percent (and sometimes it is even negative!). We thus model quasi-money demand as reflecting the possibility of substitution between domestic and foreign interest-bearing assets. Real quasi-money demand (deflated by the consumption deflator) is hypothesized to be a function of domestic interest rates, the inflation rate (which also reflects the possibility of substitution into real assets), the foreign interest rate plus the rate of depreciation, and real income. In the estimtion process, we were not able to identify any separate effect of currency depredation in addition to the effect of inflation. Willim Easly 239 This probably reflects the crawling peg system in Colombia, in whikh current inflation is the best predictor of future depreciation (the occa- sional large devaluations are usually unanticipated). We also con- strain the income elasticity of demand for quasi money to be I in the estimation.5 Thus the estimated equation thows the log of the ratio of real quasi money to CDP as a function of the real deposit rate, which is the nominal deposit rate adjusted for the ex post rise in the consump- tion deflator. The results are shown in table 5.3. The real demand for narrow money (also deflated by the consump- tion deflator) is specified to depend on the nominal deposit interest rate and real income. Because the results for a regression in levels were unsatisfactory, we specified an error-correction format. Table 5A shows the fist-stage levels regression of the log of real money on the log of real GDP, the nominal interest rate, and a time trend. The second regression is in differences form, with the lagged residual from the first-stage regression as one of the explanatory variables. Both real income and the nominal interest rate are significant, as is the residual from the first-stage regression. This specification has the intuitively appealing interpretation that the real growth in money demand responds to real income growth and changes in interest Table 5.3. Regression Results for Quasi-Money Demand, Colombia, 1973-88 Variable Coefficient error t-sfatistic Significance Constant -2.47391 0.186360 -13.2749 0.000 Real interest rate 1.43222 0.625282 2.29051 0.038 Lagged error term 0.849253 0.131997 6.43390 0.000 Equation summary No. observations 16 Sum of squared residuals 0.169050 R2 0-920 Adjusted R2 0.9142 Standard error of regression 0.109886 Durbin-Watson statistic 1.89415 F-statistic (1, 14) 160.900 Significance 0.000000 Autocwrrelation estimatin summary Standard error of rho(l) 0.13200 Fiial rho (1) 0.84925 t-value 6.434 Significance 0.000 Note: The dependent variable is demand for log (real quasi money/real GDP). 240 Colombia: Avoiding Crises through Fiscal Policy rates, with a correction for the long-run relationship between levels of money and CDP. The negative time trend in the levels regression implies a secular tendency to move away from MI, which could reflect technical change that economizes on the use of money in transactions. Piu4rE iNvwsTmENrw. As noted at the beginniing of the chapter, the ratio of private investment to CDP shows a steady long-run decline in Colombia, while the ratio of public investment to GDP has been stead- ily rising. Table 5.5 shows the results of econometric testing to see whether there is a statistically significant relationship. The ratio of public capital stock to CDP does indeed enter with a negative sign in private investment regressions and is generally significant over both long and short sample periods and in both level regressions and error-correction specifications. This suggests that there is a high. degree of substitutability between the activities in which the govern- ment invests and private sector activity. Higher public capital could either be lowering the share of private capital in the economy or actualy driving down the rate of return to private capital.6 The nega- tive private-public capital association gives statistical confirmation to the hypothesis that the decline in private investment is linked in part to the secular growth of the state in Colombia. The roles of other factors in the production function are not clearly resolved by the empirical results. The stoclk of private capital would be expected to have an ambiguous effect-increased private capital lowers the rate of return and further depresses private investment, but a higher existing stock of private capital leads to more investment to keep private capital growing at the same rate as the rest of the economy, as well as to replace depredated capitaL The coefficients on private capital are positive and significant in the error-correction results. Similarly, the coefficient on labor, which would be expected to raise the marginal product of capital and increase investment demand, differs greatly across specifications in both sign and magnitude. The ordinary least squares (oLs) coefficient has the expected positive sign, but the error-correction specification yields a counterintuitive, signifi- cantly negative coefficient. The other variable that might be expected to affect investment is the user cost of capital, defined as the real loan rate multplied by the relative price of capital goods. This variable- is of the "correct" negative sign in an oLs levels regression in which it is the sole vari- able, but it changes sign in the error-correction specification. The clear result that emerges from the empirical analysis of invest- ment is the negative relationship between public capital and private investment. This outcome suggests that the cuts in public investment William Easterly 241 Table 5.4. Regression Results for Demand for Ma, Colombia, Data from 1965-88 Standard Variable Coefficent error t-statistic Significance First stage Constant -12.9757 3.47126 -3.73804' 0.001 In (real CDP) 1.73268 0.228106 7.59594 0.000 Noniinal interest rate -0.218988 0.178090 -1.22965 0.233 rlime trend -0.0409874 0.0101437 -4.04069 0.001 Equation summary No. observations 24 Sum of squared residuals 0.0431522 R2 - 0.9727 Adjusted R2 0.9686 Standard error of regression 0.0464501 Durbin-Watson statistic 1.18149 F-statistic (3, 20) 237.672 Significance 0.000000 Seond stageb Constant -0.0342711 0.0272399 -1.25812 0.224 Difference ln(CDP) 1.66381 0.561546 2.96290 0.008 Difference interest rate -0.532078 0.238691 -2.22915 0.038 Lagged residual -0.692107 0.259542 -2.66665 0.015 Equation summary No. observations 23 Sum of squared residuals 0.0339841 I2 0.4369 Adjusted R2 0.3480 Standard error of regression 0.0422923 Durbin-Watson statistic 1.46300 F-statistic (3, 19) 4.91456 Significance 0.010791 a. The dependent variable for the regression is ln(real M,), using 1965-88. b. The dependent variable for the regression is money demand = log(real M,) - log [real M41(-)1, using 1966-88. accompanying the adjustment program may not have been so damag- ing, as long as the allocation of the reductions at the micraoeconomic level was rational. However, the counterintuitive and unstable results on other variables suggest that more research on the determinants of investment in Colombia is needed. Table 5.5. Private Investment in Colombia Private Pu4blic Real Capital capitol loan stock as stock as Laboras Tate x share of sfhare of shtare of retaliue Lagged Error GDP GDP GDP price deperndenit correclion1 Sample Durbin- Adlusted Regression Constant (percent) (percenit) (percent) tegna variable teni: rangoe Watson RZ oLs (ratio to CDF, log) -0.68b 0.16 -0.11 - - 0.75c 1926-88 1.16 0.80 (-2.58) (0.91) (-0.97) (7.15) OLS (ratio to GDP, log) -0.87C -0.27 -0.26b 0.21b 0 Q,71¶ - 1926-88 0.64 0.81 (-3,3) (-1,1) (-2.1) (2.5) (6.93) Stage 1 error .2.09c 1,11C -0.06 1925-88 0.53 0.60 correction (ratio (-8.44) (6.75) (-0.38) to GDP, log) Stage 2 error 0.01 3,39c -0.98C _ - _ _0.53c 1926-88 1.23 0.34 correction (0.86) (5.07) (-2.99) (-4.82) (variables are first differences) Stage 1 error -2.28c 0.30 -0,32a 0.35c _ _ _ 1925-88 0.66 0.67 correction (ratio (-9.85) (1.12) (-1.98) (3.68) to GDPt, log) Stage 2 error -0.07c 3.58c -0.08 _3.33c _ _ -0.71c 1926-88 1.11 0.66 correction (-4.32) (8.45) (-0,32) (-6.55) (-9.13) (variables are first differences) Stage 1 error 0.12a 0.03 -0.11c - 0.08 - 1970-88 2.37 0.74 correction (ratio (1.94) (0,52) (-4.6) (1.65) to GDP, level) Stage 2 error 0.007b 0.32C 0.36e - 0.05b _ -1.3& 1971-88 1.94 0.80 correction (2,43) (4.41) (-3.26) (2.47) (-6.4) (variables are first differences) Stage 1 error -2.8F 0.06 -0.61C - 0.698 - - 1970-88 2.14 0.73 correction (ratio (-26.63) (0.097) (-4.58) (1.49) to GDP, log) Stage 2 error .009b 4.15c -2.13c - 0.53b _.-1.31c 1971-88 2.11 0.85 correction (2,83) (5.4) (-3.34) (2.60) (-6.12) N (variables are first differences) oLS (logs) -2.08c -1.71b - - 1970-88 1.56 0.29 (-50,14) (-2.9) -Not available. a. 10 percent level of significance. b, 5 percent level of significance. c. 1 percent level of significance, Nole: For regressions with lagged dependent variables, the Durbin's h statistic is shown In the Durbin-Walson column. 244 Colomnbia: Avoiding Crises throughl Fiscal Policy PRIVAUE CONSUMPTION. Table 5.6 shows econometric results for pri- vate consumption. We regress real private constumption on real dis- posable income, real government savings, and the real interest rate. The inflation tax is subtracted from the conventional measure of dis- posable income. We experimnent with different dynamic specifica- tions, including a lagged dependent variable and error-correction equations. Degrees of freedom are scant because of limited data availability.. Disposable income is significant in all but one of the regressions.7 In specifications using logs, the elasticity of consumption with respect to disposable income is dose to unity. To test the sensitivity of the results to a larger sample size, CDP is used as a proxy for dispos- able income in some equations, with similar results. The real interest rate is surprisingly positive and significant in several equations, indi- cating that higher real interest rates tend to lower saving. Although this result- is theoretically possible, it seems rather implausible. For the simulation, we use the second equation, in which consumption simply adjusts to disposable income. The restriction of long-ran proportionality of consumption to income is not rejected by the data. GROWTh. Table 5.6 also shows the results of producdvity growth regressions, where output per worker is regressed on private and public capital stocks per worker.8 We find that the questions on the productivity of public capital suggested by the private investment regressions are not really resolved by the growth regressions. Public capital is significant in the levels regression, but there is the usual worry about spurious correlation with nonstationary variables. The magnitudes of the coefficients would imply strong increasing returms in aggregate production.9 The significance of public capital vanishes in an error-correction specification, and the magnitude of the coeffi- cient on private capital is a neodassically more plausible 0.32. Dum- mies for the Great Depression and.World War II are also significant, while the constant term inplies a rate of neutral technical progress of 1.9 percent per year. Although public capital stfill enters positively in the first-stage levels, the error-correction term itself is not significant, weakening confidence that the error-correction specification is appro- priate. The third regression shown simply relates the growth in per worker output to the growth in private capital per worker, with a constant term gain, indicating neutral technical progress. This last equation is the one used in the simulations. INTEREST RATE SPREAD. The other matter that must be addressed is the spread between loan and deposit interest rates, since the former William Easterly. 245 enters into investment and the latter into the demand for money and quasi money. We assume that the spread between the deposit rate, i4, and the loan rate, iL, is explained by the reserve requirement, r, and an exogenous component, io, which would indude profits and other costs of intermediation: (5.6) io = iL(1--i) -io. If we write the nominal loan rate as the sum of the real loan rate and the inflation rate, the nominal deposit rate can be written as a func- tion of the real loan rate and inflation, as follows:°0 (5.7) tD = (1 - r)(rL + 7r) - the real deposit rate will be given as: (5.8) io-7r=(1-r)rL-r7r-iO. EQUILIBRIUM RELxMONSHlS. We can now substitute into the equi- librium relations of equations 5.2 and 5A to determine the equilib- riuxm response of real interest rates and inflation to changes in gov- ermnent money and domestic debt financing. Equation 5.9 shows the equilibrium relation between changes in Central Bank credift to the governmient, Lq,. the inflation rate, r, and the real loan rate, 7L' (5.9) dr=PMr+1c 1[1 M] Rcs -g Ml [c + r(l - ')01 + T 1r)] ____QM'V) + rP QM[r QMr +P( - r) (M[(c + r(l - c} M1 + rIQM)3dL. An increase in inflation will be associated with more monetary financing of the deficit as long as we have not passed the maximum point of the inflation-tax Laffer curve."1 The first expression in equa- tion 5.9 says that an increase in inflation is associated with less real demand for money and quasi money but a higher nominal flow of financing. The second effect is stronger tan the first as long as we are on the upward-sloping part of the Laffer curve. We can analyze the inflation-tax-maximizing inflation rate using the estimated equations of the model. A simulation of the equations for demand for money and quasi money at different inflation rates shows the relationship between the inflation rate and seigniorage revenue Table 5.6. Private Consumption in Colombia Real Lagged Eror Interest Disposable Govemmenit Terms dependentt correction Sample Durbin- AdJf;ted Regression Constan! rate incomiie savings GOP of trade rariable term ARUl) range Watson R2 Two-stage least squares (logs)' 0.47 - O.48b _ _ - 0.49c - 1971-86 -0.74 0.996 (1.09) (1.96) (2.2) Two-stage least squares (logs)a.d -0,056c - 0.72 - - - 0.28b - - 1971-86 1.585 0.640 (-2.73) (4.79) Stage-one error correction (log) 0.32 0.140 0.800 o _ - 0.16 - - 1972-86 0.24 0.999 (1.31) (2.99) (7.86) (1.69) Stage-tvo error correction: variables -O.i02 0.17e 0.79e _ 0.24b -0.9&c 1974-86 -0.19 0.885 are first differences (-0.45) (3.88) (7.89) (2.23) (-2.8) Stage-one error correction (log) 0.30 0.18c - _ ,95r _ - 1966-88 0.563 0.997 (1.35) (2.39) . (56.10) Stage-two error correction: variables -0.0007 0.09 - - 0.96a - - -0.20 - 1967-88 1.31 0.675 are first differences (-0.10) (1,50) (6.65) (-1.19) Stage-one error correction (log) 1.260 0.29 - -0.0005 0.880 - - 7 - 1974-88 .0.937 0.992 (2.92) (3.43) (-0.11) (28.65) Stage-two error correction: variables -0,0005 0.140 _ -0.003 0.910 - - -0.30 - 1975-88 1.159 0.766 are first differences (-0.08) (2.53) (-1.15) (6.27) (-1.35) Two-stage least squares (log) -0.48" - 1,050 - -Q,Q5b - - 0.09 1972-86 1.56 0.97 (-2.28) (63.01) (-2.11) (0.35) Stage-one error correction (logs) 0.13 0.13c 0.99" 0.001 - -0,04 - - - 1974-86 1.689 0.997 (0.41) (2.37) (40.3) (0.39) (-1.82) Stage-two error correction: variables 0.0005 0.14c 0.97P -0,002 - -0.03 - -1.010 -. 1975-86 1.597 0.899 are first differences (0.13) (3.41) (9.64) (-0.98) (-1,38) (-2.59) Two-stage least squaTes (Iogs)s 0.744r 0.096 - - 0.3W 0,04b O.54e _ - 1971-88 1.969 0.998 (2.99) (1.24) (2.45) (1.84) (3.28) Private Psiblic stock of stock of World Error (rowlih regressions capital per capital per Wrr It Depression correclova Sample Durbin- Adjusted (log of outipt per torker) Calistant worker woorker dufmmy dutmy AR(11 ARM) lenn range Watson Rx -1.27e o.7e 0.48' _0.03l - - 1.4' -0.6' - 1927-88 1.97 0.997 (-3.43) (7.25) (7.0) (-1.92) (13.17) (-5.72) Stage-one error correction -0.84' 0.63' 0.49g -0.06b -0.14 - _ 1925-88 0.24 0.980 (-4.54) (12.33) (12.71) (-1.69) (-2.92) Stage-two error correction: variables 0,02' 0.32' -0.08 -0.03" -0.05 - - 0.005 1926-88 1.47 0.115 are first differences (3.72) (2.47) (-0.60) (-1.73) (-2.7) (0.08) OLS: variables are first differences 0.02' 0.31- -0.03b -0.05a - - - 1926-88 1.43 0.139 (4.65) (2.47) (1.73) (-2.86) Not available, a. Instruments: government consumption, exports. Instrumented variable: disposable income, b. 10 percent level of significance. c. 5 percent level of significance. d. In this equation it was imposed that the coefficients for disposable income and the lagged dependent variable sum to 1. e, 1 percent level of significance. f. Instruments: time trend, lagged disposable.income, per capita govemment consumption, real Interest rate. Instrumented variable: disposable income. g. Instruments: time trend, lagged cDP, per eapita government consumption. Instrumented variable: CDP. Note: In equations with a lagged dependent variable, the Durbin's li statistic Is reported; in other equations, the Durbin-Witson statistic is given. 248 Colombia: Avoiding Crise througlt Fiscal Policy (figure 5.10). We see that maximum seigniorage (defined as the change in the money base over nominal GOP) of about 2.7 percent of GDP is achieved at inflation of a little less than 100 percent. His- torically, inflation has been well below the seigniorage-naxiizming rate. The effect of a higher real interest rate on money creation is ambig- uous. This is because higher interest rates have an ambiguous effect on the demand for base money: they lower demand for narrow money but raise demand for quasi money, and base money is a linear combination of the two (with the coefficients given by the currency- to-M1 ratio and the reserve ratio). Base money is more likely to rise in response to an increase in the real interest rate the higher is the interest rate elasticity of quasi-money demand in relation to money demand and the higher is the ratio of existing quasi money to M1. Thus, equilibrium in the money market could imply either a negative or a positive relationship between the real loan interest rate and the inflation rate for a given stock of Central Bank credit to the government. Equation 5.10 shows the equilibrium relation in the credit market between real bank credit to the government, L14/P, inflation, 7r, and the real loan interest rate, rL (5.10) d -(i - r)(g1 - c)M' - r(1 - r)QM'jdir + [(1 - r)2(L - c)Ml + (1 - r)2QM' - y;ldrL. An increase in the real interest rate inaeases credit to he government because it reduces private investment and demand for credit and increases deposits in the banking system.n Higher inflation reduces credit to the government because a higher inflation rate for a given real loan interest rate implies a lower real deposit rate (from equation 5.8). Quasi-money demand is therefore reduced by higher inflation. Demand deposits are also reduced, since these are a function of nomi- nal interest rates. Thus equilbrium in the caedit market for a given stock of government debt implies a positive relationship between the real loan rate and inflation. Figures 5.11 and 5.12 show the joint determination of the real loan interest rate and inflation in the money and credit markets. Figure 5.11 illustrates the case in which the money-market equilibrium implies a negative relafionship between the real interest rate and inflation. The locus of debt equilibria is always upward sloping. An increase in government borrowing from commercal banks (Ld,1P) shifts the locus of debt equiliba upward. This implies a higher real interest rate (rti) and a lower rate of inflation (7rw). The reason for William Eastey 249 Figure 5.10. Seigniorage Revenue as a Function of the inflation Rate, Colombia Percentage of GDP 2.8 2.6 2.4 2.2 2.0. 1.6 tA |- |i 0 25 50 75 100 125 150 175 200 inflation rate (percent) Sourcc World Bank data. lower inflation is that the demand for base money is increased by higher real interest rates. High money demand implies a lower pnce level (and rate of inflation) for a given supply of money base. An increase in Central Bank credit to the government (Lg) increases the money base, which shifts the money market equilibnum curve upward. Both inflation and the real interest rate (rL2 and 7r2) increase. The real interest rate increases because higher inflation represents a tax on demand and quasi-money deposits for a given real loan rate, so tat deposits tend to fall urdess there is an offsetting rise in interest rates. Figure 5.12 ilustrates the case in which the locus of money market equiliba is upward sloping. An increase in government borrowing causes the inflation rate, as well as the real loan rate, to increase. The higher real loan rate causes the demand for money base to fall, and iation rises for a given supply of money base. An increase in Cen- tral Bank credit to the govermnent still causes an increase in interest 250 Colombia: Avoiding Crise through Fiscl Policy Figure 5.11L Effects of Changes in Defict Financing (Case 1) Real loan rate Increase in L Increase i L,,,P Debt market rL) Money market 0 2 Inflation rate rates and inflation, for tie reasons given above. Although both domestic bcrrowing and money creation cause -real interest rates-and inflation to increase, it is clear from figure 5.12 that debt financing has a proportionately larger effect on real interest rates than on inflation, compared with monetary financing. Sitmulation Results We use the model to perform counterfactual simulations within and beyond the sample period. We first calibrate the exogenous variables to reproduce the observed inflation and real interest rates over the period 1987-89.3 This period is a mixture of within-sample and out- of-sample observations, since various regressions end in 1986, 1987, and 1988. We then consider changes in the fiscal deficit and its financ- ing to evaluate how the deficit translates into changes in inflation and interest rates. Several of the regression equations have lagged error terms on the right-hand side. These are included in the model. The first simulation, shown in table 537 as differences from the base case, is an increase in public investment financed by domestic bor- rowing. The fiscal expansion of 1-2 percent of GDP in 1987, 0.8 percent William Easterly .251 in 1988, and 0.7 percent in 1989 leads to an increase in the real interest rate ranging from 3 percent in 1987 to 5 percent in 1989.14 The nse in the real interest rate causes a drop in the ratio of private investment to GDP of 0.5 percent in 1987-88 and 0.8 percent in 1989.15 Although this is not as great as the increase in public investment, growth falls because orly private investment affects growth. As pointed out above, the effect of a debt-financed fiscal expansion on inflation is ambiguous. In this simulation there is a slight increase the first year, a fall in inflation the second year, and an increase the third year This complicated pattem is the result of several offsetting factors. The increases in growth in 1988 and 1989 tend to lower infla- tion because higher growth stimulates greater demand for money, implying a lower rate of inflation for a given amount of money cre- ation. However, the increased interest rates have two offsetting effects on demand for base money: a positive effect on reserves on quasi money, and negative effects on demand for currency and on reserves on demand deposits This simulation can be interpreted counterfactually as what would have happened had the fiscal adjustment described above not taken place. Ihus, the difference between this simulation and the actual Figure 5.12. Effects of Changes in Deficit Financng (Case 2) Real loan rate Money market Increase in Leg EIncrease iniL4fP r ~ ~ ~~~~~~, ~D~elbt market rLI Inflation rate 252 Colombia: Avoiding Cnses through Fiscal Policy Table 5.7. Case of Increased Public Investment Financed by Domestic Borrowing (difference from base case, as percentage of GD?, unless otherwise specified) Chanse in ratio to GDP -197 1988 1989 National accounts (real) Private consumption 0.00 0.04 0.01 Private investment -0.50 -0.68 -0.82 Public investment 1.22 0.85 0.68 Disposable income 0.00 0.03 -0.05 Capital stock Public 0.00 1.26 2.10 Pnivate 0.00 -0.33 -0.76 Mon etay accounts Stocks Money -0.22 -0.13 -0.33 Quasi money 038 0.51 0.63 Money base -0.01 0.08 0.02 International reserves -0.01 0.10 0.03 Public sector deficit 1.22 0.85 0.68 Public deficit financing flows Foreign 0.00 0.00 0.00 Central Bank 0.00 0.00 0.00 Rest of financial system 1.22 0.85 0.67 Other liabilities of government 0.00 0.00 0.00 Stock of credit from Central Bank to government 0.00 0.02 0.01 Rest of financial system to government 1.21 1.76 2.00 Rest of financial system to private sector -0.23 -0.33 -0.44 Other variables (percentage absolute change) GoDP growth 0.00 -0.12 -0.16 miflation 0.13 -1.23 1.10 interest rates Loan rate 3.84 3.72 7.65 Real loan rate 3.00 4.00 5.00 Deposit rate 3.41 3.29 6.76 Real deposit rate 2.58 3.59 4.37 outcome represents the consequences of adjustment compared with continuing debt-financed fiscal expansion. The implication is that the fiscal adjustment had the effect of raising growth by increasing pn- vate investment and lowering real interest rates. This effect on growth is meaningful mainly in the medium run, since the model does not incorporate any effects of demand on output. William Easterly 253 Table 5.8 presents the results of a simulation of an increase in public investment financed by money creation. An increase in inflation of 15 percentage points per year is triggered by higher ratios of public investment to GOiP and by a public deficit of 0.3 percent in 1987 and 1.1 percent in 1988-89. The reason a smaller increase in the deficit leads to the same inflation rate in the first year as that associated with higher deficits in the next two years is the portfolio shift effect. An increase in inflation causes a one-time shift out of money, which Table 5.8. Case of Increased Public Investment Fmanced by Money Creation (difference from base case, as percentage of GDP, unless otherwise spedfied) Change in rtios to GDP 1987 1988 1989 National accounts (real) Private consum-ption -0.22 -0.18 -0.23 i1rivate investment -0.32 -0.15 -0.13 Public investment 0.26 1.18 1.09 Disposable income -0.33 -0.21 -0.28 Capital stock Public 0.00 0.31 1.44 Private 0.00 -21 0.30 Monetury accounts Stocks Money -1.07 -0.56 -0.46 Quasi money 0.12 0.03 0.02 Money base -0.47 -0.25 -0.21 International reserves -1.06 -1.9$ -2.80 Public sector deficit 0.26 1.18 1.09 Public deficit financing flows Foreign 0.00 0.00 0.00 Central Bank 0.26 1.18 1.09 Rest of financial system 0.00 0.00 0.00 Other liabilities of government 0.00 0.00 0.00 Stock of credit from Centrl Bank to government 0.01 0.97 1.64 Rest of financial system to government -0.08 -0.07 -0.05 Rest of financial system to private sector -0.15 -0.15 -0.15 Other wriables (percentage absolute change) GDP growth 0.00 -0.08 -0.03 Inflation 15.00 15.00 15.00 Interest rates: Loan rate 19.60 17.98 18.06 Real loan rate 1.91. 0.89 0.79 Deposit rate 17.42 15.89 15.96 Real deposit rate 0.84 0.21 0.11 254 Colombia: Avoiding Cises through Fiscal Policy sharply decreases the amount of money financing available in the first period. In the succeeding periods the quantity of money demanded grows in accordance with the new rate of inflation, without any off- setting portfolio shift. The money-financed increase in public spending raises the real interest rate, as predicted by the comparative statics set out above. This is becaus! the higher inflation tax on deposits for a given real' loan rate,requires an increase in the interest rate if deposits are to increase again and equilibrium is to be maintained. The higher real interest rates have a slight negative effect on private investment and thus on growth. The higher inflation also has a small negative effect on private consumption. Higher inflation increases the inflation tax on money balances, decreasing the after-tax disposable income of consumers. If the simulation is interpreted counterfactually, the implication is tlat the fiscal adjustment of the 1980s, as compared with continuing money-financed fiscal expansion, had the effect of lowering inflation and increasing private consumption and investnent. The final simulation we consider is a substitution of money for debt financing, leaving the deficit unchanged (table 5.9). The increase in money financing of 0.2 percent in 1987 and 1.1 percent in 1988-89 is again associated with increased inflation of 15 percent per year. The effect on the real interest rate is ambiguous, with offsetting effects of an increased inflation tax on deposits and a fall in government bor- rowmg requirements. in the first year, when the decrease in govem- ment borrowing is small, the increased inflation tax dominates, rais- ing the real loan rate by 1.3 percent. In the second and third years the larger decrease in government borrowing dominates, and the real interest rate falls first 3 and then 5 percent. The fall in real interest rates implies a rise in private investment of 0.8 percent of CDP by 1989. Private consumption again falls because of the increased inflation tax. Substituting money for debt finance is favorable for saving and growth. However, we must be c.Adious in interpreting this result: the efficiency losses associated with inflation are not captured by the model, and they could well dominate the results given here. The Real Exchange Rate and the Fiscal Deficit The model in this section emphasizes that the real exchange rate is not just the outcome of the govemment's exchange rate policy; rather, it reflects endogenous economic forces that affect the demand for and supply of tradable and nontradable goods. Of these forces, the fiscal deficit is especially important because it represents net demand pressure that is policy induced. The Colombian adjustment William Easterly 255 Table 5.9. Case of Substituting Money Creation for Debt Financing (difference from base case, as percentage of CDP, unless otherwise specified) Rato to GDP 1987 1988 1989 National accounts (real) Private consumption -0.22 -0.19 -0.26 Private investment -0.21 0.51 0.84 Public investment 0.00 0.00 0.00 Disposable income -0.33 -0.21 -L N9 Capital stock Public 0.00 0.04 -0.06 Private 0.00 -0.14 0.20 Monetary accounts Stocks Money -1.03 -0.28 -0.15 Quasi money 0.04 -0.42 -0.64 Moneybase -0.46 -0.25 -0.25 International reserves -1.06 -1.98 -2.82 Public sector deficit 0.00 0.00 O.0 Public defict financing flows Foreign 0.00 0.00 0.00 Central Bank 0.26 1.18 1.06 Rest of financial system -0.26 -1.18 -1.06 Other liabilities of government 0.00 0.00 0.00 Stock of credit from Central Bank to government 0.01 0.97 1.61 Rest of financial system to government -0.34 -1.42 -2.03 Rest of financial system to private sector -0.10 0.04 0.20 Other variables (percentage absolute change) GDP growth 0.00 -0.05 0.13 Inflation 15.00 15.00 15.00 Interest rates: Loan rate 18.72 1268 10.31 Real loan rate 1.27 -2.81 -4.67 Deposit rate 16.63 11.21 - 911 Real deposit rate 0.28 -3.06 -4.71 program initiated in 1985 achieved a substantial devaluation of the peso without a large acceleration of infltion as a result of the sub- stantial fiscal adjustment that accompanied it. Determination of the Real Exchange Rate This chapter folows closely the methodology given in chapter 2.1t The model predicts that the real exchange rate will appreciate in response to an increase in the terms of trade. An increase in the terms 256 Colombia: Avoiding Cdses through Fiscal Policy of trade shifts supply away from nontradables and demand toward nontradables, causing a real appreciation. The model also predicts that the real exchange rate will appreciate in response to a fall in the trade surplus, which increases spending in relation to iL:ome and increases the demand for nontradables, so that their relative price increases (there is a real appreciation). According to the Rodriguez models in chapter 2 of this volume, the real exchange rate is an ambiguous function of the level of govem- ment spending. An increase in government spending for a given resource surplus (and thus a given level of total spending) implies a redistribution of spending from the private to the public sector. If the government has a higher propensity to spend on nontradables than does the private sector, increased government spending implies a net increase in demand for nontradables, leading to a real appreciation. Conversely, if the government has a lower propensity to spend on nontradables (in other words, a higher propensity to spend on importables and exportables), increased spending results in real depreciation. The model is estimated. for Colombia over the period 1967-87; table 5.10 presents the results. We include a lagged dependent variable term to represent partial adjustment of the real exchange rate to changes in the fundamentals. The coefficient indicates that 39 percent of the long-run effect of a change in the fundamentals is realized the first year and 77 percent in the first three years. All the variables are significant (although the resource balance is not quite significant at the 5 percent level) and have the correct sign. A terms of trade. increase leads to an apprecation of the real exchange rate. (As else- where in this chapter, an increase in the real exchange rate signifies depreciation.) An increase in the resource surplus causes a deprecia- tion of the real exchange rate. The sign on the government expenditure variable is positive, indi- cating that increased government spending causes a real deprecia- tion. As indicated above, the sign is theoretically ambiguous. The sign found here would imply that the goverrnment devotes a lesser share of its spending to nontradables than does the private sector. In other words, government spending in Colombia is very import- intensive. Determination of the Resource Balance To determine the trade surplus, we consider a variation on the Rodriguez model that incorporates some of the details on govern- ment deficit financng from the section "Fiscal Deficits, Real Interest Rates, and Inflation," above. William Easterly 257 Table 5.10. Regression Results for Real Exchange Rate (EXCH-RL) Standard Variable Coefficient eror t-statistic Significance Constant 1.45218 1.36576 1.06327 0.303 ln(ExcH-RL(-1)) 0.612230 0.267986 2.28456 0.036 ln(TnDE) -0.221399 0.0808159 -2.73955 0.015 RSCBAL&GDP 0.0299868 0.0146600 2.04548 0.058 In(ExPToT&cDP) 0.421882 0.131091 3.21824 0.005 Equation summary No. observations 21 Sum of squared residuals 0.0548271 R2- 0.8125 Adjusted R2 0.7656 Standard error of regression 0.0585380 Durbin-Watson 1.83021 F-statistic 17.3320 Significance 0.000011 Notae. ln(Excii-RL), log (real exchange rate); In(TTRADE), o10g (terms of trade); RscuLkCrDP, resource balance (percentage of GDP); ln(EXPrOT&GDP), log (total public expenditurelGDP); PRMFSUR&C2P-SP, prim fiscal surplus (percentage GDP). Depreci- ation is up. Method, two-stage least squares; dependent variable, LN(Exc-IRL); sample period, 196747; instrumented variables: constant, LN(ENCH-RL(-1), LN(EXaC-RLn-1)), LN(TrItwE), RscBAL&ctnP(-1), LN(EXTOT& CtDP), PRMFSUR&GDP-SP. THEORETICAL DERAION. The trade surplus (ts) is given as the sum of the resource balance of the private sector (saving minus invest- ment) and the primary surplus, p, of the public sector: (5.11) ts=sp-i(p r) +p. (-) Private investment is a negative function of the real interest rat-, r. The real interest rate is determined by the equilibrium condition im the market for domestic government debt d.: (5.12) ~ds = dg (r)- * ~~~~~(+) The total derivative of equation 5.11 gives us the change in the trade surplus as a function of changes in private saving, the real interest rate, and the government primary surplus: (5.13) d(ts) = d(sp) - i5dr +J dp. We will assume in what follows that the derivative of private saving 258 GCdombia: Avoiding Cises through Fiscal Policy with respect to the real interest rate and the primary surplus is zero. (Ricardian equivalence does not hold.) The real interest rate can be determined from the government financing constraint. The government borrows a fixed percentage of GDP, f, from abroad. This is determined exogenously, either as a government policy decision or by international capital market con- straints. The residual source of financing is domestic borrowing, given by: (5.14) d =-p -f + (r-g)d5 + rdg where p is the primary surplus, dg and d,? are domestic and foreign debt, r and r* are domestic and foreign interest rates, and g is the growth rate of GIDP. In discrete time, we can write the level of govem- ment debt as: (5.15) ds = -p-f + (r -g + 1)ds (-1) + rtd(-1). The amount of govemment debt in equation 5.15 must equal the amount demanded by the public in equation 5.12. When r, p, andf are allowed to vary, the following relationship between these variables is implied: (5.16) d;dr = -dp - df + dr4dg (-1)] from which we obtain r as a function of p andf. (517) dr- -dp-df zd; - ds The real interest rate is a negative function of the primary surplus, p, and a negative function of foreign borrowing, f. Either an increase in the surplus or a shift toward foreign borrowing for a given surplus tends to relieve the pressure on domestic financial markets and decrease the interest rate. Since p and f enter symmetrically into the equation for r, a decaease in p that is exactly offset by an increase in f will have: no effect on r. In other words, an externally financed fiscal expansion has no effect on domestic interest rates. Substituting into equation 5.13, we obtain the resource balance as a function of the primary surplus and the amount of foreign borrowing: (5.18) _ _ __ _ _ d(ts)[1 + d jdp + d-D dgf The trade surplus is a negative function of foreign borrowing, f, and a positive function of the primary surplus, p (if the coefficient on f is less than one). Note the restriction that the coefficient on p be 1 plus William Easterly 259 Table 5.11. Regression Results for Resoiurce Balance (RSCBAL&GDP) Standard Variable Coeffident errur tstatistic Significance Constant 2.19633 0.507417 4.32845 0.001 PRMFSUR&GDP-SP 0.448078 0.139534 3.21125 0.005 FINExIX&GDP-2 -0.412757 0.157042 -2.62832 0.018 Equation summary No. of observations 19 Sum of squared residuals 51.4009 R2 0.6834 Adjusted R-f 0.6438 Standard error of regression 1.79236 Durbin-Watson 1.73298 F-statistic 17.2646 Significancei0.000101 Note- RSCBAL&GDP, resource balance (percentage of cDr); PRMSuR&CDP-SP, primary fiscal surplus (percentage of cGr); FRNExT&GDp-2, extemal financing (percentage of GDP). Method, ordinary least squares; dependent variable, RsCSL&cDP; sample period, 1970-88. the coefficient on f. This means that a decrease in p matched exactly by an increase inf wll reduce the resource balance one for one. Simce an externally financed fiscal expansion has no effect on the real inter- est rate, as shown above, it can only spill into the resource balance one for one. We use equation 5.18 as the basis for our estimated equation. As is shown in table 5.11, both variables are statistically significant and of the correct sign, and the other regression statistics are satis- factory. The coefficient on p minus the coefficient on f is equal to 0.86 (instead of 1, as predicted bv the theory). A formal test reveals that the violation of the theoretical restriction is not statistically significant. We can use this estimated equation to assess the behavior of the resource balance over the period 1975488. We simulate the equation for base period (1975) values and compare the results with those of the simulation that uses actual values of the primary surplus and public external financing. As figure 5.13 shows, the resource balance went from a sizable surplus in 1977 to a large deficit in 1983 before recovering to yield a surplus in 198688. We see from the figure that the extemal financing effect is the most important factor in explaining the resource surpluses of the late 1970s, with the exception of the strong effect of the primary surplus in 1978. In the 1980s the two 260 Colombia: Avoiding Crises through Fiscal Policy Figure 5.13. Decomposition of the Resource Balance, Colombia, 1975-88 Percentage of C;DP 6 4 2 a -2 -4 -6 I I t It I I t I I I 1975 1976 1977 1975 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 - Fmndng effect ------ Primury surplus Total Source: World Bankc dita. effects moved together. Extemally financed fiscal expansion was responsible for the deterioration of the resource balance the early 1980s. Begining in 1984 fiscal contraction accompanied by reduced foreign borrowing helped improve the resource balance. joint Simulation of the Real Exchange Rate and the Resource Balance The models in the first part of this chapter can be combined to get an idea of the role of fiscal variables in determining the real exchange rate through their effect on the resource balance. The simulated values of the resource balance, alternately varying the prmary sur- plus and public foreign borrowing, are substituted into the real exchange rate model and simulated. The simulated real exchange rate outcome is compared with an equilibrium in which all variables are fixed, and the difference represents the dhange in the real exchange rate attrbutable to changes in the primary surplus and public foreign borrowing, respectively. The unexplained portion of the resource bal- ance is now induded under the residual. Figure 5.14 shows the decomposition of the real exchange rate, including the indirect effects of the govemment primary surplus and public foreign borrowing. (As always, depreciation is up.) The effects William Eastery 261 Figure 5.14. Exchange Rate Decomposition, Direct and Indirect Effects, Colombia, 1975-87 Log change since 1975 0.3 0.2 ~~~---- 0.1 _ _ .0L * 0.1 * -0.2 -0.3 I 1 , , , 1975 1976 1977 1978 1979 1980 $981 1982 1983 1984 1985 1986 1987 Expenditure ; Terms of trade - - External financing - -t-- Priiay surplus Total Source. Authoes calculations. of changes in the government surplus and changes in financing tend to move together. An externally financed fiscal expansion played a major role in the appreciation of the real exchange rate in the early 1980s, just as the simultaneous reduction of the fiscal deficit and foreign borrowing supported the depreciation after 1984. This model gives some insight into the causes for some of the most important changes in the real exchange rate. The strong appreciation of 1977 was almost entirely attributable to the sharp increase in the terms of trade. However, the appreciation was exacerbated by a reduction in the government primary surplus, contrary to what would have been a sound policy of increasing the surplus to keep the economy. stable. Histories of the period often stress the attempts of monetary authorities to sterilize the reserve inflows from the coffee boom. In this model such monetary policy would work by reducing net public foreign borrowing. This, however, was not a major factor in 1977-78, at the height of the coffee boom. The 1977 experience contrasts with the policy package imple- mented in 1986 during the second, and smaller, coffee boom. The appreciation implied by the change in the terms of trade was more thar offset by the combination of an increased primary surplus with a 262 Colombiw Awiding Crises through Fisca Policy large reduction in foreign borrowing, and a real depreciation was achieved. Condusion In Colombia the management of fiscal deficits and their financing has been generally sound. Episodes of loose fiscal policy have been minor in comparison with other Latin American countries. The near-crisis of the early 1980s was addressed in a timely way through a sharp fiscal adjustment. Analysis shows this adjustment to have been a combina- tion of good luck and fundamental policy changes, with more emphasis on the latter. The means of fiscal adjustment chosen were sometimes suboptimal from the standpoint of long-run growth, and a degree of fiscal reform will eventually be needed to reverse some of the measures implemented during 1985-89. Still, the adjustment compares well with those of other Latin American counties. Two surprising results of the study are that public capital has a negative effect on private investment and that the contribution of public investment to growth is highly uncertain. This would imply that the cuts in public investment in the late 1980s were not ternily prejudicial to growth, although some specfic cuts were poorly chosen. The analysis in this chapter shows a dose relationship in Colombia between the means of financng the fiscal deficit and inaaoeconomic outcomes. The simulation model used here showed how money- financed and domestic-debt-financed fiscal deficits translate into inflation and changes in the real interest rate. Roughly speakdng, a debt-financed deficit increase of about 1 percent of GDP translates into an increase in the real interest rate of 3 to 5 percent, and a money- financed deficit inrease of about 1 percent translates into 15 percent- age points of additional inflation. We then traced the relationship between externally financed fiscal deficits and the real exchange rate and fouind that a good part of the change in the real exchange rates over 1975-87 is attributable to fiscal policy. Colombia's trademark is moderation: moderate budget deficits, moderate inflation, and decent, although not specacular, growth. This moderation enabled Colombia to be the dog that did not bark- that is, the Latin country that did not have a debt crisis. Colombia's special character continues to be evident up to the present- During 1990-92 the Colombian authorities maintained strict fiscal discipline (budget deficts of less than 1 percent of GDP) in the face of adversity- drug-related violence, an electric power shortage, and low coffee prices. Inflation accelerated somewhat, to about 30 percent, but GDP growth continued at a decent pace, 3.2 percent over 1990-92. Co- Iombia's long-run prospects have brightened with a major new oil William Easterty 263 discovery. Compared with the painful adjustment of its Latin neigh- bors, Colombia's experience is a good advertisement for a policy of low and stable public sector deficits. Appendix 5.1. Balance Sheet Identities for Mone, and Credit Market Equilibria Balance sheet of commercial banks Assets Liabilities Ldg=credit to (1- c)M1 - Pe demand government deposits Ldp= credit to private QM- PCquasi money sector OTHD = net other assets rP,- M41 - c) reserves on demand deposit rP, QM = reserves on quasi money Balance sheet of central bank Assets Liabilities Lv =a credit to the c P M1 = arrency government r(I - c)P - M1 = reserves on NFAa = net foreign assets demand OTHC = net other assets deposits r - P QM =reserves on quasi money Appendix 5.2. Defiitions of Variables, Sources, and Data Disposable income: disposable income less inflation tax (inflation rate times MI), deflated by consumption deflator. Source: for disposable income, Departamento Administrativo Nacional Estadistica (DANE), Colombia. Government saving: consolidated public sector balance as a ratio to GDP, constructed from World Bank and Banco de la Republica data. Nominal interest rate: interest rate on certificados de dep6sito a t&mino (cDT). Source: Revista del Banco de [a Rep4blica. Real interest rate: (1 plus nomiunal cDT rate) divided by (1 plus per- centage change in consumption deflator) minus 1. Private capital stock: previous year's capital stock multiplied by (1 minus depreciation rate) plus current year's private investment. (The 264 Colombia: Avoiding Crss through Fiscal Policy depredation rate is assumed to be 5 percent.) Initial year capital stock is deternlned as a ratio to GDP in that year, with the ratio given by the average over the period of the ratio of private investment to GDP divided by the sum of the average growth rate and the depreciation rate. Public capital stock: same as for public investment. Source: for public investment in 1975 prices, DANE; for public investment, 1925-88, same as for real GDP. Real GDP: for 1965-S, GDP in 1975 prices from DANE. For senes 1925-88, GDP in 1950 prices spliced data from Comisi6n Econ6mica para Amrica Latina y el Carne (cEPAL), El Desarrollo Economico de Colombia. Anexo estadistico, 1957, with ivF data for 1950-88. Real M2: M1 deflated by consumption deflator. Source: for 1964-1, Revista del Banco de la Repuiblica; for 1982-88, World Bank. Rel private consumption: private consumption in 1975 prices, from DANE. Real private investment: private investment in 1975 prices, from DANE. For 1925-88, same source as real GDP. Real quasi mone: sum of cDr and dep6sitos de ahorro tradiconal in the banking system, deflated by private consumption deflator. Source: for quasi money, Reuista del Banco de la Republica; for private consump- tion deflator, DANE. Reative price of capital: ratio of investment deflator to GDP deflator. Source: DANE. Varables for real exchange rate and resource balance regressions: EXCcH-RL: average nominal exchange rate (pesos/dollar) multiplied by (U.S. wholesale price index divided by Colombian consumer price index). Source: flF, Intemational Financial Statistics (IFS). TFRADE: Colombian export price index divided by import price index. Source: iFs. RSCBAL&GDP: resource balance (exports minus imports of goods and nonfactor services) as percentage of GDP. Source: DANE. ExProT&GDP: total consolidated public expenditure as percentage of GDP. Source: for 1967-74, Bird (1984), table 2-1; for 1975-88, World Bank data. PRMFSUR&GGDP: prmary fiscal surplus (fiscal balance exduding interest payments) as percentage of GDP. Source: for 1967-74, Garcia Garcia (1989); for 197548, World Bank data. FNExT&GDP: net external financing of public sector = net flow of medium- and long-term public debt plus net flow of short-term public debt minus change in international reserves of Central Bank. Source: for public debt, Banco de la Repiblica, "Deuda Externa de Col- ombia," July 1989; for reserves, iFs. Appendix 5.3 Data for Regressions for Consumption, Investment, and Money (millions of pesos, except as otherwise specified) Percentnoe ch1angse tn Noiiial RIeal Relalive Real Real Real private coiSltption Real {Iterest privante prc aof Real interest disposablc Year conswnjlption deflator GDP rate finvestniwvit Real M , capitna qlasi Ionncy rate income 1960 1961 . _ _ 1962 _ 1963 - - - - - - - _ - 1964 - - - _ _ _ _ _ 1965 168,362 - 235,051 0.052 - 36,147 1.0181 - - 1966 180,319 0.16263 247,360 0.052 - 35,443 1.0531 - -0,09515 1967 184,146 0.08401 257,588 0.052 - 39,849 1.0425 - -0.02953 1968 196,421 0.07639 272,871 0.052 - 42,485 1,0538 - -0.02266 1969 211,216 0.07442 289,523 0.052 - 47,264 1.0650 - -0.02087 - 1970 224,576 0,08752 307,476 0.052 37,682 50,950 1.0422 - -0.03266 248,328 1971 241,733 0.11918 325,831 0.052 38,102 50,509 1.0220 - -0.06003 259,173 1972 255,776 0.12116 350,813 0.052 34,560 56,028 1.0306 10,953 -0.06169 289,002 1973 268,183 0.19173 374,398 0.052 39,015 60,768 0.9945 12,577 -0.11725 308,725 1974 284,365 0,26605 395,910 0.167 46,181 57,382 1,0045 15,287 -0.07823 323,173 1975 292,779 0.24437 405,108 0.281 41,109 58,915 1.0000 17,507 0.02944 325,480 1976 313,199 0.20797 424,263 0.281 44,907 65,716 0.9907 21,640 0.06046 347,995 1977 325,847 0.27101 441,906 04281 38,684 67,414 0.9369 24,078 0.00786 371,026 1978 353,212 0,17953 479,335 0.265 49,598 74,485 0.9840 29,201 0.07246 397,245 1979 368,439 0.26081 505,119 0.266 50,813 73,400 1.0015 26,108 0.00412 414,867 1980 384,698 0.26237 525,765 0.302 51,280 74,344 1,0019 38,757 0.03140 432,562 1981 395,910 0.25986 537,736 0,373 54,605 71,518 1.0148 51,500 0.08980 437,738 1982 401,759 0.24731 542,836 0.379 51,687 71,907 0,9842 47,781 0.10558 444,612 1983 403,572 0.20185 551,380 0.337 54,892 74,613 0.9726 50,462 0.11245 453,812 1984 415,128 0.20446 569,855 0.347 49,222 76,427 0.9802 52,999 0.11834 465,727 1985 422,917 0.23528 587,561 0.353 44,682 79,288 1,1014 57,716 0.09530 468,961 1986 435,636 0.24118 617,527 0.312 50,652 78,432 1.0794 63,809 0,05705 488,433 1987 453,196 0,27877 654,853 0.342 - 81,543 - 65,152 0,04944 - 1988 469,692 0,29768 679,345 0.318 79,040 - 56,128 0.01566 - Not available. Souarce: See appendix 5.2. Appendix 5.4. Data for Regressions for the Real Exchange Rate and Resource Balance (percentage of cD, except as Indicated) Year EXCH-RL (itidex) FINEXT&GDP-2 7TRADE (inidex) PRMFSUR&GDP-SP EXPTOT&GOP RSCBAL&GDP 1967 109 0.49 69.03 -3.60 25.05 0.8812 1968 118 -1.16 71.35 -5.10 27.81 -1.0454 1969 119 -0.86 71.23 -7.70 . X.97 -0.4732 1970 122 2,11 87.27 -6,90 30.69 -1.2842 1971 125 1.96 81.57 -7.40 31,62 -4.0504 1972 126 0.75 85.74 -6.50 29.98 0.4567 1973 129 0.77 91.31 -7.10 28.72 2.2598 1974 135 2.33 95.29 -0.90 27.99 -1.0903 1975 142 1.51 88.13 -1.96 27,06 1.8057 1976 139 -3,44 123.64 0.08 24.80 3.1512 1977 118 -2.35 171.54 -0.47 25.90 3.6669 1978 115 -2.46 128.04 1.95 24.87 2.8274 1979 113 -3.78 114.44 -3.49 28.93 1.7713 1980: 113 -1.43 116,33 -0.81 30,24 0.6210 1981 111 2.24 98.13 -4.40 31.19 -3.5669 1982 107 4.98 95,24 -5.80 31,77 -4.2781 1983 112 7.34 97.58 -5.70 33.59 -2.7808 1984 126 5.76 101.02 -3.90 35.63 -0.5792 1985 142 3.82 100.00 -0,60 34,34 1.2825 1986 159 -2.04 121.83 3.40 34.73 6.8532 1987 165 -0.19 94.04 2.40 32.86 4.2306 1988 165 0.92 94.67 1.30 38.82 1.7062 Soiurce: See appendix 5.2. Appendix 5.5. National Accounts, Colombia, 1950 Prices (millions of pesos, except as specified) : ~~Lsbor force Pliblic Private Piblic Private (thousands Year GOP investment itveslstonet capital stock capital stock of uwrkers) 1925 2,721.5343 95,9404 460.4900 664.5795 2,249.5740 2,505 1926 2,981.3793 139.3987 554.9801 737.5203 2,579.5967 2,551 1927 3,249,9272 164.8818 673.2416 828.6500 2,994,8786 2,596 1928 3,488.6365 180.4026 813.0576 926.1877 3,508,4483 2,645 1929 3,614.2075 141.5644 716.2658 975.1333 3,873.8693 2,693 1930 3,583.1256 102.6541 456.0948 980.2741 3,942.5772 2,743 1931 3,525.9348 90.3096 369.9048 972.5563 3,918.2243 2,799 1932 3,759.6710 93.5581 450.1208 968.8588 3,976.5226 2,857 1933 3,971.0282 111.3169 420.7698 983.2897 3,999.6401 2,916 N 1934 4,220.9269 76.6657 465.8540 961.6265 4,065.5301 2,976 1935 4,324.1189 96.3735 523.8146 961.8374 4,182.7916 3,038 1936 4,552.8820 96.3014 586.4852 961.9550 4,350.9976 3,102 1937 4,623.7488 118.7524 674.1610 984.5119 4,590.0588 3,165 1938 4,924.6220 139.1100 671.1919 1,025.1707 4,802.2448 3,232 1939 5,226.7384 163.6546 761.4111 1,086.3082 5,083.4314 3,287 1940 5,339.8766 217.8692 671,2603 1,195.5466 5,246.3485 3,343 1941 5,429.3926 183.5068 668.5273 1,259.4987 5,390.2410 3,401 1942 5,440.5821 219.9627 528.8999 1,353.5115 5,380.1169 3,460 1943 5,462,9611 212.0940 581.9786 1,430.2543 5,424,0838 3,520 1944 5,832.2145 140.3372 724.4484 1,427.5661 5,606.1239 3,583 1945 6,105.7355 127.7040 950.3800 1,412.5135 5,995.8915 3,647 1946 6,692.5625 141.2757 1,132.7182 1,412.5378 6,529.0205 3,697 1947 6,952.4075 212.9603 1,324,1788 1,484.2443 7,200,2973 3,750 (Tablc couliwiues oni thefollouringpage.) Appendx 5.5 (conlinued) Laborforce Piblkc Private Puiblic PriVate ftlwOzsands Year GOP inrvesfpliewlt ipivestient capital stock caplal stock of workers) 1948 7,150.0886 213.8987 1,289.6227 1,549.7186 7,769.8903 3,805 1949 7,774.2139 149.0000 998.6378 1,543.7468 7,991.5391 3,859 1950 7,860.0000 149.0000 1,176.0000 1,538.3721 8,368.3852 3,916 1951 8,114.2169 198.8941 1,066,0352 1,583.4290 8,597.5819 3,968 1952 8,620.1581 189.2430 1,143.4916 1,614.3290 8,881.3153 4,042 1953 9,150.8679 260.9257 1,219.7600 1,713.8219 9,212.9438 4,118 1954 9,775.4576 310.7912 1,548.7471 1,853.2309 9,840.3966 4,249 1955 10,185.3741 431.5895 1,548.4001 2,099.4973 10,404.7570 4,320 1956 10,560.8805 414.3772 1,601.6127 2,303.9248 10,965.8940 4,410 1957 10,820.0538 297.8513 1,588.0066 2,371.3836 11,457.3113 4,499 1958 11,075.0711 251.6743 1,301.3892 2,385.9196 11,612.9694 4,589 1959 11,877.5556 274.6543 1,348.1685 2,421.9819 11,799.8409 4,679 1960 12,388.3197 283.0688 1,641.6588 2,462.8524 12,261.5156 4,768 1961 13,020.6818 380.1256 1,713.8705 2,596.6928 12,749.2346 4,898 1962 13,728.5968 391.9846 1,538.9918 2,729.0081 13,013.3030 5,028 1963 14,175.4755 330.2556 1,533.6688 2,786.3629 13,245.6415 5,158 1964 15,036.8317 347.4230 1,777.3894 2,855.1496 13,698.4668 5,7 S 1965 15,596.8218 355.9759 1,688.8869 2,925.6105 14,017.5070 5, :J 1966 16,422,3692 484.3217 1,903.9400 3,117.3712 -14,519.6963 5,580 1967 17,022.738C 612.2499 1,551.9266 3,417.8840 14,619.6533 5,742 1968 18,159.1000 682.9678 1,917.2792 3,759.0634 15,074.9672 5,904 1969 19,317.9507 767.4616 1,857.9738 4,150.6187 15,425.4442 6,066 1970 20,866.4951 806.3811 2,032.5455 4,541.9380 15,915.4453 6,228 1971 22,157.9497 965.5015 1,977.2848 5,053.2457 16,301.1856 6,394 1972 23,833.0701 939.4597 1,998.4871 5,487.3808 16,669.5541 6,561 1973 25,441.7407 1,029.2148 2,248.9155 5,967.8575 17,251.5142 6,727 1974 26,884,8266 981.6909 3,038.0139 6,352,7627 18,564.3767 6,893 1975 27,502.2109 994.1626 2,280.0540 6,711,6490 18,987.9930 7,060 1976 28,792.0105 1,121.3029 2,455.6954 7,161.7870 19,544.8892 7,246 1977 29,994.9546 2,051.3382 2,154.6881 8,496.9465 19,745.0883 7,433 1978 32,543.4991 1,566.7592 2,667.7940 9,214.0111 20,438.3735 7,619 1979 34,269.5561 1,395.9691 2,958.6151 9,688.5791 21,353.1513 7,806 1980 35,688.3380 1,902.9520 2,854.4280 10,622.6732 22,072.2641 7,992 1981 36,490.7182 2,163.2286 3,030.1966 11,723.6345 22,895.2343 8,233 1982 36,840.0653 2,463.1206 2,907.8303 13,014.3916 23,513.5411 8,473 1983 37,431.4338 2,652.3695 2,710,8030 14,365.3220 23,872.9900 8,714 1984 38,685.8233 2,647.7364 2,594.6957 15,576.5262 24,080.3867 8,955 1985 39,882.4947 2,313.7590 2,516,0213 16,332.6325 24,188.3694 9,195 1986 41,918.5275 2,132.5938 2,749.9664 16,831.9631 24,519.4988 9,435 1987 44,160.7923 1,789.4313 3,431.2046 16,938.1981 25,498.7535 9,675 1988 46,112.2827 1,862.4891 3,729.0426 17,106.8674 26,677.9208 9,914 Soumrce See appendix 5.2. 270 Colombia: Avoiding Crises through Fiscal Policy Notes The author is grateful to Luis Jorge Garay for advice and assistance in the construction of a historical time-series on the public sector in Colombia; to Bela Balassa, Alberto Carrasquilla, Fernando Clavijo, Vittorio Corbo, John Cuddington, Albert Fishlow, E. C. Hwa, Paolo Leme, Johannes Linn, Car- men Reinhart, and Luis Valdivieso for comments and useful discussions; and to Piyabha Kongsamut for research assistance. He also thanks participants in tte World Bank Conference on the Macroeconomics of the Public Sector Deficit, June 1991. Any errors are the responsibility of the author. - 1. The real interest rate is defined as [(1 + 90 day deposit rate)/(1 + cn inflation rate) -1] x 100. 2- Coffee prices are the main explpnatory variable behind the evolution of the coffee balance, although there is some effect of real devaluation and autonomous policy changes, such as inventory investment and changes in the producer price ratio. 3. We also examined the effect on the deficit of the real exchange rate and inflation. The revaluation of the foreign currency items of govemment reve- nue net of foreigi interest payments, a result of real devaluiation, would have implied an improvement of about 0.4 percent of iDP in the fiscal balance. However, a regression of tax revenue on the real excInange rate finds a nega- tive effect of devaluation, although it is of questionable significance. Appar- ently, the strong contraction of import volume offsets the valuation effect of real devaluation on revenue. We also tested the effect of inflation on tax revenue and found a strong Olivera-Tanzi effect, but since inflation did not change much during the adjustment program, this was not a major factor. 4. This is similar to the consistency relationship between fiscal deficits and inflation proposed by Anand and Van WVijnbergen (1989), except that we also allow for portfolio shifts between money and interest-bearing assets. 5. This restriction is emphatically rejected by the data, which would imply an income elasticity of 2.4. Because such a large elasticity would lead to implausible simulation results, we impose the income elasticity of unity. The reasons for the explosive income elasticity will be investigated in further research. For an analysis of stability of money demand, see Carrasquilla and Renteria (1990). 6. The share of private capital in output would fall with an increase in public capital if the elasticity of substitution between the two were greater than 1 in absolute value. A fall in the share of private capital would lower the share of private investment in output for a given rate of return on investment. The rate of return to private capital would fail with increases in the types of public capital that were dose to being perfect substitutes for private capital. These results are discussed in chapter 1 in this volume. Note that public capital can have a negative effect or. private investment as a result of crowding-out through the financal market if the loan rate is not a true mea- sure of the cost of funds (because of financial repression and credit rationing, for example). 7. We did not find significant differences between permanent and transi- tory disposable income. For a careful treatment of this issue, see Cuddington and Urzua (1989) and Clavijo (1989). William Easterly 271 8. For an alternative treatment of labor productivity trends see Clavijo (1990). 9. We experimented with a nonlinear regression of a constant elasticity of substitution (as) function of the two types of capital (assuming that the elasticity of substitution between capital and labor was still 1), but the esti- mated elasticity of substitution was very close to 1, suggesting that the log- linear function is appropriate. 10. Actual inflation is used throughout as a measure for expected inflation. This implies static expectations: the current inflation is expected to continue. 11. The Laffer curve measures the relationship between tax revenue and the tax rate. It shows an inverted-U shape because the tax base shrinks as the tax rate rises. In this case, tax revenue is the printing of money (seigniorage), the tax rate is the inflation rate, and the tax base is the money stock. 12. To take account of the small negative effect of a decrease in demand deposits in the banking system, technically we must require that this effect be dominated by the quasi-money and investment effects. 13. The exogenous variables "net other assets" are adjusted to reproduce the actual equ6ilbrium in the money and credit markets. All other exogenous variables retain their actual or estimated values over 1987-89. 14. The reason for the round number for the change in the real interest rate is that the simulation is actually run the other way around-by specifying an increase in the real interest rate and then calculating the change in inflation and the deficit consistent with an unchanged level of monetary financing. This greatly simplifies the computation, and the economiic interpretation of the sinulation remains the same. 15. For this simulation we use the econometric equation that relates the ratio of private investment to GDP only to the real interest rate, since tis is the only specification with the "correct"" sign- 16. For an interesting altemative approach to the real exchange rate in Colombia, see Clavijo (21990). References Anand, Ritu, and Sweder van Wijnbergen. 1989. "Inflation and the Financ- ing of Government Expenditure: An Introductory Analysis with an Appli- cation to Turkey." World Bank EconomicReuiew 3 (1): 17-38. BernLal, Joaquin. 1991. "l Politica Fiscal en los Afios Ochenta." Ensayos sabre Politica Economica (Colombia) 19: 7-42. Bird, Richard. 1984. Lntergoaemnzental Finance in Cofomit. Harvard Law School, International Tax Program, Cambridge, Mass. Carrasquilla, Alberto, and Carmen Ren teria. 1990. "'Es Inestable la Demanda por Dinero en Colombia?" Ensayos sobre Politica Economica (Colombia) 17: 23-37. Clavijo, Sergio. 1989. "Ingreso Permanente yTransitorio: QueTanto Ahorran (o Consuman) los Colombianos?" Coyntura Economica 15: 71-93- . 1990. "Productividad Laboral, Multifactorial y la Tasa de Cambio Real en Colombia." Ensayos sobre Political Economica (Colombia) 17: 73-97. 272 Colombia: Avoiding Crises through Fiscal Policy Cuddington, John T, and C. M. Urzua. 1989. "Trends and Cycles in Col- ombia's Real GoP and Fiscal Deficit." Joumal of Development Economics 30: 325-43. Easterly, William, E. C. Hwa, P. Kongsamut, and J. Zizek. 1990. "Un Modelo sobre Requisitos Macroeconomics para Adelantar Reformas de Politica." Ensayos sobre Politica Economica (Colombia) 18: 99-132. Easterly, William, Carlos Alfredo Rodriguez, and Klaus Schmidt-Hebbel. 1989. "Research Proposali The Macroeconorics of the Public Sector Defi- cit." World Bank, Policy Research Departnent, Washington, D.C. Edwards, Sebastian. 1989. Real Exchange Rates, Devaluation, and Adjustment. Cambridge, Mass.: Massachusetts Institute of Technology Press. Engle, R. F., and C. W. J. Granger. 1987. "Co-lntegration and Error Correc- tiorn Representation, Estimation and Testing." Econometrica 55 (2): 251-76. Garcia Garcia, Jorge. 1989. "Macroeconomic Policies, Crisis and Growth in the Long Run." Colombia Country Study, Revised Version, Part m. World Bank Research Project, Washington, D.C. To be published as a volume in the Comparative Macroeconomic Studies series, World Bank. Herrera, S. 1989. "Determinantes de la Trayectoria del Tipo de Cambio Real en Colombia." Ensayos sabre Politic Economica (Colombia) 15 (Junio): 60-70. imP (International Monetary Fund). Various years. International Financial Sta- tistics. Washington, D.C. World Bank. Various years. World Debt Tables. Washington, D.C. 6 Cite d'Ivoire: Fiscal Polty wvith Fixed Nominal Exchange Rates Christophe Chamley and Hafez Ghanem Ten years of fiscal mismanagement in Cote d'Ivoire have led to a state of virtual bankruptcy.' In the niid-1970s a commodity boom on the two main exports-cocoa and coffee-led the fiscal authorties to embark on a spending binge. The fiscal deficit exceeded 12 percent in 1980 and 1981, averaged 9 percent of GDP during the 1980s, and was more than 14 percent in 1990. Numerous missions of the International Monetary Fund (IMF) and the World Bank emphasized repeatedly that the level of government spending was not sustainable and suggested reforms to address the issue. The failure of fiscal policy led to a crisis of unprecedented magnitude: in April 1987 the government stopped service of its foreign debt, and public disorders forced the beginning of an emergency program in which Allassan Ouattara, former head of the Banque Centrale des Rtats de i'Afrique de l'Ouest (BcEAo), had the role of a financial administrtor. This chapter descbes how the large deficits developed and how they eventually led to the current crsis. We focus orn the impacts of these deficits on the competitive- ness and growth of the economy. The coffee and cocoa price boom of the nrid-1970s allowed policy- makers to throw macroeconomic policy on the wrong track. The price booms were transitory and should have been perceived as such. Instead, they stimulated an unprecedented rise of public expenditure that continued after the end of the boom in 1979. Severe imbalances became evident in 1980, when the internal and exteral deficits exceeded 10 percent of aGO. The numerous attempts that were made thereafter failed to put the economy back on a growth path. With the end of the commodity boom, conditions deteriorated. Real GDP has been stagnant, with no difference between 1980 and 1987. Real GDP has decreased since 1983, while the popula- tion has kept growing at a rapid pace. 273 274 Cte d'lwoire: Fiscal Policy with Fixed Nominal Exchtange Rates * Employment in the formal nonagricultural private sector decreased from 200,000 in 1982 to 146,000 (equal to 1.4 percent of the population) in 1988. * Total investment decreased from 35 percent of CDP in 1977 to less than 15 percent since 1984. The government deficit increased from 12 to 15 percent, on an accrual basis. ^ Foreign lending to the private sector has ceased for the past eight years. Since 1982 all net foreign borrowing has been carried out by the government, which has now exhausted its borrowing capability. * The foreign debt now exceeds 100 percent of cDP. No one expects this debt to be paid back, and its value on the secondary market is between 5 and 10 percent * The financial system is now stalled, with many insolvent debtors. In this chapter we investigate whether there is a relationship between the dismal performance of the economy and the large gov- errnent deficits. Such linkages are difficult to establish even with a large sample of countries. Since this study is limited to one country, the analysis has to rely on a variety of data. External Shocks In this section we review the impact of the celebrated commodity boom of 1976-77 and its aftermath. A quantitative evaluation of the external shocks that have affected the primary exports shows that the position of the country is now back to its initial state before the boom- The vulnerability of public revenues and of the whole economy to international fluctuations has increased. Commodity Prices The economy of C6te d'Ivoire has been heavily dependent on its two main exports, cocoa and coffee. This dependency has increased over the past twenty years. These commodities show three remarkable properties: 1. Prices are highly variable (figure 6.1). The significant boom of the miid-1970s now appears as an anomaly in the time-series. 2. The comovements of the prices are remarkable, since external factors that affect the supplies of the two commodities seem to be independent. For instance, the sudden rise of coffee prices in 1976 was caused by a frost in Brazil, which had no influence on the market for cocoa, yet cocoa prices also increased. 3. There is no apparent trend in the prices of cocoa and coffee. Figure 6.1 shows no evidence of a random walk. This is con- Chtistophe Chamky and Hafz Ghamem 275 Figure 6.1. External Price Shocks, Coffee and Cocoa Prices, Cote dIvoire, 1970-87 Ratio 1.2 1.6 01S * 0.2 --' .. 1970 1972 1974 1976 1978 1980 1982 1984 19861987 ------- Coffee pricelimpowt prices Cocoa pricelimport prices Sourcc World Bank data. firmed in a more detailed analysis of the properties of the time- series of the two basic commodity prices by Melhado (1990). There is therefore no statistical support for the view that the price shocks of the mid-1970s could affect the expected prices of the commodities in the long run.2 It is interesting to recall some of the perceptions during the first commodity boom. A distinction was made between the two commod- ities: the coffee boom was understood to be the result of the frost in Brazil and was perceived as temporary, but the increase in the price of cocoa was apparently perceived as permanent. A reason for this myo- pia may be the domiunant position of Cate d'Ivoire in the cocoa mar- ket In 1979 authorities had experienced three years of largely above- trend prices (see figure 6.1), and they were slow to recognize that the boom had indeed been temporary. In the mid-1980s a temporary price recovery again raised the level of expectations about the permanent level of cocoa prices. In the subsequent slump an attempt to manipulate the world price through stock retention failed. The current slump to historical lows does not 276 Cdte d'lvoire: Fisad Policy with Fixed Nominal Exchange Rates so far appear to be out of line with statistical fluctuations. However, the dominant position of Cote d'2voire has been eroded by higher cocoa production in other countries (Malaysia, for example), and the downward price trend is likely to continue. Measurement of External Shocks The continuing fiscal and economic crisis of the 1980s, which has deepened recently, has been blamed on a combination of external shocks and poorly designed and implemented fiscal policies. To dis- entangle these causes, we have computed some measurements of the external shocks that have affected the markets for primary commuodi- ties and the service of the foreign debt. Our purpose is to measure how variations in the prices and quantities of three primary commod- ities (coffee, cocoa, and wood) and the flow of the debt service affected the national income of the country. We ignore induced effects and follow an accounting approach. For the economy of CBte d'lvoire, the definition of national income Y can be rewritten as: 4 Y=E PZXk + H, - PQ,Q,-Z, where P,t represents the prices and X the quantities of exports of cocoa (i = 1), coffee (i = 2), wood (i = 3), and other commodities (i = 4). As a first approxation, the production of these goods, which are entirely exported, is assumed to require only domestic inputs. The value added in other sectors of the economy is H,. The term PQ,Qt represents the value of imports, and Zt is the interest payment on foreign borrowings. We focus on the shocks that affected the export sector and the service of the foreign debt. Suppose that all changes in exports are attnbutable to external causes and that the same property applies to the service of the foreign debt. WMe denote by a superscript 0 a value in the base year (1975, the last year before e -e commnodity boom)- The income effect of the exog- enous shocks can be decomposed into changes inr P, X, and Z by the accounting identity: 4 -~~ . DY, - YO-° (P,X - P°zn - (Z t- ) i=1 which can be rewritten as: 4 4 DYt = Z (Pit - P)Xt + E (X1t - ,a ,- (Z, -Z i=l i=l Chdlsfophe Chamky and Hafcz Ghanem 277 The income effects are thus the sum of the price effect, the quantity effect, and the effect of the debt service. Note that this expression is not a linear approximation and that the price effect in year t is mea- sured in combination with the actual level of output Xt. The output of cocoa tripled from 1975 to 1985, and a percentage point increase in prices therefore has a bigger effect in 1985 than in 1975. To normaize the changes, we divide all terms by the level of output in the base year 4 D:)Y/? = Z[(Pi, -PP)/P;J(P.A'RD 4 + > [(Xv - nxWtlut?(4Ptt - (Zt - Z7%'. The values of the components of the shocks for quantities, pnces, and the debt service are presented in figure 6.2. The decomposition of prices is shown in figure 6.3, and quantity effects are presented in figure 6.4. The following Temarks can be made about the contribution of the pnnimary sector to economic growth in C8te d'Ivoire. * As we have already seen, there is no evidence of an upward trend of the prices of the three commodities. During the second boom (1984-86) prices increased less dramatically than in the first. However, the impact of the second boom was greater because output had increased since the 1970s. * There is apparently no evidence of a contnrbution of the primary exporting sector to growth in the Ivorian economy over the past fifteen years. This is very striking because the principal policy- makers have always emphasized the role of primary commodi- ties in the development of national wealth. The increase of real output in the cocoa sector was partially offset by declines in the production of coffee and wood. Between 1975 and 1987 these three primary commodities contributed to a growth of income of less than 2 percent.3 * The primary commodity sector became more sensitive to price fluctuations as the share of wood, which has a'ess volatile price, was reduced in favor of cocoa. The evolution of the activities producing primary commodities thus made the economy more vulnerable to external shocks. * Debt service increased to 12 percent of 1975 GDP in 1989, puffing a heavy burden on future income. The structural evolution of the economy inaeased its vulnerabilit' to external shocks and the fragility of the public sector's capacity to 278 Cate d'Ivoire: Fiscad Policy with Fired Nominal Exchange Rates generate revrenues. This fragility contributed to the severity of the current crsis, which was induced by the combination of another neg- ative shock and the large indebtedness of the public sector. Public Revenues and Expenditures Fiscal policy in C6te d'Ivoire has been characterized by wide fluctua- tions of revenues, expenditures, and deficits and is constrained by the rules of the crA zone. We briefly review some of the relevant constraints and then present a brief desciption of the sources of public revenues and a definition of structural revenues. The level of structural revenues provides the basis for the computation of the structural surplus (or deficit). This is a useful indicator for the evalua- tion of the composition and the drift of expenditures that led to the present crisis. Figure 6.2. Contribution of External Shocks to GDP, C6te d'lvoire, 1970-87 Percentage difference with respect to 1975 50 40 - 30 - 20 - 10 0 -10 -20 -30 1970 1972 1974 1976 1978 1980 1982 1984 1596 19E * . Price effect - Quantity effect Total Total growth wDc) * Interest Source: VWorld Bank data. Czdstophe Chwntey and Hafez Ghanem 279 Figure 6.3. Contribution of Price Shocks to GDP, Cote d'Ivoire, 197047 Percentage difference with respect to 1975 12 10 6 4 2 0 -2 -4 -6 1970 1972 1974 1976 1978 1980 1982 1984 1986 1987 D=P (co ffee)/Pdef -s DP (cocoa)/Pdef - DP(wood)lPdef Total Sourc World Bank data. Constraints on Fiscal Policy The institLtional rules of the CFA zone put severe restrictions on domestic borrowing by the govemment: domestic govenment debt cannot exceed 20 percent of the previous year's tax revenues. The prinary purpose of this institutional rule is to prevent the govern- ment from financing expenditures by issuing a currency that is backed by the monetary union (and the French treasury). The use of permanent money growth for financing is tfhus not feasible in Cote d'lvoire. The policy constraint improves the price stability of the economy. In this environment of relative stability of pnrces and inter- est rates (compared with other Sub-Saharan African countries), the quantity of money is also stable. Attempts to estimate a demand for money proved fruitless.4 The restiction of the cfrA monetary union does not eliminate the possibility of a temporarily significant level of domestic borrowing, and this opportunity was used immediately after the first commodity 280 COe d'lwire: Fisal Policy with Fixed Nominal Exchange Rates Figure 6.4. Contribution of Quantity Shocks to GDP, Cbte d'lvoire, 1970-87 Percentage difference with respect to 1975 7 6 5 4 3 2 -1 -2 -3 I E . . . . . . . . . . . 1970 1972 1974. 1976 1978 1980 1982 1984 19861987 DQ (coffee)/Q def - DQ (cocoa)/Q def DQ (wood)/Q def -Total - Source: World Bank data. boom in 1978-79. Once the ceiling is reached, any additional domestic borrowing is feasible only if the tax base and the formal sector are growing. There has been little or no such growth since 1980. There is no institutional restriction on the level of foreign borrow- ing by the government. In the case of C6te d'Ivoire, the only limit on government borrowing is imposed by the government's creditworth- iness. All public financial resources, therefore, depend on domestic taxes and foreign borrowing. Composition of Public Revenues As in other Sub-Saharan African countries (see Tanzi 1987), the for- mal nonagricultural domestic sector represents a small share of the economy. Hence the domestic base for taxation is narrow, and most taxes are levied at the border or on the rural sector, which .sels pri- Chnistophe Chamley and Hafez Ghanem 281 mary commodities (cocoa and coffee) to a marketing board. Domestic taxes may be implicit or explicit. Thnucrr TAxEs. All the domestic production of cocoa and coffee is sold to a marketing board (Caisse de Stabilisation), which -handles exports. The gap between the border and producer prices, net of processing costs, generates implicit taxes. Producer prices varied little from 1970 until recently (in terms of the domestic price level); the marketing board, an arm of the government, has absorbed the wide fluctuations of the border prices. Such a policy of price smoothing seems reasonable because the public sector in C6te d'Ivoire has better access to foreign credit mar- kets, which can provide insurance to the economy. However, the main purpose of the marketing board has been to generate revenues for the public sector. The wide fluctuations of intemational prices have been translated into similar fluctuations of revenues of the board. Public revenues from cocoa and coffee, plotted as a fraction of cDP In figure 6.5, match closely the variations in commodity prices illustrated in figure 6.1. Bxpucrr TAXES. The government, pressed by a fiscal crsis since 1980, has attempted to increase the burden of explicit taxes. This burden has almost reached the maximum feasible level for the econ- omy. Repeated changes in statutory rates on tariffs, the value added tax (vAT), and direct taxes on personal and business income have been introduced since 1980 and have failed to augment revenues noticea- bly. The levels of the statutory rates and the structure of the economy are such that further increases in rates lead only to higher evasion, shifts of activities toward the informal domestic sector, and smuggling. In 1987 tariffs were increased by 30 percent across the board, and the standard VAT rate of 25 percent was extended to commerce. This reform led at first to higher revenues, both absolutely and as a share of GDP. For example, revenues from import duties rose from cFAF 233 billion in 1986 to cFAF 245 billion in 1987, despite a decline in the ca (cost, insurance, and freight) value of imports. The ratio of tariff revenues to imports rose from 33 to 36 percent. Similarly, revenues from domestic indirect taxes rose from CFAF 195 billion (6 percent of cDP) to CFAF 224 bilion (7 percent of CDP). As a result, total tax revenues increased from 18 to 20 percent of GDP. These increases proved short lived, however. The tax base shrank because of two effects. First, the general level of economic activity decreased: GDP in current prices fell 3.9 percent in 1987, 1.6 percent in 1988, and 3.3 percent in 1989. Thus, all categories of revenues fell in absolute termtls. 282 COte d'lvoire: Fiscal Policy nith Fixed Nominal Exchange Rates Figure 6.5. Sources of Public Revenue, C6te d'Ivoire, 197049 Percentage of GDP 40 35- 25- 20 - - \ - 15A 10 5 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1989 Total revenues Tax revenues (exduding cocoa and coffee) -- Tax revenues (cocoa and coffee) SourTce: World Bank data. Second, fraud increased and ad hoc tax and tariff exemptions were widened as economic agents tried to avoid the higher tax rates. In particular, total revenues from import tariffs decreased as a propor- tion of total revenues from 1986 to 1989, although nominal rates increased. Overall tax revenues as a proportion of GDP fell from their 1987 peak, so that by 1989 the tax-to-nDP ratio stood at the same level as in 1986 (table 6.1). The small size of the tax base in the Ivorian economy may come as a surprise to economists who are used to industrial economies. The base has shrunk still more under the tax burden- (Tis issue is exam- ined more thoroughly in appendix 6.1.) Although the tax base in the formal domestic sector is ridiculously small when compared with that of an industrial country, the level of taxes is commensurate with those observed in some industrial economies. As a consequence, the aver- age effective rate of taxation in the formal nonagrcultural sector is close to 50 percent. The ratio between public revenues (other than for cocoa and coffee) and measured aGP has not fluctuated significantly since 1975 (see Christophe Chainley and Hafez Ghanern 283 Table 6.1. Evolution of Tax Revenue, Cote d'Ivoire, 1986-90 (bilions of CFA francs) item 1986 1987 1588 1989 Total tax revenue 578 621 570 543 Import duties 233 245 214 213 Direct taxes 150 152 150 142 Indirect domestic taxes 195 224 206 188 C;P- 3,244 3,118 3,068 2,967 imports (aF) 709 672 620 658 Import duties as a share of value of irnports (percent) 33 36 35 32 Taxes as a share of GDP (percent) 18 20 19 18 Source: World Bank data. figure 6.5). This share is slowly but definitely declining because of the gradual shift toward informal activities. STRUCI-uRAL LEVEL OF PUBUC REVENUEs. A key issue in the evalua- tion of fiscal stance is the definition of a long-mn level of revenues. In an industrial country the long-run tax burden is determined by some social agreement about the fraction of the national permanent income that should be used to finance public expenditures. The value of this ratio is typically less than the maximum feasible level (at the top of the Laffer curve). It depends on the relative magnitudes of the benefits of public expenditures and the effciency cost of taxation. In Cote d'Ivoire the situation is simplified because policy reform studies and experiments during the 1980s have shown that the level of explicit taxes on the nonagricultural sector, as a percentage of GDP, cannot be raised. We therefore assume that the ratio between explicit taxes and GDp is the maximum feasible, and this level will define the structural val-ue of explicit tax revenues. The main source of fluctuations in revenue is therefore to be found in implicit taxes, which are raised almost entirely by the Caisse de Stabilisation. Consider cocoa, for instance. Revenues from its impliit tax fluctuate with the international price and with the supply, which may be affected by droughts, as in the early 1980s. In a definition of the structural level of the cocoa tax, we have to take into accomunt both effects. We define the deviation from the structural level by the formula: Rt =(pi - c)X, - (pF - ct)X where Xt is the actual real output of the crop, XIis the trend value, pt is the border price, p? is its trend value, and c, is the producer price, 284 Cote d'lvoire: Fiscal Policy with Fixed Nominial Exhange Rates including processing costs. The value of pp is taken to be equal to that in the reference year 1975, adjusted for inflation: p? = p7p where P, is the CDP deflator. The deviation of revenues from the structural level can be expressed as fractions of GDP, and we replace the preceding formula with [PtX, (p7u/Pm)X?] - cX1 - X) Rt = GDP The trend values of the quantities are found by a simple regression. We have made the same computation for coffee and wood. The Structural Surplus We use four definitions of the actual surplus:5 1. The overall surplus is the difference between total revenues and total expenditures. 2. The current surplus is the difference between total revenues and current expenditures. 3. The primary surplus is the difference between the overall sur- plus and interest payments. 4. The primary current surplus is the difference between the cur- rent surplus and interest payments. These definitions provide different instruments for evaluating the surplus, or the deficit, from a long-term perspective. For example, the current deficit is a reasonable indicator if it finances capital expendi- tures, much as when a private firm invests in a period of growth. The overall deficit is a useful indicator if public investments, unlike pri- vate investments, do not generate, at least directly, an earning capac- ity that is used to repay the debt. The values of the surplus that are obtained from the preceding definitions are represented in figure 6.6. These values are subject to large fluctuations because the structure of revenue is highly sensitive to extemal shocks. There is no reason for the government budget to be balanced in a given year, just as there is no reason for it to be balanced on any given day or in any week. To iron out the exogenous fluctuations and evaluate the fiscal stance for the medium to long run, we use the measurements of the structural surplus. The structural surplus is quite naturally defined as the difference between structural revenues and structural expenditures. We have already defined structural revenues. Expenditures in Cote d'Ivoire CItristoplae Chanmley and Hafez Ghanem 285 Figure 6.6. Budgetary Surplus, COte d'Ivoire, 1970-89 Percentage of GDP 30 25 20- 15 10 05le -10 -15 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1989 Primary current surplus * …----c Current surplus - - -- C- Primary surplus Overall surplus Source: World Bank data. are not linked to cycles as some public programs (for example, unem- ployment insurance) would be in industrial countries. Hence we make no correction on the expenditure side for the fluctuations of the economy. All expenditures are assumed to be discretionary and structural. Each definition of the actual surplus corresponds to a similar defini- tion of the structural surplus. The values of the levels of the structural surplus according to the various definitions are presented in figure 6.7, which highlights very nicely the main features of fiscal policy in Cote d'lvoire over the past twenty years. The structural definitions of the surplus provide a clearer view than do the standard definitions, especially for the inportant period of structural adjustment. Three distinct regimes, separated by relatively short transition periods of about two years each, can be observed. 1. Before 1975, the structural budget was in equilibrium. 2. In the first phase (197642) the structural deficit jumped from 286 COte d 'lvoire: Fiscal Policy with Fixd Nominal Exchange Rates Figure 6.7. Structural Budgetary Surplus, Cote d'lvoire, 1972-87 Percentage of cOP 25 - 20- 15 *--->---~ ~~~~~~- Prmr surpl-s 10 u: World Bank data. . - -a 1972 1974 1976 1978 1980 1982. 1984 1986 1987 Primarycurrent surplus - - - -* -Current surplus - .Primary surplus Overall surplus Source: World Bank data. about 0oaf COvP to 10 percent. It stayed very dlose to that value for six years, until 1982, and then declined until 1985. 3. Since 1985 the budget has been more or less in structural equibrium. The values of the surplus are computed here on an accrual basis. However, the Ivorian government has been accumulating arrears since 1987 at an annual rate of about 3 percent of GCm. More impor- tant, the service of the foreign debt (about 8 percent of CDP) has been suspended since the spring of 1987. The budgetary deficit on a cash basis is therefore lower than its accrual value. The primary surplus is equal to the surplus when the interest service of the debt is not paid. Since the nonpayment of foreign interest is at least partially structural (nobody expects the debt to be serviced entirely), the actual value of the structural surplus is probably between those of the primary sur- plus and the surplus and is therefore positive. Nevertheless, the pres-. ent crisis has forced the government further to reduce expenditures and to raise taxes on the agricultural sector. Christophe Chiamtey and Hafez Ghanem 287 To summarize, despite the most recent adjustments, the govern- ment is still running a significant deficit (see figure 6.6). This deficit is explained by adverse extemal conditions, which are worse than ever for the country. Our analysis, which uses the definition of the struc- tural surplus, allows us to isolate, at least parially, the effect of these external shocks. It shows that the contractionary stance of current fiscal policy is unprecedented in C8te d'Ivoire. Composition of Pubtic Expenditures The large structural deficits during the period 1977-84 were driven by expenditures. The composition of these expenditures (figure 6.8) is one of the main issues in the discussion of the rise and fall of the deficit. Three main features of expenditures since 1975 can be observed. 1. Current expenditures on noninterest payments increased from an average 16 percent of CDP before the boom to about 21 per- cent during the 1980s (see figure 6.9). For the ten years of macro- economic adjustment that began in 1980, one cannot detect any declning trend in the ratio of current nondebt expenditures to G3DP. Attempts at structural adjustment proved to be a complete failure in this respect. 2. The main cause of the significant deficts was the high level of Figure 6.8. Public Expenditure, Cbte d'Ivoire, 1970-89 *Percentage of GDP 40- 35 30 -/, 25- 20- 1.0 1970 1972 1974 1976 1978 1980 1982 '1984 1986 1988 1989 To tat expenditure ----Current expenditure Capital expenditure Source. World Bank data. 288 COle d'Ivoire: Fial Policy with Fixed Nominal Exchange Rates Figure 6.9. Current Public Expenditure, C6te d'lvoire, 197089 Percentage of GDP 35 - 30- 25. 20 10 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1989 Current expenditure - Current nondebt expenditure Government wages Source:AWorld Bankdata. capital expenditures, which increased from 8 to 22 percent of GDP in the three years after 1975. Much of this expenditure was wasted on inefficient projects. (The composition of public investment is discussed in appendix 6.2.) The high levels were clearly unsustainable, and the ratio of capital expenditures to GDP has declined steadily since 1979. It is remarkable that all the adjustment toward a balanced budget has fallen on capital expenditures, whereas cunrent nondebt expenditures (and the wage bill of the public sector) have continued to rise. The value of capital expenditures after the adjustment (less than 5 percent) is probably not sufficient to cover the depredation of the public capital stock. 3. The service of the (mainly foreign) debt has increased signifi- cantly, from 3 percent of GDP in 1980 to 9 percent in 1990, because of two factors: the growth of the stock of the debt and the rise in world interest rates in the 1980s. Impacts of Fiscal Policy The expansionary fiscal policy that followed the commodity boom of 1976 raises the standard issues of the "Dutch disease" and the impact Christophe Chamley and Haflz Ghanem 289 of the boom on the external deficit, saving, and investment. This section begins with a brief discussion of some theoretical issues and goes on to analyze the impact on the aggregate level of prices and on the relative prices between the main sectors of production. The Price Level and the Real Exchange Rate The doubling of the price level in Cote d'lvoire between 1975 and 1980 has to be seen in the context of worldwide inflation. (For exam- ple, France's GDP deflator increased by 62 percent during the same period.) Ncvertheless, a positive inflation differential appeared. This differential has been attributed by some observers to the high money growth rate and by others to the high level of government expenditure. THE moNElST INTERPRETAION. The money supply in COte d'Ivoire doubled between the end of 1975 and the end of 1977. This high growth has lent some credence to a monetarist explanation of the post-1975 inflation rate. Such arguments may have a firm arith- metic base; but what is the economic reasoning? The link between money expansion and prices is based on an "excess supply" of money that generates a price increase when the market for goods dears. But the policy regime is one of a fixed nomrrinal exchange rate between COte d'Ivoire and France, with free capital mobflitv-at least from Cote d'lvoire to France (and other capital markets).6 In this situation there can hardly be an excess supply of money; the relevant quantity of money for the country is the quantity in the whole cur- rency zone. There may be a situation of "excess demand" when the country is rationed for foreign liquidities. But this situation is not relevant either for the theoretical argument or for the empirical facts of the period 1975-80. If this hiterpretation is correct, the quantity of money (M2; in Cote d'Ivoire is driven by the supply from the banking system. Advances by the Central Bank have a strong impact on this supply. The Central Bank and the banicng system act, effectively, as intermediaries for some of the savings that are deposited in foreign banks. In this case variations in the quantity of money have, in them- selves, little impact on the price level. RELAIVE PRICES IN THE BOOM. During the first commodity boom and the period that immediately followed, the level and structure cf. pnces were significantly altered. Government expenditure expanded suddenly. Some of these expenditures were directed to imports, the prices of which are not sensitive to demand by Cote d'Ivoire. Other 290 COte d'Iwire: Fiscal Policy with Fixed Nominal Exchange Rates Figure 6.10. Ratio of Sectoral Prices to the GDP Deflator, Cite d'Ivoire, 1971-84 Ratio 130 - 120- 110 100 -. ....... . .. . -....... ........ 90 . 80 ¾ - ~ - 70 60- . i 1971 1972 1974 1976 1978 1980 1982 1984 Services Construction Manufaduring Average Source. World Bank data. expenditures increased the demand for nontraded goods, the supply of which cannot be adjusted easily. Hence prices increased more for services than for manufacturing. The ratios between the prices in some sectors and the GDP deflator are presented ir- figure 6.10. The curves for prices of services and prices in otha sectors of the econ- omy form a pincer: during the first two years of the boom (1976-77) services led the price increases, while industrial prices lagged behind the average increase (represented on the figure by the horizontal line). In the second phase of expansionary fiscal policy (1978-81) the price increases were carried to the other sectors of the economy. The behavior of relative prices thus fits an interpretation that is based on the rise of public sector demand. THE IMPACr OF FISCAL POLICY ON THE REAL EXCHANGE RAE. The rise of domestic prices during the first boom adversely affected the com- petitiveness of the economy. All the available indexes follow similar paths. The ratios between price indexes in C6te d'Ivoire and France are presented in figure 6.11. During the expansionary period, prices increased much faster in C6te d'lvoire, but in the period of recession Christophe Chainley and Hafez Ghanemn 291 Figure 6.11. Competitivity Indexes, Cote d'Ivoire, 1971-83 Ratio 140- 130- 120- 90 80 70 60 1971 1972 1974 1976 1978 1980 1982 1983 - Industrial products/Mnr Real exchange ratea Manufactures/Mm- Note Mn irnport price index. a. Purchasing power parity Sourcer World Bank dab. after 1980, they increased more slowly than in France.7 Note that the inflation Tate was si positive during the recession, with an average value of 7 percent. The prices in the construction and manufacturing sectors showed more variation than the average price level in the economy. At the end of the cyde of boom and bust (about 1983-84), the relative pnrces in the industrial sectors were at about the same level in relation to those in France as at the beginning of the cyde. However, the general price level in Cote d'Ivoire, as measured by the consumer price index (at), was still 20 percent higher than in France. This difference was mainly attributable to the rise in the price of services (see figure 6.10). The simultaneity of a large increase in public expenditure and a rise in the domestic price level in the late 1970s raises the important issue of the relation between public expenditures and the domestic price level or the real exchange rate. We have estimated for Cote d'Ivoire the same econometric relation that has been analyzed in other studies done under this project (see, in particular, Rodrfguez 1990 and chap- ter 2 in this volume). The basic idea is that the domestic price level 292 Cated'lvoire: Fiscal Policy with Fixed Nominal Exchange Rates (with respect to some foreign prices), adjusts to clear the market for domestic goods. An exogenous increase of demand-for example, from public expenditure-increases the demand for nontradable goods and thus raises their prices with respect to those of tradable goods. Since the domestic price level is an average of the prices of nontradable and tradable goods and the price of tradable goods is determined by foreign prices, one should observe a positive relation between exogenous demand shocks (for example, public expendi- ture) and the ratio between domestic and foreign prices (of competing goods). This relation is embodied in the equation: (6.1) lOg(RER) = ao + aD + b log(Px/Pm) where the price indexes for exports and imports are Px and PM' respectively. The real exchange rate is defined as the ratio between the domestic price level and the price of tradable goods for which the index is the price of imports. Exogenous demand shocks are repre- sented by the variable D. These shocks may arise in the private or the public sector. To examine whether public sector shocks have a greater eiffect on the domestic price level than those of the private sector, we introduce an additional variable in the preceding equation that mea- sures the differenfial effect of the public sector on prices: (6.2) log(RER) = ao + aD + b log(Px/PM) + gG where G is a variable that measures public expenditure. Exogenous shocks to total demand, D, are measured by the ratio of domestic expenditures (absorption) to income. This ratio is equal to 1 minus the ratio between the trade surplus and cDP, by the identity of national income accounts. One can therefore introduce the ratio between the trade balance and GDP in equation 6.2. The variable G is taken to be the ratio of total expenditures of the public sector to CDP. In this c.se the equation tDbe estimated is: (6-3) log(RER) = ao + a[(X - M)IY] + b log(Px/PM) + g(G/Y) (using variables mBY, LPxM, and CY) where X is exports, M is imports, G is public expenditure, and Yis GDP. This equation has been estimated in levels and first differences, and results are presented in tables 6.2 and 63.. The equation in levels has a coefficient of autocorrelation that is almost equal to 1. We will focus on the equations in first differences. A difficult issue is how to measure the exogenous shock to overall aggregate demand. We have used the balance of trade-that is, the difference between exports and imports of goods. The level of exports Christophe Chamley and Hafez Ghanem 293 Table 6.2. Regression Results for the Real Exdhange Rate, Levels, C8te d'Ivoire, 1972-89 Standard TiOt-afi ariable Coefficient error t-statistic significance DRER(1) 0.240 0.118 2.04 0.060 UTBY -1.012 0.233 -4.33 0.001 DLPXM 0.253 0.061 4.14 0.001 DGY 0.295 0.221 1.33 0.204 R2 0.797 Adjusted R1 0.753 S.E. of regression 0.035 Durbin-Watson statistic 1.498 Mean of dependent variable -0.002 S.D. of dependent variable 0.0705 Sum of squared residuals 0.0171 F-statistic 8.367 Note: Number of observations, 18; method, ordinary least squares; dependent vari- able, DRER. S.E., standard error, S.D., standard deviation. DRER,real exchange rate; TVY, surplus of the trade balance; LpXM, terms of trade; GY, ratio of government expenditure to CDO. Table 6.3. Regression Results for the Real Exchange Rate, First Differences, CMte d'Ivoire, 1972-87 Standard Two-fail VariabE Coefficient error f-statistic significance VRER(-I) 0.21 0.120 1.67 0.120 serR -1-.06 0.245 -4.32 0.001 DLUXM 0.27 0.065 4.13 0.001 DGY 0.24 0.245 0.99 0.341 ,R2 - 0.80 Adjusted R2 0.75 S.E. of regression 0.036 Durbin-Watson statistic 1.58 Mean of dependent variable 0.0050 S.D. of dependent variable 0.0736 -Sum of squared residuals 0.0156 F-statistic 16.823 Note: Method, ordinary least squares; dependent variable, oRER; S.E., standard error; S.D., standard deviation. is obviously exogenous, while the level of imports is not; the level of imports probably depends on income and on the ratio of domestic prices to import prices, which is the dependent variable itself. For this reason it is recommended that this variable be omitted from the regressors and that instrumental variables be used. The results are 294 C8te d'Iwire: Fiscl Policy with Fixed Nominal Exchange Rates Table 6.4. Regression Results for the Real Exchange Rate Using Instrumental Variables, Levels, C6te d'Ivoire, 1972-89 Standard 7wo-tail Variable Coeffcient error t-statistic significance DThY -1.29 0.317 -4.09 0.001 DLPXM 0.26 0.066 3.92 0.002 DGY 0.08 0.234 0.34 0.733 AR(1) 0.33 0.214 1.54 0.144 R2 0.76 Adjusted RZ 0.71 S.E. of regression 0.03 Durbin-Watson statistic 1.68 Mean of dependent variable -0.0027 S.D. of dependent variable 0.0705 Sum of squared residuals 0.0200 F-statistic 15.07 Note: Method, two-stage least squares; dependent variable, DRER; instruments, C, DLPXM, DGY, DnMIr, Dxr, oM2Y. S.E., standard error S.D., standard deviation. Table 6.5. Regression Results for the Real Exchange Rate Using Instrumental Variables, First Differences, 1972-87 Standard Two-tail Variable Coeffident error t-stastic significance LPXM 0.83 0.23 3.64 0.004 CY 5.13 1.92 2.67 0.022 GY 0.63 0.19 3.31 0.007 C -2.30 1.12 -2.05 0.064 AR(1) -0.64 0.32 -2.00 0.071 0.536 Adjusted R2 0.367 S.E. of regression 0.072 Durbin-Watson statistic 1.756 Mean of dependent variable 0.835 S.D. of dependent variable 0.091 Sum of squared residuals 0.058 F-statistic 3.17 Note: Method, two-stage least squares; dependent variable, RER; instruments, C, LPXM, LPXM(-1), CY, Gn-I), DMPI, DXPI, DM2Y. S.E., standard error; S.D., standwd deviation. reported in tables 6.4 and 6.5. The estimated values of the coefficients are not very different from those of ordinary least squares, and the variations of the estimated values are in the direction that can be expected from the simultaneity bias. Christophe Chamley and Ha fez Ghanem 295 Tables 6.2-6.5 show some relation between demand shocks and the real exchange rate. However, prudence is called for in interpreting the results because the surplus of the trade balance (TmY) is used as a proxy for demand shocks. Of course, it is more plausible to assume that government expenditures constitute exogenous demand shocks. Unfortunately, this variable does not appear to be significant in the preceding estimations. Although we have attempted to correct for the endogeneity through the use of instruments, it may be worthwhile to test other specifications. Another proxy for exogenous changes in demand is obtained by two separate components of total demand-private consumption and public expenditures, as measured by their ratios to GDP (CY and GY, respectively). When these proxies are included in the equation, the estimation yields the results presented in table 6.5. The equation performs less well for curve fitting, but the coefficient of government expenditures is now significant. Overall, we conclude from these results that there is empirical evidence for a positive impact of gov- ernment expenditures on the real exchange rate in the economy of COte d'lvoire. The Balance of Payments The deficit of the balance of payments is the sum of the external defcits of the private and the public sectors. We have seen that the capacity for domestic borrowing by the public sector-albeit not insignificant-is severely limited by the regime of the CFA zone. When this capacity is exhausted, the public sector has to turn abroad for its financing. In Cote d'Ivoire the capability of the private sector to run an exter- nal deficit depends on the willingness of foreign lenders to lend. Before and during the first commodity -boom, foreign loans to.the. private sector were indeed significant. However, after the completion of the first cycle (1975-82), the capability of the private sector to bor- row abroad seem to have been exhausted.8 These conditions explain the behavior of the balance of payments and its relation to the internal deficit since 1970. The raw data are presented in figure 6.12. Because the yearly reporting of the data introduces some noise, we have also given, in figure 6.13, a represen- tation of the data after smoothing through a three-year moving aver- age (with weights of 0.25, 0.50, and 0.25). There were four distinct regimes. 1. Before the first commodity boom the internal deficit was small (less than 3 percent of GDP). The deficit of the balance of pay-. ments was fairly large (about 8 percent of GDP) and reflected the 296 COte d'lvoire: Fiscal Policy with Fixed Nomiinal Exchange Rates Figure 6.12. Balance of Payments, C6te d'lvoire, 1970-88 Percentage of wODI -15 -20 1970 1972 1974 1976 1978 1980 1982' 1984 1986 1988 -Transfers C;ovenment surplus 1alance of payment Soursce World Bank data. inftlow of funds for investmnent in Moe-d'Ivoire. There was a pparently little relation between the internal and external defi- cits. The external deficit. was mainly attributable to activities in te private sector. 2. During the expansionary phase of the first boomn (1975-77) the external deficit stayed at the same level of abou t .7 percent of GDP, since the increased exports were matched by spending on f-oreign goods. 3. During the phase that immediately folIowed the b oom (1978-81) the structural deficit of the governunent did not adjust (see figure 6.7). The external deficit iIlcreased dramatically because the pri- vate and public sectors contirtued to borrow as economic activity continued to increase. lDuring this period the public sector used - up all its capability for domestic borrowing.- 4. Since 1982 the sources of foreign borrowing have dried up com- pletely for the private sector. Because of the regulation of the CFA zone, the public sector could not increase its domestic borrow- ing. lTherefore, the external and intemal deficits are identical in ffiis phase. . ~ ~ ~ ~ ~ ~ I -/ An important feature of the balance of payments in Cote d'Ivoire is the high level of unrequited transfers by the private sector (about 5 Christophe Chlamley and Hafez Ghanem 297 Figure 6.13. Balance of Payments (Smooth Data), C6te d'lvoire, 1970-88 Percentage of GDI' 0* -21 -4 -6r -16 - . - -10 1970 1972 1974 1976 2978 1980 1982 1984 1986 1988 Transfers Covernment surplus - - - Balance of payments Source: World Bank data. percent of GDP). These transactions are probably made by expatriates and individuals with high incomes. The magnitude of the transfers indicates that a large fraction of private savings is not used for invest- ment in the Ivorian economy. The estimation of the impact on the trade surplus (TBY) of the terms of trade [LPXM = log(PX/PA4)], the real exchange rate [RER = log(P/ PM)], and the ratio of the primary govermnent surplus to GDP (PSURY) are reported in table 6.6, for 1971-81, and table 6.7, for 1979-89. The equation has been estimated for two periods because of the structural break that probably occurred in 1982. An overlap was necessary because of the sparseness of the data. The results support the descrip- tion given above: * In the first period the primary government surplus has no signifi- cant impact on the trade balance. The terms of trade and the real exchange rate have an impact in the expected direction. * In the second period relative prices do not have a significant effect, and the trade surplus is driven mainly by the government surplus. Some of the statistics of the estimation (residuals and the Durbin-Watson statistic) are not very satisfactory. This may be a result of the structural break that occurred during the period of estimation. There are not enough data for a post-1983 estimation. 298 CGLe d'lvoire: Fiscal Policy with Fixed Nominat Exchange Rates Table 6.6. Effect on the Trade Surplus, Cote d'lvoire, 197181 Standard Two-tail Variabte Coefficient error t-statistic signifiocnce C 0.408 0.087 4.69 0.002 LPXM 0.137 0.050 2.70 0.031 RER -0.379 0.109 -3.46 0.010 PSURY -0.205 0.259 -0.79 0.455 R2 0.89 Adjusted R2 0.85 S.E. of regression 0.012 Durbin-Watson statistic 2.51 Mean of dependent variable 0.075 S.D. of dependent variable 0.032 Sum of squared residuals 0.001 F-statistic 20.30 Note: Method, two-stage least squares; dependent variable, TBY; instruments, C. LPxM(-1), KER(-1), PsurY(-l), M2Y(-1). SE., standard error, S.D., standard deviation. Table 6.7. Effect on the Trade Surplus, CMte d'Ivoire, 1979489 Standiird Two-tail Variable coeciet err t-statistic significance C -0.055 0.130 -0.421 0.68 LPXM -0.090 0.048 -1.877 0.10 RER 0.17 0.150 1.165 0.28 PSURY 0.93 0.201 4.640 0.00 R>82 0.89 Adjustd R2 0.84 S.E. of regression 0.021 Durbin-Watson statistic 3.23 Log Iikelihood 29.00 Mean of dependent variable 0.10 S.D. of dependent variable 0.054 Sum of squared residuals 0.003 F-statistic 18.90 Note: Method, two-stage least squares; dependent variable, TBY; instruments, C, LpxM(-1), RER(-1). PSJiY(-1), MZY(-1). S.E., standard erro S.D., standard deviation. The same equation was also estimated with the trade surplus replaced by the balance of payments (Bon) and the primary surplus of the government replaced by the total surplus divided by GDP (SuRY). The results., presented in tables 6.8 and 6.9, are similar to those in table 6.7. Christophe Chamley and Hafez Ghanem 299 Saving and Investment The saving rate has a procyclical pattern, as one would expect from any consumption theory that is not based on hand-to-mouth behav- ior. We have computed and have shown in figure 6.14 two measures of the saving rate, as a share of GDP and of grss national income (GDP minus foreign interest payments). The more significant of the two Table 6.8. Effect on the Balance of Payments, CBte d'lvoire, 197143 Standard Two-tail Variable GCfficient ermor f-statistic significance C 0.26 0.102 2.59 0.029 LPXM 0.20 0.060 3.30 0.009 RER .-0.38 0.122 -3.17 0.011 SURY 0.23 0.178 1.30 0.226 R2 0.85 Adjusted R2 0.80 S.E. of regression 0.020 Durbin-Watson statistic 2.07 Mean of dependent variable -0.107 S.D. of dependent variable 0.0458 Sum of squared residuals 0.003 F-statistic 17.58 Note: Method, two-stage least squares; dependent variable, BOPY; instruments, C, uLXM(-1), RER(-1), suiy(-1), M2Y(-I), GY(-1). S.E., standard error; S.D., standard deviation. BOPY, balance of payments; SURYJ ratio of total surplus to aDr. Table 6.9. Effect on the Balance of Payments, COte d'Ivoire, 1979-88 Standard Two-tail Variable Coeficient error f-statistic significance C 0.032 0.1180 0-27 0.795 LPXM -0.074 0.0617 -1.21 0.271 RER -0.097 0.1387 -0.70 0.507 SURY 0.898 0.1963 4-57 0.004 R2 0.922 Adjusted R2 0.884 S.E. of regrest ion 0.021 Mean of dependent variable -0.103 S.D. of dependent variable 0.064 Sum of squared residuals 0.0028 Note: Method, two-stage least squares; dependent variable, aory; instruments, C, LPXM(-1), RE(-1), SURY(-1), GY(-1), M2Y(-l). S.E., standard error; S.D., standard deviation. 300 C6tc d'Ivoir: Fiscal Policy with Fixed Nominal Exchange Rates Figure 6.14. Saving and Investment, C6te d'Ivoire, 197187 Percent 40 - 35- 30- 2-5 15- 10 5- - -S - %t ,. 1971 1972 1974 1976 1978 1980 1982 1984 1986 1987 Investment Savinga Savingb --- ---Public saving a. As a share of GD?. b As a share of gross national income. Sotrce World Bank data. indexes is the saving rate out of national income. After reaching a peak with the first commodity boom, it decreased to a historical low in the trough of the cycle (1983), recovered a little in the second boom to a level below that of the early 1970s, and reached new lows in the current crisis. Although the data are somewhat sketchy, we have attempted to estimate a consumption function (table 6.10). The dependent variable is the ratio of consumption tO CDP, which varies inversely with the saving rate. Income shocks are represented by the rate of growth of real imcome (coP minus DRNY, factor payments). The variable SURNY represents the government surplus as a fraction of national income. The ordy vanable that is significant is the terms of trade. Its positive sign can be attributed to the dominant role of the terms of trade in the business cycle of C6te d'Ivoire: during a commodity boom (with higher terms of trade), the saving rate increases. Chistophe Chamley and Hafez Ghanem 301 Table 6.10. Consumption, C8te d'lvoire, 1972-87 --Standfard Tumtail Variable CeIient error -stdaistic significance C 0.603 0.1321 4.566 0.001 LPXM -0.128 0.0444 -2.898 0.014 RER 0.040 0.1791 0.226 0.825 DRNY -0.471 0.3684 -1L280 0.227 SURIW 0.1962 0.2820 0.694 0.502 R7 0.8101 Adjusted R2 0.741 S.E. of regression 0.021 Durbin-Watson statistic 2.247- Mean of dependent variable 0.630 S.D. of dependent variable 0.041 Sum of squared residuals 0.004 F-statistic 11.73 NoM: Method, two-stage least squares; dependent variable, CKto instruments, C, uwm, UXM(-1), RER(-1), sURY(-l). S.E., standard error, S.D., standard deviation. DRNY, factor payments; suRNY, ratio of govemment surplus to national income. Investment was significantly higher than domestic saving only dur- mng a short period between 1979 and 1983. The gap was financed mainly by the foreign borrowing of the public sector. Since 1984 investment has been below domestic saving. This situation is explained by the external flow of transfers and the current impos- sibility of financing public investment through foreign borrowing. The decomposition of investmen t between the private and the pub- lic sectors may be somewhat arbitrary because parastatals are substi- tutable with private firms. We have nevertheless attempted to pro- vide such a decomposition in table 6.11. (There are no reliable data before 1980.) The determinants of the investment equation are measured in table 6.12. Income shocks have a large impact on total investment. Govern- ment deficits have a negative, relatively small and marginally signifi- cant, impact on investment. However, the dependent variable is the level of total investment. If a large fraction of the deficit is used to finance public investment, the results imply that the aowding-out effect on private investment is larger. No data are available for the direct estimation of the impact of public investment and deficits on the level of private investment. A complete analysis of the behavior of private investment would require consideration of many other factors, the most important of which is probably the profitability of capital. Such an analysis is beyond the scope of the present study. We suspect that the high level 302 Cdte d'lvoire: Fiscal Policy with Fixed Nominal Exchange Rates Table 6.11. Public and Private Investment and Consumption, CMte d'lvoire, 198047 (percentage of CDP) Item 1980 1981 1982 1983 1984 1985 1986 1987 Investment Total 0.260 0241 0.213 0.176 0.123 0.117 0.118 0.124 Public 0.115 0.087 0.066 0.061 0.043 0.031 0.038 0.036 Consumption Total 0.596 0.631 0.615 0.625 0.625 0.602 0.618 0.628 Public 0.177 0.171 0.169 0.163 0.153 0.139 0.151 0.163 Sourre: World Bank data. Table 6.12. Investment, COte d'Ivoire, 1972-47 Standard Two-tail Variable- Coefficient error tstatitic significance. C 0.004 0.011 0.42 0.677 rY(-1) 0.860 0.059 14.40 0.000 DRfiY 0-372 0.031 11.60 0.000 SURNY -0.102 0.049 -2.05 0.062 ,-Rs2 0.91 Adjusted R1 0.89 S.E_ of regression 0.01 Durbin-Watson statistics 2.07 Mean of dependent variable 0.20 S.D. of dependent variable 0.05 Sum of squared residuals 0 01 F-statistic 44.7 Note Method, two-stage least squares; dependent variable, IY. ratio of investment to cD, instrments, C IY(-1), Da(-1), suty(-1); S.E, standard error S.D., standard deviation. of domestic prices (in relation to competing countries) had a strong negative impact on the marginal product of capital in the private sector. Since 1983 there has been little intemational lending to the private sector. We cannot determine here whether this is attributable to rationing by foreign creditors or to the low profitability of new capital in C6te d'lvoire. The significance of the growth variable (DRNY) shows that in the Ivorian economy growth is a major factor in determining investment. Condlusion The economy of C6te d'Ivoire presents a case study that highlights the effects of fiscal policy in a developing country with a fixed Chrisfophe Chamley and Hafez Ghanem 303 exchange rate. Monetary policy is constrained by the regime of the CFA zone. For the past fifteen years the growth of the Ivorian econ- omy has been dramatically affected both by exogenous factors and by the responses of fiscal policy to these factors. The potential impact of a commodity boom on the relative price of tradables and nontradables is well known. This relation has been observed for Cote d'Ivoire, and it has been illustrated in particular by the evolution of intersectoral prices (figure 6.10) and by the econo- metric results of table 6.5, in which the impacts of private and public expenditures are separated. The adverse evolution of the real exchange rate has led to high external deficits. For the period covered in this study, the econometric results indicate that before 1982 the real exchange rate had a strong impact on the external balance. After that date, this relation was replaced with a straight identity between the government deficit and the external deficit. The importance of the impact of govenment expenditure on the economy has been recogrnized since the beginning of the adjustment period more than ten years ago. The implementation of policy reforms has been extraordinarily slow, however Before the crisis of 1990, which put the public sector in a state of quasi receivership, there had been no indication that the government had adjusted the level of its permanent expenditures to match the country's long-run capacity to generate revenues. Expenditures were indeed reduced signifi- cantdy in 1983-84, but these reductions affected mainly capital ex- penditures and were not accompanied by cuts in current expen- ditures. For instance, from 1980 until 1989 the wage bil increased regularly. The current situation illustrates the failure of the adjustment pro- gram of the 1980s. Simce 1980 foreign indebtedness has increased from 30 percent to more than 100 percent of GDP- This foreign borrow- ing has not served new investment; the ratio of real investment to CDP has shrunk to a level that is barely sufficient to compensate for deprecation. The expansionary fiscal stance has contributed to the downward rigidity of the domestic price level. Under the fixed-echange-rate regime the competitive position of the cotuntry has worsened sig- nificantly with respect to its direct competitors in world trade, in- cluding Malaysia and neighboring countries in West Africa (for example, Ghana and Nigeria) that are not tied by fixed nominal ex- change rates. The goverment faced the necessity of introducing drastic changes in fiscal policy only when it had to dedlare insolvency. It is unfortu- nate that the present financial constraints severely limit the resources available for future growth. 304 Chte d'Ioire: Fiscal Policy with Fixed Nominal Exchange Rates Appendix 6.1. The Size and Burden of the Public Sector The share of the tax base in the Ivorian economy is much smaller than in industrial countries, and the administrative resources that are available for the collection of revenues are limited. Of the various indexes for measuring the size of the tax base, we consider two: the size of the government labor force and the average effective tax rate on the formal nonagricultural sector The Labor Force in the Public Sector According to the available data (from the 1989 Public Expenditures Survey) the number of permanent civil servants, exduding those involved in activities that provide marketed goods and services, was about 110,000 in 1989. This number was probably not very different than in 1985, when the total number of taxpayers was about 209,000. (The number of individuals who filed tax forms was about 10,000.) The ratio between the number of civil servants,-defined as above, and the number of taxpayers who support them is therefore about 50 percent. To put the situation in perspective, we find it difficult to imagme an industial country in which one-tid of the formal labor force is employed in government activities that do not generate a market income. The Average Effective Tax Rate The formal sector pays taxes in three stages: at the border, on firms' profits, and on personal income. A significant fraction of imports is by tax-exempt government agencies. The effective average rate of the border taxes is very high; it has been estimated at about 65 percent (Horton 1990a). Some of the goods subject to border taxation are imported for the informal sector. Correcting for these leakages, as evaluated by the Wodd Bank, and using data on the taxes and the value added of the informal sector, the average effective tax rate on activities in the formal sector is about 48 percent. The high rate of effective taxation may explain the slow but persistent reduction of the real level of activity in the formal nongovemment sector since 1980. One branch of the formal sector has grown steadily, however-the bottling of beverages. Some would regard production by this sector as an index of domestic production (and demand) that is more accu- rate than CDP. Employment in the formal private sector-already small-is not growing. In fact, it declined significantly after 1980, to 197,100 in 1982, 168,800 in 1984, 162,200 in 1986, and 146,000 in 1988. Employ- Christophe Chamley and Hafez Ghanem 305 ment decreased by 50,000 in six years, at an average rate of about 5 percent per year between 1982 and 1988. The number of African employees in managerial positions increased, however, as the num- ber of non-Africans decreased. The entire reduction of the labor force in the formal sector fell on nonmanagement workers. Appendix 6.2. Public Investment The Ivorian government reacted to the coffee and cocoa boom of the mid-1970s by increasing its capital expenditures from 8 percent of CDP in 1975 to 22 percent in 1977. Private investment remained relatively stable during that perod, at about 13 to 15 percent of GDP. The increase in the share of the public sector in total investment was correlated with an apparent dedine in the return on capital. In 1981 the agricultural sector received 31 percent of public investment (table 6.13), much of it in extension services, induding recurrent costs (mainly wages). The rate of return on agricultural investments is diffi- cult to measure: the data do not indicate an increase in agricultural output and value added as a result of these expenditures. Since most of the expenditures provided extension services for coffee, cocoa, and cotton, they did not encourage diversification out of traditional agri- cultural exports, as had been recommended by numerous World Bank missions. The second largest share of total public investment-23 percent in .1981-was devoted to transport and roads. As a result of the public investment boom, C8te d'Ivoire today has the most developed high- way and road system in Africa. However, some of the superhighways are sparsely used and have a low rate of return. it 1981, 16 percent of the investment budget was allocated to edu- cation. This included the construction of the higher institutes (gnndes ecoles) of Yamoussoukro, in which the cost per student is higher than that m most Ivy League colleges in the United States. Table 6.13. Sectoral Composition of Public Investment, CGte d'Ivoire, Selected Years, 1981-90 (percent) Im -i9821 1984 1987 19-90 Agriculture 31 26 28 47 Transport and roads 23 27 29 12 Housing and urban development 13 18 17 20 Education 16 12 6 3 Other 17 17 20 18 Public ins Kstment as a share of cDP 17 8 5 4 Soure:e World Bank data. 306 Cdte d 'lvoire: Fisoal Policy with Fixed Naminal Exchange Rates When the public sector entered into productive activities-for example sugar and palm oil factories-the results were discouraging. The public companies that produce those goods do so today at a cost that is nearly double the world price. The public investment program was drastically reduced after 1982, falling from 17 percent of GDP in 1981 to 4 percent of CDP in 1990 (see table 6.13). Its composition also changed. The share of agriculture in total investment increased from 31 to 47 percent, reflecting the diffi- culty of reducing employment in the extension services. Investment in transport and road construction fell from 23 to 12 percent of the reduced total, and road maintenance is now becoming an inportant problem. Investment in education fell dramatically, from 16 percent of total investment in 1981 to 3 percent in 1990, at the same time as the population was growing at 3.5 percent a year Today the size of the public investment program is, in relative terms, about half of its 1975 (pre-boom) level. Moreover, about 30 percent of what is now defined as capital expenditure consists of salary payments to extension workers, an item that was not significant in 1975. Appendix 6.3. Costs of and Returns to Education Government expenditures on education in C6te dlvoire have been. estimated at 40 to 45 percent of tota expenditures (Mingat and Psacharopoulos 1985). Numerous studies have found that the wages of educational staff are very high in relation to those in other coun- tries at a simlar stage of development. H Kgh salaries in the education sector can be observed in all Fran- cophone countries. The unit costs of public education at the primary level are at least 50 percent higher than in Anglophone Sub-Saharan African counties, where the cost is already much larger than in other developing countries. The ratio between unit costs in Cote d'Ivoire and Anglophone countries is even greater for secondary (3.4) and higher (1.5) education. Furthermore, the education system in Cote d'Jvoire places more emphasis on secondary and higher education tan in other African countries. The share of education expenditures in GDP is thus twice as high in C6te d'Ivoire as in other Sub-Saharan African countries (10 percent compared with 5 percent) or in industrial countries (about 5 percent). Since coP per capita is higher in Cote d'Ivoire than the average m Africa, the ratio of 10 percent is quite remarkable. The high cost of staff is a consequence of institutional rigidities. When a severe budget problem occurs, the government tries to reduce expenditures on other items, such as books and supplies. Some reports indicate that the quality of education may have suffered as a result. One might expect that Ivorians are at least getting more for their money, but Christophe Clhamicy and Hafez GhuInem 307 Table 6.14. Enrollment Ratios, Sub-Saharan Africa, 1983 (percent) Level of education Country orgroup Primary Secondary Higher Cote d'Ivoire 60 15 1.9 Francophone countries 46 14 2.4 Anglophone countries 77 17 1.2 Source: Mingat and Psacharopoulos 1985. measures of output are not better than in other counties, and enroll- ment ratios are lower at the primary and the secondary levels than in Anglophone countries (table 6.14). The high level of wages in the formal sector inflates the measure- ments of the private rate of return to education, which have been found to be high in Cote d'lvoire (van der Gaag and Vijverberg 1987). In an environment in which institutional constraints cause the supply of jobs to be rationed because of the high wage rate, education may be an important signaling device for allocating employment. Indeed, van der Gaag and Vijverberg find that holding a diploma is a impor- tant determinant of an individual's rate. This situation-may lead to an overstatement of the rate of Teturn if the wage rate is used as an indicator of the productivity of learing. Notes 1. The main work on this chapter was done in 1990. Hence we do not take account of the most recent changes in economic policies, particularly the devaluation of the CFA franc. 2. Other studies on commodity prices have been undertaken by Ardeni and Wright (1990) and Cuddington and Urzua (1987). 3. The contibution of primary commodities is overstated here because the accounting procedure implies no import cost. 4. For a discussion of financial taxation in Sub-Saharan African countries, see Chamley (1990). Ivorian residents benefit from a fixed exchange rate with France and from capital mobility with European markets that is practically free for liquid assets. The fixed parity of the exchange rate (which has been' left untouched since the early 1950s) is credible because of the difficulties of changing the system for all countries of the CEA zone at the same time. In this situation there are standard theoretical reasons for the nonexistence of a demand function for CFA francs in Cote d'Ivoire. This point is discussed again in the section "The Price Level and the Real Exchange Rate." 5. There is no inflation tax adjustment because the CFA rules effectively prevent the use of seigniorage as a significant source of revenue. The average level of seigniorage between 1979 and 1988 was less than 0.5 percent of GOP. 6. Outflows of capital are, of course, discouraged. Anecdotal evidence about the remission of CFA notes from foreign deposits (for example, in 308 COte d'lvoire: Fiscal Policy with Fixed Nominal Exchange Rates Switzerland) indicates that controls cannot be very effective for holders of large deposits. Small depositors use money only for transactions and are probably not very sensitive to fluctuations in the opportunity cost of money in the range that has been observed under the CpA regime. As mentioned earlier, attempts to find an econometric relation between the level of money and the observed real interest rate have failed for Cbte d'Ivoire. 7. The fiscal policy was more expansionary in France during the early 1980s. 8. We did not have data on the foreign indebtedness of the private sector. The constraint on foreign borrowing is mentioned in various recent reports; the data presented here can be interpreted as supporting the existence of such a constraint. References Ardeni, Pier Giorgio, and Brian Wright. 1990. "The Long-Terrn Behavior of Commodity Prices."' Policy Research Working Paper 358. World Bank, International Economics Departnent, Washington, D.C. Bevan, D. L, P. Collier, and J. W. Gunning. 1987. "Consequences of a Com- modity Boom in a Controlled Economy: Accumulation and Redistribution in Kenya 1975-83." World Bank Economic Review 1(3): 489-513. Clawson, Patrick. 1989. "Government Wage Policies." Report for the World Bank's Public Expenditure Review. World Bank, Occidental and Central Africa Department, Washington, D.C. Chamley, Christophe. 1990- "Fmancial Taxation in Sub-Saharan Counties." World Bank, Country Economics Department, WVashington, D.C. Coldough, Christopher. 1980. Primary Schooling and Economic Developmet: A Review of the Evidence. World Bank Working Paper Series 399. Washingbon, D.C. Cuddington, John T., and C. MW Urzua. 1987. "Trends and Cycles in Primary Commodity Prices." Georgetown University, Department of Economics, Washington, D.C. Deaton, Angus, and Dwayne Benjamin. 1987. "Household Surveys and Pol- icy Reform: Cocoa and Coffee in the a8te d'Ivoire." Discussion Paper 134L Princeton TJniversity, Research Program in Development Studies, Prince- ton, N.J. Devarajan, Shantayanan, and Jaime De Melo. 1988. "Adjustment with a Fixed Exchange Rate: Cameroon, C6te d'lvoire, and Senegal." World Bank Economic Review 1(3): 447-87. Fields, Gary S- 1988. 'Labor Market Polcy and Structural Adjustment irn C6te d'Ivoire." Comell University, Economics Department, New York- Hinchliffe, Keith. 1985. "Higher Education in Sub-Saharan Africa." Ear Dis- cussion Paper 3. World Bank, Education and Training Department, Wash- ington, D.C. . 1986. "The Monetary and Non-Monetary Retums to Education in Africa." EDT Discussion Paper 46. World Bank, Education and Training Department, Washington, D.C. Cliristopiwe Chamley and Hafez Glianem 309 Horton, Brendan. 1990a. "C6te d'Ivoire: Commerce Exterieur, Incitations et Reglementation." World Bank, Occidental and Central Africa Department, Washington, D.C. . 1990b. "L'emploi et les Salaires." World Bank, Occidental and Central Africa Department, Washington, D.C. Melhado, Oscar. 1990. "A Note on the Time Series Characteristics of Cocoa Prices," Boston University, Economics Department, Boston, Mass. Mingat, Alain, and George Psacharopoulos. 1985. "Education Costs and Financing in Africa: Some Facts and Possible Lines of Action." EDT DiscuS- sion Paper 13. World Bank, Education and Training Department, Washing- ton, D.C. Riveros, Luis A. 1989. "International Differences in Wage and Nonwage Labor Costs." Policy Research Working Paper 188. World Bank, Country Economics Department, Washington, D.C. Rodriguez, Carlos Alfredo. 1990. "The Macroeconomics of the Public Sector Deficit: The Case of Argentina." Policy Research Working Paper 632. World Bank, Country Economics Department, Washington, D.C. Tanzi, Vito. 1987. "Quantitative Characteristics of the Tax Systems of Devel- oping Countries." In David Newbery and Nicholas Stern, eds., The Teauory of Taxation for Developing Countries. New York: Oxford University Press. UNDP (United Nations Development Programme) and World Bank. 1980. Afri- can Economic and Financial Data. New York. UNICEF (United Nations Children's Fund). 1990. The State of the World's Chil- dren 1990. Oxford, U.K.: Oxford University Press. Van der Gaag, Jacques, and Wim Vijverberg. 1987. Wage Detertmnants in Cote d'lvoire. living Standards Measurement Study Working Paper 33. Wash- ington, D.C.: World Bank. World Bank. 1983a. "Comparative Education Indicators." Education and Training Department, Washington, D.C. 1983b. Kenya: Growth and Structural Change. 2 vols. A World Bank. Country Study. Washington, D.C. .1988a. Education in Sub-Saharan Africa: Policies for Adjustment, Revitalia- tion, and Expansion. A World Bank Policy Study. Washington, D.C 1988b. "Cote d'Ivoire: La Mobilisation des Resources Intemes en Vue d'une Croissance Stable." Occidental and Central Africa Department, Washington, D.C. 1990a. "C6te d'Ivoire: Exanen des Depenses Publiques." Occidental and Central Africa Department, Washington, D.C. 1990b World Tables. Bltimore, Md.: Johns Hopkins University Press. 7 Ghana: Ad ustment, Reform, and Gro Roumeen Islam and DeboTah Wetzel When Ghana became independent in 1957, its resource base was large, its population was well educated, and its economic infrastruc- ture was strong. By 1983, however, the economy was a shambles, with income per capita more than 10 percent lower than in 1957. In April 1983 the government initiated the Economic Recovery Pro- gramme (mRP) in an attempt to rescue the economy. Since then, the economy of Ghana has made steady progress: the average annual growth rate of real GDP during 1984-90 was 4.5 percent, compared with -0.5 percent during 1978-83. This chapter investigates the role that fiscal deficits have played in Ghana's economic dedine and renewal. Our main conclusion is that Ghana's high money-financed fiscal deficits, in conjunction with the direct controls imposed on the economy, caused severe macro- economic imbalances and reduced growth until 1983. Lower fiscal deficits, liberalization of the economy, and access to foreign financing led to improved economic performance after 1983. The first section presents an overview of economic policy and the control regime that has been in place since the early 1960s. It then discusses some of the principal factors that have affected Ghana's deficit. The second section focuses on fiscal deficits and financial markets. It considers the effect of inflationary financing on the demand for money and quasi money and for long-run- seigniorage, and it addresses the sustainability of the deficit. The third section assesses the impact of the fiscal deficit on private consumption and investment. The fourth section develops the relationship between fiscal deficits, the trade balance, and the real exchange rate. The fifth section presents our condusions. Economic Policy and Fiscal Deficits since the 1960s In this section we provide an overview of economic policy during the past three decades. We go on to look at the effect of economic policy on the magnitude of the fiscal deficit in Ghana. 310 Roumeen Islan and Deborah Wetzel 311 An Overview of Economic Policy From 1961 to 1966, under President Kwame Nkrumah, who had come to power with independence in 1957, Ghana's economic strategy con- sisted of massive investment by the state in the hope of p:opelling the country into higher growth.' The strategy focused on national plan- ning, import substitution, and the development of public enterprises. In order to contain both the budget and current account deficits, a series of controls was put in place in 1961. Foreign exchange, price, and financial sector controls were implemented. Credit was allocated by the government, with import-substituting industries benefiting at the expense of agriculture. Quantitative restrictions on imports were used to control demand. The first column in table 7.1 provides summary statistics on the performance of the economy during the period 1961-66. (See appen- dix 7.3 for annual data.) Investment during this period averaged 17.3 percent of GDP, and the average central government deficit was 6.4 percent of GDP. During 1967-71 there was some econonic liberalization, much higher growth, lower inflation, and a lower central govemment defi- cit. The National Liberation Council, or NLC (1967-69), and the Prog- ress Party (1969-71) attempted to restore internal and external eco- nomic balance with the aid of a stabilization loan and technical assistance from the International Monetary Fund (IMF). In 1970 the government took advantage of high world cocoa prices to permit a rapid expansion of imports and public expenditures. However, strong demand-resulting from increased income-and import liberalization led to an acute current account deficit. After a coup by Lieutenant-Colonel I. K. Acheompong and the National Redemption Council in 1972, liberalization stopped and con- trols were reimposed. Another series of attempted coups led in 1975 to the establishment of the Supreme Military Council and in 1978 to the regime led by Lieutenant-General Frederick Akuffo. In 1979 a coup brought to power Lieutenant Jerry Rawlings, who implementc-l a currency reform to reduce money in circulation and to control infla- tion. After a brief period of elected govenment, Lieutenant Rawlings took over in 1981 and initiated the ERP in 1983. The period 1978-83 was thus one of prolonged economic and politi- cal crisis. Annual average growth of per capita GOP during this period was -2.7 percent, while inflation averaged 73.2 percent. The real aterest rate was negative, the real exchange rate became highly over- valued, and the black-market premium averaged 9.5 percent. Invest- ment averaged only 5 percent of GDP. The results of the control regime were shortages of inputs, reduced investment and exports, rent-seeking, corruption, and the development of an extensive black- 312 Ghana: Adjustment, Reform, and Growvthi Table 7.1. Summary Statistics, Ghana, 1961-90 (period averages) item -961-66 1967-71 1972-77 7978-83 1984-90 Aggregate variable Growth of real GDP (percent) 3.0 4.3 0.3 -0.5 4.5 Growth of GDP per capita (percent) 0.4 2.0 -2.3 -2.7 1.2 Inflation (percent) 11.8 3.9 41.4 73.2 29.7 Real discount rate (percent) -5.9 2.6 -20.0 -31.4 -4.2 Real exchange rate (cedis per U.S. dollar)" 94.3 95.6 77.7 18.2 77.1 Black-market premium 1.6 1.7 2.7 9.5 1.8 Composition of output (percentage of GDP) Private consumption 76.5 75.2 76.3 86.6 82.9 Public consumption 121 14A 12.3 9.4 9.4 Gross fixed capital formation 17.3 11.4 9.8 5.0 10.9. Change in stocks -0.5 0.9 0.5 -0.1 0.1 Exports 19.2 18.9 17.7 7.0 16.0 imports 24.6 20.8 16.6 7.9 19.3 Private fixed investment - 8.2 5.7 3.0 4.5. Public fixed investment - 4.1 4.7 1.9 6.5 Monetary system (percentage of GDP) At - 16.8 14.3 183 15.2 12.3 Quasi money 3.9 5.4 7.6 5.2 3.9 Balance of paymenzts (millions of U.S. dollars) Trade balance -38.5 25.3 102.3 47.4 -6L9 Current account -103.9 -82.9 -12.1 -99.7 -79.4 Capital account 89.6 54.9 8.7 85.7 176.0 Errors and omissions -1.6 0.5 -4A -76.8 -19.1 Position above the line -15.9 -27.6 -7.8 -90.7 77.6 Total change in reserves -15.9 9.8 -15.5 39.0 - 46.1 Centnd government accounts (percentage of GDP) Central govemment revenue 17.0 17.6 13.1 6.6 13.3 Central govement expenditure 23.4 21.0 23.1 12.8 -13.6 Central government defict -6.4 -3.4 -10.0 -6.1 -0.3 Domestic finance - 1.7 8.9 5.6 0.0 Extemal finance - 1.2 0.1 OA 0.3 Other - 0.6 1.0 0.6 0.0 -Not available. a. The real exchange rate is calculated as the cedildollar rate multiplied by the wholesale price index (1985 = 100) and divided by the Ghanaian cp1 (1985 = 100). A fall in the real exchange rate indicates an appreciation, and a rise indicates a depreciation. Note: See appendix 7.3 for details on calculations and souces. Averages are unweighted means. Roumeen IsLam and Deborah Wetzel 313 market economy. The control regime also placed severe constraints on the government's ability to finance its deficits. Ghana's poor record in debt repayment, combined with a hostile attitude toward foreign capital, left the country with Little access to foreign finance until the ERP was put into place. With the onset of the ERP in 1983, a period of stabilization and adjustment under the auspices of a structural adjustment program began. The government devalued the cedi and removed controls on foreign exchange operations. Price controls and the licensing regime on imports have been eliminated, the govemment has called for the "rationalization" of the public sector, and trade taxes have been reformed. More recently the government has undertaken reform of the financial sector. As a result, real GDP growth averaged 4.5 percent between 1984 and 1990 (see table 7.1), and inflation has been brought down. Real interest rates are positive, the real exchange rate has depreciated significantly, and the black-market premium has been dramatically reduced. Investment has increased, but this is largely attnbutable to increased public investment. In summary, under the ERP Ghana's economic decine was arrested, and the country began to move toward steady economic progress. Measuring the Fiscal Defiit -In Ghana there are no consistent data on local government accounts and state-owned enterprises. An attempt was made in this study to construct a measure of the consolidated public sector deficit using the accounts available, indluding the accounts of the Social Security and National Insurance Trust, the Cocoa Marketing Board, and the Ghana hidustral Holding Corporation. (See Islam and Wetzel 1991 for details on these accounts.) The result was not significantly different from the central government accounts. THE CENTRAL GOVERNMENT DEFcIT. Figure 7.1 shows total central government revenue and grants, total expenditure (induding net lending), and the deficit, defined conventionaUlly as revenue and grants minus expenditure of the central government.2 Only in fiscal 1970/71 and in the years since 1986 has the central government bud- get been in surplus. Note, however;, that because the central government accounts pre- sented in figures 7.1 and 7.2 exclude capital expenditure that is financed through external project loans, the expenditure category is underestimated. Data on these capital expenditures are only available from 1984; their inclusion yields the following figures (as percentages of GDP) for total expenditure and the central govenmment fiscal deficit: 314 Ghana: Adjustment, Reformn, and Growth Figure 7.1. Central Government Revenues, Expenditures, and Deficits, Ghana, 1960/61 to 1990 Percentage of GDP 25 - 20 15 5 -0 - -20 4 I I I I 1960/61 1965/66' 1970/71 1975/76 1980181 1985 1990 -0 Revenue -4--talExpenditure ~-Deficdt Note: years through 1982- axe fiscal years; calendar years axe used thereafter. Source- World Bank data. 1984 7985 1986 1987 1988 1989 2990 Capital expenditure financed through project lending 0.7 1.2 3.1 2.9 3.0 2.4 3.0 ToIul central goverrunment expenditures 10.9 15.2 17A 17.2 17.3 16.8 16.6 Central government surplus (+) and defict 1>) -2;5 -3A -3.3 -2.4 -2.6 -1.7 -2.9 If this foreign-financed expenditure is taken into account, the fiscal surpluses that appear in the official accounts after 1986 become defi- cits that range from about 2 to 3 percent of GDP. Figure 7.2 shows central government expenditure and its decom- position by economic category as a percentage of total expenditure. The breakdown of expenditure shows dearly the dominance of con- Roumeen Islnrm and Deborah Wedez 315 Figure 7.2. Decomposition of Central Government Expenditure, Ghana, 1969/70 to 1989 Percentage of total expendihtre 100 90 80 70 60 50 40 30 20- 10 1969/70 1973/74 1977/78 1981/82 1985 1989 E= consumption 22 Transfers CM Net lending EJ nterest - Gross fixed 90 Other capital formiation Note: Years thtrough 1982 arm fisat years; calendar years are used thereafter Sour=e World Bank data. sumption spending over spending on capital formation. The share of the latter continued to fall until 1983; in absolute terms this dedine was the mnore dramatic because total spending fell during the same period. This led to a rapid decline in the country's infrastructure and helps to explain the country's negative growth over the period. Under the MP' gross fixed capital formation by the central government rose from 1.10 to 7.17 percent of GDp.3 As illustrated in figure 7.1, central government revenue (induding grants) as a share of GDP reached a peak in 1970171, at 20 percent of GDP and fell to a low of 5.4 percent of GDP' in 1982. At the root of the instability and the declinLe in revenues was the government's depen.- dence on export duties from cocoa as its principal source of revenue. Until 1981 the export duty~ on cocoa typically.provided between 25 and 40 percent of the governument's revenues. 316 Glana: Adjusht nt, Reworm, and Growthi Figure 7.3. DirectTaxes, Taxes on Goods and Services, and International Trade Taxes, Ghana, 1966/67 to 1989 Percentage Of CDP 10 9 7 6 5 4 3 1966/67 1971/72 1976/77 1981/82 1986 1989 -C--- Direct taxes i- Goods and semrices -a--- Inbernational trade taxes Note: Years through 1982 are fiscal years; calendar years are used thereafter. Sourwc World Bank data. Figue 7.3 presents direct taxes, taxes on goods and services, and international trade taxes as a percentage of GDP. Each of these taxes declined as a share of cDP between the late 1960s and the early 1980s. The volatility of trade taxes is immediately obvious from the figure. The govenmment's domestic policies were responsible for the decine in these taxes. In certain years-for example, 1978179-free on board (Foe) prices for cocoa were higher, yet revenues from export taxes were much lower.4 The decline in revenue is largely attributable to the decline in official cocoa production in the 1970s caused by the fall in both nominal and real producer prices: during the 1960s producers switched to new crops or smuggled cocoa out.5 The Alien Compli- ance Act of 1969 forced many workers to leave the country, which in turn caused labor shortages-6 Appendix 7.4 presents some estimates of the revenue that has been lost as a result of smuggling. The revenue obtained by the govern- ment per metric ton of cocoa was applied to an estimate of total Roumeen Islam and DebomIh Wetzd 317 anmual smuggling of cocoa. The average total revenue lost over the penod 1970171 to 1979180 is 0.5 percent of cDIP. Thus the primary explanation for the loss in revenue from export taxes was not smug- gling but the decline in cocoa production. Since 1984 all three tax measures-direct taxes, taxes on goods and services, and international trade taxes-have increased as a percent- age of wrp. The inaease in revenue from direct taxes (about 20 per- cent) and from taxes on goods and services (about 25 percent) is attributable to improved tax administration combined with growth. E-sEcrs OF ECONOMIC AND POLICY VARIABLES ON THE CENTRAL GOV- ERNMEy DEIC Irr. Marshall and Schmidt-Hebbel (1989) have devel- oped an accounting framework that allows for the decomposition of the changes in the deficit into changes attnbutable to feedback effects (for example, c;P growth and hiflation rates), to extenal variables, and to fiscal policy variables. The results of the application of their methodology to the Ghanaian central government deficit are shown in appendix 7.5. Figure 7.4 shows the change in the deficit, excluding dhanges attributable to domestic macroeconomic variables and changes attrib- utable to extemal variables. In most years almost the entire change in the defict is explained by fiscal policy variables, in particular the wage bilL (For example, in 1985 the wage bill caused a change in the deficit of 2.5 percent of cGD..) Transfers and spending on goods and services both had large effects on the deficit in some years; the latter had its greatest effect in 1984, when it inaeased the deficit by 2.0 percent of CDpU The years in which there is a divergence (for example, 1981182) are years in which high inflation-more than 100 percent in 1977 and 1983-had feedback effects on tax revenue. In the estimation of the tax functions, inflation was found to have a significant negative impact on direct tax revenue and on "other indirc tax revenue" (indirect taxes minus import duties and export taxes, whidh were estimated separately). There is thus an Olivera-Tanzi effect for both direct and other indirect taxes. MEASURNG THE PUBLIC SECrOR DEFICIT FROM FINANCIAL FOws. Using data from the monetary survey on foreign liabilities and on the claims against the central govenment, nonfinancial public enterprises, and the Cocoa Marketing Board, we derived a measure of the total outstanding debt stock of the public sector. Differencing these data on stocks provides us with a measure of the financing flows to the public sector and hence with an alternative measure of the fiscal deficit.7 As illustated by figure 7.5, this measure shows a pattern similar to that of the central government deficit 318 Ghana: Adjustment, Refbrm, and Growth Figure 7.4. Change in the Central Government Deficit, Ghana, 1972173 to 1988 Percentage of CDP 9 1 7 -I' 6 ft 5 4 I 3 2-1 -2 -3 -4t -5 1972/73 1977/78 1982 19871988 ----a---- Change in deficit +- Change in deficit (excluding extemal ard fiscal policy variables) NoLe Years through 1982 are fiscal years; calendar years are used thereafter. Source World Bank data. The differences between the measures are attrnbutable to three fac- tors. First, the total public sector measure includes the foreign financ- ing of capital expendture arising from project loans, which is not incduded in the central goverrunent accounts. Second, in the central government measure grants are treated as revenue-or as above the line-whereas in our second measure they are considered a compo- nent of financing. Third, the second measure indudes financing flows to public entities and to the Cocoa Board. A comparison of the two measures shows that prior to 1984 the finacing measure of the deficit is in some years smaller than the conventional measure. Before 1983 both grants and external financing of capital expenditure are small or zero, and in some years financial flows to public entities are negative. After 1983, however, the two measures are very smillar. If we move the grant component of central goverrunment revenue below the line and include capital expenditure Rowneen Ilam and Deborah Wetzd 319 Figure 7.5. Public Sector Deficit on the Basis of Fmancal Flows, Ghana, 1967-90 Percentage of cDP' 4- 3- 0 - -2- -3- -~5- -6- -7- -8 -9 -12I 1967 1970 1973 1976 1979 1982 1985 1988 1990 Source: World Bank data financed through external lending, our central government measure is quite dose to our public sector measure.f Fiscal Deficits and Financial Markets In this section we eamine domestic financing, inflation, the demand for assets, and seigniorage. We go on to establish model equations and to look at the demand for money and quasi money. We then investigate the effect of inflaton on seigniorage and the susiability of the deficit. Domestic Financing of the Public Sector Deficit Figure 7.6 Miustrates the financial flows to the public sector (including nonfinancial public enterprises and the Cocoa Marketing Board), bro- ken down into flows from the monetary authority and the domestic banks (or to the monetary authority and the domestic banks, if the flows are negative), Flows to the public sector from the prvate sector are negligible and have not been indluded in the figure. Because of the imited development of the banking system in Ghana, the govern- 320 Ghana: Adjustment, Reform, and Gmwth Figure 7.6. Domestic Financing of the Public Sector Deficit, Ghana, 1971-90 Percentage of Mor 10 9 8 7 6 4 3 2 0- -1 -2 -3 -4 -5 1971 1975 1979 1983 1987 1990 -a--- By monetary authority i By domnestic banks Source: World Bank data. ment's ability to borrow fom domestic banks has been weak, averag- ing about 2 percent of GDP. Until 1984, instead of borrowing from domestic bariks the government relied on borrowing from the mone- tary authority for the greatest part of its finance. Money financing has caused high inflation. Figure 7.7 illustrates inflation rates, measured by changes in the consumer price index, (Cr1) during the 1971-90 period, with rates of more than 100 percent in 1977, 1981, and in 1983. (See appendic 7.3 for yearly rates.) These inflationary peaks also mark years in which shortages in the food supply caused food prices (which have accounted for 50 percent of the ca) to xise. A simple regression of the inflation rate on borrowing from the central bank and dummy variables for the years of drought and fires indicates a significant relationship between monetary finance and inflation.9 The high levels of inflation that were brought about, at least in part, by the monetary financing of the defict in turn affected other areas of the economy- Since the government controlled nominal interest rates, Roumeen Iskm and Deborah Weizef 321 Figure 7.7. CPI Inflation Rate, Ghana, 1971-90 Percent 130 120 110 100 90 80 70 60 50 40- 30- 20 10 -~~~ 1_0 Y * 1. . . 1971 1976 1981 1986 1990 Source: World Bank data. high inflation implied negative real interest rates for deposit and lending (figure 7.8) and a fall in holdings of financial assets.10 Nega- tive real interest rates and taxes on financial intermediation led the banking sector to institute stringent requirements for credit to the private sectoL Aryeetey and Gockel (1990) argue that the system of financial con- trols combined with high iflation encouraged the development of informal credit markets. They estimate that the formal sector now controls about 55 percent of the financial savings mobflized per month. Financial disintermediation and a movement into infonnal credit markets were also encouraged by the 1979 currenry conver- sion. Cuxrency notes outside banks were exchanged at a ratio of 10 old cedis to 7 new cedis for amounts of less than 5,000 cedis and at a ratio of 10 to 5 for amounts of more than 5,000 cedis. As a result of the currency conversion, the money supply was reduced by about 12 percent of broad money. The conversion, along with other measures (demonetization of the 5O-cedi note, for example), undermined confi- dence in the banking system. Financial disintermediation and credit constraints also had important implications for economic growth: pri- rate investment was constrained by a lack of liquidity. Monetary 322 Ghanr Adjustment, Refonn, and Growth Figure 7.8. Nominal and Real (Twelve-Month) Deposit Rates, Ghana, 1971-90 Percent 30 - . .. ' 20- 1.0 -10- -20- -30- -40; -50 -1971 1976 1981 1986 1990 - Nominal deposit rate i Real deposit rate Not Real interest rates are calculated as I[(O+z-)f(l+p)I- 11 100, where i is the nominal interest rate and p is the cpt inflation rate. Sourmc World Bank data. policy was rendered impotent-the discount rate became ineffective as a policy tool, and banks consistently held excess reserves. Iflation, the DemandforAssets, and Seigniorage As figure 7.6 shows, the government relied heavily on borrowing from the central bank through money creation and seigruiorage in order to finance its deficit. This method had inflationary conse- quences Figure 7.9 illustrates actual seigniorage revenue in Ghana. The traditional definition of seigniorage revenue is the change in high-powered money (curency plus reserves) divided by cip. Clhamey (1991) argues that in economies that are subject to interest controls and high inflation the government effectively has a 100 per- cent reserve requirement and that in this case the expansion of M2 over cGDP is a better measure of seigniorage revenue. As expected, we see that the second definition of seigniorage revenue is lager than Roumeen Islam and Debormh Wetzel 323 Figure 7.9. Two Definitions of Seigniorage Revenue, Ghana, 1971-90 Percentage of GW 10 9 8 7 6 - 5 4- 3- 2 1971 1974 1977 1980 19S3. 1986 19891990 i On basis of M., On basis of H, high-powered money Snnre Calculated from World Bank data. the traditional measure, sometimes by as much as 5 percent of GDP. In general, however, the two measures follow similar patterns. (See Chamley 1991, pp. 6, 19, for a review of financial taxation in countries that are subject to financial controls.) THE MODEL. In order to evaluate the effect of inflation on seig- niorage, we use a simplified version of the Easterly, Rodriguez, and Schmidt-Hebbel (1989) framework. The equations of the model are set out in equations 7.1-7.6.1" Equation 7.1 is the budget identity which states that the government deficit must be equal to the sum of external finance (EF), domestic credit to the government from the central bank (0Cd, and government borrowing from the banking system (L3). External finance is assumed to be exogenous .'u Equations 7.2 and 7.3 are the portfolio demands of the private sector, which determine the private sector's demand for currency and the allocation of remaining assets between quasi money and foreign currency. 324 Ghana: Adjustmen2t, Reforn, and Growth (7.1) DEF EP* + Dcs + 4 (7.2) M fir, ds Y) (7.3) QM = g(r, id, Y) where vr is the rate of inflation, id is the real interest rate on deposits, and Y is real income. Dots represent the change per unit of time. The demand for foreign currency is determined as a residual. (7.4) LS = 1(- r)( - c)M1 + (1- r)QM] - ip (7.5) DCg = H-NFA-NOL (7.6) H = rQM + k + [+ -C)]M,. Equation 7.4 defines loans from the banking system to the govern- ment, which is the residual after private credit needs have been met; Lr credit to the private sector, is taken as given. The reserve ratio on deposits is r, and z is the currency-to-M1 ratio. Equation 7.5 defines do- mestic credit to the goverment from the central bank as high-powered money (H) minus net foreign assets of the central bank (NFA) minus any additional net liabilities of the central bank (NOL). High-powered money, equation 7.6, is equal to reserves on deposits and currency. THE DEMAND FOR MONEY AND QUASI MONEY. The basic equation used to estimate money demand is: Lh& = a0 + ail7t + asdt + 3Y't + e where LM equals the log of real M1, 7r is the CI inflation rate, idt is the real interest rate on twelve-mont deposits, Y is the log of real GDP, and e is the residual. The equation was estimated for the 1960-90 period. Dummy vari- ables were indludecl for the currency conversion in 1979 and for the onset of the ERP. Because these esfimates indicated the presence of autocorrelation of the errors, a lagged dependent variable was incor- porated into the equation. The results appear in table 7.2. The short-rin semielasticity of money demand with respect tco infla- tion is -0.28; the long-ran semielastidty is -0.77. As expected, the controlled real deposit rate has no significant effect on money demand. Real income is not significant. The currency conversion in 1979 had a negative impact on money demand, but the EBRP has had no significant impact. The lagged dependent variable is highly significant. Demand for quasi money is based on a similar equation but also incorporates the U.S. real rate on treasury bflls adjusted for black- market exchange rate depreciation. Only the lagged dependent vari- able and the cm were found to be significant. The short-run elasticity of quasi-money demand with respect to inflation is -0.60, and the long-mn elasticity is -4.1. Roumeen Islam and Debomrh Wclzel 325 Table 7.2. Regression Results, Money and Quasi Money, Ghana Variable n2) ) Quasi monoey -a 1.95 1.61 1.63 (0.89) (0.77) (3.13) al(1r) ~~-0.41 - -0.28 -0.60 (-1.07) (-2.96) (-8.00) a2(id) -0.29 (-0.35) *3(Y) 0.14 0.21 (0.70) (1.18) a4(DuM79) -0.24 -0.23 (-1.87) (-1.87) ar5(DUM83) -0.09 (0.70) ao6[Ml(-l)] -0.67 -0.64 (4-18) (-4.79) a7[QM(-1)J 0.85 (16.32) R2. 0.8547 0.8514 0.9260 F-statistic 22.54 35.82 168.86 Note: Figures in parentheses are t-statistics. A blank denotes amission of the specifc variable from the regression. THE EFFEcr OF INFLAIION ON SEICNIORACE. We use the model set out above to assess the impact on seigniorage revenues of the rela- tionship between ination and money demand.13 The base case and three scenarios are considered. Figure 7.10 illustrates the base case that simulates seigniorage reve- nues, given the actual rate of inflation, and the case in which inflation is 20 percent lower than its actual rate. For most years seigniorage revenue would have been higher had inflation been lower. This is particularly the case in the late 1970s. Figure 7.11 shows our base case with inflation maintained at 30 percent throughout the period. In some years, particularly the years of very high inflation (for example, 1981 and 1983), seigniorage reve- nue would have been considerably higher had the inflation rate been about 30 percent. After 1986, however, the revenues of the two con- verge. This is explained by the fact that during these years inflation was actually about 30 percent. The final scenario (figure 7.12) shows that had the currency conver- sion measures of 1979 not taken place, seigniorage revenues would have been higher, on the order of 1 to 5 percent per year. Thus, while 326 Ghana. Adjustment, Reform, and Growth Figure 7.10. Seigniorage Revenue: Base Case and with Inflation 20 Percent Lower than Actual Rate, Ghana, 1977-90 Percentage of GDP 10- 9 A 7- 6- 5 4 3 - 2- 1977 1980 1983 1986 1989 1990 - Base case i Inflation 20 percent lower an actual SoUtrce: Calcilated from World Bank dab. the currency reform had a one-time negative effect on inflation, it also reduced seigniorage revenue. Continued deficits and the need to finance them led to money aeation after 1979 and also to higher inflation. Using the long-run money demand equation, we have calculated the inflation-tax Laffer curve (figure 7.13). This figure shows the seig- niorage revenue received forgiven inflation rates; as inflation reaches about 125 percent in Ghana, seigriiorage revenue as a function of COP begins to decline.?4 In most years inflation in Ghana has been below this rate. THE SUSTAINABIUTY OF THE DEFICIT. In recent years the public sec- tor defict has stabilized at about 4 percent of GDP (using our second measure, based on financing flows to the public sector). To assess the sustainability of the deficit given the government's macroeconomic objectives of lower inflation and higher growth, we follow van Wijn- bergen and others (1992). We start with a version of the basic equation used in the analysis of the previous section: PD +IiB+Ei*F*Ik+b+EP* Roumeen Islam and Deborah Wezed 327 Figure 7.11. Seigniorage Revenue: Base Case and with Inflation at 30 Percent, Ghana, 1977-90 Percentage of CDP 10 9 7 6 4 3 2 I 1977 1980 1983 1986 1989 1990 Base case -I- Inflation at 30 percent Source: Calculated from World Bank data. where PD is the nominal primary deficit of the total public sector, i is the domestic nominal interest rate, B is the domestic public debt stock, E is the nominal exchange rate, i* is the nominal foreign inter- est rate, F* is the stock of foreign debt in foreign currency units, H is total base money, and D is credit to the government from the banking system- Each component of the equation is expressed as a ratio to GDP. Dots indicate absolute changes per unit of timne, and hats indi- cate relative changes. The equation can be rewritten as: PD + (r + P)B + i*F=f + H(P + Yt) + + B(P + fY) + P* + Ft(-e + P* + i) The sustainable deficit is based on the condition of nonincreasing public liabilities-a condition in which k = 0 = B = P. After imposing this condition and rearranging, the expression for the sustainable primary deficit is: PD-H(P + f) + B(% -r) + F*jY'-r*-a) 328 Ghana: Adjustment, Rejbnn, and Growth Figure 7.1. Seigniorage Revenue, Base Case and without the Currency Conversion, Ghana, 1977-90 Percentage of GDP 12 11 10 9 _ 8 7- 6- 5 4- 3- 1977 1980 1983 1986 1989 1990 -a-_ Base case i Without currency conversion Source: Calculated from World Bank data. where r and t are real domestic and foreign interest rates and e is the real exchange rate. Using the ratios of GcP for the above liabilities for 1989, table 7.3 sets out the sustainable deficit for the base case and three scenarios: lower inflation, high growth, and low growth com- bined with higher interest rates and a depreciation. In the base case, growth is 5 percent and inflation is 25 percent. Noniinal interest rates (calculated implicitly, using interest payments and the total stock of domestic and foreign debt) are 26 percent for domestic debt and 8 percent for foreign debt. The real exchange rate remains constant. In the base case the sustainable primary deficit is 278 percent of GDP, and the sustainable total public sector deficit is 5.32 percent of GDPA15 If we reduce inflation to 15 percent and all other variables (with the exception of real interest rates) remain the same, the sus- tainable primary deficit is 1.72 percent of GDP, and the sustainable total public sector deficit is 4.26 percent of GDP. With higher growth (8 percent), the sustainable total public sector deficit rises to 6 percent of GOP. Roumeen Islam and Deborah Wetze 329 Figure 7.13. Seigniorage Revenue for Given Inflation Rates, Ghana Percentage of GDP 8- 7~~ _- 7- 6- 5- 4- 2- 0 20 40 60 80 100 120 140 160 180 200 Inflation rate (percent) Source: Authors calculations. Finally we consider a scenano in which growth is reduced to 2.5 percent of GDP, inflation remains at 25 percent, both domestic and world interest rates inaease, and a 5 percent depreciation of the real exchange rate occurs. In this case the sustainable total public sector deficit is reduced to 4.52 percent of Grp. Given the results in table 73, the current deficit, 4 percent of GDP, appears to be sustainable. The low-inflation scenaio constrains the sustainable deficit to 4.26 per- cent of GDP, but higher growth allows it to increase. Fiscal Deficits, Private Consumption, and Private Investment Since the early 1960s pnrvate consumption has risen substantially as a share of GDP (see figure 7.14). In 1983 private consumption was 91 percent of GDP? As economic conditions worsened during the 1970s, more and more effort went into subsistence and providing for the present at the expense of saving and investment for the future. Since 1983 private consumption has fallen, but it still remains high. Private investment has been low even by developing country standards; in 330 Ghana: Adjustment, Reform, and Growth Table 7.3. Sustainability of the Public Sector Deficit, Ghana Base Loner High E mIt Item e025 iflation growth deprecation CDP growth ) 0.05 0.05 0.08 0.025 Domestic inflation (P) 0.25 0.15 0.25 0.25 Domestic nominal interest rate (i) 0.26 0.26 0.26 0.30 Domestic real interest rate (r) 0.01 0.11 0.01 0.05 Foreign noniinal interest rate (i*) 0.08 0.08 0.08 0.10 Foreign inflation (PW) 0.05 0.05 0.05 .0.05 Foreign real interest rate (rt) 0.03 0.03 0.03 0.05 Real exchange rate depreciation (e) 0.00 0.00 0.00 0-05 Inflation tax (HP) 0.02105. 0.01263 02105 0.02105 Seigniorage (HF) 0.00421 0.00421 0.006736 0.002105 Domestic debt effect (B(Y - r)] 0.000932 -0.00139 0.001631 -0.00058 Foreign debt effect [F )- 0.001804 0.001804 0.00451 -0-00225 Foreign debt capital gain (-F) 0.00 0.00 0.00 -0.00451 Sustainable primary defict 0.027996 0.017246 0.033927 0.01 80 Interest on foreign debt (i*f) 0.023452 0.023452 0.023452 0.02706 Interest on domestic debt (ib) 0.001864 0.001864 0.001864 0.00233 Sustainable public sector deficit 0.053312 0.042562 0.059243 0.045197 Note Public sector liabiities as a percentage of GDP, December 1989, are as follows: total base money, 0.0842; net foreign debt, 0.0902; net domestic debt, 0.0233. Source: For public sector liabilities listed in note, see appendix 7.3. the period 1965-88 private investment as a share of GDP peaked at about 10.5 percent (figure 7.15). The peaks and troughs have coTre- sponded to years in which public investment was very high or there was a coup or change in government. Even after the implementation of the ERi, private investment levels remained low. Fiscal Deficits and Piuvate Conswnption The equation estimated in order to assess the impact of fiscal deficits on private consumption is: Ct = a0 + cx1Ypt + cr2Cpubi + Se35pub, + c4Sfort + a59rf + c4LCt + e Roumeen lam and Deborah Wefze 331 Figure 7.14. Private Consumption, Ghana, 1961-90 Percentage of CDr 91 ! 90- 88- 86- 84- 78- 76 - 74- 722 1961 1966 1971 1976 1981 1986 1990 Source World Bank data. where C, is current private consumption, Yp, is current disposable income defined as gross national product (ONP) minus taxes, Cpub, is public consumption, Spub, is public savings (the central govement fiscal balance is used as a proxy), Sfort is a measure of foreign savings (exports minus imports) in the economy, r, is the real lending rate, and LCt, liquidity constraints, is proxied by credit to the private sec- tor. All variables other than r are represented as percentages of GDP. The results of our specifications are set out in table 7.4. All variables other than the real interest rate and the liquidity constraint are found to be significant. As previously noted, financial markets are weak,. and Ghanaians generally do not borrow to finance consumption. It is thus not surprising that these two factors are not significant. The second equation in table 7.4 drops the hisignificant variables. The relationship between current private consumption and current dis- posable income is almost one-to-one, which supports the Keynesian hypothesis concering private consumption- In low-income econ- omies, where credit markets are far from perfect, it is unlikely that most individuals will have substantial savings that act as buffers dur- ing low-income periods. Hence, current private consumption is tied dosely to cunrent disposable income. 332 Ghana: Adiustment, Refarm, and Growth Figure 7.15. Private Investment, Ghana, 1965-88 Percentage of CDP 10 9 8 7- 6 5 4 3 2- 1965 1969 1973 1977 1981 1985 1988 Source World Bank data. Public consumption is negatively related to private consumption; during the znid-1970s, when govenmuent consumption was very high, private consumption was at its lowest. This implies that some of the government's expenditure was passed on to the private sector in the form of goods or services. As this government provision stopped, private sector consumption rose to make up for the lost goods and services. Public savings is also shown to be significantly related to private consumption. If public savings increases-for example, if the deficit goes down-private consumption increases. The results imply that deficits do affect private consumption decisions and that Ricardian equivalence does not hold.16 Foreign savings, defined as exports minus imports, has a negative effect on private consumption. This relationship captures the impact of import controls on private consumption, as well. Reducing imports through the use of controls improves the trade balanoe and (assuming tiat exports remain the same) reduces private consumption. The import regime in Ghana tended to control consumer goods more than capital goods, thus squeezing private consumption. Roumeen Islam and DeboYUh Wetzel 333 Table 7.4. Regression Results, Private Consumption, Ghana Va rfS - EQNI EQN2 to 11.43 0.966 (0.77) (0.09) at(Yp) 0.88 0.965 (6.07) (9.70) az(Cpub) -0.55 -0.416 (-286) (-2.35) a3(Spub) .0.117 0260 (1-97) (3.31) a,(Sfrr) -0.85 -0.837 (-6-29) (-6.83) ct5(r) 0.02 (L09) a6LC) -0.29 (-1.68) R7 - 0.9751 0.9689 Durbin-Watson 2.27 2.02 Not Rigures in parenthweses are t-sztistics. A blank denotes omission of the specific variable from the regresston. In summary, public consumption over the years has substituted for private consumption. In the 1970s, when government expenditues were high, private expenditure remained relatively low- As the gov- emnment tried to reduce its deficit by controlling public expenditure, private consumption rose sharply. Fiscal Deficits and Private Invesm nt The investment equation that we have estimated for Ghana seems to confirm what we would expect from our analysis of Ghana's econ- omy. In the equation, the flow of credit to the private sector was used as a proxy for liquidity constraints faced by investors- As table 7.5 shows, public investment, corporate tax revenues, and the liquidity constraint variable are all statistically significant. The negative coefficent for public investment seems to indicate that pub- lic sector investment in Ghana has mostly substituted for private investment. This result is not surprising since, traditionally, the gov- erriment's policies have not encouraged private investment. Using corporate tax revenues collected by the government as an indicator of the tax burden faced by private firms, we find that this indicator has a positive, rather than the expected negative, sign. 334 Ghant Adjustment, Reform, and Growth Table 7.5. Regression on Private Investment, Ghana Item Coefficient Constant -0.002 (-0.10) PBICDGD -1.102 (-2.76) 0CDC;GDP .0.557 (2-42) RNT3 -0.0001 (-0.32) crrcoco 3.974 (2.27) RZ 0.561 Durbin-Watson 2.04 Note: Figures in parentheses am t-statistics. Since corporate revenues are correlated with profits we can think of these revenues instead as a proxy for profits rather than taxes. Unfor- tunately we do not have separate figures for corporate pxofits and cannot isolate their effects on investment. The flow of credit to the private sector has a positive coefficient, as expected. This result tells us that one way to encourage higher levels of private investment in Ghana would be to ease the economywide credit ceilings or to reallocate credit from the government to the pri- vate sector. As noted at the begining of this chapter, the government has controlled the allocation and overall level of credit since the early 1960s. As expected, the real interest rate, which has been negative, does not have a substantial effect on private investment.17 Using the equation in table 7.5, we conducted some counterfactual simulations to determine the effect on private investment of (a) vary- ing the supply of aedit to the private sector and (b) reducing public sector investment. Figure 7.16 shows actual and simulated ratios of private sector investment to GDP. In the first simulation, public sector investment in real terms is held at 1 percent of GDP (INvSIMI); in the second, private sector credit (iNvsnm2) is increased to 20 percent of GDP. In the first case investment as a percentage of GDP would have been 12.5 percent rather than 5.5 percent in 1988! From 1978 to 1983 private investment as a proportion of GDP would still have been rather low, but in most other years there would have been a definite improvement During 1978-83 the black-market premium was very high and rising, and the parallel market was flourishing. It is possible Roumeen istam and Debonh Wetzel 335 Figure 7.16. Pnrvate Investment Simulations, Ghana, 196748 Percentage of rCDP 14 13 12 10 9 8 :7 6 5 4 1967 1970 1973 1976 1979 1982 1985 1988 -X-- Actual -i---=-Simulation 1 INsI1.) S--imulation 2 (INVSIM2) Source: Estimated from World Bank data. that during this period legal economic activities were diverted to the unofficial economy and that the low investment ratios may not really be indicative of overall private investment activities in the economy.- When the supply of credit to the private sector is raised to 20 percent of cDP and maintained there, we find that investment would have been a much higher proportion of CDP in all years. In 1983, for instance, 12.1 percent rather than 2.9 percent of GDP would have been accounted for by private investment. For the period 1977-85 availabil- ity of credit seems to be the constrainig factor, during these years even lowering public sector investment does not have a great effect on private sector investment. Fiscal Deficits and the External Sector This section examines the implications of Ghana's fiscal deficits for the real exchange rate, the black-market premium, and the trade bal- ance. The fixed exchange rate system, in conjunction with high fiscal 336 Ghana: Adjustment, Reborm, and Growth deficits and strict capital controls, led to the development of a large black market in foreign exchange in Ghana. This section incorporates models of smuggling and of the parallel market in foreign exchange. It draws on studies by May (1985), Lizondo (1987a, 1987b), and Pinto (1989) to analyze the effect of the fiscal deficit on the extemal sector of Ghana's economy. Tfhe Parallel Marketfor Foreign Exchange Ghana's paraIlel market for foreign currency has been growing sub- stantially since the mid-1960s with the increase in cocoa smuggling. The decline in export earings during the late 1970s and early 1980s, coupled with strict foreign exchange controls, exacerbated the situa- tion (see, for instance, May 1985). Figure 7.17 shows the evolution of the average and end-of-period black-market- premiums from 1969 to 1987. In 1983 the black-market exchange rate was 76-58 cedis to the U.S. dollar, but the official rate was 275 cedis to the dollar. Table 7.6 presents estimates of the percentage of total cocoa smuggled out of Ghana during 1960-82 and the relative size of the parallel economy with respect to the formal economy. Estimates of the demand for currency in the black market show a steady increase during these years. During the decade before the ERP w.-is initiated, when the public sector deficit was rising, the offical trade balance sank from a surplus of US$212.9 million in 1973 to a deficit of US$60.6 mfllion in 1983. Figure 7.18 shows a positive trade balance and a negative fiscal bal- ance in years of surplus. Another important factor in the worsening of the trade balance was the deterioration mi Ghana's terms of trade, which amounted to about 47 percent between 1973 and 1983. In addi- tion, an overvalued exchange rate, combined with weak demand for Ghana's manm export, cocoa, led to a dramatic decline in export eam- ings, from US$1,066 mfflion in 1979 to US$439 million in 1983. Between 1983 and 1986 the cedi was devalued from 2.75 cedis to the U.S. dollar to 90 cedis to the U.S. dolar. A dual exchange rate system existed, with some transactions made under the fixed-rate system and others at government-managed auctions. This system was meant to, and actually did, reduce the spread between the official and paral- lel market rates- During 1983-86 exports grew 76.1 percent in U.S. dollar terms, while the public sector deficit fell from US$476 million to US$72 mril- lion. The trade balance improved from a deficit of US$60.6 million to a surplus of US$60.9 million, and the black-market premium fell: higher fiscal deficits had also been associated with higher black- market premiums. The black-market real exchange rate depreciated Roumeen Islam and Debonzh Wetz! 337 Figure 7.17. Average and End-of-Period Black-Market Premiums, Ghana, 1969487 Premium, log scale Rb - e)(el 4.0 3.5 3.0 2.0 1.5 05 0 1969 1972 1975 1978 1981 1984 1987 -a--Average premium -I-End-of-period premiumu Source: World Bank data. during this period (figure 7.19).38 The model on which our estima- tions are founded (developed in detail in appendix 7.1) i-S based on papers by May (1985), Lizondo (1987a, 1987b),, and Pinto (1989). THE monaE. As in Pinto (1989), the governmnent purchases foreign exchange from exporters at a fixed rate of e cedis to the U.S. dollar and sells part of its foreign exchange to importers at rate e. The remainder of the foreign exchange is used to finance government consumption. With a black-market foreign exchange rate of b cedis to the U.S. dollar, the marginal cost of foreign exchange is b; note how- ever, that the government obtains foreign exchange at rate e. Thus the official exchange rate e acts as a conduit for real income transfers between the government and the private sector-in particular, exporters. When the government is a net buyer of foreign exchange (as in Ghana), the black-market premium acts as a source of revenue for the government. If the implict tax on exports is lowered (the official exchange rate is devalued so that it is brought closer to or Table 7.6. Cocoa Production, Smuggling, and the Parallel-Market Economy, Ghana, 1960161 through 1982 Cocoa prodiuctio Smtuggled cocoa Illegal money Paraltel-market econiony (thowsands of (ithousands of (mlfo llos Millions As prcenitage Year metric to?s) metric tons) Year of ced(s) of cedis of GDP 1960/61 430 10 1961(62 409 8 1962/63 413 14 1963164 428 11 1964165 538 14 1965166 401 17 1965 0.01 0.08 0 1966167 368 17 1966 1,66 9.71 0.64 1967/68 415 21 1967 1.22 7.64 0.51 1968/69 323 17 1968 0.87 5.71 0.34 1969/70 403 25 1969 1.23 8.50 0.42 1970171 413 31 1970 0.96 7.15 0.32 1971/72 454 37 1971 1.37 10.72 0.43 .1972173 407 42 1972 1,27 7.73 0.27 1973i74 340 34 1973 1.05 6.53 0.19 1974175 376 30 1974 3.65 24.54 0.53 1975176 396 38 1975 8.61 45.47 0.86 1976/77 320 40 1976 15.74 72.68 1.11 1977178 271 45 1977 118.74 582.96 5.22 1978179 265 50 1978 282.71 1,543.73 7.36 1979 483.15 3,243.21 11.51 1980 1,195.62 10,024.37 24.45 1981 1,313.16 12,427.07 16.21 .__________________ _ __1982 2,741.99 27,827.27 32.41 Soulrce: May 198!i. Roumeen Islam and Deborah Wetzel 339 Figure 7.18. Trade Surplus and Fiscal Deficit, Ghana, 1970-87 Millions of US. dollars 1.3 - 1.2- 1.0- 0.6- 0.4- 02 0- -0.2 }0.3 1970 1973 1976 1979 1982 1985 1987 Trade surplus Fiscal deficit Souirce: World Bank data.. equals the marginal cost of foreign exchange), total exports wil increase. However, the basic analysis in this section regarding the effects of higher government expenditure on the black-market pre- mium is valid both when the government is a net purchaser of foreign exchange and when it is a net seller. This is because in this model an increase in govemment expenditure, other sources of finance being given, must be financed by money creation. An increased stock of money will be held only at a higher premium, regardless of whether the government is a net purchaser or seller of foreign exchange. (The formal analysis is more complex in the net-seller case.) The govern- ment also gets revenues from conventional taxes, t, which are fixed, and foreign aid to the government, A. The inflation tax finances the residual requirements of the government net of the implicit tax on exports and net of conventional taxes and foreign aid. In this formulation, an increase in public sector expenditure increases reliance on the inflation tax. We assume that the inflation tax is given by tnmt where 7r is the rate of inflation and m = Mle. (M is nominal money balances.) At the steady state the rate of inflation 340 Ghana: Adjustmnent, Reform, and Growth Figure 719. Official and Black-Market Real Exchange Rates, Ghana, 196987 Cedis per US. dollars 220- A - -- 200 180 160 140 120 :100 60 40 20 1969 1972 1975 1978 1981 1984 1987 -a-Real exchange rate F- Black-market exchange rate Source. World Bank data. equals the official rate of depreciation of the currency (Aele). The government s budget corstraint at the steady state is: l;7J) (gT - t-A) = m Aele where gT is total govermnent expenditure in dollars, which is fixed and exogenously given and includes expenditure on both tradables (g3) and nontradables (gN). That is, g7 glNPIe + g, g = ZgT z < The real fiscal deficit is defined as gT- t. Seigniorage from the infla- tion tax is equal to Ae/em, at the steady state. Given that the marginal cost of foreign exdhange to the private sector is b and that the private sector's loss is the governmnent's gain, we can rewrite the preceding equation ns: gT = Mlb Aele + et/b + g(b-e)Ib + eAb (7.8) =mAeI) i-th6+g(1-I/r)+Al Roumeen Islam and Deborah Wetzel 341 (see Pinto 1989), where X is the black-market premium, elb = ti is the real tax burden on the private sector, and eA/b = Ail is tle real value of aid flows to the private sector, and similarly for the expres- sions involving government spending and the inflation tax. The expression gT(b - e)lb represents the implicit tax on exports (Pinto 1989). The government sets e arbitrarily, and since it does not have reserves to deplete in maintaining this exchange rate, the official exchange market is rationed by capital controls and restrictions on commercial transactions. As in Lizondo (1987a, 1987b), we assume that there is no official net foreign asset accumulation by the govem- ment; thus It = 0, where Pt is the change in official reserves. The change in the stock of money is given by: (7.9) =P+b where b) is the change in domestic credit and S{ is the change in the nominal money supply. Production, exports, and imports. The model is presented in detail in appendix 7.1. Agents in the economy produce exportables (X) and nontradables (N) and trade in both the official and black markets. For smuggled exports and imports, foreign exchange is valued at its mar- ginal cost, b. The higher is the black-market premium, the greater are the benefits from smuggling. However, the greater is the volume of smuggling, the greater are the risks of detection and penalties. Thus exporters and importers allocate trading between the official and unofficial markets until the marginal benefits equal the marginal costs, which indlude the costs of bnibes to evade detection and penal- ties. We assume that all markets are perfectly competitive. Balance of payments. We assume that agents in the economy hold only non-interest-bearing assets. The balance of payments in the economy is shown in equation 7.10: (7.10) =% X- I -& where P is the total accumulation of foreign assets. Consumption. We assume that agents consurme only nontradable goods and that they consume a constant fraction y1 of their wealth.19 Wealth is defined as the sum of domestic and foreign currency: (7.11) M+bF=W. Therefore consumption spending is given by: (7.12) PNCN = (M + bF). Money demand. Money demand (M) is given by equation 7.13: (7.13) Md =X(b/b)W X< 0 where X is a decreasing function of the expEcted rate of depreciation in the black market (Lizondo 1987a, 1987b). Assuming that agents 342 Ghana: Adjustment, Refonn, and Growth possess perfect foresight, the expected and actual rates of inflation are equal and, at the steady state, are equal to the rate of depreciation of the domestic currency. Assuming that private agents can adjust their portfolio composition to the desired level instantly, equation 7.13 can be rewritten as: (7.14) m + tie) ,F 1- XC(@J + tie) where m is Mle and q is ble. THE STEADY sTuAE. The dynamic equations of our system after sub- stitution are: (7.15) P = L,- L1(, Px, gP, -zgT - I(m F, zg7) C7.16) m =S -t-em -A and equation 7.14. Equations 7.14-7.16 can be solved for the steady state of the economy. The steady-state solution for the black-market premium is: (7.17) * = 0(gT, t, e, A) 'kg, > n *tcO o rk c O < °- The derivative of the black-market premium with respect to gov- erinent spending is of particular interest: an increase in goverrment expenditure will raise the black-market premium. This is because an inaease in government consumption reduces the supply of doUars available for the private sector and the black-market price of foreign exchange (which is the marginal cost of foreign exchange) wfll be pushed up. An increase in conventional tax revenues or aid flows will reduce the premium. This is because the higher the revenues from taxes or aid, the lower will be the reliance on inflationary financing and the lower will be the money stock. To restore portfolio balance, the premium must fall. An improvement in the terms of trade will inaease the supply of dollars to the private sector and therefore lower the black-market premium. An increase in the rate of official depredation has ambiguous effects: the direction of the effect depends on the inflation elasticity of the share of domestic money in wealth, N. There are two distinct effects of an increase in the expected rate of depreciation Ae/e. Sup- pose Ae/e is raised. The differential rate of return between cedis and U.S. dollars rises, making dollars more attractive. The greater demand for foreign exchange wil tend to raise the black-market pre- mium. However, another effect works in the opposite direction: with the real deficit given, a smaller cedi base would be required for the inflation tax. The steady-state inflation tax is given by aeIeM, where M is the nominal money supply. Suppose beIe rises but the inflation elasticity of X(n) is less than unity. Then total inflation-tax revenues Roumeen Islam and Deborah Wetzel 343 will rise, the supply of dollars will increase, and the premium will fall. If, however, ir is greater than unity, inflation-tax revenues will fall and the premium will rise. The effects of a devaluation. A devaluation, by raising the nominal exchange rate, will immediately lower the real money stock, MIe = m. At the orginal premium, there will be a portfolio imbalance; agents will have a greater proportion of foreign assets in their portfolio than they desire. Portfolio balance is restored because (a) the black-market premium falls immediately to reduce the share of foreign assets in wealth, and (b) as the premium rises to its new steady-state level, the rate of inflation is higher and the desired ratio of foreign assets in wealth (1 - X) inaeases.20 Thus a devaluation, given gT, will tempo- rarily reduce the black-market premium. The real exchange rate and the trade balance. The official real exchange rate (R), which depends on the black-market premium, is defined as the relative price of tradable to nontradable goods: ePZIPN. Substitut- ing for the black-market premrium, we get R = R(gT, Px, A, t, bde). An increase in government spending appreciates the real exchange rate through two channels: (a) an increase in spending that repre- sents partly an increase in spending on nontradables will raise the relative price of nontradables, and (b) an increase in government spending wil also raise the black-market premium. This again wi tend to make imported inputs more expensive and wil therefore raise the price of nontradables that make use of these inputs. An increase in the price of exports, Px, will depreciate the real exchange rate. An ncrease in the rate of depreciation wil lower the premriium because the inflation elasticity of the demand for money in Ghana is less than unity, and the real exchange rate wfll appreciate. An increase in aid flows or tax revenues lowers the pmium and thus the costs of imports and nontradables. This puts downward pressure on the price of nontradables and depreciates the real exchange rate. Note that at the steady state the trade balance is zero. If agents held interest-bearing foreign assets, private agents' accu- mulation of foreign assets would consist of the sum of net interest payments plus the trade surplus or deficit. P would be the current account. In this case the trade balance would not be zero at the steady state. Although it is useful to consider the steady-state determinants of the real exchange rate and the black-market premium, a look at the non-steady-state relationships between the premium, the trade bal- ance, the real exchange rate, and public sector expenditure is particu- larly useful, as in equations 7.18 and 7.19. The reason is that the analysis of the steady state assumes that the rate of inflation equals 344 Ghana: Adjustment, Rejbrm, and Growth the rate of official depredation. This obviously was not true for Ghana during the past decade and a half, when the official exchange rate was kept fixed, with some discrete devaluations, but the rate of inflation was very high. (7.18) - = mt, mF, z, P, grT). An increase in the stock of money will require an inaease in the black-market preminum to restore portfolio balance. An increase in the stock of foreign assets implies an increase in private sector wealth and therefore an increase in private sector consumption- The higher con- sumption demand raises the black-market premium. An increase in government expenditure raises the black-market premium, as before. An improvement in the terms of trade will have an ambiguous effect on the black-market premium. On the one hand, it will tend to raise the productior. of tradables im relation to nontradables. On the other hand, by raising wages, it will induce substitution toward imported inputs and away from labor in the nontradable sector, and imports wil rise. This may put upward pressure on the black-market premium, depending on the relative short-run responses of exports and imports, given the trade balance. An improvement in the trade balance is accompanied by a lower premiium. This is because an improvement in the trade balance implies a larger supply of foreign exchange in the black market. (7-19) t = F(PM, m, E z, gr)- The trade balance depends positively on the terms of trade. The reason is that an increase in the terms of trade raises export revenues at every level of exports and also raises the production of tradables by inducng agents to shift away from nontradable to tradable produc- tion. An increase in the stock of money could have either a negative or a positive effect on the trade balance. In the first case, a higher stock of money implies greater wealth for the private sector and therefore greater consumption demand for nontradables. This demand puts upward pressure on the relative price of nontradables, and agents shift production toward nontradables. Since imports are used in nontradable production, imports must rise, and the trade balance tends to deteriorate. In the second case, an increase in the stock of money held by the private sector, given the trade balance, is associated with a higher black-market premium. The higher premium reduces imports-and exports, but less so-and improves the trade balance. (This is the condition for saddlepath stability.) The result makes sense if we recall that the overall trade balance reduces to the unofficial trade balance, since by construction the official sector is always in balance. All other things being equal, an incease in public PRoumeen Itm and Deborah Wetzel 345 sector expenditure has a negative effect on the trade balance by rais- ing imports. The official real exchange rate can also be written in terms of the terms of trade, the trade balance, the stock of real money balances, the stock of foreign currency, and public sector expenditure. An improvement in the trade balance leads to an appreciation of the real exchange rate. This is because a higher trade surplus implies lower imports, higher exports, or both. If the lower imports or higher exports come at the expense of production of nontradables, the rela- tive price of nontradables wfll rise, causing an appreciation of the official real exchange rate. If total public sector expenditure increases, given that the share of imports in total expenditure is fixed, part of the increase wil fall on nontradables. This will lead to real exchange rate appreciation. An increase in the stock of real balances or of foreign currency wil increase wealth and therefore consumption. This will put upward pressure on the price of nontradables and will lead to real exchange rate appreciation. Some empirical obsernations. The effect of the devaluations and other policy changes that occurred in Ghana from 1983 on can be analyzed through this modeL As discussed before, a one-shot devaluation is expected to reduce the premium temporarily by lowering the real money stock instantaneously. Thus, as expected, the black-market premium fell in 1983. In subsequent years the exchange rate regime did not change significantly. There were a number of discrete deval- uations in 1983-87, tax collection methods improved, and aid flows to Ghana started to rise. In this way the government gradually switched to alternative means of financn a lower level of government expen- diture. All these events would, in this model, lower the black-market premium. This in fact happenedc the black-market premium started to fal in 1983 and continued to fall dramatically thereafter, as the fiscal deficit decreased (see figure 717). Unification. In 1987 the government undertook polcies aimed at unifying the official and black markets for foreign exchange. In view of the fact that the fiscal deficit had been falling, as had reliance on money financing, there was not a substantial differential between the black-market and official exchange rates. In fact, by 1989 the black- market premium had fallen dramatically. In the context of the current model, unification could lead to an increase in the rate of inflation under certain circumstances. The higher is the black-market premium before unification, the larger is the implicit tax on exports and therefore the more important is the black-market premium as a source of revenue for the government and the higher is the postunification rate of inflation. In Ghana's case 0 (7.24) Xi = L where P1 = world price of exports PI = world price of imports, normalized to unUty PN = price of rnontraded goods X = total exports I = total imports lo = imports through officid channels XO, = exports through official channels C1(X.,) = cost of smuggling exports C2(..) = cost of smuggling imports R(4) = cost of importing through official channels, induding rent- seeking Roumee -lJam and Deborah Wetzel 355 LI = labor employed in the production of nontradable goods. LO = labor employed in the production of exports going through official channels Lu = labor employed in the production of exports going through illegal channels. The cost functions have the following properties: (7.25) C"(Xu) > 0 C;(Xu) > 0 R'(4,) > 0 C(4) > 0 * ~~~C;(I.,) > O Re"(1 > °. The first-order conditions for the maximization problem are: (7.26) 515L1 = 0 a aPN N = WL1 (7.27) 5/I8 = 0 PN -N = [bl(1 -a)][1 + Cq(I - lIOl (7.28) IoCi (7.29) StSLu = 0 -eP. + bPW - b-(XC,X) 0 or XI= XU(+, PX) (7.30) 5/3L0 = 0* eP, = w. Balance of Payments The balance of payments equation in the text can be rewritten as: 17.31) F=P..X - I-g = Pj: L- Lo -I - gi. Equations 7.31 and 7.23, and first-order conditions 7.26 and 7.27 can be used to solve for L and I in terms of F, g, I, P., and 0: (7.32) L1L,(Px, g1. 1) (7.33) -I=I(Px, &, (AlF) The Real Exchange RRate and the Trade Balance The official real exchange rate, ePx,/PN can be rewritten as a(LtI-), since PN = ePLjIaN = ePJa(LjII)I-a. The black-market real exchange rate is defined as: (7.34) bPxIPN = (pxalepr)(la-1)II) = Oa[-L-4¶1)t] Using equation 7.15 and first-order conditions 7.26 and 7.27, we can express the trade balance as: (7-35) F = Px[ L, (0, P,w gTj]- z I(m, Fr z, gr). 356 Ghana: Adjustment, Reform, and Growth Appendix 7.2. Dynamics. The system of equations (t i, j) is linearized around the steady state: p -~a -ai _ _ -F-dm- ° _ _ m = 0 -e 0 dm aL1 aI p( - iz so x (1-X)4o ->0 (1 -X)2 > 0. For saddlepath stability, the determinant of the preceding matrix must be greater than 0. A sufficient condition for saddlepath stabilty is that , > . Appendix 7.3. Time-Series Data for Principal Variables, Ghana, 1961-90 Item, 1961 1962 1963 1964 1965 1966 1967 1968 1969 Aggregate varfables CDP (millions of current cedis) 932 997 1,101 1,237 1,466 1,518 1,504 1,700 2,001 Growth of real CDP 6.17 4.83 3,47 2.15 1.36 0.09 -3.00 6.43 5.88 Growth of GDP per capita 5.09 3.62 2.29 -3.23 -3.09 -2.06 -5.04 4.11 3.62 Infiationa 6.25 5.88 5.56 15.79 22.73 14.81 -:'.68 10.71 6.45 Nominal discount rateb 4.5 4.5 4,5 4.5 4.5 7,0 6.0 5.5 5.5 Nominal lending rateb 8.25 8.25 8.25 8.25 8.25 8.25 8.5 8.5 8.5 Nominal deposit rateb 3,25 3.25 3.25 3.25 3.25 3.375 3.375 3.375 3.375 Real discount rateb -1.65 -1.31 -1.00 -9.75 -14.85 -6.81 16.80 -4.71 -0.89 Real lending rateb 1.88 2.24 2.55 -6.51 -11.80 -5.72 20.13 -2.00 1.92 Real deposit rateb -2.82 -2.49 -2.18 -10.83 -15,87 -9.96 14.45 -6.63 -2.89 Official exchange ratec 0,71 0.71 0.71 0.71 0.71 0.71 0,86 1.02 1.02 Real exchange rateC 117.02 110.83 104.70 90.43 75.15 67.64 90.54 99.23 96.88 Real black-market exchange ratec 121.23 155.17 168.57 149.38 148.33 130.67 186.06 180.87 160.46 Black-market premiumC 1.04 1.40 1.61 1.65 1.97 1.93 2.06 1.82 1.66 (Table continies on thefollowing page.) Appendix 7.3 (continued) lten, 1961 1962 1963 '964 1965 1966 1967 1968 1969 Cornipositioti of outpiut (percentage of CDP) Private consumption 78.6 75.8 75.8 72.7 77.3 79.1 77.2 72.6 74.6 Public consumption 10.8 11.1 11,4 11.8 14.5 13.0 15.0 16.8 14.2 Gross fixed capital formation 20.5 16.8 18.0 17.1 18.1 12.9 11.6 11.0 9.8 Change in stocks -1.9 -1.0 -0.6 1.0 -0.2 -0.1 -1.3 0.1 2.0 Exports 23.9 21.9 19.4 18.2 17.1 14.6 17.4 20.3 19.7 Imports 31.9 24.7 24.0 20.9 26.7 19.6 19.9 20.8 20.3 Memoranda: Private fixed Investment 0.0 0.0 0.0 0.0 5.2 6.5 5.3 6.9 8.5 Public fixed investmentd 0.0 0.0 0.0 0.0 12.7 6.3 5.0 4.2 3.2 Monetary sumttrary (miltlionis of cedi, end of period)e Net foreign assets -6 -67 -77 Net domestic assets - 468 488 513 Claims on government - - 251 271 255 Claims on public entities - - . 25 25 47 Cocoa financing - - - - - - 128 126 120 Claims on private sector - - - - - - 64 67 92 Net other assets - - - - - 0 0 0 Revaluatio n account Ml (percentage of GDP) 14,66 15.53 14.06 17.69 16.37 16.23 15.96 15.17 14.42 Quasi money (percentage of cD,P) 2.93 3.65 3.31 3.68 4.09 4.43 5.20 5.48 4.95 Balance of paynetn ts (millions of u.s. dollars) Trade balance -52.2 9.8 -29.6 -0.1 -118.2 -40.4 18.8 39.8 49.7 Current account -103,5 -33.7 -80.1 -55.4 -222.9 -127.8 -84.9 -56,1 -60.1 Capital account 86,1 45.9 51.3 48.3 228.3 77.7 8.0 48.3 67.2 Errors and omissions -4.4 -8.1 -12,0 2.9 9.8 2.0 4.6 1.3 -34.0 Position above the line -21.8 4.1 -40.8 -4.2 15.2 -48.1 -72.3 6.5 -26.9 Total change in reserves 21.8 -4,1 40,8 4.2 -15.2 48.1 49.0 -1.5 18.6 Central governiment accounts (percentage of GDP) Central government revenue 16.41 15.69 19.35 19.27 15.16 15.98 18.75 15.36 16.92 Central government expenditure 25.91 25.3 24.96 24.66 17.68 21.63 23,1 19.35 20.6 Central government deficit -9.50 -9.61 -5.61 -5.39 -2.52 -5.65 -4,35 -3.99 -3.68 Domestic financing - - - 1.82 3.41 2.74 2.58 External financing 4.35 1.38 0.89 1.49 Other -0.52 -0.44 0.36 -0.39 (Table conltitines on tlhefollowing page withl years 1970-79.) Appendix 7.3 (contintued) temn f1970 1971 1972 1973 1974 1975 1976 1977: 1978 1979 Aggregate variables CDP (millions of current cedis) 2,259 2,501 2,815 3,501 4,660 5,283 6,526 11,163 20,986 28,222 Growth of real GDr 6.76 5.56 -2.49 15.25 3.39 -12.86 -3,52 2.26 8,48 -3.17 Growth of CDP per capita 4,65 2.58 -4.96 11.57 1.03 -15.16 -7,64 1.27 5.05 -0.67 Inflationa 3.03 8.82 10.81 17.07 18.75 29.82 55.41 116.52 73.09 54.52 Nominal discount rateb 5,5 8.0 8.0 6,0 6.0 8.0 8.0 8,0 13.5 13.5 Nominal lending rateb 8,5 13.0 13.0 10.0 10.0 12.5 12.5 12.5 18.5 18.5 Nominal deposit rateb 3.38 8,0 8.0 5.5 5.5 8.0 8.0 8.0 13.0 13.0 Real discount rateb 2.40 -0.76 -2.54 -9.46 -10.74 -16.81 -30.50 -50.12 -34.43 -26.55 Real lending rate' 5.31 3.84 1,98 -6,04 -7.37 -13.34 -27.61 -48.04 -31.54 -23.31 Real deposit rateb 0.33 -0.76 -2.54 -9.89 -11.16 -16,81 -30.50 -50.12 -34.72 -26.87 Official exchange ratec 1.02 1.03 1.33 1.16 1.15 1,15 1.15 1.15 1.76 2,75 Real exchange rater 97.80 93.55 114.35 96.05 95.36 80,24 53.80 26,38 25.19 28.62 Real black-market exchange ratec 166.20 137.77 143.73 123,84 143.12 138.71 135,95 211.06 127.90 161.96 Black-market prermiumc 1.70 1,47 1.26 1.) 1.50 1.73 2.53 8.00 5.08 5.66 COmposition of oiulput (percentage of GDP) Private consumption 74.4 77.4 74r8 75.0 78.2 73.3 79.2 77.4 84.7 83.1 Public consumption 12.8 13.0 12.6 10.9 12.2 13.0 12.2 12.6 11.3 10.3 Gross fixed capital formation 12.0 12.4 8.7 7.7 11.9 11.6 9.8 9.4 5.1 6.7 Change in stocks 2.1 1.7 -1.6 1,4 1.1 1.1 -1.0 1.7 0.3 -0.2 Exports 21.3 15.8 20.7 21.4 18.3 19.4 15.7 10.5 8.4 11.2 Imports 22.7 20,2 15.2 16,4 21.8 18.4 16.0 11.6 9.7 11,2 Memoranda: Private fixed investment 10.4 9.8 3.4 6.3 9.4 7.3 2.1 5.4 2.1 3.8 Public fixed investmentd 3.7 4.3 3.7 2.7 3.6 5.4 6.8 5.7 3,3 2.7 Monetary suimmnsary (Oiluions of cedi, enid of period? Net foreign assets -57 -149 -46 9E -91 49 -157 -149 -898 -529 Net domestic assets 514 701 799 839 1,215 1,560 2,106 3,229 6,056 6,540 Claims on government 295 283 349 372 570 903 1,584 2,780 4,500 4,906 Claims on public entities -12 108 107 122 196 101 124 141 234 290 Cocoa financing 97 152 170 119 175 188 259 271 867 1,340 Claims on private sector 127 175 174 207 294 365 395 572 751 799 Net other assets 7 -17 -1 19 -554 3 -256 -535 -296 -795 Revaluation account - - - - - - Ml (percentage of cDr) 13.35 12.71 16.31 15.30 14.08 18.76 21.96 21.37 19.59 16.59 (Table conlntines oS thefollowring page.) Appendix 7,3 (continued) 1e1m 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 Quasi money (percentage of GoP) 5.37 6,14 7.29 6.56 6.60 7.15 7.26 5.83 4.79 4,48 Balantce of paymtci;ts (inrillio,is of U.S. dollars) Trade balance 51.9 -33.6 161.4 212,9 -29.2 150.4 88.8 29.4 112.5 262.6 Current account -67.7 -145.8 108,2 126.7 -171,5 17.6 -74.0 -79.7 -45.9 122.0 Capital account 53.6 97.2 -80.6 -12.6 40.6 82t4 -36.5 59.0 103.0 19.8 Errors and omissions 16.6 14.0 5.4 -9.5 -14.1 6.3 -26.7 12.3 -119.7 -106.0 Position above the line 2.5 -34.7 33,1 104.6 -144,9 106.3 -137.3 -8.4 -62.7 35.8 Total change in reserves 8.0 -25.0 -77.7 -86.8 83.3 -10.6 59.7 -61.1 -117.5 51.9 Cenitral govenmtent accowdts (percen fagc of GDP) Central government revenue 20.43 16.36 12.39 14.31 16.18 13.79 12,15 9.57 8.89 8.52 Central government expendlture 19,68 22.10 19.02 19.38 25,89 28.98 24.82 20.49 17,65 13,47 Central government deficit 0,75 -5.74 -6.63 -5.07 -9.71 -15,19 -12.67 -10.92 -8.76 -4.95 Domestic financing -2.39 1.98 2.79 5,75 4.62 17.95 10.64 11.84 4,25 4.13 External financing 1.46 0.63 1.04 -0.34 0.05 -0.01 0.06 -0.04 -0.03 0.82 Other 0.18 3.13 2.80 -0,34 5.04 -2.74 1,97 -0.89 4.55 0.01 lieni 1980 1981 1982 1983 1984 7985 1986 1987 19888 1989s 19908 Aggregate variables GDP (millions of current cedis) 42,852 72,626 86,451 184,038 270,561 343,048 511,373 746,000 1,051,196 1,417,214 1,956,090 Growth of real cnr 0.00 -1.79 -7.20 0.70 2.64 5.09 5.20 4.80 5.60 5.11 3.00 Growth of GrDP per capita -2,33 -4.81 -10.43 -3.10 -1.26 2.37 2.54 2.14 0.66 1.34 0.38 Inflatlo, 50.15 116.50 22.31 122.85 39.65 10.31 24.56 39.81 31.40 25.20 37.20 Nominal discount rateb 13,5 19.5 10.5 14.5 18.0 18.5 20.5 23.5 26.0 26.0 33.0 Nominal lending rateb 18.0 18.0 18.0 19.0 22.5 23,0 23.0 26.0 30.3 30.3 30.3 Nominal deposit rateb 13.0 19.0 9.0 12.5 16.0 18.0 20.0 22.3 22.0 20.0 21.5 Real discount rateb -24.41 -44.80 -9.66 -48.62 -15.51 7.42 -3.26 -11.67 -4.11 0.64 -3.06 Real lending rateb -21.41 -45.50 -3.52 -46.60 -12.28 11.50 -1.25 -9.88 -0.84 4.07 -5.03 Real deposit rateb -24.74 -45.03 -10.88 -49.52 -16.94 6.97 -3.66 -12.56 -7.15 -4.15 -11.44 a Official exchange ratec 2.75 2.75 2.75 8.83 35.99 54.36 89.20 153.73 202.35 270.00 326.33 Real exchange ratee 21.77 10.97 9.15 13.35 39.89 54.37 69.53 87.96 91.61 102.53 93.56 Real black-market exchange ratec 150.40 104.73 205.09 115.79 107.14 131.25 144.12 124.74 114.08 119.24 Black-market premiumc 6.90 9.55 22.43 8.67 2.69 2.41 2.07 1.42 1.24 1.16 Comrtpostifott of ouitpitt (percentage of owp) Private consumption 83.9 87.2 89.8 90.8 86.1 83.0 81.2 81.8 79.4 83.7 85.2 Public consumption 11.2 8.8 6.5 8.6 7.3 9.4 11.1 10.0 10.0 10.3 8.1 Gross fixed capital formation 6.1 4,7 3,5 3.8 6.9 9.5 9.3 10.4 10.9 13.5 15.8 Change in stocks -0.5 --0.2 -0.2 0.0 0.0 0.0 0.1 0.1 0.1 0.1 0.1 Exports 8.5 4.8 3.3 6.1 7.5 9.7 16.0 21.2 20.7 20.6 16,7 Imports 9.2 5.3 3.0 9.3 7.7 11.6 17.7 23,4 21.0 28.1 25.8 (Table continiues oti t1le follozwing page.) Appendix 7.3 (continuled) Itcm 1980 1981 1982 1983 1984 1985 1986 1987 1988s 1989s 1990' Memoranda: Private fixed investment 4.2 2.7 2.3 2.9 4.4 5.4 2.0 2.5 2.9 5.6 8.4 Public fixed investment" 1.4 1.9 1.1 0.9 2.5 4,2 7.3 7.9 8.0 7.9 7.4 Monetary suniniary (Qnillions of cedi, end of period?e Net foreign assets -358 -1,160 -1,024 -16,676 -31,440 -48,702 -130,656 -141,707 -141,303 -127,903 -86,376 Net d ;1mestic assets 8,404 12,943 16,004 22,601 39,237 66,464 94,590 105,335 116.631 102,918 76,149 Claims on government 6,526 10,655 11,064 21,059 24,170 27,176 29,647 22,220 1i,008 -10,454 -34,012 Claims on public entities 400 441 527 813 676 4,795 5,274 8,610 10,422 19,674 24,592 Cocoa financing 1,559 2,950 5,553 521 3,580 13,545 16,889 16,471 21,000 23,747 17,058 Claims on private sector 943 1,345 1,562 2,841 12,153 21,178 37,455 46,946 58,183 61,140 80,218 Net other assets -1,024 -2,448 -2,702 -2,633 -1,342 -230 5,325 11,088 16,018 8,811 -11,707 Revaluation account - - - 16,158 29,930 41,884 133,135 179,768 224,990 276,189 332,332 Ml (percentage of CDP) 14.20 12.96 12.96 8.94 10.12 12.81 12.87 12.74 13.23 13.06 11.09 Quasi money (percentage of cDIp 4.32 3.06 4.20 2.22 2.68 3.40 3.93 4,99 4.77 3.88 3.39 Balance of paynsents (ntilhlionls of U.S. dollars) Trade balance 195,3 -243.6 18.3 -60.6 32.9 -36.3 60,9 -124.7 -112,4 -191.8 Current account 29.2 -420.8 -108.6 -174.1 -38.8 -134.2 -43.0 -96.9 -65.8 -97.5 Capital account 40.9 108.1 123.6 119.1 206.9 84.9 63.4 255.7 209.0 236 Errors and omissions -100.4 24.0 -32.8 -125.8 -132.5 63.4 -81.2 -18.7 37.9 17.1 Position above the line -30.3 -288.7 -17.8 -180.8 35.6 14.1 -60.8 140.1 181.1 155.6 - Total change in reserves 93.4 -16.1 -23.8 246.4 7.3 54.8 24.1 .394.0 -76.2 -127.4 Cenltral government accoil(is (percenitage of GDP)f Central government revenue 5,68 5.72 5.37 5.58 8.37 11.75 14.40 14.89 14.63 15.14 13.68 Central government expenditure 13.76 11.57 11.72. 8.27 10.16 13.96 14.34 14.34 14.26 14.41 13.51 Central government deficit -8.08 -5.85 -6.35 -2.69 -1.79 -2.21 0.06 0.55 0.37 0.73 0.17 Domestic financing 7.32 7.06 5.65 2.16 1.12 1.18 1.04 -0.39 -0.59 -1.08 -1.39 External financing 0.83 0.00 0.25 -0.53 0.67 1.03 -1.10 -0.16 0.21 0.35 1. 22 Other 0.00 -1,20 0.45 - - - - - - Not available. a. Annual change In consumer price Index; b. The discount rate is the Central Bank rate, the lending rate used is that for unsecured loans, and the deposit rate used is that on twelve-month deposits. Real rates are calculated as t(1 + i)1(1 + p) -13100, where i Is the nominal Interest rate and p is the cPI inflation rate. Interest rate data are from the & Quarterly Digest of Statistics (QDS) and World Bank data. c. The official exchange rate given Is cedis per U.S. dollar (period average). The real exchange rate is calculated as the cedildollar rate multiplied by the U.S. wholesale price index (1985 - 100) and divided by the Ghanaian cpi (1985 = 100). A fall In the real exchange rate indicates an appreciation, and a rise indicates a depreciation, The real black-market exchange rate Is the black-market cedis per U.S. dollar rate (from Pick's currency yearbook) multiplied by the U.S. wholesale price index and divided by the Ghanaian cri. The black-market premium is the ratio of the black-market rate to the official rate. d. Data prior to 1965 are from the 1958 System of National Accounts. Prior to 1981, estimates of private and p ..lic fixed investment are fromn Huq (1989); thereafter they are based on World Bank estimates, e. Monetary data are from World Bank data and include secondary banks after 1983. f. Data for 1961-81 are for fiscal years (that is, 1962 represents the 1962163 fiscal year). Data for 1961 and 1963 represent fifteen-month periods. Data for 1961 through 1965 are from Killick (1978), p. 150. Data for all other years are from World Bank data and the Quarterly Digest of Statistics. g. Data for 1988 through 1990 are subject to revision, Appendix 7.4. Central Government Revenue Lost as a Result of Smuggling, Ghana, 1960161 through 1981182 Estimated total Official Reveniue to Revente lost sitniggled cocoa FOB price CMB costs produter pnrce govenmtetit lthrough swuggling Revenue lost Fiscal (thousands of (lnew cedis (iew cedis (nero ceais (nev cedis (thorusands of tirwugh i1niugglilig year metict bos) per ietric toil) per metric toil) per metric tou) per metric tOn)' inetnc lonsS)b (percethage of GDP) 1960161 10 342 35 224 83 855 0.1 1961/62 9 318 35 224 59 555 0.1 1962/63 -3 337 39 220 78 -226 0.0 1963164 -12 357 43 202 112 -1,344 -0.1 1964165 -9 278 42 187 49 -451 0.0 1965166 1 262 43 187 32 42 0.0 1966167 57 396 47 224 125 7,138 0.5 1967168 47 562 46 254 262 12,262 0.8 1968/69 27 761 49 284 428 11,342 0.6 1969/70 34 818 45 293 480 16,272 0.8 1970/71 34 643 45 293 305 10,462 0.4 1971172 32 688 88 293 307 9,947 0.4 1972173 22 824 146 366 312 6,833 0.2 1973/74 14 1,294 211 439 644 9,080 0.2 1974/75 27 1,688 314 489 885 23,541 0.5 1975176 34 1,526 397 585 544 18,714 0.3 1976177 35 2,596 336 732 1,528 53,480 0.6 1977/78 64 3,942 1,071 1,333 1,538 98,893 0.6 1978/79 61 10,396 1,492 2,667 6,237 379,210 1.5 1979180 56 9,120 2,120 4,000 3,000 168,000 0.5 1980181 62 6,300 3,573 4,000 -1,273 0 0.0 1981182 46 5,000 5,200 12,000 -12,200 0 0.0 Note: FOB, free On board; CMD, CocOa Marketing Board. a. Fos price minus CMB costs minus official producer price. b. Revenue to government per metric ton (column 5) multiplied by total smuggled cocoa (column 1). Source: Stryker 1990, pp. 266, 308, 313. Appendix 7.5. Decomposition of the Changes in the Central Government Deficit According to Changes in Economic and Policy Variables, Ghana, 1972/73 through 1988 (ratio to GDP) IIe1 c1972/73 1973/74 1974175 1975/76 1976177 1977178 1978179 Changes attributable to domestic variables 0.000 -0.003 .0008 0.015 -0,016 -0.007 0.006 Real rGDP (Y) -0,001 -0.011 0.003 0.006 -0.026 -0.011 0.013 Exports ( 0.003 -0.004 -0.005 0.002 -0.003 -0.008 -0.003 Imports (IM) -0.001 0.008 0.002 -0.002 0.001 0.000 -0003 Domestic nominal interest rate (di) -0,001 -0.002 0.004 0.002 0.000 0.005 0.003 Domestic inflation (fP) 0.000 0.001 0.001 0.000 0.003 0.003 -0.002 Real exchange rate (Rfls) 0.001 0.006 0.004 0.007 0.008 0.004 0.000 Changes attributable to foreign variables Foreign nominal Interest rate (di*) 0,002 0.001 0.000 -0.001 o.o00 o.o00 0.001 Changes attributable to fiscal policy variables -0.007 0.020 0.006 0.055 -0.021 -0.030 -0.019 Foreign debt (D*IPt) 0.001 0.000 0.000 0.000 0.000 0.000 0.000 Domestic debt (DIP) 0.000 0.000 0O000 0.002 0.001 -0.001 -0.002 tv Wage bill (WBid) 0.003 -0.001 0.013 -0.007 -0.021 0.003 -0.001 Goods and services (05(P) -0,004 0.019 -0.002 0.004 -0.012 -0.001 -0.0t Transfers and subsidies (TSIP) 0.005 0.001 0.006 0.026 -0.013 -0.006 -0.015 Investment (IP) 0.005 0.001 -0.007 0.003 0.017 -0.013 0.013 Policy dummy: direct taxes (dD7478) 0.000. 0.000 0.003 0.000 0.000 0.000 -0.003 Policy dummy: post-1984 direct taxes (dD84) 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Policy dummy: import duties (dD7275) -0,010 0.000 0.000 0.009 0.000 0.000 0.000 Policy dummy: Import duties (dD77) 0.000 0.000 0.000 0.0 0.000 -0.015 0.014 Policy dummy: 19580-82: export tax (dD8082) 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Policy dummy: export tax (dD7374) O.W0 -0.018 0.000 0.017 0.000 0.000 0.000 Policy dummy: export tax (dD7879) 0.0, 0.000 0.000 0.000 0.000 0.000 -0.016 Policy dummy: other indirect taxes (dD73) 0.000 0.012 -0,011 0.000 0.000 0.000 0.000 Nontax revenues (NTIP) -0,007 0.007 0.003 0.000 0.007 0.002 0.001 Grants (grants/P) 0.000 -0.001 0.001 0.000 0.000 0.000 0.000 Sum of explained changes -0,005 0.018 0.015 0.069 -0.037 -0.037 -0.012 Unexplained changes 0.024 -0.028 0.006 0.014 0.013 0.020 -0.009 Change in deficit 0.018 -0.010 0.020 0.083 -0.025 -0.018 -0.022 (Table conilinues ot tihefollowuinrg page willt years 1980/81-88.) Appendix 7.5 (continued) Itent 1980181 1981/82 1982 1983 1984 1985 1986 1987 1988 Changes attributable to domestic variables 0.001 0.013 -0.003 0.019 0.002 0.005 0.011 0.010 0.005 Real CDP (Y) -0.003 0.015 0.009 0.000 -0.001 -0.001 -0.001 0.000 0.001 Exports (t) -0.001 0.000 0.000 0.000 0.000 0.004 0.003 0.002 0.004 Imports (IM) 0.000 -0.001 -0.002 0.001 0.003 0.001 0.003 0.004 0.001 Domestic nominal interest rate (d) 0.005 -0.001 -0.007 0.009 0.003 0.000 0 001 0.002 -0.001 Domestic inflation (dP) 3.002 -0.002 -0.002 0.011 -0.009 -0.003 0.003 0.000 -0.001 Real exchange rate (ntR) -0.001 -0.001 0.000 0.000 0.006 O0r03 0.003 0.003 0 000 Changes attributable to foreign variables Foreign nominal interest rate (di*) 0.001 0.000 0.000 0.000 0.000 -0.002 -0.002 0.000 0,001 Changes attributable to fiscal policy variables 0.028 -0.030 -0.006 -0.043 0.008 0.038 0.006 0.007 0.004 Foreign debt (D*IP*) 0.000 0.000 0.000 -0.001 0.004. 0,003 0,004 0.001 0.000 Domestic debt (DIP) -0.003 -0.001 0.001 -0.005 -0.002 0.000 0.000 -0.003 -0.003 Wage bill (WBIP) 0.004 -0.012 -0.003 -0.006 O.0O 0.025 0.012 -0.001 0.002 Goods and services (GSIP) -0.003 -0.002 0.000 -0.004 0.020 0.001 -0.005 0.003 0.003 Transfers and subsidies (TSIP) -0.003 -0.009 0.002 -0.006 -0.005 0.007 -0.002 0.002 0.000 Investment (IIP) 0.001 -0.007 -0.002 -0.002 0.006 0.00) -0.002 0.006 0.006 Policy dummy. direct taxes (dD7478). 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0,000 0.000 Policy dummy: post-1984 direct taxes (dD84) 0.000 0.000 0.000 0,000 -0.007 0.000 0.000 0.000 0.000 Policy dummy, import duties (dD7275) 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Policy dummy: Import duties (dD77) 0.000 .0,000 00000 00 0.000 0.000 0.000 0.000 0.000 Policy dummy: 1980-82: export tax (dDS082) 0.015 0.000 0.000 -0.017 0.000 ..000 0.000 0.000 0.000 Policy dummy: export tax (dD7374) 0,000 0.000 0.000 0.00 0.000 0.000 0.000 0,000 0.000 Policy dummy: export tax (4D7879) 0.016 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Policy dummy: other indirect taxes (dD73) 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Nontax revenues (NTIP) 0.001 O.101 -0.005 -0.001 -0.005 -0.005 0.002 0.001 0.000 Grants (grantslP) -0.001 0.000 0,000 0,000 -0.003 -0.002 -0.003 -0.001 -0.004 Sum of explained changes 0.030 -0.020 -0,009 -0.024 0.010 0.041 0.015 0.017 0.010 Unexplained changes 0.002 -.003 -0.003 0.005 -0.019 -0.037 -0.038 -0.022 -0.008 Change in deficit 0.032 -0.023 -0.013 -0.019 -0.009 0.004 -0.023 -0.005 0.002 Note: As an example, a figure such as 0.021 should be interpreted as increasing tihe deficit by 2.1 percent of GDP. Roumeen Islam and Deborah Wetzel 369 Notes At the time of writing, Deborah Wetzel was at St. Anthony's Coflege, Oxford, U.K. She participated in this project as a consultant to the World Bank. 1. For a detailed account of economic policy through 1972, see Leith (1974) and Killick (1978). Rimmer (1992) provides an in-depth account of economic policy from preindependeence days to the present. Stryker (1990) provides a detailed account of the political economy, with particular emphasis on agri- cultural policies. 2. Net lending has primarily been to public sector entities, and very little of it has been repaid. It therefore is considered expenditure rather than financing. 3. Adding foreign-financed capital expenditure increases central govern- ment gross fixed capital formation from 1.8 percent of GcP in 1984 to 5.27 percent of GDP in 1993. 4. The export tariff that goes to the central goverrnment is determined as the mOn price minus Cocoa Marketing Board costs minus the price paid to producers. 5. A newly planted cocoa tree takes five years to begin producing and about eight years to reach maxiinum yield. Depending on maintenance, a tree wil produce for thirty years. Hence, policies in the 1960s affected the volume of production in the 1970s. 6. Revenue from the export tax on cocoa was zero in 1980181 through 1982- because the FOB price during these years, when converted into cedis, was not sufficient to meet Cocoa Marketing Board costs and payments to producers. Hence there was nothing left to go to the government 7. See the appendix tables for the data on financial stocks. Note also that smce 1983 a revaluation account has been included in the monetary survey. This account essentially serves as credit to the monetary authorities to offset losses brought about by the devaluation of the cedi. In 1990 the government issued long-term bonds to cover this liability. 8. Making these changes gives the following Lgures (as a percentage of cop): 1984 1985 1986 1987 1988 1989 1990 Central government deficit -2.83 -3.88 -3.79 -3.16 -3.73 -3.52 -4.36 On basis of financing flows -1.86 -3.43 -3.05 -4.17 -3.88 -3.89 -3.70 (Central government deficits exdude grants but include capital expenditure financed through project lending.) 9. The estination results are: r =27.92 + 1.81BORCB + C13DUM. (2.05) (7.8) Adjusted R2 = 0.828 Durbin-Watson statistic = 1.69 The data used are the cpi inflation rate and World Bank data on public sector borrowing from the central bank for 1971-90. 370 Ghana: Adjustment, Reform, and Growth 10. For a detailed review of the Ghanaian financial sector and the controls to which it is subject, see Aryeetey and others (1990). 11. Note that although the economy is credit-constrained, the identities in the equations still hold. Lending from the banking system to the private sector, Lp, is taken as given (for example, credit to the private sector is allocated). 12. For simplicity we have assumed that domestic and foreign interest payments are zero. 13. Note that we define seigniorage in these simulations as the change in high-powered money over CDP, rather than as the change in M. over cDr. 14. The seigniorage-maximizing inflation rate is based on the long-run (steady-state) equation and does not incorporate any short-term influences. As a result, it is higher than the seigniorage-maxinmizing rate implicit in the scenarios presented earlier, which incorporzate short-run influences. 15. These figures are based on the assumption that there was no increase in the end-1989 stocks of total base money, net foreign debt, and net domestic debt. 16. Ricardian equivalence assesses whether an inaease in the deficit as a result of a decrease in (neutral) taxes affects private consumption. See Haque and Montiel (1987), Haque (1988), Leiderman and Blejer (1988), and Rossi (1989) for tests of Ricardian equivalence in developing countries. To date, the results have been mixed. 17. The real rate of interest was estimated by the relationship r = I[(I + O)/ (1 + r)l - 11100, where r is the real interest rate, i is the nominal interest rate, and r is the rate of inflation calculated using the cm'. 18. For studies on black markets and their effects on the economy (and vice versa) see May (1985), Pinto (1989), and Chhlbber and Shafik (1990). The real exchange rate has been defined as the relative prices of tradable and nontrad- able goods. 19. The assumption that agents consume only nontradables is a simplifying one; the condusions do not depend an this assumption, but the algebra is less complex. 20.X = MO + e): with e given, any increase in qS will lower X. 21. We have used rational expectations in this model. Thus the official rate of depreciation equals the expected and actual rate of inflation at the steady state. 22. For variables related to the stock of wealth, lagged values are -used, since it seems reasonable that the stock of wealth at the end of the preceding period determines consumption in the present period. 23. Terms of trade was found to be insignificant and was dropped from the equation. If the surn of the real money stock and the private sector's stock of foreign assets (wealth) is used, that sum is significanL When the money stock and the stock of foreign assets are entered separately, only the stock of for- eign assets is significant. 24. The error process in this equation has the form e, = Ole,1 + 02e4-z + Ut. This process is stationary if 01 + 02 C 1, 02- 1 < 1, and-1 c <02< 1. These conditions are satisfied for our equation. References Ahmad, Naseem. 1970. Deficif Financing, Inflation and Capital Formation: The Ghanaian Experience 1960-65. Munich, Germany: Weltform Verlag. Roumeen Islam and Deborah Wetzel 371 Amoako-Adu, Ben. 1991. "Demand for Money, Inflation and Income Veloc- ity: A Case Study of Ghana (1956-1986)." Sauings and Development 1 (15): 53-65. Aryeetey, Emest, and Fritz Gockel. 1990. "Mobilizing Domestic Resources for Capital Formation in Ghana: The Role of Informal Markets."" African Economic Research Consortium, Nairobi. Aryeetey, Ernest, Yaw Asante, Fritz Gockel, and Alex Kyei. 1990. "Mobiliz- ing Domestic Savings for African Development and Diversification: A Ghanaian Case-Study." International Development Centre, Oxford Uni- versity, Oxford, U.K. Buiter, Willem H. 1983. ''Measurement of the Public Sector Deficit and Its Implications for Policy Evaluation and Design." International Mdonetary Fuind Staff Papers 30 (June): 307-49. - . 1985. "A Guide to PubLic Sector Debt and Deficits." Economic Policy 1 (November): 33-79. 1988. "Some Thoughts on the Role of Fiscal Policy in Stabilization and Structural Adjustment in Developing Countries." NIBER Working Paper 2603. National Bureau of International Research, Cambridge, Mass Chairey, Christophe. 1991. 'Taxation of Financial Assets in Developing Countries." World Bank, Country Economics Department, Washington, D.C. Cahhibber, Ajay, and Nemat Shafik. 199. "Exchange Reform, Paralel Mar- kets and InMation in Africa: The Case of Ghana." Policy Research Working Paper Series 427. World Bank, Country Economrics Department, Washing- ton, D.C. Easterly, Wiliam. 1989a. "A Consistency Framework for Macroeconomic Analysis." Policy Research Working Paper Series 234. World Bank, Coun- try Economics Department, Washington, D.C. - 1989b. "Fical Adjustment and Deficit Fmancing during the Debt Crisis." Policy Research Working Paper Series 138. World Bank, Country Economics Department, Washington, D.C - 1989c. "How Much Fiscal Adjustment Is Enough? The Case of Col- ombia." Policy Research Working Paper Series 201. World Bank, Country Economics Department, Washington, D.C Easterly, William, Carlos A. Rodriguez, and IGlaus Schmidt-Hebbel. 1989. "Macroeconomiics of the Public Sector Deficit: Research ProposaL." World -Bank, Country Econonmcs Department, Macroeconomic Adjustment and Growth Division, World Bank, Washington, D.C. Ghana, Central Bureau of Statistics. Various years. Quarterly Diget of Statfis tics. Accra. Green, Reginald. 1987. StabilRzation and Adjustment Policies and Programmes: Ghana Case Study. Helsinki: World Institute for Development Economics Research. Haque. Nadeem UI. 1988. "FLscal Policy and Private Sector Saving Behavior in Developing Economies.'" International Monetary Fund StaffPapers 35: 316- 35. Haque, Nadeem Ul, and Peter J. Montiel. 1987. "Ricardiaia Equivalence, Liquidity Constraints, and the Yaari-Blanchard Effect: Tests for Developing 372 Ghana: Adjustment, Reform, and Growth Countries." imF Working Paper. Intemational Monetary Fund, Washing- ton, D.C. Huq, M. M. 1989. The Economy of Ghana. New York: St. Martin's Press. Islam. Roumeen, and Deborah Wetzel. 1991. "The Macroeconomics of Public Sector Deficits: The Case of Ghana." Policy Research Working Paper 672. World Bank, Country Economics Department, Washington, D.C. he, Alain, and Guillermo Ortiz. 1987. "Fiscal Rigidities, Public Debt, and Capital FRight." International Monetary Fund Staff Papers 34 (2): 311-32. KiGlick, Tony. 1978. Development Economics in Action: A Study of Economic Poli- cdes in Ghlana. London: Heinemnann. Leiderman, Leonardo, and Mario I. Blejer 1988. "Modeling and Testing Ricardian Equivalence: A Survey." International Monetary Fund Staff Papers 35 (March): 1-35. Leith, J. Clark. 1974. Foreign Trade Regimes and Economic Development: Ghana. Vol. 2. New York: Columbia University Press. Lizondo, J. Saul. 1987a. "Exchange Rate Differential and Balance of Pay- ments under Dual Exchange Markets." Journal of Development Economics 26: 37-53. -. 1987b. "Unification of Dual Exchange Markets." Journal of International Economics 22: 57-77. Marshall, Jorge, and Klaus Schnddt-Hebbel. 1989. Economic and Policy Determinants of Public Sector Deficits. Policy Research Working Paper 321. World Bank, Country Economics Department, Washington, D.C May, Ernesto. 1985. Exchange Contnols and Parallel Market Economies in Sub- Saharan Africa: Focus on Ghana. World Bank Staff Working Paper 711. Wash- ington, D.C. Mansfield, Charles. 1980. "Tax-Base Erosion and Inflation: The Case of Ghana." Finance and Development 17 (3): 31-34. Pinto, BnarL 1989. "Black- Market Premia, Exchange Rate Unification and Inflation in Sub-Saharan Africa." Worid Bank Economic Review 3 (Septem- ber): 321-38. Rimmer, Douglas. 1992. Staying Poor. Ghana's Political Economy 1950-90. Oxford, U.K.: Pergamon Press. Rossi, Nicola. 1989. "Government Spending, the Real Interest Rate, and Liquidity Constrained Consumers' Behavior in Developing Countries." In Mario L Blejer and Ke-young Chu, eds., Fiscal Policy[ Stablization, and Growth in Developing Countries. Washington, D.C.: lnternational Monetary Fund. Stryker, J. Dirck 1990. "Trade, Exchange Rate and Agricultural Pricing Poli- des in Ghana." In Anne 0. Krueger, Maurice Schiff, and Alberto Valdes, eds., The Political Economy of Agricltdral Pricing Policy. Vol. 3. Baltimore, Md.: Johns Hopkins University Press. Tanzi, Vito. 1982. The Underground Economy in the U.S. and Abroad. Lexington1, Mass.: Lexington Books. Tanzi, Vito, and Mario l. Blejer. 1983. "Fiscal Deficits and Balance of Pay- ments Disequilibrium in IMF Adjustment Programs." In Joaquin MuAns, ed, Roumeen Nslam and Deborah Wetz1 373 Adjustment, Canditionality and Intenfational Financing. Washington, D.C.: International Monetary Fund. van Wijnbergen, Sweder, Ritu Anand, Ajay Chhibber, and Roberto Rocha. 1992. External Debt, Fiscal Policy, and Sustainable Growth in TurTky. Baltimore, Md.: Johns Hopkins University Press. 8 Morocco: Reconciling Stabilization and Growth Riccardo Faini The recent economic history of Morocco resemnbles the histories of many African countries. The origins of Morocco's economic diffi- culties can be traced to the commodity (phosphate) boom of the. mid-1970s, which coincded with rising government expenditure and an unprecedented expansion of the public investment program. This in turn signaled the end of the conservative fiscal policies of the past. The sudden reversal in the terms of trade in the late 1970s-a result of the plunge in phosphate prices and the second oil shock-prompted Morocco to resort increasingly to external capital markets in order to maintain an unabated level of public expenditure. However, the con- tinued deterioration of the terms of trade, the unanticipated rise in international interest rates, and the severe drought of 1980-84 eroded debt service capacity and precipitated a major foreign exchange crisis in 1983. In response to this cisis, Morocco launched a medium-term program of economic reform and, in consultation with the Interna- tional Monetary Fund (tMF) and the World Bank, introduced a com- prehensive set of stabilization and structural adjustment measures. Since 1983 Morocco has made great progress in alleviating both inter- nal and external disequilibria. The overall budget deficit has been reduced from 9.2 percent of CDP in 1982 to 4.9 percent in 1988; during the same period, the current account shifted from a deficit of 12 percent of cDP to a surplus of 2 percent.1 There are several interesting features in Morocco's adjustment. Despite the severity of the ce ,is and the size of the adjustment under- taken, growth has remaineu fairly high, at least in relation to other highly indebted countries, and inflation has been subdued. As mea- sured by the consumer price index (cPI), inflation was equal to 6.2 percent in 1983. It increased to 12.5 percent in 1984 but then steadily declined, to 2.3 percent in 1988 and 3.1 percent in 1989. During the same period (1983-89) real GDP growth averaged 4.3 percent.2 This nerformance seems to contradict the conventional wisdom that large budget deficits foster inflation- The inflation record is particularly surprsing if we also consider that Morocco achieved a 20 percent real depreciation during the 1980s. 374 Ziccardo Faini 375 In this chapter we try to uncover the reasons underlying the perfor- mance of the Moroccan economy. We argue that wage moderation and judicious monetary policies were instrumental in restraining inflation. With one brief exception, in 1983, the monetary authorities eschewed any inflationary financing of the budget deficit. The strat- egy of avoiding inflationary financing could succeed only because the wide-ranging system of credit and monetary regulations worked to channel domestic fhnds toward the treasury at relatively low cost. The prospects for continuin-g such a strategy are not favorable, how- ever. It appears that the growth performance can be attnbuted to an exceptional export response to the new trade regime and to a set of favorable supply shocks, induding a string of record agricultural har- vests and the collapse of real oil prices. In the next section we study the evolution of the budget and its different components and examine how the defict was financed. In the subsequent section we argue that the reluctance of Morocco's policymakers to monetize existing budget defidts is well explained by the sharply unfavorable tradeoffs between higher monetization and inflation in Morocco, as is suggested by the estimation of a system of asset demand. Next, we analyze the implications of continuing bud- getary disequilibria for investment and saving decisiona. We find that such implications may be substantial, even though they may not work exclusively through traditional interest rate channels. We then assemble the various pieces of econometric evidence collected in the chapter to study the impact of fiscal policy in the context of a macro- econometric model. The final section offers conclusions. The Budget Defict: Evolution and Financing The budget in Morcco continued to deteriorate during the second half of the 1970s. Some tmid attempts at macroeconomiic stabilization only temporarily restrained the burgeoning financial needs of the treasury. However, at the beginnin of the 1980s the fiscal repercus- sions of the fall in Morocco's terms of trade, the increasing burden of foreign debt attendant on the steep rise of intemational interest rates, and the growing reluctance of foreign commercial banks to provide continuing financing to the treasury brought the issue of f&:al responsibility to the forefront. The budget defict, measured on a cash basis, showed a slow but steady decline starting in 1981. After increasing (as a percentage of GrP) from an average annual value of 3.2 percent d.ring 1971-74 to 13 percent during 1976481, it fell to 8 percent in 1984-85, stabilized at about 5 percent at the end of the decade, and declined again in more recent years (table 8.1). Cash-based measures of the budget defict can be quite misleading when the treasury resorts to the use of financial arrears to cope with 376 Morocco: Recondling Stabilization and Growth Table 8.1. Budget Defict (Cash Basis, before Debt Relief), Morocco, 1971-91 (percentage of air) Year Djefiit Year Deftit 1971 2.99 1981 13.64 1972 4.01 1982 9.22 1973 2.01 1983 11.52 1974 3.91 1984 8.07 1975 9.51 1985 857 1976 18.09 1986 5.71 1977 15.84 1987 6.28 1978 11.28 1988 5.81 1979 10.14 1989 4.84 1980 9.01 1990 4.44 1991 3.81 Source: World Bank data. mounting financial difficulties. This is what happened in Morocco up to 1985. Because cash-based deficts do not allow for the accumulation of financial arrears, they fail to reflect the full pressure that fiscal policy exerts on available resources. Conversely, they overestimate the size of the fiscal problem when financial arrears are being decumulated. After 1985 adjustment programs m Morocco involved a sizable reduction in the stock of arrears. It is more appropriate to treat the accumulation of arrears as a source of (involuntary) finance and to measure the deficit on a payment-order basis instead. If this is done for Morocco, the deficit picture changes substantially. Both the initial inacrease and the subsequent decine in the budget deficit are more pronounced when a payment-order basis rather than a cash basis is used (table 8.2). The reason is simply that arrears were accumulated up to 1985 and decumulated afterward (except in 1989). A further correction relates to interest payments. As column (3) of table 8.2 shows, interest payments have increased steadily as a share of GOP, thereby compounding the budget difficulties. Yet allowance should be made for the fact that under inflationary conditions a potentially conspicuous share of interest payments represents early amortizaton of outstanding debt and, as such, should be considered a (negative) financing component. For Morocco historically low inflation rates mean that this correction is not going to produce dra- matic effects. Its iInpact is nonetheless substantial, albeit dedlining. From column (4) we can see that capital gains on domestic debt attrnb- utable to inflation were equal to 3.1 percent of GDP in 1984. Because of the drop> inflation, they dedined to 1.3 percent in 1988. As a result, the operational budget, which only includes real interest payments, registered a relatively modest deficit in 1986 but showed some im- R-RiTrdo Faini 377 Table 8.2. Budget Deficit, Morocco, under Various Definitions, 1983-91 (percentage of cDP) Capital Capital Payment- gains on gains on Cash order Interest domestic forign Opetional Primary basis Izsis payments debt dbt deFdt deficit Year (1) (2) (3) (4) (5) (6) ( 1983 115 12.1 4.9 0.0 0.0 12.1 7.2 1984 8.1 11.2 6.1 3.1 0.5 8.1 5.1 1985 8.6 9.6 6.2 3.1 1.9 6.6 3.4 1986 5.7 5.4 5.9 3.6 1.8 1.8 -0.5 1987 6.3 5.9 5.9 2.0 3-0 3.9 -0.1 1988 5.8 4.7 6.3 1.3 4A 3.4 -1.6 1989 4.8 6.0 6.2 1.7 8.5 4.2 -0.2 1990 4.4 3.4 6.2 2.2 5.1 2 -2.8 1991 3.2 3.0 5.6 2.0 4.3 1.0 -2.6 Sou mr: Author's computadons, based an Wodd Bank data. provemenrt over the following years- We have also corrected for valu- ation effects in the stock of foreign debt by multiplying the outstand- mg stock at the beginning of the period by the excess of domestic inflation (measured by the change in the GDP deflator) over the rate of devaluation (taken with respect to the nominal effective exchange rte). We feel much less confident about this correction insofar as it overlooks valuation changes resulting from, for instantce, cross- currency fluctuations, which appear to have played a substantial rule in determining the evolution of foreign debt indicators for Morocco. We nonetheless report the resuts of such correction (column 5). Its impact was basicafly insignificant in 1984 but grew over time until 1989, as domestic inflation was no longer matched by a correspond- ing devaluation. The last measure we consider is the primary deficit, which excludes all interest payments. By excluding a component that is to a large extent beyond the control of fiscal authorities, the primary deficit provides a more accurate indicator of the effort to redress existing fiscal imbalances. The evolution of the primary deficit highights the adjustment effort on the fiscal front. The primary deficit was equal to 7.2 percent of GDP ft .'983 and has steadily dedined since then, reach- ing a surplus of 1.6 percent of C;DP in 1988 and 2.6 percent in 1991. This amounts to a turnaround of almost 10 percentage points of GDP. The improvement in the overall and the operational deficits is less significant because of the increased burden of nominal and real inter- est payments, respectively. - IIi summary, Morocco has undeniably taken a decisive step on the road to redressing its fiscal imbalances. The substantial achievements 378 Morocco: Recnciling Stabilization and Grmuth in the overall budget reflect an even more significant improvement in the primary deficit that more than compensates for the rising burden of nominal interest payments. The favorable evolution of the budget was also facRitated by low and even negative real interest rates on domestic debt. This situation, however, is coming to an end, and the treasury is now forced to offer positive and high real rates of return to convince domestic investors to finance its deficit. Despite the drop in inflation, the average nominal interest rate on domestic debt increased from 4.3 percent in 1983 to 5.7 percent in 1986 and reached 7.6 percent in 1989. The possibility that the average cost of servicing domestic debt will keep rising rapidly in the near future underscores the need for continuing fiscal restraint. ne Primary Deficit In this section we examine how the sizable reduction in the primary deficit was achieved. It is apparent from the data in tables 8.3 and 8.4 that the burden of adjustment fell mostly on public investment, which registered a major decrease, from 11.5 percent of CDP at the onset of the adjustment program in 1982 to only 3.6 percent in 1986. It then recovered, but only marginally, to about 6 percent in 1987 and 1988 and 7.5 percent in 1989.3 The drop in public investment was therefore the major contributing factor in the process of fiscal retrenchment. This is an unfortunate feature of the adjustment pro- cess that Morocco shares with many other developing countries. We shall see later how the decline in public investment may have nega- tively affected the propensity of the private sector to invest Table 8.3 shows how other items contributed to fiscal restraint. Expenditure on goods and services fell somewhat, reflecting mostly the drop in public employees' real wages and the more sober trend in public employment. Simple econometric analysis shows that expen- diture on goods and services is not related to inflation (see note 5), suggesting that inflation was not paramount in reducing public sector real wages. Fmally, we find that the share of subsidies in GDP fell substantially, mainly because the dedine in imported food prices reduced the need for government intervention. This is, incidentally, an interesting example of the direct impact of terms of trade fluctua- tions on the budget. A comparison between the evolution of the primary deficit, as reported in table 8.2, and the behavior of public expenditure in table 8.3 suggests that taxation did not contribute noticeably to the inprovement in the budget. As a matter of fact, taxation as a percent- age of GDP fluctuated around 21 percent until 1987 (table 8.4). Had it not been for the windfall revenue attendant on the petroleum levy introduced in 1986, fiscal revenues would actually have registered a Rkccardo Faini 379 Table 8.3. Central Government Expenditures (before Debt Relief), Morocco, 1971-89 (percentage of GDP) Year Current CGpital Goods and serics interest Subsides 1971 14.12 3.72 12.51 1.01 0.06 1972 14.35 4.06 1261 1.14 0.61 1973 14.14 4.03 12.25 1.08 0.80 1974 19.25 5.77 12.20 0.85 6.21 1975 20.18 12.66 13.62 0.95 5.62 1976 19.48 18.89 14.93 1.20 3.35 1977 18.58 18.93 14.44 1.50 2.63 1978 18.89 13X36 15.00 1.90 1.99 1979 19.46 12.89 15.05 2.19 221 1980 21.24 9.31 14.19 254 4.51 1981 24.60 12.33 15.42 3.88 5.30 1982 22.91 11.54 14.75 3.61 4.55 1983 23.72 9.64 15.05 4.86 3.81 1984 2.4.00 8.09 13.63 6.08 4.29 1985 23.09 7.18 12.63 6.23. 4.23 1986 20.56 3.63 12.17 5.90 2.49 1987- 20.38 6.10 12.64 5.89. 1.84 1988 20.83 6.47 12.63 6.17 2.03 1989 21.51 7.54 13.08 62 . 221 Source: World Bank data. major decline over the period 1983-88. There are several reasons for this unsatisfactory evolution. This was a time when, with the support of the World Bank and the IMF, the Moroccan government imple- mented far-reaching reforms m the system of both direct and indirect taxation. For instance, the value added tax (vxr), introduced in 1986, was expected to provide in the medium n.m a more efficient and more reliable source of revenue.. Because of implementation problems, it was accompanied on its introduction by a revenue shortfall. Sim- ilarly, the overhaul of the system of direct taxation with the phasing- out of a set of income taxes differentiated according to income sources and their consolidation into a unique tax was not likely to contnrbute to the budget in the very short run. Overall, the impact of reforms on the tax system was mnost likely to be felt only in the medium run. Some promsing signs can already be detected from the evolution of fiscal revenues in 1988 and 1989.4 Other factors may also have helped to determine the evolution of tax revenues. For instance, it is often argued that high inflation reduces the real value of tax revenues because of delays in revenue collection. This effect is unlikely to be significant in Morocco because of relatively low inflation levels, and it may also have been offset by the bracket creep mechanism whereby taxpayers in an unindexed 380 Morocco: Recondling Stabilization and Growlit Table 8.4. Tax Revenue, Morocco, 1971-89 (percentage of cDr) Taxes on Taxes on Total Income goods and international Petroleum OCP Year revenues taxes setrces trade kew contribution 1971 14.85 2.83 7.60 2.58 0.00 0.00 1972 14.40 2.81 7.40 2-41 0.00 0.00 1973 16.16 3.21 7.45 2.88 0.00 0.35 1974 21.11 3.29 6.80 3.62 0.00 5.77 1975 23.33 7.02 7.32 4.33 0.00 2.72 1976 20.29 4.01 7.69 4.18 0.00 1.85 1977 21.67 5.14 8.33 5.19 0.00 0.94 1978 21.20 5.34 8.01 4.67 0.00 0.61 1979 22.25 5.36 7.99 5.48 0.00 0.61 1980 20.51 4.51 7.84 5.56 0.00 0.90 1981 22.57 4.77 7.64 6.19 0.00 1.52 1982 22.05 4.12 8.41 6.23 0.00 0.76 1983 21.27 4.38 8.77 5.31 0.00 0.91 1984 20.89 4.44 8.51 5.01 0.00 1.16 1985 20.65 4.58 8.38 t46 0.00 1.27 1986 18.84 4.03 7.07 3.62 2.44 0.13 1987 20.68 4.48 7.66 3.56 2.78 0.39 1988 22.74 4.81 7.78 4.07 3.26 0.25 1989 23.06 5.13 8.07 4A48 2.71 0.08 Note: ocPr, Office Cherifien des Phosphates. Source: World Bank data. system are taxed at increasing rates because of the impact of inflation on nominal incomes. Again, econometric evidence does not suggest a significant relation between the share of tax revenues in cDp and the level of inflation.5 Another noticeable factor that affected the evolution of govermment revenues was trade reform. In 1984, with the support of the World Bank, the Moroccan government, in an attempt to rationalize the trade regime and reduce protection for import-substituting produc- tions, chose to reduce substantially import duties and gradually phase out the "special import tax," an across-the-board tariff. The revenue shortfall caused by this measure was estimated to reach 4 percent of CoP. Continuing budgetary difficulties prompted a reas- sessment of the situation. In 1987 the special import tariff was inaeased from 5.0 to 12.5 percent, and its name was changed to "fiscal import duty." This policy shift is reflected in the increase in trade taxes between 1987 and 1989 (see table 8.4). The last column of table 8.4 highlights the impact of terms of trade fluctuations on fiscal revenues. In Morocco production of and trade Riccardo Faini 381 in phosphate products (the main foreign exchange earners for the country) are controlled by the Office Cherifien des Phosphates (ocp), a public enterprise. The latter contributes to the treasury bud- get through tax and dividend payments. As showni in table 8.4, these contributions reached a peak in 1974-75, concomitantly with the boom. in phosphate prices. They have been declining since then (except for a small rebound in 1981), adding substantially to b,td- getary difficulties. Financing the Budget Deficit The impact of budget deficits on macroeconomic conditions is to a significant extent a function of the mode of financing of the deficit itself. It is therefore essential to take a closer look at the way budget imbalances have been financed in the past. Prior to 1983 foreign bor- rowing financed nearly 60 percent of the treasury defict. The avail- ability of foreign finance came to an abrupt halt in 1983, precipitating a major payment crisis. Between 1985 and 1989, despite a substantial amount of debt relief, foreign finance accounted, on average, for only 33.3 percent of the treasury's financial needs. The goverment was forced to rely to an unaprecedented extent on Central Bank borrowing during 1983 and, to a lesser extent, 1984. This was also the period when the government inaeasingly resorted to the accumulation of arrears as a source of involuntary finance from the private sector. Yet the increased monetization of the deficit would unavoidably have provided fuel for inflation. It was the firm commitment by monetary authorities to eschew inflationary financing that prompted a major revision in the financing strategy. Starting in 1984 the government increasingly began tapping noninflationary domestic sources of financing- Nonbank sources, which accounted for a negligible share of the treasury's financial requirements in 1983-84, increased their share to 38.1 percent, on average, between 1986 and 1989. The trea- surv's reliance on voluntary domestic lending was exacerbated by the need to reduce the sizable stock of financial arrears. The partial liber- alization of domestic interest rates, which was associated with a major shift in the private sector portfolio composition from currency to time deposits, assisted this strategy. At the same time the ceiling on credit to the economy forced commerdal banks to channel a sub- stantial part of these financial resources to the treasury, thereby creat- ing a steady source of finance for government deficits. Another important component of the financing strategy was the direct sales of treasury bills and bonds to the nonbank sector. The disadvantage of this strategy was its increased cost to the trea- sury. Time deposits were excluded from the base on which obligatory investments in treasury bonds are calculated, to make these deposits 382 Morocco: Reconciling Stabilization and Growth more palatable to commercial banks. Together with the reduction of arrears and the need to offer attractive (posttax) returns on direct issues of government bonds, this measure induced a significant increase in the cost of domestic financing for the treasury. This is clearly reflected in the rapid increase in the average real interest rate on government domestic debt. Further institutional developments are likely to reinforce this trend if existing proposals to reduce liquidity and portfolio requirements for the banking sector are inplemented. In the final section of this chapter, we assess the macroeconomic impact of furthering the process of financial liberalization. Overall, Morocco has been quite successful in coping with formnida- ble financial difficulties. The foreclosure of foreign borrowing in 1983 represented a major shock to the economy and to the treasury, more significant perhaps than the rise in international interest rates. The two-pronged response of Moroccan policymakers relied on a sharp reduction in the treasury's financing needs and a shift in the composi- tion of finance in hopes of avoiding inflationary pressures. A key to the success of this strategy was the availability of relatively cheap sources of domestic finance, mostly determined by a complex array of financial regulations. However, the difficulties of relying at the mar- gin on financial repression as a cheap source of funds have already prompted a further reassessment of existing strategies and, perhaps more crucially, highlighted the sharp tradeoffs that continuing bud- get disequilibria entail. Today the major question facing Morocco's policymakers is whether the country's fiscal policy stance is con- sistent with the maintenance of low inflation, the resumption of investment and growth, and the external payments constraint Consider the plausible case in which foreign finance will not increase substantially in the medium run. Under these circumstances, the failure to persevere on the road to fiscal discipline may entail severe macroecononic consequences. Even the perpetuation of pre- sent financing strategies with a relatively unchanged budget deficit would soon run into severe problems. First, the budget deficit may not be compatible with the- maintenance of low inflation. Simple cal- culations show that in 1989 the sustainable primary budget should register at least a surplus of 2.2 percent of GDP. Similar calculations for the overall budget indicate a maximum deficit equal to 3.3 per- cent.6 Second, the budget is still extremely vulnerable to international interest rates and terms of trade shocks. For instance, the revenue from the petroleum levy, which in 1988 still represented 3.3 percent of GDP, is likely to vanish under increasing oil prices. Third, the cost of domestic finance is likely to rise steeply in the future. Again, the treasury budget would be extremely vulnerable to such an evolution. If the cost of servicing domestic debt were to increase to competitive Riccardo Faini 383 market rates, this would add to the budget an extra burden equal to about 2 percent of Gc::1.7 Could Morocco resort at the margin,to larger monetization of the deficit? By international standards Morocco's usual inflation is low, and an increase in the inflation tax might be a palatable alternative. However, besides the danger of tampering with monetary policy and damaging the credibility of the Central Bank's commitment against inflation, there is actually little room for a substantial contribution to budget financing from increased monetization of the deficit. We will show in the next section that, for Morocco, estimated elasticities from a system of asset demand indicate an extremely unfavorable tradeoff between inflation and monetary financing. Continuing reliance on noninflationary sources of domestic finance is therefore essential for keeping inflation in check. However, this strategy could lead to large increases in domestic real rates of interest. Besides putting into jeop- ardy budget equilibria, this would most likely crowd out investment. Furthermore, as discussed in a later section, even if, under the most optimistic scenario, the increase in interest rates did not materialize and the perpetuation of a system of financial regulations guaranteed a source of inexpensive finance for thae treasury, the impact of unabated budget deficits on investment and growth would still be significant because of their effects on credit markets. Asset Demand, Seigniorage, and the Inflation Tax According to the standard wisdom, when conventional lending dried up after 1982, developing countries had to make a twin tansfer of resources: abroad, to foreign creditors (insofar as new lending had fallen much below the required service on outstanding exteral debt); and at home, to the government, which in several developing coun- tries had assumed the burden of servicing foreign debt. Together with the increase in interest rates and the fall in terms of trade, this has often meant the disruption of budgetary equilibra, already jeopar- dized in many cases by unsustainable fiscal policies. The inflexibility of the tax system, the downward rigidity of fiscal expenditure (with the notable exception of public investment), and the thinness of domestic financial markets left local poliynmakers with little choice but to monetize the fiscal deficits, with sometimes calamitous conse- quences for the inflation rate. Under this interpretation, Morocco's exceptional inflation record is undoubtedly puzzling. In the preceding section we saw how financial repression, by channeling low-cost funds to the treasury, was inst-rL- mental in a strategy of reliance on noninflationary sources of budget financing. Yet the question remains: why do Moroccan policymakers remain staunchly opposed to even a limited monetization of the defi- 384 Morocco: Reconciting Stabilization and Growth cit and a greater reliance on the inflation tax? Inflation in Morocco is extremely low-3.1 percent in 1989-even by the most stringent inter- national standards. A recent report (United Nations-World Bank 1990), while not explicitly advocating an increase in inflation, argues nonetheless that there is some scope for boosting revenues from the inflation tax. The issue, however, cannot be solved on purely theo- retical grounds because the inflationary implications of deficit mone- tization depend on the response of money base demand to changes in inflation and interest rates. In what follows, we argue that the out- look for greater monetization of the budget deficit in Morocco is alto- gether unfavorable. If we also consider the destabilizing effects on Morocco's social fabric and the loss of Central Bank credibility that higher inflation would imply, we can perhaps understand the firm commitment of the country's poicymakers to price stability. The scope for increased government revenue from seigniorage and the inflation tax is determined by the private sector's choice of assets. We rely on a standard Tobin's portfolio approach to analyze the demand for currency, demand deposits, and time deposits Other assets are not included in our analysis on the grounds that, with the exception of real assets, they play a minor role in the private sector's choices. In what follows we assume that the demand for each asset is related to its own return, the other assets' returns, and the level of income. As a proxy for the return on real assets, we use the (expected) inflation rate, which, however, is not observable. We assume that expectations are formed rationally and depend on the set of information available at time t -1, which is defined to include the lagged level of prices, the money supply, the wage rate, and the exchange rate. According to our estimates, the last variable does not contribute significantly to the prediction of future prices. This prelim- inary evidence suggests that for Morocco the passthrough from the nominal exchange rate to domestic prices is fairly weak, which accounts for the fact that a sizable real depreciation did not translate into higher inflation. Similarly, the significant impact of wages on pnces supports the claim that wage moderation (the real minimum wage declined by 7.3 percent from 1983 to 1988, after inaeasing by 17 percent in the three previous years) was instrumental in containing inflation. The fitted value from the price equation is used as an estimate for expected inflation. In the process of estimation, we rely on Pagan's (1984) procedure and use our proxy for expected inflation as an instrument for actual inflation. This procedure is designed to yield consistent standard errors for the coefficients. We still have to deter- mine whether the system of asset demand should be expressed in nominal terms (with prices included among the explanatory variables and the relevant homogeneity assumptions tested) or in real terms. Riccardo Faini 385 We take the first course for the sake of generality and assume that asset demand does not respond fully and instantaneously to changes in prices. We also allow for the possibility of lagged adjustment to variations in income and interest rates. Therefore the estimating equation for a generic asset Mi reads: (8.1) In M = ao + a,l In p, + a2 In p, - + a3 In Yt + a4 ln Yt1 + asi + a't-l + a77r? + aB In Ml,- where p and Y denote the price level and the income level, respec- tively, i represents the vector of asset retums, and r is expected inflation. We would expect that in the long run the price elasticity of asset demand is equal to 1; asset demands are homogeneous of degree 1 in prices. This implies that a, + a2 1 - 08. We can repara- metrize the previous equation to test this restriction: (8.2) ln(Mi/p), = ao + (a, - 1) In p, + a3 In Y, + a4 In Yt-, + aSit + a6i_l + a77ri + (a8 + a2) In p-1 + as ln(Mi/p)..-1 If long-mn price homogeneity holds, we have: (8.3) ln(Mi/p)t=aO- (a2 +a8)Alnpt +a3hnYt +a4lnY.1 a5it + a6itf1 + a77t + as In(Mifp)..1. Equation 8.3 shows that imposing long-run price homogeneity in a context in which assets are expressed in real terms adds a backward- looking inflation term, A In p, to the equation itself. This term how- ever does not reflect any substitution effects between real and finan- cial assets but only lagged adjustment to price changes. Traditional money demand equations, in which inflationary expectations are sim- ply modeled in a backward-looking fashion, may therefore mis- takerly interpret the significance of A In p as evidence of substitution toward real assets, whereas only a process of dynamiic adjustment is involved. Our approach allows us to distinguish between these two effects by separately evaluating the statistical significance of A In p and r. Admittedly, the coexistence of these two effects may be diffi- cult to justify on purely theoretical grounds. In the estimation process we indude in the vector of asset returns (i) only the interest rate on time deposits, itd. Some demand deposits are also remunerated, but their return moves in an almost perfectly collinear way with that on time deposits. We have therefore included the remunerated component of demand deposits in the aggregate of time deposits. We can now descnbe the results of the estimation of the system of asset demand. We start from the more general dynamic specification in equation 8.3 and restrict the model to obtain 386 Morocco. Reconciling Stabilization and Growth a parsimonious representation of the data-generating process. We report only the final equation. All equations are estimnated over the 1974-88 period. The last obser- vation is saved to test for out-of-sample stability. Given the limriited number of degrees of freedom, all our results should be interpreted with considerable caution. We begin with the demand for currency (t-statistics are in parentheses): A ln(cuR/p,) = 0.46 - 1.10itd + 0.30|1n Y, -ln(CUR,-1IPt-A (2.69) (-6.39) (3.95) -0.60 ln(p/p,_p.4 (-4.30) Durbin-Watson statistic = 2.01 Standard error of regression (SER) = 0.013 Lagrange multiplier = 0.25 Hendry = 1.47 The demand for currency (cuR) is significantly related to the inter- est rate on time deposits (itd) and to income.$ Actual inflation appears in the equation with a short-run elasticity of 0.60. As mentioned earlier, this is not necessarily an indication of significant substitution possibilities between currency and real assets but may simply reflect the fact that nominal demand for currency fully adjusts to changes in the price level only in the long run. The coefficient on expected infla- tion is not significantly different from zero and bears the wrong (posi- tive) sign. Jointly estimating the price and the currency equations (and testing for the expectational restrictions) did not improve the results. The variable r therefore has been eliminated from the final equation. Finally, the hypothesis that nominal demand for currency is unit elastic in the long run with respect to the price level and to income is not rejected by the data (tla = 0.94). As diagnostic tools we rely on the Hendry test for out-of-sample stability and the Lagrange multiplier test for serial correlation. Both are distrbuted as X(1). They do not provide any indications of misspecification. We now tum to the equation for demand deposits (DD). ln(DD/pt) =-11.65 - 8.46itd - 1.03 lin(p/pt-,) + 1.20 In Y. (2.73) (-2.85) (-2.49) (3.22) + 0.76InYt-1. (2.97) Durbin-Watson statistic = 1.43 sER = 0.033 Lagrange multiplier = 1.56 Hendry = 0.81 Riccardo Faini 387 Demand deposits have been defined to exclude savings deposits. Following this modification, we find that both inflation and the return on time deposits significantly affect demand deposits. The long-run income elasticity is equal to 1.96 and is significantly different from 1. Once again, expected inflation does not contribute in a statistically significant way to the equation. The usual battery of tests shows no sign of misspecification. The last component of our menu of assets is time deposits (TD): A ln(TD,/P) -1.54 + 6.94i' + 0.43[ln Yt -n(TD,_l)]. (1.93) (1.64) (2.43) Durbin-Watson statistic = 1.97 SER = 0.09 Lagrange multiplier = 0.02 Hendry 0.18 As expected, time deposits respond positively, albeit not very sig- nificantly, to an increase in their own rate of return. The dynamic specification of the equation is very simple. There is no evidence of lagged adjustment to prices, as indicated by the insignificant coeffi- cient on actual inflation. On the same ground, expected inflation has been excluded from the final equation. It was not possible to reject the hypothesis that the demand for time deposits is unit elastic with respect to income (F1 9 = 0.54). The Lagrange multiplier and Hendry tests fail to point to any misspecification problems. These results indicate -that the demand for monetary assets in Morocco is strongly influenced by the pattern of returns. The esti- mated semielasticities on the interest rate suggest potentially conspic- uous shifts in portfolio composition in response to variations in the structure of interest rates. There is, however, less indication of strong substitution possibilities with respect to real assets. Our proxy for expected inflation did not prove significant in any of the three equa- tions. It is not apparently a problem of multicollinearity, insofar as the coefficient on actual inflation was quite well determined even in the more general specifications, nor is it a problem of statistical methodol- ogy. Our approach should provide, at a mniumum, consistent esti- mates of the coefficients and of their standard errors. As mentioned earlier, more efficient simultaneous estimation methods, which also alow for the expectational restrictions, fail to change the basic find- ings. Perhaps our proxy for expected inflation is not well specified. Alternatively, expectations may to a large extent have an adaptive form. To compute the amount of monetary financing that the government cart count on, we allow for the distinction that must be made in Morocco between reserve requirements, which apply only to demand 388 Morocco. Reconciling Stabilizationi and Growtlh deposits, and liquidity requirements, which force commercial banks to invest a fixed share of their deposits in low-yield treasury bills. Recent financial sector reforms have substantially increased the inter- est rate paid on liquidity requirements, which now approaches mar- ket rate levels. We therefore indlude liquidity requirements in the domestic public debt. The narrowly defined monetary base (MB) is equal to: (8.4) MBt = CURt + rrddDD, where rrdd is the reserve requirement coefficient for demand deposits. At any point of time the amount of monetary financing is equal to the change in the monetary base. As a share of nominal GDP, the amount of monetary financing (AMB,IptY) is equal to the rate of change of mone- tary base (AMBt/MB1), multiplied by the GDP share of MB, (MBtIp1Yt). We can now use equation 8.4, together with our estimates of the demand for currency and demand deposits, to evaluate the relation- ship between inflation and the amount of monetary financing. We focus on a steady-state situation and arbitrarily impose the condition that demand deposits be uriit elastic with respect to income. As a result, the growth rate of MB will be equal to the sum of output growth (n) and inflation '7r). Similarly the GDP share of MB will depend on the level of nominal interest rates and, possibly, of inflation: (8.5) AMB/pY = (r + n)[(cuw + rrddDD)(pYj = (r + n)[Ae- rrddBe-ftr where the coefficients A and a (B and 63) are computed from our estimates of the currency (demand deposits) equation. In assessing the values of a and ,B, we face a basic ambiguity. Changes in the inflation rate can have a direct impact on the demand for monetary base to the extent that they lead to corresponding variations in the nominal interest rate. We assume this to be the case. But changes in r also have a direct effect on the demand for currency and deposits. If we interpret the significant coefficient of actual inflation in our esti- mates simply as evidence of lagged adjustment, this effect should not reasonably play a role in a steady-state analysis. If, however, we believe that expectations are adaptive and our results reflect the exist- ence of significant substitution possibilities with real assets, inflation should have an independent effect. Note that in the latter case changes in inflation will exert a larger impact on the demand for monetary base. In what follows we allow for both possibilities. We can use equation 8.5 to compute the effect of a I percent change in the inflation rate on the quantity of monetary financing as a share of cop. We take cDp growth t be equal to 4 percent. From equation Riccardo Faini 389 Table 8.S. Impact on Inflation of a 1 Percent Increase in the Share of Monetary Financing in GDP, Morocco (percent) Initial inflationl Low elastieities High elasticities 4 7.5 7.6 6 9.8 11.1 8 12.9 16.9 10 16.7 26.4 8.5 it is dear that this derivative is a function of the initial share in CDP of currency and demand deposits, which in turn depends on the inflation rate itself. We can also compute the inverse derivative, which measures the increase in the inflation rate attendant on a I percent increase in the cop share of monetary financing (table 8.5). Our estimates suggest that the inflationary impact of higher monetiz- ation increases very rapidly and that the tradeoff between inflation and inflation tax worsens substantially as inflation increases. This provides considerable support to the choice of Moroccan policy- makers not to rely on inflationary forms of deficit finance. Investment and Saving Decisions The Investment Choice A high rate of investment represents a basic condition for sustained increases in econormic growth over the long term. From 1982 to 1987, however, the share of investment in (current prices) GDP in Morocco has been steadily falling, causing increasing concern about the long- run perspectives of the economy. Looking at the constant-price ratio between investment and GDP would only accentuate the fall in invest- ment because of the increase in the real price of investment goods attendant on the real depreciation. Perhaps of greater concern is that the drop in capital accumulation is generalized, involving both public and private investment- Contrary to initial expectations, the fiscal retrenchment that took a toll on public investment was not compen- sated by a matching increase in private investment. Is the fall in investment a significant cause for concem? It could be argued that most of the fall in investment can be predicated on the higher cost of capital. By encouraging less capital-intensive projects (and removing the previous bias against labor-intensive production), an increase in the cost of capital would allow the same level of growth to be achieved with a lower volume of investment. Under this interpreta- tion the drop in investment would only represent the outcome of adjusting to a new constellation of factor prices. Empirical evidence 390 Morocco: Reconciling Stabilization and Growth for a relatively large -sample of developing countries (Faini and de Melo 1990) suggests that cost of capital considerations can account for only a small fraction of the fall in investment and that other factors must therefore be at work. In what follows, we shall assess the rele- vance of this approach for the case of Morocco. We rely on a simple model of a firm. This firm is assumed to maximize its net worth, subject to a standard neoclassical production function. In contrast to the traditional setup however, financial choices are assumed to have a significant bearing on real decisions.9 We model the impact of financial variables by assuming that external equity financing is unavailable and that the firm must rely on two altemative sources of investment finance: retained earnings and bank debt. Given that the entrepreneurs' discount rate is assumed to be larger than the risk-free interest rate-otherwise the firm would accu- mulate financial assets-debt is the favored source of investment finance. (Another possible reason for this choice is tax considera- tions.) An internal solution for the optimal debt choice of the firm still exists, though, provided we assume that higher debt, in relation to the firm's capital stock, is associated with increasing agency costs. Finally, because of constraints in the credit market, at each point of time the firm's ratio of debt to capital stock is bounded by an upper constraint. Ex-pressing this formally: (8.6) max E[(1)t/(1 + fllp,-Yt - wNt - qt,1 + B, - (1 + r)Bt, - c(Bt, qKt)] subject to (8.TJ Bt s B qt K, (8.8) Y, = F(Kt, Nt) (8.9) K, (I. -8)Kt4, + It where, in standard notation, Y, N, K, and I represent the levels of production, employment, capital stock, and investment, respectively, and B denotes the outstanding stock of debt. The function c( ), with c1 > 0, c11 > 0, and c2 < 0, is the agency cost function, which is assumed to be quadratic. The output price is indicated as p, the wage as w, and the price of investment goods as q; r and i denote the interest rate and the discount rate, respectively. We assume that 1 + i = (1 + r)(l + k), where k is a multiplicative risk prenmium. Finally, 4 is the depreciation rate. Equation 8.7 is the leverage constraint that defines the maxmum amount of debt as a time-varying proportion (B*) of the capital stock, while equations 8.8 and 8.9 descnbe the production relationships and the capital accumulation identity, respectively. Suppose first that the Riccardo Faini 391 debt constraint is not binding. It can be easily shown that, at an optimum, debt will be a fixed proportion y of nominal capital stock. In turn, -y is a function of the risk premium k and of the parameters affecting the position of the agency cost function.'0 Note that -y does not depend on the interest rate. As a matter of fact, variations in r (or in the discount rate z) affect the level of investment, not the composi- tion of its financing. The choice of -y in turn affects the demand for capital, which is otherwise determined in a standard way. If we con- sider the case in which the debt constraint is binding (B* < y), finan- cial conditions have a direct impact on the demand for capital. Under linear homogeneity conditions in production, it can be shown that the optimal capital stock will be a function of output, the real cost of capital, and the availability of debt. Represented formally: (8.10) X, = K (Y,, c,tp,, B,) where c, = qt (1 - I) [1(1 - r) + 6]. The parameters r and A denote the corporate tax rate and the percentage reduction in q induced by the system of fiscal and financial incentives available to investors in Morocco. Full details for the calculations of r and ;L are provided in World Bank (1990). A problem with the formulation of equation 8.10 is that it contains one obviously endogenous variable, the level of output (Yt). It is assumed that because of delivery lags, firms must determine their desired level of output, and therefore their investment, one period in advance. They wiUl therefore need to predict, on the basis of available information, the optimal level of capacity output for the following period. This will determine in turn their demand for investment goods. We also assume that expectations about the determinants of the capacity decision can be simply modeled by a first-order auto- regressive process. The expected optimal level of output (t-,Y,) will therefore be equal to: (8.11) t =-lY = Y(p,-1, w,_1, c,_ l, pub,_., MS,-,) where the information set has been augmented to indlude both Ipub (the level of public investment) on the ground that this may affect the production relationship, and MS (the stock of money), as a further predictor of prices. In estimating equation 8.11 we must allow for the fact that t-PY' is not observable. However, under rational expecta- tions, it will differ from actual output Y only for a random term uncorrelated with any available information at t -1. We can therefore use the actual production level Y, as the dependent variable and take the fitted value of equation 8.11 as the estimate of expected output. By estimating equation 8.11, it is now possible to take into account the endogeneity of Yin equation 8.10. We again follow Pagan's (1984) suggestion and take the estimated value of Y(YJ in equation 8.11 as 392 Morocco; Reconciling Stabilization and Growtit an instmmernt for the actual level of output in equation 8.10. Both equations 8.10 and 8.11 are estimated over the 1972-bd period. We first present the estimates of equation 8.11: In Y, = 7.01 - 0.30 ln(w/p),_ - 0.26 ln(c/p),_ + 0.43(MSIt.,. (29.9) (-0.48) (-1.81) (1.63) + 0.52(1pub/Y)t,_ + 0.59timc - 0.04D1 (1.72) (8.08) (-2.84) R2 = 0.99 Durbin-Watson statistic = 2.02 SER = 0.02 Lagrange multiplier = 0.21 -Hendry = 0.01 where DI is a dummy variable that takes a value of 1 in correspon- dence to the agricultural negative supply shocks in 1981, 1983, and 1987. The wage rate does not contribute significantly to the equation. This may be attributed to measurement errors (we use an indicator of the minimum wage) or to the fact that labor is not a significant con- straint. Instead, the cost of capital appears to play a more significant role, together with public investment and the money supply. As a diagnostic tool, we rely on the Lagrange multiplier test for serial correlation; this test does not indicate any significant problems. Moving now to the investment equation, we find that: A in it -3.55 + 1.94 a in Y, - 0.69 ln(InY),1 (-1.96) (2.21) (-2.24) - 0. 76 ln(c/p)t + 5.67(BpIY)t (-2.0) (1.84) R2= 0.50 Durbin-Watson statistic = 1.11 SER = 0.12 Godfrey = 0.02 Sargan = 0.18 where I represents private investment and BpIY, the ratio of firms' credit to GDP, is used as a proxy of the stringency at an aggregate level of the debt constraint. The equation has been estimated by an instru- mental variable procedure, with the fitted value from equation 8.11 as an instrument for Y. The restriction that investment be unit elastic with respect to output has been tested (t12 = 0.5) and imposed on the data.2 The estimation of equation 8.10 yields two interesting results. First, investment is significantly affected by the real cost of capital. Fiscal policy can therefore affect the investment decision through its impact Riccardo Faini 393 on interest rates or, more directly, by changing the set of fiscal and financial incentives available to investors. Second, the level of invest- ment in the economy also depends on the availability of credit. The joint presence of both the cost of capital and credit availability may appear redundant (or even contradictory), until we recall that invest- ment in this model is not fully determined by the stock of credit that financial intermediaries are willing to extend to firms. Even if firms are credit constrained, a change in the interest rate r will still affect investment through its impact on the discount rate i and thus on the choice of retained earnings. By varying their retention behavior, firms can relax the credit constraint somewhat. Fiscal policy therefore will affect investment by influencing either the interest rate or credit avail- ability. This latter channel, as we shall see later, can play a crucial role in determining the macroeconomic outcome of different fiscal poli- cies. It is also plausible, of course, to interpret the joint significance of credit and interest rates in equation 8.10 as implying that only a subset of firms is credit constrained. Finally, by putting together the estimation results for equations 8.10 and 8.11, we can argue that public investment bears a complementary relationship to private investment. Indeed, an increase in public investment will lead to higher-capacity output (equation 8.11) and, through this channel, to higher private investment. By severely cutting public imvestment, fiscal policy may have contributed in the past to the stagnation of private investment. T'he Saving Decision A steady supply of domestic savings will ensure that a sustained rate of investment would not be incompatible with existing constraints on external payments. It is therefore essential to gather an adequate understanding of the determinants of saving behavior. Unfor- tunately, the measurement of saving is beset with difficulties, to the extent that consumption is computed residually in Morocco's national accounts. We can rely on two alternative measures of pnrvate saving, one derived as the difference between private disposable income and private consumption, the other based on the saving- investment identity for the economy. Because of statistical inconsis- tencies, the two procedures do not yield the same result. In what follows we use the first measure. We find that after peaking during the phosphate price boom and dedining afterward, the average pro- pensity to save increased steadily during the 1980s. To model the behavior of saving, we first take a simple permanent income approach. Under well-kn( wn conditions, the maximization of intertemporal utility by the representative consumer will imply that 394 Morocco: Riconciling Stabilization and GrowtJi aggregate consumption (denoted as C) is simply equal to a proportion of permanent income (YP). That is, (8.12) C = kYp. Note that C should be defined to include only the consumption of nondurable commodities. Lack of data precludes this important refinement. For estimation purposes, at least two issues need to be addressed. First, we must specify an indicator of permanent income. In what follows, we simply regress the actual value of real disposable income on a time trend and take the fitted value from such an equa- tion as an estimate of YP. Second, we need to recognize that the parameter k will not in general be fixed but will depend on the real interest rate and, possibly, the real exchange rate. The irnpact of both variables on the propensity to consume, however, is theoretically ambiguous. Our estimated equation reads: A In C, = 0.13 + 0.48(ln Y, - Ct,) - 0.17r, - 0.044 In X, (1.08) (6.27) (-1.65) (-2.41) R2= 0.73 Durbin-Watson statistic = 2.25 SER = 0.015 Godfrey = 0.75 Sargan = 8.59 where r and X denote the real interest rate and the real exchange rate, respectively.12 The latter is defined so that an increase in X implies a real depreciation. The equatiorn is estimated over the 1972-88 period. In the estimation, the restriction that the long-run elasticity of con- sumption with respect to permanent income be equal to 1 has been tested and imposed in the equation (t14 = 0.81).13 Our results indicate that both an increase in the real interest rate and a real depreciation will lead to a decline in the propensity to consume. This approach highlights two channels through which fiscal policy affects private saving and consumption behavior. First, fiscal policy may influence, through taxes and transfers, the volume of disposable income that consumers can spend. In this framework a temporary tax increase will have a more limited effect than a permanent increase. Second, fiscal policy may affect private saving by influencing the level of interest rates. Our estimates suggest that an expansionary fiscal policy will crowd out both consumption and investment demand through the interest rate channel. A third channel (not alowed for in the preceding estimates) may be at work if rational private agents take fully into account the future tax liabilities associated with bond- financed deficits. The implication is that a shift from tax to debt finance of a given volume of public expenditure should be neutral insofar as it would be matched by offsetting behavior by private Riccardo Fainti 395 agents. This is because private agents would be perfectly aware of the future tax liabilities and the consequent reduction in their permanent disposable income that the increased deficit entails and would reduce their consumption correspondingly (according to the Ricardian equiv- alence hypothesis). There are several reasons why this proposition may not hold. For example, private agents may discount the future at a different rate than the government, and the presence of capital market imperfections may hinder the intertemporal swnoothing of consumption by private agents. Evidence on these issues for develop- ing countries is limnited (see, however, Haque 1988; Rossi 1988; Haque and Montiel 1989; Nam 1989; Deaton 1990). In what follows we amend the previous specification to allow for the possibility that by reducing perceived disposable income, govern- ment deficits may have a negative impact on consumption. We do not expect this effect to be particularly strong in Morocco because of the absence of a consumer credit market and the likely pervasiveness of liquidity constraints on households. To capture this effect, we intro- duce liquid assets (currency plus bank deposits) into our equation. We also allow for inflationary effects on the grounds that a high level of inflation will lead to capital losses on liquid acsets and impart an upward bias to the traditional measure of disposable income. Further- more, an increase in -inflation should be associated with greater uncertainty and should lead, through this channel, to an increase in precautionary saving. If we estimate this more complete model, we find that: A In C = 0.46(ln Y,-1 - In C,.1) + 0.06(ln M, - ln C,t-) (8.19) (3.92) - 0.09 In X, -0.12 A In Pi - 0.13(defJGDP), (-4.66) (-1.64) (-2.40) R= = 0.83 Durbin-Watson statistic - 2.16 SER = 0.012 Godfrey = 0.08 Sargan = 6.26 where, in standard notation, p indicates the price level (so that A In p approximates the inflation rate), M denotes the stock of liquid assets, and def/GDP indicates the ratio of the budget deficit to GDP. In this specification, which closely follows Hendry and von Ungern Stem- berg (1981), consumers adjust their expenditures to ensure constant steady-state equilibrium ratios of consumption to disposable income and to liquid assets. These long-run equilibrium ratios are affected by the real exchange rate, by inflation, and by budget deficits. The signif- icant impact of budget deficits on consumption in Morocco may seem 396 Morocco: Reconciling Stabitization and Gwwth surprising, but it was already found by Haque and Montiel (1989) and Schmidt-Hebbel and Muller (1990). We have tried to assess whether this effect could be attributed to the fact that government expenditure on goods and services substitutes for private consumption, but the high collinearity between the deficit variable and govemment expen- diture prevented such a test. Finally, both higher inflation and a depreciating real exchange rate have a negative impact on consumption. The real interest rate is no longer statistically significant and has been excluded from the equa- tion. We can only speculate why the real interest rate no longer seems to be a significant determinant of consumption behavior. One expla- nation might be that in the previous specification the real interest rate variable was actually picking up the effect of budget deficits. Indeed, as our last estimates suggest, a larger budget deficit lowers consump- tion and is also likely to be associated with rising interest rates. Over- all, this new specification appears to indicate an even stronger role for fiscal policy in irnfluencing consumption behavior. Modeling the Impact of Fiscal Policy A comprehensive analysis of fiscal policy requires an economywide model. In this section, building on the estimates already presented and on previous work by Faini, Porter, and van Wijnbergen (1989), we rely on a simple macroeconometric model to evaluate the impact of fiscal policy in Morocco. A full presentation of the model goes beyond the scope of this chapter but can be found in Faini, Porter, and van Wijnbergen (1989). Table 8.9 in the appendix provides some details about the model. The model is based on a simple variant of the aggregate supply-aggregate demand open-economy framework. Its salient features include the emphasis on supply behavior, the model- ing of import demand under rationing, and the analysis of both exter- nal and domestic debt dynamics. In what follows, we discuss only some of the main comparative static properties of the model. We then present a number of simulations on the impact of fiscal policy. Throughout this section, it is assumed that the economy is rationed on international capital markets. As a result, the financeable current account defict is given at each point of time. As in the model by van Wijnbergen (1989), two equilibrium conditions play a crucial role in determining the equilibrium in the economy. The first is the goods market equilibrium (equation 33 in table 8.9), which can be read as requiring that demand and supply for domestically produced goods be equal. The second condition is the current account constraint, whereby the excess of private and public saving over total investment must be equal to the exogenously given level of the current account. This condition can bc derived by combining equations 12-13, 25-26, Riccardo Faini 397 and 33-34 in table 8.9. The real exchange rate and the real interest rate. will move to equilibrate the goods markets and ensure that the cur- rent account constraint is not violated. Following van Wiinbergen (1989), the model admits a simple graphical representation. A decline in the interest rate will lead to an excess demand for domestic goods and a current account deficit. Under plausible conditions (fulfilled in our model), a real depreciation (an increase in X) will be required to meet the current account constraint, while for the goods market equi- librium, X will need to decrease. As a result, the current account constraint has a negative slope and the goods market equilibrium has a positive slope (figure 8.1). The impact of fiscal policy will depend to a substantial extent on whether the credit constraint on investment is binding. Consider the first case, in which firms are rationed on the credit market (that is, Bt c y). Desired private investment cannot be fully financed from existing sources. Equilibrium can be achieved through (a) an increase in private saving, (1) a drop in notional private investment, or (c) a fall in the demand for financial assets, which frees private saving for investment finance. As shown in the preceding section, changes in private saving and investment wil be brought about by variations in Figure 8.1. Macroeconomic Equilibrium with a Current Account Constraint r (real interest rate) AL Current account constraint Goods market equilibrium k (real exchange rate) 398 Morocco: Reconciling Stabiliation and Grwlth the discount rate, i. Indeed, the excess demand for fumds will put pressure on the rate at which households and firms make their saving and investment decisions. The discount rate will rise, presumably through the operations of an informal credit market, so as to bridge the gap between retained earnings, household saving, and invest- ment demand. Under this interpretation it is the rate on the second- ary credit market that plays a crucial role in the attainment of macro- economic equilibrium. Changes in the official interest rate are much less effective. Actually, by increasing the demand for financial assets these changes may aggravate the situation of excess demand for funds (van Wijnbergen 1983). In what follows, we also assume that the differential between the secondary market and the official interest rate is kept constant, presumably by monetary authorities, in an attempt to keep the official rate from falling too much off line with respect to actual credit market conditions. Even under these assump- tions, the official credit market wil not reach equirium. As shown in the preceding section, the composition of investment finance (as desired by the firm) does not depend on the interest rate but only on the risk premium. The firm, therefore, is still off its notional demand for credit, since it is unable to achieve the desired composition of its investment finance. Finally, both changes in secondary market inter- est rates and availability of bank credit will affect investment demand- The situation is, of course, much simpler if we abstract from the possibility of credit rationing. Under this condition the interest rate on bank loans will adjust to dear the credit market What is more crucial is that investment demand wil depend on a price signal-the level of interest rates-and not on quantity signals (the availability of aedit). We begin by considering such a case through three simple simulation exercises.14 Column (1) in table 8.6 presents the base solution. In the first com- parative static exercise (column 2) we study the impact of an increase in government spending by 3,000 million dirhams (DH), approxi- mately 1.7 percent of cDP. Foreign saving and monetary financing are assumed to be unchanged. From equation 26 in table 8.9 we see that most of the financing wfll rely on domestic debt. The increase in government spending will cause an excess demand for domestic goods, leading to a decline in saving and thus a current account imbalance. The goods market schedule wil shift upward, as will the current account locus. Indeed, for a given real exchange rate, the interest rate needs to increase to stinulate net saving and eliminate excess demand for domestic goods. The current account schedule, however, will shift less. As a matter of fact, higher interest rates lead to a fall in absorption, which is fully reflected in a current account improvement, whereas the reduction in the excess demand for domestic goods depends also on the marginal propensity to spend on Riccardo Faini 399 Table 8.6. Simulation of the Impact of Fiscal Policy, Morocco Domestically Domestically Externally financed Base financed financed Financial inrease in run incrase in C increa in C liberalization Jpub Variable (1) 2) (3) (4) (5) r 3.0 4.3 4.2 4.7 4.0 X 100.0 99.2 94.9 94.6 100.5 LY 18.8 17.7 18.4 18.0 17.9 BDIY -5.3 -7.1 -7.1 -8.1 -6.9 6.3 6A 6.9 7.0 6.3 Natet G. government expenditure, r, real interest rate; X, real exchange rate; I/Y, private investment as share of cur; BD/Y, budget deficit as share of cwr; Y, growth rate; Ipub, public investment DR. Moroccan dirhams. Simulations: 1, base case; 2, increase in govemrnment current consumption of DH 3,000 milion; 3, as in 2, plus a current account deterioration of DH 3,000 mllion; 4, as in 3, plus abolition of the liquidity requirement on demand deposits; 5. increase in public investment of DH 3,000 million. such goods, which is less than 1. As expected, in the new equilibrium (column 2), the real exchange rate wil appreciate (because of the excess Jemand for domestic goods) and the real interest rate will increase (because of the fall in saving). Growth inreases somewhat, but the impact of the expansionary fiscal policy is reflected mostly in a crowding-out of investment. Capital accumulation fals because of higher interest rates, although the effect of these rates on the cost of capital is mitigated by the real appreciation. Table 8.6 ondy presents the effect of a fiscal expansion. Over the medium run the supply impact of the drop in investment offsets the expansionary demand effects of higher public spending: it takes only three years for the initial positive output effect to be reversed. Table 8.7 provides some details about the simulation results over a four-year period. Suppose now that the increase in governent expenditur-^ is financed from abroad. We model this by assuming that the current account is allowed to deteriorate by an amount equal to the increase in public consumption. The outcome of this simulation is presented in column (3) of table 8.6. The higher current account deficit is equiva- lent to an increase in foreign saving, which fully offsets the fall in domestic saving. Diagrammatically, the current account schedule will not shift. We would then expect the interest rate to increase less than in the case in which no extra financing from abroad was available. This is indeed what we find in column (3). The real interest rate is now equal to 4.2 percent, as against 4.3 percent in the previous sim- ulation. As expected, the real exchange rate appreciation is more pronounced (5.1 compared with 1 percent in the previous simula- 400 Morocco: Reconciling Stabilization and Growth Table 8.7. Simulation of the Impact of Fiscal Policy, Morocco, 1988-92 Variable 2968 19B9 1990 1991 1992 Base case Y 167,235.922 177,819-613 187,583.305 197,081.872 207,245.196 ER 11000 0.999 1.030 1-061 1.046 R 5.0 3.0 3.6 3.8 4.3 rtalvY 16.3 18.8 18.2 17.4 16.9 BD/Y -3.9 --5.3 -5.9 -6-3 -7.3 Y' 10.4 6.3 5.5 5.1 5.2 Domesticaly financed increase in government current expenditure Y 167,235.922 177,978.036 187,673.768 197,033.446 207,042.953 ER 1.000 0.991 1.018 1.047 1.031 R 5.0 4.3 4.8 5.1 5.8 IPRIV/y 16.3 17.7 16-9 16.1 15.6 BDIY -3.9 -7.1 -7.8 -8.3 -9.4 r 10.4 6.4 5.4 5.0 5.1 Extermally financed increase in govemment current expenditure Y 167,235392 178,844.371 188,661.661 198,082.640 208,168.847 ER 1.00 0.948 0.976 1.010 1.001 R 5.0 4.2 4.5 4.6 5.1 piPrvIY 16.3 18A 17.8 17.1 16.6 BDIY -3.9 -7.1 -7.5 -7.8 -8.7 -r 10.4 6.9 5.5 5.0 5.1 Externally financed increa in government current expenditure plus financia libe alat ion Y 167,235.922 178,903.272 188,682.372 198,059.977 208,099.632 ER 1.000 0.945 0.972 1.006 0.995 R 5.0 4.7 5.0 5.1 5.7 nIIvIY 16.3 18.0 17.4 16.7 16.1 BD/Y -3.9 -8.1 -8.3 -8.8 -939 1r 10.4 7.0 5.5 5.0 5.1 Domesticatly financed inrease in public investment Y 167,235.922 177,722.467 187,580.96S 197,162.728 207,392-948 ER 1.000 1.005 1-035 L066 1.050 R 5.0 4.0 4.4 4.7 5.4 ipluvlY 16.3 17.9 17.1 16.3 15.7 BDIfY -3.9 -6.9 -7.6 -8.2 -9.3 10.4 6.3 5.5 5.1 5.2 Note: Y, real cDP at factor cost (miMlions of real dirhams); ERreal exchage rate (1988- 91); R, real interest rate (percent); IPmv, private investment as a percentage of GDP; BD, budget deficit as a percentage of GDP; r, growth rte of real GDP. tdon). Both the greater real appreciation and the more limited in- crease in the real interest rate stimulate investment, in comparison with the case in which foreign financing was unavailable. G:P growth is therefore more sustained. We find that the level of CDP is Riccardo Faini 401 always higher than its base-case value over the full simulation period (see table 8.7). We next consider the case in which a foreign-financed increase in government expenditure is accompanied by a process of financial liberalization. We model this process as a reduction in the liqdity requirement imposed on demand deposits. There are certainly bene- ficial influences from a more competitive and less heavily regulated financial sector, but these are not modeled in our setup. There are also some macroeconomic costs, as shown in column (4) of table 8.6. The direct impact of the reduction in the liquidity requirement is to increase interest expenses on public domestic debt and force the gov- ernment to rely to a larger extent on domestic credit markets. Interest rates increase, with a further expansionary effect on pubLic spending. The simulation indicates that the budget deficit would increase as a share of GoP by a full percentage point, with a crowding-out effect on private investment. Against the long-run benefits from financial liber- alization, therefore, should be placed the macroeconomic costs stem- ming from a further deterioration of the government budget. 'rhe results would change somewhat if the expansionary fiscal pol- icy stance were characterized by higher capital rather than current government expenditure. Column (5) of table 8.6 shows the impact of an increase of DlR 3,000 million in public investment. The resulting defiit is assumed to be financed domestically Comparison of column (5) with column (2) shows that the impact on output of larger capital expenditures is smaller in colum (5). Indeed, because of the rela- tively Iarger import content of investment, demand for domestic goods increases by less, and the current account deterioration is more pronounced. The real exchange rate therefore must depreciate rather than appreciating as it did in colum (2). Ir. turn, the real depreciation has a negative impact on output. In the medium run, though, the larger volume of public investment boosts both investment and sup- ply. The conrtrast with the effect of an increase in current expenditure is worth stressing. Following an increase in public investnent, output in the medium run is systematically larger than in the base case (see table 8.7). Consider now the case in which the interest rate on the primary credit market is not allowed to dear the loan market. It is assumed that B* c %y that is, firms are rationed on the official credit market. As shown in the preceding section, both credit availability and the (sec- ondary credit market) interest rate wil affect investment demand. The base-case simulation is presented in column (1) of table 8.8. Sup- pose again that govremment current consumption is increased by DH 3,000 millon. The new equilibrium is reported in column (2). As expected, an expansionary fiscal policy brings both a real appreciation and higher interest rates. Compared with table 8.6 (where credit mar- 402 Morocco: Recornciling Stabilization and Growth Table 8.8. Simulation of the Impact of Fiscal Policy, Morocco: The Credit Rationing Case Domestically Ext wely financed financed Financial Base run incraw in C increase in C lbemlization Varikbit (1) (2) (3) (4) r 0.9 1.2 1.9 19 ?100.0 98.6 94.6 94.0 flY 17.1 15.2 16.5 15.6 BDY -5.6 -7.4 -7A -8A 6.5 6.6 7.1 7.2 BVY- 18.7 17.0 18.5 17.5 - Note: G. government expenditure; r, real interest rate; X, real exchange rate; IY, private investment as share of CDP; MDY. budget defidt as share of GDP,; r', goWth rate; B/ credit to the private sector as share of GDP. Snulations: 1 base case; z increase in government current consumption of DH 3,000 million; 3. as in 2, plus a current account deterioration of DH 3,000 milion; 4, as in 3. plus abolition of the liquidity requirement on demand deposits. kets were allowed to clear), however, the impact on the level of inter- est rates is more limited, whereas the effect on investment is substan- tial (almost 2 percentage points of CDP compared with a 1 percentage point drop in table 8.6). The reason behind this result is relatively simple. Under credit rationing, the crowding-out effect of a larger volume of government spending takes place mostly through lower availability of credit to the private sector. As a matter of fact, we find that the share of private sector credit in GDP decines by almost 2 percentage points following the fisml expansion. There is, therefore, less of a need for interest rates to increase to restore macroeconomic equilibrium. Consider, finally, the case in which the increase in government expenditure is financed from abroad. We would again expect the interest rate to increase less than when foreign financing was unavail- able. This is not, however, what we find in column (3). The real interest rate is now equal to 1.9 percent, compared with 1.2 percent in the preceding simulation. The reason for this somewhat surprising result again hinges on the crucial role the credit market plays in this model. What is happening is that the possbility for the government to finance a larger portion of its deficit abroad reduces the pressure that government borrowing puts on credit markets, leaving a larger share of total domestic credit for the private sector. Investment, there- fore, is not crowded out through credit rationing and decines less than in the previous simulation. A larger increase in the interest rate and a more sustained real appreciation are required to achieve equi- librium in the goods market. When, as in the first simulation, the Riccardo Faini 403 higher government expenditure was financed domestically, most of the crowding-out of investment took place through credit rationing, with more limited effects on the interest rate. The differences in out- come between the two simulations highlight the crucial role that con- ditions in the credit market play in determining the impact of fiscal policy. Conclusions Morocco has made great progress toward macroeconomic and fiscal stability. Yet the need to consolidate and broaden the achievements to date remains paramount. We have argued that even the financing of a relatively unchanged budget defict may pose major problems. It is undikely that foreign finance will increase substantialy in the medium run. Monetary financing does not seem to be a palatable alternative, given its highly inflationary implications. Finally, increasing reliance on domestic financial markets is likely to lead to a steep increase in interest costs for the treasury, with a destabilizing effect on the evolu- tion of the mnain public debt indicators. Our simulations also suggest that an increase in government current expenditure crowds out investment, so that the short-run benefits on output of an expansion- ary fiscal policy -will be outweighed by its long-run negative impact on growth. It is essential, therefore, that the commitment toward fiscal discipline remain unshaken. At the same time a determined effort is required to implement effective reforms in the tax and public expen- diture system, to avoid having the brunt of fiscal adjustment again fall mosty on public investment. Our results indicate that because of I he complementarity between public and private investment, reduc- tions in public capital expenditures would lead to lower growth, cre- ating a detrimental effect on tax revenues and the budget. At a more general level, we have argued that the effects of budget deficits cannot be measured by looking orly at their impact on aggre- gate demand through changes in the interest rate. As shown by Blinder and Stiglitz (1983), Blinder (1987), and Bernanke and Blhnder (1988), a large part of the impact of fisca poliy may be felt through the credit market. An expansionary fiscal policy may exacerbate the pervasiveness of caedit-rationing effects on investmnent demand, with a limited impact on the level of the interest rate. Our estimates sup- port the claim that, even after controlling for the cost of capital, the availability of credit plays a significant role in influencing the demand for investment. As suggested by our simulations, the credit channel in turn wil be a crucial factor in determining the macroeconomic outcome of different fiscal policies. Finally, we argue that the impact on macroeconomic equilibria, in particular on the government bud- 404 Moroxco: Reconciling Stabilizatiotn and Growth get, should be a relevant factor in assessing the speed of financial liberalization. Appendix. The Model In this appendix we present a brief description of the macro- econometric model used to simulate the impact of fiscal policy. A full presentation of the model goes beyond the scope of this chapter and can be found in Faini, Porter, and van Wiinbergen (1989). The presen- tation is organized by economic agent (firms, households, govern- ment, and so forth) rather than by the more usual approach based on a distinction between markets (goods, labor, and money markets). This should permit a more critical evaluation of the mnicroeconomic foundations of the model. Table 8.9 can be used as a guide to the main relationships induded in the model. The first block of equations focuses on the firm. The approach used to specify the behavior of the representative firm was described in "Investment and Saving Decisions," above, when analyzing the investment decision. In equation 1 in the table, invest- ment (1) is a function of expected output (Ye), the cost of capital (c) and, possibly, depending on the way the model is specified, the availability of credit (Bp). The cost of capital (equation 2) is a function of the tax (A and r) and depreciation (6) parameters, as well as of the interest rate, r, and the price of investment goods, pi. In turn pi (equa- lion 3) is equal to a weighted average of the price of domestic (p,,) and imported capital goods, with the latter depending on the real exchange rate (I, defined as the ratio of foreign to domestic prices) and on the tariff rate of domestic capital goods (r,W,,). The specification of expected output is taken from equation 8.11 in the text. After choosing total investment, the firm decides how to allocate it between domestic and foreign capital goods as a function of their relative prices and of the extent of quantitative restrictions on imports of capital goods, q, (equation 5). Short-run choices can be described as follows. Output supply (Y) depends on total capital stock and the real exchange rate, where the latter acts as a proxy for variable costs. A real depreciation, that is, an increase in X, will boost wage and intermediate input costs and lower supply. Given that capital stock data are not available in Morocco, we take a first quasi difference of the original supply function and esti- mate equation 6 of table 8.9. In the estimation we take the primary and the government sector output levels to be exogenous. After out- put supply is determined, the demand for intermediate inputs (M") and the supply of exports (X) can be simply descrbed in equations 7 and 8, respectively, as a function of Y, the wage rate (w), and the relevant price variables (Pm,, and ps). We do not, however, assume the Riccardo Faini 405 country to be small in export markets. The price of exports, px, depends therefore on export volume, world demand (WD), and for- eign competitors' prices (p*) 15 The choices of households are described in equations 10-13 of table 8.9. We make a separability assumption in which consumers first determine the level of saving and aggregate consumption and then allocate total consumption between domestic and foreign comrnodi- ties. Aggregate private consumption (equation 10) is modeled as in the previous section. In the present version of the model we take the specification in which private consumption depends on disposable income (Yd), the real interest rate (r),. and the real exchange rate (X). The allocation of aggregate consumption between domestically and foreign pzoduced commodities is described by a simple (constrained) linear expenditure system. We allow for the fact that a subset of foreign consumption goods cannot be freely imported, and we study the impact of these rationing measures on total imports of consump- tion goods. Our approach parallels that used for investment imports and draws on Bertola and Faini (1990). As with investment goods, we do not allow for any rationing effect on aggregate consumption. The resulting equation is highly nonlinear but can be simply described (equation 11) as stating that imports of consumption goods depend on total consumption, relative prices, and the extent of quotas on such import categories (q). Equation 12 defines disposable income as the sum of GDP, net factor income from abroad (NFI), interest on public domestic debt (cDINT) and foreign debt (GxNT); transfers (TR) minus total taxes (T); and monetary financing (Mf). Note that interest payments on foreign debt (GcxNT) are already induded, with a nega- tive sign, in NFI. Then NFI + GXINT is a measure of net factor pay- ments from abroad accruing to the private sector (mainly workers' remittances). Finally, equation 13 defines private saving. The behavior of the government sector is descrbed very simply by a set of accounting and technical identities. As far as taxation is con- cerned, we distinguish between trade (equation 14) and income (equation 16) taxes. Import duties are estimated endogenously in the model by applying the relevant duty rate to imports of consumption, investment, intermediate, and other commodities. We also allow for the temporary levy on petroleum (equation 15), which was intro- duced in 1986 and should be gradually phased out over the next few years. Information on the revenue likely to be generated by this levy was provided by UvF sources. Finally, the income tax rate is deter- mined residually from total revenue data. Equation 17 defines total tax revenue. Government expenditures are also fairly disaggregated. We distinguish between government investment (equation 19), gov- emnment current spending on goods and services (equation 20), cur- rent transfers (equation 18), and interest payments on both domestic 406 Morocco: Reconciling Stabilizationi and Gmwth Table 8.9. The Complete Model, Morocco Equation item Finns (1) I = IY(YrcBp) Private investment demand (2) c = p,(l - IA) Ir(l - T) + 6J Cost of capital (3) Pi = 0dPdI + (1 - ,d)(1 + rJX,)X Price of investment goods (4) Ye = YELW-w.1), c(-1), Jpub(-1), MS(-1)] Expected output (5) Mt = MA(I, Pd/p,ni' q,) Imports of investment goods (6) Y = Y[Y(-1), 1, k X(-1)1 Supply of goods (7) MA = MA4(Y wIpnm) Imports of intermediate goods (8) X = XS(Y, p/w) Export supply of manufactured goods (9) P. = pJX. WD. p*) Export demand of manufactured goods Households (10) C = C(Y", r, X) Private consumption (11) M= M(C, PdJPm qj) Imports of consumption goods (12) Yd Y + NFI + GDINT + cXINT + TR Disposable income -T-MF (13) Sp = - C Private saving (14) TM, = AMW, Import taxes (t = n, c, ao i (15) TP- tpY Petroleum tax (16) TY=t Y Other taxes (17) T= S TM + TP + TY Totaltaxes (18) TR = tY Transfers (1) ]7tu =pub Y Public investment (20) G =gY Government expenditure on goods and services (21) GDINT = iddBI(-l) Interest payments on domestic debt (22) Gxirr = iZ,dXD(-1) Interest payments on foreign debt (23) Aim = Ar + Mr Interest rate on domestic debt (24) Sg = T - GDINT - GraNT -fl-G Govemment saving +MF (25) XD + XD(-1)XTh(-1) - CA External debt (26) ABS-CA = 1p,b - S Government budget constraint Credit and the money markets (7) CUR = CUR(Y, i', r) Demand for currency (28) DD = DD(Y, jid, 7r) Demand for demand deposits Riccardo Faini 407 Table 8.9 (continued) Equation -Ite Credit and the money mnarkets (continued) (29) TD = TD(Y, itd, 7) Demand for time deposits (30) MF = ACUR + r4r* ADD Monetary financing (31) B,, = DD + TD - R + Bg, Credit to the private sector (32) 4?d = Ar + Ar Interest rate on time deposits National income identities (33) Y= C+ G + X -M + 1 + 1,,b - ETM Goods market equlibrium (34) CA = X-M + NFI Current account Notes: X, real exchange rate; r, real interest rate; MS, money supply: q1, extent of quantitative restrictions (QRs) on imports of type i; Br aedits to firms; p4,, foreign currency price of imports of type j; p,. domestic currency price of imports of type ; pt, domestic price of good j; P,, export price; ri, tariff rate on good j; r, corporate tax rate; w, wage rate; WD, world demand; NFI, net factor income from abroad; 7r, inflation rate; R1 bank reserves; nrr reserve requirement for demand deposits. a. Not implemented in the present version of the model. debt (B. in equation 21) and foreign debt (XD in equation 22). We can therefore account in our simulations for debt dynamics. The accu- mulation of foreign debt is determined by the current account deficit and by real exchange rate variations that induce a capital gain or loss in the value of outstanding external debt (equation 25). The change in domestic debt (AB.) is equal to that part of the budget deficit which cannot be financed abroad or by seigniorage and the inflation tax (equation 26). Simple manipulations show that ABg can also be expressed as the difference between private saving and private investment. The last item in the consolidated govemrnment Central Bank budget identity is revenue from seigniorage and the inflation tax (equation 24). Government revenue from seigniorage and the inflation tax is deter- mruned by the private sector's choice of assets. Following the analysis in "'The Budget Deficit: Evolution and Financing," above, demands for currency (equation 27), for demand deposits (equation 28), and for time deposits (equation 29) are simply modeled as functions of income (Y), inflation (1r), and the interest rate on time deposits (it"). In equation 30, monetary financing (MF) is defined as the change of narrowly defined monetary base-that is, of currency and of demand deposits, the latter multiplied by the required reserve ratio (rrdd). Finally, conditions in the money market will help determine, through the banks' balance sheet identity, the equilbrium in the credit mar- ket. The sum of demand and time deposits defines total commercial bank liabilities, which are allocated on the assets side among com- pulsory reserves (R), credit to the private sector (B,,), and credit to the 408 Morocco: Reconciling Stabilization and Growthi government (Bg) (equation 31). It is assumed that banks do not hold free reserves. Credit to the government has already been determined by the government budget constraint for a given current account deficit and given monetary financing. From equation 31 we find that credit to the private sector is determined residually. Depending on the way the model is specified, the availability- of credit may in turn influence investment demand. In such a case fiscal policy will affect investment not only through its impact on interest rates, but also, more directly, by influencing the availability of credit. To close the model, we need to specify the equilibrium condition in the goods markets (equation 33) where total demand for domestic goods (the sum of consumption, investment, government spending on goods and services, and the resource balance) must equal aggre- gate supply (. A second condition relates to the current account constraint (equation 34), which is identically defined as the sum of the resource balance and net factor income from abroad (NFI). It is easy to verify that Sp +s-CA=I+Ipub. That is, the sum of private, government, and foreign saving (the negative of CA) is, in equilibrium, equal to total investment. Table 8.9 presents the complete model in summary form. There are, altogether, 34 equations and 35 endogenous variables-32 endo- genous left-hand side variables (allowing for the fact that Y appears twice and the monetary base, MB, is exogenous), plus the real exchange rate, the real interest rate, and the inflation rate. In the text, we present a number of simulations on the impact of fiscal policy. The common assumption underlying aU the simulations is that the econ- omy is rationed on international capital markets. As a result the financeable current account deficit is given at each point of time; that is, CA is exogenous. Notes This chapter is an offspring of joint work with John Porter and Sweder van Wijnbergen. I am very grateful to both of them for stimulating discussions and suggestions. The chapter has also greatly benefited from the extensive comments of Klaus Schmidt-Hebbel. I would also like to thank Tobias Muller and Klaus Schmidt-Hebbel for providing me with crucial data and a preview of their paper, which has been a constant source of ideas and insights. Finally, I am very grateful to Roberto Fumagalli for skillful research assistance. The responsibility for any remaining errors is mine alone. 1. The current account worsened in 1989, mostly because of delivery prob- lemns with phosphoric acid exports. Similarly, the budget deficit registered a small slippage in the same year because of an increase in public investment. Both the current account and the budget deficit improved substantially in the Rkccardo Faini 409 following years. In 1991 the current account deficit was 1.0 percent of GDP and the budget deficit was 3.2 percent of GDP. 2. Between 1989 and 1991 coP increased at an annual average rate of 4.4 percent. There was, however, a significant increase in inflation, with the consumer price index increasing by 8.0 percent in 1991. Much of the worsen- ing inflation performance can be attributed to the depreciation of the nominal exchange rate. 3. Public investment fell again to 7.1 percent of cDP in 1990 and 6.1 percent of cOP in 1991. 4. Similarly, in 1990 total revenues as a percentage of GDP increased to 23.8 percent of GDP, accounting for a large share of the improvement in the bud- get. However, most of the increase in tax revenue must be attributed to higher taxes on international trade flows, which climbed from 4.1 percent of GDP in 1988 to 4.9 percent in 1991. 5. We have run two simple regressions of total government revenues (R) and government expenditure on goods and services (GS) on the following . variables: real CDP (Y), the real exchange rate (X), and the inflation rate (v). For both regressions, the hypothesis of a unit-elasticity with respect to Y was not rejected at very comfortable significance levels (F1,10 = 0.04 and F1,10 = 0.33, respectively): In T 2.59 + In Y - 0.40 Ar - 0.24X. (3.81) (-1.0) (-1.57) R2 = 0.22 Durbin-Watson statistic = 2.56 SER = 0.03 In GS = -5.3 + In Y + 0.06r-0.74X. (-7.99) (0.15) (-4.99) R2 = 0.70 Durbin-Watson statistic = 1.75 sEn = 0.03 (Figures in parentheses are t-statistics.) These results indicate that while infla- tion plays no significant role in both equations, a real appreciation (that is, a drop in X) wfll lead to higher revenues and expenditures. Presumably, if the real appreciation is brought about by higher tariffs and better terms of trade, both factors should also lead to higher revenues. In turn, more buoyant revenues may prompt the government to increase the expenditure on goods and services (CS). 6. The calculations are based on standard formulations (Buiter 1985; Anand and van Wijnbergen 1989). In computing the sustainable aggregate deficit (ds), we follow Haque and Montiel (1991): d= (r + n)(b + b* + m) - el b* where, with standard notation, ir and n denote the inflation rate anid the growth rate of CDP, b, b*, and m indicate, respectively, the ratios of domestic debt, foreign debt, and monetary base to GDP, e is the nominal exchange rate, and a prime denotes a proportional rate of change. The sustainable primary deficit (pds) is equal to: 410 Morocco: Reconciling Stabilization and Growth pd = (n - r)b + (n -P - X')b! + (r + n)m where r and r' are the domestic and foreign real interest rates, respectively, while X indicates the real exchange rate (defined so that an increase in X implies a real depredation). 7. Recall that at the end of 1988, the stock of domestic debt was equal to approximately 40 percent of GDP. The computation assumes, quite conser- vatively, that the average cost of servicing domestic debt at market rates could rise by 5 percentage points from its present level of 6 percett. 8. Equation 8.3 has been reparametrized to test for a unitary elasticity of currency with respect to income. Take the simple case in which a4 = 0. The long-run elasticity of Mip with respect to Y is then equal to a31(1 - as). By subtracting the lagged value of ln(M/p) from both sides of equation 8.3, we can test whether a3l(1 - as) = 1, that is, whether a3 = 1 - aby simply checdking whether the coefficients of In Y, and ln(M/p),_. sum to zero. The same specification is used for the time-deposit and the consumption equations. 9. For a seminal contribution to the empircal evaluation of the relationship between capital market imperfections and investment decisions, see Fazzari, Hubbard, and Petersen (1988). 10. The formal solution to the firm's optimization problem is available from the author. 11. As diagnostic tools we rely on the Godfrey and Sargan tests for residual autocorrelation and overidentifying restrictions, respectively. These are the appropriate procedures in the estimation context of instrumental variables. Both tests are distnbuted as )F with 1 degree of freedom. They do not provide any indications of nisspecification. 12- The equation was estimated by an instrumental variable procedure to allow for the possible endogeneity of income, r, and B. The Sargan test is distnbuted as x2 with 5 degrees of freedom. 13. See note 8. The lagged value of disposable income was never statis- tically different from zero and was therefore dropped from the final equation. 14. We do not examine in the simulations the effect of financing the larger volume of government spending through money creatinn. The impact of monetary finaning has already been investigated in "Asset Demand, Seig- niorage, and the Inflation Tax," which pointed to the firly unfavorable tradeoff between inflation defict monetization in Morocco. The results were derived from a simple partial equlibrium rather than an economywide model. It should be recalled, however, that in our macroeconometric model the level of real interest rates is virtually determined without reference to the money market, the only exception being the impact of the inflation tax on disposable income and, thereby, on consumption. Empirically, this effect appears to be negligible. As a result, changes in the growth of money supply wil be reflected mostly in inflation. 15. In the empirical implementation of the external sector block of the model, only manufacturing exports, net tourist receipts and consumption, investment, and intermediate goods imports are endogenous. All the other components of the resource balance (mainly exports of phosphate rocks and agricultural goods and imports of food and petroleum) are assumed, in the RkcardoFaini 411 present version of the model, not to depend on price incentives and are, as a result, projected exogenously. In a future version of the model it would be important to endogenize the flow of workers' renmittances (a component of net factor income from abroad), by relating it to the level of the interest rate and the (expected) movements of the exchange rate. References Anand Ritu, and Sweder van Wijnbergen. 1989. "Inflation and the Financing of Government Expenditure: An Introductory Analysis with an Applica- tion to Turkey." World Bank Economic Review 3 (1): 17-38. Bernanke, Ben S., and Alan S. Blinder. 1988. "Credit, Money, and Aggregate Demand." ruER Working Paper 2534. National Bureau of Economic Research, Cambridge, Mass. Bertola,. Giuseppe and Riccardo Faini. 1990. "Import Demand and Non-Tariff Barriers: The Impact of Trade Liberalization." Journal of Development fio- nomics 34: 26946. Blinder, Alan S. 1987. "Credit Rationing and Effective Supply Failures." Economic journal 91: 327-52. Blinder, Alan S., and J. Stiglitz. 1983. "Money, Credit Constraints and Eco- nomic Activity." American Economic Review, Papers and Proceedings 73 (May): 297-302. Buiter, Willem H. 1985. "A Guide to Public Sector Debt and Deficits." Eco- nomic Policy 1 (November): 13-79. Deaton, Angus. 1990. "Saving in Developing Countries: Theory and Review." Proceedings of fhe World Bank Annual Conference on Development Economics 1989, pp. 61-96. Washington, D.C.: World Bank. Faini, Riccardo, and Jaime de Melo. 1990. "Adjustment, Investment and the Real Exchange Rate." Economic Policy 5: 492-519. Faini, Riccardo, John Porter, and Sweder van Wijnbergen. 1989. "Trade Lib- eralzation, Budget Deficits and Growth." World Bank, Europe, Middle East, and North Africa Department, Washington, D.C. Fazzari, F. M., G. Hubbard, and C. Petersen. 1988. "Financing Constraints and Corporate Investment." Brookings Papers on Economic Activity 1: 141- 206. Haque, Nadeem Ul. 1988. "'Fisa Policy and Private Saving Behavior in Developing Economies.'" Intemrational Monetary Fund Staff Papers 35: 316- 35. Haque, Nadeem U1, and Peter MontieL 1989. "Consumption in Developing Countries: Tests for Liquidity Constraints and Finite Horizons." Review of Ewonomics and Statistics 71 (August): 408-15. - . 1991. "Macroeconomics of the Public Sector Deficit The Case of Palistan." Policy Research Working Paper 633. World Bank, Washington, D.C. Hendry, David, and T. von Ungern Stemnberg. 1981. '"Liquidity and Inflation Effects in Consumers Expenditures." In Angus Deaton, ed., Essays in The- ory and AMesurement of Consumer Behaviour. Cambridge, U.K.: Cambridge University Press. 412 Morocco: Reconciling Stabilization and Growth Nan, Sang-Woo. 1989 "MWhat Determines National Saving? A Case Study of Korea and the Philippines." Policy Research Working Paper 205. World Bank, Washington, D.C. Pagan, Adrian. 1984. "Econometric Issues in the Analysis of Regressions with Generated Regressors," Interational Economic Review 25 (February): 221-47. Rossi, Nicola. 1988. "Govemment Spending, the Real Interest Rate and the Behavior of Liquidity-Constrained Consumers in Developing Countries." International Monetary Fund StaffPapers 35:104-40. Schmidt-Hebbel, Klaus, and Tobias Muller. 1990. 'Private Investment under Macroeconomic Adjustment in Morocco." In Ajay Chhibber, Mansoor Dai- lami, and Nemiat Shafik, eds., Rezviing Private Investment in Developing Countries. Amsterdam: North-Holland. United Nations-World Bank. 1990. "Morocco 2000: An Open and Competi- tive Economy." Trade Expansion Program. van Wiinbergen, Sweder. 1983. "Interest Rate Management in Lcs," Journal of Monetary Economics 12( 3): 433-52. -. 1989. "Growth, External Debt and the Real Exchange Rate in Mex- ico." Policy Research Working Paper 257. World Bank, Europe, Middle East, and North Africa Department, Washington, D.C World Bank. 1990. "Morocco: Sustained Investment and Growth in the 1990s." Europe, Middle East, and North Africa Department, Washington, D.C. 9 Pakistan: Fiscal Sustainability and Macroeconomic Policy Nadeem Ul Haque and Peter J. Montiel Over the past two decades Pakistan has experienced fiscal deficts that have been very large in relation to the size of its economy, by international standards. The country's authorities have made repeated attempts, indluding several adjustment programs, to deal with fiscal imbalances over this period, but they have achieved only temporary successes. The deficit of the federal and provincial govern- ments combined averaged about 6.75 percent of gross national prod- uct (GNP) during the 1980s and amounted to 7.75 percent in fiscal 1991192, the most recent year for which data are available.1 As in many developing economies, the deficit remains high because of the govemment's political and administrative inability to mobflize addi- tional resources and cut current expenditures. Weaknesses in the tax system have led to an inelastic tax struchwe and a heavy reliance on trade taxes for revenues. Moreover, with defense expenditures consti- tuting about 25 percent of expenditures, interest payments 15 per- cent, and admiinistration (including social services) another 15 per- cent, a large fraction of expenditures is not amenable to large cuts. The burden of expenditure cuts, therefore, has often fallen on devel- opment expenditure, at the cost of much-needed investments in infrastructure. In spite of the similarities to conditions elsewhere, the macro- economic consequences of fiscal deficits in Pakistan have apparently been quite dissimilar to those in other developing countnes with fiscal deficits of comparable magnitude. Specifically, Palistan has experienced neither hypernflation nor debt rescheduling. As mea- sured by official figures, growth has remained quite strong through the past two decades, inflation has not been high, and the current account deficit, at an average of about 2.5 percent of GNP, has remained largely financeable and has not posed debt-servicing prob- lems for the country. For these reasons, Paksan presents an interest- ing contast to some of the other countries in this book. This chapter focuses on why Pakistan's extremely high observed fiscal deficits proved relatively benign during the decade of the 1980s-at least in comparson with the experience of several Latin American countries. 413 414 Pakistan F&Is Sustainability and Macroeoomic Poticy The analysis concentrates on the 1980s because the political setting was relatively stable during this period under the martial law govem- ment that assumed power in 1977 and held it until the election of a new democratic government in 1988. In addition, Pakistan undertook a comprehensive revision of its macroeconomic data in 1989, and available data compiled under the new methodology go back only to fiscal 1980/81. Thus, for the purpose of empirical work, time series of useful length are avaflable only under the old methodology, which was applied up to fiscal 1987188 Moreover, although macroeconomic data are available for the period after 1988, as of the time of writing, information for the period 1989190 to 1991192 remains provisional or incomplete. The first section of this chapter begins with a brief overview of macroeconomic developments in Pakistan over the past two decades. This is followed by a more detailed look at fiscal developments that is intended to address the question of why fiscal deficits have remained consistently high. The macroeconomic consequences of these deficits are examined in the subsequent three sections, which describe calcu- lations of equilibrium deficits, estimates of the effects of fiscal policy variables on other macroeconomic relationships, and counterfactual simulations of alternative historical fiscal policies. The main conclu- sions are presented in the last section. Overview The distinguishirg feature of Pakistan's recent macroeconomic his- tory has been the country's relatively good macroeconomic perfor- mance, in spite of fiscal deficits that were high by intemational stan- dards. This is not to say that fiscal deficits of the magritudes observed have not had harmful effects or that performance could not have been improvred with lower deficits. But there is no evidence in Pakistan of the recurring acute macroeconomic crses-as manifested in extended periods of negative growth of income per capita, hyperinflation, and inability to service external debt-that have characterized many other developing countries with comparable fiscal performance. As figure 9.1 shows, the deficit (as a percentage of GNP) remained very high over a period of nearly two decades, amounting, on aver- age, to about 7 percent of GNP during 1972-883; Although some fiscal adjustment was achieved during fiscal 1977178 to 1981182, when the extremely high deficits of the 1974-78 period were nearly cut in half, the fiscal deficit was on a rising trend during most of the 1980s. Over these years Pakistan's fiscal deficit averaged nearly twice that of Asian countries as a group, according to the International Monetary Fund's World Economic Outlook (MF 1990). Nadeen LUI Hapm and Peter Mmntiel 415 Figure 9.L Consolidated Deficit of the Federal and Provincial Governments, Pakistan, FiLscal 1972/73 to Fiscal 1987/88 Percentage of GNP 10 - 9 B 7 6 5 4 3 1972/73 74/75 76/77 78/79 80/81 82/83 84/85 86/8787/88 Sourc= Country case study. Figure 9.2 shows that inflation performance in Pakistan appears to have been remarkably good, whether measured by the consumer price index (cr1), the wholesale price index (WPI), or the GDP deflator The peak inflation rate during this period, in 1973(74, amounted to 33 percent, as measured by the GDP deflator. After 1976 the inflation rate averaged less than 10 percent per year, falling in recent years to about 5 percent- At the same timne, economidc growth has been robust (figure 9.3), averaging more than 7 percent during the 1980s and never fall- ing below 3 percent per year over the entlire 1972-SB period. As a result, real GNP per capita exhi'bited a continuously rising trend dur- ing this time, with a cumulative increase of about 60 percent?3 Turning to the external sector, figure 9.4 depicts the ratio of the current accounLt to GNP. After a peak de-ficit of more than 8 percent of GMNP in 1974/75, Pakistan achieved a substantial current account adjt.stment, registering a smallI surplus Wy 19-2-i83. Although the country benefited during this period from a substantial increase in external receipts in the form of worker's remiittances (primarily front Pakistani workers employed in Middle Eastern oil-exporting coun- tries), it is noteworthy that this boon contributed to current account adjustmnent rather than to an imnport binge. Expressed as a proportion of exports of goods antd nonfactor services, Pakistan's external debt service ratio remained relatively low, even by Asian standards-about 416 Pakistan: Fiscl Srcstainabi7ity and Macroeconomic Policy Figue 92 Inflation, Pakistan, Fiscal 1972/73 to Fiscal 1987/88 Percent 35 - 30- 25- 20- 10 .5 1972/73 74/75 76/77 78/79 80/81 82/83 84/85 86/87 87/88 Consumer price index - - - - Wholesale price index - CD)P deflator Source: Country case study. 17.5 percent until around 1981182 (see figure 9.5). Expressed in pro- portion to a more relevant measure-exports of goods and services, which indudes remittances-the average debt service ratio for the same period was substantially lower, about 12 percent. This reflects not only the current account adjustment but also Pakistan's access to extemal funds at concessional rates. After 1981182, however, Pakistan's debt service ratio increased sharply, with the higher measure stabilizng at about 23 percent of exports of goods and nonfactor services after 1983/84. As is shown below, this reflects a shift in the composition of external financing toward increased borrowing at market rates. A final notable feature of Pakistan's macroeconomic performance over the period examined concerns the rather unusual behavior of national saving and investment. These variables, expressed in pro- portion to cNP, are plotted in figure 9.6. In spite of a substantial increase in saving and investment rates over the 1974-76 period, levels of saving and investment in Paldstan have been rather low, not only by developing country standards but also in view of the coun- Nadeem Ul Haque and PeterJ. Montid 417 Figure 9.3. Rate of Growth of GDP, Pakistan, Fiscal 1972173 to Fiscal 1987188 Percent 9 - 8 7 - 6- 4- 3- I , 1972173 74/75 76/77 78/79 80/81 82/83 84/85 86/87 87/88 Source Country case study. Figure 9A. Ratio of Current Account to GNP~, Pakistan, Fiscal 1972173 to Fiscal 1987/88 Percent 2- 0 -2 -4. -6 -8 1972173 74/75 76/7 78/79 80/81 82/83 84/85 86/87 87/88 Sou re:- Country case study. 418 Pakistan: Fiscal Sustainab1ityandMacroeconomic Policy Figure 9.5. Ratio of Debt Service, Pakistan, Fiscal 1972/73 to Fiscal 1987/88 Percent 25- 20 i5~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~4 :z5 - 1972/73 74/75 76/77 78/79 80/81 82/83 84/85 86/87 87/88 Debt service/exports of goods and nonfactor services - - - - -Debt service/exports of goods and services Sourae: Country case study. t's high growth rate. The implication is that measured incremental capital-output ratios (icoRs) have been remarkably low (figure 9.7). Fiscal Policy in Pakistan The origin of fiscal deficits in Pakistan is similar in many ways to that in other developing countries. In brief, an upsurge of extemally financed development spending during the early to mid-1970s, pri- marily in the form of investment by public enterprises, proved rela- tively permanent, and the public sector was unable to generate the revenues-either through taxation or from the direct return to the investments undertaken-to dlose the fiscal gap thereby created. For the purposes of this section, it is useful to split the description of fiscal policy in Pakistan into three periods: the period of democratic rule under the Bhutto government (1972173 to 1976177), the early years of the subsequent martial law government (1977178 to 1981182), and 1982183 to 1987188. The section closes with a brief description of events since 1988. Nadeem Ul Haque and Peter J. Montiel 429 Figure 9.6. National Saving and Investment, Pakistan, Fiscal 1972/73 to Fiscal 1987/88 Percentage of GNP 20- 15L-tt 1972/73 74/75 76/77 78/79 80/81 82/83 84/85 86/87 87/88 - - -- Investment Saving Source Country case study. The Bhutto Period The Bhutto government, which took power after the 1971 war with India and the separation of Bangladesh, saw as its mandate the rapid economic and social development of the former West Pakistan- It was given a strong impetus in this direction by the combination of numli- tary defeat and plentiful external financing at concessional terms- primarily from Middle Eastern oil producers, which at that time were reaping the windfalls of the first oil price shock. To an extent that is difficult to quantify. spending during these years was influenced by factors such as flood relief and the attempt to provide a countercyclical offset to the negative output effects of the oil price shock. (Pakistan is a net oil importer.) However, in view of the circumstances described above, the bulk of the spending increase (from 18 percent of GNP in 1972173 to 24 percent in 1975176; see table 9.1) must be understood as the result of a conscious policy choice- that is, as an explicit intention of fiscal policy. During these years the share in GNP of development expenditures-consisting of investment by the federal and provincial governments, capital transfers to local 420 Pakistan Fisal Sustainability and Macroeconomic Policy Figure 9.7. Incremental Capital-Output Ratio, Pakistan, Fiscal 1972173 to Fiscal 1987/88 Ratio 4.5 - 4.0- 3.5- 3.0 - 2.5- 1972/73 74/75 76/77 78/79 80/81 82/83 84/85 86/87 87/88 Source Country case study. governments and state enterprises, and production subsidies-more than doubled. Public sector investment was devoted to the develop- ment of the chemical industry, as well as to cement, fertilizers, engi- neering, petroleum, steel, and production of a vegetable substitute for ghee, or drified butter. This period also saw the nationalization of the seven principal manufacturing and industrial groups, as wel as of banking, insurance, shipping, and educational institutions. Within the category of current expenditres, subsidies (classified as "other"' current expenditure in table 9.1) increased by an average annual rate of 39 percent in the first half of the 1970s. The principal subsidy over the period was on atta (whole wheat flour), which was distnbuted at ration depots, primarily in urban areas, at prices below those required to cover the cost of wheat imports. By the niid-1970s, therefore, the public sector had a significantly larger role in Pakistan than at the beginning of the decade. Unfortunately, this expansion of the role of the public sector was not matched by an corresponding rise in revenues. In fact, although expenditures of the federal and provincal governments increased by more than 6 percent of GNP from 1972W73 to 1975/76, total revenues increased by only 1 percent of cNP during the same period. (As table 9.1 indicates, the gain came primarily from an increase in domestic indirect taxes.)4 The result, of course, was an increase in the fiscal Nadeem Ul Huque and Peter. Mon tiel 421 deficit, to about 5 percent of GNP. Since foreign loans were plentiful and were avaflable at favorable terms, almost three-quarters of the deficit financing during these years was external. Early Years of the Martial Law Government The martial law government that assumed office in July 1977 initiated a short-term stabilization program supported by an IMP standby fad!- ity. This program attempted to address some of the structural prob- lems of the economy and to correct perceived financial imbalances, the most significant of which was the fiscal deficit. An important goal of the new government was to deemphasize the role of the public sector in the economy- This change in regimes coincided with an exteral shock-the sharp curtailment of financial assistance from the United States because of Pakistan's nudear program-and these developments ushered in a period of relative fiscal retrenchment. Between 1976X77 and 1980/81 the fiscal deficit fell from more than 8 percent of GNP to less than 5 percent. The fiscal improvement during this period took the form of both cuts in expenditure and incaeases in revenue. The spending reduc- tions can safely be treated as exogenous policy measures, since they represented an avowed policy goal of the new government. More- over, the bulk of the spending cuts (more than 2 percent of GNP over the period) occurred in development spending, consistent with the govemment's explicit goal of curailing public involvement in pro- ductive activities and leaving these to the private sector. An effort to reduce current expenditures met with only limited success; current spending remained roughly constant as a share of GNP over the period.5 The total contribution of revenue increases to the fiscal adjustment (amounting to 1 percent of GNP) was about half that of spending cuts. However, despite the serious effort to improve tax administration, only about half of the revenue increase appears to have been the result of exogenous fiscal policy measures. To assess the extent to which discretionary revenue measures may have contrbuted to the fiscal adjustment, we report in table 9.2 the results of very simple regressions that relate the three components of total tax revenue (direct taxes, taxes on international trade, and other indirect taxes) to their primazy determinants. In the case of direct taxes and indirect taxes other than trade taxes, we took the primary determinant to be the tax base (proxied by nominal GNP). For trade taxes the determi- nants consist of the tax base (in the form of exports and imports) and the share of workers' remittances in GNP.6 The latter variable greatly improves the fit of the regression. We take it as a proxy for the compo- Table 9.1. Consolidated Accounts of the Federal and Provincial Governments, Pakistan, Fiscal 1972/73 to 1987188 (percentage of GNP) ltem 1972173 1973174 1974175 1975176 1976/77 1977178 1978/79 1979180 1980/81 1981182 1982183 1983/84 1984185 1985/86 1986187 1987188 Total expenditures 17.9 20.9 24.1 24.2 22.7 21.7 23.4 21.6 21.2 20.5 21.6 21.7 22.4 22.8 24.2 24.1. Current 13.4 14.7 15,5 15.0 13.0 13.6 14.6 13.0 12.6 12.8 14.3 15.6 16.1 16,1 17.3 17.4 Consumption 11.4 9.6 10.6 11.4 10.8 10.1 9.7 9.3 . 9.7 9.9 10.5 11.2 11.2 11.3 12.1 12.7 Total interest 1.8 1.8 1.6 1.8 1.8 1,8 1.9 2.0 2,0 2.2 2.8 3.1 3.2 3.4 3.7 3.8 Foreign 1.1 0.9 0.9 0.8 0.9 0.9 1.0 0.9 0.8 0.8 1.1 1.1 1.1 1.1 1.2 1.1 Domestic 0.7 0.9 0.7 1.0 0.9 0.9 0.9 1.1 1.2 1.4 1.7 1.9 2.0 2.3 2.5 2.7 Other 0.2 3.3 3.3 1.8 0.4 1.6 3.0 1.7 1.0 0.7 1.0 1.4 1.8 1.4 1.5 0.9 Development 4.5 6.2 8.6 9.3 9.7 8.1 8.8 8.6 8.6 7.6 7.3 6.1 6.4 6.8 6.9 6.7 Investment 2.0 2.7 3.0 3.3 3.5 2.8 2.9 2.4 2,6 3.1 2.7 2.7 2.7 2.6 2.8 3.0 Other 2.4 3.5 5.6 5.9 6.2 5.4 5.9 6.3 6.0 4.6 4.6 3.4 3.7 4.1 4.1 3.7 Total revenues 14.2 15.7 14.8 15.4 14.4 14.3 15.1 15.8 16.4 15.5 15.2 16.2 15.4 15.8 15.9 16.7 Tax revenue 11.0 11.8 11.4 11.7 11.4 11.5 12.0 12.9 12.9 12.4 12.2 11.6 10.8 10.7 10.2 10.9 Direct taxes 2.2 1.9 1.7 2.1 2.0 1.8 1.9 2.2 2.5 2.6 2.3 2.0 1.9 1.7 1.7 1.9 Indirect taxes 8.8 9.9 9.7 9.6 9.4 9,7 10.1 10.7 10,4 9.8 9.9 9.6 8.9 9.0 8.5 9.0 Export duties 1.6 2.1 0.9 0.6 0.1 0.2 0.1 0.2 0.2 0.1 0.1 0.1 0.1 0.2 0.0 0.2 Import duties 2.3 2.6 3.4 3.3 3.9 4.2 4.7 4.8 4.5 4.2 4.5 4.6 4.4 4.1 4.1 4.2 Other 5.0 5.3 5.4 5.8 5.5 5.3 5.2 5.7 5.7 5.5 5.3 5.0 4.4 4.7 4.4 4.6 Nontax revenue 3.2 3.7 3.3 3.3 2,7 2.6 2.7 2.4 2.7 2.6 2.5 4.0 4.1 4.6 5.2 5.1 Interest receipts 0.8 0.8 0.7 0.9 1.3 1.0 0.9 1.0 1.0 0.9 1.0 0.9 1.0 0.9 1.4 1.2 Other 2.4 2.9 2.6 2.4 1.4 1.6 1.7 1.4 1.7 1.7 1.6 3.1 3.2 3.7 3.9 3.9 Surplus of autonomous bodies. 0.0 0.1 0.2 0.4 0.3 0.3 0.5 0.6 0.7 0.5 0.6 0.6 0.5 0.5 0.4 0.7 Overall deficit 3.7 5.2 9.3 8.8 8.3 7.4 8.3 5.8 4.8 4.9 6.4 5.5 7.1 7.1 8.3 7.4 Bank financing 0.0 0.5 1.8 2.9 3.9 2.6 4.1 2.5 2.0 0.4 1.5 1.4 1.8 3.3 0.9 1.6 State Bank of Pakistan -1.1 -0.6 3.1 0.9 3.7 3.0 3.5 1.2 -0.6 3.0 -1.3 4.5 0.8 0.6 0.9 1.3 Scheduled banks 0.0 1.1 -1.3 2.1 0.2 -0.4 0.6 1.3 2.5 -2.7 2.8 -3.2 1.0 2.7 0.0 0.3 External financing 3.2 3.7 7.0 5.1 3.8 3.3 3.2 2.8 2.6 1.5 1.3 1.1 1.0 1.5 1.8 1.6 Domestic nonbank 0.9 1.1 0.5 1.4 0.6 1.5 1.0 0.6 0.3 3.0 3.6 3.0 4.3 2.3 5.7 4.2 Menwioranidutin itemtis: Deficit/cop (percent) 4.60 6.96 12.00 10.59 9.84 9.28 9.82 7.52 6.20 6.19 7.97 6.91 8.73 8.64 9.81 5.70 Composition of deficit financing 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 Bank financing 0.00 0.09 0.20 0.33 0.47 0.36 0.49 0.43 0.40 0.08 0.23 0.25 0.25 0.47 0.11 0.21 External financing 0.87 0.70 0.75 0.57 0.46 0.44 0.39 0.47 0.53 0.31 0.20 0.20 0.14 0.21 0.21 0.22 Domestic nonbank 0.23 0.20 0.05 0.15 0.08 0.20 0.12 0.10 0.07 0.61 0.57 0.55 0.61 0.33 0.68 0.57 Note: Items may not sum to totals because of rounding. Source:. Pakistani Economik Sumey, various issues. 424 Pakistan: Fiscal Sustainability and Macroeconomkc Policy Table 9.2. Determi ts of Tax Revenues, Pakistan Direct taxes Trade taxes Other Item (7) (2) indirect taxes Constant -3.72 -0.47 -1.93 (6.35) (-0.76) (-3.13) GNP 0.98 0.93 0.92 (21.00) (29.80) (20.33) Inflation rate -0.22 (-0.32) Ratio of imports to 0.50 GNP (3.58) Ratio of exports to 0.21. GNP 12) Ra;io of remittances 0.14 tO GNP (3.22) R2 - 0.97 0.99 0.99 Note: All variables except the inflation rate are in log form. The data are annual, and the regressions are estimated for the period 1972173 to 1987188. Figures in parentheses are t-statistics. A blank denotes the omission of the specific variable from the regression. sition of imports, on the hypothesis that an increase in remittances increases the share in total imports of dutiable imports such as con- sumer durables. These simple regressions account for almost all of the variation in tax receipts. When their residuals are taken as measures of discretion- ary tax changes, total discretionary tax measures account for about 5.5 percent of total tax receipts in 1979/80 and about 7.5 percent in 1980181 and in 1981/82. At their peak, these increases in discretionary revenue amounted to less than one-sixth of the total increase in tax revenues. The remainder of the increase in tax revenues is accounted for by an increase in trade taxes; both exports and imports rose rap- idly during these years. A real exchange rate depreciation fueled by the depreciation of the U.S. dollar against the currencies of Pakistan's trading partners in the late 1970s gave a boost to exports, while a substantial increase in remittances as a consequence of the second oil shock gave rise to an import boom at the end of the 1970s. The Period 1982183 to 1987/88 The imnprovements mn Pakistan's fiscal stance did nct turn out to be permanent. After 1981182 the fiscal deficit began to increase once Nadeem U HaqueandPeterj. Montie? 425 again and by 1986/87 had reached 8.5 percent of GNP, a level compara- ble to those of the mid-1970s. Although the composition of revenues changed somewhat during the period-slippage in the collection of direct taxes and of domestic indirect taxes tended to be offset by increased nontax revenues, primarily in the form of profits on the distribution of oil products7-the share of public sector revenue in GNP showed no trend. The increase in the deficit arose from the expenditure side. This was so in spite of a continued contraction in the share of development expenditures in GNP; this type of spending peaked at 9.75 percent Of GNP in 1976177 and then declined to 8.5 percent in 1980181 and to 6.75 percent in 1987/88. Part of the increase in spend- ing was from higher public consumption in the form of public sector wages and salaries and defense spending. The share of defense spending in total public expenditure rose from 21 percent in 1978179 to 27 percent in 1984185 (see Kemal 1987). But by far the most rapidly increasing category of spending during the 1980s was total public sector interest payments. The increase in the share of interest payments reflects a conscious change in the composition of deficit financing after 1980181. In an effort to keep inflation in check and to tap directly what was per- ceived to be a plentiful supply of private saving originating with remittance inflows, the government limited its borrowing from the domestic banking system after 1980181, and domestic nonbaank bor- rowing became the residual source of finance. When the interna- tional debt crisis curtailed the availability of external financing in 1981182, external funds were also replaced by domestic nonbank borrowing. Thus the combination of an aversion to inflationary finance, reduced availabflity of extemal funds, and increased public consumption brought about a substantial increase in domestic non- bank borrowing. As wfll be shown later, the rising stock of internal debt could be absorbed domestically only by offering higher interest rates, and the combination of higher debt stock and increasing inter- est rates caused interest payments to mount over time. Total public sector interest payments, which were 2 percent of GNP in 1980/81, had almost doubled as a percentage of GNP by 1987/88, accounting for about two-thirds of the increase in the deficit-to-GNP ratio over that period. For any given year, therefore, this component of the deficit increase reflected past financing decisions rather than current policy In summary, the upsurge in fiscal deficits in Pakdstan during the 1980s was the result of two policy choices: an increase in public con- sumption in the face of a political inability to raise commensurate revenues, and a change in the financng mix from domestic bank borrowing and external financing to domestic nonbank borrowing. 426 Pakistan: Fiscal Sustainability and Macroeconomic Policy Recent Developments As indicated earlier, recent revisions in Pakistan's macroeconomic data complicate any attempt to present a cohesive picture of the entire period 1972-92. Data after 1987/88 are compiled under a different methodology than was used earlier and therefore are not directly comparable. Nevertheless, recent developments can be described by using 1987188 as a benchmark and focusing on changes since that time. The most significant feature of the recent period is a notable improvement in the fiscal picture between 1988,89 and 1990191. As a fraction of GNP, the fiscal deficit decined from about 8.5 percent in 1987/88 (as against 8.75 percent under the earlier methodology) to about 5.75 percent in 1990/91. Most of the adjustment came from the expenditure side, as total spending of the federal and provincial gov- ernments contracted by about 2 percent of GNP. Three-quarters of this adjustment was achieved by restrining current expenditures, offer- ing some hope for a permanent fiscal adjustment. However, in 1991192 the government appears to have missed its fiscal deficit target of 5 percent of GDP by a wide margin, and recent figures suggest a return to the range of 7 to 8 percent that characterized the period up to 1987188. The period of fiscal retrenchment was accompanied by a growth slowdown. The growth rate of real coP, which amounted to 6.5 per- cent in 1987/88 under the new methodology, slowed to 4.75 percent in 1988/89 and 1989/90 before rising to more than 5.5 percent in 1990191. Preliminary figures for 1991)92 suggest that growth has again risen above 6 percent. It would be simplistic, however, to attribute the growth slowdown exclusively to the fiscal retrenchment, since Pakistan was buffeted by several shocks during these years. Not the least of these shocks were a continuous and severe contraction in worker remittances from abroad after 1986/87 (associated with low oil prices and war-related dislocations in the Middle East) and political uncertainty at home, culminating in 1990 in the replacement of the government of Benazir Bhutto by that of Nawaz Sharif. Other broad macroeconomic indicators behaved in a more stable fashion after tiis period. The rate of inflation has remained low (in the 10 percent range, as measured by the cpi), and the ratios of the current account of the balance of payments to GNP and of debt service to exports of goods and nonfactor services have remained broadly unchanged since 1990. Overall, recent developments appear to be consistent with the experience of the 1980s. The latest data suggest that no permanent change was achieved in Pakistan's fiscal policy under the two demo- cratic governments that followed the end of martial law, and the Nadeem Ul Haque and Peterl. Montiel 427 country's macroeconomic performance reflects a continuation of pre- vious trends. Deficits and Inflation As indicated previously, Paldstan has operated for the better part of two decades with fiscal deficits that, by international standards, are quite large in relation to GNP. In other developing countries fiscal deficits of smaller magnitude have been blamed for a number of adverse macroeconomic developments, chief among them being a high rate of inflation. By contrast, Pakistan has performed relatively well in a maaoeconomic sense, with a high average rate of eco.aomic growth, low inflation, and a relative absence of major exteral imbal- ances. The key questions that arise in connection with Pakistan's fiscal policy are thus the following: What macroeconomic effects have Pakistan's sustained high fiscal deficits had on its economy? Why in Pakistan have high deficits not been associated with inflation of Latin American proportions? In popular discussions, the lirik between deficts and inflation in developing countries is quite direct. In the absence of secondary secu- rities markets, open-market operations are not an important mone- tary policy tool in such countries. Since government borrowing from the central bank expands the supply of base money, the rate of growth of the money supply is taken to depend primarily on the size of the fiscal deficit. With the rate of inflation in tum being determined by the rate of growth of the money supply, the link between deficits and inflation follows. Although variations in velocity, the availability of other modes of financing, and several other factors tend to complicate matters, there is nevertheless a valid long-run relationship, emerging from the solvency constraint of the public sector, between fiscal deficits and inflation. This is easiest to show formally. (The discussion that follows draws heavily on Buiter 1985.) Let b denote the real stock of the debt of the public sector (induding the central bank) to the domestic private sector. Let FG denote public external debt; d, the real primary fiscal deficit; and m, the real stock of base money, al measured as ratios to GNP. Also, let rB denote the real interest rate on domestic borrowing; rF, the real interest rate paid on foreign debt; rt, the real interest rate prevailing in international capital mar- kets (external interest rate plus rate of depreciation minus domestic inflation); n, the rate of growth of real CNP; and s, the nominal exchange rate. The public sector's budget constraint can then be written: 428 Pakistan: Fiscal Sustainability and Macroeconoomic Policy (9.1) b + sYG + m = d + (r, - n)b + (rF+& -n)sfc-(r + n)m where -r is the rate of inflation. (A dot [I] over a variable denotes a time derivative; a hat [^] denotes a proportional rate of change.) This can be transformed into: (9.2) b + sf0 = J + (r-n)(b + sfG)-[r + (r + n)m] where J = d + (TB-rt)b + (rF + - r*)sffc is the adjusted primary deficit-that is, the primary deficit, plus the excess interest paid on domestic debt and foreign debt over that prevailing in international capital markets. The initial net worth of the public sector is given by -(b + sfG), and the public sector will be solvent if the present value (calculated using the growth-corrected interest rate, r - n) of its anticpated future debt service is at least equal to its net debt; that is: (9.3) PViz + (r + n)m-d] b + SfG- The resources available to the public sector for servicing debt consist of future seigniorage revenue, given by m1 + (-r + n)m, and future adjusted primary surpluses, given by -d. Note that, other things being equal, an increase in the present value of the stream of future deficts requires an increase in the present value of the inflation tax, m. It is in this present-value sense that higher fiscal deficits are related to higher inflation. Note also that a number of factors influence the present value of the inflation tax associated with a given path of the primary deficit. The following observations are germane: * The relevant value of the primary deficit is the adjusted deficit- that is, the deficit adjusted to take account of differences between the interest rate on domestic borrowing and the actual interest paid on external borrowing, on the one hand, and the marginal cost of extemal funds, on the other. Access to domestic or exter- nal funds at favorable rates reduces the present value of the inflation tax associated with a given unadjusted deficit. * The amount of seigniorage required to finance a given path of the adjusted deficit is smaller, the smaller is the initial net stock of debt. * Given the amount of seigniorage required, the larger are the rates of growth of output and the greater is the secular growth in the money-to-income ratio, the smaller is the requisite inflation tax. Application to Pakistan During i980-88 the consolidated defict of Pakistan's federal and pro- vincial governments averaged about 6.5 percent of GNP (see table 9.1). Nadeem UlIHaque and Peter). Montid 429 Figure 9.8. inflation and Growth of Base Money, Pakdstan, Fiscal 1980/81 to Fiscal 1987188 Percent 30 - 25- 20 - 15 - 10 U-- - - I - _ I 0- 1980/81 1981/82 1982/83 1983184 1984/85 1985/86 1986/871987/88 Growth oF base money hInflation Sourrce: Country case study. Since the 1980 stock of base money amounted to about 12 percent of cNP, financing of this deficit solely through the issuance of base money would have implied an inflation rate of more than 54 percent per year during the 1980s, even before allowing for erosion of the monetary base through a rise in velocity. After allowing for such erosion, reliance on the inflation tax to finance a deficit of this magni- tude would imply inflation rates of Latin American proportions. The actual growth of base money and domestic prices (measured by the GDP deflator) in Pakistan is depicted in figure 9.8. During 1980-88 both the growth of base money and the domestic inflation rate fell substantially short of what the simple analysis would predict. The observations of the preceding section can be used to explain why. Rearranging equation 9.2 permits us to write: (9.4) + i~ + m t ah + n)(b + sfc + m) - gsf<: =dA where d = d + (r0 + ar)b + (rF + -r)sfc is the conventional deficit-MNP ratio. The variable dA represents the inflation- and growth-adjusted 430 Pakiistan: Fiscl Sustainability and Macroeconomic Policy Figure 9.9. Actual and Equilibrium Deficits as a Share of GNP, pakistan, Fiscal 1980/81 to Fiscal 1987)88 Percentage of cNP 8.5 - _ - 8.0 - 4% 7.5 - - 6.0~ -r t- 5-5 - 7.0 _ 1980/81 1981J82 198P2/83 19?83184 1984,J85 1985186 1986/871987>J88 - - - -Actual deficits, Equobrium deficits Soutrce: Countr case study. deficit ratio-tat is, the actual deficit minus the portion that can be financed without altering the ratio of total debt bo- GNP- When the valxiue of this expression is 0, the conventional deficit can be accommo- dlated without reqiig macroecononiic adjustments-includig adjustment of the rate of inflation-because the requlisite amount of financing willpresumably be forthcoming. In other worcls, d = [(7r + n)(b + sfG + m) - f;G] is the deficit ratio that is consistent with macroeconomidc equMliriumn, with inflation rate Xr and real growthi rate n. Figure 9.9 depicts the ratios of the actual defidt, d, and estiated deequflinrum" defid:t, d*, for Pakidstan in the years 1980-87, using smnoothed values of the g;rowth rate, n. and the rate of inflation, ir, and setting th:e rate of exchange rate depreciation equal to Tr to calcu- late the equiIibrium deficit- The equfi-brum deficit estimate should be seen as the midpoint of a range of estimates. It is based on our estimiate of the end-1979J8 stock of net (interest-bearing and non- interest-bearing) public sector debt, as well as on the assumption that all such debt was wilhingly held.8 As figure 9-9 shows,the equw- hllbium deficit was indeed at a relatively high value for Pakistan dur- ng this penod (about 5.5 percent of GNP, on average), in spite of an Nadeem Ul Haque and Peterj. Montiel 431 Figure 9.10. Public Sector Liabilities as a Share of GNP, Pakistan, Fiscal 1980181 to Fiscal 1987188 Percentage of GNP 70 - /__ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 65 60 55 50 45 1980/81 1981/82 1982/83 1983/84 1984185 1985/86 1986/871987/88 Source Country case study. inflation rate (for the GDP deflator) averaging a little more than 7 percent. This is primarily attrbutable to a very high rate of growth of real output (about 6 percent per year), which permitted a fairly rapid expansion of both interest-bearing and non-interest-bearing debt without recourse to inflationary finance. The ratio of the actual deficit to C;NP fell below this equilbrium value for the first two years of the period 1980181 to 1987188 (see figure 9.9). For the remainder of the period the deficit averaged about 7 percent of GNP. Thus the adjusted deficit was significantly smaller than the actual defict during the perod; it amounted to about 1.5 percent of GNP, on average. It is not surprsing, therefore, that, as indicated in figure 9l0, the ratio to cNP of the liabilities of Pakistan's consolidated public sector (consisting of base money, domestic debt, and external debt) was fairly stable for the first half of the 1980s, in spite of a fiscal deficit amounting to almost 6 percent of GNP. Only in the last two years covered by the analysis (1986187 to 1987188) did this ratio increase rapidly, consistent with a substantial increase in the adjusted deficit during the second half of the 1980s. In short, this analysis suggests that low inflation can be reconciled with large deficits in Paldstan because of the economy's very high growth rate. Rapid economic expansion made it possible to finance 432 Pakistan: Fiscal Sustainability and MzcroEconomkc Policy large fiscal deficits by issuing debt, without extensive reliance on the inflation tax. Because of the economy's high growth rate, the relation- ship of the stock of public sector liabilities to ONP remained fairly stable until recent years in spite of the large fiscal deficits recorded during the 1980s. Some Qualifications Although the foregoing analysis follows fairly conventional lines, its application to Pakistan-and possibly to many other low-income devel- oping countries-is Problematic. The reason is that the calculation of an equilbrium deficit is done from the financing side of the budget on the assumption that if private agents willinly hold an initial stock of daims on the public sectot they would be wiling to expand the real value of daims at the rate of growth of real output. Complications arise wher. the initial stock of debt is not in fact willingly held-by private agents on market terms. It woiud then be misleading to assume tat the market would be wiling to accept a steady growth of dains on the public sector under prevailing macroeconomic conditions--that is, without disrupting macroeconomic equilibrium. In Paldstan ;s case, as with many other developing countries, there are at least two reasons why the stock of public sector debt cannot be treated as wilingly held by market agents. First, as pointed out above, much of Pakistan's external debt was acquired at concessional terms on a bilateral basis and thus contains a substantial grant ele- ment. Second, some of Paksas domestic debt is held by- financial institutions, parly to satisfy reserve requirements. During much of the period with which we are concerned, for example, commercial banks faced a required "liquidity" ratio of 30 percent, which had to be satisfied with government securities (see Morshed 1987)_ To the extent that such institutions would have required a higher rate of return to hold these securities willingly, this requirement subjects them to an implicit tax. As figure 9.11 illustrates, interest rates paid by Pakistan on both external and domestic debt were much below inter- national interest rates (measured by the London interbarnk offered rate, or uBOR) during most of this period. It cannot be safely assumed tat either set of creditors-foreign governments or domestic financial iisdtutions-would be willing to see their claims on the government of Pakistan expand in real terms at the rate of growth of domestic output. To the extent that such an assumption is unwarranted, the equilibrium deficit reported above would be overestimated, and thus the inflation- and growth-adjusted deficit would be underestimated. . To assess the potential importance of these factors, we have per- formed some rough calculations to correct our estimate of dt for non- market lending. To do so, we require estimates of the total amount of Nadeem LUt Haque and PeterJ. Mantiel 433 Figure 911. Interest Rates on Domestic and External Debt, Pakistan, Fiscal 1980(81 to Fiscal 1987188 Percent 16 14 12 - 10 - 8 - _- -- - 6__ 4 2 . . . _ _ .......... ----- 0 | I 1980/81 1981/82 1982/83 1983/84 1984/85 1985/86 1986/87 1987/88 London interbank offered rate (lIBOR) - - - - Domestic debt -- External idebt Source Country case study. claims on the Pakistan government that domestic and foreign market agents would have willingly held initially (that is, at the beging of fiscal 1980181)- We assume that to hold this debt wllingly extemal agents would have required a return equal to L1BOR and that domestic agents would have required an interest rate corresponding to uncov- ered interest parity (uIP), using uBOR as the foreign counterpart rate. The proportional differences between the average values of these rates and the average interest rates paid by the government on its external and on its domestic debt during the 1980(81 to 1987(88 period were taken as estimates of the grant element associated with foreign loans and the implicit tax rate on domestic securities, respectively. Applying these proportions to the debt stocks outstanding at the end of 1980/81 yields our estimates of the claims that would have been wfllngly held by market agents. Using these as the basis for our equilibrium deficit calculations yields a corrected equilibrium deficit, denoted dc, of 2.1 percent of GNP for the period 1980/81 to 1987188, reflecting a scaling down in the estimate of voluntary lending from market agents if there is no change in prevailing macaoeconomic conditions. 434 Pakistait Fiscal Sustainabilily and Macroeconomic Policy Figure 9.12. Actual and Corrected Deficits as Shares of GNP, Palkistan, Fiscal 1981182 to Fiscal 1987/88 Percentage of GNP 9- 8 7 - 6 5 4 - 1981/82 1982183 1983/84 1984/85 1985/86 1986/87 1987/88 Actual deficit, d Corrected inflation- and growth-adjusted deficit, d0 - - - - Corrected equilibrium deficit dc Soure: Country case study- It would be incorrect, however, to recalculate the adjusted deficit as d- d, because Palistan was in fact able to borrow at favorable rates during the 1980s. In other words, the grant component of foreign lending and the implicit tax associated with domestic borrowing would in any event have reduced the govenmment's need to raise funds at market rates. In essence, these fumds simply represent unconventional sources of revenue. We therefore corrected the acual deficit by adding as revenues the grant component of foreign borrow- ing and the imnplicit tax component on domestic borrowing.9 Because this step essentially involves redassifying financing as revenue entries, the corrected actual deficit, d, is substantially reduced in relation to d. The corrected inflation and growth-adjusted deficit, calculated as d9L = c-d, as wel as d3 andd*, are plotted inL figure 912. As is dear from the figure, these corrections decrease the size of both the actual and equilbnum deficits but have little effect on the difference between them. In other wards, the corrected deficit is not substan- tially affected. Whereas our previous estimates may exaggerate the Nadeem Ul Haque and Peterj. Montiel 435 Figure 9.13. Composition of Deficit Financing, Pakistan, Fiscal 1980/81 to Fiscal 1987188 Percentage of GNP 9- 8 -/ 7 6 4 3 2 0- 1980/81 1981/82 1982/83 1983/84 1984/85 1985/86 1986/871987/88 Total deficit 4--- Domestic financing External financing Source Country case study. amount of financing that private agents would have been willing to sup- ply to the government during the 1980s, they also exaggerate-by an approximately equal amount-the size of the deficts that required financing. Our previous conclusions and their implications for the mac- roeconomic effects of Pakistan's fiscal deficits therefore continue to hold. Conclusions As shown in figures 9.9 through 9-12, fiscal deficits in the second half of the 1980s have begun to substantially exceed the equilibrium values calculated for 1980-87. Deficts of such magnitude can indeed be expected to exert significant effects on financial markets. However, even these higher deficits have not been associated with an inflation- ary upsurge. An explanation for this is suggested in figure 9.13, which shows how recent deficits have been financed. The height of each curve measures the amount of financing from specific sources. The lowest curve measures the flow of external financing, the middle curve adds to this domestic financing, and the top curve, which rep- resents the total defict, adds money financing. As is evident from the figure, recent years have witnessed a rapid increase in domestic bor- rowing (the gap between the lowest curve and the miiddle one). As a 436 Pakistan: Fiscal Sustainability and Macroeconomic Policy Figure 9.14. Base Money, Domestic Debt, and External Debt as Shares of GNP, Pakistan, Fiscal 1980181 to Fiscal 1987/88 Percentage of CNP 35 - 30 - 25 -................. ....................................... 20 - _ ' a. 15- _ - 10 - b_ ~ ~ ~~ ~~ ~~ ~~~~~~ , I* , - . 1980/81 1981/82 1982/83 1983/84 1984/85 1935/86 1986/87 1987/88 -- ..-- External debt -Domestic debt Base S-ource: Country case study. result, while the ratios of base money and external debt to cNp have remained roughly stable, the ratio of domestic debt to GDP has risen noticeably (figure 9.14). The macroeconomic effects of this financing policy are investigated in the next section. To sunnarize, a simple explanation linking fiscal deficits to inflatio fails to hold in Pakistan. Three reasons emerge. First, the measured deficit overstates the "true," economiicaly meaningful deficit because foreign financing contained a substantial grant element and domestic financing contained an implicit tax element that should be treated as above the line in the general government budget. This aspect of deficit financing in Pakistan reduces the pressure exerted by fiscal deficits on financial mar- kets. Second, the economy grew very rapidly during the 1980s, so that equilibrium defidits proved extremely large by internatinal standards. Third, when the corrected inflation and growth-adjusted defiits in fact became large in the second half of the 1980s, primary reliance was placed on domestic debt financing rather than on the iflation tax. Nevertheless, defiits of the magnitude experienced during the second half of the 1980s can be expected to have other maacoeconomiuc effects, and these are addressed in the remainder of the chapter. Nadeern Ul Haque and Peter]. Montiel 437 Figure 9.15. Private and Public Investment, Pakistan, Fiscal 1964/65 to Fiscal 1987188 Percentage of GNP 13 12- 10 9 8 ¾.- 7 -. . 6- -5- 1964/65 68/69. 72/73 76/77 80/81 84/185 87/88 66/67 70/71 74/75 78/79 82/83 86/87 Private investment Public investment Source. Country case study. The Effects of Fiscal Policy on Economic Behavior This section examines the effects of fiscal policy in general (rather than fiscal deficits spedfically) on the behavior of economic agents in Pakistan. Recent research has shown that both the investment and the consumption decisions of economic agents may be directly affected by policy variables such as government consumption and investnent. Such decisions also are known to respond to financial variables that are themselves affected by the mode of government financing. Consumption and Investment As was shown in the Overview to this chapter, total investment as a ratio of GNP in Pakistan has averaged about 15 percent. Figure 9.15 shows the separate behavior of the ratio of public and private invest- ment to GNP over the period 1964/65 to 1986/87. The strong invest- ment drive initiated by the Bhutto government in 1973 is dearly dis- cernible, and it is evident that private investment, although it 438 Pakistan: Fiscal Sustainability and Macroeconomic Policy Figure 9.16. Private and Public Consumption, Pakistan, Fiscal 1964/65 to Fiscal 1987/88 Private consumption Public consumption (percentage of GNP) (percentage of GNP) 85 >--15 14 80 13 ..0 70 f 90 68/69 I 72/73 I 76/77 8 80/81 1 84/85 1 87/88 66/67 70/71 74/75 78/79 82/83 86/87 Public consumption Private consumption Source: Country case study. mcreased steadily after 1973174, did not return to the high levels that it had attained prior to the period under study. Thus, since 1972/73 public investment has represented the bulk of total investment. Whether the high level of public investment has served to catalyze private investment since 1973174 by means of infrastructural develop- ment or whether private investment was aowded out during most of the period by the elevated level of public investment is an issue that will be examined econometrically. Figure 9.16 shows the movements of private and public consump- tion as a ratio to GNP between 1963164 and 1987/88. The first signifi- cant point to note is that Pakstan has exhibited a fairly high average propensity to consume: the ratio of total consumption to GNP aver- aged about 85 percent per year over the period. It may be suspected that the high fiscal deficit played a role in producing this result. However, the data also suggest an underlying relationship between public and private consumption. As public consumption inaeases, private consumption appears to decine. Such compensating behav- ior would appear to lend credence in the case of Pakistan to the Ricardian view that there is a tendency for aggregate consumption to NAdeem Ul Haque and PeterJ. Montiel 439 Table 93. Tests for Unit Roots, Paldstan: Consumption Augmented Indicator Durbin-Watson Dickey-FdlIr Dickey-Fuller Consumption 0.088 -0.629 -0.427 Disposable income 0.094 -0.086 1.049 Government consumption 0,196 -0.843 0.996 Govemment deficit 0.268 -0.196 -3.052 Permanent income 0.082 0.316 1.621 Permanent public sector saving 0.736 -1.529 -1.19 Note: All variables are logs of real per capita values. Critical values are, for the cointegrating residuals Durbin-Watson (cRnw) test, 1.1, and for the Dickey-Fufler test, -2.61. maintain its level regardless of the level of public consumption (see David and Scadding 1974; Haque and Montiel 1989). CONSUMPTION. The estimation of the consumption function was conducted with annual data for the period 1963-87 derived from var- ious issues of the Pakistan Economic Survey. We assume that consump- tion can be explained by permanent income, disposable income, and possibly several fiscal variables (public consumption, the fiscal deficit, and permanent public sector saving). Disposable income (denoted yD in per capita terms) is defined as GNP minus total tax revenue. Public sector saving (PBs) is defined as govemment revenues minus current expenditures, which include interest payments on government debt. Preliminary investigations revealed that the disposable income process could be represented by an autoregressive moving average (ARMA) (1, 1) specification. Consequently, permanent income per capita, yP, was generated as the predicted value of the following equation: (9.5) yg = -2730.86 + 1.07ytd - 0.26MA(1). Before estimating the consumption function, unit root tests were conducted on the aggregate variables that were to be included in the consumption function. Table 9.3 presents the results of the unit root tests for consumption, disposable income, general government con- sumption, the public sector deficit, permanent income, and the per- manent public sector deficit. Test results for the Durbin-Watson, Dickey-Fuller, and augmented Dickey-Fuller tests, as well as the criti- cal values for the tests, are presented in the table. The null hypothesis that a unit root is present was rejected only in the case of the public sector deficit (by the augmented Dickey-Fuller test). Since the pres- ence of unit roots in almost all the variables suggests that conven- tional estimation approaches may yield misleading results, an alter- native approach, which resulted in an error-correction specification, was used. 440 Pakistan: Fiscal Sustainability and Macroeconomic Policy The cointegrating regression, which captures long-run equilibrium relationships between consumption and some of its determinants, was based on the permanent income approach, incorporating both the income variables mentioned above and intertemporal relative price variables such as the rate of interest and the rate of inflation. Additional fiscal policy variables-as indicated above-were also included. Instrumental variables were used for estimation to contend with endogeneity issues. The instruments used included the lagged values of permanent income, the government deficit, government consumption, and the rate of inflation. The estimated cointegrating equation is as follows (with t-ratios in parentheses): (9.6) c4 = -045 + 1.35ytP - 0.56ct - 1.11rt (-0.41) (6.33) (-2.86) (-2.83) R2 = 0.93; Durbin-Watson statistic = 1.8 where cr is private consumption, ytP is permanent income obtained using equation 9.5, and cs is government expenditure, all measured in log real per capita terms, while r is the annual inflation rate mea- sured by the cn. The equation appears to fit well-it explains 93 percent of the variation in the log of private consumption per capita- and, as the Durbin-Watson statistic shows, the residuals are non- autocorrelated. The signs and magnitudes of the coefficients are in keeping with economic theory. The coefficient of permanent income, for example, is not significantly different from unity. Increases in government consumption do appear to lead to a reduction in private consumption, as do increases in the rate of inflation. We could not find significant roles for the other fiscal variables listed above or for the real interest rate. Using the residuals from the cointegrating regression in equation 9.6 as the error-correction term (denoted cc), an error-correction speci- fication of the consumption function produced the following result (see Davidson and others 1978): (9.7) A{ = 0.23 + 0.19At + 058Ay - 0-4A14! - 0.22ect-1. (1.40) (1.10) (2.45) (-327) (-1.67) R2 = 0.93; Durbin-Watson statistic = 2.01 Instrumental variables induded the error-correction term, govern- ment consumption, lagged inflation, lagged disposable income, and lagged permanent income. Judging by the usual criteria, the equation seems to fit welL The error-correction term, ec-l, is significant at the 10 percent level, suggesting that private consumption does not adjust fully to its long-ran desired level in the first period. Only a fifth of the total adjustment is completed in the first year. The negative and sig- nificant coefficient of Ac, shows that even in the short run, increases Nadeem Ul Haque and PeterJ. Montiel 441 in government consumption are negatively related to private con- sumnption. The coefficient of the change in disposable income is posi- tive and statistically significant, suggesting that consumption behav- ior may be influenced by liquidity constraints. The evidence suggests that fiscal policy may have affected private consumption in Pakistan primarfly through direct substitutability between private and public consumption, trough tax policy, and through indirect effects operating through macroeconomic variables such as the level of real income and the rate of inflation. INvEsTMErN. The private investment equation was estimated using annual data for the period 1972173 to 1987188 from the Pakistan Eco- nomic Suroey- Capital stock series for both the public and private sectors, denoted Kg and KPO, respectively, were constructed using an initial (1971172) economywide capital-output ratio of 2, an initial share of 30 percent for public capital stock, and depredation rates of 10 percent overall and 5 percent for public capital stock. A rental cost of capital (:) series was constructed by dividing the product of the real rate of interest and the investment deflator by the GDP deflator. We take private investment in Pakistan to be determined by the sizes of the private and public capital stocks, the le-vel of real output, and the rental rate on capital- We could find no evidence of a credit availabil- ity effect, in spite of the repressed financial conditions prevailing in the economy for most of the period. Unit root tests were again con- ducted for the level variables and are presented in table 9.4- The tests suggest that unit roots cannot be ruled out for most of the relevant variables. Consequently, the approach adopted in ftis case was simi-- lar to that for private consumption. The cointegrating regression was estimated as follows: (9-8) KPIY = -0.07 - 1.26rK + 2.09KGJY - 0.09DUM (-0.25) (-5.33) (3.78) (-2.79) R2 = 0.90; Durbin-Watson statistc = 2.26 where DUM is a dummy variable for the immediate post-Bhutto (1977- 81) period.10 The estimates support the hypothesis that the government capital stock is positively correlated with private sector capital accumulation. The infrastructural buildup brought about by government investment appnrs to facilitate private investment. At the same time, an increased real rental cost of capital depresses private investment. The remaining variables did not prove statistically significant. Once again. the residuals from the cointegrating regression were used in computing the tests for cointegration. For our cointegrating regression, the cointegrating residuals Durbin-Watson (CRDW), the Dickey-Fuller, and the augmented Dickey-Fuller statistics were esti- 442 Pakistan: Fiscal Sustainability and Macroeconomic Policy Table 9.4. Tests for Unit Roots, Pakistan: Investment Ausmented Test Durbin-Watson Diekey-Fuller Dickey-Fuller Private capital stock 0.026 5.31 -1-41 Govemrnment capital stock 0.099 -1.74 Real rate of interest 0.51 -1.51 -1.33 Real output 0.15 -0.29 -0.65 - Not available. Note. See note to table 9.3. mated to be 1.8, -4.85, and -3.93, respectively. These values imply that the null hypothesis of unit roots in the residuals can be rejected and that equation 9.8 is a cointegrating form. Consequently, an error- correction specification is warranted in this case as well. The results for the error-correction estimation are:1" (9.9) A(KPIY) =0-0)2 - 0.27ec- + 0.90A(K;/Y) (-4.10) (-1-53) (1.94) O.lAr-K .O3ADUM. (-O-71) (-2.20) R2 = 0.45; Durbin-Watson statistic = 1.64 This equation obviously leaves much of the variation in the ratio of the capital stock to output unexplained, and some of the individual coefficients are not estimated very precisely. The coefficient of the error-correction term ec-1 suggests that the private capital stock adjusts slowly to its long-run desired level. The positive and signifi- cant coefficient of A(KG/Y,) indicates that even in the short run, an increase in the government capital stock or a positive level of govern- ment investment induces an increase in private investment. Con- sistent with the theoretical hypothesis, the rental return on capital is negatively related to private investment. Our examination of investment behavior, therefore, shows that fis- cal policy has both direct and indirect effects on private investment in Pakistan. The direct effect appears to operate by expanding domestic infrastructure through public investment. More indirect effects are traceable through the effects of fiscal policy on interest rates. Output To complete the simulation model used in the next section, a specif- cation of the determinants of real output growth is required. Produc- tion is assumed to follow a Cobb-Douglas technology with three Nadeem UL Haque and PeterJ. Montidl 443 inputs: the two capital stocks-govemment ancd private-and labor. Using population as a proxy for labor and assuming constant returs to scale, the production function was estimated in per capita form as: (9.10) y = 5.838 + 0.076ks + 0.268kp + 0.82MA(l) (17.732) (2.304) (13.048) (3.842) R2 = 0.90; Durbin-Watson statistic = 1.87 where kg and kP denote per capita values of the public and private capital stocks. As expected, the coefficients of both the public and the private capital stocks are significant and positive. The magnitudes of the coefficients are also reasonable, suggesting that the total share of capital in output is about 30 percent, which leaves 70 percent of total output as the share of labor. Financial Sector In order to model the domestic financial sector, it is assumed that the household sector's total financial assets (denoted as A) consist of domestic currency (C), domestic deposits (D), government bonds (BP), and foreign currency assets (F"). Household financial wealth (W) therefore consists of the holdings of these four assets minus house- hold debt to the banking system (LpB): (9.11) W = A-Lpc where (9.12) A=C+D+BP+sFP. Since these assets are substitutes in individual portfolios, individual demand for each asset is a function of the asset's own rate of return as well as of the returns available on other assets.2 Domestic currency, which pays no interest, is demanded for transaction purposes. All other assets are assumed to be held for portfolio reasons. The total size of the portfolio to be allocated among these remaining assets thus consists of financial wealth net of currency plus credit from the bank- ing system. As is well known, in this framework only three asset demand equations need be estimated, since the adding-up constraint (equation 9.12) yields the properties of the demand for the remaning asset. The demand for currency was specified as a function of the nominal interest rate on deposits-the dosest substitute for currency-and income (real GNP). Demand for domestic goverrnent bonds as a share of the allocable portfolio was taken to be a function of the rate of return on those bonds, as well as on competing assets- the nominal rate of interest on domestic deposits and the return on foreign assets (that is, the foreign market interest rate corrected for changes in the exchange rate)-?3 The household asset demand system 444 Pakistan: Fisal Sustainability and Macroeconomic Policy Table 9.5. Financial Asset Demand Functions, Paldstan Log of the Mtio of Log of the ratio of curency in Log of the ratio of domestic deposits circulation to domestitpublic toJobrign Item Wealth debt to wealth currncy holdings Interest rate on -0.079 0.271 deposits (-5.21) (5.098) Interest rate on public 0.058 sector debt (2-186) Interest rate on foreign -0.002 currency assetsr (-0.354) Log of the ratio of 0.996 mcome (GNP) to (3.875) wealth Exchange rate -0.008 depreciation (-2.95) Dummy for 1970-72 1.505 (4.67) Lagged dependent 0.932 variable (5.787) MA(1) 0.97 (3.962) Constant -1517 -OA64 L635 (-1.373) (-1.130) (3.269) R2 0.81 0.91 0.64 Durbin-Watson 1.85 2.34 227 Note: Figures in parentheses are f-statistics. A blank denotes the ominssion of the specific variable from the regression. a. London interbank offered rate (UBoR) plus expected exchange rate depreiation- is completed with a currency-substitution equation (in which the ratio of domestic deposits to foreign currency-denominated assets is taken to be a function of the three rates of return). This is equivalent to the altemative approach of estimating either the demand for domestic deposits or foreign currency assets and determining the remaining demand as a residual. The approach adopted was preferred, how- ever, because of our interest in obtaining direct estimates on currency substitution comparable to those existing in a wide body of empirical literature. The results of the estimations for the financial sector have been collected in tabIe 9.5. The nominal interest rates on both deposits and income are significant and carry the expected signs in the acrenqr demand equation. Moreover, as expected, the income elasticity is dose to urnity. The demand for government bonds is positively and significantly related to the interest rate on those bonds and is nega- Nadeem Ul Haque and PeterJ. Montiel 445 tively related to the deposit interest rate. Interestingly enough, deposit interest rates do not significantly affect the demand for these bonds, and they were therefore dropped from the equation. Strong evidence of partial adjustment behavior was found for this equation. In the currency-substitution equation both the deposit rate and the expected change in the exchange rate are significant and of the expected sign. The effect of the exchange rate change is surprisingly small. In the deposit equation a dunmy variable for the period of the Bangladesh war was found to be significant and is therefore included. To investigate the indirect effects of fiscal deficits on the real sector operating through the financial variables descrbed above, the next section embeds the equations just estimated in a general equilibnrum model. Policy Simulations The preceding section identified several direct channels through which fiscal policy may have affected macroeconomic outcomes in Pakistan. These indude the direct crowding-out of private by public consumption, the effects of taxation on household resources, and the effects of increases in the public capital stock on economic growth, both directly and indirectly, through a complementary relationship with private investment. The nature of these relationships suggests one set of Teasons why fiscal deficits may not have greatly inhibited Pakistan's growth performance: public dissaving in the form of con- sumption may have been in part offset by private saving, thereby liting the effects of changes in public consumption on the resources available for investment. Simultaneously, public investment has itself been directly productive and may have tended to stimulate private investment. These relationships, however, are only part of the picture. Not only are these direct channels of influence dynamic in nature, so that their long-run implications may differ from contemporaneous effects cap- tured in regression coefficients, but indirect channels of influence must also be taken into account. These channels include, in particu- lar, the possibility of financial crowding-out because domestic bor- rowing to finance fiscal deficits may tend to raise domestic interest rates, thereby raising the rental cost of capital. To capture dynamic effects as well as these indirect interactions, in this section we analyze counterfactual scenarios under two alternative fiscal policies with a view to assessing how the fiscal adjustment would have affected the performance of Paldstan's economy during the 1980s. The model used for the simulations embodies the behavioral equa- tions estinated in the preceding section. These include the perma- nent income equation (9.5), the consumption function (9.7), a private 446 Pakistan: Fiscal Suslainahbily and Macroeconomic Policy investment function derived from equation 9.9, the growth equation (9.10), and a set of financial sector equations consisting of identities 9.11 and 9.12, as well as the three asset demand functions reported in table 9.5. The model is completed with several additional identities. For brevity, these are reported in table 9.6. The workings of the model can be described as follows: public sector consumption, investment, and tax revenues are taken to be policy-determined fiscal variables.14 Monetary policy variables con- sist of the supply of base money, borrowing by the public sector from the commercial banks, lending by the central bank to the commercial banks, and the required reserve ratio. We treat public external bor- rowing as an exogenous variable. As can be seen from the public sectcr budget constraint in table 9.6, the implication is that domestic nonbank borrowing is the residual mode of financing for the public secton This would seem to be the appropriate assumption for Palistan during the 1980s. Two other policy variables deserve mention. Pakistan maintained a system of administered interest rates in the commercial banking sys- tem until July 1, 1985, when all deposit and new financing operations of the banks were placed on an Islamic noninterest basis. The new profit-loss system implied a move toward farly flexible market- related rates of remuneration for deposits and charges for bank loans. However, although for most of the period over which our asset demand functions were estimated bank interest rates were essentially a policy instrument, we treat them as endogenous in the simulation exercises. Implicitly, we are assuming that the monetary authorities mranaged these interest rates so as to maintain financial market equi- librium. In fact, these rates do exhibit substantial year-to-year varia- tion over the sample period. Finally, Pakistan maintained a fixed exchange regime for most of the period under review. In 1982 this regime was modified in favor of a managed exchange rate, under which the authorities undertake frequent small devaluations of the rupee. While the exchange rate has been managed with an eye on the effective real exchange rate of the rupee (and a substantial real depreciation has been achieved since 1982), price stability has also been an objective of the authorities. For the purpose of the simulations, therefore, we treat the nominal exchange rate as a policy instrument. The model is solved as follows: at the beginning of each period real output is a predetermined variable, given as a function of the inher- ited private and public capital stocks. Beginning-of-period asset stocks are also predetermined because they are given by last period's government financing decisions and allocations of the private sector portfolio. The domestic-currency value of the private sector's stock of foreign assets, however, is also affected by the official exchange rate Nadeem 111 Haque and PeterJ. Montied 447 Table 9.6. Identities for the Simulation Model, Pakistan 1. Public sector budget constraint 4--H + AB + S AF0 + (ALEB - AL& = DEF) 2. Public sector deficit DEF = (CC + c - TP + i8B. + iFSFCI + iCS(L6 -LE)-1 3. Public sector capital acumulation Kc = 0 + (1 -t9K1 4. High-powered money H= C+ rrD 5. Commercial banks' balance sheet L5B = (1 - rr)D LB - Bcs + LEB 6. Deposit interest rate i° = [11(1 - rrfliCB 7. Household dispable income iBBP1 + i£SF, + iD, -iCBL5B1 Y" =Y +Z -T+ 8. Household budge constrint AW = (YD - cr-MP + (S-&pE1 9. Private investment IP = (K['Y)- (I - P)Kt, 10. Renta cost of capital (IBK - + 65}PK 11. Reative price of capital 12. Equilbrium condition for public-sector securities B =BP+Bc 13. Trade balance TB = Y - CP-CG - IP Irc Note: Variables not previously identified are defined as folows: FG foreign debt of the public sector, L = commercial bank lending to the public sectora L& = Central Bank lending to commercial banks; H = high-powered money; e = total publc sector securities outstanding; DEF ' public sector deficit; Ic = public invesment, T = taxes; ifl = interest rate on public sector securities; i" = interest rate on foreign debt; icE = commercial bank lending rate, rr = reserve ratio; YD = household disposable income; Z = foreign remittances; 6P = rate of depreciation on private capital stock; SC = rate of depreciation on public capital stock; PK = price of capital goods; TB = trade balance; to = fixed positive parameter. 448 Pakistan: Fiscal Sustainability and Macroeconomic Policy during the period. For these assets to be wiflingly held, the price level (which affects the demand for currency), the interest rate on public sector securities, and the deposit interest rate all adjust endogenously to achieve equilibrium levels. The interest rate on public sector securi- ties, in turn, determines the rental rate on capital. (The expected rate of inflation is treated as an exogenous variable in these simulations.) The rental rate on capital, together with public consumption and investment decisions and other contemporaneous exogenous deter- minants of private disposable income, determine private consump- tion, investment, and saving, as well as the fiscal deficit and the trade balance. Public sector financing decisions will then determine the inaements during the period to the domestic components of the private sector's asset portfolio that are to be carried over to the next period. The total size of the portfolio depends on prvate saving and the amount of lending that banks amr able to make available to the private sector after satisfying the public sector's financing needs. Any discrepancy between the increase in the private sector's portfolio and the total new liabilities issued by the public sector and the banks is accumulated by the private sector in the form of foreign assets. With private and public capital stocks determined from this period's net investment by the xespective sectors, next period's output is deter- mined, and the model is ready to be solved again. We have used this model to undertake two different simulation exercises. The first represents an alternative way of financing histori- cal fiscal deficits, while the other Tepresents a deficit reduction scenario. In the first simulation we examine the macroeconomic conse- quences of limiting the buildup of domestic debt by the public sector after 1982/83 by increasing the use of money financing. Specifically, we reduce the flow of new domestic debt in each year from 1983184 to 1987188 by 10 percent and assign the role of residual financing to the issuance of base money. The results are presented in figure 9.17. It is obvious that reduction of the flow of debt by 10 percent in each period keeps the stock of debt below its baseline value, but the per- centage deviations from that value vary over time. Additional money financing would have implied larger price increases than historically observed (panel E of figure 9.17). But since domestic interest rates would have been lower (panel C), private investment would have increased (panel F), and as a result real GDP would have attained higher levels in the short run (panel D). Although factor income would have been higher, lower domestic nterest rates and higher prices would have squeezed private disposable income, leading to lower private consumption (panel D). With higher output and lower private consumption, the trade deficit would have fallen, in spite of the increase in private investment (panel B). In fact, lower debt and Figure 9.17. Macroeconomic Effects of a 10 Percent Reduction in Domestic Debt: Simulations for Pakistan A. Domestic debt of the government Percentage deviation -0.02-\ -0.04- P. Ratio of the deficit and trade defidt to GDP Percenth'!.e deviation 0.04 . Ratio of trade =deficit to GDP -0.06 - . . . , _~~~~., -.- -0-08 C Interest rai_ on government debt Level deviation 0.2 -OA- -0.7 -1.0 1982 1983 1984 1985 1986 1987 (Figure continues on thefollowing page.) 449 Figure 9.17 (continued) D. Real private consumption and real GDP Percentage deviation 0.2- 0.03 - _ - _ _Real rDP 0.06- -0.01 - 4.01- 9<\ /\Re~~Ral pnrte consumpfion -0.08- \ .\_ -0.1-,,, E Price level Percentage deviation 0.075 0.060 - 0.045 - 0.030- 0.015 0 F. Real private investment Percentage deviation 0.03 0.01 -0.01 1982 1983 1984 1985 1986 1987 Source: Country case study. 450 Nadeenm Ul Haque and PeterJ. Montiel 451 lower interest rates would have implied a lower value of the ratio of the fiscal deficit to GNP. In short, the mode of financing actually chosen seems to have operated as intended to contain the conse- quences of the deficit for the price level, but at the expense of some- what lower investment-and therefore slower economic growth- than would otherwise have been observed during this period. Our remaining simulation covers the same period but assumes a 10 percent reduction in the fiscal deficit. As in the baseline simulation, debt issuance is the residual mode of financing. Deficit reduction is brought about by means of a reduction in public sector investment. Figure 9.18 shows that reducing the deficit by 10 percent in each year would have required larger and larger reductions in public invest- ment in relation to the baseline. As is evident from panel D, this would have implied progressively larger reductions in real output, both because of the lower public capital stock and because of the induced decrease in the private capital stock, since the smaller public capital stock would have depressed private investment (panel F). Crowding-in through lower interest rates does not materialize in this case because the lower public capital stock represents a substantial negative supply shock, which raises prices (panel E) and thus actually increases the domestic interest rate (panel C). Both reduced output and higher prices depress private consumption (panel D). However, the reductions in public and private invesrtment, together with the decrease in private consumption, do succeed in reducing the trade deficit (panel B), in spite of the lower level of output. Overall, the macroeconomic effects of Pakistan's deficits from 1983184 to 1987188 depend on the nature of the counterfactual fiscal policy. It appears that reducing the deficit by cutting public invest- ment, which has tended to be a favorite vehicle for deficit control in Pakistan, could have had favorable trade balance effects, but at a cost to economic growth and with little payoff in terms of price-level objectives. The way in which the historical deficits were financed also seems to have had an important effect on the economy's macro- economic performance during the 1980s. According to our simulation results, altering the composition of deficit financing from doinestic borrowing to the issuance of money would have had fairly predict- able effects: shifting to more money financing would have meant lower interest rates and higher growth in the short run.'5 Conclusions The underlying causes of Pakistan's high fiscal deficits during the period 1972/73 to 1987/88 were not dissimilar to those in other developing countries. The deficits reflected explicit policy choices, as weU as political and administrative problems, and they were facili- Figure 9.18. Reduction in Government Deficit through a Reduction in Government Investment Expenditure, Pakistan A. Real government investment Percentage deviation -0.05 -0.10 -0.15 -0.20 -0.25- i,, B. Ratio of the deficit and trade deficit to GOP Percentage deviation 0.15 -0.30 - Ratio of trade .. -A5 - deficit to GDP - -0.60 - -0.75 - C Interest rate on govemment debt Level deviation 0.85 0.65 - 0.45 0.25- 0,05 -0.15 - _ _ _ _ 1982 1983 1984 1985 1986 1987 452 D. Real private consumption and real GDP Percentage deviation O- - \Real private onsumption <.OI- - ~~~~~~Real GDP 0.a 0.02_ 1 , . ff -0.1~~~~~~~~~~~~~~~~~~~~~- 43.01 ~ ~ F Reall GDP' ietmn -0.3_ , | j . 1982 1983 1984 1985 1986 1987 Source. Country case study. 453 454 Pakistan: Fiscal Sustainability and Macroeconomic Policy tated by the avaflability of extemal financing. Among the key policy decisions contributing to Pakistan's fiscal performance during the period were the Bhutto government's decision in the early 1970s to substantially erdarge the role of the public sector and the decsions of subsequent governments to maintain high levels of defense spending and consumer subsidies and, in the 1980s, to rely heavily on nonbank domestic borrowing as a source of finance. Throughout the period, a key political problem has been the inability to levy significant taxes on various politically powerful econoniic sectors. Coupled with the administrative difficulties that would be posed by greater reliance on income taxation, this inability has prevented the emergence of a reve- nue base for financing the chosen expenditure levels. Nevertheless, the ready availability of external finance at concessionary rates has permitted the resulting deficits to be financed without a fiscal explosion. As has been emphasized throughout the chapter, a distinctive fea- ture of the Pakistani experience during the period under review has been the coexistence of very large fiscal deficits for long periods of time with an economic performance that was relatively satisfactory and crsis-free with respect to growth, inflation, and the external accounts. To some extent, this reflects a statistical musiot The grant element associated with external borrowing and the tax element attached to domestic borrowing are, in effect, sources of revenue for the general government but are treated in official statistics as financ- ing items. Although our crude estimates for the magnitude of the effects of these elements on the measured deficits should be consid- ered as an upper bound, this factor may in fact have accounted for sizable proportions of the deficit at various times during the past two decades. Growth itself also explains the absence of high inflation, since the assocated expansion of the base for both conventional taxes and seigniorage made it possible to finance in a noninflationary way equilibrium deficits that were significantly larger than could have been financed in a slow-growth economy. Nevertheless, since the early 1980s fiscal deficits in Paidstan have dearly exceeded such equilibrium values. These deficts have been financed by domestic nonbank borrowing, resulting in increasing ratios of domestic public debt to crNP and, until quite recently, in rising interest rates on such debt. Our simulations indicate that although relying on this source of finance may have mitigated the inflationary consequences of the deficits, it was done at the expense of some aowding-out of private investment, implying slower growth than would otherwise have been observed. Controlling the deficit over the period would perhaps have made it possible to generate more favorable macroeconomic outconmes, at least with respect to growth and the external accounts. This outcome, however, would not Nadeem Ul Haque and PeterJ. Montiel 455 have been possible if the deficit reduction were brought about in a manner commonly relied on both in Pakistan and elsewhere-that is, through reducing public investment. The question of alternative modes of deficit reduction has become an issue of increasing relevance to Pakistan in recent years because fiscal deficits of the magnitudes observed up to 1987188 have become more difficult to finance. Not only are the sources of external financ-- ing at concessional rates dependent on the vagaries of the world oil market and political developments in the Middle East, but the accu- mulation of domestic debt and the increased costs of borrowing at home both require the generation of lower primary deficits. The alter- native, money financing, runs the risk of moving Pakistan's macro- economic performance closer to that of the countries with high fiscal deficits that are discussed elsewhere in this book. Notes At the time of writing, both authors were on the staff of the International Monetary Fund (imF). The views expressed are the sole responsibility of the authors and do not represent the views of the rMF. The authors are gateful to William Easterly, E. Ahmed, Mohsin S. Khan, and Malcolm D. Knight for their comments on an earlier draft and to Ravina Malkani for providing excellent research assistance. 1. The dates in this chapter refer to fiscal years, which in Pakistan run from July 1 to June 30. 2. Deficit as a percentage of GNP is the broadest deficit measure available for Pakistan and will be used throughout the chapter Because of the importanca of workers' remittances in Pakistan's economy, GNP rather an gross domes- tic product (GDP) is used as the scale variable. 3. To some extent the official price and output figures could be misleading. The quality of the output data is open to question, particularly in the pres- ence of what is said to be a substantial underground economy. As for prices, the consumer price index encompasses a number of goods that have been subject to price controls for some time. Data problems, however, are unlikely to account for more than a very minor part of the discrepancy between the macroeconomic performance of Pakistan and that of, say, several large coun- tries in Latin America. 4. Indirect taxes account for about 80 percent of total tax revenues in Pakistan, and foreign trade taxes represent about half of total indirect tax reve- nue. Adiniiistrative problems have hampered the collection of direct taxes, and the taxation of agricultural incomes has not been politically feasible. 5. The previous govemment, however, had already achieved a substantial reduction in public consumption during its last year in office. 6. Since the coefficient of log GNP in equation 2 in table 9.2 is essentially unity, this equation in effect regresses the share of trade taxes in GrNP on the share of exports and imports in GNP, as well as on the share of remittances in GNP. 7. These show up as "other nontax revenues" in table 9.1. 456 Pakistan: Fiscal Sustainability and Macroeconomic Policy 8. Uncertainty surrounds not just the netting-out procedure used to calcu- late net debt for 1979/80 but also the degree to which the debt can be consid- ered to be "willingly held" by creditors. Complications arise in the latter regard because some of the domestic debt was held by domestic banks sub- ject to asset supervision, and much of the external debt consisted of bilateral lending. 9. More precisely, the procedure involved subtracting the product of the grant element and the flow of foreign loans, as weli as the product of the impLict tax rate and the flow of domestic borrowing, from the actual deficit 10. This variable is intended to capture confidence effects associated with the change in economic policies as a result of the change in political regimes. 1U. The instruments used in the estimation induded the error-correction term, the lagged growth rate, the ratio of the government capital stock to cDp, and the time trend. 12. The framework used is a variation of the Tobin (1969) general- eqtulibrium approach. 13. The economy is assumed to be fairly open, in accordance with the findings of Haque and Montiel (1990,1991). 14. This implidtly assumes that tax rates are adjusted to offset deviations in the tax base from baseline values. 15. For additional information on the infuence of foreign financing on fiscal policy in Pakistan, see Haque, Husain, and Montiel (1991). References Buiter, Willem HK 1985. "A Guide to Public Sector Debt and Deficits." Eco- nomic Policy 1 (November): 13-79. David, Paul, and John Scadding. 1974. "Private Saving, Uitrarationality, and Denison's Law." Journal of.Political Economy 82 (March-April): 225-49. Davidson, James F. H., David F. Hendry, Frank Srba, and Stephen Yeot 1978. "Econometric Modeling of the Aggregate rune-Series Relationship Between Consumers' Expenditure and Income in the UK." Economik Jour- nal 88 (December): 661-92. Haque, Nadeem U, and Peter J. Montiel. 1989. "Consumption in Developing Countries: Tests for Liquidity Constraints and Finite Horizons." Reuiew of Economis and Statistics 71 (August): 408-15. 1990. "How Mobile Is Capital in Developing Countries?" Economic Letters 33: 35942. - . 199L "Capital Mobility in Developing Countries-Some Empirical Tests." World Development 19 (10): 1391-98. Haque, Nadeem U1, Aasim Husain, and Peter J. Montiel. 1991. -"An Empiri- cal 'Dependent Economy' Model for Pakdstan." w Working Paper 911102. International Monetary Fund, Washington, D.C. RAF (International Monetary Fund). 1990. World Economic Outlook. Washing- ton, D.C. Kemal, A. R. 1987. "Fiscal System of Pakistan." World Bank, Development Research Department, Washington, D.C. Nudeem U1 Haque and Peter J. MonUid 457 Khan, Mohsin S. 1990. "Macroeconomic Policies and the Balance of Pay- rnents in Pakistan: 1972-1986." IMF Working Paper 78. International Mone- tary Fund, Washington, D.C Morshed, Nader. 1987. "Pakistan Banking System and Capital Markets." World Bank, Development Research Department, Washington, D.C. Pakistan. Various issues. Pakistan Economic Survey. Islamabad. State Bank of Pakistan. Various issues. Annual Report. Islamabad. Tobin, James. 1969. "A General Equilibrium Approach to Monetary Theory." Jounmi ofMoney, Credit, and Banking 1 (February): 15-29. 10 Zimbabwe: Fiscal Disequilibria and Low Growth Felipe Morande and Klaus Schmidt-Hebbel Since independence in 1980, Zimbabwe has shown significant im- provements in areas such as education, health, and smallholder agri- culture. However, these social achievements have not been matched by improvements in the overall standard of living. Large fiscal imbal- ances, a vast and loss-making state-enterprise sector, and lack of an adequate market-based incentive structure have severely affected macroeconomic stability and growth during the past twelve years. This inadequate polcy framework has also hampered Zimbabwe's ability to cope with recurrent droughts, of which the recent 1992 drought was the most severe. As a result of policy mistakes and bad luck, CGDP growth per capita was zero during 1980-92. Public sector imbalances have been at the heart of the country's difficulties. The deficit of either the central govermnent or the consoli- dated nonfinancial public sector (cNr*s, comprising general govern- ment and public enterprises) shows double-digit levels in almost every year since 1981 (table 10.1). Some sustained fiscal adjustment was achieved in 1987, when the deficit was brought down from about 14 percent of wDp to 10 percent. Additional fiscal adjustment has been pursued since 1991 in the framework of a structural adjustment pro- gram but has not been reflected in declirdng fiscal deficits because of the massively adverse budgettay impact of the 1992 drought Public deficit financing has gone through various phases: foreign financing in the early 1980s, domestic debt financing in the mid- to late 1980s, and a resurgence of foreign financing since 1990. In the absence of significant private capital flows, the public sector deficit has been the driving force behind Zimbabwe's current account defi- cits and the derived steep increase in total foreign debt, from 15 percent of GDP' in 1980 to 69 percent in 1992. Strongly declining current account deficts were associated with a cumulative 56 percent depredation of the real exchange rate between 1981 and 1985. Between 1985 and 1990 the real exchange rate did not change much, while the current account deficit was small. As a result of the partial trade liberalization since 1991 and, principally, the re- cent drought, the current account deteriorated in 1991 and reached a 458 Felipe Morandu and Klaus Schmidt-Hebbel 459 record deficit of 20.7 percent of GDr in 1992 The real exchange rate deprecdated by a cumulative 23.1 percent between 1990 and 1992. Monletary policy followed a relatively prudent course until 1990, with low to moderate inflation rates. An expansionary monetary pol- icy in 1991 (reversed in late 1991) started an inflationary cycle fueled by subsequent drought-induced price increases, leading to an un- precedented 46 percent inflation rate in 1992. Macroeconomic policy is only partly to blame for Zimbabwe's un- satisfactory economic performance. In fact, a large array of trade and factor market distortions has hampered investment and growth. Domestic price controls have been prevalent in agriculture, utilities, and transport. Interest rate controls have kept interest rates on bank deposits and public debt at levels that have rarely exceeded domestic inflation rates. Foreign trade has been subject to large barriers in the form of quantitative and nonquantitative trade restrictions. Rationing of foreign exchange by the goverrment is institutionalized through a system of foreign exchange allocation to importing sectors on the basis of noneconomic criteria. These barriers have maintained signifi- cant import substitution inherited from the preindependence period at the cost of severely hampering the economy's productive effi- ciency. Consumer imports, foreign capital flows, and, in particular, outflows of private capital have been greatly restricted. A large and generally inefficient public enterprise sector includes many loss- making firms, which contribute significantly to the overall cwws defi- cit. Private investment is constrained by state licensing and controls that raise the cost of doing business in Zimbabwe and by an uncertain policy and property-rights framework. Private consumption and investment have been crowded out by a public sector able to finance part of its deficit by relying on nonmarket mecharisms to generate and make use of a significant private sector surplus. When the public sector requires more resources from the domestic private sector to finance its deficit, it restricts allocation of foreign exchange to the pnvate sector, constraining aggregate private investment and consumption. During the past decade high saving and low investment were reflected in large private surpluses. By exerting strict controls on private capital flight and placing more government debt in domestic banks and nonbanking financial institutions, the gov- ernment ensures that the private surplus is channeled toward domestic financial markets. This allows the government to capture these private resources at typically negative real interest rates. Crowding-out of private investment and declining public invest- ment, combined with a low quality of investment projects as a result of the distorted incentive structure, have substantially affected Zim- babwe's growth potential and performance. In addition, to the extent that imported intermediate and capital goods are not perfect substi- Table 10.1. Macroeconomic Indicators, Zimbabwe, 1980-92 lIdicator 1980 1981 1982 1983 2984 1985 1986 1987 1988 1989 1990 1991 1992a Aggregate intdicators cGD growth (percent) 10.6 12.5 2.6 1.6 -1.9 6.8 2.6 -1.5 9.6 4.6 2.1 4.9 -8.3 Capacity utilization (percent) 79.7 89.3 88.0 85.4 80.5 84.3 85.8 83,8 88.8 - - - - cpr Inflation (percent) 5.5 13.1 10,7 23.1 20.2 8.5 14.3 12.5 7.4 12.9 17.4 24,3 46.2 Real exchange rate (1980 = 100)b 100.0 104.6 110.3 123.4 131.9 163.1 150.2 138.2 145.1 158.0 164.4 192.9 202.4 Nominal interest rate on CNPPS domestic debt (percent) 4.4 5.9 7.8 7.7 8.0 10.4 12.5 13.0 13.3 13.3 - - - Nominal interest rate on three-month bank deposits (percent) 3,4 8.9 8.6 11.9 10.1 9,3 9.6 9.2 9.4 9.1 9.8 14.2 25.0 Composition of outputi (percenttage of GDP) Resource balance -3.0 -7.3 -5.9 -3.2 0.6 1.2 4.4 4.1 5.2 2.4 0.3 -3.9 -12,6 Exports 30.3 25.2 22.0 21.3 26.7 29.9 30.9 31.2 30.0 30.2 30.1 32.7 33.8 Imports 33.3 32.5 27.9 24,5 26.1 28.7 26.5 27.1 24.8 27.8 29.9 35.6 44.6 Total consumption 84.2 84.2 84.8 84.5 83.7 85.4 81.9 76.8 73.3 76.1 78.6 82.9 90.6 Private 64.5 67.0 65.0 66.1 62,4 63.2 60.1 52.7 47.0 - 55.7 59.2 - Public 19.7 17.2 19.8 18.4 21,3 22.2 21,8 24.1 26.3 - 22,9 23.7 - Gross domestic Investment 18.8 23.0 21.1 15.9 18.9 21.0 19A4 19.1 21.5 21.5 21.1 21.0 22.0 Gross fixed capital investment 15.3 18.6 19.6 19,6 18.5 16.1 15.8 15.5 17.9 17.4 - - - Private 10.6 13.3 10.0 8.2 10.6 7.9 8.4 7.8 9.0 - - - Public 4.7 5,3 9.9 11,4 7.9 8.2 7.4 7.7 8.9 - - Change in stocks 3.5 4.4 1.2 -3.7 0.4 4.9 3.6 3.6 3.6 4,1 - Government deficit and debt (percenitage of GDP) CNFPS (fiscal year data) Deficlt 9,1 13.5 13.1 14.4 12.7 14.3 14.4 10.9 10.4 Foreign debt 12.0 17.6 23.3 27.0 33.3 42.2 40.6 411 38.0 Domestic debt 43.4 37.2 33.7 31.3 35.7 35.5 36.6 41.7 42.9 - - - - Central government (calendar year data) deficit (including net lending) - - - - - - - - 10.0 9.9 9.8 11.3 16.0 Monetary aggregates (percetntage of GDP) Base money 6.9 7.1 7.3 6.2 6.7 7.5 7.2 7.0 7.4 7.6 7.9 7.9 - Ml 18.4 15.3 15.9 11.9 13.5 14.3 13.3 13.7 14.0 14.1 14.8 13.8 - Quasi money 16.8 16.3 17.7 14.9 15.2 16.4 13.7 18.1 16.3 17.7 15.3 12.6 - Cuirrenit account and foreigni debt (perce|ntage of GDP) Current account deficitc 5.6 11.5 11.1 8.4 3.7 3.5 1.0 0 -1.2 1.2 4.4 11.0 20.7 Total foreign debtd 14.7 19.6 27.3 35.1 42.3 55.5 52.7 53.1 42,1 43.7 48.6 54.6 68.7 - Not available. Note: CNFrS, consolidated nonfinancial public sector. a. Estimated. b. Defined as the ratio between the product of the Z$/US$ nominal exchange rate and by the U.S. cr1, divided by the Zimbabwe cpl. c. Excludes official capital grants. d. End-of-year total (public and private) foreign debt outstanding and disbursed as a ratio of annual GDP. Sourwe; Reserve Bank of Zimbabwe; Zimbabwe Ministry of Finance; Schmidt-Hebbel 1990; and World Hanl data. A'2 Zimbabwe: Fiscal Disequilibriaand Law Growth tutes for domestic goods, import compression and exchange controls have lowered the utilization of exsting capacity. An adjustment program was initiated in 1991 to tackle both macro- economic disequiLibria and structural distortions. More stringent fis- cal and monetary policies were put in place, supported by a devalua- tion of the nominal exchange rate. Progress has been made in the areas of domestic price liberalization, private investment deregula- tion, and trade liberalization. However, much more progress is re- quired in liberalization of domestic goods and financial markets, as well as in external tiAde, restructuring and privatization of state enterprises, and deregulation of private sector production and invest- ment. These measures should complement further fiscal adjustment at the general-government level. Only when these conditions are met-and after allowing for the significant period required to estab- lish the credibility of the new policy framework-can Zimbabwe start on a path of sustained growth and poverty reduction. This chapter attempts to disentangle the macroeconomic implica- tions of Zimbabwe's public deficits and poliqr framework, following the methodology laid out in Easterly, Rodrfguez, and Schmidt-Hebbel (1989).1 Most of the empirical analysis covers the 1965-89 period, with a strong emphasis on the last decade- As a first step, the next section identifies the main macroeconomic and fiscal policy vanrables that have contributed to CNFPs deficts, focusing on the sensitivity of the budget to its main deteminants. The results show the overrding influence of fiscal policy variables as determinants of deficits, in companson with domestic or extemral variables beyond the direct control of fiscal policy- makes A subsequent deficit-sustainability analysis suggests that cur- rent public deficits are unsustainable in the sense that they imply exploding public debt profiles. We then assess the impact of public deficits orn monetary and finan- cial markets. The empirical evidence from simulations based on a dynamic portfolio framework suggests that excessive debt financing only postpones inflation or-in line with Sargent and Wallace's (1981) "unpleasant arithmetic"-the current combination of high public def- icits and low-to-moderate inflation rates is unsustainable in the longer term. This confirms the simple steady-state results of the preceding section. We next go a step further by analyzing the impact of the public sector on private consumption and investment. Qear evidence of crowding-out is presented, especially for the post-1980 pedod. Crowding-out is not only indirect, through higher interest rates, but is also a direct result of semicompulsory placement of public debt in financial institutions. Again the issue is one of longer-run sus- tainability, as permanently high fiscal deficits inhibit growth, leading to a further deterioration of fiscal stance. Fdipe Morandi and Mlaus Schmidt-Hrebbed 463 We then deal with the effect of the public defidt and its financing on external accounts, in particular the trade deficit and the real ex- change rate. The Rodriguez model in chapter 2 in this volume is modified to take into account the foreign exchange allocation mecha- nism and the binding constraints on capital movements. These fea- tures make the levels of public sector deficits and public sector spend- ing more relevant than deficit financing for the behavior of the trade balance and the real exchange rate in Zimbabwe. If public defcits crowd out private investment, potential output and growth prospects are affected. In the final sections we tackle this issue, as well as the effects of other policy distortions, and present our conclusions. Determination and Sustainability of Public Sector Deficits A first step in studying the macroeconomics of public sector deficits is to identify the main macroeconomic and fiscal policy variables that have contributed to those deficits and to assess their sustainability under altemative macroeconomnic scenarios. These exercises are car- ried out in this section, which begins with a discussion of public deficits and liabilities in the 1980s. Consolidated Public Sector Deficits and Balance S.heets Zimbabwe's nonfinancial public sector comprises the central govern- ment, local authorities, and public enterprises.2 The financial public sector consists mainly of the Reserve Bank of Zimbabwe and the Post Office and Savings Bank. These financial institutions do not carry out quasi-fiscal operations and do not show significant deficits or sur- pluses. Hence most of the analysis will be restrcted to the public sector deficit of the cNRFPs. Consolidated balance sheets were con- structed, however, for both the nonfinancial public sector and the total public sector (ns), the latter being defined as the nonfinancial public sector and the two public financial institutions mentioned above. The decomposition of the public deficit is referred to the cNFPs; the subsequent sustainability analysis is carried out for the asset and liability holdings of the Tps. Tables 10.2 and 10.3 report data on the deficits and net liabilities of the CNFPS and the TPs. Figure 10.1 shows the evolution of the primary and total nominal deficits during the 1980s, while figure 10.2 does the same for Zimbabwe's domestic and foreign debt ratios. After inde- pendence, public sector deficits grew from less than 10 percent of cGP to 13-14 percent of GDP and stayed at that level for six years, until 1986/87. The increase in the primary deficit initially took place mainly in the public enterprise sector, but a rise in the central government Table 10.2. Consolidated Public Sector Deficit, Zimbabwe, Fiscal 1980181-1988/89 (ratio to fiscal year CDP) Item 1980/81 1981/82 1982/83 1983/84 1984/85 1985/86 1986/87 1987188 1988189 CNFPS deficit 0.091 0.135 0.131 0.144 0.127 0.143 0.144 0.109 0.100 1. Primary deficit 0.067 0.100 0.087 0.090 0.063 0.071 0.069 0.033 0.022 Central government 0.054 0.024 0.019 0.046 0.037 0.035 0.055 0.018 0.009 PLA 0.014 0.075 0.069 0.043 0.026 0.036 0.014 0.015 0.013 2. Net interest payments 0.024 0.036 0.043 0.055 0.064 0.072 0.075 0.077 0.078 Foreign debt 0.007 0.011 0.018 0.032 0.035 0,033 0.029 0.027 0.025 Domestic debt 0.017 0.025 0.025 0.022 0,029 0.039 0.046 0.050 0.053 Consolidated TPS deficit 0.088 0.131 0.126 0.143 0.126 0.138 0.137 0.103 0.093 1. Primary deficit 0.067 0.100 0.087 0.090 0.063 0.071 0.069 0.033 0.022 Central government 0.054 0.024 0.019 0.046 0.037 0.035 0.055 0.018 0.009 PLA 0.014 0.075 0,069 0.043 0.026 0.036 0.014 0.015 0.013 2. Net interest payments 0.021 0.032 0.039 0.053 0.064 0.067 0.068 0.070 0.071 Foreign debt 0.004 0.011 0.018 0.034 0.038 0.033 0.029 0,0.i 0.024 Domestic debt 0.016 0.021 0.020 0.019 0.025 0.034 0.040 0.045 0.047 Note: rPL, public enterprises and local authorities; Tis, total public sector. Source: Schmidt-Hebbel 1990. Table 10.3. Public Sector Liabilities, Zimbabwe, 1980-88 (ratio to calendar year CDP) JOlne juniie Junte Jlunle JUlle Junzte iJunie Junte June Itemn 1980 1981 1982 t983 1984 1985 1986 1987 1988 CNFPS liabilities 0.541 0.544 0.568 0.617 0.746 0.813 0V793 0.864 0.830 1. Cash 0.013 -0.004 -0.004 -0.041 -0.065 -0.045 -0.030 -0.044 -0.030 2. Net foreign debt 0.120 0.176 0.233 0.270 0.333 0.422 0.406 0.411 0,380 3. Net domestic debt 0.434 0.37f 0.337 0.313 0.357 0.355 0.366 0.417 0.429 4, Equity 0.000 G.007 0.007 0.007 0.009 0.009 0.009 0.008 0.008 Consolidated TPs liabilities 0.541 0.544 0.567 0.617 0.746 0.813 0.793 0.864 0.830 1. Base money 0.058 0.061 0.064 0.057 0.059 0.058 0.060 0.064 0.064 2. Net foreign debt 0.074 0.169 0.237 0,289 0.366 0.419 0.393 0.393 0.365 3. Net domestic debt 0.423 0.318 0,270 0.260 0.309 0.312 0.319 0.375 0.380 4. Other liabilities -0.014 -0.004 -0.003 0.010 0.012 0.024 0.021 0.031 0.021 Souirce: Schmidt.Hebbel 1990. 466 Zimbabwe: Fiscal Disequilibda and Low Growlh Figure 10.1. Consolidated Fiscal Sector Deficit, Zimbabwe, Fiscal 1980(81 to 1988/89 Percentage of GDP 14 - 12- 10 8 6 4 2 1980/81 81/82 82/83 83/84 84/85 85/86 86/87 87/88 88/89 _- CNFPS deficit Primary deficit ConsolidatedTPSdeficit Note: CNFPS, consolidated nonfinancial public sector; Tis, total public sector. Source: Table 102. primary deficit soon followed. Increasing nominal interest rates on domestic debt ;nd rising debt-output ratios during 1980-85 explain the increasing net interest payments throughout the 1980s and up to the present. However, begining in 1987/88 a parfial fiscal adjustment in the central government reduced the deficit by 3.5 percentage points in that fiscal year and by 1I additional percentage point in 1988/89. The financing requirements of high public sector deficits have con- tributed to a steady and massive rise in public (cNn's or TPs) lia- bilities, from 54.1 percent of GDP in June 1980 to 86.4 percent in June 1988. It is interesting to note how the composition of public debt has been changing. In the early 1980s public deficits relied massively on foreign financing, raising nPs foreign debt fiom 7.4 percent of GD? in 1980 to a peak of 41.9 percent in 1985. (A similar change was observed in CNFPS debt ratios.) This made possible a reduction in domestic debt from 42.3 percent of CDP in 1980 to 26 percent in 1983. A strong reversal of the composition of debt financing occurred afterward, when the foreign debt ratio fell by a couple of percentage points while the domestic debt ratio increased to reach a level in 1988 only slightly Felipe Morandiand Klaus Schmidt-Hebbel 467 Figure 102 Con0iJlidated Public Sector Liabilities, Zimbabwe, 1980-88 Percentage of GDP 60 40 20 0 1980 1981 1982 1983 1984 198 1986 1987 1988 --_ NFlPS net foreign debt + NFPS net doDmesfic debt ConsoLidatec TPS liabilities TPS net foreign debt --- ;TiS net domnestic debt Note: NFS, nonfinancial public sector' T;S total public sector Source: Table 10.3. below that of 1980. Monetary fina-ncing of TPS deficits was relatlively small and stable throughout the 1980s. Base money, after increasing slightly in the early 1980s, has remained stable at about 6.4 percent of Economic and Policy Deteminants of Public Sector Deicits The following decomposition of public deficits in Zimbabwe dunng the 1980s compares the role of macxoeonozic domesZic and foreign vrariables with that of fiscal policies in generating the initial expan- sionary phase and the subsequent paruial fiscal adjustment. The ntethodolog.., which is based on the framework developed by Mar- shall and Schmidt-Hebbel (1989), is presented in the appendix to MOrande and Schmidt-Hebbel (1991). The main budgetary items of the CNF~PS deficit are first identified. By making Use of estimated tax revenue furnctions,3 the Fisher equation for domesfic interest rates, and simnple ariable nsformatios, it is possible to identify the ef- fects of the mnain macroecononic and policy variables on the deficit. This permits measurement of the sensitivity of Zimbabwe's public Table 10.4. Decomposition of Changes in CNFPS Deficits with Changes in Economic and Policy Determinants, Zimbabwe, Fiscal 1981182-1988189 Itengi 7981/82 1982183 1983184 1984/85 1985/86 1986187 1987/88 2988/89 Clantiges altribltable to doitestic variables -0.029 -0.004 -0.001 -0.016 -0.021 0,007 0.001 -0.024 Real cDP rc (denominator effects) -0.009 -0.003 0.000 -0.004 -0.007 -0.001 -0.004 -0.009 Real cor rc(economic effects) -0.016 -0.004 0,000 -0.005 -0.010 -0.001 -0.005 -0.012 Real Imports rc -0.012 0.004 0,007 0.000 -0.004 0.002 0.008 0.000 Domestic real Interest rate c 0.003 -0.004 0.014 0.029 -0.007 -0.009 0.006 -0.005 Domestic inflation c 0.006 0,007 -0.013 -0.020 0.015 0.010 -0.005 0.005 Real exchange rate rc -0.00! -0.003 -0.009 -0.018 -0.008 0.005 0.001 -0.003 Chlanges attributable to foreign variables Foreign nominal Interest rate c 0.000 0.003 0.007 -0.003 -0.008 -0.004 -0,001 0.000 Chnilges attribiutable to policy variables 0.067 -0.002 -0,016 -0.007 0.047 0.017 -0.016 -0.010 Foreign real debt rc 0.004 0.003 0.001 -0.001 0.006 0,003 0.000 -0.002 Domestic real debt ac -0.001 -0,002 0.000 0.003 0.000 0,004 0.005 0.005 Wage bill rc 0.010 0.001 0.000 0.006 0.011 0.016 0.036 0.004 ci Goods and services expenditure rc 0.008 0.002 0.006 -0.002 0.007 0.013 0.000 0.006 Transfers and subsidies rc 0.008 0.011 0.003 -0005 0.012 -0.010 -0.051 0.001 1980 political regime c -0.014 0.000 0.000 0.000 0.000 0.000 0.000 00 1988 direct tax regime c 0,000 0.000 0.000 OM 0.000 0.000 -0.024 0.000 1981 indirect tax regime c -0.028 0.000 0.000 0.000 0.000 0.000 0.000 0.000 1982 customs duties regime c 0.000 -0.022 0.022 0.000 0.000 0.000 0.000 0.000 1983 customs duties regime c 0.000 0.000 -0.037 0.000 0.000 0.000 0.000 0.034 1988 customs duties regime c 0.000 0.000 0.000 0.000 0.000 0.000 0.000 -0.052 PLA primary deficit rc 0.015 0.006 -0.014 0.001 -0.011 -0.002 0.012 0.001 NFrS Investment ri 0.064 0.000 0.003 -0.009 0.022 -0.007 0.007 -0.007 Explained sum of changes 0.038 -0.003 -0.010 -0.026 0.018 0.020 -0.016 -0.034 Changes attributable to other variables 0.006 -0.002 0,023 0.008 -0.001 -0.020 -0.028 0.026 Actual change in cNrrs deficits 0,044 -0.005 0.013 -0.017 0.017 0.000 -0.045 -0,008 Note: The figures reflect the annual changes In CNFPS deficitS (as ratios to cDr) caused by proportional changes (rc is rate of change) or absolute changes (c) in the corresponding variables. Souirce: Authors' calculations based on data from Zimbabwe Ministry of Finance. Felipe Morandf and Klaus Schmidt-Hebbel 469 budget structure to changes in macroeconomic and policy determinants. Table 10.4 reports the changes in the CnFps deficit according to their underlying macroeconomic and policy causes. To illustrate the useful- ness of this approach, the principal variables behind the partial fiscal adjustment of 1987188 to 1988189 are identified next. GDP growth was the main macroeconomuc variable contributing to deficit reduction during 1987/88 and 1988/89. Its positive effect on tax bases (the "economic effect" in table 10.4) reduced the deficit by 0.5- 1.2 percentage points of GDP, in addition to the 0.4-0.9 percentage point reduction resulting from the simple fact that the deficit and every budgetary item are expressed as ratios to GDP (the "denomina- tor effect"). Other macroeconomic variables (apart from imports, whose decline in 1987-89 increased the deficit) tended to cause only minor changes. Among fiscal variables a stabilization effort was reflected by signifi- candy lower transfers and subsidies in 1987188 and inaeased revenue from customs duties in 1988/89. However, other variables under the control of policymakers contributed to an increase in the deficit the budgetary wage bil expanded significantly, and, to a lesser extent, higher expenditure on goods and services and a higher deficit of public enterprises and local authorities (PLA) raised the CNnPS deficit. In addition, the secular rise in domestic debt increased domestic interest payments. WVhat was the role of fiscal policies in companson with the influ- ence of exogenous variables during the 1980s in Zimbabwe? Compu- tation (from table 10.4) of the average contribution of diff;erent vari- ables to the explained variation in the deficit over the 1981182 to 1988189 period yields the following results: domestic variables, 7 per- cent; external variables, -11 percent; and fiscal policy variables, 110 percent.4 These figures confirm the massive predominance of fiscal policy changes in both the cyclical variation and the trend change of cNFps deficits in Zimbabwe. Both domestic and external variables play a secondary role in shaping deficits-external variables even con- tribute negatively to deficit changes. This contrasts with the large contribution of fiscal policy variables, which actually offset the influ- ence of foreign interest effects. We condude that policymakers are responsible for both the fiscal deterioration in the early 1980s and the partial fiscal adjustment since 1987/88. The decomposition performed above also permits identification of the structural sensitivity of Zimbabwe's cNFS defict to its main determinants. Table 10.5 reports measures of the responsiveness (or semielasticities) of the deficit to changes in underlying macro- economic and policy variables, computed as absolute changes in the CNFPS deficit in response to a 1 percent (or I percentage point) change 470 Zimbabwe: Fiscal Disequilibria and Low Growth Table 10.5. Sensitivity of CNFPS Deficits to Changes in Economic and Policy Determinants (percentage points of GOP) Change in detenninant Change in CNFPS deficit Domesti variable 1 percent growth of real GDP Denominator effect -0.16 Economic effect -0.21 1 percent growth of real imports -0.10 1 percentage point increase in domestic real interest rate 0.40 (0.30) 1 percentage point inaease in domestic inflation 0.31 1 percent increase in real exchange rate -0.06 (-0.04) Foreign variable 1 percentage point increase in foreign nominal interest rate 0.25 (0.18) Policy variable 1 percent growth in foreign real debt 0.03 (0.01) 1 percent growth in domestic real debt 0.05 (0.02) 1 percent growth in real wage bill 0.14 (0.10) 1 percent growth in expenditure on goods and services 0.07 1 percent growth in transfers and subsidies 0.10 Change in political regime, 1980 -1.4 Change in direct tax regime, 1988 -2.4 Change in indirect tax regime, 1981 -2.8 Change in customs duties regime, 1982 -2-2 Change in customs duties regime, 1983 -3.5 Change in custorns duties regime, 1988 -5.2 1 percent growth hi NFPS investment 0.10 (0.05) Note. The changes in CNFPS deficts were obtained by dividing the 1987188-1988189 change in the deficit caused by the corresponding economic or policy determinant by the change in the corresponding determinant. Values for 1981182-1982183 that differ from the 1987/88-1988/89 levels are given in parentheses. The changes in political and tax regimes are measured by the changes in the corresponding dummies estimated by the tax revenue functions. Source: Authors' calculations based on data from Zimbabwe Ministry of Fmance. in the corresponding determinants during the recent past.5 These elasticities reflect the share in the budget of the respective budgetary variable (which may change over time) and, in the case of the behav- ioral tax revenue functions, the size of the corresponding coefficients (estimated as time-invariant parameters). The deficit appears to be quite sensitive to changes in macroeco- nomic variables. Its semielasticity with respect to GDP (-0.37, the sum of the denominator elasticity of -0.16 and the economic elastic- ity of -0.21) is surpassed only by that of the domestic real interest Felipe Morandi and Klaus Schmidt-Hebbel 471 rate (0.40). The responsiveness with respect to inflation is also rela- tively high, at 0.31, but is lower than the real interest rate semi- elasticity because the negative effect of inflation on the deficit through higher direct taxes partly offsets its positive effect through higher domestic nominal interest payments. Slightly lower is the semi- elasticity of the deficit with respect to foreign nominal interest rates (0.25). Finally, the deficit is only weakly responsive to the real exchange rate. A 1 percent real depreciation reduces the deficit by 0.06 percentage point of cDp; in other words, the strong effect on the deficit through higher interest payments on foreign debt is almost offset by the higher tax revenue, as both direct and indirect tax pay- ments are boosted by a depredation. Among policy variables, changes in tax regime that lead to higher tax burdens or impose stricter controls on evasion tend to affect the cNFPS deficit significantly. Expenditure variables (by decreasing weight) include the wage bill, public investment, transfers and sub- sidies, and expenditure on other goods and services. Fialy, al- though it has been omitted from the table, a change in the PLA primary noninterest current deficit is obviously of enormous impor- tance, as it and the cNFPS deficit change one-for-one. The preceding discussion has shed light on the sensitivity of Zim- babwe's public finances to the major macroeconomic and fiscal policy determinants of the deficit Future fiscal programming and stabiliza- tion efforts could be based on this kind of quantitative framework, which complements the usual policy considerations with a dear determination of the effectiveness of poicy instruments. Sustainable Deficits We now turn to the derivation of bounds for sustainable deficits of the public sector. The sustainability concept applied here refers to the time path of public liabilities for given demands for these liabilities by the domestic private and foreign sectors.6 The analysis starts with the standard budget constraint of the con- solidated Tns in current prices. By simple manipulation, the primary (that is, noninterest) deficit of the IPs as a share of GDP (pd) can be financed by the following sources: (1021) pd =h+b+jI4&+lh(?+9) +bW- r) + r§ - r*e) + ol(.P + y where h is the ratio of total base money to GDP, b is the ratio of domestic public debt to GDP, f iS the ratio of foreign public debt to GDP, of is the ratio of other public liabilties to GDP, P is the CDP deflator, y is real cDP, r is the domestic real interest rate, rt is the foreign real interest rate, and e is the real exchange rate. Dots over 472 Zimbabwe; Fi5caI Disequilrbra and Low Grwuht variables denote absolute (not relative) rates of change per time unit, and hats denote percentage rates of change? On the basis of the steady-state notion of the ratios of fixed public liability to GDP, a sustainable deficit is defined as a level consistent with the maintenance of constant holdings of public liabilities, in proportion to GDP, by domestic private and foreign creditors. Under this condition (h = =b = = , a primary deficit level that can be sustained over time has to be financed by a combination of the four sources identified in equation 10.1: inflation tax and seigniorage from GDP growth; the excess of domestic growth over the domestic real interest rate; the excess of domestic growth over the foreign real interest rate (adjusted for real exchange rate depredations); and infla- tion tax cam seigniorage on other liabilities.8 Table 10.6 presents simulation results for sustainable public sector deficits in Zimbabwe, consistent with the structure of its public fi- nances and with its recent evolution of macroeconomic variables. The first part of the table summarizes the recent evolution of macro- economic variables requmied for applying equation 10.1. This helps to identify relevant values for the base. The ratios to GDP of the four main liabilities of the consolidated rrs for June 1988 (obtained from table 10.3) are presented in the second part of table 10.6. They are used as the relevant (constant) liability ratios for the simulations re- ported in the third part of the table. Three scenarios are considered. The base scenario assumes GDP growth, real interest rates, and a real exchange rate constancy broadly consistent with Zimbabwe's perfornance during the late 1980s. Under the high scenario, growth exceeds the corresponding base- case value bv 1 percentage point, and the domestic real interest rate falls short of the base case by I percentage point. The low scenario has lower cap growth, higher real interest rates, and a real exchange rate depredation of 7 percent. This combination could reflect a -reform scenario featuring both financial reform (and higher real interest rates on public debt) and trade liberalization (which could require a more depreciated real exchange rate); transiional adjustment costs would be reflected in lower short erm growth Changes in growth ant interest rates have the strongest effects on the sustainable defict level simply because stocks of domestic and foreign debt are large in relation to base money and other public liabilities. In addition, capital losses on foreign debt from the real exchange rate devaluation can severely lmit sustainable deficts, as shown in the low scenario. Under the base case the sustainable primary deficit is estimated at 1.7 percent of GDP. The defict increases slightly, to 2.9 percent of GDP, under the high case and declies significantly, to -4.2 percent of GDp, under the low scenario. Sustainable nominal deficts vary accordingly. Felipe MorandiE and Klaus Schimidt-Hebbel 473 Table 10.6. Estimation of Sustainable Public Deficits, Zimbabwe Variable 1986/87 1987188 1988189 Macroeconomic variable GDP growth -0.038 -0.014 0.044 Domestic inflation 0.172 0.148 0.137 Domestic nominal interest rate 0.130 0.130 0.135 Domestic real interest rate 0.042 0.018 0.002 Foreign nominal interest rate 0.075 0.072 0.072 Foreign inflation 0.022 0.042 0.045 Foreign real interest rate 0.013 0.030 0.028 Domestic devaluation 0.015 0.044 0.129 Real exchange rate depreciation -0.114 -0.053 0.037 Scenario Sustainable nonfinacial public sector deficit Base High LoW GDP growtht 0.040 0.050 0.020 Domestic inflation 0.110 0.110 0.110 Domestic nominal interest rate 0.140 0.130 0.170 Domestic real interest rate 0.030 0.020 0.060 Foreign nomi-nal interest rate 0.080 0.080 0.090 Foreign inflation 0.040 0.040 0.040 Foreign real interest rate 0.040 0-040 0.050 Real exchange rate depreciation 0.000 0.O0 0.070 Inflation tax 0.007 0.007 0.007 Seigniorage from growth 0.003 0.003 0.001 Domestic debt effect 0.004 0.011 -0.015 Foreign debt effect 0.000 0.004 -0.011 Foreign debt capital gain 0.000 0.000 -0.027 Effect of other liabilities 0.003 0.003 0.003 Sustainable pimary deficit 0.017 0.029 -0.042 Interest payments on foreign debt 0.029 0.029 0.033 Interest payments on domestic debt 0.053 0.049 0.065 Sustainable nominal deficit 0.099 0.107 0.056 Note The underlying ratios of public sector liabilities to cDr! are those of the total public sector for June 1988, reported in table 10.3. Sounr Authors' estimations based on data from Reserve Bank of Zimbabwe, Zim- babwe Ministry of Fmance, and World Bank. The actual 1988/89 primary defidt of 2.2 percent of GDP (see table 10.2) is at the midpoint of the base and high scenarios and exceeds the low-case dUict by a large amount (6A percentage points of GDP). The nominal deficits of the base and unfavorable cases are similar to the actual TPS and CNn'S nominal deficits, but again, the low scenario shows a sustainable nominal deficit that is almost 5 percentage points below the latter measures. 474 Zbnbabwe: Fiscal Diseilria and Low Growth We may condude that while public sector deficits in the range observed during the late 1980s in Zimbabwe may be sustainable from the limited perspective of the ratio of constant liability to CDP and under macroeconomic conditions ranging from normal to favorable, they are clearly unsustainable under adverse macroeconomic shocks or when significant devaluations are required in response to policy changes. This is precisely what occurred in 1990-92, when the combi- nation of low growth and real devaluations, aggravated by the drought-induced increase in the deficit, led to a deterioration of the fiscal stance. It is not surprising that total public debt has been rising massively since 1990. Deficit Fmancing and Financial Markets Having identified the determinants of public deficits and their sus- tainability, we now invert the focus of causality by looking at the effects of public deficits on Zimbabwe's macroeconomic performance. This section discusses the macroeconomic impact of public sector deficits on financial markets, deriving and estimating a version of Easterly's (1989) model. The model emphasizes the determination of the real interest and inflation rates, with money demand as its main behavioral component. The framework used here incorporates addi- tional features distinctive to the Zimbabwean economy, reflecting three dimensions of financial control exercised by the government to capture private surpluses for financing public deficits: controls on foreign exdhange and capital flows, compulsory placement of public debt in financial iistitutions, and partial interest rate controls. (For a similar view see Chluibber and others 1989.) Foreign exchange rationing effectively constrains private demands for consumer, capital, and intermediate goods, leading to a private sector aggregate surplus of investment over saving. The government exerts strct control on capital outflows to prevent the private sector from shifting its surplus abroad. Domestic banking and financial markets have rather smoothly intermediated the private sector surplus to the public sector as a result of several regulations that force financial intermediaries to pur- chase public debt. In fact, the financial system is exceptionally deep for a country at Zimbabwe's stage of development.9 A number of institutions comprise the monetary sector (the Reserve Bank of Zim- babwe, two discount houses, firve commercial banks, and accepting houses), and monetary assets exceed 30 percent of CDP. The non- monetary sector (building societies, finance houses, the Post Office and SavL-.gs Bank, insurance companies, and pension funds) is sig- nificantly larger than the monetary sector, in part because institu- tional investors (insurance companies and pension funds) have been Felipe Morandt and Klaus Schmidt-Hebbel 475 capturing most private long-term savings. Financial institutions (in particular, institutional investors and the Post Office and Savings Bank) are required to hold significant shares of their portfolio in the form of public sector liabilities. Nominal interest rates have been kept low by the combination of controls on external and domestic financial flows that has already been described. However, partial interest rate controls have also con- tributed to the low cost of private funds. Some nominal interest rates are free but tend to follow controlled rates. Both categories of nominal rates have been relatively stable, with the exception of large tempo- rary bursts in the early 1980s and early 1990s, in the aftermath of inflationary shocks. In addition, it is argued that the monetary authority manages required reserve ratios to help maintain stable and low interest rates whenever net private credit demand is inconsistent with stability (Chhlibber and others 1989). As a result of such nominal stability, real interest rates have fluctuated much more, in tandem with inflation. On average, real interest rate" naiJ on public debt and banking deposits have been negative. In sum, the public sector has a major captive source of private funds in the institutional investors, which have financed a significant share of the centrl government deficit. What is left of private sector saving is distributed among deposits at monetary institutions, the Post Office and Savings Bank, and building societies. Other non- monetary institutions, such as finance houses, are of less importance. A snall share of private saving is directly channeled to treasury bills and government stocks and bonds, all of which are issued by the central government. Model Structure and Estimation Results The model starts by relating base money creation to the consolidated government budget constraint. Taking into account the main features of Zimbabwe's financial system and its public sector budget structure and financing, such a dependence is summarized as: (10.2) ht = gt'- _ -lS, + (1 + xt))-'(z..l) where all variables are in real terms (that is, deflated by the domestic CMi), b, is public sector bonds in private sector hands, l, is government stock and bonds plus treasury bils in the banking system, gr is the government's primary deficit net of the part financed by foreign debt and the placement of domestic debt with institutional investors, h is base money, and i8 is the common interest rate paid by all types of public debt. Additionally, z,;1 = ht-, + (1 + iy4-)(bn.l t i + _), and 7r, is the inflation rate. (See Morande and S:hmidt-Hebbel 1991 for a detailed discussion.) Here and below, time periods are denoted by subscripts t-1, t, and t+1L 476 Zimbabwe. Fiscal Disequilibria and Low Growth The nonfinancial private sector voluntarily holds three broad assets: money, interest-earning deposits in the banking system, and public sector bonds. Domestic residents are allowed to hold neither foreign assets nor foreign liabilities, a prohibition that seems to be binding. Private bank loans to the private sector are netted out of the demand for interest-earning deposits. These three asset demands fol- low a standard portfolio setup. However, interest-bearing deposits can be seen as an indirect way for the nonfinancial private sector to hold public debt (1_). The substitution of public debt and money from the banking system's balance sheet for interest-bearing deposits yields a portfolio of two financial assets: public debt and money. The demand for the latter is: (10.3) ml = m(r,, 0, and v, is a zero-mean constant-variance stochastic term. Estimation of equation 10.4 for inflation permits the derivation of values for expected inflation according to the rational-expectations hypothesis. We estimate a linear version of equation 10.4 in which inflation and the percentage variation of H, WU and E are measured in annual terms, for consistency with the estimated money demand.'l The empirical results are reported in table 10.7. The short-run semi- elasticities of money demand with respect to the real interest rate and inflation are significantly different from zero and are similar to each other, as expected: -1.41 and -1.13, respectively. The long-run values of the semielasticties are -4.55 for the real interest rate and -3.65 for expected inflation.2 The elasticity with respect to pri- vate net financial assets, which is also significantly different from zero, is 0.26 in the short run and 0.84 in the long run. Siilar results are found in Elbadawi and Schmidt-Hebbel (199la), using a general equilibrium setup. The results reported for inflation include as independent variables the one-quarter-lagged percentage variations of base money and the exchange rate (in addition to the lagged dependent variable).13 Strong pnce inertia is evidenced by the large role of the one-quarter-lagged inflation rate. Simulation ResultsforAlternative DeficitFinancingForms On the basis of the results reported in table 10.7, we perform simiula- tions of government policies related to the size and financing of the public deficit. For convenience the estimated money demand is in- verted, and the inflation equation is restated as follows: (10.7) r -0.71 In mt + 0.18 In nfa2 - 0.80et+1 + 0.49 ln mtn-L - 0.71E, -(10.8) rt = 0i128dHt.. + 0.092dEt.. + 0.7237r,- + Pt. The policy simulations are based on equations 10.7 and 10.8, the government budget constraint (equation 10.2), and the assumption of 478 Zimbabwe: Fiscal Disequilzlbia and Low Growth Table 10.7. Estimation Results for Money Demand and Inflation, Zimbabwe, 1980-88 Vanable Coeffaent Money demand (dependent variable, m) Constant -0.18 -(0.56) In r -1.41 (-Z76) -1.13 (-2.33) In naf 0.26 (2.37) m_l 0.69 (7.27) Adjusted R2 0.85 F-statistic 47.7 Q 19A4 Inflation rate (dependent variable, r) Constant -0.003 (-0.18) dH-1 0.128 (4.0) - dE-1 0.092 (3-1) 0.723 (7.4)- Adjusted R2 0.73 F-statistic 35.3 __ Q 23.5 Note: The estimation method is ordinary least squares. Numbers in parentheses are t-statistics. Q is the Ljung-Box statistic, marginal significance levels for the Q values are 0.63 (for Q = 19.4) and 056 (for Q = 23.5). rational expectations. They consider both direct and indirect dynamic feedback effects- For instance, the sensitivity of the real interest rate with respect to changes in real money is such that a 1 percent incease in the latter variable at time t causes a contemporaneous reduction in the real interest rate of 0.7 percentage point and then an increase of 0.49 percentage point in period t + 1. However, if changes in m cause changes in base money, there are several other indirect mechanisms of transmission, such as the effect of the change in base money on 5nfla- Felipe Momndi and Klaus Schmidt-Hebbel 479 tion in t + I (equation 10.8) and changes in the real value of private net financial assets, that will modify the effect starting in period t + 1. The following exercises simulate a temporary primary deficit, fi- nanced alternatively by issuing base money and domestic debt. The simulations consider a one-time 10 percent increase in G' (the nomi- nal adjusted primary deficit) during the first period of a sixteen- quarter simulation horizon, from 1990:1 to 1993:4. At time 0 (1989:4), before the change in G' occurs, the system is assumed to be at a steady-state position, with all variable changes set at zero. In addi- tion, initial real interest and inflation rates are equal to zero. A. A 20 percent increase in the primary deficit, financed by issuing base money (dH, >0). According to the relative magnitudes of G' and H, a 10 percent increase in G; requires a 2.83 percent increase in H,. Since there are no further increments in G', dHr,. = 0, for s >0. The nominal interest rate remains constant, which is not an unrealistic assumption. The simulation results are reported in the first part of table 10.8. Fig- ure 10.3 shows the path of the real interest and inflation rates. On impact, the real interest rate decines significantly. Higher base money raises real money balances, since inflation is not affected until the second quarter (1990:2). By 1990:2 the positive lagged effect of real balances on the real interest rate brings the latter back to a level dose to its initial value. This effect is offset, however, by the rise in inflation during that quarter. Afterward, the persistence of a positive inflation rate dominates the determination of the real interest rate, even though the same positive inflation rate lowers real balances and net financial assets, thus raising r. In the end the price level rises by a cumulative 1.31 percent, about half of the init increase in base money. The final effect on the real interest rate is a reduction of 1.3 percentage points. B. A 10 percent increase in the primary deit, financed by domestic debt, with monetization of interest payments and debt repayment. In this simula- tion the government resorts to issuing new debt to finance the in- creased deficit in 1990:1 and, beginning in the second quarter, issues base money to pay the interest on the new debt. Issuing debt has a strontg positive effect on the real interest rate but no effect on infla- tion, since no change in base money has occurred. However, the money-financed interest payments (starting in 1990:2) gradually re- duce the real interest rate because of the forces that were discussed in the previous simulation. Higher base money also leads to inflation from 1990:3 onward. The debt issued in 1990:1 matures fifteen qua:- ters later, in 1993:3, and is repaid by issuing base money. This portfo- lio shift causes a large decline in the real interest rate in 1993:3 and an inflationary burst in the subsequent quarter. Although government debt could stay at its increased level for a long time, with interest Table 10.8. Simulation Results for Different Budget Financing Policies, Zimbabwe -Polfcy aild quaJrter dG' dH dB r dtHi d nfa dr r db(d) A. 10 percent increase in the primary deficit (G') fitnnced by issulitg base, nioney 1990:1 10 2.83 0 0.0000 2.8300 0.8094 -1,8636 -1.8636 -0.6524 1990:2 0 0 0 0.3679 -0.3679 -0.1052 1.3346 -0.5290 0.4479 1990:3 0 0 0 0,2649 -0,26A9 -0.0758 -0.2177 -0.7467 -0.0731 1990:4 0 0 0 0.1907 -0.1907 -0.0545 -0.1568 -0.9035 -0.0526 1993:1 0 0 0 0.0099 -0.0099 -0.0028 -0.0082 -1.2857 -0.0027 1993:2 0 0 0 0.0071 -0.0071 -0.0020 -0.0059 -1.2915 -0.0020 1993:3 0 0 0 0.0051 -0.0051 -0.0015 -0.0042 -1.2958 -0.0014 1993:4 0 0 0 0.0037 -0.0037 -0.0011 -0.0030 -1.2968 -0.0010 B. 10 percent increase in tihe primr1ary deficit (G') fiianced by domiiestic debt, witi mtronetizalion of interest payments and debt repayment 1990:1 10 0 1.3 0.0000 0.0000 0.6637 0.6221 0.6221 0.2088 1990:2 0 0,2819 0 0.0000 0.2819 0.0806 -0.1856 0.4365 -0.0623 1990:3 0 0.2819 0 0.0366 0.2453 0.0701 -0.0527 0.3838 -0.0177 1990:4 0 0.2819 0 0.0630 0.2189 0,0626 -0.0744 0.3094 -0.0250 1993:1 0 0.2819 0 0.1274 0.1545 0.0442 -0,1273 -0.726i -0.0427 1993:2 0 0.2819 0 0.1283 0.1536 0.0439 -0.1281 -0,8541 -0.0430 1993:3 0 3.1119 -1.3 0.1291 2.9828 0.8531 -1.9923 -2,8464 -0.6686 1993:4 0 0 0 0,4975 -0.4975 -0.1423 1.3912 -1.4552 0.4669 C. 10 percentincrease in the primary deficit (C') financed by donmestic debt will itnterest payweutS,finaniced byfurtherdebt uni1til 1993:3, when out- statiding debt is monietized 1990:1 10 0 1.30 0 0 0.6637 0.6221 0,6921 0.2088 1990:2 0 0 0.13 0 0 0.0664 0.0622 0.6843 0.0209 1990:3 0 0 0.13 0 0 0.0664 0.0622 0.7465 0.0209 1990:4 0 0 0.13 0 0 0.0664 0.0622 0.8087 0.0209 1993:1 0 0 0.13 0 0 0.0664 0.0622 1.3686 0.0209 1993:2 0 0 0.13 0 0 0.0664 0.0622 1.4308 0.0209 1993:3 0 6.495 -2.99 0 6.4948 0.3311 -4.5517 -3.1208 -1.5275 1993:4 0 0 0 0.8443 -0.8443 -0.2415 3.063 -0.0579 1.0279 Note: Variables are defined as folloiws: dG', percentage change in C'; dH, percentage change in EH; dB, percentage change in 8; w, inflation rate; dim, percentage change in real base money, Ih; dnfa, percentage change in real private net financial assets, utfa; dr, change in real Interest rate, r; db(d), percentage t change in demand for government bonds. Simulation results for 1991:1-1992:4 are omitted for brevity. 482 Zimbabwe: Fiscal Disequilibria and Low Growthi payments financed through the creation of base money, the situation tends toward an unsustainable path of positive, slightly increasing inflation rates. At some point it will be convenient to pay back the debt with a higher but dedining inflation rate. In the end the debt- financing strategy ends up being more inflationary after sixteen quar- ters (with a 1.88 percent cumulative inflation) than the alternative of resorting directly to monetization from the beginning. C. A 10 percent increase in the primary deficit, financed by domestic debt, with interest payments financed &yfitrther debt until 1993.3, when total debt is monetized. While the debt stock is increasing, the real interest rate also rises, but the price level is not affected since no change in base money has taken place (figure 10.4). This changes drastically when the government debt is paid back in 1993:3; the real interest rate falls significantly and inflation rises. After that both variables follow the path of the first simulation, but the size of the effects is different. The increase in base money has to be sufficiently large to pay back the accumulated debt, implying a drastic inflationary surge: cumulative inflatior in just five more quarters (up to 1994:4) exceeds significantly cumulative inflation in 1990-94. Figure 10.3. Effects of Primary Deficit on Inflation and the Real Interest Rate in Zimbabwe: Simulation A Percent 0.40 0.20 -/ . = - _ . _ .- 020=. 0.00 -020 -1.40 -0.60 -L.20 -1.60 -1.80 -2.00 - 1990 1991 1992 1993 - *--- Inflation -.~-.. t Real interest rate Source Author5' sinulations. Felipe Mormndi and Klaus Schmidt-Hebbel 483 Figure 10.4. Effects of Primary Deficit on Inflation and the Real Interest Rate in Zimbabwe: Simulation C Percent 1.50 0.50 0.00 -0.50 -2.00 -2.50 -3.30ITIIIIII;II;_ 1990 1991 1992 1993 1994 - Real interest rate Inflation Source. Authors' simulations. Despite their simplicity, the reported simulations show the dynamic sensitivity of the main endogenous variables-inflation and the real iterest rate-to govenmment policies for financing a budget deficit. Inflation ensues in all cases, but cumulative inflation never matches the inaease in base money. It is not the nonneutranity of the simulation system that is behind this result but the simulation assumption of a constant nominal exchange rate. Indeed, the empirical estimates of the coefficients of the inflation equation add up to almost 1. Moreover, when equation 10.4 is esfimated with the homogeneiof-degree-one restriction imposed on its coefficients, results do not change much. To account for nominal exchange rate devaluations, however a simulation was performed for the first case, adding a devaluation of 2.83 percent in 1990:1 and distinguishing between an almost neutral case (the one presented above) and a fully neutral case by imposing the restrction that the estimated coefficents of equation 10.4 add up to 1. The qualita- tive results do not differ much from those reported above. Real interest rates follow very different patterns that depend on how the government finances its deficit and on the assumption of fixed nominal interest rates. The simulations illustrate Sargent and Table 10.9. Public and Private Sector Saving and Investment, Zimbabwe, Fiscal 1980(81-1988189 hongtl 1980181 1981182 1982/83 1983184 1984185 1985186 1986187 1987188 1988/89 Cuirretnt price investment and savitlg (milliotns of Zifnbanive dollars) Foreign saving 270.8 451.8 440.8 293.1 166.2 91.1 1.7 -45.8 -30.8 Gross national saving 544.6 611.1 610.0 810.4 1,173.8 1,453.5 1,658.1 2,026.5 2,544.1 Central government -186.4 -89.9 -55.1 -279.8 -306.6 -386.1 -575.6 -155.6 -170.4 PLA 30.6 -45.0 -92.8 51.3 105.3 152.3 242.1 157.5 132.0 Nonfinancial public sector -155.8 -134.9 -147.9 -228.5 -201.3 -233.8 -333.5 1.9 -38.4 Private sector 700.4 746.0 757.9 1,038.9 1,375.1 1,687.3 1,991.6 2,024.6 2,582.5 Gross domestic investment 815.4 1062.8 1,050.9 1,103.5 1,340.0 1,544.5 1,659.8 1,980.7 2,513.3 Central government 65.1 122.2 191.9 208.7 203.2 221.2 293.1 485.0 523.0 PLA 136.2 394.2 412.0 479.6 447.2 643.5 615.3 587.4 600.0 Nonfinancial public sector 201.3 516.4 603.9 688.3 650.4 864.7 908.4 1,072.4 1,123.0 Private sector 614.1 546.4 447.0 415.2 689.6 679.8 751.4 908.3 1,390.3 Nonfinancial public sector deficit 357.1 651.3 751.8 916.8 851.7 1,098.5 1,241.9 1,070.5 1,161.4 Private sector deficit -86.3 -199.5 -311.0 -623.7 -685.5 -1,007.4 -1,240.2 -1,116.3 -1,192.2 Ratios of inwestrnent and saving ratios to GDP Foreign saving 0.069 0.094 0.077 0.046 0.025 0.012 0.000 -0.005 -0.003 Gross national saving 0.139 0.127 0.106 0.128 0.175 0,190 0.193 0.207 0.219 Central government -0.048 -0.019 -0,010 -0.044 -0.046 -0.050 -0.067 -0.016 -0.015 PLA 0.008 -0.009 -0,016 0.008 0.016 0.020 0.028 0.016 0.011 Nonfinancial public sector -0,040 -0.028 -0,026 -0.036 -0.030 -0.031 -0,039 0,000 -0.003 Private sector 0.179 0.155 0.132 0.163 0.205 0.220 0.231 0.207 0.223 Gross domestic investment 0.209 0.221 0.183 0.174 0.200 0.202 0.193 0.202 0.217 Central government 0.017 0.025 0.033 0.033 0,030 0.029 0.034 0.050 0.045 PLA 0,035 0,082 0.072 0.075 0.067 0.084 0.071 0.060 0.052 Nonfinancial public sector 0.052 0,107 0.105 0.108 0.097 0.113 0.105 0.110 0.097 Private sector 0.157 0.113 0.078 0.065 0.103 0.089 0.087 0.093 0.120 Nonfinancial public sector deficit 0.091 0.135 0.131 0.144 0,127 0.143 0.144 0.109 0.100 Private sector deficit -0,022 -0.041 -0.054 -0.098 -0.102 -0.132 -0.144 -0.114 -0.103 Source: Reserve Bank of Zimbabwe, various years; Schmldt-Hebbel 1990. Felipe Morandt and Klaus Schmidt-Hebbel 485 Wallace's (1981) "unpleasant monetarist arithmetic"; debt financing of government deficits in Zimbabwe would ordy postpone inflation to the future. Of course, this is true as long as government debt cannot be increased beyond some point. The issue, then, is when this point is achieved; the answer will depend on both macroeconomic condi- tions (that is, the extent to which private consumption and private investment can continue to be restricted) and, to a lesser degree, on conditions in financial markets. We have found that cihanges in deficit-financing decisions have some effects on real interest rates that could destabiize financial markets, especially when nominal interest rates are fixed for long periods of time, as assumed here. Finaly, the implications of a purely monetary policy that is effected through, say, the required reserve ratio, should be equivalent to those of an increase in base money. Both impinge on the government bud- get constraint sooner or later. Crowding-Out or Private Consumption and Private Investment This section goes a step further in analyzing the macroeconomic im- plications of public sector deficits by assessing the impact of the pub- lic sector on private sector spending- The focus is on the sensitivity to fiscal variables of private consumption and investment, in addition to the indirect effects of fiscal variables through interest rates, inflation, and private disposable income. How private saving and capital for- mation are affected by fiscal policies has significant implications for both short-term stabilization and long-run growth prospects. Table 10-9 presents data on Zimbabwe's saving and investment record, by sectors, from 1980/81 to 1988189.14 Between 1981/82 and 1987/88 a mnajor extemal adjustment took place that was reflected in a reduction of the current account deficit by 10 percentage points of GDP. This improvement came exclusively from the private sectoL Whereas until 1986187 the cNRpS defict hovered around 14 percent of CDDP, the private sector surplus rose from 4.1 percent of CDP in 1981182 to 14A percent of Gw in 1986/87. In fact, in 1986187, when the public deficit matched its previous record, 100 percent of that deficit was financed by the private sector As mentioned above, a partial public sector adjustment was initiated in 1987188, bringing about a reduction of 3.5 percentage points in the deficit in that year and an additional 0.9 percentage point dedine in 1988189. The private sector benefited directly from this decline, with a similar xeduction in its required surplus. To generate a surplus that financed 100 percent or more of the public deficit after 1986/87, the private sector significantly raised its saving rate: since 1984185 the rate has exceeded 20 percent of GDP, 486 Zimbalbwe: Fiscal Disequilibria and Low Growth which is larger than the economy's gross domestic investment. This private saving rate is extremely high for a developing economy-a counterpart of very low private consumption rates that barely ex- ceeded 50 percent of GDP. High private saving, channeled through Zimbabwe's developed financial system to the public sector, is a re- sult of restrictions on private consumption (particularly on imported consumer durables) and on formal or illegal capital outflows, coupled with a perception by the private sector tat the domestic financial system is stable. Aggregate or domestic gross investment did not show a strong downward trend during the 1980s. However, when in 1986/87 the public deficit returned to its 1983/84 peak, the domestic investment rate was a couple of percentage points lower than in the period 1980181 to 1981182, when the high deficits started. Conversely, when fiscal adjustment took place after 1987188-1988189, the domestic in- vestment rate recovered by 2.4 percentage points of GDP. The sector composition of investment changed significantly with the fiscal ex- pansion of the early 1980s. In fact, the deficit increased approximately one-to-one with public investment, while private investment fell. With fiscal adjustment after 1986187, both the absolute level and the share of private investment in domestic capital formation recovered, with a rise of more than 3 percentage points in the private investment rate, while public investment did not suffer significantly. The fact that both total investment and the share of private invest- ment recovered under fiscal adjustment is an encouraging sign. Growth-which was rather modest throughout the 1980s-depends on the quantity and quality of investment, and a higher share of private investment is likely to improve quality- Hence additional in- vestment gains, particularly in the private sector, could be positively influenced by continued fiscal adjustment. Fiscal adjustment should rely on additional gains in public saving, over and above the increase of the public saving rate from -3.9 percent in 1986187 to -0.3 percent in 1988/89. Private Conswnption This section, based on a framework developed by Corbo and Schmidt- Hebbel (1991), addresses the effects of public policies on consumption in Zimbabwe. Private consumption, expressed as a ratio to private disposable income, depends on neodassical determinants (perma- nent disposable income, interest rates, and relative prices), Keyne- sian variables and borrowing constraints (current income, consumer credit, money, and foreign saving), public saving, inflation, and pub- lic spending on private goods. The presence of (permanent) public saving reflects two very different hypotheses- The first, Ricardian Felipe Morandt and Klaus Schrnidt-Hebbel 487 equivalence, states that private consumption increases one-to-one with permanent public saving.'5 The second, direct crowding-out, asserts that under an institutional arrangement whereby the public sector captures private saving either directly or through domestic fi- nancial markets, current private saving is crowded out one-to-one by current public saving. In the case of Zimbabwe the second interpreta- tion is probably more relevant than the assumption of rational, forward-looking pnrvate consumers who internalize the public sec- tor s intertemporal budget constraint. The following specification for the ratio of private consumption to private disposable income reflects these variables.16 In addition it allows for testing the simple Keynesian, permanent incorne. and Ricardian or direct-cowding-out hypotheses.)7 c ~ ~ lY PSq (10.9) Pt +0 + 1 -+a + 33 r + 04rd + p DY,, DYt DY,,1i -+R DYri + 37D + ,aFs + /3 + l where C. is private consumption expenditure, DYp is current private disposable income,16 PDYp is pernanent private disposable income, PSg is permanent public saving r. is the consumption-based real interest rate, rc is the expected rate of change of the private consump- tion deflator, Pc,,, and P0, are the deflators for imported and national private consumption goods, respectively, cr'r is the sum of public expenditure on privately appropriated services and direct transfers to consumers,19 H is base money, FS is foreign saving, CC is bankdng sector credit to consumers, and 7ij is a stochastic error term.20 Expected signs of the coefficents are g 03, (2, 17, & '39 > 0; 06 < 0; 1t134 $s ' 0. Table 10.10 reports the main estimation results for equation 10.9, using annual data for the 1965-88 period. The complete specification yields not very satisfactory results for both expectational altema- tives. Most variables are not statistically significant, and two bor- rowing constraints (consumer credit and base money), have oppo- site signs from those expected a priori, although the coefficients are not significant. Less surprising is the low significance of the infla- tion and interest rates, which have ambiguous a priori signs. As in many other developing countries (see, for instance, the cross- country studies by Corbo and Schmidt-Hebbel 1991, Giovanrrui 1985, and Schmidt-Hebbel, Webb, and Corsetti 1992), interest rates are not sgnificant. This may be either because substitution, income, and wealth effects offset each other or because direct crowding-out Table 10.10. Estimation Results for Private Consumption, Zimbabwe, 1965-88 Durbil- Eqiation Constatlt PDYpJDYp PSJ/DYp PrIP, CcDYp H/DY, FS/DYJ, D74 D6586 p R2A Walsot Static expectatiotis .1 OLS 1.10 0.01 0.21 -0.53 -0.30 -0.15 -4.64 -0.86 0.08 0.62 1.91 (8.4) (0.1) (0.4) (-0.8) (-0.4) (-2.3) (-1.3) (-1.1) (0.4) 1,2 ML 0.61 0.12 0.67 -0.06 0.06 0.72 0.50 1.61 (7.6) (1.7) (3.3) (-2.0) (2.4) (5.0) Partial perfect foresigh t 2.1 OLS 1.06 0.02 0.08 -0.39 -0.07 -0.16 -3.9 -0.93 0.06 0.62 1.97 (8.0) (0.1) (0.1) (-0.6) (-0.1) (-2.1) (-0.7) (-0.8) (0.3) 2.2 ML 0.66 0.09 0.05 -0.03 0.03 0.13 0.16 1.70 (10.6) (1.7) (0.2) (-0.8) (1.8) (0.6) Note: The dependent variable is the ratio of private consumption to private disposable income (CplDYp). Numbers in parentheses are I-statistics. Estimation methods are ordinary least squares (oLs) and maximum likelihood with first-order autoregressive residuals (NIL); p Is the first-order residual correlation coefficient, and R2A is adjusted R2. The results that Include cmr are not reported here because of the high positive value of its coefficient, which seriously affects the signs and significance levels of other variables. A blank denotes the omission of the specific variable from the regression. Felipe Morandd and KJaus Schmidt -Hebbel 489 of public savings dominates the role of intertemporal substitution in consumption. We next followed a different approach, concentrating on the Key- nesian (current income), permanent income, and Ricardian direct- crowding-out (public saving) determinants. WrT.en dummies for the 1987/88 structural decline in private consumption and the 1984 outlier are added to these variables, the results reported in rows 1.2 and 2.2 in table 10.10 are obtained. Both the overall fit and the separate signif- icance of the contributing variables are more acceptable under the static-expectations alternative for permanent income and permanent public saving. The magnitude of current income is surprisingly high compared with permanent income-a feature that is even more extreme under the partial perfect foresight specification. In fact, the relative magni- tudes of current and permanent income (0.61 and 0.12) are higher than the 0.60 and 0.24 values obtained for tiirteen developing coun- tries with a similar methodology. (See the panel data results reported in Corbo and Schmidt-Hebbel 1991.) This suggests that current income is a stringent borrowing constraint that effectively limits inter- temporal consumption smoothing. By contrast, public saving strongly affects private consumption in Zimbabwe under the static-expectations alternative. The fact that the current public saving alternative (the measure for permanent public saving under static expectations) is significant whereas the three-year moving average of current and future public saving is not (under partial perfect foresight) confirms the initial presumption that it is direct crowding-out of private saving by public saving and tnot Ricardian anticipation of future taxes that is behind this high value. * The main conclusion of our results points toward the over- whelming dominance of the direct effects on private consumption of public sector defidts (or dissaving) over other indirect effects of deficit financing (via interest or inflation rates). An increase in the deficit of I Zimbabwe dollar (Z$), caused by a corresponding rise in public consumption, reduces private consumption by Z$0.67, with- out significant additional effects through interest and inflation rates.21 Private Investment Private investment is specified to depend on neoclassical profit and cost variables, borrowing constraints, and risk determinants. To avoid spurious correlation, nonstationary variables are scaled to GDP, yield- ing the following equation for the ratio of private investment to GDP:27 Table 10.11. Estimation Results for Private Investment, Zimbabwe, 1965-88 Equationi Constant P/P RIL Kpr11Y KS,/_Y PCOTIY PRO/Y FSfY HIY 1 OLS -0.09 -0.01 -0.27 0.22 -0.27 1.68 0.28 0.16 -0.13 (-0.8) (-0.1) (-1.5) (1.1) (-1.1) (2.4) (2.0) (1.7) (-0.2) 2OLS 0.11 -0.13 -0.42 -0.18 0.23 0.47 0.33 1.07 (1.2) (-3.7) (-2.6) (-1.3) (1.2) (3.5) (3.9) (3.5) 3 ML 0,16 -0.16 -0.41 -0.21 0.25 0.40 0.34 1.29 (2.4) (-6.5) (-3.3) (-2.4) (2.0) (4.1) (5.1) (6.0) 4 OLS 0.09 -0.12 -0.45 .-0.12 0.14 0.49 0.31 1.08 (2.2) (-6.5) (-5.7) (-2.8) (2.1) (4.8) (5.6) (6.3) 5 ML 0.14 -0.14 -0.49 -0.14 0.15 0.44 0.30 1.31 (2.2) (-6.8) (-5.7) (-2.8) (2.1) (4.8) (5.6) (6.3) Durbin- FC/Y VPWp VRIL VY D7375. D84 p R2A Watson 1 OLS 0.18 0.66 1.68 -0.0001 0.01 0.02 0.95 2.26 (1.7) (0.9) (0.9) (-1.9) (1.5) (1.0) 2oLS -0.37 1.12 0.02 0.03 0.93 2.76 (-0.5) (0.5) (2.3) (1.4) 3 ML -0.62 -0.86 0.02 0.06 -0.70 0.99 2.40 (-1.1) (-0.4) (4.1) (2.3) (-3.5) 4 OLS 0.02 0.05 0.94 2.74 (2.3) (4.1) 5 ML 0.02 0.06 -0.66 0.98 2.32 (4.0) (5.2) (-3.2) Note: The dependent variable is the ratio of private Investment to GDP (I4Y). Numbers in parentheses are 1-statistics; p is the first-order residual correlation coefficient, and R2A is adjusted RZ. Estimation methods are ordinary least squares (oLs) and maximum likelihood with first-order autoregres- sive residuals (ML). A blank denotes the omission of the specific variable from the regression. Felipe MomndV and Klaus SchArnidt-Hebbel 491 I,,, r /"4,,,Pi PCoT, Kg,... PRrRO FCT (10.10) = -IPUCK,, PMPKIC, v gPCO, - H, t Y, Yr Y, Y, VUCK,, VY, *ye t )1 where Iv is private fixed-capital investment, Y is GDP, UCK is the user cost of capital,23 PUCK is estimated permanent UCK, MPK is the margi- nal product of capital (approximated by the average product of capital equal to the ratio of current-period GDP to lagged end-of-period pri- vate sector capital stock), PMPK is estimated permanent MPK, Pj1piP , is the price ratio of imported and national private investment compo- nents, COT is corporate tax revenue, PcoT is estimated permanent COT, - is lagged end-of-period public sector capital stock, PRO is corporate profits, FC is banking credit flaws to firms, H is base money, FS is foreign saving, VUCK' is the coefficient of variation of ucKc, and VY is the coefficient of variation of GcDP. Expected signs of the corresponding partial derivatives are shown below each variable. Expected investment inflation is based on an autoregressive struc- ture. All expected permanent variables are specified according to par- tial perfect foresight or static expectations. The two coefficients of variation, which reflect risk and the incidence of investment, are defined as five-period moving coefficients based on two periods back, the current period, and two periods into the future. A linear form of equation 10.11 was estimated for Zimbabwe using ratios of annual private investment to GDP for the 1965-88 period (table 10.11). Some differences arise between equation 10.10 and the reported results. First, better results were obtained when the user cost of capital was split into its two components, the relative price of investment goods (PSIP) and the sum of the relevant real interest rate and the depredation rate (muL). Second, for L, as well as for other variables involving estmates of permanent values (the relative price of investment goods, the marginal product of capital, and corporate tax revenue), only the static-expectations results are reported. - The results are very satisfactory. Most neoclassical, borrowing- constraint, and uncertainty variables present the expected signs and are highly significant. The results for the most general specification are reported in row 1 of table 10.11, although not many degrees of freedom are left over. Of all the variables, only the ratio of corporate tax revenue to output has an opposite sign-with a significant coefficient-from what was expected a priori. This variable, in addi- tion to the ratio of firm credit to output and the coefficient of variation of CDP, is deleted from the next set of results. 492 Zirnzbabwe: Fiscal Disequilipria and Low G rowtII The two components of the user cost of capital are highly significant. The magnitudes of their signs differ: private investors in Zimbabwe react three times as strongly to the real interest rate as to the relative price of investment goods. The ratio of private capital stock to output (the inverse of the current average product of capital) presents the correct sign but achieves acceptable significance levels only under the maximum likelihood estimations, correcting for residual first-order cor- relation. In addition, the magnitude of the ratio is small in relation to that of the real interest rate. The significant role of the ratio of the public capital stock to output (with a coefficient similar in magnitude and significance to that of the ratio of the private capital stock) suggests a strong complementarity between public and private capital in Zimbabwe. This crowding-in of rpivate investment by public capital is an important result that reflects the importance of the composition of public expenditure for the coun- try's growth prospects. Two flow variables (firm profits and foreign lending as reflected in the current account deficit) and one stock variable (base money) sug- gest the significant role of the borrowing constraints faced by private investors. This result is not surprising for a period dominated by inter- est rate controls, which were partially relaxed only in recent years. Even under complete domestic financial liberalization, one should expect that borrowing constraints will affect private capital formation, in addition to the influence of completely deregulated interest rates. Finally, there is only weak evidence for the effect of our uncertainty proxies on private capital formation. In the most general specification (row 1 of table 10.11), the coefficient of variation of GDP affects private investment negatively and significantly. In rows 2 and 3 the variation of the relative price of investment goods (vPIP) is negative but weak; its coefficient does not achieve acceptable significance levels. The results support the notion that public deficits and their struc- tures affect private investment in Zimbabwe through various direct and indirect channels. Real interest rates have a strong negative influ- ence on private investment. Hence domestic debt financing of public deficits, which tends to push up interest rates-as observed during the 1980s in Zimbabwe-has a significant crowding-out effect. Public investment, by contrast, has a significant crowding-in effect. For each 1 percentage point of GDP increase in public investment (which raises the ratio of public capital stock to GDP by a similar amount), private investment rises by 0.15-0.25 percentage point of GDP. External Accounts, Real Exchange Rates, and the Public Deficit The next step in analyzing the nacroeconomic implications of public sector deficits is to evaluate their effects on Zimbabwe's external Felipe Morandd and Klaus Sebmidt-Hebbed 493 accounts and the real exchange rate. These variables are influenced to a large extent by the foreign exchange allocation system. In deciding how to allocate foreign exchange, the foreign exchange allocation commission first makes a projection of the availability of net foreign capital inflows, subject to the government's foreign debt target. The commission then projects total exports on the basis of different assumptions regarding foreign prices and domestic supply condi- tions, such as future harvests of the main crops. These projections of total available foreign exchange resources provide the basis for. the allocation of foreign exchange to importers of capital, intermediate, and consumer goods, which proceeds according to sectoral and his- torical criteria. The comniission's efforts to decrease dependence on foreign lend- ing led to a significant reduction of the current account deficit during the 1980s, from 10.3 percent of GDP in 1982 to a small surplus in 1988. As mentioned above, this external adjustment fell entirely on the private sector, since the public deficit did not decline below 10 percent of GDP throughout the 1980s. The government, however, does not only manage the quantitative mechanism of centrally allocating foreign exchange; it also controls the nominal exchange rate and sets import tariffs and quantitative trade restrictions. The exchange rate policy could be important for the foreign exchange projection for total exports; exports are sensitive to the real exchange rate, which in turn can be affected by the nominal exchange rate policy. Trade taxes, however, seem to have been set more to raise fiscal revenue than to protect national production or limit imports. Despite this, custom duties have increased significantly during the past decade, contributing to reduced imports. The Model In estimating the relationship between extemal variables, such as the trade balance and the real exchange rate, and fiscal policies, the two- stage framework of Rodriguez (chapter 2 in this volume) is amended in two ways. First, less emphasis is placed on the accumulation of net foreign assets (or debt) as the long-term driving force because of the lack of access by Zimbabwe's private sector to foreign financial mar- kets. Second, the deterrnination of the trade balance involves a two- step procedure. At the start of each year the government projects the trade balance on the basis of the difference between income and absorption, which is equivalent to specifying a behavioral relation. At the same time, the government projects total exports on the basis of estimated values of the appropriate relative prices-the terms of trade and the real exchange rate for exports. Given the projected trade balance and export levels, the government instructs the foreign Table 10.12. Estimation Results for Real Exchange Rates, Trade Surplus, and Exports, Zimbabwe, 1965-88 Durbin- Dependent variable Constatit ln(G/Y) ln(GN/G) ln(TS_-/M In(TM) tA p R2A Watson 1. Export real exchange rate (ex)a -0.94 -0.52 -0.23 0.06 0.37 -0.26 0.58 0.61 1.78 (-4.13) (-3.84) (-1.58) (2.15) (1.77) (-1.15) 2. Import real exchange rate (em)a -0.55 -0.38 -0.11 0.06 -060 0.15 0.68 0.53 1.79 (-2.16) (-2.74) (-0.85) (2.02) (-2.93) (0.93) Durbin- Conlstanit ird ln(ODrjM In(TSIY)-L In(CA/Y) ln(B-h/Y 7Iax p RzA Watson 3. Trade surplus as share of GDP (TSfY) -4.67 -3.47 --0.65 0.38 -0.24 0.46 0.78 1.76 (-0.94) (-0.62) (-2.78) (2.16) (-2.59) (1.09) 4. ln(TSIY) -3.62 -2.3 -0.58 0.35 -0.23 -0.56 0.76 1.75 (-0.82) (-0.44) (-2.36) (1.95) (-2.28) . (-0.40) Durbih- Constant ln{ex) ln(TCT lntY/YP) Rho RZA Watson 5. Export share of GDP (XIY) -1.25 0.54 -0.43 0.87 0.70 1.78 (-15.6) (3.8) (-1.*) 6. In (XIY) -1.31 0.25 -0.52 0.78 0.55 1.68 (-17.3) (1.3) (-1.5) - Note: Numbers In parentheses are I-statistics; p is the first-order residual correlation coefficient, and R2A is adjusted R2. The simulation method is generalized least squares, with a maximum-likelihood procedure to correct for first-order autocorrelation and instrumental variables to correct for possible simultanelty bias. A blank denotes the omission of the specific variable from the regresE'on. a. The deflator is the average wage index. Felipe Morandd and Klaus Schmidt-Hebbel 495 exchange allocation commission to allocate projected available foreign exchange to imports. Naturally, projected and actual foreign exchange resources differ as a result of unexpected changes in the exogenous variables driving the trade balance and exports. The goverrnent can also affect the trade balance through its exchange rate policy, at least in the short term. The extent of these effects can be tested in the empirical setup that follows. These amend- ments result in a model comparable to that of Rodriguez in one important respect: fiscal policies are still reflected by the trade balance and real exchange rate equations. The relative prices of (or relevant exchange rates for) exports and imports are specified as: (10.11) ex (PXIPN) = ex(7T* tM, TSIY, G/Y, GN1G) (10.12) em (PM/PN) = em(T, tM, TSIY, GIY, GNfG). The ratio of the trade surplus to GDP is given by: (10.13) TSP! = ts(ird, ODG0Y, CAPY, B11IY, irtax. (-) C-) (-) C-) C-)- Finally, the export function is specified as: (10.14) X/Y = xQTT, Y/YP) or X/Y = x'(ex, YIYP) where ex is the real exchange rate for exports, em is the real exchange rate for imports, T'S is the trade surplus, X is total exports, Y is CDP, Px is the price index of exports, PM is the price index of imports, PN is the price index of nontradable goods, TP is foreign terms of trade, TT is domestic terms of trade, tM is the average tariff rate, G is govem- ment spending (public consumption plus public investment),24 GO is government spending on nontradable goods, i is the average domes- tic interest rate, ird is the uncovered interest rate differential, ODG is the operational public sector deficit, CA is the capital account surplus, B-1 is lagged-end-of-period domestic public sector debt stock, YP is potential GDP, and rtax is the inflation tax. Empirical Results Equations 10.11 through 10.14 were estimated in log-linear form, using annual data for 1965-88 (see table 10. 12).25 Both real exchange rates show significantly negative elasticities with respect to the share of government spending in CDP, which confirms the theoretical prediction. In addition, the share of govem- ment spending on nontradable goods negatively affects both real 496 Zimbabwe: Fiscal Disequlfibria and Low Growth exchange rates, although this effect is not significantly different from zero. Foreign terms of trade also exhibit the expected sign in both equations, but they affect more significantly the real impart exchange rate. The implicit tariff rate also shows expected signs but again is not significantly different from zero. (In the real export exchange rate, the theoretical sign is ambiguous.) Finally, the effect of the one-year- lagged trade surplus is small but significant and shows the expected sign.26 Two alternative specifications were estimated for the trade surplus, depending on how the government finances its deficit. First the debt- output ratio (B.lJY) was specified following Rodriguez (chapter 2 in this volume). We then tried inflationary financing by indluding the inflation tax (7rtax)- The overall adjustment is slightly better for the former option, but the coefficient of &11Y has the incorrect sign. In terms of other individual variables, all coefficients show the expected signs in both equations. The interest rate differential is not signifi- cantly different from zero in both estimated versions of equation 10.13. The results were not improved by using an alternative defini- tion of ird based on the actual implicit interest rate paid by Zimbabwe on its foreign debt. Most interesting are the estimated effects of the operational deficit of the public sector, on the one hand, and the ratio of the capital account surplus to GDP (CAIY), on the other. The latter variable is used as a flow proxy for net foreign assets (NFA) in the Rodriguez setup, and it seems even more appropriate than NFA in the Zimbab- wean context, given the way the government allocates foreign exchange. As one could expect from a theoretical point of view, the effect of CAIY on the ratio of trade surplus to GDP is negative: the more foreign funds flow in, the more finanacng is available for imports without resort to increased exports. What the estimated elas- ticties suggest is that an increase in capital inflows does not bring an equal decline in the trade surplus but a substantially smaller reduc- tion27 This result reflects government policy to save some of those capital inflows in the form of foreign reserve accumulation and so reduce net foreign indebtedness. In the case of the operational deficit of the publc sector, the estima- tions indicate that an increase of 1 percentage point of GDP in this variable lowers the trade surplus by 0.6 percentage point of GDP. This confirms the theoretical presumption: the rise in ODr, inaeases absorption and thus, for a given income level, reduces TS. However, the form by which this defict is financed does not influence the ratio of trade surplus to GDP to any significant extent. Neither the out- standing stock of public sector debt nor the inflation tax are statis- tically significant. The reason is clear: during the mid- to late 1980s most of the public deficit was financed by issuing domestic debt, Felipe Momndi and Klaus Schnzidt-Hebbel 497 which is held either compulsorily-as is true of the share held by institutional investors-or, in the case of private savers, because alter- native portfolio choices are lacking. In addition, private saving has significantly increased since the early 1980s because of restrictions on imports of foreign goods. Private investors have not been so much affected by the public indebtedness process as by the restrctions on the acquisition of foreign capital goods imposed by the foreign exchange allocation system. In the end, it is not surprising that the increase in public debt has not been reflected in a lower trade surplus. The two are, temporarily at least, disconnected from each other- The low significance of the inflation tax is even less surprising, as the public sector has resorted to debt to finance its deficits. Exports were also specified according to two different versions. The first is more in agreement with the spirit of the Rodriguez model because it uses the domestic terms of trade as the relevant relative price variable, while the second specifies the export real exchange rate. Both versions indlude the ratio of current to potential CDP as an additional explanatory variable. The reasoning is that the higher this ratio, the lower is the share of exportable goods produced that effec- tively ends up in foreign markets. Both the sign of this variable and the signs of either measure of relative prices are correct, although Y/YP does not reach conventional significance levels. The reported empirical results confirm the influence of public defi- cits and spending in Zimbabwe's external sector. They reflect the particular way in which Zimbabwe's government regulates imports through the foreign exchange allocation conunission and restricts capital flows. Growth Prospects This section assesses Zimbabwe's growth prospects in connection with the earlier discussion on the macroeconomic effects of public sector deficits. A behavioral function for the relative output supply (the ratio of actual to potential GDP) is specified, following a standard neodlassical setup in which output depends on relative factor prices and prices of intermediate goods.28 This sheds light on the effect of domestic investment and the incentive system on this ratio and on future growth prospects. The final discussion of the effects of public deficits and structural distortions focuses on the overall performance of the Zimbabwean economy and its growth prospects. Growth A neoclassical output supply is specified for GD? by substituting con- ditional factor demands into an aggregate production function that Table 10.13. Estimation Results for Relative -Output Supply, Zimbabwe, 1966-88 * ~~~~~~~~Durbi!n- Eqialtioti Y a 61 _2 R2A Walson 1, NLLS -0.10 0.45 0.92 0.19 -0.13 -0.25 0.92 2.26 (-4.7) (4.7) (6.7) (1.2) (-5,8) (-9.2) 2, NLTSLS -0.11 0.44 0.85 -0.12 -0,25 0.91 .2.12 (-2,9) (2.6) (3.1) (-3.7) (-5.6) 'IO 3. NLLSMS -0.11 0.40 0.80 -0.12 -0,24 0.91 2.12 co (-5.9) (4.6) (0.14) (-5.9) (-9.3) Note: The equation Is: ln(ylyp) = y + X la In(P/WeM-') + (I - a) ln(P/P,,)I + t (rp- 0.05) + 61°D + 6.20 The first dummy is 1.0 for 1974,1975, 1976, 1980, 1984 (O otherwlse) and the second dummy, for the stronger recessionary years, is l for 1977, 1978, and 1979 (0 otherwise). Estimatlon methods are nonlinear least squares (NLLS) and nonlinear two-stage least squares (NLTSLS). The NLTStS estimation uses the following list of instruments: the constant; the lagged values of the logarithms of the productivity-adjusted real wage, the real price of Intermediate imports, and the dependent variable; and the contemporaneous values of the two dummies, the real Interest rate, and the log of the ratio of public expenditure to potential output. Numbers in parentheses are I-statistics, Felipe Morand and Klaus Sdimidt-Hebbel 499 depends on capital, variable factors (labor and working capital), and intermediate imports. By substituting potential output for capital, aggregate supply can be restated as the ratio of actual and potential output levels, defined as a function of the real wage adjusted for productiity gains, the real exchange rate relevant for intermediate imports, the real interest rate relevant for working capital, and period-specific dunmnies for Zimbabwe's conflictive preindepen- dence period. if PI (10.15) ln(- e+ aln- t+(1-a)ln-& + f(rj,-0.05) + ZFiPD5 where y is actual CDP, yp is potential GDP, P is the GDP deflator, W is the nominal unit wage, Plmpir,t is the price of intermediate imports, t is time, L1 represents supply-specific dwnnies, and rp is the real inter- est rate relevant for production decisions (that is, taking into account the nominal lending rate and expected inflation of the GDP deflator).29 The results of equation 10.15 are reported in table 10.13. The first row shows the results for the complete specification, with a positive but not significant coefficent for the real interest rate, which is dropped from the subsequent estimations. The second row reports nonlinear two-stage least-squares (NTSLS) results to take care of possi- ble sinultaneity biases caused by the nonindependence of the real wage and the real price of imports stemming from the interaction of aggregate supply and demand. The results are not very different from the OLS results reported in the third row, in terms of both the excel- lent overal fit and the individual coefficients. The price-elasticity of aggregate supply is relatively low; it is 0.44 in the NrrsLs equation. This result implies that aggregate demanid shocks (for a given aggregate demand elasticity) will have a strong relative price respornse and a weak output effect. The coefficient at (the share of labor in gross output net of capital value added) is very high and significant, reflecting a strong weight of the real product wage in comparison with the real exdhange rate in determing short-run out- put. Finaly, 81 and 62 reflect the relative intensity of supply disrup- tions, which coincide mostly with the international oil shocks and the preindependence period of civil war. Deficits, Distortions, and Growth Zimbabwe's economic fragility increased during the 1980s and early 1990s, culminating in the particularly intense drought of 1992. But the country's zero per capita growth in twelve years is a symptom of 560 Zimnbaae: Fiscal Disequilibria and Low Growth deeper problems. Major fiscal disequilibria, a large and inefficient state enterprise sector, and a distorted incentive structure have led to a weak performance in private investment and exports. How are these problems reflected in the empirical results reported shown in the preceding sections? A sharp contraction in economic activity occrred in the early 1970s as oil shocks and domestic turoil hit the economy. After indepen- dence in 1980 a hesitant recovery began, initially financed by external debt. When, after 1983, foreign capital inflows fell, a major reduction in current account deficits was achieved during 1982-86 and was preserved until 1989. To finance the large deficit, the government turned to the private sedor, both high private saving and low private investment contnbuted to the large private surplus required to finance public deficits in the range of 10-14 percent of GDP. Since 1990, however, the public sector has again been turning to foreign financing as private surpluses dry up. Because the decline in private investment after the early 1980c was not matched by higher public investment, total investment in fixed capital has been decreasing as a share of cGP. It is in this sense that public deficit financing has been detrimental to Zimbabwe's growth prospects. But private inrvestment in Zimbabwe is not only crowded out by public deficits; it is also affected by an environment not sup- portive enough of private business. The scarcity of financial resources available for private investment, restricons on private business, a heavy tax burden, and linited access to intermediate and capital goods imports as a result of foreign exchange quotas and uncertain property rights inhibit private capital formationL These interventions affect both the level and the productivity of investment. Indeed, they generate a distorted relative price and overall incentive structure that induces investment to flow to sectors in whidh socal returs are low. Therefore both fiscal stabilizaticon and structural reforms are required to attain a higher quantity and quality of investment and improve Zimbabwe's growth prospects. Condusions and Policy Implications The Zimbabwean case is most interesting because of the coexistence of persistent large public deficits with a private surplus that is sizable m comparison with those in comparable developing countries. Pri- vate investment has been squeezed to less than 10 percent of GDP and private consumption to about 60 percent of GDP through a combina- tion of foreign exchange and capital fiow restrictions with compulsry public debt placement in a relatively developed-albeit managed- financial market. This has been possible at moderate inflation rates, Felipe Morandd and Klatis Schmidt-Hebbel 501 reflecting a generally conservative monetary policy and low, typically negative, real interest rates. This chapter has identified various determinants and consequences of public deficits in Zimbabwe. The results point to ten implications related to defict financing and the role of macroeconomic and fiscal policy variables in determining deficits. The large financing requirements of public deficits have led to a steady and massive buildup of total public liabilities, from 54 percent of GDP in 1980 to 81 percent in 1987, and well beyond 100 percent in 1992. Issuing money has traditionally been a minor source of defict financing, with the exception of the recent past, when the inflation tax rose substantially. * Among macroeconomic variables (and in decreasing order), real CDP growth, real import growth, and a real exchange rate devaluation have a negative impact on the public sector deficit. Also in decining order, higher domestic real interest rates, domestic inflation, and for- eign nominal interest rates boost the defict. Among central govern- ment policy variables, cuts in the wage bill, transfers and subsidies, public investment, and expenditure on goods and services affect the deficit. Further policy measures on the revenue side, such as tax reforms and reductions in the public enterprise deficit, have major and immediate effects on public finance. Reductions in outstanding domestic and foreign debt stocks affect deficits with a lag by lowering interest payments. * Measuring the role of fiscal policy variables in relation to vari- ables beyond the direct control of policymakers, we conclude that fiscal policy had an absolutely predominant role in shaping the size of deficits throughout the 1980s. Therefore policymakers-not domestic or foreign shocks-are to blame for the fiscal deterioration in the early 1980s and are to be praised for fiscal adjustment, such as the partial deficit reduction in the late '1980s. Bounds for sustainable public deficits were derived by relating pri- mary deficits and interest payments to financing sources: money, domestic debt, and foreign debt Sustainability was defined in the sense of holding constant the 1988 ratio of total public sector 1 iabilities to CDEP. The main conclusions follow. - Under a base scenario showing a macroeconomic environment similar to that of the recent past, the sustainable primary deficit is estimated at 1.7 percent of GDP. It increases to 2.9 percent of GDP under a high case of higher growth and lower real domestic interest rates. The corresponding nominal (primary plus interest payments) deficits are 9.9 and 10.7 percent of rOP, slightly lower than the actual 10 to 11 percent of GDP for the nominal deficit observed in 1987-91. 502 Zimbabwe: Fisca Disequilibri and Low Growth * Under a low-case scenario of low growth, high real domestic and foreign interest rates, and a real exchange rate depreciation of 7 per- cent per year, the sign of the sustainable deficit is reversed; a primary surplus of 4.2 percent of GDP (or a nominal deficit of 5.6 percent of GoP) is required to avoid unbounded growth of public sector lia- bilities. Since 1990, Zimbabwe has been experiencing such an adverse scenario. Low growth, real exchange rate depreciations, and rising drought-induced deficits are leading to a rapid buildup of total public sector debt. Next we reversed the focus by looking at the macroeconomric effects of public deficits. On the financial market implications of deficit financing, we concluded the following. e Since 1983 the govemment has intervened in financial markets in order to generate a private sector surplus that can be used to finance its large deficits. Reliance on domestic debt financing seems to be an optimal choice because it avoids inflationary finance. However, our simulations indicate that this solution is not sustainable in the long term; larger inflation should be anticipated eventually, when domes- tic financial markets become less wiing to absorb additional public liabilities- In spite of financial sector regulation and the large policy- induced private sector surplus, a moderate upward trend of real interest rates was observed during the 1980s as a result of increasing public debt (see also Elbadawi and Schmidt-Hebbel 1991b). On the question of how the large public deficts and their composi- tion have affected private saving and investment, Zimbabwe's recent experience suggests the following conclusions. * Between 198V82 and 1987188 a major external improvement was achieved, turning a current account deficit of 9A percent of GDP into a balanced account. This improvement relied exdusively on the private sector, as the public sector deficit hovered around 10-14 percent of GDP. Both an increase in private saving and a decline in private investment were behind the rise in the private surplus Private saving exceeded 20 percent of GDP and exceeded the economy's domestic investment between 1984185 and 1988189. Low private consumption was made possible through a combination of consumer import repression, strict controls on capital outflows, and a perception that the fiancial system was stable. The empirical results show that pri- vate consumption is strongly influenced by current private disposable income and by public saving. However, after 1989 the private sector surplus shrank, and in 1992 there was actually a deficit * The declining private investment up to 1986187 implies lower aggregate capital formation and, probably, a lower efficiency of domestic investment, which contributed to Zinbabwe's meager Fedipe Mormndi and Klaus Schmidt-Hebbel 503 growth. The effect on private investment of the partial fiscal adjust- ment in 198789 is encouraging because it allowed a recovery of 2.4 percentage points of the gross domestic investment rate. Our results indicate that real interest rates have a strong negative influence on private investment. Hence domestic debt financing of public sector deficits, which (as observed in the 1980s) tends to push up interest rates, has a significant indirec crowding-out effect. This is in part offset by crowding-in of private investment as a result of higher pub- lic investment. To what degree does the fiscal deficit spill over abroad? * Public deficits and public spending strongly affect the trade sur- plus and relative export and import prices in Zimbabwe. An increase of 5 percentage points of aDP in the pubic deficit lowers the trade surplus by 3 percentage points of GDP, causing an apprecation of the real exchange rate. The particular way in which Zimbabwe's govern- ment administers imports through the foreign exchange allocation comission and the binding restraints placed on capital movements are the central factors in this outcome. Finally, on growth prospects and required policy reforms, our find- ings are as follows. Crowding-out of private investment and declning public invest- ment, combined with the low quality of investment projects resulting from the distorted incentive structure, have substantially affected Zimbabwe's growth performance and potential. In addition, to the extent that imported intermediate and capital goods are not perfect substitutes for domestic goods, import compression and exchange controls have reduced utilization of existing capacity. The 1991 adjust- ment program has started to deal with macroeconomic disequllibria and structural distortions. The main conclusion of this study is that this adjustment program has to be deepened substantially along two lines. First, much more fiscal adjustment is needed to achieve macro- economic and financial stability and to foster higher growth. This point is strongly supported by a recent study on macroeconomic adjustment in Zinbabwe (Elbadawi and Schmidt-Hebbel 1991b). The simulation results in this study, which uses a macroeconomic general equilibrium model show that fiscal reform would help Zimbabwe. achieve a sustainable debt path, a dedine in interest rates paid on public debt, and a recovery of private consumption and mivestment. Second, much more progress is required in liberalizing domestic goods and financial markets as wel as foreign trade, restructuring and privatizing state enterprises, and deregulating private sector pro- duction and investment. Only when these conditions are met-and 504 Zimbabwe: Fiscal Disequilibria anid Low Growthi after allowing for a significant lapse of time required to raise the credibflity of the new policy framework-will Zimbabwe be able to start on a path of sustained growth and poverty reduction. Notes The authors thank Rob Davies and Lloyd McKay for illuminating discussions in Harare. They are also grateful to Bela Balassa, Ibrahim Elbadawi, Sarshar Khan, Steve O'Connell, Jiom Rattso, Ragnar Torvik, and Michael Walton for their conmments on previous drafts. Efficient research assistance was pro- vided by Maria Cristina Almero-Siochi, Rodney Chun, and Heinz Rudolph. Responsibility for any remaining errors is the authors'. 1. Among recent papers on Zimbabwe's macroeconomnic situation and prospects are Chhibber and others (1989); Dailani and Walton (1989); Khadr and others (1989); Davies and Rattso (1990); Elbadawi and Schnmidt-Hebbel (1991a, 1991b); Davies, Rattso, and Torvik (1993); and Mehlum and Rattso (1993). 2. The 1980-89 data presented in this section are based on Schmidt-Hebbel (1990), the most comprehensive attempt to date to build consistent consoli- dated and stock-flow data for nonfinancial public sector deficits and for nonfi- nancial. and financial public sector balance sheets. For a detailed discussion, see the above reference and Morand6 and Schmidt-Hebbel (1991). A first application to Zimbabwe of a framework for macroeconomic consistency in current and constant prices for a six-sector disaggregation (for 1981 and 1987) can be found in lChadr and Schmidt-Hebbel (1989a, 1989b). An application of the RMSM-X macroeconomic consistency model for a five-sector disaggrega- tion to Zimbabwe, covering the 1985-87 historical period and the 1988-95 projection period, was done by Khadr and others (1989). A significant exten- sion of the former, in terms of behavioral specification, sector disaggregation, and period coverage, is the macroeconomic general equilibrium model for Zimbabwe by Elbadawi and Schmidt-Hebbel (1991a, 1991b), with base year 1988 and simulations covering 1988-95. 3. Tax revenue functions were estimated separately for direct taxes, indirect taxes, and customs duties. The results in Morande and Schmidt-Hebbel (1991) show that tax revenues depend on relevant tax bases (ax' and imports) and tax reforms. An interesting findikg is that inflation boosts income tax payments as a result of income bracket creep, without any evidence of negative Olivera- Tanzi effects of inflation on any tax revenue category. 4. The average relative contribution of each group of deficit determinants to the explained variation of the deficit is computed as 889 1 I818g IZ dvxsignd,) f jd1l LI-81182 j1i-81182 where di is the explained change of the deficit in period i (the third-from-last line in table 10.4) and dvi is the change in the deficit caused by variable category v (the total variations resulting from domestic, foreign, and fiscal policy variables in table 10.4). Felipe Morandd and Klaus Sclhmidt-Hebbel 505 5. The semnielasticities were computed for 1987188 to 1988189. If the semi- elasticities changed over the 1980s, the values for the early 1980s (1981182 to 1982183) were added in parentheses after the 1987188 to 1988189 values. 6. This follows work on fiscal sustainabiity developed by Buiter (1983, 1985) and van Wijnbergen (1989), with applications such as that by van Wijn- bergen, Rocha, and Anand (1988) to Turkey. 7. The real exchange rate, denoted by e, is defined as (E P*WP), where E is the nominal exchange rate and Pt is the foreign price index. The domestic and foreign real interest rates are defined according to the Fisher equations in their simplified linear form. 8. Note that this equation determines the size of the primary deficit from the effects of inflation, output growth, the real exchange rate, and domestic and extemal interest rates on interest payments and deficit financing sources. Hence the direct effects of the first three variables on the primary deficit, considered in the preceding section, are omitted here. The same holds for.the simulations carried out below. For instance,- whereas a higher real devalua- tion worsens the sustainable deficit calculation in table 10.6 because of higher external real interest payments (for a given primary deficit), the net total effect-through both higher interest payments and a lower primary deficit- was shown in table 10.5 to reduce the deficit, as a result of the dominating effect on the primary deficit. 9. The pattem and depth of the financial system were inherited from the preindependence period and have remained intact, preventing the develop- ment of informal credit markets. This has been the result of a combination of factors: a strict regulatory framework, relatively conservative monetary and exchange rate policies, and high confidence in public debt as a result of strict debt servicing. 10. As an alternative to the money market equilibrium condition for deter- mining the real interest rate, one can use the equilibrium condition in the public sector bonds market, which also depends on the demand for money. .That is, b = bd nfa' - (lls)m(r, w%-1 nfa), where s is the money multiplier. This equation, in combination with the government budget constraint in equation 10.2, determines the amount of monetary financing of the deficit. 11. Some variable measurement and data features merit a discussion. Money is seasonally adjusted M1; the nominal interest rate is a weighted average of public sector stock and bonds, annual interest rates, and deposit rates at commercial banks (also on an annual basis); and net financial assets is the sum of private sector deposits in the financial system, including com- pulsory savings in pension funds and insurance companies. All series are deflated by the consumer price index (cn) of the rich, which is less affected by (at times pervasive) price controls than the cpi of the poor during the sample period. The same cPm of the rich is also used for actual and expected inflation. The data frequency is quarterly, and the sample period, determined by data availability, is 1979:1 to 1988:3 for most estimations. 12. In the reported results we do not restrict b1 to be equal to b2, although this was also tried. The estimates were similar but were not so dose as to reject the hypothesis that they are significantly different from each other. 13. Coefficients of other variables in equation 10.4, such as wages, were not significantly different from zero. 506 Ziminabrwe: Fiscal Disequilibrin anid Low Growth 14. The fiscal year macroeconomic aggregates of table 10.9 (foreign saving, national saving, gross domestic investment, and GCD) are consistent with calendar year data from national accounts. 15. Note that this refers to the separate effect of public saving (taxes minus current public expenditure) on consumption, in addition to the effect of dis- posable income (gross income minus taxes). 16. All nonstationary variables are scaled to current private disposable income to reduce the incidence of spurious correlation. An alternative pro- cedure, combining cointegration tests and dynamic error-correction models, is not feasible because of the short time series. 17. Three simple null hypotheses are tested with this specification: (a) Keynesian: l,B > 0° (¾ 6- 62 = 0; (b) permanent income hypothesis without Ricardian equivalence: 1,B > o, a1 = 02 = 0; and (c) Ricardian equivalence or direct-crowding-out hypotlheses: O = 0, (1 = 02 > 0° 18. For a precise definition of disposable income and public saving, see Corbo and Schmidt-Hebbel (1991). 19. Privately appropriated services paid by the government are measured as the sum of public expenditure on education and health. These, plus direct transfers to consumers, could reduce private consumption (if they are substi- tutes for the latter) or increase it (if they strongly crowd in complementary private expenditure categories). 20. We specify two alternatives for the expected permanent values of public saving in equation 10.9 and other variables in the investment equation (10.10) that follows. The first alternative is partial perfect foresight, defined as the simple average of the variables for the current period and two periods into the future. The second altermative is the static-expectations specification, which assigns a 100 percent weight to the current value. Similar assumptions are made with respect to expected consumption inflation (and expected investment inflation, below). A first altemative takes actual inflation between today and tomorrow as the relevant proxy for ratio- naly expected inflation. The second alternative is adaptive expectations, specifying the expected price change either from an autoregressive moving average (ARmA) backward-looking process or by assigriing 100 percent weight to the actual price change between yesterday and today1 consistent with static expectations. Permanent private disposable income is defined as the ratio of permanent to current cDP (see the last section of this chapter) multiplied by current private disposable income. 21. Similar results can be found in Elbadawi and Schmidt-Hebbel (1991a) in an estimated general equilibrium model for Zimbabwe. Our approach allows for more freedom in choosing lags and variables, but it is less explicit on potential feedbacks among variables. 22. This setup is similar to a private investment model implemented for Morocco by Schwnidt-Hebbel and Miiller (1992). 23. ucK is defined as the product of the investment deflator ratio to the CDP deflator (P,IP) and the sum of the real interest rate relevant for investment and the depreciation rate (Rn.). 24. Goverrunment spending has a negative effect on both real exchange rates when more than half of G is made up of nontradable goods. FPlipe Morandi atid Klaus Sclmnridt-Hebbel 507 25. The application of logarithms to variables with negative values necessi- tated adding a constant to them. In estimating equations 10.11. and 10.12, nontradable prices-the deflator in the definition of both ex and em-were proxied alternatively by the average wage index and the domestic price level. Results were clearly better when using the former, on *which the results reported in table 10.13 are based. Prices of exports and imports were proxied by the corresponding national accounts deflators. C, GNI TS, and Y are mea- sured at current prices. GN, government spending on health, housing, and education, stands for spending on nontradable goods. rr* was constructed as the ratio of export and import deflators, adjusted by the average tariff rate implicit in custom duty revenues. This implicit average tariff rate is also present in the regression as tM. The expected rate of devaluation was assumed to be equal to the actual rate, a perfect-foresight approximation of the rational-expectations hypothesis. The foreign interest rate is LIBOR, and the domestic interest rate is a weighted average of active rates in the financial sector. 26. The use of the lagged trade surplus rules out a potential simultaneity bias. We also used the current account deficit as an alternative to the lagged trade surplus, without success. 27. Since we are considering percentage rates of change, there could be some differences between 1 percent of CAIY and 1 percent of TSIY. On average, however, these differences are not large. 28. This section draws heavily on Elbadawi and Schmidt-Hebbel (1991a). 29. Note that equation 10.15 is not a reduced-form market equilibrium equation but a structural aggregate supply form. Hence no demand variables are included. The specification is homogeneous of degree zero in absolute prices. The real wage is adjusted for Harrod-neutralproductivity increases at an annual rate of A = 0.008, its 1965-72 trend growth rate, which is deemed to be representative for a normal period of productivity-related wage increases when the economy is operating at levels close to full employment. From 1972 to 1979 real wages stagnated; after 1979 they grew strongly, probably reflect- ing both the partial recovery of output and the change in political regine. References Balassa, Bela. 1989. 'The Effects of Interest Rates on Saving in Developing Countries." Policy Research Working Paper 56. World Bank, Office of the Vice President, Development Economics, Washington, D.C. Barro, Robert F. 1974. "Are Government Bonds Net Wealth?" Journal of Politi- cal Economy 82(6): 1095-1117. Bernheim, B. Douglas. 1987. "Ricardian Equivalence: An Evaluation of The- ory and Evidence." In NBER Macroeconomics Annual. Cambridge, Mass.; Massachusetts Institute of Technology Press. Buiter, Willem H. 1983. "Measurement of the Public- Sector Defiut and Its Implications for Policy Evaluation and Design." International Monetary Fund Staff Papers 30 (2): 30749. . 1985. "A Guide to Public Sector Debt and Deficits."' Eronomic Policy I (November): 13-79. 508 Zimbabwe: Fiscal Disequilibria and Low Growth Chhibber, Ajay, Joaquin Cottani, Resa Firuzabadi, and Michael Walton. 1989. "Inflation, Exchange Rates, and Fiscal Adjustment: The Case of Zim- babwe. Policy Research Working Paper 192. World Bank, Country Eco- nomics Department, Washington, D.C. Corbo, Vittorio, and Jaime de Melo. 1989. "External Shocks and Policy Reforms in the Southern Cone: A Reassessment." In Guillermo A. Calvo, ed., Debt, Stabilization and Development: Essays in Memory of Carlos Diaz Ale- jandro. Oxford, U.K.: Basil Blackwell. Corbo, Vittorio, and Klaus Schmidt-Hebbel. 1991. "Public Policies and Sav- ing in Developing Countries." Journal of Developing Economics 36 (1): 89-115. Dailaffi, Mansoor, and Michael Walton. '9S9. "Private Investment, Govern- ment Policy, and Foreign Capital in Zimbabwe." Policy Research Working Paper 248. World Bank, Country Economics Department and Southern Africa Department, Washington, D.C. Davies, Rob, and Jom Rattso. 1990. "Macroeconomic Policies for Medium Term Development: The Zimbabwe Case Study." University of Trondheim, Department of Economics, Trondheim, Norway. Davies, Rob, jorn Rattso, and Ragrar Torvik. 1993. "The Macroeconomics of Zimbabwe in the Eighties-A ccE-Model Analysis." University of Trondheim, Norway, Department of Economics. Dornbusch, Rudiger. 1983. "Real Interest Rates, Home Goods, and Optimal External Borrowing." Journal of Polifical Economy 91 (1): 141-53. . 1985. "Overborrowing: Three Case Studies." In Gordon W. Smith and John T Cuddington, eds., International Debt and the Developing Coun- tries. A World Bank Symposium. Washington, D.C. 1989. "Capital Flight: Theory, Measurement and Policy Issues." Mas- sachusetts Institute of Technology, Department of Economics, Cambridge, Mass. Easterly, William. 1989. "Fiscal Adjustment and Deficit Fnancing during the Cost Crisis." In Ishrat Husain and Ishac Diwan, eds., Dealing with the Debt Crsits. Washington, D.C.: World Bank. Easterly, Wlilliam, Carlos A. Rodriguez, and Klaus Schmidt-Hebbel. 1989. "Research Proposal: The Macroeconomics of the Public Sector Deficit" World Bank, Country Economnics Department, Washington, D.C. Elbadawi, I. A., and Klaus Schmidt-HebbeL 1991a. "Macroeconomic Struc- ture and Policy in Zimbabwe: An Analysis with an Empircal Model (1965- 1988)." World Bardn, Policy and Research Department, Washington, D.C. . 1991b. "Macroeconomic Adjustment in Zimbabwe: 1988-95 Simula- tions with a RMsM-XX Model." World Bank, Policy and Research Depart- ment, Washington, D.C. Fry, Maxwell J. 1988. Money, Interest and Banking in Economic Development. Baltimore, Md.: Johns Hopkins University Press. Giovannini, Alberto. 1985. "Saving and the Real Interest Rate in LDCS." Journl of Deveopment Economics 18 (August): 197-217. Hall, R. E. 1978. "Stochastic hnplications of the Life-Cyde Permanent Income Hypothesis: Theory and Evidence." Journal of Political Economy 86(6):-971- 87. Felipe Morande and Klaus Scd:midt-Hebbel 509 Haque, Nadeem UI, and Peter J. Montiel. 1989. "Consumption in Developina Countries: Tests for Liquidity Constraints and Finite Horizons." Review of Economics and Statistics 71(3): 408-15. Hayashi, Fumio. 1982. "The Permanent Income Hypothesis: Estimation and Testing by lnstrumental Variables." Joumal of Political Economy 90(5): 895- 916. -. 1985. "Tests for Liquidity Constraints: A Critical Survey." tNR Work- ing Paper 1729. National Bureau of Economic Research, Cambridge, Mass. Hubbard, R. G., and K. L. Judd. 1986. "Liquidity Constraints, Fiscal Policy, and Corsumption." Brokings Papers on Economic Activity 1:1-50. e. (International Monetary Fund). 1990. "Zimbabwe: Recent Economic Developments." Policy Framework Paper. Washington, D.C. Khadr, AUi, and Maus Schmidt-Hebbel. 1989a. "A Method for Macro- economic Consistency in Current and Constant Prices." Policy Research Working Paper 306. World Bank, Country Eoonomics Department, Wash- ington, D.C . 1989b. "A Framework for Macroeconomic Consistency for Zim- babwe." Policy Research Worldng Paper 310. World Bank, Country Eco- nomics Department, Washington, D.C. lhadrz AlB, L McKay, Klaus Schmidt-Hebbel, and J. Ventura. 1989. "A RMSM-X Model for Zimbabwe." World Bank, Country Economics Depart- ment, Washington, D.C. Kiguel, Miguel A., and Nissan Liviatan. 1989. "The Old and the New in Heterodox Stabilization Plans: Lessons from the 1960s and 1980s." Policy Research Working Paper 323. World Bank, Country Economiucs Depart- ment, Washington, D.C. eiderman, Leonardo, and Mario I. Blejer. 1988. "Modeling and Testing Ricardian Equivalence: A Survey." Intematirnal Monetany Fund Staff Papm 35 (1): 1-35. Marshall, Jorge, and IKaus Schinidt-Hebbel. 1989. "Economic and Policy Determinants of Public Sector Deficits." Policy Research Working Paper 321. World Bank, Country Economics Department, Washington, D.C. Mehlum, H., and Jom Rattso. 1993. "'Import Compression and Growth Restrictions in Zimbabwe." University of Trondheim, Norway, Depart- ment of Economics. Morand6, Felipe, and Klaus Schrnidt-Hebbel. 1991. 'Ma:croeconomics of Public Sector Deficits: The Case of Zimbabwe." Policy Research Working Paper 688. World Bank, Country Economics Department, Washington, D.C Reserve Bank of Zimbabwe. Various years. Quarterly Economic and Statistical Review. Harare. Rossi, Nicola. 1988. "Government Spending, the Real Interest Rate, and the Behavior of Liquidity-Constrained Consumers in Developing Countries." International Monetary Fund Staff Papers 35 (1): 10440 Sargent, Thomas J., and Neil Wallace. 1981. "Some Unpleasant Monetarist Arithmetic." Federal Reserze Bank of Minneapolis Quarterly Review 9: 15-3L Schmidt-Hebbel, Klaus. 1987. "Terms of Trade and the Current Account under Uncertainty." Andlisis Econ6mico 2 (1): 67-89. 510 Zimbabwe: Fiscad Dsequilibria and Lov Growth Schiidt-Hebbel, ICaus. 1990. "Zimbabwe: The Need for. Fiscal Adjust- ment." World Bank, Country Economics Department, Washington, D.C. Schmidt-Hebbel, Klaus, and Tobias Miller. 1992. '"rivate Investment under Macroeconomic Adjustment in Morocco." In Ajay Chlubbez Mansoor Dai- lami, and Nemat Shafik, eds., Reviving Private Investment in Developing Countries. Amsterdam: North-Holland. Schmidt-Hebbel, Klaus, Steven B. Webb, and Giancarlo Corsetti. 1992. "Household Saving in Developing Countries: First Cross-Country Evi- dence." World Bmak Economic Review 6 (3): 52947. Summers, Lawrence H. 1985. "Issues in National Savings Policy." NsER Workdng Paper 1710. National Bureau of Economic Research, Cambridge, Mas. Tanzi, Vito, Mario I. Blejer, and Mario 0. Teijeiro. 1987. "Infation and the Measurement of Fiscal Deficits." International MAonetary Fund Staff Papers 34 (December): 711-38. van Wijnbergen, Sweden. 1989. "External Debt, Inflation and the Public Sec- tor: Toward Fiscal Policy for Sustainable Growth." World Bank Economic Review 3 (3): 297-320. van Wijnbergen, Sweder. Robert Rocha, and Ritu Anand. 1988. "Inflation, External Debt, and Fmancial Sector Reform: A Quantitative Approach to Consistent Fiscal Policy." Policy Research Worling Paper 261. World Bank, Latin America and the Caribbean Country Department IL Washington, D.C. World Bank. 1990. Adjustment Lending Policies for Sustainable Grwwth. Policy and Research Series 14. Washington, D.C. - . Various years. World Development Report. New York. Oxford Unrversity Press. Zimbabwe, Central Statistical Office. Various issues. National Accounts- Harare. Zimbabwe, Minstry of Fmance. Various issues. Financial Statements. Harare. CD The Political Economy of Fiscal Defict Redudion Vito Tanz The project that resulted in the case studies presented in this volume has generated a wealth of interesting conclusions and analyses. As one who has been deeply interested in fiscal policy for many years, I can appreciate the importance of the results. The managers of the project and the authors of the papers should be congratulated on the outcome- This project is as close to a scientific experiment as, per- haps, is possible in economics. I hope that similar "experiments" will be carried out in the future by the World Bank on other important policy issues. In these remarks, I shall focus not on the technical aspects of the macroeconomics of public sector deficits-since these have been exhaustively studied in the case studies-but on the political and institutional reasons why promoting policies aimed at reducing large fiscal deficits has proved su difficult in many countries. These reasons deserve muc-h more attention from economists than they have received in the past. By now, many economists would agree that, especially in develop- ing countries, the public sector, far from being the "balancing factor"' advocated by Keynes, has often been an accomplice-if not the main culprit-in generating major macroeconomic imbalances. This reality must be kept in mind when the government is caled on to pursue "4stablizg"J fiscal poicies. Macroeconomic Imbalances and Government Response A question to be addressed first is the following: if, in fact, the gov- ernment has been part of the macroeconomic problem., what is the lielhood that it wi become part of the solution? Two views are possible: an optimistic and a pessimistic one. The optmistic view is that policymakers are fast learners and have enough political flex- ibility to be winig and able to change their minds and their policies when the situation requires these changes. For some countries-it is to be hoped that the number is increasing-this is a realistic view. These are the countries that manage to stay out of major economic 573 514 The Polcat Economy of Fisc Deict Reduc6on difficulties or that manage to make quick policy corrections when required. The pessimistic view is that policymakers are not fast leamers and that governments are not very flexible: policymakers do not learn from other countries' mistakes or even from the mistakes made by their predecessors. They learn mostly from their own mistakes, and they learn slowly and sometimes not very well. Furthermore, by the time they have learned from their own mistakes, they may have convinced themselves that the political situation does not allow them any mom for policy changes.' If the pessinistic view represents the reality of a significant number of countries, the policymakers of these countries face a credibility problem. This problem comes about because what may be considered mistaken polices today are often the policies on which the govern- ment ran in past elections or which it espoused when it took power. It is dfficult for governments to run away from or renege on their past policies and still be credible. Such behavior might bring into question the government's claim to be the one capable of running the country. For this reason govenmments tend to deny, for a long time and even to themselves, that they have made and are still making mistakes. They naturally tend to rationalze the situaion by blaming undesirable developments on the polices of their predecessors or, more often, on world or other exogenous developments. And, in fact, governments do often inheritbad situations. A recurrent theme is that what is happening to the economy is the result of unfavorable intemational economic developments: reces- sions in industrial countries; high intemational interest rates caused by large fiscal deficits in industi countries; marketing practices of industri countries that restrict access to those markets for particular products; and so on. Natural events are also frequently a convenient part of the rationalization. While these factors may contribute to eco- nomic difficulties and should thus not be ignored, they are rarely the whole or even the main explanation for poor economic performanceS2 Policymakers set on achieving the "social good" can escape the credibility problem by claiming that their previous choices were cor- rect but that new developments require a change of course. In this case, the policymakers can argue that the goals they are pursuing are the same as their earlier ones but that achieving these goals requires different policies. These policynakers will retain their credibility if they (a) explain dearly the reason for the change in policies, (b) are seen to be sincere, and (c) are seen to be competent and capable of choosing and implementing the appropriate policies. The extent to which a government can convincingly blame exteral events for the failure of its own economiic policy choices depends crucially on the existence of asymmetric information between the Vito Tarei 515 policymakers and private agents; the governrment's easier access to data and information and its greater expertise in interpreting them may allow it to mislead people by overstating the importance of adverse external events. Therefore, knowledgeable and independent irLsttutions with technical expertise and data availability comparable to the government's can play an independent role. The more devel- oped and democratic is the country, the more likely it is that such institutions exist, and thus the more difficult it wil be for the govern- ment to mislead the people about the reasons for its poor economic performance. Governments unwilling to take serious corrective actions occa- sionally resort to the use of jargon as a substitute for policies. For example, they may announce that they will "rationalize public expen- diture" or "improve tax adm rtioi" These "commitments" may occasionally be induded in formal agreements with international organizations. The inevitable question is: if public spending was in need of being "rationalized" (whatever that means), or if tax admin- istration needed to be improved, why wasn't this done earlier? Periods of crisis, when real wages may have been reduced and civil servants may be demoralized, are rarely the best time to make these changes. One is led to suspect that at times the promises may not reflect real, specific commitments. For example, after maldng strong statements about improving tax administration, the govermnent may not even bother to discuss this intention with the ranldng tax admin- istrators, who, as the ones who will implement the changes, wil determine whether tax administration really improves. Such an example relates again to the issue of observability of eco- nomic policy actions. In order to be acedible, a commiitment should be dearly defined and should promise an action or a result that can be readily observed so that the private sector can easily see when the government deviates from the promrised course of action. The gov- ernment will then know that eventual misconduct will be spotted and "punished' at the next elections or the next opinion polL Vague and general promises wil be discounted by the people because of the much greater difficulty in verifying whether they are carried out. A specific commitment is easier to monitor and is therefore more credible. Macroeconomic Policy, Credibility, and Biases in Official Pronouncements Biases in official pronouncements or forecasts have important impli- cations for eronomic policy. For example, in official statements about future growth, inflation, fscal deficits, and so on, the future often looks rosier tan the present, and the solution to current macro- 526 The Political Economy of Fiscal Deficit Reduction economic difficulties, regardless of how big they are, always seems to be just over the horizon. This problem affects developing countries in particular but is not limited to them. Consider, for example, the pro- jections about the federal fiscal deficit in the United States made throughout the 1980s: each year the federal budget was going to be balanced three or four years in the future. Another (admittedly extreme) example was a public speech in August 1989 in which the president of a Latin American country forecast that the rate of infla- tion for that country would be negative by the end of that year. In December 1989 the consumer price index rose at a monthly rate of 40 percent. In January 1990 it rose again, at a monthly rate of 79 percent. A survey of the official pronouncements or even the official forecasts of many countries would indicate that an optimistic bias is common. Furthermore, the size of the bias is generally not isignificant. Official optimism in economic forecasts is not without important policy consequences. A legitimate question is whether the policy- makers really believe in their own official pronouncements or make them for political reasons. If they believe in them, it is difficult to see how the same individuals who make the projections can pursue the policies necessary for remedying the macroeconomic imbalances- policies that are both correct and adequate to the situation. If they do not believe in their projections, that raises questions about the gov- ernment's credibility and reputation and about the expectations that individuals will form about the future. When forecasts are system- atically biased, either the policymakers are not competent or they are not sincere- The second interpretation appears more plausible: gov- emments try to gain popularity by making sanguine forecasts that underestimate the senousness of the problems and overestimate poli- cymakers' ability to deal with them. This behavior is bound to create significant credibility problems. Optimistic forecasts lead to insufficient policies and vice versa, and, of course, if people come to believe that the forecasts are biased, the government's credibility and individuals' faith in the effectiveness of its policies will be weakened.3 The result will be a magnified negative impact on the effectiveness of poliaes. Governments have to be con- sistent; they cannot forecast a rosier future and at the same time demand "blood, sweat, and tears" from their citizens.4 We find here a common cause of inadequate macroeconomic policies. Careful research on the connection between rosy forecasts and insufficient policies would be useful. I am not aware of any study that has specifi- cally analyzed this connection. Good economic policies, however, requiLre more than unbiased forecasts. They require competence, honesty, willingness to acknowl- edge past mistakes, and precise, down-to-earth explanations of the problems and the proposed solutions. It would be refreshing and Vito Tanzi 517 salutary if a president or prime minister were to face the Ctizens and state dearly: "Ladies and gentlemen, we goofed; we goofed badly. We take full blame, and we will take corrective action. Tne medicine will hurt a lot, but it is necessary. We wfll make sure tat the burden of adjustment is shared equitably. I will explain now, and I promise to explain in the future, exactly what we are going to do and why."5 People will understand that sacrifices are necessary if poicymakers can explain to them dearly and sinply why the sacrifices must be made and can convince them that the burden of adjustment wil be fairly shared.6 Popular support wfll be necessary to override the opposition of powerful groups that wil try to shift onto others the burden of the adjustment. The problem of credibility may persist even after such an expla- nation. A government that hi the past has not been sincere about the real macroeconomic situation and has made major mistakes will not have a good reputation. As we learn from Aesop's fable, the boy who cried wolf too many times when there was no wolf was not believed when he finally told the truth. Governments may learn from past mistakes, but if they are slow leamers, by the time they have learned they may have lost much of their credibility. As a consequence, policies that require credibility to be effective wil lose their effectiveness. It will thus be necessary to rely on altema- tive policies that require less credibility to be effective. These alte- niative policies may be less than optimal, raising the cost of the adjustment- For example, it may be necessary to shift up front the policy pack- age contained in the adjustment program rather than follow an opti- nmal sequencing path. It may also be necessay to push through some measures before they have been fully prepared. This may be inter- preted as a signal of firm commitment. When private agents are uncertain as to whether the government is really committed to the reform, the government can try to signal its intention by paying immediately a cost that only a govement with good intentions would be wiling to bear. As a consequence of these changes, the cost of adjustment will rise.7 The above discussion may explain why the ability of a new govern- ment to solve the country's problems is likely to decine with the passing of time, even though the government acquires more informa- tion and experience with time.8 Beging with the more costly mea- sures can be an effective signal that might increase credibflity, have a favorable effect on expectations, and thus raise the probability of success. But those policies must be seen as durable, and major techni- cal mistakes must be avoided. A govemment that waits too long to face the macroeconomic situation realistically wil have less chance of success when it finally decides to take action. 518 The Potitial Economy of Fisal Deficit Reduction It might be argued that a good reputation is not sufficient to generate credibility. At least two other conditions are required: (a) the policy proposed should be intemally consstent-that is, it should pursue nonconflicting goals with adequate instruments, and (b) to avoid the "time consistency problem," future incentives and constraints should be so designed that the government wi not find it in its interest to deviate from the preannounced course of action. Moreover, in particu- lar cases-as, for example, when a new government takes office-a good reputation may not even be a necessary condition for credibility. Because private agents have not yet had any experience with the gov- emmnent on which to base their assessments, the characteristics of the policy itself, or the way it is announced, wil be crucial. The Control Issue One reason why governments occasionally pursue economic policies that are unsustainable over the longer run is that in too many instances they come to power with little experience andwithunrealis- tic commitments. Often, this course of action leads to misguided policies or makes the new policymakers abstain from quicidy intro- ducing policies that would solve the existing problems. But let us assume that the government has credibflity or has somehow regained it. In this situation the problems encountered in the pursuit of sound macroeconomic policies may be of a different nature. Here I shall focus on the extent to which the policymakers who are responsible for economic policy control (a) the decisionmaking process and (b) the instruments of economic policy. The issue of control over the decsionmaking process and the instruments of economic policy has attracted little attention on the part of mainstream economists. Perhaps the formal or mathemaical approach that dominates modern maacoeconomics has distracted us from what is, admittedly, a messy area. Formal relationships that simply establish a stable and, often, single-valued relationship between a policy instrment and a policy objective ignore the issues I wish to discuss here.9 An example is an expression such as R = fl() that implies a direct and single-valued relationship between tax reve- nue, R, and a statutory tax rate, t. In this necessarily brief discussion the focus wfll be on fiscal instu- ments, but simlar issues may arise in connection with other instru- -ments. The discussion will be divided into two parts: control over policymaking and control over policy instruments. Control over Policmaking It is converient to start by focusing on the nerve centers where the basic decisions in macroeconomic policy are or should be made. In Vito Tanzi 519 most countries these are the office of the minister of finance or eco- nomics and the office of the governor of the central bank. In countries where the central bank is not independent, the former plays the leading role. The minister of finance and the governor of the central bank should have in mind the public interest (however defined) when they formu- late their economic policies.10 This, of course, raises questions about the individuals in charge-questions that relate to their economic sophistication, their biases, their honesty, their ability to withstand pressures from organized interest groups, their political ambitions (which may influence their short-mn actions), and so forth. In discussions of economic policy, economists take for granted that the poicymakers have the technical background to understand the finer points of economnics and the wisdom and independence to apply that sophistication to the solution of current economic problems. However, some of these individuals have no background, or only a limited one, in economics. Thus their ability to disciminate between good and bad economic advice may not be as great as is generally assumed.ll This helps explain why at times the people in charge of economic policy seem to pay undue attention to advice that good economists would consider obviously poor.12 The technical ability of the minister of finance and his immediate advisers can determine the extent to which technical errors are made. But assume, for the sake of argument, that the minister of finance is competent enough to choose good economic policy, or at least that he has able advisers who help him sort out the good policies-13 Fiscal decisions (budgetary cuts and, to a lesser extent, tax increases) must often be sold to the rest of the cabinet before they become official government policy. Here we face another common difficulty. While the minister of finance has (or at least should have) the public interest in mind, the other ministers wil usually have more parochial or par- ficular interests which it is, essentially, their job to promote or repre- sent. It is therefore natural for them to assess the proposed policies from their own angle: will the policies help the interests of the par- ticular groups they represent? For these ministers, fiscal retrench- ment is a negative-sum game. To the extent possible, each will try to shift the burden of deficit reduction onto some other part of the public sector, even when they agree that a reduction is desirable."4 The minister of defense wi be interested in protecting defense spending; the minister of planning, in protecting public investment; thoe minis- ter of education, in protecting educational expenditure; and so forth. A minister who is forced by his responsibilities to play Scrooge among ministers who want to play Santa Claus is not going to be very popular And unless he gets strong support from his superiors, he will not have much political power. At times his recommendations 520 The Political Economy of Fisca Deficit Reduction will be distorted and emasculated before they become government policy. At other times he will be voted down, or his budgetary instructions may be ignored. This is one reason why budgetary over- runs on the part of some ministries are so common.'5 In many coun- tries some of the other ministers are politically more powerful than the minister of finance and thus have the clout to neutralize or even ignore his decisions. (The same reasoning points to the importance of a politcally independent central bank.) To get leverage over other ministers, the minister of finance will have to sell his proposed policies to the people above him-the presi- dent or the prime minister; indeed, he will need their full support. But presidents and prine ministers, who are rarely economists, often take a political rather than an economic view of policymaking. They worry more about the next six months than about the medium run, and in the short run political and economic objectives may appear to conflict. Furthermore, the voice of the minister of finance will be just one among many. Other ministers will try to get the ear of their superiors in order to obtain backing for their own positions. Unlike decisions concerning the exchange rate or monetary devel- opments, which usually do not require legislative approval, many fiscal decisions must be approved by the legislature. A legislature may simply refuse to go along with the policies that the government proposes.16 Its members represent special interests or at least certain geographic areas, and they often see their role as protecting those interests, even though in the abstract they may share broad govern- mental objectives such as reduction of the fiscal deficit. When the proposed policies conflict with their particular interests, they wil try to prevent the policies from being enacted or wil try to water them down. This explains why it is so difficult to enact major tax reforms in countries with powerful and independent legislatures. Unfor- tunately, fiscal decisions are always decomposable into many subdivi- sions related to tax rates, kinds of tax, specific spending cuts, and so forth, and it is therefore impossible to avoid friction among groups. Even when there is no legislature to deal with, the difficulties out- lined above are still significant; they just take different forms. Control over Policy Instuments Up to now I have discussed problems encountered in the fonnuzlation of policy, which exist in all countries, industrial and developing. This section deals with issues of poliqr implementon that are particularly relevant to developing countries. Economists often suffer from what could be called the "rich country syndrome." To use an analogy, they generally assume that if an architect draws a sketch of a house, that house can be built to specifications and without particular problems Vito Tanzi 521 as long as the financing is there. In other words, they take for granted that all the required skills and materials exist and are available and that tie architectural directions will be followed faithfully. Applying the analogy to policy change, the assumption is that when an econo- mist sketches a proposed tax reform and the proposal is accepted by the government and becomes law, the actual or effective tax system (as distinct from the statutory one) will be reformed accordingly. However, this assumption is often unrealistic. Between conception of a policy-the decision to enact a particular policy-and its effective implementation there are many steps, and each is a potential trap that can prevent the policy from having its full effect or can even change its effect. The final outcome is often somewhat different from the original intention. As a consequence, it becomes difficult to establish a formal relationship between a change in the policy instrument (for example, in a tax rate) and a change in some objective (say, increased tax revenue).17 In the implementation stage of policy reform, policy decisions must often be decomposed and must then be carried out by lower-level government employees. For decisions to be implemented at the lower levels as decided at the top, at least three conditions must be met. First, the signals that the policymakers said downward must be very dear. The subordinates must have no doubts about exactly what the policymakers want. Making a decision is different from conveying that decision to those who have to implement it. When the signals sent are confused or conflicting or are seen as timid, they are not going to raise to action those in charge of maldng the policy opera- tional. The policymakers must put their full prestige and authority on the line when they convey the policy decisions. Second, the policymakers must have the power to force those immediately in charge of the various operational departments to push through the decisions. There have been cases in which the minister of finance, intent on reducing the fiscal deficit, was unable to get the cooperation of the director of taxation or the director of the budget and could not remove those individuals from their important jobs. Third, the incentive structure for those who must implement the policy decisions must be such that they do not have a strong interest in sabotaging the decisions. Powerful, entrenched bureaucracies often have enough power to determine the success or failure of the policies chosen by the political leaders. For example, freezes on wages may be circumvented by faster promotions, and reductions in spending may be defeated by accumulation of arrears. Tax reform provides examples of some of these problems. It is frustrating to see sinple tax proposals become distorted during the drafting stage to such an extent as to be unrecognizable. Once the laws are drafted and approved, extensive new regulations are 522 The Political Economy of Fiscal Deficit Reduction needed; here and dunang the approval process, further pitfalls can be encountered. Finally, the new lauws must be administered and enforced, and how they are implemented depends on the incentive structure of the tax administrators. A new law that cannot be admin- istered is worth not much more than the paper on which it is written. At times, the tax admirnstration will not have the information or the means to administer the new laws. For example, in some coun- tries laclc of gasoline may prevent tax inspectors from visiting tax- payers' prenmses. Often the tax reform wiUl change the power struc- ture within the tax administration, and the losers will try to ambush the changes. And of course, the more complex the tax laws become, the more widespread evasion or avoidance will be, occasionally with the help of corrupt tax inspectors (who may become even more cor- rupt if their wages are too low). In some instances, politically power- ful taxpayers will see to it that the new laws are not fully imple- mented, or the tax administration will be reluctant to go after powerful taxpayers. What the govemment really controls (with the cooperation of the legislature) is the statutory tax rates and some statutory definition of tax bases. However, the effective tax rates and the effective tax bases may change in ways that are not predictable.18 Thus an equation indicat- ing that tax revenue, R, is a function of the tax rate, t, may not be very meaningful since, depending on the circumstances, a given change in the rate might generate a whole range of revenue outcomes. As a result, forecasting the effect of changes in tax rates on tax revenue is very difficult. Similar problems occur with respect to policies aimed at changing public spending. Because of such problems, governments find it easier to change the look of the fiscal situation than its substance. Condusion I have tried here to call attention to issues that do not lend themselves to easy formal treatment but that nonetheless are very important in determining the success or failure of macroeconomic policy and that help explain why some countries appear to have major difficulties in reducing fiscal imbalances. These issues have not so far received the attention that they deserve, and they could benefit from more formal study. Notes This chapter is an edited and expanded version of remarks delivered at the roundtable session of the World Bank Conference on the Macroeconomics of the Public Sector Deficit, held June 20-21, 1991. The views expressed are strictly personal and are not official views of the International Monetary Vito Tanzi 523 Fund. Comments received from Marco Annunziata and Karim Nashashibi were much appreciated. 1. It is a common experience to hear policyrnakers admit that a situation requires policy changes (for example, a reduction in the size of the fiscal deficit) but that political realities would make it impossible for them to imple- ment such changes. In recent years discussions of policies have often been replaced by discussions about implementation of those policies. 2. International developments should be similar to random shocks: they should help in some periods and hurt in others. Policymakers, however, rarely attribute improvements in their country's macroeconomic situation to positive external developments. 3. For example, investors will not believe that these policies will improve the situation and thus will not invest; individuals who have taken their money out of the country will not repatriate it; and so forth. 4. Of course, they could promise a rosier future as a result of drastic policy changes. But this would imply thlat the forecast is conditional on those policies. 5. Interestingly enough, after the rosy inflation prediction mentioned above proved completely wrong, the president started considering drastic policy changes and talking about "surgery without anesthesia." As a conse- quence, the credibility of the government improved, and so did the economic situation. 6. Governments have a tendency to try to hide or minimnize the required sacrifices or, worse, to blame them on international institutions that require "demand adjustment." 7. This shift up front of the adjustnent effort will mean that less efficient but more quick! imnplemented policies will be preferred. It may also mean that temporary policies may be chosen over more permanent polices. Thus, a strong adjustment effort based on temporary (that is, nondurable) policies may not solve the credibility problem. (On this subject, see Tanzi 1989, 1990.) 8. It is often stated that the first one hundred days are crucial for a new govermnent. 9. That formal approach was pioneered by Jan Tmbergen for industrial countries and has been extended, without the necessary qualifications, to developing countries. Econometric models depend heavily on it. Even in industrial countries it is questionable whether the simple relationships between instruments and objectives ever exist. The recent literature on models of voting and lobbying is probably a first step in the right direction. However, the issues I raise here extend beyond issues of voting and lobbying. 10. Here is another crucial problem that raises serious doubt about the validity of many theoretical analyses of macroeconomic policy: how can we define the public interest? Society is made up of different groups of individ- uals with different and often conflicting interests. To reconcile these particu- lar interests within some definition of public interest (or social welfare func- tion), we need to make arbitrary value judgments about the relative importance of different groups. This is what the traditional approach to mac- roeconomiics often does, failing to analyze how the strategic interaction between various groups of individuals influences the decisionmaking process in economic policy. We must recognize that in particular circumstances what 524 The Political Economy of Fiscal Deicit Reduction we call poor policy may simply be a policy that has given excessive weight to the interests of some groups- Thus, in some sense, it may stil be a "rational" policy. 11. Of course, to the extent that the ministtr and the governor surround themselves with competent economists and rely on the advice of these advisers, their own lack of economic sophistication becomes less relevant. 12. It would be interesting to study whether the technical competence of policymakers has made a difference for macroeconomic adjustment. My impression is that it has. Of course, political considerations explain some of the choices made. 13. This is itself a strong assumption: economists of equivalent quality and reputation may disagree as to the appropriate policies in particular circumstances. 14. As a U.S. senator once put it: "Don't tax him, don't tax me, tax the fellow behind the tree." The argument presented here is, of course, highly relevant to legislators' reactions to proposed cuts. A current example is the way U.S. legislators who had argued over many years for cuts in defense spending reacted to proposed closings of military bases in their own districts ("When Defense Cuts Hit Home: Lawmakers Look for Creative Reasons to Keep Bases Open," Washington Post, March 22, 1993, p. 1). 15. The existence of effective institutions that permit the minister of finance to monitor most expenditures on a timely basis is particularly important These institutions are often nonexistent or highly inadequate. 16. The relationship between the government in power and the majority in parliament will play an important role. 17. The greater the number of policy changes taking place, the more com- plex the relationship becomes. 18. They also change as a consequence of the impact of macroeconomic policies on tax bases (see Tanzi 1991, ch.8). References Tartzi, Vito. 1989. "Fiscal Policy, Growth, and the Design of Stabilization Programs." In Mario I. Blejer and KCe-young Chu, eds., Fiscal Policy, Stabi- lization, and Growfh in Developing Countries. Washington, D.C.: Interna- tional Monetary Fund. -----. 1990. "Fscal Issues in Adjustment Programs." Rcerrhe Economiche 2- 3:173-94. . 1991. Public Finance in Developing Countries. Aldershot, U.K.: Edward Elgar, - Appendix and Indexes Statistical Appendix William Easterly and Klaus Schmidt-Hebbel This appendix presents annual data on public sector deficits, the monetary sector, and the financial sector for a large and varying sam- ple of member countries of tlhe Organization for Economic Coopera- tion and Development (oEcD) and developing countries for pedods of varying length, from 1965-90 to 1980-89. A brief description of the data methodology for each table follows. Public sector surplus or deficit (table A. 1). Coverage of the public sector in OECID countries is for the general government. Coverage for devel- oping countries may be for the central government, general gover- ment, or the consolidated nonfinancial public sector. Sources: For OECD countries, OECD, OECD Econmic Outlok (various issues). For developing countries, tvr, International Financia Statisics (various issues), ECLAC data, and World Bank data, except for the following: Brazil 1979-86, from Polak (1989); Colombia 1967-83, from Garcia and Guterman (1988); Dominican Republic 1970-79, from Greene and Roe (1989); Indonesia 1980-84, from Reisen and Trotsenburg (1988); Korea 1976-79, from Tanzi (1985); Korea 1980-84, from Reisen and Trotsen- burg (1988); Mexico 1979-86, from Polak (1989); Peru 1979-8, from Polak (1989). Seigniorage (table A.2). Annual seigniorage is calculated as the sum of monthly increases in the constant-price monetary base as a ratio to constant-price annual CDP. The monthly increase in constant-price monetary base is defined as the monthIly change in monetary base divided by the monthly consumer price index (any. Hence annual seigniorage (ST) is: S= (M, - Mt..Ic)PI,Jf GDPT where Mt is current-prce money base at the end of month t, acpi is the consumer price index in month t, and CDrT is constant-price gross domestic product in year T. Source: tma, International Financial Statis- tics (various issues). Inflation (table A.3). Inflation is defined as the annual percentage rate of change between annual average cps for the preceding and the 527 528 Statitsical Appendix current year. Source: imF, International Financial Statistics (various issues). Real interest rate (table A.4). The annualized real interest rate (r,) is calculated as the geometric mean of ex-post monthly real interest rates and is defined as: = '( + ,)1(1. + rt)] - 1 where i4 is the annualzd monthly nominal interest rate and rt is the annualized monthly CPI inflation rate. The nominal interest rate is the average deposit (or passive) rate in the banking system for maturities that vary between 30 days and 18 months. Sources: For OECD cOUIn- tries, IMF, International Financial Statistics (various issues), except Tur- key 1966-85, from World Bank data. For developing countries, sources are as follows: for Argentina, Brazil, Chile, COte d'lvoire, Ecuador, India, Mexico, Pakistan, Peru, Zaire, and Zambia for the years 1965484, World Bank data; forjamaica, Malaysia, Morocco, Phi- lippines, Sri Lanka, and Venezuela for the years 1965-85, for Nigeria for the years 1966-84, for Korea, Malawi, Sierra Leone, and Thailand for the years 1966-85, for Indonesia for the years 1969-84, and for Colombia for the years 197048, World Bank data; for Argentina 1985- 88, Brazil 1985-86, Chile 1987-87, Indonesia 1985-89, and Morocco 1986-88, Easterly (1989); for Zaire 1985-87 and India 1985, de Melo (1988); for Zimbabwe 1970-88, Centrl Barn of Zimbabwe; for Ghana 19674-8, Mexico 1985-89, and Zimbabwe 197948, country case studies listed in the references to chapter 1; for all other figures, IMF data. Taxes from financial repression in ten countries (table A.S). This table presents estimates from various sources for the implicit tax revenue frm financial repression, using methodologies specified in the tables. Sources: oECD data and nAF, Intemational Financial Statistics (various issues). References Chaney, Christophe, and Patrick Honohan. 1990. "Taxation of Financial Intermediation." Policy Research Working Paper 421. World Bank, Courn- try Economics Department, Washington, D.C. De Melo, Jaime. 1988. "Computable General Equihbrium Models for Trade Policy Analysis in Developing Countries: A Survey." World Bank, Country Economics Department, Washington, D.C. Easterly, William. 1989. "Fiscal Adjustment and Deficit Financing during the Debt Crisis." In Ishrat Husain and Ishac Diwan, eds., Dealing with the Debt Crisis. A World Bank Symposium. Washington, D.C. Garca, Jorge, and Lia Guterman. 1988. "Medicid6n del Daficit del Sector Pdblico Colombiano y su Financiaci6n 1950-1986." Ensayos de Politica Eco- n6mica 14 (December). William Easterly and Klaus Schamidt-Hebbe 529 Green, Duty D., and Terry L Roe. 1989. Trade, Exchange Rate, and Agrcultural Priing Policies in thze Dominican Republic. World Bank Comparative Studies. Washington, D.C. IMP (International Monetary Fund). Various issues. International Financial Sta- tistics. Washington, D.C. OECD (Organization for Economic Cooperation and Development). Various issues. OECD Economic Outlook. Paris. Polak, Jacques J. 1989. Financial Policies and Development. Development Centre Studies. Organization for Economic Cooperation and Development, Paris. Reisen, Helmut, and Axel van Trotsenburg. 1988. Developing Country Debt: The Budgetary and Transfer Problem. Development Centre Studies. Organiza- tion for Economic Gooperation and Development, Paris. Tanzi, Vito. 1985. "Fiscal Management and Extemal Debt Problems." In Hassanali Mehran, ed., Extermal Debt Management. Washington D C: Inter- national Monetary Fund. Table A.1. Consolidated Public Sector Surplus (+) or Deficit (-), 1970-90 (percentage of CDP, except as specified) Countlry 1970 . 1971 2972 1973 1974 1975 1976 1977 1978 1979 1980 1981 OECD coauilres Australia 2.8 2.3. 2.1 -0.2 2.3 -0.6 -2.9 -0.8 -2.8 -2.6 -1,9 -1.2 Austria 1.2 1.5 2.0 1.3 1.3 -2.5 -3.7 -2.4 -2.8 -2.4 -1,7 -1.E Belgium -2.1 -3.0 -4.2 -3.8 -2,9 -5.3 -6.0 -6.3 -6.7 -7.5 -9,2 -13.1 Canada 0.8 0.0 0.0 0.9 1.9 -2.5 -1.8 -2.5 -3.1 -2.0 -2,8 -1.5 Denmark 3.2 3.9 3.9 5.2 3.1 -1.4 -0,3 -0,6 -0.4 -1.7 -3,3 -6.9 Finland 4.3 4.5 3.9 5.7 4.6 2.7 4.9 3.2 1.4 0.4 0,3 1.2 France 1i1 0.8 0.8 0.8 0.1 -2.2 -0.6 -0.8 -2.1 -0.8 0.0 -1.9 Germany, Fed. Rep, 0.2 -0.2 -0.5 1.2 -1.3 -5.6 -3.4 -2.4 -2.4 -2.6 -2,9 -3.7 Greece -0.1 -0.9 -0,3 -1.4 -2.2 -3.4 -2.6 -2.1 -1.7 -2.5 -2.9 -10.9 Ireland -3.6 -3.5 -3,2 -3.8 -6,9 -11.1 -7.4 -6.6 -8.6 -11.0 -12.2 -13.3 Italy -4.0 -5.9 -8.6 -7.9 -7.8 -12.9 -9.8 -8.6 -10.4 -10,2 -8.6 -11.6 tn Japana 1.7 1.2 -0,1 0.5 0.4 -2.8 -3.7 -3.8 -5.5 -4.7 -4.4 -3.8 Netherlands -1.1 -1.0 -0.4 0.6 -0,5 -3.0 -2.7 -1.8 -2.8 -3.7 -4.1 -5.5 Norway 3,2 4.3 4,5 5.7 4.7 3.3 2.5 1,2 -0.1 1.3 5.7 4.7 Spain 0,0 -1,0 -0.1 0.8 -0.4 -0.5 -1.1 -1,4 -2.4 -2.2 -2,6 -3.9 Sweden 4,6 5,3 4.4 4,1 2.0 2.8 4.7 1.7 -0.5 -3.0 -4.0 -5.3 United Kingdom 2.9 1.3 -1.3 -2.7 -3.9 -4.6 -5.0 -3.4 -4.4 -3.3 -3.4 -2.6 United States -1.1 -1,8 -0,3 0.5 -0.3 -4.1 -2.2 -1.0 0.0 0.5 -1.3 -1.0 Average OECD 0,8 0.4 0.1 0.4 -0.3 -3.0 -2.3 -2,1 -3.1 -3.2 -3.3 -4.6 Developing countries Argentina -1.9 -4,4 -5.8 -7.5 8.1 -15.1 -11.7 -5.1 -6.8 -6.1 -7.5 -13.3 Bangladesh - - - - - - - - - - -14.5 -11.8 Bolivia - - - - -8.4 -9,0 -7.8 Brazil - - - - - - - - - - -1.2 -3.4 Bulgaria - - - - - - - - - . - Burkina Paso - - - - - - - - - Chile -2.1 -7.5 -8.1 -19.0 -5.5 -2.1 4.0 0.4 1.4 4,6 5.4 0.4 Colombia -6.9 -7.4 -6,5 -7.1 -0.9 0.9 -1.9 -2.7 -1,2 -4,0 -5.8 -6.8 C6Ie d'lvoire -3.4 -4.5 -1.3 0.0 -0.9 -2.3 -12.4 -3.6 -8.4 -10.3 -12.2 -11.8 Dominica - - - - - - - - - Dominican Rep. -0.3 0.8 1.4 -0.4 -2.6 1.9 -0.7 -1.3 -5.4 -5.1 -6.5 -5.8 Ecuador -4.4 -4.3 -1.9 3.1 0.8 -2.2 -3.3 -8.3 -6.2 -2.0 -4.6 -5.6 Ghana 1.3 -2.7 -3.5 -3.2 -5. -13.2 -9.2 -7.5 -7.1 -4.0 -6.0 -7.4 Honduras -8.5 -8.7 -8,5 -6.1 -6,9 -8.3 -8.3 -5.6 -6.5 -7.5 -8.5 -8.5 India -3.4 -3.7 -4.6 -2.9 -3.5 -4.6 -5.6 -5.4 -3.7 -5.0 -6.4 -7.3 Indonesia -3.0 -2,5 -2.4 -2.3 -1.5 -3.5 -4.2 1.0 0.4 1.3 2.6 0.1 Jamaica -2.5 -3,7 -4,1 -5.3 -7,8 -7.9 -15.5 -14.5 -16,7 -15.3 -20.8 -16.6 Jordan - - - - - - - - - - - - Kenya 0.0 0.0 -3.8 -5.1 -2.8 -4.8 -5.9 -4.1 -3.8 -7.0 -6.3 -9.9 Korea, Rep. of -0.8 -0.3 -3.8 . -0.5 -2.2 -2.0 -1.4 -1.7 -1.3 -1.4 -3.2 -4.6 Malawi - -6.5 -7.7 -6.7. -4.8 -6.7 -5.5 -3.6 -11.6 -13.6 -16.4 -12.1 Malaysia - - -9.7 -5.9 -5,6 -8.3 -7.1 -9.3 -6.6 -32 -11.9 -17,0 Mexico -3.5 -2.2 -4.4 -5.4 -5.7 -8.4 .-7.2 -4.9 -5.1 -6.7 -8.3 -11.6 Morocco -3.0 -3.0 -4.0 -2,0 -3.9 -9.5 -18.1 -15.8 -11.3 -10.1 -9.0 -13.6 Nigeria - - -0,7 -1,5 6.4 -6.5 -6.5 -1,8 -0.9 -1.1 -3.6 -9.1 Omanm- - - - - - - - - - Pakistan - - -3.7 -5.2 -9.3 -8.8 -8.3 -7.4 -8.3 -5.8 -4.8 -4.9 Paraguay -1.1 -2.2 -2.9 -0.6 0.2 -5.0 -3.3 -0.8 0.0 1.2 0.1 -2.9 Peru - -1.6 -0.9 -2.9 -1.9 -2.7 -3.7 -2.8 -4.5 -1.5 -5.4 -6.7 Philippines - - - - - - - - - - -7.5 Poland - - - - - - - - - -1.9 -10.7 -2.5 Sierra Leone - - -4.4 -10.5 -7.9 -6.9 -9.3 -11.6 -12.5 -9.3 Sri Lanka -6.2 -7.1 - -5.0 -3.1 -6,3 -8.2 -4.4 -12.1 -11.8 -18,3 -12.4 Ihailand -5.2 -5.4 -5.3 -3.1 0.7 -2.2 -5.0 -4.5 -5.3 -5.3 -6,3 -7.1 Trinidad and Tobago - - - - - - - - - - - Turkey -3.5 -4.9 -3.1 -2.6 -2.6 -1.9 -2,8 -8.7 -10.6 -9.0 -11.9 -6.1 Venezuela -1.2 0.4 -0,2 1.3 3.9 1.4 -2.5 -1.6 -3.3 3.8 4.4 3.6 Zaire -2.2 -7.3 -7.0 -9.6 -1.8 -11.3 -21.9 -11.4 -10.9 -5.2 -1.9 -7.2 Zambia - -13.1 -16,8 3.4 -21.7 -14.0 -14.4 -10.3 -12.2 -16.4 -15.6 Zimbabwe -7.1 -6.5 -6.9 -8,1 -4,3 -7.8 -4.1 -4.8 -3.2 -1.3 -9.1 -13.5 (Table contitnes on thefollowing page wiltl theyear 1982.) Table A.1 (con titnied) _ Average, (1970-90,. Conlitry 1982 1983 1984 1985 1986 1987 1988 1989 1990 except as specified) OECD countries Australia -1.0 -4.5 -4.0 -3.3 -3.0 -1.1 0.7 1.2 2.3 -0.8 Austria -3.4 -4.0 -2.6 -2.5 -3.7 -4.3 -3,1 -2.7 -1.1 -1.8 Belgium -11.2 -11.4 -9.3 -8.7 -8.8 -7.2 -6.8 -6.5 -6.1 -7.0 Canada -5.9 -6.9 -6.5 -6.8 -5.5 -4.4 -2.6 -3.4 -3.0 -2.7 Denmark -9.1 -7.2 -4.1 -2.0 3.4 2,5 0.3 -0,4 -0.5 -0.6 Finland -0.6 -1.7 0.4 0.1 0.8 -1,2 1.4 2.7 2.5 2.0 France -2.8 -3.1 -2.8 -2.'3 -2.7 -1.9 -1.8 -1.4 -1,2 -1.2 Germany, Fed. Rep. -3.3 -2.5 -1.9 -1.1 -1.3 -1.8 -2.1 0.2 -0.8 -1.8 Greece -7.6 -8.6 -10.2 -14.0 -12.7 -12.0 -14,5 -17.8 -17.2 -6.9 Ireland -14.1 -12.0 -10.1 -11.8 -11.6 -9.2 -2,6 -2.8 -1.1 -7.9 Italy -11.3 -10.7 -11.6 -12.5 -11.7 -11.1 -10.9 -10.2 -10.2 -9.8 uj Japana -3.6 -3.7 -2.1 -0.8 -0.9 0.7 2.1 2.7 3.1 -1.3 r% Netherlands -7.1 -6.4 -6.3 -4.8 -6.0 -6.5 -5.0 -5.1 -5.1 -3.7 Norway 4.4 4.2 7.5 10.4 5.9 4.8 3.1 1.0 1.2 4,0 Spain -5,6 -4.8 -5.5 -7.0 -6.1 -3.2 -3.1 -2.1 -2.0 -2.6 Sweden -7.0 -5.0 -2.9 -3.9 -1.3 4,2 3.4 5.3 4.6 0.7 United Kingdom -2.4 -3.3 -3.9 -2.7 -2.4 -1.2 1.1 1.3 0.7 -2.1 United States -3.5 -3,8 -2.8 -3.3 -3.4 -2.4 -2.0 -2.0 -1.3 -1.7 Average OECD -5.3 -5.3 -4.4 -4.3 -3.9 -3.1 -2.4 -2.2 -2.0 -2.5 Developinig cotritiies Argentina -15.1 -15.2 -11,9 -6.0 -4.7 -5.5 -7.4 -0.5 1.6 -7.5 Bangladesh -13.6 -13.0 -10.4 -10.5 -9.9 -8.4 -7.1 -7.2 -8.1 -10.4 (1980-90) Bolivia -14.7 -19.1 -27.4 -9.1 -3.4 -7.8 -.6.7 -4.3 -3.3 -9.4 (1979-90) Brazil -4.8 -7.0 -3.5 -2.7 -4.3 -3.7 -5.5 -4.3 1.2 -4.1 (1980-90) Bulgaria - - - -0.8 -2.6 1,2 -0.9 -0.6 - -0.7 (1985-89) Burkina Faso - - -6.2 -3.1 -6.9 -5.5 -6,1 4.1 - -4,0 (1984-89) Chile -3.9 -3.5 -4.6 -2.9 -2.0 -0,2 3.6 3.8 0.5 -1.8 Colombia -8.9 -8.5 -6.3 -3.5 -0.3 -2.0 -2.1 -1.8 -2,2 -4.1 C4te d'lvaire -15.9 -11.4 -1.7 2.0 -2.4 -7.3 -13.5 -14.4 _ -6.8 (1970-89) Dominica - - -3,9 1.4 3.2 3.5 2.3 -8.8 - -0.4 (1984-89) Dominican Rep. -7.0 -5.3 -6.7 -0.9 -5.6 -3.8 -5.6 - - -3.1 (1970-88) Ecuador -6.7 0,0 -0.6 1.9 -5.1 -9.6 -5.1 -1.6 0.9 -3.1 Ghana -4.1 -2.3 -0.5 -1.9 2.2 3.4 2.2 - - -3.6 (1970-88) Honduras -12.5 -12.7 .-11.4 -8.2 -6,3 -6.1 -6.7 -7,2 -6.3 -8.1 India -7.0 -7.5 -7.7 -9.1 -8.6 -8.6 -8.9 -9.3 -7.0 -6,2 Indonesia -4.1 -2,6 0.9 -2,9 -4.8 -2.0 -0.7 -2.7 -0.5 -1.6 Jamaica -15.3 -13.6 -15.1 -13,2 -5.6 -5.4 -13.4 -4.3 -3,5 -105 Jordan - -4.7 -8.5 -8.7 -4.7 -14.3 -15,3 -6.6 -14.2 -9.6 (1983-90) Kenya -6.9 -3.9 -4.4 -4.9 -5.3 -6.6 -4.3 -4.7 -5.5 -4.8 Korea, Rep. of -4.3 -1.6 -1.4 -1.2 -0.1 0.9 -1.6 0.2 -0.7 -1.4 Malawi -8.7 -7.0 -4.9 -9.6 -6,6 1.2 -0.4 -1.8 - -7.0 (1971-89) Malaysia -17.9 -14.8 -11.1 -5.9 -10.3 -5.6 -2.7 -4.2 -6.1 -8,6 (1972-90) Mexico -15.4 -6.0 -2.9 -3.3 -7.0 -5.4 -10.7 -8.9 -3.9 -6.4 Morocco -9.2 -11.5 -8.1 -8.6 -5.7 -6.1 -5.5 - - -8.3 (1970-88) Nigeria -8.4 -10.1 -4.2 -1.8 -2.8 -8.7 -10.8 -8.1 -6.1 -4.4 (1972-90) W Oman - -0.2 -3.2 -2.1 -16.8 1.9 -7.4 0.2 - -3.9 (1983-B9) Pakistan -6.4 -5.5 -7.1 -7.1 -8.3 -7. -7.8 -6.4 -6.1 -6.8 (1972-90) Paraguay -2.7 -5.3 -6.4 -2.6 -1.8 -2.9 -3.1 0.6 1.2 -2.1 Peru -7.2 -9.5 -6.8 -3.7 -6.2 -7.5 -8.1 -13.6 -8.1 -5.3 (1972-90) Philippines -6,5 -4.6 -3.3 -2.1 -4,8 -2,6 -3.0 -4.3 -i'S -4.3 (1981-89) Poland -0.9 -0.5 0.0 -0.3 -0.8 -0.9 -9.0 - - -2.8 (1980-89) Sierra Leone -10.5 -13.7 -7.1 -10,9 -10.8 -3.8 -4.0 -5.9 -2.9 -8.4 (1974b90) Sri Lanka -14.0 -10.6 -6,8 -9.7 -10.1 -8.7 -12.6 -10.8 -7.7 -9.3 Thailand -7.6 -5.7 -7,3 -8,6 -6.6 -3.2 0,2 0.8 1.2 -4.3 Trinidad and Tobago - - - - -7.4 -10,9 -7.4 -6 4 -4.8 -7.4 (1985-89) Turkey -6.1 -6.6 -9,2 -5,6 -4.3 -4.4 -3.9 -3.8 -3.7 -5.3 Venezuela -5,6 0.4 8.6 5.0 -7.5 -5.4 -9.4 -1.3 0.2 0.6 Zaire -10.5 -4.1 -0.7 1.2 -1.7 -3.1 -4.6 1.2 - -6,0 (1970-89) Zambia -16.8 -8.9 -9.3 -19.8 -28.5 -10.8 -12.1 -9.4 -8.2 -13.4 (1972-90) Zimbabwe -13.1 -14.4 -12.7 -14.4 -14.4 -10.9 -10.4 -9.9 -9,8 -8.9 - Not available. a. Percentage of GNP. Table A.2. Seigniorage, 1965-89 (percentage of GDP) Couniry 1965 1966 1967 1968 1969 1970. 1971 1972 1973 1974 1975 1976 1977 OECD countries Austria 0.3 1.0 1.2 0.2 0.5 0.7 1.4 1.7 0.2 1.0 1.8 0.5 0,5 Belgium 1.1 0.5 0.3 0.4 0.0 0.4 1.0 1.s 1.6 0.5 0.7 0.7 0.9 Canada 0.6 0.5 0.3 0.5 0.3 0.3 0.8 0.9 0.9 0.7 0.8 0.5 0.6 Denmark - - 0.4 1.7 0.4 -0.6 0.2 0.2 0.1 0.1 1.6 0.2 -0.6 Finland 0.0 0.4 0,4 0.3 -0.1 0.1 0.3 0.4 0.3 0.4 04 0.0 0.2 France 0.9 0.6 1.0 0.9 -0.1 0.5 1.0 3.0 0.8 1.4 -2.3 0.5 0.6 Germany, Fed. Rep. 0.6 0.8 -0.5 0.9 0.3 1.8 1.4 2.5 0.8 -0.1 0.3 1.0 0.8 ,, Greece 1.5 2.6 1.8 1.9 0.6 1.0 2.9 3,0 2.8 3.1 3.1 2,9 3.2 t Italy 2.2 1.6 2.0 1.4 2.0 1.7 3.0 3.1 5.2 3.1 3.1 2.9 2.4 Japan 0.7 0.9 1.3 1.3 1,3 1.1 1.1 2.2 2,6 1 3 0.3 0.7 0.7 Netherlands i.l 0.8 0.3 0.2 0.6 0.5 0.4 0.7 0.4 0.5 0.8 0.6 0.6 Norway 0.9 0.8 0,4 0.7 0.7 0.3 1.0 0.4 0.5 1.1 1.3 1.0 0.9 Spain 1.5 1.4 1,2 1.1 1.2 1.3 2.6 1.8 2.1 2.3 1.9 1.5 2.0 Sweden 0.3 0.4 0.5 0.5 0.5 0.2 0.8 0.6 0.6 2.3 -0.1 0.7 0.6 Turkey - - - - 2.0 2.3 3.4 4,1 3.2 3.1 3.5 2.5 5.0 United States 0,5 0.5 0.4 0.6 0.3 0.5 0,8 0,1 0.7 0.5 0.4 0.4 0.4 Developiig countries Argentina 2.6 4.1 0,4 3.5 2.7 1.9 5.6 2,9 5,0 6.5 13,2 4,9 1.1 Bangladesh - - - - - - - 0.0 0.1 1,4 Bolivia 1.7 2.3 0,5 0.6 1.1 1.2 1.9 1.6 2.9 2.0 1.4 3.5 2,4 Brazil 4.7 2.1 1.9 3.1 2.0 1.4 2.3 0.5 2.2 1.9 1.8 2.0 2.1 Burkina Faso 0.3 0.4 0.0 0.3 1.1 0.2 -0.1 0.5 1.1 1.1 2.1 1.0 1.3 Chile - - - - 3.3 10l5 - Colombia - - - - 1.4 0.9 1.7 2.2 1.5 1.7 2.7 2.9 C8te d'lvoire 0.2 1.4 0.8 0.7 1.2 2.0 1.2 1.4 1.0 3.3 1.4 1.8 3.3 Dominican Rep. 2.7 -0.9 0.0 1.8 1.7 1.3 1.6 0,9 2.4 4,5 -0.9 -0.1 2.3 Ecuador 0.3 1.1 0.5 1.8 1.7 3.1 1.2 3.1 4.1 3.9 -0.1 2,0 - Ghana - - - 0.3 1.6 1.3 0.2 4.1 2.4 3.3 5.0 5.6 6.7. Honduras 0.7 0.2 0.5 0.7 1.0 0.8 0.6 0.2 1.3 -0.2 0.8 2.3 1.2 India 1.0 0.8 0.7 0.6 1.2 1.1 1.2 1.0 2.2 0.5 0.5 1.6 2.1 Indonesia - - - - - 1.3 1.4 3.1 2.3 2.9 2.3 1.6 1.8 Jamaica - - - - - 1.0 1.7 0.8 1.6 1.2 1.2 1.1 1.3 Jordan - - - - - - - - - - - 8.0 6.2 Kenya - - - 1.7 3.2 1.1 -0.9 0.8 1.0 1.9 -0.9 1.3 3.6 Korea, Rep. of - - - - - 3.1 -0.4 3.3 3.6 1.9 2.8 2.5 3.4 Malawi - - - - - - - - - - - - - Malaysia 0.8 1.1 -1.4 0.6 1.6 1.0 0.8 3.1 3.5 1.6 0,3 2.0 1.7 Mexico 0.3 0.6 0.8 1.0 0.7 0.4 0.8 4.2 2.4 3.2 3.2 -0.8 8.5 Morocco 1.5 -0.1 1.5 1.6 1.6 1.0 1.3 2.3 1.9 2.3 1.9 .2.8 2.0 Nigeria 0.1 0.6 -0.4 -0.7 1.8 2.0 0.3 0,4 0.9 4.4 3.6 2.2 1.9 Pakistan 2.3 2.3 0,1 0.9 1.7 2.0 2,1 3.4 2.1 0.0 1.2 2.3 2.8 Paraguay 1,1 0.3 3.1 -0.9 1.1 1,7 1.3 2.2 2.6 1.5 - 2.1 3.1 Peru 1.4 0.5 0.9 0.5 0.7 3.3 0.5 2.7 . 0.6 3,2 0.5 2.9 1.6 Philippines - - - 1.0 0.5 0.6 1.6 0.6 1.2 0.7 0.6 1.2 Sri Lanka 0.9 -0.8 1.5 1.0 0.2 0.3 1.4 1.8 1.7 0.7 -0.6 1.9 3.5 Thailand 1.2 1.3 0.8 0.9 0.6 0.9 1.1 1.7 1.5 1.2 1.0 1.1 0.8 Trinidad and Tobago -0.4 0.9 0.0 1.5 -0.3 0.6 1.6 0.6 0.8 3.4 3.2 2.7 0.2 Venezuela - - 0.7 0.9 0.4 0.2 1.5 0.9 1.7 2.4 3.3 2.0 2.3 Zaire - - - 4.4 0.8 2.0 -1.7 2,8 1.7 2.3 4 8 8.6 4.2 Zambia - - 1.1 2.3 3.7 -4.1 1.2 2.0 -0.5 3.4 2.3 -0.2 Zimbabwe - - - - - - - - - - - - (Table continues on titefollowing page withl Ike year 1978.) Table A.2 (contfined) Counlry 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 OECD countries Austria 1.6 0.2 0.7 0.7 0,4 0.7 0.2 0.1 0.7 -0.3 0.1 1.2 Belgium 0.8 0.3 0.1 0.2 0.0 0.3 0.1 -0.1 0.4 0.2 0.1 0.2 Canada 0.6 0.4 0.5 0.1 0.2 0.1 0.1 0.2 0.3 0.3 0.2 0.1 Denmark 0.3 0.5 0.1 0.3 0.1 0.2 0.3 4.6 -2.4 0.7 0.2 -0.2 Finland 0.5 1.4 1.1 -0.1 0.4 0.9 1.5 0.6 -0.1 2.2 1.2 1.5 France 0.8 0A4 0,9 0.3 1.0 0.3 0.6 1.0 0.1 0.7 -0.1 -0.0 Germany, Fed. Rep. 1.3 0.5 -0.4 -0.2 0.5 0.5 0.3 0.3 0.6 0.8 1.0 0.6 Greece 3.0 1.5 3.5 6.4 4.6 1.6 4.8 1.1 2.8 3.7 - - Italy 3.4 1.7 1.7 1.6 2.0 2.1 1.9 2.4 1.1 1.2 1.3 1.8 Cal Japan . 1.2 0.6 0.5 0.2 0.5 0.5 0.7 0.3 0.7 0.8 1.2 1.3 uf. Netherlands 0.5 0.5 0.5 0.1 0.5 0.8 0.4 0.3 0.3 0.7 0.9 1.3 Norway 1.0 0.7 0.6 -0.1 0.4 0.2 0.9 0.4 -0.1 0.4 -0.1 0.3 Spain 2.4 3.6 1.8 1,5 2.5 11.7 0.7 0,8 1.5 4.9 1.4 3.1 Sweden 0.9 2.4 -0.9 0.8 0.5 0.1 0.3 0.1 1A 0.3 1.1 0.7 Turkey 4.6 4.9 3.4 3.9 3.2 3.1 3.6 2.9 1.9 2.7 4.2 - United States 0.7 0.4 0.3 0.2 0.3 0.3 0.3 0.5 0.8 0.3 0.4 0.2 Deeloping countiies Argentina 3.3 1.4 6.5 4.4 7.6 6.2 5.8 1.5 3.2 4.9 - - Bangladesh 1.2 0.7 0.9 0.5 0.5 1.7 1.9 0.8 0.5 2.6 0.8 0.9 Bolivia 1.2 1.0 3.2 1.4 10.4 8.3 9.9 - - - - Brazil 1.8 2.9 1.7 1.7 1.8 1.8 2.4 2.2 - Burkina Faso -0.3 1.9 0.6 1.5 0.7 3.5 4.0 -1.2 4.4 Chile - 2.4 2.2 -0.6 -1.8 0.5 0.9 - - - Colombia 4.4 2.6 2.6 2.1 1.5 - 1.8 - - 1.2 Cote d'lvoire 3.1 0.4 0.1 0.9 -0.6 0.7 1.8 2.5 0.5 - - - Dominican Rep. 08 0.4 -0.4 1.9 0.2 1.4 2.1 -0.2 6.5 -0.6 7.2 2.6 Ecuador 1.5 2.6 1.7 0.6 0.6 1.5 2.2 1.3 1.9 2.2 2.9 2.1 Ghana 7.8 2.4 2.9 4.2 1.2 2.9 2.8 2.2 3.3 2.5 3.4 Honduras 1.4 1.0 0.1 0.4 - - 0.0 0.4 0.9 0.6 1.0 1.9 India 2.1 2.4 1.7 1.2 2.0 1.4 1.9 2.7 2.0 2.4 2.1 - Indonesia 0.7 1.8 2.0 0.9 0.3 1.3 0,6 1.1 1.4 0.7 -0.4 1.4 Jamaica 0.6 1.3 3.0 -0.5 -0.9 4.3 7.3 4.3 2.6 2.8 - 3.6 Jordan 5.8 8.9 7.3 4.6 3.9 3.9 0.7 1.4 2.3 1.6 8.4 6.7 Kenya 0.6 1.8 0.1 0.3 1.6 -0.2 0.7 1.0 2.4 1.5 0.1 1.2 Korea, Rep. of 2.9 2.0 -1.1 -1.0 1.8 0.4 0.2 0.1 0.7 2.2 1.7 2.1 Malawi - - 1.5 1.7 1.2 0.7 4.1 0.4 5.3 5.6 1.1 -1.2 Malaysia 1.7 1.6 1.9 1.2 1.9 0.5 0.4 0.9 0.6 0.7 1.3 2.8 Mexico 3.5 4.2 4.7 5.4 11.0 6.2 5.6 1.5 3.0 2.7 - - Morocco 2.1 2.3 0.7 17 0.9 1.7 1.5 1.0 2.7 0.7 2.0 2.5 Nigeria -0.4 1.3 5.1 -0.6 0.8 0.2 0,3 0.6 0.5 1.3 2.7 2.6 8 ; Pakistan 2.2 3.7 2.4 1.2 2.3 1.9 2.2 1.2 2.7 2.9 1.7 2.5 Nl Paraguay 3.6 2.5 2.7 1.8 0.1 3.2 1.9 1.8 2.8 3.4 1.9 - Peru 3.0 5.3 6,3 3.5 3.6 6.0 7.3 10.3 3.4 5.5 9.6 6.9 Philippines 1.3 1.2 0.6 0.5 0.3 - - - 1.0 1.1 2.7 Sri Lanka 1.2 1.9 1.7 1.4 1.9 2.1 1.3 2.3 0.3 1.0 2.4 .0.6 Thailand 1.3 1.2 1.1 0.5 0,9 0.8 0.4 0.7 0.8 1.7 1.1 1.2 lTinidad and Tobago 0.4 4.6 0.6 1.8 5.2 -0.8 -1.1 0.9 -2.4 -2.6 - - Venezuela 1.0 0.9 0.5 1.4 1.7 3.3 2.7 0.4 -2.3 1.8 0.3 2.6 Zaire 7.5 0.1 6.4 5.2 9.3 5.4 5.7 3.1 5.7 10.2 Zambia 0.5 0.4 1.0 1.3 1.4 1.2 1.5 1.4 8.0 4.0 10.1 Zimbabwe - - 1.9 1.8 1.2 0.2 0.6 1.3 1.0 - - a. For developing countries, 1970-89. Table A.3. Inflation Rates, 1965-90 (percent) Country 1965 1966 1967 1968 1969 1970 2971 1972 1973 1974 1975 1976 1977 OECD coiuilries Australia 4.0 3.0 3.2 2.7 2.9 3.9 6.1 5.9 9.5 15.1 15.1 13.5 12.3 Austria 4.9 2.2 4.0 2.8 3.1 4,4 4.7 6.4 7.5 9.5 8.4 7,3 5.5 Be'gium 4.1 4.2 2.9 2.7 3.7 3.9 4.3 5.4 7.0 12.7 12.8 9.2 7.1 Canadi 2.5 3.7 3.6 4.1 4.5 3.4 2.8 4.8 7.6 10.9 10.8 7.5 8.0 Denmark 5.5 7.1 8.2 8.0 3.5 6.5 5.9 6.6 9.3 15.3 9.6 9.0 11.1 Finland 4.8 3.9 5.6 9.2 2.2 2.7 6.5 7.1 11.0 16.7 17.8 14.4 12.7 France 2.7 2.6 2.8 4.6 6.1 5.9 5.5 6.2 7.3 13.7 11.8 9.6 9.4 Germany, Fed. Rep. 3.2 3.6 1.6 1.6 1.9 3.4 5.2 5.5 7.0 7.0 5.9 4.3 3.7 Greece 3.1 5.0 1.7 0.3 2.5 2.9 3.0 4.3 15.5 26.9 13.4 13.3 12.2 Ireland 5.0 3.1 3.2 4.7 7.4 8.2 9.0 8.6 11.4 17.0 20.9 18.0 13.6 Italy 4.2 2.4 3.9 1.5 2.2 5.1 4.9 6.0 10.6 19.2 17.1 16.6 17.0 en Japan 6.6 5.1 4.0 5.4 5.2 7.7 6.3 4.9 11.6 23.2 11.8 9.4 8.2 > Netherlands 5.8 5.8 3.5 3.7 7.4 3,7 7.5 7.8 8.0 9.6 10.2 9.1 6.5 Norway 4.3 3.3 4.4 3.5 3.11 0.6 6.3 7.2 7.4 9.4 11.7 9.2 9.0 Portugal 3,5 5.0 5.5 6.1 7.4 4.5 7.5 8.9 10.4 28.0 20.4 18.2 27.1 Spain 13.1 6.2 6.5 4.9 2.2 5.7 8.2 8.3 11.4 15.7 17.0 15.0 24.5 Sweden 5.0 6.4 4.3 1.9 2.7 7.0 7.4 6.0 6.7 9.9 9.8 10.3 11.5 Turkey 5.9 4.4 6.8 0.4 7.9 6.9 15.7 11.7 15.4 15.8 19.2 17.4 27.1 United Kingdom 4.8 3.9 2.4 4.7 5.5 6.4 9.4 7.1 9.2 15.9 24.2 16.5 15.9 United States 1.7 3.0 2.8 4.2 5.4 5.9 4.3 3.3 6.2 11.0 9.1 5.7 6.5 Developitng couitries Argentina 28.6 31.9 29.2 16.2 7.6 13.6 34.7 58.4 61.2 23.5 182.9 444.0 176.0 Bangladesh 7.9 8,0 5.4 2.7 5.0 3.0 -0.5 40.7 49.0 54.8 21.9 2.3 4.8 Bolivia 2.9 7.0 11.2 5.5 2.2 3.8 3.7 6.5 31.5 62.8 8.0 4.5 8.1 Botswana - - - - - - - - - - 12.0 11.7 13.2 Brazil 65.7 41.3 30,5 22.0 22.7 22.4 20.1 16.6 12.7 27.6 29.0 42.0 43.7 Burkina Faso -0.7 2.4 -4.3 -0.3 9.7 1.8 2.1 -2.9 7.6 8.7 18.8 -8.4 30.0 Burundi - 4.4 -1.1 6.1 4.0 -0.2 3,9 3.8 6.0 15.7 15.7 6.9 6.8 Camneroon - - - -1.1 5.9 4.0 8.1 10.4 17.2 13.6 9.9 14.7 Central African Rep, Chad Chile 28.8 23.1 18.8 26.3 30.4 32.5 20.0 74.8 361.5 504.7 374.7 211.8 91.9 China - - - - - - - - - - 0.3 2.5 Colombia 3.5 19.9 8,2 5.8 10.1 6.8 9.1 13.4 20.8 24.3 22.9 20.2 33.1 Congo 4.1 5.3 3.5 3.8 -0.7 1.5 4.1 9.8 3.5 5.4 17.4 7.2 14.0 Costa Rica -0.7 0.2 1.2 4.1 2.6 4r7 3.1 4.6 15,2 30.1 17.4 3.5 4.2 Cdte d'lvoire 1.3 5.5 2.3 5.3 4.5 9.4 -1.5 0.3 11.1 17.4 11.4 12.1 27.4 Czechoslovakia - - - - - - -0.4 -0.3 0.3 0.5 0.7 0.8 1.3 Dominica 2.5 2.5 1.0 5.4 4,2 12.4 3.6 3.7 12.1 34.4 19.9 10.9 86.5 Doominican Rep. -1.9 0,3 1.2 0.0 1.0 3.8 3.6 8.6 15.1 13.1 14.5 7.8 12.9 Ecuador 3.1 5.5 3.8 4.3 6.3 5.1 8.4 7.9 13.0 23.3 15.4 10.7 13.0 Egypt 14.8 9.0 0.7 -1.7 3.4 3.8 3.1 2.1 5.1 10.0 9.7 10.3 12.7 El Salvador 0.5 -12 1,5 2.5 -0.2 2.8 0.4 1.5 6.4 16.9 19.1 7.0 11.8 Ethiopia - -1.4 0.8 0.2 1.4 10.1 0.5 -6.1 8.9 8.6 6.6 28.5 16.7 Fiji - - - - - 4.1 9.1 22.0 11.1 14.5 13.1 11.4 7.0 Gambia, The 12 0.2 1A4 4.2 5.0 -2.0 3.1 8.7 6.9 9.2 25.9 17.0 12.4 Ghana 26.4 13.2 -8.4 7.9 7.3 3.0 9.6 10.1 17.7 18.1 29.8 56.1 116.5 Guatemala -0,8 0.7 0.5 1.9 2.1 2,3 -0.5 0.5 13.8 16.5 13.2 10.7 12.3 Haiti 2.3 8.3 -2,9 1.3 1.4 1.4 9.6 3.2 22.7 15.0 16.8 7,0 6.5 Honduras 3.2 1.8 2.1 1.9 1.3 2.9 2,2 3.6 5.2 12.8 8.4 4.9 8.4 Hungary - - - - - - - - 3.4 1.8 3.8 5.2 3.9 India 9.5 10.8 13.1 3.0 0.6 5.1 3,1 6.5 16.9 28.6 5.7 -7.6 8.3 Indonesla 305.5 14.5 106.0 128.8 15.5 12.3 4.4 6.5 31.0 40.6 19.1 19.9 11.0 Iran 2.2 -0.4 1.6 0.7 3.6 1.7 4.2 6.4 9.8 14.2 12.9 11.3 27.3 Israel 7.7 7.9 1.7 2.1 2.4 6.1 12.0 12.9 20.0 39.7 39.3 31.4 34.6 Jamaica 2.6 1,9 3.0 5.9 6.3 14.7 5.3 5.4 17.7 27.2 17.4 9.8 11.2 Jordan - - - - - 5.9 4.8 7.7 11.1 19.4 12.0 11.5 14.6 Kenya 3,6 5.0 1.8 0.4 -0.2 2.2 3.8 5.8 9.3 17.8 19.1 11.4 14.8 Korea, Rep. of - - 10,6 10.9 12.5 16.1 13.4 11.7 3.2 24.3 25.3 15.3 10.2 Lesotho - - - - - - - - - 13.4 14.2 11.4 16.7 Madagascar 4.2 3,2 0.8 1.0 3.8 2.9 5.4 5.6 6,1 22.1 8.2 5.0 3.1 Malaysia -0.1 1.0 4.6 -0.2 -0.4 1.8 1.6 3.2 10.6 17.3 4.5 2.6 4.8 Malta 1.6 0.5 0.7 2.0 2.3 3.7 2.3 3.4 7.7 7.3 8.8 0.6 10.0 (Table conzinies on Ihefollowing p4ge.) Table A.3 (continuied) Colnttry 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 Developing corut tries (conitin ued) Mauritius 1.7 2.5 1.9 6.9 2.3 1.6 0.3 5.4 13.5 29.1 14.8 12.9 9.2 Mexfco 3.6 4.2 3.0 2.3 3.4 5.2 5.3 5.0 12.0 23.8 15.2 15.8 29.0 Morocco 3.5 -1.0 -0.7 0.4 2.9 1.3 4.2 3.8 4.1 17.6 7.9 8.5 12.6 Nicaragua - - - - - - - - 27.0 13.3 7.5 2.8 11.4 Niger 4.4 10.6 0.4 -2.9 10.6 1.1 4.2 9.8 11.8 3.4 9.1 23.5 23.3 Nigeria 4.1 9.7 -3.7 -0.5 10.2 13.8 16.0 3.5 5.4 12.7 33.9 24.3 13.8 Pakistan 5.6 7.2 6.8 0.2 3,2 5.3 4,7 5.2 23.1 26.7 20.9 7.2 10.1 Panama 0.5 0.2 1.4 1.6 1.8 3.1 1.9 5.4 6.9 16.3 5.9 4.0 4.6 Papua New Guinea - - - - - - - 6.1 8.3 23.2 10.5 7.7 4.5 Paraguay 3.8 2.9 1,4 0,7 2.1 -0.7 4.8 9.5 12.5 25.2 6.8 4.6 9.3 Peru 16.4 8.8 9.8 19.1 6.2 5.0 6.8 7.2 9.5 16.9 23.6 33.5 38.1 Philippines 3.1 4.9 5.7 2.0 1.3 15.3 21.4 8.2 16.6 34.2 6,8 9.2 9.9 Poland - - - - - - 1.1 -0.1 2.5 7.1 2.3 4.4 4.9 Rwanda - - 1.5 3.1 0.5 0.5 0.5 3.1 9.4 31.1 30.2 7.2 13.7 Senegal - - - 0.1 4.0 2.8 3.9 6.2 11.3 16.6 31.7 1.1 11.3 Sferra Leone 4.7 4.3 4.9 1.6 3.1 6.4 -1.3 5.5 5.7 14.4 19.9 17.2 8.3 Somalia 13.0 -3.3 -0.3 3.4 6.3 0.9 -0.6 -3.0 6.5 18.2 19.4 14.2 10.6 South Africa 3.9 3.6 3.4 2.1 3.2 4.1 5.7 6.5 9.5 11.6 13.5 11.2 11.1 Sri Lanka 0.2 -0.2 2.2 5.9 7.5 5,9 2.7 6.3 9.6 12.3 6.6 1.3 1.2 Sudan -2.4 1.7 11.0 -10,0 12.6 4.0 1,3 13,6 15.3 26,2 24.0 1,7 17.1 Syria -3.9 3.8 6.5 2.8 -2.0 4.6 5,6 2.1 20.4 15.5 11.5 11.4 12.0 Tanzania - 9.8 12.2 15.6 16.4 3.5 4.7 7.6 10.5 19.2 26.5 6.9 11.6 Thailand 0.2 4,0 4.3 1.8 2.5 -0.1 0,5 4.8 15.5 24.3 5.3 4.1 7.6 Togo - - -2.3 0.3 6.0 4.5 6.5 7.7 3.6 12.8 18.0 11.6 22.5 Trinidad and Tobago 1.8 4.1 2.1 8.2 2.4 2.5 3.5 9.3 14.8 22.0 17,0 10.7 11.7 Tunisia 6.6 4.0 3.0 2.5 4.1 1.1 5.7 2.1 4.6 4.2 9,6 5.3 6.7 Uruguay 56.6 73.5 89.3 125.3 21.0 16.3 24.0 76.5 97.0 77.2 81.4 50.6 58.2 Venezuela 1.7 1.8 0.0 1.3 2,4 2.5 3.2 2.8 4.1 8.3 10.3 7.6 7.8 Yugoslavia 33.3 25.5 6.6 5.1 9.1 9.5 15.7 15.9 19.5 22.0 23.5 11.2 14.7 Zaire -2.7 15.8 36.9 53.3 6.2 8.0 5.8 15.8 15.6 29.5 28.7 80.4 68.9 Zambia 8.1 10.2 5.0 10.8 2.4 2.7 6.0 5.1 6.5 8.1 10.1 18.8 19.8 Zimbabwe 2.5 3.1 2,4 1.4 0.4 2.1 3.0 2.8 3.1 6.6 10.0 11.0 10.3 Co;,ntnj 1978 1979 1980 1981 1982 1983 1984 1985 1956 1987 1988 1989 7990 OECDc countries Australia 7.9 9,1 10.1 9,7 11,1 10.1 4.0 6.7 9.1 8.5 7.2 7.6 7.3 Austria 3.6 3.7 6.3 6.8 5.4 3.3 5.7 3.2 1.7 1.4 1,9 2.6 3.3 Belgium 4.5 4.5 6.7 7.6 8.7 7.7 6.3 4.9 1.3 1.6 1.2 3,1 3.5 Canada 8.9 9.1 10.2 12.5 10.8 5.8 4.3 4.0 4.2 4.4 4.0 5.0 4.8 Denmark 10.0 9.6 12.3 11.7 10.1 6,9 6.3 4.7 3,7 4.0 4.6 4.8 2.6 Finland 7.8 7.5 11.6 12.0 9.6 8,4 7.1 5.9 2.9 4.1 5.1 6.6 6.1 France 9,1 10.8 13.3 13.4 11.8 9.6 7.4 5.8 2.5 3.3 2.7 3.5 3.4 Germany, Fed. Rep. 2.7 4.1 5.4 6.3 5,3 3.3 2.4 2.2 -0.1 0.2 1.3 2.8 2.7 Greece 12.5 19.0 24.9 24.5 21.0 20.2 18.4 19.3 23.0 16.4 13.5 13.7 20,4 Ireland 7.6 13.2 18.2 20,4 17.1 10.5 8.6 5.4 3.8 3.1 2.2 4.0 3.4 Italy 12.2 14.8 21.0 17,9 16.5 14.7 10,8 9.2 5.9 4.7 5.0 6.3 6.4 Japan 4.2 3.7 7.7 4.9 2.7 1.9 2.3 2.0 0.6 0.0 0.7 2.3 3.1 Netherlands 4.1 4.2 6.5 6.7 5.9 2.8 3.3 2.2 0.1 -0.7 0.7 1.1 2.5 Norway 8.2 4.8 10.9 13.6 11.4 8.4 6.3 5.7 7,2 8.7 6.7 4.6 4.1 Portugal 22.7 23,6 16.6 20.0 22.7 25.1 28.9 19.3 11.7 9.4 9.6 12.6 13.4 Spain 19.8 15,7 15.6 14.5 14.4 12,2 11.3 8.8 8,8 5.3 4.8 6,8 6.7 Sweden 9.9 7,2 13.7 12.1 8.6 8.9 8.0 7.4 4.2 4.2 5.8 6.4 10.5 Turkey 45.3 58.7 110.2 36.6 30.8 31.4 48.4 45,0 34.6 38,8 75.4 69,6 60.3 Unitea Kingdom 8.2 13.5 18.0 11.9 8,6 4,6 5,0 6.1 3.4 4.1 4.9 7.8 9.5 United States 7.6 11.3 13.5 10.3 6.2 3,2 4.3 3.6 1.9 3.7 4.0 4.8 5.4 Deueloping contlries Argentina 175.5 159.5 100,8 104.5 164,8 343.8 626.7 672.1 90,1 131.3 343.0 3,080.0 2,314.0 Bangladesh 5.3 14.7 13.4 16.2 12.5 9.4 10.5 10.7 11.0 9.5 9,3 10.0 8.1 Bolivia 10.4 19.7 47,2 28.6 133,3 269,0 1,281.0 11,750.0 276,3 14.6 16.0 15.2 17.1 Botswana 9.0 11.7 13.6 16,4 11.1 10.5 8.6 8.1 10.0 9.8 8.4 11.6 11.4 Brazil 38.7 52.7 82.8 105.6 97.8 242.1 197.0 226,9 145.2 229.7 682.3 1,287.0 2,938,0 Burkina Paso 8.3 15.0 12.2 7.6 12.1 8.3 4.8 6,9 -2.6 -2.8 4.4 -0.5 -0.5 Burundi 23.9 36,5 2.5 12.0 5,8 8,4 14.4 3.6 1.9 7.1 4.5 11.6 7.0 Canmeroon 12.5 6.6 9.6 10.7 13.3 16.6 11.4 1.3 7.7 6.0 8.6 0.0 10.7 (Table continues on the followifg page.) Table A.3 (cotintined) Cosuntry 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 7988 1989 1990 Developinig countries (continued) Central African Rep. 13.3 14.6 2.5 10.4 2.2 -7.0 -4.0 07 0.0 Chad - - - 20.3 5,2 -13.1 -6,0 15.5 -4.9 0.6 Chile- 40.1 33.4 35.1 19.7 9.9 27.3 19,9 30.7 19.5 19.9 14.7 17.0 26.0 China 0.9 2.0 7,4 2.5 2.0 1.9 2.7 11.9 7.0 8.8 20.7 16.3 1.4 Colombia 17.8 24.7 26.5 27.5 24.5 19.8 16.1 24,0 18.9 23.3 28,1 25,8 29.1 Congo 10,5 8.1 7.3 17.0 12.8 7.8 12.7 6.1 2.5 2.3 3.7 4.1 -4.8 Costa Rica 6.0 9.2 18.1 37.1 90.1 32.6 12.0 15.1 11.8 16.8 20.8 16.5 19.0 C6te d'lvoire 13.0 16.6 14,7 8,8 7.3 5.9 4.3 1.8 7.3 04 7.0 1.0 -0.8 Czechoslovakia 1.6 3.9 2.9 0,8 5.1 0.9 0.9 2.3 0.5 0.1 0.1 1.4 10.0 Dominica -36.7 25.2 25.2 13.3 4.4 4.1 2.2 2.1 3.0 5.3 1.8 6.8 1.4 Dominican Rep, 3.5 9.2 16.8 7.5 7.6 4.8 27.0 37.5 9.7 15,9 44.4 45.4 59.4 Ecuador 11.7 10.3 13.0 16.4 16.3 48.4 31,2 28,0 23.0 29.5 58,2 75,6 48.5 Egypt 11.1 9.9 20.7 10,3 14.8 16.1 17,0 12.1 23.9 19.7 17,7 21.3 16.8 El Salvador 13.3 14.6 17,4 14.8 11.7 13.3 11.5 22.3 31.9 24.9 19.8 17.6 24.0 Ethlopia 14.3 16.0 4.5 6,1 5.9 -0.7 8.4 19.1 -9.8 -2.4 7.1 7.8 5.2 Fiji 6.1 7.8 14.5 11.2 7.0 6.7 5.3 4.4 1.8 5.7 11.8 6.2 8.2 Gambia, The 8.9 6.1 6.8 5.9 10.9 10.6 22.1 18.3 56.6 23.5 11,7 8.3 12.2 Ghana 73.1 54.4 50.1 116.5 22.3 122.9 39.7 10.3 24.6 39.8 31.4 25.2 37,3 Guatemala 8.3 11.3 10.8 11,4 0.3 4,5 3.4 18.7 36,9 12.3 10,8 11.4 41.2 Haiti -2.7 13.1 17,8 10.9 7.4 10.2 6.4 10,6 3.3 -11.4 4,1 6.9 21.5 Honduras 5.7 12.1 18.1 9.4 9.0 8.3 4.7 3.4 4.4 2.5 4,5 9.9 23.3 Hungary 4.? 9.0 9.3 4.5 7.0 6.4 8.7 7.0 5.3 8.2 16.3 16.9 29.0 India 2.5 6.3 11,4 13.1 7.9 11.9 8.3 5.6 8.7 8.8 9.4 6.2 9.0 Indonesia 8.1 18.3 18,0 12,2 9,5 11.8 10,5 4.7 5.8 9.3 8.0 6.4 12.5 Iran 11.7 10.5 20.6 24,2 18,7 19.7 12.5 4.4 18.4 28.6 28.7 22.3 7.6 Israel 50.6 78.3 131.0 116.8 120.4 145.6 373.8 304.6 48.1 19.8 16.3 20.2 17.2 Jamaica 34.9 29.1 27.3 12.7 6.5 11.6 27.8 25.7 15.1 6.7 8.3 14.4 22.0 Jordan 6.9 14.2 11.1 7.7 7.4 5.0 3.8 3.0 0.0 -0.2 6.6 30.0 16,2 Kenya 16.9 8.0 13.9 11.8 20.4 11.5 10.2 13.1 3.9 5.2 8,3 12.9 15.6 Korea, Rep. of 14.5 18.3 28,7 21.3 7.2 3.4 2.3 2,5 2.8 3,0 7.1 5,7 8.6 Lesotho 12.5 16.0 15.7 12,4 12.1 17.5 11.0 13.3 18.0 11.8 11.5 14.7 11.6 Madagascar 6.5 14.1 18.2 30.5 31.8 19.3 9.9 10.6 14.5 15.0 26.8 9.0 11.8 Malaysia 4.9 3.7 6.7 9.7 5.8 3.7 3.9 0.3 0.7 0.9 2.0 2.8 2.6 Malta 4.? 7.1 15.7 11.5 5.8 -0.9 -0.4 -0.2 2.0 0.4 0.9 0.8 3.0 Mauritius 8.5 14.5 42.0 14.5 11.4 5.6 7.4 6.7 1.6 0.5 9.2 12.7 13.5 Mexico 17,5 18.2 26.4 27.9 58.9 101.8 65,5 57.7 86.2 131,8 114.2 20.0 26.7 Morocco 9.7 8.3 9.4 12.5 10.5 6.2 12.4 7.7 8.7 2.7 2.4 3.1 6.8 Nicaragua 4.6 48.2 35.3 23,9 24.8 31.1 35.4 219.5 681.4 911.9 10,205.0 48.7 75.8 Niger 10.1 7.3 10.3 22.9 11.6 -2.5 8.4 -0.9 -3,2 -6.7 -1.4 -2.8 -0.8 Nigeria 21.7 11.7 10.0 20,8 7.7 23.2 39.6 5.5 5.4 10.2 38.3 50.4 7.4 Pakistan 6.1 8.3 11.9 11.9 5.9 6.4 6.1 5.6 3.5 4.7 8.8 7.8 9.1 Panama 4.2 8.0 13.8 7.3 4.3 2.1 1.6 1.0 -0.1 1.0 0.3 0.0 0.6 Papua New Guinea 5.8 5.8 12.1 8,1 5.5 7.9 7.4 3.7 5.5 3.3 5.4 4.5 6.9 Paraguay 10.6 28.3 22.4 14.0 6.8 13.4 20.3 25.2 31.7 21.8 22.8 26.4 38.2 Peru 57.8 66.7 59.1 75.4 64.4 111.2 110.2 163.4 77.9 85.8 667.0 3,399.0 7,482.0 Philippines 7.3 17.5 18.2 13.1 10.2 10.0 50.3 23.1 0.8 3.8 8.8 12.2 13.4 Poland 8.1 7.0 9.4 21.2 100.8 22.1 15.0 15.1 17.7 25.2 60.2 251.1 555.4 Rwanda 13.3 15.7 7.2 6.5 12.6 6.6 5.4 1.7 -1.1 4.1 3.0 1.0 4.2 Senegal 3.4 9.7 8.7 5.9 17.4 11.6 11.8 13.0 6.2 -4.1 -1.8 0.4 0,3 Sierra Leone 10.9 21.3 12.9 23,4 26.9 68,5 66,6 76.6 80.9 178.7 34.3 60.8 110.9 Somalia 10.0 24.3 58.8 44.4 23.6 36.0 91.2 37.8 35.8 28.2 81.9 - - South Africa 10.2 13.1 13.8 15.2 14.7 12.3 11.5 16.3 18.6 16.1 12.8 14.7 14.4 Sri Lanka 12.1 10.7 26.1 18.0 10,8 14.0 16.6 1.5 8.0 7.7 14.0 11.6 21.5 Sudan 19.2 31.1 25.4 24,6 25.7 30.6 34.1 45.4 24.5 20.6 64.7 63.5 68.5 Syria 4.8 4.6 19.3 18.4 14.3 6.1 9.2 17.3 36.1 59.5 34.6 114 19.4 Tanzania 11.3 13.8 30.3 25.6 28.9 27.1 35.3 34.1 32.4 29.9 31.2 25.8 19.7 Thailand 7.9 9.9 19.7 12.7 5.3 3.7 0.9 2,4 1.8 2.5 3.9 5.4 5.9 Togo 0.4 7.5 12.3 19.7 11.1 9.4 -3.5 -1.8 4.1 0.1 -0.1 -0.8 1.0 Trinidad and Tobago 10.3 14.7 17.5 14.3 11.6 15.2 13.3 7.6 7.7 10.8 7.8 11.4 11.0 Tunisia 5.3 7.8 10.0 8,9 13.7 8.9 8.4 8.0 5.8 7.2 6,4 7.4 6.5 Uruguay 44.5 66.8 63.5 34.0 19.0 49.2 55.3 72.2 76.4 63.6 62.2 80.4 112.5 Venezuela 7.1 12.4 21.5 16.2 9.6 6,3 12.2 11.4 11.5 28.1 29.5 84.2 40.8 Yugoslavia 14.1 20.7 30.9 39.8 31.5 40.2 54.7 72.3 89.8 120.8 194.1 1,240.0 583.1 Zaire 48.8 101.1 46.6 35.4 36.7 76.5 52.2 23.8 46.7 90.4 82,7 104.1 81.3 Zambia 16.3 9.7 11.6 13.0 13.6 19.6 20.0 37.3 51.8 43.0 55.6 127.9 117.5 Zimbabwe 5.7 18.2 5.4 13,2 10.6 23.1 20.2 8.5 14.3 12.5 7.4 12.9 17.4 -Not available. Table A.4. Real Interest Rates, 1965-89 (percent) Coittry 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 OECD Countries Australia - - - - - 0.1 -1.8 0.5 -6.8 -7,1 -4.4 -5.0 -0,2 Austria - - - - - - - - - - - - - Belgium -0.6 0.5 0,2 0.6 0.2 2,7 -0.9 -3.3 -2.8 -7.7 -5.1 -1.8 -0,8 Canada - - - - - - - - - - -1.6 3.3 -1.8 Denmark - Pinland - - - - - - - - - - - - France - - - - - -1.4 -1.8 -2.9 -4.3 -9.0 -3.5 -2.1 -1.8 Germany, Ped. Rep. - - - - - - - - - Greece 0a3 0.6 7.2 3.2 3.5 2.0 2.8 -0.7 -19.0 -3.2 -5.3 -2.5 -3.8 Ireland -0.7 -0.9 0,5 -1.3 -3.0 -5.0 -4.5 .-4.5 -5.0 -8.7 -7.8 -10.8 -4.2 Italy - - - - - - - - - - - - 5 N~'theJapan -2.4 -0,4 -1.6 0.2 -2.2 -3.9 -0.7 -1.8 -12.0 -13.0 -2.3 -5.6 -1.1 Netherlands - - - - - - - - - - - - - Norway_ __ _ Spain - Sweden - 1,3 2.5 3.8 2.9 -0.6 -0.6 -0.7 -1.9 -4.4 -1,7 -2.6 -3.7 Turkey - -5.6 -9.9 -3.0 -8,4 -5,6 -9.9 -3.0 -8.4 -8.3 -8.9 -6.8 -24.6 United Kingdom -0.2 0.9 1.6 -0.4 1.1 -2,5 -4,5 -3.3 -2.4 -7.9 -14.3 -6.6 -6.5 United States - - - - - - - - - - - - - Average OECD -0.7 -0.5 0.1 0.4 -0.8 -1.6 -2.4 -2,2 -7.0 -7.7 -5.5 -4.1 -4.9 Developing cousniries Argentina -20.4 -15.4 -13.4 0.4 3.0 -9.8 -18.8 -27.2 -17.0 -16.3 -72.4 -65.1 -19.2 Bangladesh - - - - - - -7.6 1.6 Bolivia - - - - - - - - - - - - - Brazil -4.3 -8.6 5.5 5.2 3.5 8.1 8.3 9.9 7.4 -5.0 -2.9 -9.4 -1.5 Burkina Faso - - - - - 2,0 -4.2 15.7 -8.8 5.2 -8.3 5.3 -15.5 Chile -10.0 -0P9 -5.0 -9.2 -7.9 -11.1 -3.1 -52,9 -78.8 -52.6 -18.5 3.9 21.7 Colombia -8,8 -7.9 -1.6 -0.6 -6.5 5.0 -5,0 -2.6 -6.9 -5,9 2.2 -3.9 -3.8 Cte d'lvoire 1.0 -0.5 0.0 -6.3 2.7 -3.3 3.9 1.2 -8.8 -12.0 -4.6 -7.2 -13.2 Czechoslovakia - - - - - 2.7 3.1 3.0 2.4 2.1 2.0 1.9 1.4 Dominica Ecuador -1.1 2.0 -1.5 1.4 -1.3 -3.1 1.1 1.0 -10.0 -10.5 -4.2 -4.1 -1.1 Ghana - - 13.6 -3.6 -3.1 0.9 -2,8 -3.2 -10,6 -10.9 -17.0 -31.0 -50.2 Honduras - - - - - - - - - - - - - India -2.9 -7.6 -0.5 7.0 0.2 2.6 0.5 -4.8 -16.7 -7.5 21.2 -1.0 3.0 Indonesia - - - - 46.2 16.5 23.7 -4.9 -8.9 -10.3 -3.0 1.6 0.8 Jamaica 1.7 1.0 0.9 -0.7 -4.0 3.5 2.5 -1.1 -14.5 -6.6 -3.2 3.5 -2.1 Kenya - - 1.5 3.0 3,6 1.9 -3.7 0.5 -10.8 -9.4 -12.6 -2.3 -13.1 Korea, Rep. of 12.9 14.0 13.7 11,6 8.9 7.2 8.4 3.2 -9.1 -8.3 4.0 4.7 Malawi - 1.8 1.1 -5.1 0.4 -4.9 -1.7 2.6 -2,7 -8.5 -8.6 6.1 -1.4 Malaysia 5.0 3.7 -0.6 9.1 3.7 5.4 3.1 2.0 -10.1 -2.7 6.8 3.6 1.1 Mexico 3.3 2.6 3.3 4.7 3.8 -0.5 1.5 0.8 -11.7 -7.9 0.2 -12.9 -6.4 Morocco - - 4,5 1.2 2.5 0.5 -1.6 0.4 -6.6 -8.7 -0.6 -7.0 -3.2 Nigeria - 3.0 7.3 -4.6 -9.3 -9.7 -6.7 4.7 -8.3 -18.6 -19.8 -9.8 -13.2 Oman - - - - - - - - - - - - 5Q Pakistan -3,0 -4.3 -4.2 3.5 4.1 -2.0 1.4 0.6 -9.4 -20.1 -11.7 2.8 0.1 W' Peru -6,9 -0.6 -9.9 -2.6 1.2 1.3 -0.6 2.6 -5.9 -10.2 -13.7 -21.9 -12.4 Philippines 5.7 -0,9 -2.1 5.0 2.6 -8.3 -4.2 3.4 -15.2 -15.7 7.7 5.0 2.1 Poland - - - - - - - - - - Sierra Leone - -1.6 -2.0 3.7 0.1 -4.1 8.5 -0.7 -1.5 -12.1 -6.2 -13.5 -2.4 Sri Lanka 2.3 3.2 -2.5 -3.1 -2.0 1.2 -1.1 0.7 -8.4 -5.4 6.3 4.9 13.3 Thailand - 2.1 3.9 5,7 4.0 8.3 5.7 -1.8 -10.9 -8.3 3.4 4.4 -0.8 Trinidad and Tobago - - - - - - - - - - Venezuela 4.9 6,6 7.5 4.9 5.1 3.7 4.8 4.5 1.7 -3.7 -0.4 0.5 -0.1 Zaire -2.9 -10.3 -41.7 -8.1 -8,1 -3.7 3,7 1.0 -7.8 -10.6 -14.9 -35,7 -26.6 Zambia -4.9 -3.1 -0.3 -5.9 3.0 0,2 -0.3 --0.6 -2.9 -2.1 -6.5 -14.4 -7.6 Zimbabwe - - - - - 1.1 0.2 0.4 0.1 -3.1 -6.1 -6.9 -6.6 (Table conibrtmes on 1hefollowing page wills tihe year 1978.) Table A.4 (conitinued) Country 1978 1979 2980 1982 1982 2983 1984 2985 1986 1987 1988 1989 OECD coin:tnries Australia 0.7 -1.6 -0.6 -0.8 1.3 1.9 7,0 2.0 3.8 6.3 3.9 7.0 Austria 1.3 0.3 -1.6 -1.3 0,3 0.4 -1,0 1.1 2.5 1.4 09 0.1 Belgium 0.6 0.3 a 1 -o.6 -0.6 -0.5 2.0 2.6 4.7 3.5 2.6 1.5 Canada 0.4 2,1 1.5 5.4 4.1 3.2 6,1 3.9 3.9 3.4 5.3 6.6 Denmark - -0.9 -0.1 0.7 3.6 3.7 3.2 4.5 2.2 2.9 3.1 3.3 Finland - - - -0.8 -0.3 0.2 2,9 3.6 3.8 3.3 1.8 -1.3 France -2.7 -5.3 -6.4 -5.4 -0.4 -0.9 0.9 2.0 3.0 2.1 1.9 2.3 Germany, Fed. Rep. 0.6 -0.3 2.3 2.8 2.8 1.9 2.8 2.7 4.8 2.2 1.5 2.4 Greece -1.4 -10,3 -9.3 -6.5 -3.8 -4.5 -2.3 -7.6 -1.2 -0.3 2.9 2.1 Ireland -1.5 -4,3 -5.3 -9.7 0.3 -0.9 1.1 2.0 3.2 3.1 0,9 -0.1 Italy -0.7 -6,9 -7,1 -3,2 -0.7 0.3 1.3 -0.7 4.5 1.7 1.2 0l.6 Japan -1.1 -2,. -1,3 -0.2 1.7 2.0 0.8 2,1 2.2 1.0 0.8 -0.3 mh Netherlands 1.1 0.7 -0,7 -1.1 1.5 1.0 1.3 2.4 4.0 4.3 2.2 2.2 Norway - 0A4 -7,6 -6.1 -5.9 -1.7 -0.6 4.2 1.9 4.3 5.5 5.2 Spain - -5.1 -1,9 -2.6 -1.5 0.1 3.0 2.1 0.7 4.2 3.0 2.5 Sweden 0.0 -1.8 -2.5 3.0 1.6 0.6 1.9 5.9 6.1 3.6 2.7 - Turkey -18.0 -33,7 -35.6 9.0 10.2 5.8 -3.1 3.1 8.6 -12.7 -14.9 -9.1 United Kingdom -2.1 -4.7 -0.9 -1.2 3.2 1.1 1.7 3.1 3,1 1.3 -2.1 -1.8 United States - - - - - - - - - - - - AverageoEcD -1.6 -4,3 -4.5 -1.0 1.0 0.8 1.6 2,2 3.4 2.0 1.3 1.4 Developitng colltnes Argentina -14.7 -9.5 -4.4 9.4 -27.7 -30.2 -39.0 -21.9 -12.0 -14.8 9.2 - Bangladesh 0.9 -9.4 -3.9 -6.4 5.5 1.1 1.5 -0.8 1.3 1.7 4.1 3.2 Bolivia - -20.2 -4.8 2.6 -67.1 -67,3 -90.8 - - Brazil 3.2 -15.8 -19.1 -4.5 4.6 -10,0 6.9 -0.8 -1.8 -5.9 -13.3 223.9 Burkina Paso -13.1 5.2 -3.9 8.0 1,2 9.3 - - - - - 7.7 Chile 27.0 5.0 5.2 28.0 23.3 5.3 1.7 4.1 1.4 3.1 - - Colombia 3.1 -4.6 0.7 2.1 3,2 9,9 8.6 5.6 8.6 5,5 4.8 Cdte d'lvoire -11.4 -7.6 -4.4 -0.3 3.6 1.1 5.2 - - - - - Czechoslovakia 1.1 -1.1 -0,2 1.8 -2.3 1.7 1.7 0.3 2.1 2.6 2.5 0,9 Dominica -5.0 -22.4 -14.3 -2.9 0.9 2.3 2,9 1.4 1.8 2.0 - 0.3 Ecuador -2.9 -0.4 -2.0 -6.2 -9.2 -24.9 -6.1 -2.7 -4,7 -5.4 -27.8 -9.1 Ghana -35.3 -27.5 -25.4 -45.3 -10.5 -49.3 -16.4 6.8 -3.8 -12,2 -9.0 Honduras - - - - 1.7 3.2 6.4 5,4 6.3 6.5 1.8 India 2.5 -4.8 -4.5 -0.7 -1.7 -2,8 2.8 0.1 - - - - Indonesia 2.5 -13.0 -6.6 1.9 -0,6 3.6 8.1 12.7 5.0 7.8 11.5 11.8 Jamaica -25.0 -7.9 -14.3 5.0 1.8 -2.0 -12.9 -1.8 7.5 8.4 8.6 1.6 Kenya -7.5 -3.7 -6.5 -8.8 -1.0 3.0 0.8 0.8 7.0 4.4 0.3 2.1 Korea, Rep. of 1.9 -2.1 -7.8 7.0 3.0 5.9 6.4 6.6 8.6 3,8 2.6 4.7 Malawi 2.3 -10.6 -2.6 1.6 -1.6 -4.9 0.7 115 -1.0 -15.7 -10.2 13.9 Malaysia 2.5 2,1 0.8 0.7 5.0 5.2 7.7 7.9 5.7 2.6 - - Mexico -2.7 -5.9 -6.6 0.2 -25.3 -11.5 -6.2 -0.5 -7.9 -21.7 3.6 22.1 Morocco -2.5 -1.8 -1.1 -4.2 3.1 -2.3 2.3 1.4 3.9 6.0 6.9 - Nigeria -7.7 -0.1 -13.4 -4.1 -5.0 -21.4 -8.7 8,0 -3.8 3,1 - Oman - - - - - - 9.0 9.0 8.3 7.5 7.6 8.7 Pakistan 2.1 2.6 -2.3 -1.8 2,9 3.8 3.2 - - - Peru -23.1 -20.1 -17.6 -11.4 -10.4 -31.1 -24.3 - - - Philippines 3.7 -8.8 -3.2 3.0 5.4 -9.7 -23.4 15.8 11.6 0.7 2.1 0.1 Poland - - -6.8 -20.5 -45.1 -11.4 -6.9 -7.0 -9.8 -18.4 -27.6 - Sierra Leone -2,7 -11.4 -0.9 -11.6 -11.6 -30.5 -37.1 -45.5 -45.5 -48.0 -15.9 -35.3 Sri Lanka -1.7 0.3 -3.7 1.5 15.8 0.5 13.3 20.2 2.9 1.2 -1.6 1.2 Thailand 0.2 -5.2 -3.8 0.6 10.2 8.8 13.4 9.4 7.9 5.6 6.2 3.2 Trinidad and Tobago - - - -4.5 -4,1 -6.7 -6.4 -1.3 -3.4 -2.1 - -2.8 Venezuela 0.9 -8.8 -6,4 2,7 7.6 6.4 -5.0 3.9 -3.4 -22.3 -19.6 -28.6 Zaire -40.4 -33.0 -1.2 -15.0 -7.8 -35.3 13.7 -3.7 -6.6 -51.6 - - Zambia -4.7 -2.5 -2.8 -5.0 -5.2 -10.3 -7.6 -27.2 -12.5 -24.8 -32.1 - Zimbabwe -2.5 -13.7 -1.9 -2.9 -5.1 -18.4 -3.0 0.8 -6.5 -1.4 0.1 -6.4 -Not available. Table A.5. Taxes from Financial Repression in Ten Countries (percentage of ctP) Stuidy ard coutitry 980 1981 1982 1983 1984 1985 1986 1987 1988 1989 Average Tax ont fina ticial initertiediatiorn (excilidinig inflaiion tax) Easterly (1989)" Argentina 0.1 0.1 0.7 1.7 0.3 -2.9 0.1 - - - 0.0 Chile 0.2 1.8 1.4 0.0 0.0 0.1 - - - - 0.4 t Colombia -0.2 -0,2 -0.2 -0.8 -0.9 -0.6 - -0,3 Mexico 0.7 -0.2 5.5 3,9 1.7 0.1 5.1 - - - 2.0 Morocco 0.6 1.0 0.0 0.8 0.2 0.3 -0.6 - - 0.4 Thailand 0.3 -0.0 -0,8 -0.7 -1.2 -1.0 -0,8 - - - -0.4 Giovannini and de Melo (1990)b Colombia 0.2 0.2 0.2 0.3 0.4 - - - - 0.3 Mexico - - - - 0.5 0.8 10.8 11.0 - - 5.8 M.;torocco 1.1 5.5 2.9 3.7 4.7 3.3 - - - - 2.3 Pakistan - - 3.6 2.9 - - 3.3 Thailand 1.2 0.9 0.2 0.5 1.7 -0,9 - - - - 0.4 Zimbabwe 5.8 4.6 9.2 6.7 7.4 -0.5 - - - - 5.5 Chamley and Honohan (1990)c Cote d'lvoire - 0.0 0.0 0,0 0.0 0.0 0.0 0.0 0.0 - 0.0 Ghana - 0.0 2.4 0.0 0.1 0.4 0.5 0.2 0.2 - 0.5 Tax on time deposits; estimnations fromti tlis studyd Argentina 0.8 -1.4 3.4 3.5 4.M 2.2 1.6 2.1 -1.2 - 1.7 Chile -0.4 -4.7 -3.8 -0.5 -0.1 -0.4 -0,1 -0.4 - - -1.2 COte d'lvoire 0.4 0.1 -0.2 0,0 -0.4 - - 0.0 Colombia 0.0 -0,1 -0.1 -0.5 -0.4 -0.3 -0.4 -0.2 -0.2 - -0.2 Ghana 0.9 1.3 0.4 0.9 0.3 -0.1 0.1 0.4 0.3 - 0.5 Morocco 0.1 0.4 -0.2 0.3 -0.2 -0.0 -0.4 -0.7 -0.8 - -0.2 Mexico 1.3 0.2 3.9 1.9 1.2 0.2 1.2 2.9 - - 1,6 Pakistan 0.4 0.3 -0.3 -0.4 -0.3 - - - - - -0.1 Thailand 1.2 0.1 -3.2 -3.1 -5.7 -4.2 -3.6 -2.5 -2.7 -1.2 -2.5 Zimbabwe 0.5 0.6 1.0 2.6 0.6 0.0 1.0 0.4 - - 0.8 Not available. a. The tax is calculated as the negative of the real interest rate multiplied by government domestic debt outside the central bank as a percentage of cDp. (The inmplicit long-run real interest rate is assumed to be zero.) b. The tax is calculated as the difference between the foreign and the domestic Interest rate multiplied by government domestic debt outside the central to bank as a percentage of GDP. c. The tax is calculated as I minus the domestic real interest rate multiplied by government debt outside the central bank as a percentage of GDPI. d. The tax is calculated as the OECD aver.3e real interest rate (0.9 percent) minus domestic real interest rate multiplied by time deposits as a percentage of GOp. The data used are from IMF, Inteniational Finnaicial Stnftstics, various years. Author Index Almansi, Aquiles, 136 de Haan, J., 24 Anand, Ritu, 68, 270 n4, 326,409 de Melo, Jaime, 390 n6 de Melo, Martha, 16, 68, 73 n19, Arrau, Patricio, 218 n4 n20 Aryeetey, Ernest, 321 de Pablo, Juan Carlos, 106, 164 nl Aschauer, David, 59 Diaz-Alejandro, Carlos, 150 Auemheimer, Leonardo, 82 Dombusch, Rudiger, 15, 16, 72 n14, 73 niS, n20, 86, 88, 106, 150, 183, Balassa, Bela, 219 nl5 219 n16 Barro, Robert F., 16, 54-55, 72 n13 Bernanke, Ben S., 403 Easterly, William, 16, 17, 39, 40, 73 Bermheim, B. Douglas, 16 n20, 183, 194,197, 20, 323, 462, Bertola, Giuseppe, 405 474 Blanchard, Olivier, 71 n5 Edwards, Sebastian, 60,72 nl3, 164 Blejer, Mario I., 15, 16, 49, 59, 79, nl 82 EIbadawi, Dbrahim A., 477,502, Blinder, Alan S., 403 503,504 n2, 506 n21 Buiter, Willem H., 68, 69,183, 409 n6, 427 Faini, Riccardo, 390, 396, 404, 405 Ferndndez, Roque, 106, 136, 164 Cagan, Phillip, 40 nl Calvo, Guillermo A., 88, 106, 164 Fischer, Stanley, 15, 17, 38,72 nn5 nl Flavin, Marjorie, 68 Carr, Jack, 184 Frenkel, Jacob A., 82, 88 Cavallo, Domingo, 102,150 Fry, Maxwell J., 219 nlS Chamley, Christophe, 73 n20, 307 n4, 322,323 Garcia Garcia, Jorge, 225 Cheasty, Adrienne, 15 . Gelb, Alan, 25 Chhibber, Ajay, 68, 326, 474 475 Giavazzi, Francesco, 71 n5 Corbo, Vittorio, 16, 57, 73 n21, 219 Giovannini, Alberto, 16, 57, 73 n19, nl5, 486, 489 n20, 219 ni5 Corsetti, Giancarlo, 16, 57, 219 n14, Gockel, Fritz, 321 n15 Grilli, Vittorio, 68 Darby, Michael R., 184 Hamilton, J., 68 Dai d, Paul, 439 Haque, Nadeern LTI, 16, 57, 219 Davidson, James F. H., 440 n14, 395, 396, 409 n6, 439 Deatnn, Angus, 395 Harberger, A. C., 86,150 551 552 Authorlndex Hayashi, Fumio, 16 Rodrfguez, Carlos Alfredo, 16, 17, Hendry, David, 395,440. 60, 63, 84, 86, 87, 88,106,114, Honohan, Patrick, 73 n20 136, 150, 164 nl, 194,197, 205, Horton, Brendan, 304 256, 291, 323,462,493,496 Rossi, Nicola, 219 n14, 395 Islam, Roumeen, 313 Sachs, Jeffrey, 16 Sargent, Thormas J., 15, 462, 483 Kemal, A. Ri, 450 Scadding, John, 439 Khadz, Ali, 504 n2 KChan, Mohsin S., 16, 59, 67 Schenone, Osvaldo H., 108 Kiguel, Miguel A, 15, 49 Schmidt-Hebbel, Klaus, 16, 17,33, 40, 57, 73 n21, 194, 197, 199, 205, Leidermn, Leonrdo, 1679, 82219 n14, n15, .317, 323, 396, 462, Leiderman, Leonardo, 16, 79 82 . 467, 475, 477, 486, 489, 5o,502 , Liviatan, Nissan, 15,49 Lizondo, Samuel, 67,336,337, Seater, n J.16 Serv&n, Luis, 16 S harif, Nawaz, 426 Marshall, Jorge, 17, 33, 218 n7, 317, Sar Ly, N6 467 Sjaastad, Larry A., 86, 150, 164 nl Solimano, Andrds, 16,197 Mayro, Paulo, 34 Srba, Frank, 440 May, Emnesto, 336, 348 Stiglitz, Joseph, 403 Mingat, Alan, 3 Sturzenegger, Fede0co, 7 n14, 73 Montiel, Peter J., 16, 57, 219 n14, nl8 395, 396,409 n6, 439 Morandd, Felipe, 467, 475, 504 n2, Tabellit, Guido, 72 nl 113 Tanzi, Vito, 280, 523 n7, 524 ni8 Morshed, Nader, 432 tmbergen, Jan, 523 n9 Mujica, Patricio, 183 Mhuler, Tobias, 199, 396 -van der Gaag, Jacques, 307 van Wijnbergen, Sweder, 68, 69, Nam, Sang-Woo, 395 183, 270 n4, 326, 396, 397, 398, 404, 409 n6 Pagan, Adrian, 384 Vijverberg, Wim, 307 Pagano, Marco, 71 n5 von Ungem Stemnberg, T., 395 Patel, U., 68 Penaa, Angel, 150 Wallace, Neil, 15, 462, 485 Pinto, Brian, 336, 337 Webb, Steven B., 16, 57, 219 n14, Porter, John, 396, 404 nl5 Psacharopouos, George, 306 Werner, Martin, 68 Wetzel, Deborah, 313 Rama, Martin, 16 Wilcox, David, 68 Razin, Assaf, 82 Wolf, Holger, 72 n14, 73 nl8 Reinhart, Carmen M., 16, 59 Reynoso, Alejandro, 73 n2O Yeo, Stephen, 440 Rimmer, Douglas, 369 nl Rocha, Roberto, 68, 326 Zelhorst, D., 24 Subject Index Agricultural marketing boards, 104,106-07, 143,155-56,157; reforms of, 35, 38, 313 public sector primary surpluses Argentina: Austral plan in, 53, 128, in, 102; public spending 142; Banco Hipotecario (state (consumption) in, 102, 104, 108, mortgage bank) of, 117; bank 162-64; quasi-fiscal deficits and deposits in, 122, 136-39, 139-45; expenditures in, 6, 19, 114-16, BoNEx plan in, 101, 126; Bunge 118; real exchange rate in, 149- and Born plan in, 53,126,142; 54; Rod rigazo inflation in, 122; currency devaluations in, 6-7, saving in, 5; stabilization plans 106; currency overvaluation in, in, 53, 101-02,124,126,128, 142, 102; debt meltdowns in, 53, 106- 143, 164 nl; terms of trade for, 07; deficit financing in, 3, 9, 116- 151, 152, 153, 155; trade balance 2-2; demand for real cash in, 9, 154-56. See also Central balances in, 128, 130-32; dollar- Bank (Argentina) denominated instruments (LEDOL Asset portfolio model, 183-85, 213- and BONEX) in, 101, 142-45; 15 "dollarization" in, 6-7,107; domestic debt of, 9, 124; Bank deposits: in Colombia, 230, economic growth in, 9-10, 102, 244-45; interest elasticity of 104, 106; Erman plan in, 53, 142; demand in Argentina and, 139- external debt of, 9, 82, 153; -45; interest rates in Argentina financial markets in, 135-36; and, 136-39; Islamic noni-nterest fiscal policies in, 34,35; foreign basis for, 446; plazs ffjos, 122. See shocks in, 26; indexation in, 106; also Interest rates inflation and hyperinflation in, 3, Banque Centrale des Etats de 6, 46, 101-02, 104,12; inflation l'Afrique de l'Ouest (BcEAo), tax in, 117-19; interest rates in, 273 122, 136-39; measuring CNFPS of, Barro-Ricardian hypothesis on 105-06,108-09,114; money consumption, 64-65. See also creation (seigniorage) in, 43, Ricardian equivalence 114, 117-19, 120-22; money hypothesis demand in, 126, 128, 130-32, Black market: fiscal balances and, Primavera plan in, 53, 124,126, 24, 336-37, 339-46; in Ghana, 128,142; private consumption in, 316-17, 336-46; methodology for 158-62; private credit as share of analyzing, 337-46 GDP in, 53; private investment Bolivia, 46 and saving in, 4, 5, 162-64; Budget deficit. See Public sector public sector primary deficits in, deficit 553 554 Subject Index Cagan money demand, 40,47,48 78, 183; real exchange rate in, 3; Central Bank (Argentina): CNrFs seigniorage in, 43; short-run financing by, 109, 114; converted demand for public debt in, 185- into currency board, 101-02; 88; state pension system in, 218 domestic liabilities of, 124; n4; structural adjustment in, 6, financial markets regulated by, 10; sustainable public sector 135-36; interest Tates on deposits deficit in, 179-80; termns of trade at, 137-39; LEDOL (dollar- for, in. 173, 202, 205, 206, 207. denominated bill) issued by, 143; See also Central Bank (Chile) quasi-fiscal expenditures of, 114- CNFPS. See Consolidated 16 nonfinancial public sector Central Bank (Chile): estimated Cocoa prices: in Cote d'lvoire, 273- deficit of (1991), 179; losses from 78; Ghana's black market and, quasi-fiscal operations of, 6, 19, 336-46; Ghana's public spending 167,178; public policy and, 311; Ghana's tax revenue innovations of 1980s and, 172- and, 316-17 73; quasi-fiscal programs of, 177- Coffee prices: Colombia's trade 78,183 balance and, 270 n2; in COte Chile: alternative financing d'lvoire, 273-78, 303, 305; strategies and inflation in, 188- relative prices and, 289-90 91; CNFPS deficit in, 167, 175; Colombia: adjustment program of crowding-out and -in of private 1985-89 in, 231, 233-36; bank consumption in, 191,194-95, deposits in, 230, 244-45; Central 197,210; crowding-out and -in of Bank of, 263; consumption private investment in, 4, 191, deflator in, 237-39; crowding-out 197, 199,201-02, 210; data of private investment in, 4, 231, sources on, 215-17; 240,270 n6; deficit financing in, decomposition of CNFPS in, 175- 233, 237-41, 244-45, 250-54, 262; 77; deficit financing in, 167-68, definitions of macroeconomic 171-74; deficit financing model variables for, 263-69; fiscal deficit and simulations for, 180, 183-89, and real exchange rate in, 254- 194; deficit financing and private 62; fiscal policies in, 33; foreign consumption in, 191,194,197; shocks in, 3,26-27; govemment deficit financing and private balance of, 227; inflation and investment in, 191, 197, 201-0Ž; public sector deficit in, 8-9, 28; domestic debt in, 178, 180,183; macroeconomic management in, fiscal policies in, 33, 35; fiscal 225-31, 262-63; money demnard policies and trade balance in, in, 239-40; private consumption 202-03, 205-07; foreign shocks in, 244; producthvity growth in, in, 25-27; indexation in, 6; 244; seigniorage in, 47; structural inflation and public sector deficit deficit in, 233, 235; sustainable in, 28, 180; price controls in, 46; public sector surplus in, 20; private sector saving in, 194, 197, terms of trade for, 255-56, 261 201, 210; public sector primary Commodity booms, 25. See also deficit in, 179,180, 183-84, 217 Cocoa prices; Coffee prices n2; public sector primary surplus Commodity marketino boards, 35, in, 10, 20, 179480; public saving 38 in, 194,195, 197, 219 ni8, nM9; Consolidated nonfinancial public quasi-fiscal deficits in, 19,178; sector (CNFmPs): in Argentina, 105- quasi-fiscal operations in, 177- 06, 108-09,114, 156; in Chile, Subject Index 555 167-68,175-77; macroeconomic 454; in Zimbabwe, 459,462, 485- shocks and deficits of, 32; in 86,489, 491-92, 502-03 Zimbabwe, 458, 466-71. See also Currency devaluations in Public sector deficit Argentina, 6-7, 106 Consumer price index (crl). See Currency overvaluation, 2; in Inflation Argentina, 102 Consumption: Keynesian hypothesis on, 54, 55, 73 n2l, Debt: Argentina's domestic, 124, 194-95, 331, 487, 489, 506 n17; 126; Argentina's meltdowns of, Ricardian hypothesis on, 16, 54, 53, 106-07; deficit financed by 55, 57, 64-65, 67, 194,197, 332, extemal, 79, 81-82; deficit 370 n16, 438, 486-87, 489. See also financed by intemal, 82-83; as a Private consumption; Public deficit measure, 68; inflation spending; entries under individual and, 28, 38; interest rates and, countries 137-38; macroeconomic variables Cate d'lvoire: balance of payments correlated with, 23-25; output of, 295-99; Caisse de Stabilisation and, 31-32; public sector of, 281, 283; Central Bank of, 289; solvency and, 67-68; real interest coffee and cocoa booms in, 9, rate and, 29-30; sustainable, 69- 27378, 303, 305; constraints on 70; trade balance and financing fiscal policy of, 279-8t; of, 80-84. See also Public sector consumption function in, 300; deficit deficit financing in, 9, 295-98; Deficit financing: by Argentina, 3, defining fiscal surplus of, 284-87; 9, 119; asset portfolio model educational expenditures in, derivation for, 213-15; in Chile, 306-07; fiscal policies of, 34, 35, 167-68, 171-74; in Colombia, 302-03; foreign shocks in, 9, 25, 233, 237-54; constrints on Cote 27, 274-76, 278-79; inflation and d'lvoire's, 279-80; by Cdte real exchange rate in, 289-95; d'lvoire's private sector, 295-98; public sector spending in, 9, 287- Ghana's use of domestic, 319-22, 88; rates of private saving and through inflation tax in investment in, 299-302; tax Argentina, 116-22; methods and revenues in, 281484, 304-05; consequences of, 17; model and terms of trade for, 297, 300 simulations for, in Chile, 180, Crowding-out of private consump- 183-89, 194; of Morocco's deficit, tion, 5, 16, 17, 24, 54-55, 57, 73 381-83; in Pakistan, 425, 434, n21, 81; in Chile, 191, 194-95, 448, 451; private consumption in 197, 210; in Pakistan, 445; in Argentina and, 158-62; private Zimbabwe, 485-87,489. S.e aLso consumption in Chile and, 191, Argentina, private consumption 194, 197; private investment in in; Ghana, private consumption Argentina and, 162-64; private in; Morocco, private consump- investment in Chile and, 191, tion in 197,201-02; public sector Crowding-out of private expenditure in Argentina and, investment, 2, 4, 5, 16, 17, 24, 158-62; tax versus debt, 79-80; 59-60, 65; in Chile, 191, 197, 199, trade balance and, 80-84; in 201-02,210; in Colombia, 231, Zimbabwe, 474-79, 482-83, 485. 240, 270 n6; in Cote d'lvoire, 301; See also Seigniorage in Morocco, 383, 394, 399, 401, Denmark, 46 402-03; in Pakistan, 438,451, Dependent-economy model, 60-63 556 Subject Index Developing countries: control over 18; through tax reform, 521-22, policymaking in, 518-20; 524 n14 correlation of real exchange rates Fiscal deficits: black market and, and deficits in, 60-63; credibility 24, 346-53; Chile's fiscal policies and bias issues in, 515-18; and, 33-36, 174; Colombia's external shocks in, 25-27; fiscal financing of, 233, 237-38, 250-54, adjustment issues for, 16; 262; conclusions regarding, 1-2; government response to current account and, 24-25; macroeconomic imbalances in, domestic macroeconomic 513-15; implementation of fiscal shocks and, 28-32; financial policy in, 520-2Z; insensitivity of liberalization and, 29-30; foreign savings to real interest rate in, financing and, 30; foreign shocks 29 nIS; reduced external and, 25-27; in Ghana, 33, 313- financing by indebted, 183; 19, 329-34; inflation and, 28, seigniorage in, 42; tax reform 119-22; interest rates and, 29-30, in, 521-22; twin transfer of 49T '3, 66; investment and, 24; resources in, 383 macroeconomic adjustment and, Direct crowding-out hypothesis, 15; macroeconomic variables 55, 57,60,197,487,489 and, 23, 25; in Morocco, 375-83; Disposable income: defining, 439; in Pakistan, 7, 23, 427-37, 438, fiscal policy impact on, 54-57 439, 445, 448, 451; real exchange "Dutch disease," 288-89 rate and, 3, 24-25, 6063, 2542; seigninrage and, 23-24, 47-49, Economic forecasts, 515-18 54-66; trade balance and, 154-56; Econonic growth, 31-32,46-47; in trade surplus and, 61-63, 67, in Argentina, 9-10, 102, 104, 106; in Zimbabwe, 33. See also Debt Ghana, 310, 311; in Pakistan, Fiscal policies: in Chile, 207-11; 415; in Zimbabwe, 497,499 Colombian adjustment program Educational expenditures, 306-07, and, 231, 233-36; in C&te 506 n19 d'Ivoire, 279-0, 28-30; fiscal Effective tax base and rate, 522 adjustment using, 15-18, 65; External deficits: interest rates on fiscal deficits due to, 1; under Pakistan's, 433; public sector Ghana's Economic Recovery deficits and, 16-17, See ao Trade Progranune, 310, 313, 352-54; 'balance; Trade surplus implementation issues External sector. Ghana's fiscal concerning, 520-22; interest rates defict and, 335-53; impact of and debt and, 145-49; models for external prices on Chile's, 202- analysis of, 157-64, 396-403, 493- 07. See also Trade balance 97; in Morocco, 34, 399-400, 402- External shocks. See Foreign shocks 04; in Paldstan, 418-27; Pakistan's private and public Financial repression. See Interest investment and, 43742; private rates consumption and, 54-57, 66; Fiscal adjustment, 15-17, 65; private investment and, 57-60, currency overvaluation and, 2; 66-67; public sector deficits and, fiscal policies and, 33-34; 33-38, 174; as response to implementation of, 520-22; macroeconomic imbalances, 513- political economy of, 513-22; 15; tax revenues and, 211-13. See selection criteria for study of, 17- also Fiscal adjustment Subject Index -557 Fiscal surplus. See Fiscal deficits; and, 152; Argentina's private Structural surplus or defict credit and, 53; Cote d'lvoire's Foreign interest rates, 26. See aLso educational expenses as share of, Interest rates 306-07; COte d'Ivoire's tax Foreign shocks: blamed for revenue as share of, 281483; econonmc conditions, 514-15; foreign shocks and C8te cessation of U.S. aid to Pakistan d'lvoire's, 278-79; growth of, 46- and, 421; contnbution to C6te 47; public sector deficits as share d'Ivoire's CWD by, 278-79; COte of, 24; seigniorage spikes and, d'Ivoire commodity prices and, 46-47; Zimbabwe's prospects for 274-76; deficits and, 25-27; growth of, 497, 499; Zimbabwe's measurement of, 276-78 public sector deficit as share of, 471-74. See also Gross national Ghana: asset demand in, 322-24; product (cuP) of Pakistan balance of payments in, 341-42; Gross national product (aN?) of bladk market in, 336-53; causes Pakistan: deficits as share of, of inflation in, 320-26; cocoa 430, 434, 436; growth of, 415; smuggling in, 316-17, 336, 34 public sector liabilities as share devaluation in, 4, 345; domestic of, 431; use of, 455 n3 financing of public sector defiit in, 319-22; economic and fiscal Hypernflation in Argentina, 6; policies in, 6, 33, 3 ; maximum revenue aised prior Economic Recovery Programme to, w political crisis and, in, 310,313,352-54; fiscal deficit 1013-2 See atso Inflation in, 313-19,326-30,329-34; foreign shocks in, 25-27; cDP . growth in, 310, 311; inflation in, Oruanizal our SfE 6, 28, 311, 320-26, 345-46; C eon and Develpment interest rates in, 4; measurnng (oo)ecount fiscal deficit of, 313-19; (In ona)countries methodology for case study on, ration: aIternatve financing 354-8; money demand in, 32:2- strategies and, 188,191; in 26, 336, 341, 343, 353; private Argentina, 3, 6, 46, 101-02,122; consumption in, 330-33; private in Chile, 180; in Colombia, 227- investment in, 4; public savings ~ 2 237-38, 245, 248; in C6te i, 331,332; real exchange rate d'lvoire, 289-95; determining in, 3; seigniorage (money impact of deficit on, 121-22, 427- aeation) in, 3, 43, 47, 322-24; 36; in Ghana, 320-26,345-46; in sustainable primary deficit in, Morocco, 374, 384-89; in 19-20, 326-30; tax revenues in, Pakistan, 427-36; public sector 316; terms of trade for, 336, 342, deficits and, 15; rejection of fiscal 344, 345, 346, 347, 348, 351, 352, approach to, 1; seigniorage and, 370 n23 3-4, 17, 3840, 4749, 245, 248, Government expenditure. See 325-26; sensitivity of deficits to, Public spending 28, 38, 65-66; in Zimbabwe, 477- Gross domestic product (GDP): 79, 482-83, 485. See also Inflation Argentina's, 102,104; tax Argentina's CN}PS expenditumes Inflation tax: in Argentina, 117-19; as share of, 108-13; Argentina's assets markets and, 8889; in nominal government spending Colombia, 245, 248-50; 558 Subject Index Inflation tax (continued) Macroeconomic variables: deficits computing maximum, 132-35; correlated with other, 23-25; fiscal defidt and, 119-22; real sensitivity of deficits to domestic, exchange rate and, 89-95; tax 28-32 reduction financed by, 83 Mexico, 3,-4, 34, 35, 43, 46, 49-50, Interest-bearing deposits, in 59, 60, 61, 62, 63 Argentina, 122, 139. See also Bank Money creation. See Seigniorage deposits Money demand: in Argentina, 126, Interest rates: Argentina's 1977 128, 130-32,; in Colombia, 239- reform and, 122; for bank 40; in Ghana, 322-26, 336, 341, deposits in Argentina, 136-39; in 343, 353; seigniorage Laffer curve Chile, 180; Colombia's deficit and misspecified, 38-40; in financing and, 237-41, 244-54 Zimbabwe, 477-79 demand for deposits and, 139- Moroo: alternative defict 45; effect of deficit on real, 16, measures in, 20; asset demand in, 29-30, 49, 54; fiscal deficits and 38-8; cowding-out of private repression of, 4, 49-53, 66; investment in, 383, 394, 399, 401, foreign, 26; Islanic system's 402-03; deficit financing in, 381- nonuse of, 446; Pakistan's 83; economic history of, 374-75; domestic and external debt and, estimating expected inflation in, 433; tradeoff between debt policy 384-89; evolution of fiscal deficit and, 145-49. See also Bank in, 375-83; fiscal adjustment in, deposits 20; foreign shocks in, 25-26; International Monetary Fund (IMr}: impact of fiscal policies on, 34, data from, 227,414; Morocco 399400, 402-04; impact of fiscal and, 374 technical advice to policies on investment and saving Ghana by, 311; warnings to C6te in, 389-96; inflation and public d'Ivoire by, 273 sector deficit in, 28; measuring Investment. See Private investment; defict in, 18; methodology for Public investment; entries under case study of, 404-08; model for individual countries analysis of fiscal policies of, 396- 403; model for investment and Keynesian hypothesis on saving decisions in, 389-96; consumption, 54, 55, 73 n21, primary public sector surplus in, 194-95, 331, 487, 489, 506 n17 20; private consumption in, 394- 96,405,- private saving in, 5, 393- Labor force: in Cote d'Ivoire, 304; 96; tax revenues of, 379480; terms Ghana's Alien Compliance Act of trade for, 8, 374, 375, 378, 380, of 1%9 and, 316 382, 383, 409 n5 Laffer curve, 3, 38-40, 65-66, 72 nl3, 147, 148, 245, 271 nll, 283, 326 National saving, 2, 419. See ls Loan rates, in Colombia, 230, 244- Private investment; Public 45. See also Deficit financing investment Neodassical solvency approach, Macreconomic adjustment, 15. See 68 also FLscal adjustnent Net foreign assets, 206, 238-40 Macroeconomic imbalances, 513- Neutral tax, 79 15. See also FLscal adjustnent Macroeconomic shocks, 32. See also Olivera-Tanzi effect, 3, 28,72 n9, Foreign shocks 211, 317,504 n3 Su4ject Index 559 Orgarization of Economic 54-5$ 5r7, 73 n21, 81; disposable Cooperation and Development income and, 487489; fiscal (oEwD) countries: seigniorage in, polcies and, 54-57,66; Ghana's 42, 47; statistical data for, 23 fiscal deficit and, 330-33; in Output, deficits and, 31-32. Seealso Morocco, 394-96, 405; Pakistan Gross domestic product and crowding-out of, 445; Pakistan's fiscal policy and, Pakistan: consumption in, 437-41; 437-42,448-51; public sector crowding-out of private deficts and, 16; Zimbabwe consumption and investment in, and crowding-out of, 485-87, 438, 451, 454; debt service ratio 489 of, 415-16, 418; deficit financing Private investment: in Argentina, by, 434; determinants of real 162-64; Chile and crowding-out output growth in, 442-43; of, 191,197, 199, 201-02, 210; determinants of tax revenues of, Colombia and crowding-out of, 424, 455 n4; domestic financial 231, 240, 270 n6; in C6te d'Ivoire, sector model in, 443-45; 299-302; crowding-out of, 2, 4, 5, economic history of, 414-18; 16,17,24,59-60, 65; Ghana's fiscal policy during Bhutto fiscal deficit and, 333-36; impact govemment in, 419-21, 426, 454; of deficit financing on, 162-64; fiscal policy (1982-8 in, 424-25; impact of fiscal policy on, 57-60, fiscal policy simulation model 66-67; methodology for for, 445-51; fiscal poliqy under estimating Zimbabwe's, 489, martial law in, 421, 424; foreign 491-9Z- Morocco and crowding- shocks in, 25, 421; impact of out of, 383, 394,399,401,402-03; fiscal polices on, 34, 437-45; in Pakistan, 416, 418; Pakistan inflation and public sector deficit and crowding-out of, 438,451, in, 428-36; macroeconomic 454; Pakistan's fiscal polices effects of debt reduction in, 448, and, 437-42 public sector 451; macroeconomic indicators deficits and, 1-2, 16; Ziimbabwe for, 426-27; primary public sector and crowding-out of, 485-86, deficit in, 23; private saving and 489,491-92,502-03 investment in, 416,418,438,441- Private saving: in Argentina, 163; 42, 451, 454; public sector deficit in Chile, 194, 197, 201, 210; in in, 7, 413-14, 420-21, 424-25, Colombia, 238, 244, 257-58; in 426,427-32 Cate d'voire, 297; in Morocco, Permanent income hypothesis, 53, 393-96, 405; in Pakistan, 425, 55, 73 n2l, 194-9, 219 n8, 393- 445,448; in Zimbabwe, 475,497, 94,487,489,50C n17 500 Peru, 46 Productivity: in Colombia, 244; Primary defict. See Public sector Ghana's black market and, 353- deficit 54; growth prospects in Primary surplus, 1819, 20-23; in Zimbabwe of, 497-500, 507 n29; Colombia, 257-61; in C6te wage rate and, 307 d'Ivoire, 284, 286, 297-98, 300 Publicconsumption. See Public Private consumption: in Argentina, - spending 158-62; in Chile, 191, 194-96, Public debt. See Debt 197, 210; in Colombia, 244; in Public enterprises, 30-31, 35 Cote d'lvoire, 299-302; Public investment: in Colombia, crowding-out of, 5, 16, 17, 24, 24, 230, 235-36, 252, 253-54; 560 Suect Index Public investment (continued) prmary, 18-20,23,69-70; in in COte d'Ivoire, 274, 284,299- Zimbabwe, 32,458-59, 463,466, 302 305-06; crowdinig-out of 467, 469-74, 504 n2. See also private investment by, 1-2, 16;. Consolidated nonfinancial public tink between private and, 1-2, sector; Fiscal deficits; Fiscal 28; in Morocco, 378, 383, 389- poUcies 93; in Pakistan, 437-42, 451 Public spending: in Argentina, 102, Public revenues. See Tax revenues 104,108; in a8te dIvoire, 287- Public saving, 2; in Chile, 194, 195, 88; methodology for analyzing, 197,219 n18, n19; in Ghana, 331, 84-7, 89-95; in Morocco, 379, 332; in Pakistan, 439; in 399, 401; in Pakistan, 425, 445- Zimbabwe, 486-89, 502, 506 nl5, 46,455 n5 nI8, n20 Public sector. measuring solvency Quasi-fiscal deficits, 19; in of, 67-68; measuring sustainable Argentina, 6, 19; in Chile, 19,178 deficits in, 69-70. See also Public Quasi-money demand, 23-40 investment; Public sector deficit; Public spending Rate of returrL See Real interest rate Public sector deficit: in Argentina, Real exchange rate: Argentina's, 101-02; in Chile, 183; in 1549- Chile's trade balance Colombia, 225,231,262, and, 202-07; in Colombia, 229- descrbed, 79; domestic 30,255-57; C6te d'Ivoire and macroeconomic shocks and effect of inflation on, 289-95; Zimbabwe's, 32, economic deficit devaluation and, 30-3S growth and, 31-32; external dynamc general eqitibrium deficits and, 16-17: exterral model fotr 89-95; fiscal deficits effects of, 149-56; foreign shocks and, 24-25, 60-63, 254-62; fiscal and, 25-27; imnplications of policy and, 290-95; Ghana's domestic financing of, 183-85; trade balance and, 34345, 348- inflation and, 15, 28,121-22, 51; joint sinulation of resource 427-36; inflationary financing of, balance and, 260-62; short-run 116-22-, interest rates and, 16; determination of, 84-7; macroeconomic imbalances theoretical framework for and, 2; measurement of, 18; determining, 149-S54 measuring Argentina's, 108-09, Zimbabwe's deficit and, 492-97 114; measuring Ghana's, 313-19; Real interest rate: deficit financing in Morocco, 374-81, 384, 395, and, 16; determining 4, 403, 407, 408 nl; output Colombia's, 258; under financial shocks and, 31-32; in Pakistan, reform or repression, 50, 66; 413-14,420-21,424-25,426,427- fiscal deficits and, 49, 54; 32; private consumption and sensitivity of deficits to, 29-30; investment and, 16; real Zimbabwe's deficit and, 479, exchange rate and, 30-31; 482-83, 485. See also Interest rates reduced public investment and, Resource balance. See Trade 451; response of private sector balance to, 53-60; as share of cnr, 24; Ricardian equivalence hypothesis: short-run demand for, 185-88; on consumption, 16,54,55, 57, sustainability of Ghana's, 19-20, 64-65, 67, 194, 197, 332, 370 n16, 326-30; sustainabllity of 438,486-7, 489; debt-financed Zimbabwe's, 471-74; sustainable tax reductions and, 79, 82-83; Subject Index 561 empirical evidence on, 16, 219 in Morocco, 379480; in n4; private sector saving and, 5; Zimbabwe, 469, 504 n3 trade balance and, 81 Terms of trade, 25, 26, 55, 86, 87; Seigniorage: Argentina's revenue for Argentina, 151,152, 153, 155; from, 118-19; cFA rules on use of, for Chile, 172, 173, 202, 205, 206, 307 n5; Chile's revenue from, 207; for Colombia, 255-56, 261; 188; in Colombia, 227-28; for Cbte d'Ivoire, 297,300; for Colombia's debt financing and, Ghana, 336,342,344,345,346, 255; Colombia's public 347, 348, 351, 352, 370 n23; for investment and, 253-54; Morocco, 8, 374, 375, 378, 380, Colombia's real interest rate and, 382, 383, 409 n5; for Zimbabwe, 248-49; computing maximum 493,495,496,497 revenue from, 132-35; Thailand, 3, 27, 34,-35, 47, 49, 50, decomposition of, 70; fiscal 59,60.61,62 balances in relation to, 23-24; in lEade balance: Argentina's, 9,154- Ghana, 42, 322-24, 325-26, 329; 56; asset markets and inflation as inflation and, 3 4,17, 38-40, 47- variables of, 8849; Chile's fiscal 49, 245, 248, 325-26; public policies and, 202-07; coffee investment financed by prices as variable of, 270 n2; Colombian, 253-54; spikes in, Colombia's, 256-60; deficit 46-47; steady-state versus one- financing and, 8044; fiscal shot, 40-47- See also Deficit deficitand, 154-56; Ghana's financing fiscal deficit and, 336-46; Seigniorage Laffer curve, 38-40, 271 Ghana's real exchange rate and, Shock-absorber hypothesis, 184 343-45; simulation of real State-owned enterprises (soEs), 30- exchange rate and, 260-62; 31 Zimbabwe's, 493, 495-97. See also Statutory tax base and rate, 522 Trade surplus Structural surplus or deficit: in brade surplus: accumulation of net Colombia, 233,235; in C6te foreign assetz linked to, 206-0; d'lvoire, 278, 284-87 Colombia's, 256-60; C6te Sustainable public sector deficit. d'Ivoire's fiscal policy and, 297- See Public sector deficit 98; debt-financed deficit and, 83; Sustainable public sector surplus, derivation of equation for, 215- 21-22. See also Primary surplus, 16; fiscal deficits link to, 61-3, Structurl surplus or deficit 67; Ghana's, 349-5S1; Zimbabwe's, 495-97, 503 Tax cuts: financed by external debt, 81-2; financed through inflation Value added tax (vAT), 281, 379. See tax, 83; financed by internal also Tax revenues debt, 79, 82-83; private consumption and, 54-55 World Bank: Morocco and, 374; Tax reform, 521-22 warnings to Cate d'Ivoire by, 273 Tax revenues: in Cote d'Ivoire, World Bank research project: 281-84, 304-05, 307 n4; conference on, 5522; country case determinants of Palcistan's, 424, studies of, 6-10, 11 nl; main 455 n4; financing through conclusions of, 1-2, 10-11; neutral, 79; fiscal policies and, methodology for, 2-5,17-18 211-13; in Ghana, 316; Laffer curve measurement of, 271 nll; Zaire, 46 562 Subject Index Zimbabwe: alternative deficit of, 458-59, 462; economic impact financing methods in, 477-79, of deficit financing in, 474-77, 48243; consolidated public 502; fiscal policies in, 33, 34-35; sector deficit of, 33, 463, 466-67; foreign shocks in, 25-27; growth crowding-out of private prospects inr 497-500, 503-04 consumption in, 485-87,489; 507 n29; implications of case crowding-out of private study conclusions for 500-04; investment in, 459, 462, 485-86, inflation and public sector deficit 489, 491-92, 502-03; deficit in, 28; interest rates in, 4; financing in, 474-79,48243,485, macroeconomic shocks in, 32, 502; deficit's effect on trade money demand in, 477-79; balance and real exchange rate private saving in, 5; sustainable in, 492-93, 495-97, 503; public sector deficit in, 471-74; determinants of public sector tax revenues in, 469, 504 n3; deficits in, 467,469-71; domestic terms of trade for, 493, 495, 496, (gross) investment in, 486, 497, 497 502, 506 n14; economic history