11027 Lance Taylor • Edmar L. Bacha• Eliana:Ã. Cardosoe* Frank J. LySy -' 1 4 A Wórld -Bank-mResearchPublitation-, ýl Return to BROC, 18-203 Models of Growth and Distribution for Brazil LANCE TAYLOR EDMAR L. BACHA ELIANA A. CARDOSO FRANK J. LYSY A WORLD BANK RESEARCH PUBLICATION Published for the World Bank Oxford University Press Oxford University Press NEW YORK OXFORD LONDON GLASGOW TORONTO MELBOURNE WELLINGTON HONG KONG TOKYO KUALA LUMPUR SINGAPORE JAKARTA DELHI BOMBAY CALCUTTA MADRAS KARACHI NAIROBI DAR ES SALAAM CAPE TOWN @ 1980 by the International Bank for Reconstruction and Development / The World Bank 1818 H Street, N.W., Washington, D.C. 20433 U.S.A. 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 permission of Oxford University Press. Manufactured in the United States of America. The views and interpretations in this book are the authors' and should not be attributed to the World Bank, to its affiliated organizations, or to any individual acting in their behalf Library of Congress Cataloging in Publication Data Main entry under title: Models of growth and distribution for Brazil. (A World Bank research publication) Includes bibliographies and index. 1. Brazil-Economic conditions-1945- - Addresses, essays, lectures. 2. Income distribution-Brazil- Addresses, essays, lectures. 3. Brazil-Economic policy-Addresses, essays, lectures. I. Taylor, Lance. II. International Bank for Reconstruction and Develop- ment. III. Series: World Bank research publication. HC187.M584 330.981'063 80-13786 ISBN 0-19-520206-6 ISBN 0-19-520207-4 (pbk.) Contents List of Tables vi Preface x Acknowledgments xii 1. Introduction Lance Taylor 3 Political Economy 4 Forced Saving and Growth 6 Forced Saving in the Models for Brazil 8 Toward More Egalitarian Policies? 12 Toward Better Models? 14 References 15 2. Selected Issues in Post-1964 Brazilian Economic Growth Edmar L. Bacha 17 Growth Record 18 Foreign Capital 26 Role of the State 29 Balance of Payments 32 Economic Policymaking after the Oil Crisis 41 Perspectives 44 References 45 3. Theoretical Framework for Identity-Based Planning Eliana A. Cardoso and Lance Taylor 49 A Simple Model 50 A Model for Brazil 60 References 75 4. Brazilian Growth and Distribution in the 1960s: An Identity-Based Postmortem Eliana A. Cardoso 77 Kaldorian Base Growth Paths 78 Effects of Wage Reduction on Growth, Inflation, and Labor Share 82 iii iv Contents Effects of Labor Tax Reduction on Growth and Labor Share 88 Exports and Growth 89 Conclusions 91 Appendix A: Data and Sources 92 Appendix B: Failure of the Neoclassical Specification 94 References 100 5. Planned and Possible Growth in the Late 1970s: Some Identity-Based Complications for Brazil Eliana A. Cardoso and Lance Taylor 102 A K-Specification Growth Path 103 An N-Specification Growth Path 107 Wages and Inflation 109 Growth, Distribution, and the Trade Gap 112 The Second National Development Plan 118 Appendix: Data and Sources 120 References 126 6. The General Equilibrium Income Distribution Model Frank J. Lysy and Lance Taylor 128 A One-Sector Version 129 Sketch of a Solution Algorithm 136 Extension to Many Sectors and Income Recipients 137 References 139 7. Formal Statement of the General Equilibrium Model Frank ]. Lysy and Lance Taylor 140 Cost Functions and Input Coefficients 141 Employment and Income Levels 150 Sectoral Consumption Functions 153 Investment Functions 156 Commodity Balances 157 National Accounting 158 Closing the Model 161 Appendix: Solving the General Equilibrium Model 162 References 171 8. Data for the General Equilibrium Model and a Base Solution Frank ]. Lysy and Lance Taylor 173 Interindustry Flow Data 174 Employment and Labor Payments 183 Payments to Capital 188 Distribution of Capital Incomes 191 Income Levels by Consumer Class 192 Cost Function Parameters and Rates of Technical Change 196 Summary Description of the Base-Year Data 198 CONTENTS V Data for Simulations over Time 204 Results from the Base Solution of the Model 209 References 221 9. Income Distribution Simulations, 1959-71 Frank]. Lysy and Lance Taylor 224 Tests of Model Sensitivity to Parameter Changes 227 Shifts in Investment Demand Levels 238 Changes in Government Expenditure Policies 247 Modifying Tax Rates and the Foreign Exchange Rate 255 Changes in Labor Supplies by Education Type 265 Other Redistributive Policies 270 Export Effects on Employment 284 Summary and Conclusions 288 References 295 10. Brazilian Income Distribution in the 1960s: "Facts," Model Results, and the Controversy Edmar L. Bacha and Lance Taylor 296 Six Hypotheses of Brazilian Income Distribution 297 Wrong Measurement Dismissed 302 Kuznets Effects 306 Skill Differentials 309 Wage Squeeze 314 Profits and Wage Spread 329 Appendix: Brazilian Distributional Data 336 References 337 Index 343 Figures 2-1. Gap between Potential and Actual Output, 1947-77 20 2-2. Ratio to Exports of C.i.f.-Trade Deficits, 1959-77 35 4-1. Inflation Rate and the Labor Share 83 5-1. The Tradeoff between Trade Deficit and Employment Levels in the K Specification 113 8-1. Growth Rates of Sectoral Output: Actual and Calculated by the Model 218 9-1. Constant Elasticity of Substitution (CES) Isoquants 233 Tables 2-1. Gap between Potential and Actual Output, 1947-77 19 2-2. Vargas Foundation Capacity Utilization Indexes for Manufacturing Industry, 1968-77 21 2-3. Minimum Wages and Cost of Living in Rio de Janeiro, 1963-77 23 2.4. Patterns of Asset Ownership of the 5,113 Largest Nonfinancial Enterprises, 1974 27 2-5. Ratio to Exports of C.i.f.-Trade Deficits, 1959-77 34 2-6. Foreign Capital: Demand and Supply, 1970-77 36 2-7. External Indebtedness and International Reserves, 1969-77 38 2-8. Wharton and Banco Garantia Forecasts for Net External Debt and Significant Debt Ratios 41 3-1. Symbols in the GNP Model 51 3-2. Level Forn of the GNP Model 51 3-3. Log-Change Form of the GNP Model 52 3-4. Symbols in the Brazil Model 62 3-5. Level Form of the Brazil Model 63 3-6. Log-Change Form of the Brazil Model 67 4-1. Value of Key Variables in Kaldorian (KS) Base Solutions 79 4-2. Inflation Rate and Labor Share for Different Employment and Wage Growth 84 4-3. Effects of Wage Reduction on Growth, Inflation, and Labor Share 86 4-4. Effects of Labor Tax Reductions on Growth and Labor Share 88 4-5. Effects of Export Reduction on Growth and Labor Share 90 4-6. Values of Exogenous Growth Rates in the Base Solutions: Kaldorian Path A and Neoclassical Path a 95 vi TABLES Vii 4-7. Comparison of Observed Macrovariables and Model Results: Kaldorian Path A and Neoclassical Path a 96 4-8. Values of Key Variables in NS Base Solutions 98 4-9. Comparison of Observed Macrovariables in 1966-69 and Model Results When Growth Rate of Money Wages (w') = 0.3 99 5-1. Values of Key Variables in Base Solutions with 3 Percent Annual Employment Growth 104 5-2. Model Responses to Different Wage Growth Rates 110 5-3. Model Responses to Enforced Import Substitution with 3 Percent Annual Employment Growth 116 5-4. Comparison of Observed 1973 Macrovariables and Model Results 124 6-1. Symbols in the One-Sector Model 130 7-1. Symbols in the Multisector Model 142 7-2. Jacobian of Excess Labor Supplies, 1971, Base Run 168 7-3. Matrix of Elasticities of Excess Labor Demands with Respect to Wage Changes 169 8-1. Interindustry Accounting for 1959 176 8-2. Consumption Parameters for the Four Consumer Classes 184 8-3. Tax Rates and Receipts, 1959 186 8-4. Employment and Labor Payments in the Base Year 188 8-5. Makeup of the Aggregate Capital Stock by Sector and Depreciation Rates 190 8-6. Capital Cost Data for 1959 192 8-7. Distribution of Income from Profits 193 8-8. Aggregation of Income Recipients to Consumer Classes 194 8-9. Elasticities of Substitution and Rates of Technical Change 197 8-10. National Accounts for 1959 200 8-11. Size Distribution of Income for the Economically Active Population: A Comparison of Base-Year Data and Fishlow's Estimates 203 8-12. Government Purchases and Exports 206 8-13. Labor Supplies and Government Employment 207 8-14. Animal Spirits Parameters by Sector 208 8-15. Output, Price, Expenditure, and Wage Variations: Base Solution 210 8-16. Functional Income Distribution Measures: Base Solution, Shares of GNP 211 8-17. Saving, Taxes, and Size Distribution Measures: Base Solution 212 8-18. Wage and Employment Structure: Base Solution 214 8-19. Changes in Output, 1959-71, by Sector 216 8-20. Changes in Economically Active Population (EAP), 1959-71, by Sector 217 viii Tables 8-21. Producers' Prices and After-tax Profit Rates, by Sector 220 8-22. Aggregate Distributional Data 221 9-1. Catalog of Solutions to the General Equilibrium Model 225 9-2. SEN Solutions-Output, Price, Expenditure, and Wage Variations: 1971, Wages Variable 228 9-3. SEN Solutions-Functional Income Distribution Measures: 1971, Wages Variable 229 9-4. SEN Solutions-Saving, Taxes, and Size Distribution Measures: 1971, Wages Variable 230 9-5. Labor Demand in SEN 1 as a Proportion of Labor Demand in the Base Solution 232 9-6. Real Consumption in 1959 Prices in SEN 1 and the Base Solution 234 9-7. INV Solutions-Output, Price, Expenditure, and Wage Variations: 1971, Wages Variable 239 9-8. INV Solutions-Functional Income Distribution Measures: 1971, Wages Variable 240 9-9. INV Solutions-Saving, Taxes, and Size Distribution Measures: 1971, Wages Variable 241 9-10. GOV and TRANs Solutions-Output, Price, Expenditure, and Wage Variations: 1971, Wages Variable 246 9-11. Gov and TRANS Solutions-Functional Income Distribution Measures: 1971, Wages Variable 248 9-12. Gov and TRANS Solutions-Saving, Taxes, and Size Distribution Measures: 1971, Wages Variable 250 9-13. TAx and RATE Solutions-Output, Price, Expenditure, and Wage Variations: 1971, Wages Variable 256 9-14. TAx and RATE Solutions-Functional Income Distribution Measures: 1971, Wages Variable 260 9-15. TAx and RATE Solutions-Saving, Taxes, and Size Distribution Measures: 1971, Wages Variable 262 9-16. EDUC Solutions-Output, Price, Expenditure, and Wage Variations: 1971, Wages Variable 266 9-17. EDUC Solutions-Functional Income Distribution Measures: 1971, Wages Variable 267 9-18. EDUC Solutions-Saving, Taxes, and Size Distribution Measures: 1971, Wages Variable 268 9-19. DIST 1-3 Solutions-Output, Price, Expenditure, and Wage Variations: 1971, Wages Variable 271 9-20. DIST 1-3 Solutions-Functional Income Distribution Measures: 1971, Wages Variable 272 9-21. DIST 1-3 Solutions-Saving, Taxes, and Size Distribution Measures: 1971, Wages Variable 273 9-22. Changes from the Base Levels of Output, by Sector, as a Result of Income Redistribution in DIST 2 276 9-23. DIST 3 Employment as a Proportion of Base-run Employment 278 TABLES ix 9-24. DIST 3 Producers' Prices and Profit Rates as a Proportion of Their Base-run Values 281 9-25. Base-run Values of Wages in 1968 and 1971 as a Proportion of 1965 Wages 283 9-26. DIST 4 Employment Demands in 1968 and 1971 as a Proportion of 1965 Demands 283 9-27. DIST 4 and EXP 1 Solutions-Output, Price, Expenditure, and Employment Variations: 1971, Wages Fixed 285 9-28. DIST 4 and EXP 1 Solutions-Functional Income Distribution Measures: 1971, Wages Fixed 286 9-29. DIST 4 and EXP 1 Solutions-Saving, Taxes, and Size Distribution Measures: 1971, Wages Fixed 287 9-30. Employment Elasticities from Changing Export Demands 288 9-31. Summary Statistics: Ratios of Variable Wage Solutions to Base Solution, 1971 290 9-32. Summary Statistics: Ratios of Fixed Wage Solutions to Base Solution, 1971 292 10-1. Comparison of Income Distribution by Income Deciles, 1960 and 1970 305 10-2. Structural Characteristics of the Income-earning Population, 1960 and 1970 308 10-3. Decomposition of the Change in the Variance of Logs of Incomes, 1960 and 1970 310 10-4. Comparison of Income Distribution by Educational Level for the Income-earning Population, 1960 and 1970 311 10-5. Guideline Indexes of Real Wages, 1957-74 316 10-6. Share of Workers in the Industrial Sector with Monthly Earnings in the Neighborhood of the Highest Minimum Wage, 1965-71 319 10-7. Behavior of Workers' and Managerial Earnings in a Group of Large Industrial Firms, 1966-72 320 10-8. Brazil: Rural Wages Outside Sao Paulo and Highest Minimum Urban Wage, 1966-73 321 10-9. Sao Paulo: Rural Wages and Minimum Urban Wages, 1962-74 321 10-10. Rio de Janeiro City Manufacturing Sector: Results of Regressions of Median Wages on Selected Variables, 1952-73 322 10-11. Sao Paulo and Belo Horizonte: Infant Mortality and Minimum Wages, 1952-73 324 10-12. Concentration Ratios in Manufacturing Industry and Explanatory Variables, 1966-72 330 10-13. Manufacturing Industry Gini Coefficients, 1966-72: Regression Coefficients and Standard Errors 331 10-14. Breakdown of Gross Value Added in Manufacturing, 1939-72 332 Preface THE RELATION BETWEEN GROWTH and the distribution of its bene- fits has become a major concern of economists and policymakers in recent years. Simon Kuznets was the first to observe from cross-country data that the distribution of income shows an initial tendency to worsen among countries with increasing income, but subsequently to improve at still higher income levels. This em- pirical finding has been confirmed by later writers, although there is much variation among individual countries. The World Bank has conducted research studies and supported others (of which the present volume is an example) to investigate the extent to which observed changes in the distribution of in- come during the process of growth can be attributed to the policy regime in the country concerned, as opposed to being an inevi- table consequence of growth. Such studies have required the development of new tools of economic analysis, specifically the endogenous-price multisectoral model, or computable general equilibrium model. Such a model is described and used in this volume, which contributes as much to the methodology of eco- nomic analysis as to the understanding of the distributional con- sequences of rapid economic growth in Brazil. A major finding of this work by Professors Taylor, Bacha, Car- doso, and Lysy is that public policy contributed to the deteriora- tion in the distribution of income over the period studied. Wage policy (and particularly holding down the level of the minimum wage) and the concentration of educational expenditures at the tertiary level are highlighted. This finding stands in contrast to the conclusion drawn by Irma Adelman and Sherman Robinson x PREFACE xi in their study of Korea (Income Distribution Policy in the Devel- oping Countries [Stanford, Calif: Stanford University Press, 1977]), which was also supported by the World Bank as part of its program of research on growth and distribution. Adelman and Robinson concluded that the overall size distribution of income was not greatly affected by particular policy interventions, nor even by combinations of policies. The extent to which these differences in main findings are due to variations in the country background, in the extent and inten- sity of policy interventions postulated, or in modeling ap- proaches is not yet clear. Korea, in contrast to Brazil, is a small country with a relatively egalitarian distribution of income, so that the scope for further policy intervention may be limited. The macroeconomic specification of the model employed for the two countries also is different. The Adelman-Robinson study em- ployed an essentially neoclassical model, whereas Taylor and his co-workers experimented with what have become known as dif- ferent closing rules. They demonstrate that how macroeconomic equilibrium is obtained between supply and demand, whether with neoclassical or neo-Keynesian closing rules, can lead to profound differences in the macroeconomic story. The objectives of the present study and the broad approach taken were discussed at a conference in Bellagio in 1973, as part of an overall assessment of the Bank's program of research on distribution and country modeling. The execution of the study and the conclusions drawn from it, however, were the responsi- bility of the authors. The volume makes an important contribu- tion both to the practice of macroeconomic modeling and to the understanding of the evolution of the Brazilian economy over a period of rapid economic growth. JOHN H. DULOY . Director, Development Research Center The World Bank Acknowledgments THIS BOOK HAD ITS ORIGINS in a collaborative research project begun in the early 1970s to investigate prospects for growth and income distribution in Brazil through the use of computable planning models and political economic analysis. In addition to the four authors listed on the title page, Francisco Lopes, Dioisio Carneiro, and Pedro Malan made substantial contributions to the research. Sympathetic commentary over the years by Hollis Chenery, Albert Fishlow, Roger Norton, Jack Duloy, Peter Clark, and Rudolfo Hoffman is gratefully acknowledged. Most of the results described herein were complete by 1977, but various delays postponed publication of the book until 1980. The work was undertaken at several places, mostly the Uni- versity of Brasilia, Harvard University, Massachusetts Institute of Technology, Stanford University, and the World Bank. We are grateful to all these institutions for their support, and especially to the Bank for providing the bulk of the finance necessary. Jane Carroll edited the manuscript for publication. Raphael Blow prepared the charts, Brian J. Svikhart supervised produc- tion of the book, and Ralph Ward and James Silvan indexed the text. LANCE TAYLOR Xui Models of Growth and Distribution for Brazil  1 Introduction Lance Taylor THE EXPERIENCE OF BRAzIL in the last fifteen or twenty years provides a classic case for the study of economic change. Follow- ing several years of economic stagnation and political unrest, the military coup of 1964 inaugurated first a period of rigorous eco- nomic stabilization and then almost a decade of very rapid growth, accompanied by a marked increase in income inequality. Rapid expansion in exports and in the openness of the economy occurred simultaneously with increases in the depth and so- phistication of financial intermediation. The functional income distribution apparently shifted toward capital, and at the same time the role of state enterprises increased enormously. An unprecedented marriage of private capitalism and state interven- tion produced a growth spurt unmatched in recent Latin Ameri- can history. Regrettably, the pattern of income distribution was such that much of the population did not benefit from this growth. The papers in this volume explore the Brazilian experience from the point of view of political economy, using computable general equilibrium income distribution models. The investiga- tion has focused on the interactions of growth and income distri- bution, though the length of the ongoing research project, which began in 1972, has allowed time for both intellectual enrichment and change. The purpose of this volume is to put together the 3 4 Introduction and Summary authors' current (early 1978) ideas about the Brazilian economic model' and to point to fruitful lines of future research. This chapter provides an outline of the book to ease the reader's task in following our own evolving ideas about economic change in Brazil. Political Economy Our observations on political economy are concentrated in chapters 2 and 10. The first of these is an interpretative review of aggregate Brazilian growth since 1964. Chapter 10 covers the extensive debate about the causes of the deterioration in Brazil- ian income equity that has been observed since 1960. The results of models discussed in other chapters and new statistical evi- dence are used to weigh the merits of the various explanations of distributional change that have been proposed. The major points made in the second chapter are: * The 1968-74 boom in Brazil was not an economic "miracle" but conformed rather closely to the cyclical growth pattern of the economy since World War II. * The success of inflation control policies since 1964 can be attributed to the massive wage squeeze applied after 1964, espe- cially during the first year of the military regime. Simulations with a macroeconomic model in chapter 4 show quite clearly that while more rapid growth of the money wage would have added to inflation, it would also have improved the functional income distribution. * Foreign capital played an important dynamic role in the economic upsurge of the late 1960s but also was responsible for some of the most pressing problems of the period. * Government intervention in the economy increased, though the "inflationary" cash deficit declined. Recent research by Car- doso (1978) shows that wage control was a key factor in permit- ting this more positive fiscal role. * The balance of payments during the 1968-73 upswing was not sound. In effect, the 1974 international oil crisis only hastened the time when Brazil could not maintain the potential 1. Other research results following directly from the project are reported by Bacha (1976, 1978, and 1979), Cardoso (1977 and 1978), Cardoso and Taylor (1979), Taylor and Bacha (1976), and Taylor and Lysy (1979). 6 Introduction and Summary level education may have led to slower growth because of the high cost of training university graduates. * Other papers, written by Americans but widely circulated by official agencies in Brazil, argue that on relevant welfare scales income distribution may in fact have improved during the 1960s. It is shown that these approaches misinterpret the available data. * Our own view, already mentioned, is that wage control played a substantial role in income concentration, and this per- spective is argued more fully in chapter 10. * The wage squeeze hypothesis also fits naturally with the idea that institutional factors, including a more complex hierar- chical structure in management and rapid expansion of govern- ment employment in "supergrades" contributed to the spread in wages. Real wages at the bottom of the earnings scale may have been held down by surplus labor, while in a rapidly growing economy the lid was off at the top. Institutional evidence sup- porting this explanation is provided. * The last two hypotheses complement the macroeconomic observation that "forced saving"-shifts of the income distribu- tion away from low-saving classes during periods of accelerating investment and economic growth-must have played a role in shaping Brazilian income distribution during the boom. The pro- cesses of forced saving are built into the formal models described in this book, and they may partially explain the Brazilian experi- ence, as discussed in the following sections. Forced Saving and Growth In any economy, saving must equal investment in terms of cur- rent prices for macroeconomic equilibrium to exist. The question at hand is, What adjustments within the system permit this iden- tity to be observed? Sen (1963) was perhaps the first to point out explicitly that expected patterns of growth and distribution may depend essentially on how a descriptive model of the economy is made algebraically determinate, or closed. Following Sen and others, we can distinguish two closure rules that have been im- portant in the literature: the neoclassical and the neo-Keynesian. The main neoclassical assumption is that something approxi- mating full employment of factors of production is maintained by freely varying prices. Real wages or rents paid to factors are determined by their marginal productivity, so that with full em- FORCED SAVING AND GROWTH 7 ployment all earnings flows are determined from the side of supply. With earnings given, the total amount saved is deter- mined by behavioral parameters, and investment is assumed to adjust to saving. Expressed with more brevity, the neoclassical position is that "productivity and thrift" determine income distri- bution and investment in the short run, and the growth pattern of the economy in the long run. According to the neo-Keynesian position, investment is deter- mined by expectations, nonmarket considerations, and possibly the rate of interest. It is certainly not affected by the supply of saving generated by marginally determined wages and profit rates in the short run. Rather, factor payments and employment levels adjust to permit the supply of saving to equal investment demand. In particular, if real investment rises, the income distri- bution will shift toward economic agents with higher savings propensities to permit the necessary finance to appear. The extra saving is in some sense unintended and is called "forced" for that reason. As already mentioned, processes of forced saving are built into our Brazilian models in various ways. A verbal example of the process will illustrate what happens.3 Consider an economy in which labor and capital can substitute freely along a well-behaved production function and in which raw material imports are also required to support production. The cost of imports and the money wage are assumed to be fixed, and some labor is unemployed. If investment demand rises, output will go up as more workers find jobs. If exports are fixed in the short run, the balance of payments deficit will become larger, since a higher production level requires more raw material im- ports. The rate of profit (marginal product of capital) will increase since there are decreasing returns to the existing stock of capital in the short run. The, real wage, however, will fall, since higher profits will be passed along in a higher price of output and the money wage is fixed. Now look at what occurs in terms of saving and investment. Real investment demand has gone up, and the current value of investment has gone up even more because of the price in- creases. The extra saving comes from increases in real profits that are partly hoarded and from the increased balance of payments 3. The following example is worked out in algebraic form in Taylor and Lysy (1979). 8 Introduction and Summary deficit, or "foreign saving." But real wages have fallen, and the real wage bill could have gone either up or down, depending on the elasticity of substitution. The income distribution has shifted to favor national profit recipients and foreigners over workers. Forced saving at the expense of labor has occurred. Forced Saving in the Models for Brazil The macroeconomics of the computable models for Brazil de- scribed in chapters 3 through 9 can be interpreted along the saving-investment lines just sketched out. The first three of these chapters deal with a one-sector model designed to fit available Brazilian national accounts information, while chapters 6 through 9 describe a much more ambitious multisectoral simula- tion. In the second exercise, the macro effects of forced saving are complemented by microeconomic stories involving shifts in re- source allocation and consumption demand among different pro- ducing sectors and economic groups. The macroeconomic model is based on national income ac- counting identities, with variant specifications closing them along neoclassical and neo-Keynesian lines similar (but not identical) to those described above. In the neoclassical closure, investment varies to meet available saving, the sum of the gov- ernment current surplus, the balance of payments deficit, and savings flows from capital and labor incomes. These are deter- mined by factor supplies and by profit and wage rates that follow from standard conditions of marginal productivity. In the neo- Keynesian closure, investment is taken as exogenous, and the marginal productivity conditions are dropped so that the relative returns of labor and capital can vary freely to permit forced saving to equal investment. In both closures, the money wage rate and labor employment are set exogenously (though varied para- metrically), and the price level is determined from the sum of costs of labor, capital, and raw material imports.4 4. The growth of money supply is determined within the model by the govern- ment deficit and foreign trade surplus, and there is a weak feedback from varying prices to consumer demand through the interest rate. Prices are mainly deter- mined by costs, however, and not by the money market. This treatment corre- sponds to most diagnoses of Brazilian inflation in the 1960s (see chapter 2), and its implicit assumption of endogenicity or passivity of monetary policy is sup- ported by the econometric causality tests reported by Cardoso (1977). FORCED SAVING IN THE MODELS 9 The one-sector model is used to simulate Brazilian historical experience in the 1960s (chapter 4) and to make conditional fore- casts for the 1970s (chapter 5). The first main conclusion in chapter 4 is that with plausible values for elasticities of substitu- tion and other parameters, the neoclassically closed model can- not be made to fit Brazilian 1960-70 data,'5 largely because the volume of real investment generated endogenously by the model is far too low. In part because its investment levels are set exoge- nously, the Keynesian model can be used to replicate the past. In this closure, it is shown that the fall in the real legal minimum wage had a major role in shifting the income distribution away from labor and in slowing inflation. Similarly, rapid export growth may have led to deterioration in the functional income distribution by demanding more extreme savings efforts via infla- tion and a fall in real wages. Chapter 5 extends these results to conditional forecasts for the 1970s. In both versions of the model more rapid growth of the money wage would lead to greater inflation, a higher labor share, and possibly higher output growth as well. The appropriate tradeoffs are also mapped out. The distributional effects of import substitution and export promotion policies are then analyzed, and it is shown that in the neo-Keynesian model there is a very costly tradeoff between improvements in the balance of pay- ments and the labor share. This problem disappears in the neo- classical model, where substitution of domestic resources for im- ports permits accelerated depreciation of the cruzeiro to close the payments gap. Which model more accurately captures the pres- ent situation is a matter of judgment, but our own is that the neoclassical substitution assumptions are probably far too opti- mistic. Chapters 6 through 9 present a much more detailed approxima- tion to the Brazilian economy in the form of a multisector, multilevel-of-skill general equilibrium model of the functional income distribution. The theoretical structure is described in chapters 6 and 7, the data base in chapter 8, and a number of simulation exercises in chapter 9.6 5. The simulations for the 1960s are based on an old set of Brazilian national accounts which have since been superseded. The results for the 1970s are based on the new accounts. 6. In more detail, the model has twenty-five input-output sectors and dis- tinguishes six levels of labor skills, largely by level of education. Capital stocks 10 Introduction and Summary Given the complexity of the model, it is useful to think about its solutions at two levels. When the model is solved under varia- tions in parameters or exogenous variables, there is first a mac- roeconomic adjustment to bring saving in line with investment (or, what amounts to the same thing, to make aggregate supply equal aggregate demand). The mechanism is based on changes in the values of saving and investment resulting from movements both in relative prices and in the overall price level in relation to payments fixed in nominal terms-the exchange rate, certain gov- ernment taxes and transfers, and remunerations of proprietors or entrepreneurs associated with production in certain sectors (especially small owner-operators in agriculture and the ser- vices). When aggregate demand falls, for example, all prices fall, and the income distribution shifts in favor of the proprietors and recipients of transfer payments. Government nominal expen- ditures decline (its purchases are fixed in real terms), but so do tax revenues and certain subsidies; the impact of price deflation on government saving is complex and of uncertain sign. The balance of payments deficit in domestic currency may also re- spond either negatively or positively, depending on import changes and relative price variation. Finally, with a fall in prices the value of a given real investment demand vector will fall, and the sum of saving from all sources must adjust accordingly. The usual result is that deflation improves the income distribution, calculated from all payments to labor (four to six skill types in each sector), capital income recipients by sector, and proprietors. This macro savings-investment adjustment is accompanied by numerous microeconomic changes in the equilibrium solu- are fixed in the short run in each sector, so that in addition to labor payments by skill types and sector, twenty-five quasi-rent flows are distinguished in the func- tional distribution. Payments to and employment of both labor and capital are determined by marginal productivity rules, assuming constant elasticity of substi- tution production functions in the sectors. The 120-odd income flows in the model are treated separately in calculating income distribution, but they are aggregated to four household classes by income and sector specialization in arriv- ing at final consumer demand patterns via direct addilog complete expenditure systems. The model is completely closed with respect to saving and investment, with the former coming from the government surplus, balance of payments defi- cit, corporate retained earnings, depreciation allowances, and households. The specification is relatively complex, but of a type becoming more common in the literature. For other examples, see Adelman and Robinson (1977) and McCarthy and Taylor (1980). FORCED SAVING IN THE MODELS 11 tion-that is, shifts in employment levels by sector and consumer class, in costs and prices of particular sectors, in profit rates, tax revenues, and so on. A large number of such results are presented in chapter 9. The more interesting ones follow: * In sensitivity analyses, it appears that the elasticity of labor supply from sectors such as agriculture and tertiary activities with surplus labor strongly affects economic equilibrium. More inelastic labor demand curves in these sectors lead to reduced growth and shift the income distribution toward capital. Among consumers, more diverse income elasticities of demand (a strong taste for luxuries by the rich, for example) lead to greater income inequality. Both results are consistent with the "structuralist" diagnosis of Latin American development problems and with the focus of formal development models such as that of Taylor and Bacha (1976). * Under the model's pricing assumptions, changes in labor costs are passed along into higher prices, and the wage structure is kept stable by assumed high elasticities of substitution among skill types. Under such circumstances, increases in employment taxes in certain sectors, for example, are largely passed along into higher market prices and do not affect employment through neo- classical factor substitution. The employment shifts that are ob- served really come from consumer responsiveness to higher prices in the taxed sectors and are not large. Increases in profit taxes, on the other hand, are borne by recipients of profit incomes and improve the income distribution. * Redirection of transfer payments from richer to poorer seg- ments of the population does not affect relative prices unfavor- ably and thus is equalizing. Resulting changes in consumer de- mand from income effects substantially modify the mix of sectoral outputs, suggesting once again that divergence in de- mand patterns between rich and poor plays a major role in condi- tioning income inequality. * Upgrading the educational attainment of any group in the labor force is equalizing, since it leads to an increase in aggregate supply, which is spread toward poorer people. Because of the high elasticities of substitution among types of labor skill, up- grading does not lead to large shifts in relative wages, but the shares of the affected groups in national income change mark- edly. Expanded education of low-skill workers would be both more equalizing and less costly in terms of gross national product (GNP) than would an expansion of the higher educational system. 12 Introduction and Summary * For all skill types, employment elasticities with respect to overall export increases are about 0.1. Low-skill groups benefit more from expansion of agricultural or food exports, however, while those with high skills benefit from export of the relatively high-technology products that the government has favored. In general, the summary measures of income inequality such as the Gini coefficient do not show large improvements-a few percentage points at most-as a result of equalizing policies such as those described above. A combination of policies, however, would in part be additive and might lead to fairly noticeable improvements. At least according to the model, income inequal- ity can be lessened by policy. The problem is that many Brazilian governmental actions seem to have taken the other direction. Toward More Egalitarian Policies? The general view we arrive at from the numerical results, the available data, and less easily formalized considerations of politi- cal economy, is that the "model" of growth adopted by Brazilian economic technocrats under the military regime has led to dete- rioration in relative income equity. The income distribution could be made more equal through a variety of policies. Specifi- cally, we can draw on the results just summarized above to make several observations. * From the one-sector model, it appears that a decline in money wages reduces the labor share and favors profit incomes; it may even retard the rate of economic growth. Less rigorous control of money wages might therefore benefit poorer classes, though at the cost of a higher rate of inflation. The crucial govern- ment policy variable here is the minimum wage, but it is argued on institutional and econometric grounds in chapter 10 that the entire wage structure is sensitive to the minimum, so that the government does have the ability through wage policy to help or harm a substantial proportion of the working population. * Within the multisectoral framework, it appears that reduc- tion of structural divergences in the economy (such as bottle- necks in the labor supply or highly diverse income elasticities of demand for different products) would lead to increased income equality. One obvious policy (though it does not fit neatly with the neoclassical production and consumption economics of the MORE EGALITARIAN POLICIES? 13 general equilibrium model) would be control of the expansion of sales of "modern" or "luxury" commodities by multinational and other firms operating in Brazil. Some of the political and institu- tional difficulties implicit in such a strategy are discussed in chapter 2. * Well-designed tax and transfer policies could substantially improve the income distribution. Results from both the small and large models suggest that reductions in taxes on wages would not act as any particular spur to employment, but that shifting from taxes on wages to taxes on profits would lead to increased income equality (with the proviso that real investment is maintained). Redirection of transfer payments toward poorer segments of the population would clearly reduce inequality and shift production patterns away from luxury commodities and employment pat- terns away from the household-servant sector. Whether such shifts would be institutionally feasible given existing patterns of control of Brazilian industry is a moot point. * Redistribution of assets toward poor people (for example, in agriculture) would improve their income position, and general equilibrium secondary effects would be neutral or equalizing in distributional terms. Again, the pattern of consumer demand would shift away from luxuries and toward wage goods. * Stimulation of investment to break bottlenecks in sectors producing capital goods and intermediate goods would help equalize relative price shifts, so long as the general equilibrium model's competitive pricing assumptions are broadly applicable to Brazil. * As mentioned above, education to develop labor skills would lead to more rapid growth of output and income equaliza- tion. But both growth and equity would benefit more from educa- tional programs directed toward the illiterate and people with low educational attainments than from support of universities and other institutions training the highly skilled. * Consideration of the employment-generating effects of ex- ports from different sectors suggests that policies stimulating sales abroad of relatively high-technology goods may be mis- placed. On the basis of the model, a collection of policies of this type would lead to observable but not radical improvements in distri- butional equality, but such limited results are in the nature of the analysis conducted here. General equilibrium economic compu- 14 Introduction and Summary tations may provide rough guidelines for taking small but visible equalizing steps. They say very little about the design of larger political and economic changes in Brazil. The relevant political question is, of course, whether the ruling Brazilian model would permit policy measures to bring about even a small amount of economic equalization. In chapters 2 and 10 it is argued that the dominant forces in the post-1964 regime-large government enterprise and multinational manage- ment, the bureaucracy, and the 10 to 20 percent of the Brazilian population that has reaped the fruits of the boom-had scant interest in income redistribution. It may be that this coalition is weakening, and more active redistribution policies will be pos- sible. But this is a topic of investigation beyond the scope of the present volume. Toward Better Models? Has anything been learned that can help future model builders? It seems clear from the results here and in Adelman and Robin- son (1977) that some sort of forced saving mechanism holds the key to macro adjustment in general equilibrium models in which saving must equal investment and in which output responds to aggregate demand. Income gainers and losers under any policy change can be traced by asking how the payments flows they receive will adjust to an assumed savings-investment disequilib- rium. Two sets of considerations constrain these responses. First, how well can specific groups maintain the real value of their payments flows as prices change? Such a question leads into issues of economic power and coercion which economists have ignored for decades. Second, insofar as relative income gainers and losers are determined by differences in small and hard-to- estimate parameters such as savings rates and Engel elasticities, detailed projections of the effects of policy changes on the func- tional distribution are beyond the realm of feasibility. A purely macro analysis with a model based on available data can, however, give substantial insight, as the results from chapters 3 through 5 demonstrate. Under the circumstances, my own judgment is that a large model such as that in chapters 6 through 9 adds little to what one can learn from macro analysis based on a relevant class-based REFERENCES 15 categorization of the functional income distribution, and from detailed partial equilibrium analysis of important areas of the economy such as education.' Useful research programs could be built around applied studies in that direction. Other efforts could be devoted to socioeconomic processes that cumulate to lead to large distributional shifts over time. The Taylor and Bacha (1976) model of Belindia is a first step along such lines, but there are undoubtedly many other roads for alert, politically observant economists to explore. References Adelman, Irma, and Sherman Robinson. 1977. Income distribu- tion policies in developing countries: A case study of Korea. Stanford: Stanford University Press. Bacha, Edmar L. 1976. Os mitos de uma dcada: Ensaios de economia brasileira. Rio de Janeiro: Paz e Terra. _. 1978. Brazil's balance of payments before and after the oil crisis. Paper prepared for the UNCTAD/UNDP project on the bal- ance of payments adjustment process in developing countries, July 1976. Brasilia: University of Brasilia; processed. _. 1979. Notes on the Brazilian experience with minideval- uations, 1968-76. Paper presented at the Ford Founda- tion/Central Bank of Barbados seminar on developing coun- tries and the international financial system, Barbados, January 11-14, 1978. Journal of development economics 6(Decem- ber):463-82. Cardoso, Eliana A. 1977. Moeda, renda e inflapdo: Algumas evid&ncias da economia brasileira. Pesquisa e planejamento econbmico 7(2):423-34. _. 1978. Growth and real wages: Modelling the Brazilian economic miracle. Cambridge: Massachusetts Institute of Technology, Department of Economics; processed. 7. For example, the results on shifts in taxes and transfer payments in the large Brazilian general equilibrium model depend essentially on its macroeconomic specifications, and the direction of the results is certainly not affected by sectoral detail. In contrast, the results on the equalizing effects of education do not have to be formally derived in a general equilibrium framework, though they should be interpreted in light of macro balance. 16 Introduction and Summary Cardoso, Eliana A., and Lance Taylor. 1979. Identity-based planning of prices and quantities: Cambridge and neoclassical models for Brazil. Journal of policy modelling 1(Janu- ary):83-111. Kuznets, Simon S. 1955. Economic growth and income inequal- ity. American economic review 45(March): 1-28. McCarthy, F. Desmond, and Lance Taylor. 1980. Macro food policy planning: A general equilibrium model for Pakistan. Review of economics and statistics 62(February):107-21. Sen, Amartya K. 1963. Neoclassical and neo-Keynesian theories of distribution. Economic record 39:53-64. Taylor, Lance, and Edmar L. Bacha. 1976. The unequalizing spiral: A first growth model for Belindia. Quarterly journal of economics 90:197-218. Taylor, Lance, and Frank J. Lysy. 1979. Vanishing income re- distribution: Keynesian clues about model surprises in the short run. Journal of development economics 6(March): 11-30. 2 Selected Issues in Post-1964 Brazilian Economic Growth Edmar L. Bacha THIS CHAPTER ADDRESSES SELECTED ISSUES of Brazilian economic growth under the post-1964 military regime.' The aggregate growth record is summarized in the first section, using the con- cepts of potential GDP and GDP gap. Inflation control policies are also discussed in this section, with emphasis on the roles of the wage control program, monetary policy, and mechanisms for monetary correction. After a review of the role played by foreign capital, the pattern of state intervention in the economy is ana- lyzed. A discussion of the balance of payments situation follows, with stress on the import-intensive nature of the post-1964 growth pattern. The section on macroeconomic policymaking since the oil crisis shows the difficulties that Brazilian policy- makers felt in trying to satisfy multiple objectives in a context where the limits of the economically possible became much nar- 1. Parts of this chapter are drawn from Bacha (1977 and 1978). Readers not familiar with the economy of Brazil may benefit from reading Furtado (1963) and Baer (1965) for historical background; and Fishlow (1973 and 1974), Baer (1973), and Malan and Bonelli (1977) for complementary views on post-1964 Brazilian economic growth. A discussion of income distribution issues is postponed to chapter 10. 17 18 Post-1964 Economic Growth rower than before. Factors affecting the short- and medium-term perspectives for the economy are summarized in the last section. Growth Record The available figures for real aggregate output in Brazil provide a picture of an economy vigorously expanding at an average 7.5 percent potential growth rate in the post-World War II period.2 Actual growth follows a cyclical pattern, with 1952-56, 1961-67, and the period from 1974 on characterized by a slowdown of growth rates and 1947-52, 1956-61, and 1967-74 characterized by rapid economic upswings. The cycles peak in 1952, 1961, and 1974. Following a common procedure, we obtain a potential out- put path by connecting the peak output levels with an exponen- tial growth rate line (see table 2-1 and figure 2-1). The growth rates of potential output accelerate slightly: they are 7.2, 7.4, and 7.7 percent a year, respectively, in 1947-52, 1952-64, and 1964-74. This pattern is in line with a historical trend of accelerating growth rates in the Brazilian economy since the beginning of the century.3 In this framework, the recent growth experience is best de- scribed as a vigorous economic recovery rather than an economic miracle. Its duration and the high growth rates attained can be explained by the broad gap between actual and potential growth since 1962. The economy was able to grow at an average 11.2 percent rate for seven years from 1967 to 1974 (which compares favorably with the 8.6 percent average growth rate maintained for 2. The output series for 1949 through 1976 was constructed by the economics department of Brazil's Central Bank, by interpolating between the "old" (1947 through 1972) and the "new" (1949, 1959, and 1965-76) GDP estimates of the Vargas Foundation. The values for 1947 and 1948 were obtained by applying to the Central Bank's 1949 value the same growth rates as in the "old" Vargas Foundation estimates. The value for 1977 is from Conjuntura Econ^Mica, Feb- mary 1978. 3. Haddad (1977) estimates the following yearly growth rates for aggregate output since 1900: 1900-02 to 1910-12: 3.96 percent 1910-12 to 1920-22: 3.97 percent 1920-22 to 1930-32: 4.04 percent 1930-32 to 1940-42: 4.71 percent 1940-42 to 1945-47: 5.49 percent. GROWTH RECORD 19 Table 2-1. Gap between Potential and Actual Output, 1947-77 (1970 actual output = 100) Year Potential output Actual output Gap (1) (2) [(1) -(2)]/(1) 1947 22.4 22.4 0.000 1948 24.0 24.0 0.000 1949 25.7 25.6 0.004 1950 27.6 27.3 0.011 1951 29.6 29.0 0.021 1952 31.7 31.7 0.000 1953 34.1 32.6 0.046 1954 36.6 36.0 0.017 1955 39.3 38.6 0.018 1956 42.2 40.0 0.055 1957 45.4 43.4 0.046 1958 48.7 46.9 0.038 1959 52.3 49.7 0.052 1960 56.2 54.6 0.029 1961 60.4 60.4 0.000 1962 65.0 63.8 0.019 1963 70.1 65.0 0.078 1964 75.5 67.1 0.125 1965 81.3 69.1 0.177 1966 87.6 71.7 0.222 1967 94.3 75.2 0.254 1968 101.5 83.6 0.214 1969 109.4 91.9 0.190 1970 117.8 100.0 0.178 1971 126.8 113.3 0.119 1972 136.6 126.6 0.079 1973 147.1 144.2 0.020 1974 158.3 158.3 0.000 1975 170.6 167.3 0.020 1976 183.7 182.7 0.005 1977 197.9 191.3 0.035 Source: Economics department ofthe Central Bank of Brazil, based on the Vargas Founda- tion national accounts. five years from 1956 to 1961), because during the six previous years, from 1961 to 1967, on average it had grown at only 3.7 percent a year (which compares unfavorably with the 6.0 percent yearly average experienced for four years in the previous downswing, from 1952 to 1956). Such statistical considerations suggest that the basic structural parameters limiting the country's 20 Post-1964 Economic Growth Figure 2-1. Gap between Potential and Actual Output, 1947-77 (1970 actual output = 100) 200 potential growth rate have changed only slightly in the past few years, and that the interruption of the 11 percent growth path in 1974 had to do as much with the limitations of domestic capacity as with the international oil crisis. The above description is supported by a recent study of Suzi- gan and others (1974), in which a potential output series is con- structed for the industrial sector in 195572. This is done by applying cumulated investments to the 1961 peak industrial output-capital ratio. The quotient between actual and potential output, which has the value 1.0 in 1961, reaches a low figure of 0.79 in 1965, increases to 0.84 in 1966, and recedes to 0.80 in 1967. Thereafter, it starts a continuous upward trend, climaxing with a value of 0.98 in 1972, the last year in their series. The cyclical nature of the 1967-74 experience is confirmed by the behavior of the quarterly Vargas Foundation capacity utilization GROWTH RECORD 21 Table 2-2. Vargas Foundation Capacity Utilization Indexes for Manufacturing Industry, 1968-77 Year January April July October 1968 n.a. n.a. n.a. 0.83 1969 n.a. 0.85 0.83 0.86 1970 0.86 0.85 0.87 0.86 1971 0.86 0.86 0.87 0.87 1972 0.87 0.86 0.87 0.89 1973 0.89 0.90 0.90 0.90 1974 0.90 0.89 0.89 0.87 1975 0.87 0.87 0.87 0.87 1976 0.89 0.89 0.89 0.87 1977 0.87 0.86 n.a. 0.82 n.a. Not available. Source: Conjuntura Econ6mica, various issues. index in the manufacturing sector (table 2-2). The initial value for this indicator is 0.83 in the fourth quarter of 1968, the following numbers show an uninterrupted upward trend, until a maximum value of 0.90 is attained, kept constant from the second quarter of 1973 to the first quarter of 1974. The successful economic recovery after 1967 was generated by conventional expansionist monetary and fiscal policies. Roberto Campos, the economic czar in government from 1964 through 1967, meant to save Brazilian capitalism from inflationary men- tality by squeezing aggregate demand. His successor, Antonio Delfim Netto, among other "bizarre structuralist ideas" (as Campos has put it in Simonsen and Campos 1974, p. 66), inter- preted Brazilian inflation in 1966-67 as cost determined.4 Acting on this diagnosis, Delfim Netto introduced direct price controls and, keeping a tight rein on wage deals, promoted active credit expansion for firms, consumers, and government.' Expansion started with consumer durables and government expenditures, reached private investment subsequently, but left the wage goods sectors dependent on the behavior of export outlets since domestic markets did not recover significantly. Suzigan and others (1974) suggest that recovery proceeded in two 4. See Delfim Netto and others (1967). 5. The shift in government emphasis and economic philosophy is well docu- mented by Fishlow (1973), and by Simonsen in Simonsen and Campos (1974, chap. 5). 22 Post-1964 Economic Growth stages: the first, from 1966 through 1969, was commanded by the expansion of consumer durables; and the second, from 1970 through 1973, centered on the dynamism of the capital goods industry. An unusual characteristic of the post-1964 period was that in- flation was successfully kept under control until 1972, despite the pressure of increasing demand, after having been signifi- cantly reduced from 87 percent a year in 1964 to 24 percent in 1967. A tight incomes policy seems to have been the main factor for this success. From January 1964 through February 1965 the Rio de Janeiro cost of living index went up by 91 percent, but the government in March 1965 allowed a minimum wage increase of only 57 percent from the levels established in February 1964. From February 1965 to February 1966 the cost of living went up by 44 percent, but the minimum wage was readjusted in March 1966 by only 27 percent. In March 1967 the minimum wage went up by 25 percent when the cost of living increased by as much as 37 percent from February 1966 to February 1967. After 1967 the wage squeeze was reduced, but only in the 1974-76 period did legal minimum wage readjustments consistently surpass ob- served cost of living increases (see table 2-3). The lag of wages behind price increases under declining rates of inflation was made possible after the substitution of a military- backed regime for an elected government in April 1964. Pre- viously, during the populist Goulart government, competitive bidding for popular support had led to a mobilization of the urban proletariat, with a multiplication of strikes and an acceleration of inflation rates. The 1964 regime brought about an immediate change in policy toward labor associations. Labor leaders were arrested and charters of some unions revoked. Usually, an inter- vener from the Labor Ministry was appointed, with control later turned over to "loyal" union members. The 1967 constitution prohibited strikes in essential activities, which are left undefined in the legal text. Strikes also were supressed under the provisions of the 1967 national security law. Free bargaining between labor unions and employers' associations was suspended, and all wage agreements were to be regulated by a wage formula dictated by the government. The formula was made up of three components. The first was designed to compensate for past inflation, reestab- lishing the average real wage observed in the preceding twenty- four months. The second anticipated future inflation, attempting to maintain during the following twelve months the same average GROWTH RECORD 23 Table 2-3. Minimum Wages and Cost of Living in Rio de Janeiro, 1963-77 Minimum wage Cost of living Percentage Index Percentage Date Cruzeiros change (1965-67 = 100) change January 1964 31.5 - February 1964 42.00 - February 1965 60.3 91.4 March 1965 66.00 57.1 February 1966 86.8 43.9 March 1966 84.00 27.2 February 1967 119 37.1 March 1967 105.00 25.0 February 1968 145 21.8 March 1968 129.60 23.4 April 1969 185 27.6 May 1969 156.00 20.4 April 1970 226 22.2 May 1970 187.20 20.0 April 1971 274 21.2 May 1971 225.60 20.5 April 1972 325 18.6 May 1972 268.80 19.1 April 1973 368 13.2 May 1973 312.00 16.1 April 1974 461 25.3 May 1974 376.80 20.8 April 1975 580 25.8 May 1975 532.80 41.4 April 1976 817 40.9 May 1976 768.00 44.1 April 1977 1,180 44.4 May 1977 1,106.40 44.1 Note: The cruzeiro was devalued steadily over this period. Source: Conjuntura Econ6mica, various issues. real wage as in the preceding period. A productivity adjustment factor was the third component of the formula. In theory, these rules not only guaranteed the purchasing power of wages but also maintained constant the wage share in output. In practice, expected future inflation was consistently underestimated until the late 1960s, whereas the allowance for productivity increases was much smaller than the observed growth of GDP per capita 24 Post-1964 Economic Growth through the mid-1970s. (For details, see Carvalho 1973, and Fish- low 1974). During Campos's term as economic minister, the wage policy was supplemented by an orthodox monetary policy in an attempt to reduce the inflation rate faster. The negative effects of this policy on real output promoted Delfim Netto to adopt an alterna- tive approach to monetary policy, which, in the words of Mario Simonsen, started to be passively conducted in accordance with the prin- ciple that real output growth should not be affected by liquidity crises ... This resulted in an expansion of the means of pay- ment at rates above what would normally be accepted by a believer in the quantity theory of money ... Actually, this pas- sive monetary policy was compatible with inflation control only because the Government started to adopt an intense pol- icy of price controls.6 Inflation control through a rigid incomes policy with monetary policy following the purely passive role of providing for the "needs of trade" was baptized "gradualism" by Brazilian techno- crats. Structuralist economists from the United Nations Eco- nomic Commission for Latin America (ECLA), however, had warned since the early 1950s that shock treatment stabilization policies, as recommended by the International Monetary Fund (IMF), would result in sharp output losses with meager gains in combating inflation. They also stressed that inflation should be fought by the eradication of structural cost pressures rather than by demand management (Grunwald 1961). Given the populist ideological underpinnings of their doctrines, structuralist econo- mists did not acknowledge that authoritarian regimes could succeed in substantially reducing inflationary pressures without impairing growth prospects by rigidly controlling wage earners' claims to higher nominal incomes. Hence, gradualism can be 6. See Simonsen and Campos (1974, pp. 85, 86, 114). "Passiveness" of mone- tary policy is hardly a novelty in modem Brazilian economic history. Using quar- terly data for 1954-69, Cardoso (1977) has shown that the econometric evidence for a "needs of trade" causation sequence running from nominal GNP to money stock is at least as strong as the evidence for the reverse sequence of most macro- economic models, running from money stock to nominal income. The conclusion that the money supply is passive is strengthened when the causation tests pro- posed by Sims (1972) are performed using quarterly series of inflation rates and growth rates of money stock for the 1946-74 period. GROWTH RECORD 25 viewed as a belated, though twisted, admission of the correctness of ECLA economists' criticisms of the IMF brand of monetarism as a means of fighting inflation in Latin America. The success of gradualism also promoted orthodox Brazilian economists to re- view their macroeconomic models. For example, the inflation model of Simonsen (1970) is embedded in a demand-constrained output-determination process and stresses cost determination of price levels through a feedback mechanism which is only moder- ately qualified by the pressure of demand on capacity.' The inflation control policy was accompanied by extensive reforms in the financial sector, the tax system, and the exchange rate regime, giving them more flexibility and increasing overall saving and the availability of foreign exchange. With wages under control, exchange rates, interest rates, tax collections, and public enterprise tariffs and prices could be raised without im- pairing private sector profit rates. The costs were borne by the public at large through a "corrective inflation" that fed only par- tially into legally determined wage levels. The principle of monetary correction-of nominal interest rates or the value of debt instruments for expected or actual inflation-was first introduced for government bills in 1965. The extension of monetary correction to other financial instruments eventually allowed a considerable expansion of financial inter- mediation in the country, which previously was constrained by rigid ceilings on interest rates (Fishlow 1974). The crawling peg as a system of exchange rate adjustment was introduced in August 1968, following the lead of Chile and Colombia (Bacha 1979). Previously, the exchange rate was de- valued at long intervals and in large steps to keep up with a steadily growing domestic price level. Consequently, there were large fluctuations in the real exchange rate, which favored spec- ulative activities and forced the government to adopt strict foreign exchange controls. These hindered export activities and made medium-term financial transactions more difficult. Under the crawling peg, the cruzeiro is devalued in small steps at fre- 7. Cavallo (1977) vindicates the structuralist model for Argentina, showing that for industrial goods the short-run contractionary effect of credit restriction is stronger on supply than on demand. Consequently, in the short run, credit restric- tion increases excess demand and aggravates inflationary pressures. Cavallo's theoretical model may help explain the stagflationary consequences of the Cam- pos economic policies in Brazil. For more detailed empirical evaluations of these policies, see Morley (1971) and Fishlow (1973). 26 Post-1964 Economic Growth quent but uncertain intervals. The incentive to speculate in foreign exchange is minimized, and risk-averse entrepreneurs can be attracted to export activities as foreign exchange restric- tions are lessened. The consistent pattern of minidevaluations decreased the probability of financial losses arising from unpre- dictable large devaluations. Hence, this policy also helped local entrepreneurs gain access to international financial markets. Under the Geisel government, after 1974, wage readjustments were put in line with observed inflation rates. This policy de- cision closed the main escape valve of the monetary correction system (including the crawling peg) and consequently made the Brazilian inflation rate extremely vulnerable to the oil crisis of late 1973. This crisis occurred at the peak of a domestic cyclical expansion, when the high growth rates of the "miracle" period started catching up with available input capacity. Scarcities of raw materials and manpower became apparent in Brazil's main urban centers in 1972-73 (Singer 1977). This coincided with an exhaustion of surplus capacity in the industrial sector, as indi- cated both by the behavior of the Vargas Foundation index of capacity utilization (table 2-2) and by the proportion of industrial- ists considering demand to be "strong" rather than "normal" or "weak" in the Vargas Foundation quarterly industrial survey. Overheating of the economy is clearly indicated by this index. In October 1973, 60 percent of the respondents considered demand to be "strong," as compared with an average of 22 percent of the respondents in the final quarter of the four previous years.8 Excess demand in output and factor markets together with the increase in the prices of oil and other tradable products resonated with the monetary correction system and nearly tripled the rate of inflation between 1972 and 1976. The next to the last section below analyzes in more detail the post-1973 macroeconomic problems of the country. Foreign Capital The outward-looking economic strategy followed after 1964 relied heavily on the domestic expansion of multinational corpo- rations (MNCs) for its dynamism. As reported in Suzigan and 8. See Conjuntura Econbmica, various issues. The relevant data are collected in Bacha (1978). FOREIGN CAPITAL 27 others (1974), leading sectors of industrial growth in the post- 1966 period were transport equipment, electrical machinery and appliances, mechanical industry, rubber, chemicals, and nonme- tallic mineral products. Foreign capital has a dominant influence in the first four sectors, as shown in table 2-4. Chemicals grew under the impulse of Petrobr6s, the state petroleum company, and nonmetallic minerals benefited from the boom in construc- tion activity initially promoted by the National Housing Bank Table 2-4. Patterns of Asset Ownership of the 5,113 Largest Nonfinancial Enterprises, 1974 Percentage share Total net assets National (millions Public Foreign private Sector of cruzeiros) enterprises enterprises enterprises Mining 9,637 62 12 26 Manufacturing 161,571 20 29 51 Nonmetallic 7,551 2 35 64 Metallic 27,711 34 12 54 Mechanical 8,293 1 46 53 Electrical 6,476 0 61 39 Transport equipment 15,155 4 63 33 Wood 8,782 0 9 91 Furniture 577 0 0 100 Rubber 1,834 6 61 33 Leather 685 0 11 89 Chemicals 40,162 55 23 22 Textiles 12,411 0 13 87 Food 16,910 1 31 68 Beverages 3,571 0 14 86 Tobacco 2,095 0 99 1 Printing 2,143 0 2 98 Miscellaneous 8,211 0 47 53 Agriculture, forestry 4,825 1 3 96 Construction 18,317 15 3 82 Public utilities 97,836 88 7 6 Transport 19,052 78 1 21 Other 78,784 90 8 2 Commerce 30,735 1 5 95 Services 84,656 27 4 69 Total 407,557 37 15 48 Note: Percentages may not add to 100 because of rounding. Source: "Quem e Quem na Economia Brasileira," Visao, August 31, 1975, p. 29. 28 Post-1964 Economic Growth and later financed by low-interest loans from the Eurodollar market.9 Foreign capital also has a monopoly in the tobacco industry and participates heavily in the modem sectors of food and other manufacturing activities. The overall participation of foreign firms in the total value of Brazilian fixed assets is not large. The "Who's Who" data on 5,113 nonfinancial firms published by Visio, a Brazilian business magazine (see table 2-4), estimate the foreign share of fixed assets to be 15 percent. At issue, however, are bigness, technological dynamism, access to credit sources, asset concentration, and market dominance. Being in control of new technological developments in industrial processes and products, the subsidiaries of mecs can condition the milieu in which they operate. This occurs especially when domestic firms expand passively in response to the growth of their market, and the government "pragmatically" adapts its policies, infrastruc- ture investment, and the country savings potential to the growth needs of the leading privately controlled sectors.10 This seems to have happened in post-1964 Brazilian experience. For example, the extraordinary growth of the foreign-owned car industry stim- ulated the expansion of private Brazilian and foreign firms pro- ducing spare parts and metallic products; determined the rhythm of oil imports, refinery construction, and road building; condi- tioned the style of urban planning; and channeled a large share of private saving to finance the purchase of cars. The car industry in its multiple ramifications epitomizes the premature affluent society brought to Brazil by the confluence of interests associated with the uNcs. According to Furtado (1972), the Brazilian experience also illustrates another fundamental distortion stemming from an un- controlled expansion of mNcs' activities. This relates to the conse- quences of adapting the pattern of domestic final demand to the expansion needs of the mNcs. The products sold by these firms conform to the market needs in industrial countries, where per capita incomes are five to ten times higher than in Brazil. To 9. The interrelationship between "dynamism" and foreign ownership comes up even more clearly when only the largest 318 industrial firms are taken into account, as in Von Doellinger, Faria, and Cavalcanti (1974). In their sample of big firms, the share of total assets owned by foreign enterprises in mining and manu- facturing is 40 percent rather than the 29 percent that can be calculated from the data in table 2-4. 10. This theme is further explored in Bacha (1976, Essay no. 1.3). ROLE OF THE STATE 29 produce an attractive market for these firms in Brazil, income must be highly concentrated on the top and, for a given family income, demand needs to be twisted toward international goods. At a minimum, this conclusion is not negated by the fact that in a sample of twenty-one countries from all continents Brazil has the largest share of income going to the top 20 percent of the population (Atkinson 1975). More broadly based international comparisons of income distributions in Chenery and others (1974) also find Brazil among the countries with the highest levels of income inequality. The demand twist toward foreign-controlled goods has been referred to in the literature as the "international demonstration effect." It is stressed by Wells (1976), whose criticism of the Furtado thesis is based on data showing that ownership of con- sumer durables is widespread among urban groups in the coun- try, rather than concentrated in the top 5 or 10 percent of urban families. Wells's point (which should be viewed as a complement to rather than a substitute for Furtado's hypothesis) is supported by the results of consumer budget surveys conducted in Sao Paulo in 1958 and 1970. They show a distinctly higher share of expenditures on consumer durables in workers' family budgets (and a correspondingly lower share of expenditures on food items) in 1970, in spite of slightly lower real family incomes (DIEESE 1974). In a subsequent paper, Wells (1977) criticizes these survey results and apparently changes his mind on the empirical importance of the demand twist toward durables. A definite empirical judgment on these questions is made difficult by the paucity of hard data. Role of the State As a result of the political impasse of the early 1960s, both Presi- dents Janio Quadros (1961) and Jodo-Goulart (1962-63) failed to enlarge the revenue base for the expansion of state economic activities. Public sector expenditures rose from 20 percent of GDP in 1955 to 27 percent in 1962, and gross investment of the public sector (including state enterprises) represented 39 percent of total capital formation in 1962, up from 25 percent in 1955 (Silva 1971). Nonetheless, the tax base was inelastic and the fiscal ma- chinery outdated. Exporters benefited from the establishment of a single exchange rate in 1961, but with the extinction of the 30 Post-1964 Economic Growth multiple exchange rate system the government lost a revenue source that in 1960 had generated income equal to 15 percent of its regular tax collections. The acquisition of the surplus over exports of bumper coffee crops from 1958 to 1960 also imposed severe financial strains on the government. As a consequence, deficits soared. As a proportion of GDP, the federal government budget deficit increased from less than 1 percent in 1955 to nearly 5 percent in 1962 (Ministry of Planning 1964). Concur- rently, government agency or public enterprise programs for road building, oil exploration and refining, electricity supply, com- munications, and iron ore mining and steel production, among others, lagged behind planned levels as strictly enforced price controls undermined the financial capabilities of these agencies. The 1964 regime, by doing away with the participation of the urban working class in the power structure, resolved the political impasse in clearly defined ways. The state was made solvent at the new high levels of expenditure with the costs being borne by the public at large through the mechanism of corrective inflation. The 1964-67 wage crunch was the crucial element of this opera- tion. At a more explicit level, a substantial tax reform was imple- mented, the fiscal machinery streamlined, and a "realistic" price policy initiated for the output of public services and government firms. The success of the reforms allowed federal government ex- penditures to grow at a faster rate than GDP since the mid-1960s. The government share in output averaged 9.5 percent in 1966-70. This increased to 10.1 percent in 1971-75 and to 10.5 percent in 1976-77. Federal government revenues expanded at an even faster rate. Budget deficits became increasingly smaller after 1967, and equilibrium was achieved in 1973. From 1974 onward, a small budget surplus was generated. Throughout the period, except for 1968 and 1977, net sales of government bonds and bills either more than compensated for the deficits or else added to the surpluses. Consequently, a continuous downward pressure on the money stock was exerted by the operations of the monetary authorities with the federal Treasury." The emphasis on administrative decentralization, efficiency, and profit making resulted in a significant expansion and diversi- fication of the activities of public enterprises, as well as in new mechanisms of financial coordination through holding compa- 11. The relevant data are in Conjuntura Econbmica, various issues. Derived statistics are compiled in Bacha (1978). ROLE OF THE STATE 31 nies such as EletrobrAs, Telebrds, SiderbrAs, and PortobrAs at the national level, and Fepasa, Sabesp, and Telesp at the level of the state of SAo Paulo. As of 1974, nineteen of the twenty largest, and forty-five of the hundred largest Brazilian corporations were state-owned. Only eleven of those forty-five are not in public utilities. The growth rate of state corporations has been suffi- ciently high, however, to increase their share in the total profits of the hundred largest corporations from 54 percent in 1968 to 63 percent in 1974. The proportion of net assets of the same hundred largest corporations belonging to state enterprises went up from 67 to 74 percent between 1970 and 1974. Thus, despite a lower profit rate, the state enterprises, through their access to ear- marked taxes, compulsory loans, and general government funds, apparently managed to grow at a faster pace than their private counterparts.'2 This complex growth pattern of public capital goes beyond federal activities. State and local governments have been seized by a desenvolvimentista (developmentalist) mentality that may eventually cause them to assert the position of their own firms in the marketplace, even in the face of opposition from private in- terests. 3 In spite of eventual conflicts, it can be argued that pub- lic investments were meant to play only a supporting role in the expansion of private initiative, particularly that of foreign origin. According to Tavares and Serra (1973), the crucial element guar- anteeing the economic dynamism of the period was the high level of "organic solidarity" achieved by the productive activities of the state and the Ncs.14 Together, they formed "an integrated nucleus of expansion," by dividing the tasks between them. The state supplied the domestic market with basic inputs and external economies at low cost, which the mNcs used to expand in both the domestic and export markets.'" 12. The average profit rate for state enterprises was 9 percent over the 1968-74 period, as compared with average profit rates over the same period of 12.5 and 15.8 percent, respectively, for the private national and foreign firms included in the hundred largest Brazilian corporations. See Visao, August 31, 1975. 13. A good case in point is the state of Minas Gerais, which has recently moved toward industrialization with its own public corporations, such as Camig, Ca- semg, Cemig, Frimisa, and Usiminas. 14. The "organic solidarity" between the state and the mcs is illustrated by the industrial incentives policy of Minas Gerais, which succeeded in attracting to this mineral-rich region subsidiaries of Fiat, Krupp, Peugeot, and Terex (a divi- sion of General Motors), among others. 15. Additional information on the role of the state in the economy can be found in Baer (1973), Baer, Kerstenetzky, and Villela (1973), Maneschi (1972), and Silva 32 Post-1964 Economic Growth The state enterprises need not necessarily follow the logic of depend ncia, however, as their importance in the economy grows, their internal linkages strengthen, and the country's growth needs in a changing international environment lead them into conflict with the mNcs. To put it more specifically, at least since the First National Development Plan (1971), there have been some attempts within the government to strengthen the industrial basis of the country through import substitution in heavy industries. Nonetheless, the external financing conditions in this period were so favorable that the alternative of furthering growth with external inputs imposed itself rather naturally. Domestic and international conditions after 1974, however, have led the government to favor a program of import substitution in capital goods and basic industrial inputs (such as ferrous and nonferrous metals, fertilizers, petrochemicals, and pulp and paper) based on expanded domestic sources of energy. The same conditions imposed restrictions on further growth of import- intensive consumer durables, such as automobiles, which were on center stage in the 1968-73 economic boom. As a conse- quence, state enterprise activities became more important for the economy, whereas foreign corporations based in the domestic market started playing a less relevant role than before. Before discussing the changing economic roles of state enter- prises and mNcs in post-1974 Brazil, the evolution of the payments balance needs to be considered and economic policy- making in the country since the oil crisis analyzed. Balance of Payments A battery of tax and credit incentives coupled with the system of minidevaluations proved very successful in promoting Brazilian exports of both nontraditional primary products and manufac- (1972). In view of the recent revision of Brazilian national accounts, it should be pointed out that the estimates presented in these sources for the share of government-related activities in CDP are considerably overestimated for the more recent years. For example, Baer, Kerstenetzky, and Villela assert that the share of government and public enterprises in gross fixed investment was 60.7 percent in 1969, whereas Conjuntura Econbmica, vol. 29, no. 6 (1975), p. 83, estimates this share as 43.4 percent in 1970. According to the latter source, the share of public expenditures (excluding state enterprises) in GDP was 22.4 percent in 1970, whereas Baer, Kerstenetzky, and Villela estimated this share as 34 percent in 1969. BALANCE OF PAYMENTS 33 tured goods. Revised and friendly legislation on foreign capital welcomed immense inflows of Eurodollar loans and renewed inflows of direct investment and official loans that had come to a standstill in 1963.16 The outward-looking post-1964 growth pattern is better charac- terized by its import intensity than by export promotion policies or access to international financial markets. This is vividly illus- trated in table 2-5 and figure 2-2, where the proportional resource deficit (equal to the ratio to exports of the difference between imports and exports of goods and commercial services) is plotted for the 1959-77 period. A distinction is made between the current deficit (with exports and imports measured in current dollars) and the basic deficit (with exports and imports calculated in con- stant 1970 dollars). The current deficit is influenced by terms of trade movements, whereas the basic deficit maintains the terms of trade constant at its 1970 value.'7 The cyclical movement of both resource deficits replicates the behavior of the GDP gap in figure 2-1 except that a trough is reached in 1965 rather than in 1967, and the 1974 peak is much higher than the value observed in 1962. The large difference in the 1974 peak value of the current deficit can be attributed to the oil crisis, since the deficits in 1970-73 are not much larger than those observed in the 1959-62 period. The values for the basic deficit indicate a balance of payments performance in the early 1970s worse than in the late 1950s and early 1960s. Under con- stant terms of trade, a trendless resource surplus of about 7.5 percent of exports is observed in 1959-62, whereas a rapidly growing deficit averaging 10 percent of exports is indicated for 1970-73. Projecting forward the average growth rate in 1965-73 of the ratio of imports to exports in constant 1970 prices, we conclude that a gap of the magnitude observed in 1974, under current dollar prices, would occur in 1977 or 1978, under constant 1970 16. These trends are reviewed in detail by Von Doellinger, Faria, and Caval- canti (1974). 17. The choice of 1970 as base year would seem inadequate, as the terms of trade in this year are the highest value in the series, except for 1973, 1976, and 1977-the first being the best year in memory for Brazilian commodity exports, and the last two being heavily affected by a temporary but sharp shortage of world coffee. In the 1960s, however, the terms of trade were negatively affected by a worldwide excess supply of coffee. Barring such events, in the 1980s Brazil might be able to maintain terms of trade with average values similar to those observed in 1970, which is the reason this year was chosen as a base for the index. Table 2-5. Ratio to Exports of C.i.f.-Trade Deficits, 1959-77 (imports and exports in millions of U.S. dollars) Exports of Imports of Exports of Imports of goods and goods and goods and goods and Current commercial commercial Basic Terms of commercial commercial deficit services in services in deficit trade Year services (X) services (M) ([M -XIX) 1970 prices (X*) 1970 prices (M*) ([M* -X*]/X*) (1970=100) 1959 1,324 1,348 0.018 1,568 1,456 -0.071 90.9 1960 1,320 1,428 0.082 1,593 1,582 -0.072 91.5 1961 1,458 1,428 -0.021 1,679 1,555 -0.074 94.4 1962 1,264 1,437 0.137 1,658 1,552 -0.064 81.8 1963 1,459 1,447 -0.008 1,921 1,526 -0.216 79.5 1964 1,482 1,211 -0.183 1,641 1,319 -0.196 98.4 1965 1,646 1,034 -0.372 1,813 1,111 -0,387 98.0 1966 1,806 1,420 -0.214 2,060 1,493 -0.275 91.9 1967 1,731 1,576 -0.090 1,976 1,623 -0.179 89.8 1968 1,980 2,006 0.013 2,286 2,026 -0.114 86.0 1969 2,445 2,273 -0.070 2,748 2,317 -0.157 90.2 1970 2,908 2,874 -0.012 2,908 2,874 -0.012 100.0 1971 3,074 3,686 0.199 3,172 3,543 0.117 92.8 1972 4,177 4,772 0.142 3,829 4,299 0.123 98.2 1973 6,455 7,084 0.097 4,317 5,097 0.181 107.9 1974 8,292 14,060 0.696 4,372 6,593 0.508 88.3 1975 9,083 13,482 0.484 4,604 6,215 0.350 90.5 1976 10,530 13,631 0.294 4,601 6,230 0.354 104.8 1977 12,689 13,421 0.058 4,614 5,797 0.256 118.8 Source: Central Bank of Brazil, Boletim Mensal. BALANCE OF PAYMENTS 35 Figurc 2-2. Ratio to Exports of C.i.-Tradc Deficits, 19,59-77 0.6- 0.4 -- C;urrenft lcIIit O ,, /Basic deficit 1960 1965 1970 1975 Source: Table 2-5. prices. This result stresses the import-biased nature of post-1964 growth. It suggests that the oil crisis only anticipated in time a c.i.f.-trade deficit that would occur a few years hence, even under the unusually high 1970 terms of trade. The pre-1974 growth pattern could be maintained without seri- ous balance of payments dislocations, once the 1965 import slack was absorbed, only because the terms of trade improved signifi- Table 2-6. Foreign Capital: Demand and Supply, 1970-77 (millions of U.S. dollars) Item 1970 1971 1972 1973 1974 1975 1976 1977 Foreign capital demand 1,234 2,157 2,691 3,361 9,043 8,873 9,022 8,385 Debt service plus profit remittances 1,025 1,270 1,722 2,385 2,821 3,818 5,026 6,800 Resource gap 209 887 969 976 6,222 5,055 3,996 1,585 Foreign capital supply 1,779 2,687 5,130 5,539 8,106 7,923 10,214 8,845 Direct investment (net) 132 168 318 940 887 895 1,010 800 Loans and financing 1,433 2,037 4,299 4,495 6,891 6,530 7,921 8,345 Short-term capital (including errors and omissions) 214 482 513 104 328 498 1,283 300 Additions to foreign reserves +545 +530 +2,439 +2,178 -937 -950 +1,192 +460 Sources: Central Bank of Brazil, Boletin Mensal; and Conjuntura Econbmica, February 1978. BALANCE OF PAYMENTs 37 cantly between 1968 and 1973. This means that to maintain a high GDP growth rate in the future, Brazil needs to reduce the income elasticity of its import demand, because it cannot hope to increase continuously the ratio of exports to GDP. This is in line with the results of the simulations in chapter 5, according to which a less import-intensive strategy is a precondition for a high steady rate of GDP growth in the near future. The growth perspec- tives for the near term are more troublesome because of the balance of payments burden of the large foreign debt that has accumulated since the oil crisis. In table 2-6 the trends in Brazil's balance of payments can be followed since 1970, when the country began to experience a consistent resource gap. This table reshuffles the external ac- counts to indicate the balancing of foreign capital supply and demand. Capital demand is divided in two components: the re- source gap (that is, the c.i.f.-trade deficit plus tourism and govern- mental and other nonfactor services) and debt service plus profit remittances. Capital supply has three sources: net direct invest- ment, loans and financing, and short-term capital. The movement of foreign reserves is the balancing item between foreign capital supply and demand. Two periods can be distinguished in this table, 1970-73 and 1974-77. During the first period, capital demand increased at a rate of nearly 40 percent a year, but capital supply more than matched this growth rate. " As a result, foreign reserves soared from US$1.2 billion to US$6.4 billion between December 1970 and December 1973 (table 2-7). In 1974, as a result of a US$6.2 billion gap in the resource account, foreign capital demand increased by 170 percent with respect to its value in 1973. In both 1974 and 1975 foreign re- serves declined nearly US$1 billion. Subsequently, the resource gap declined significantly as a result of import controls, high coffee prices, and expanded manufactured exports. Nevertheless, under the pressure of rapidly increasing debt service, capital demand was kept at a very high level of more than US$8.0 billion 18. Foreign capital demand would have reached US$9.2 billion in 1977, had it continued to grow at the rate of40 percenta year after 1973. This is slightly higher than the observed value of capital demand in 1974 and confirms the results of the forecasting exercise above, according to which the oil crisis merely anticipated by three years the deficit that Brazil would have experienced in 1977. (Note that "billion" is used throughout this book to mean "thousand million.") Table 2-7. External Indebtedness and International Reserves, 1969-77 (millions of U.S. dollars at year end) Private Total foreign International suppliers' Currency International Year debt Governmental, institutionsb credits loans, Otherd reserves 1969 4,403 1,104 364 447 1,605 879 656 1970 5,295 1,246 456 611 2,285 698 1,187 1971 6,622 1,403 576 845 3,193 604 1,723 1972 9,521 1,504 762 1,136 5,528 591 4,183 1973 12,572 1,687 972 1,442 7,849 621 6,416 1974 17,166 2,151 1,388 1,812 11,211 603 5,269 1975 21,171 2,430 1,655 1,980 14,561 546 4,041 1976 25,985 2,756 1,993 2,414 18,194 628 6,544 1977 32,037 2,864 2,355 3,773 21,528 1,517 7,256 a. Includes U.S. Agency for International Development (AID), Export-Import Bank, U.S. Public Law 480, Canadian Wheat Board, and others. b. Includes World Bank, Inter-American Development Bank, and International Finance Corporation. c. Federative Republic of Brazil, Resolution 63, Law 4131, and Instruction 289. d. Includes compensatory loans plus foreign debt generated by the acquisition of foreign assets in the field of electricity and telecommunications plis consolidated public debt plus bonds. Sources: Central Bank of Brazil, Boletim Mensal; and Conjuntura Econbmica. BALANCE OF PAYMENTS 39 a year, and capital supply more than met the increased re- quirements. Consequently, by December 1976 foreign reserves reached their previous peak and continued to grow to US$7.3 billion in December 1977. Loans and financing were by far the most important source of capital supply (luring the entire period. According to table 2-7, in 1970 official debt (that is, that owed to governments and interna- tional institutions) was 32 percent of total Brazilian foreign debt. In the same year, private sources (that is, private suppliers' credits plus currency loans) represented 55 percent of total debt. In 1976-77 official sources were only 10 percent of the additions to foreign debt, while private credits and currency loans repre- sented 80 percent of the total. The size of foreign debt increased substantially from US$5.3 billion in 1970 to US$17.2 billion in 1974 and then to US$32 billion in 1977. The rate of expansion of net debt (gross debt minus international reserves) was equally pronounced: it rose from US$4.1 billion in 1970 to US$11.9 billion in 1974 and then to US$24.8 billion in 1977. As a proportion both of exports and of GDP, net debt doubled in value between 1973 and 1977, when it was twice as large as exports and equal to 15 percent of GDP. Until 1974 the ratio of net debt service to exports was reasonably stable at an average value of 0.35. After 1974 debt services started growing very rapidly to reach 52 percent of exports in 1977.19 Between 1974 and 1977 strict import controls were applied, and the quantum of imports declined by nearly 13 percent, while real GDP increased by slightly more than 20 percent. This is a surprising result in view of the strong upward trend of the import coefficient during the "economic miracle." The excessive level of imports in 1974 seems to have been one factor contributing to this result. When the average income elasticities of imports ob- served in the 1965-72 period were applied to the real output growth in 1974, "excess" 1974 imports were estimated at more than US$1 billion, or nearly 10 percent of the import bill (Bacha 1978). This excess is shown in the national accounts by a change in inventories amounting to 7.5 percent of GDP (in constant 1970 cruzeiros). The largest previous figure in the series is 4.3 percent in 1973. In no other year since 1966 did inventory changes 19. The figures in this paragraph are derived from official sources in Bacha (1978). 40 Post-1964 Economic Growth amount to more than 2.8 percent of GDP. The conclusion is that a part of the increase in imports in 1974 was additions to inven- tories made in anticipation of import restrictions after the oil price increase. A second reason for the success of import controls was a de- cline in the rate of fixed investment from 25 to 22 percent of GDP between 1974-75 and 1977. This was accompanied by a shift toward domestically produced components in the investment projects approved by CACEX, the foreign trade department of the Central Bank of Brazil, allowing capital goods imports in 1977 to be 34 percent lower than in 1974.20 The decline in imports under conditions of an increasing rate of domestic absorption also was accompanied by a rekindling of inflation rates. For these reasons, observers agree that imports will have to start growing again if GDP growth is to be maintained with declining rates of domestic inflation. Government spokes- persons have suggested that ongoing projects of import substitu- tion will significantly reduce the income elasticity of import demand. Two recent independent forecasts are more cautious about likely patterns of the demand for imports. For the 1978-82 period Banco Garantia (1978) and Wharton (1978) anticipate growth rates of output in the neighborhood of 7 percent a year.2' They both assume income elasticities of im- port demand equal to 1.44. (Observed elasticities were 2.02 be- tween 1965 and 1972, and 1.68 between 1968 and 1972.) Exports are projected to grow at a rate of 11.6 percent a year by Wharton and at 10.4 percent a year by Garantia. (Past growth rates of real exports were 10.1 and 12.1 percent a year, respectively, in 1965-72 and 1968-72.) In both cases foreign reserves are as- sumed to be kept constant at their December 1977 levels. Banco Garantia and Wharton forecasts for net external debt and significant debt ratios are shown in table 2-8. In their scenarios, net debt continues to grow at relatively fast rates, with the ratio of net debt to exports and the debt service rate tending to sta- 20. CACEX promotes agreements between importers and local suppliers of capital goods to ensure an "adequate" participation of domestic industry in in- vestment projects benefiting from fiscal incentives. The dollar share of national firms in these investment agreements increased from 58 to 77 percent between 1974 and 1977. 21. Actually, 6.7 percent in the case of Garantia, and 7.1 percent in the Wharton forecasts. In both cases GDP growth rates start in the range of 5-6 percent and then converge to 7-8 percent a year. AFTER THE OIL CRISIS 41 Table 2-8. Wharton and Banco Garantia Forecasts for Net External Debt and Significant Debt Ratios Item 1978 1979 1980 1981 1982 Wharton Net debta (billions of U.S. dollars) 27.7 32.5 37.5 42.3 47.0 Net debt/exports 2.2 2.4 2.4 2.3 2.3 Net debt service/exports 0.59 0.60 0.66 0.66 0.65 Banco Garantia Net debt, (billions of U.S. dollars) 27.9 31.3 35.3 39.6 44.4 Net debt/exports 2.3 2.2 2.3 2.3 2.3 Net debt service/exportSb 0.64 0.58 0.63 0.63 0.65 a. Gross debt = net debt + US$7.3 billion. b. Values calculated by the author, assuming the same ratio of net debt services at year to net debt at end of year t - I as estimated by Wharton. Sources: Wharton (1978); Banco Garantia (1978). bilize at levels not yet experienced in recent Brazilian history. Notice, particularly, that net debt services will tend to absorb 65 percent of exports during the period. (In the recent past, the highest previous rate was 52 percent, observed in 1977.) In summary, moderately high GDP growth rates in the near future are likely to cause considerable balance of payments dislo- cations. The external finance may be available as a counterpart to the surpluses of the oil-producing countries, and its deployment in the Brazilian economy may be a good investment in the long run. But it is not clear that the international banking community will have the nerve to sit still while Brazil's debt service in- creases continuously until reaching the all-time ratio of 65 per- cent of total exports. On the basis of recent trends, which are discussed in the next section, it is more likely that Brazilian policymakers will see fit to reduce the growth rate of domestic absorption rather than face the external economic and political costs of a rapidly expanding foreign debt. Economic Policymaking after the Oil Crisis Since 1974 the macroeconomic policies of the federal govern- ment have followed a stop-and-go pattern, indicative of the diffi- culties of meeting policy objectives when the options were more 42 Post-1964 Economic Growth narrowly limited than before. The planning document of the government published in 1974 effectively ignored the con- straints imposed on the economy by the emergence of domestic capacity limitations and the quadrupling of oil prices in 1973. It forecast growth rates of output of 10 percent a year for the 1975-80 period, assuming that inflation rates would decline and that the balance of payments would be kept in equilibrium (Federative Republic of Brazil, 1974).22 The projections of the planning document notwithstanding, monetary and fiscal policy followed a cautious path in 1974. Partly as a result of international reserve losses, the money sup- ply in real terms expanded by only 5.7 percent on the average between 1973 and 1974. Federal government outlays increased modestly by 5.5 percent in real terms during the same period. Prices appeared to respond perversely to the deceleration of de- mand growth. Under the previous government, some critical prices had been frozen to hide real inflation rates. Dislodging this "inflation backlog" when the full impact of higher oil prices was being felt meant that prices increased faster in 1974 than in 1973. The GDP price deflator went up by 20.5 percent in 1973 and by 31.5 percent in 1974. Fiscal and monetary restraints under conditions of autono- mously rising prices and a higher propensity to import soon showed in the behavior of manufacturing output. In relation to the same period in 1973, growth of industrial consumption of electricity in Rio de Janeiro and SAo Paulo decelerated continu- ously from 16.6 to 4.0 percent from the first to the fourth quarter of 1974. When this growth rate of consumption declined further to 1.3 percent a year in the first quarter of 1975, the government decided that it could not stand an economic recession on top of the political problems it was facing as a consequence of the vic- tories of the opposition party in the senatorial elections of November 1974. The decisions were made to expand aggregate demand, as a means of stimulating industrial production, and to fight the bal- 22. Contemporaneous macrosimulation exercises by Cardoso and Taylor (see Cardoso 1975, and chapter 5 of this book) pointed out the internal inconsistencies of the plan targets. Later on, the minister of planning (Velloso 1975) offered the following argument in defense of his growth forecasts: "If, in August 1974, a target growth rate of 4-6 percent had been established for 1975, despair would have been total. And this lower growth rate would have materialized not in 1975 but in 1974 itself." AFTER THE OIL CRISIS 43 ance of payments deficit in the short run by means of rigid import controls and, in the medium run, by programs of export promo- tion and import substitution emphasizing self-sufficiency in so- called basic inputs (steel, petrochemicals, nonmetallic minerals, fertilizers, and pulp and paper). These decisions are reflected in the behavior of the money base and of government expenditures. In spite of international reserve losses, the money base grew by 47 percent between June 1975 and June 1976, while federal government expenditures in the first half of 1976 were 81 percent higher than in the first half of 1975. (In June 1976 wholesale prices were 39 percent above those of June 1975.) Industrial output responded swiftly to the demand expansion, growing by 10.8 percent between 1975 and 1976. Inflation rates responded both to the demand pressure and to the cost tensions arising from the import control program. After increasing slowly from 31.5 to 32.7 percent between 1974 and 1975, the implicit price deflator rose by 41.3 percent in 1976. At the end of 1976 the government concluded that reheating of the economy had gone too far. The rekindling of inflation rates was imputed to the excessiveness of the demand expansion, al- though independent cost-push pressures seem to have been equally important. For example, Munhoz (1978) suggests that the cost increases brought about by the import control program added 8.9 percentage points to the rate of inflation in 1976, which is about the difference in the rate of increase of the GDP price deflator between 1975 and 1976.23 Interpretation of the causes of inflation may be disputed, but the recent events made policymakers aware of the necessity to revise downward the ambitious investment program outlined in 1974. The government proclaimed a new policy alternative, the strategy of progressive deceleration, which, according to a dia- gram in Velloso (1977, p. 122), contemplated a period of GDP growth in the neighborhood of 5 percent a year extending through 1980. 23. Munhoz calculates the expenditure increases that were due to higher mark- ups on oil sales to domestic consumers, higher tax rates on oil and fuel, additional importtaxes, the creation of the import deposit, and the freeing of interest rates. These represented a Cr$90 billion cost increase in 1976, or 129 percent more than the value of these expenditures in 1975. Multiplying this figure by the share of such expenditures in GDP in 1975 (6.9 percent), we arrive at the 8.9 percentage points mentioned in the text. 44 Post-1964 Economic Growth Perspectives Brazilian political scientists have noted that, lacking popular sup- port, the post-1964 authoritarian regime tried to legitimize itself through its economic performance (Lafer 1975; Soares 1978). From 1967 to 1972 this task was facilitated by the existence of unused industrial capacity, an initial trade surplus, and the strongest boom in world trade and international capital markets since World War II. Not having to worry about labor union move- ments, Brazilian policymakers had an easy life collecting the dividends of an expansionary policy after inflation reached manageable levels in 1967. When the "narrow limits of the possible" (Taylor 1974) made their presence felt after 1972, Brazilian economic policymaking seesawed in a frustrating attempt to make high GDP growth rates compatible with declining inflation rates and balance of pay- ments equilibrium. Future scenarios were difficult to visualize because the economic troubles came at a time of political unrest. The extent of public dissatisfaction was made obvious in the November 1974 elections when the poorly organized opposition party easily won in sixteen out of twenty-one senatorial races. On another political front, at the same time as the House of Repre- sentatives, at the request of the opposition party, was carrying on an investigation of the "extravagant" role of uNcs in Brazil, the most prestigious Brazilian newspaper and newsmagazines launched a fierce campaign against the "excessive" state inter- vention in the economy. In the background of this dispute stands the question of under whose command the large import substitu- tion and export promotion projects now under consideration are going to be executed. In the development of the Northeast petrochemical complex an ingenious formula was found to combine the interests at stake. As reported by Araujo and Dick (1974), nearly all firms are jointly owned by a national group, a foreign group, and Petroquisa (a Petrobris subsidiary). The foreign group always holds a minority interest, and its share is not larger than that of Petroquisa. U.S. and German firms balked at the deal, but adventurous Japanese capital signed up for the job. The government, however, is only reluctantly coming to realize that a new growth phase will not get under way without a new division of tasks between state enter- prises and foreign firms. Lacking balanced economic growth as a legitimization device, REFERENCES 45 the government will also have to come to grips with the issues of political liberalization and popular participation to enable the country to go through the difficult transition period without po- litical turmoil. The sharp tradeoffs between GDP growth, balance of payments equilibrium, and inflation rates, the power balance between state enterprises and multinational corporations, and the institutionalization of the political regime are problems inherited from the previous growth phase. From their solution will derive the pattern of Brazil's development in the 1980s. References Araujo, J. T., and V. Dick. 1974. Governo, empresas multinacio- nais e empresas nacionais: 0 caso da ind6stria petroquimica. Pesquisa e planejamento econbmico 4(3):629-54. Atkinson, A. B. 1975. The economics of inequality. Oxford: Clar- endon Press. Bacha, Edmar L. 1976. Os mitos de uma decada: Ensaios de economia brasileira. Rio de Janeiro: Paz e Terra. . 1977. Issues and evidence on recent Brazilian economic growth. World development 5(1-2):47-67. . 1978. Brazil's balance of payments before and after the oil crisis. Paper prepared for the UNCTAD/UNDP project on the bal- ance of payments adjustment process in developing countries, July 1976. Brasilia: University of Brasilia; processed. . 1979. Notes on the Brazilian experience with minideval- uations, 1968-76. Paper presented at the Ford Foundation/ Central Bank of Barbados seminar on developing countries and the international financial system, Barbados, January 11-14, 1978.Journal of development economics 6(December):463-82. Baer, Werner. 1965. Industrialization and economic develop- ment of Brazil. Homewood, Ill.: Irwin. Published in Portu- guese as: A industrializagio e o desenvolvimento econ6mico do Brasil. Rio de Janeiro: Vargas Foundation, 1975. . 1973. The Brazilian boom, 1968-72: An explanation and interpretation. World development 1(8):1-16. Baer, Werner, Isaac Kerstenetzky, and Annibal V. Villela. 1973. The changing role of the state in the Brazilian economy. World development 1(11):23-34. 46 Post-1964 Economic Growth Banco Garantia. 1978. Carta mensal (Sao Paulo, June). Cardoso, Eliana A. 1975. 1982: Algumas projecbes condicionais para o Brasil. M.A. thesis, University of Brasilia. _. 1977. Moeda, renda e inflagao: Algumas evid6ncias da economia brasileira. Pesquisa e planejamento econ5mico 7(2):423-34. Carvalho, L. 1973. Principios e aplicaqao da politica salarial p6s-1964. Ensaios para discussao no. 8. Brasilia: University of Brasilia, Department of Economics. Cavallo, D. 1977. Stagflationary effects of monetarist stabiliza- tion policies. Ph.D. dissertation, Harvard University. Chenery, Hollis, and others. 1974. Redistribution with growth. London: Oxford University Press. Delfim Netto, Antonio, and others. 1967. Andlise do comporta- mento recente da economia brasileira. Working paper of the Joint Advisory Group to the Ministries of Finance and Plan- ning. Partly reproduced as an appendix to Ministry of Planning, Diretrizes do governo: Programa estrategico de desenvolvimento. Rio de Janeiro: Departamento de Imprensa Nacional, 1967. DIEESE (Departamento Intersindical de Estatistica e Estudos S6cio-Econ6micas). 1974. Familia assalariada: Padrao e custo de vida. Estudos s6cio-econbmicos 2 (January). Federative Republic of Brazil. 1974. Projeto do II plano nacional de desenvolvimento (1975-1979) [Second national develop- ment plan]. Rio de Janeiro: Servigo Grafico, Instituto Brasileiro de Geografia e Estatistica (IBGE). Fishlow, Albert. 1973. "Some reflections on post-1964 Brazilian economic policy. In A. Stepan, ed. Authoritarian Brazil. New Haven: Yale University Press. _. 1974. Indexing Brazilian style: Inflation without tears? Brookings papers on economic activity 1:261-82. Furtado, Celso. 1963. The economic growth of Brazil. Berkeley: University of California Press. -. 1972. Analise do 'modelo' brasileiro. Rio de Janeiro: Civilizagao Brasileira. Grunwald, Joseph. 1961. The "structuralist" school on price sta- bility and development: The Chilean case. In A. Hirschmann, ed. Latin American issues: Essays and comments. New York: Twentieth Century Fund, pp. 95-123. REFERENCES 47 Haddad, Claudio. 1977. Crescimento do produto real brasileiro, 1900-1947. In F. R. Versiani and J. R. M. de Barros, eds. Formação econômica do Brasil: A experiência da indus- trialização. São Paulo: Edição Saraiva. Lafer, C. 1975. O sistema político brasileiro. São Paulo: Editora Perspectiva. Malan, Pedro, and Regis Bonelli. 1977. The Brazilian economy in the seventies: Old and new developments. World Develop- ment 5(1-2):19-46. Maneschi, Andrea. 1972. The Brazilian public sector. In Riordan Roett, ed. Brazil in the sixties. Nashville, Tenn.: Vanderbilt University Press. Ministry of Planning. 1964. Programa de ação econômica do gov- erno, 1964-1966. Rio de Janeiro: Serviço Gráfico, Instituto Bra- sileiro de Geografia e Estatística (IBGE). Morley, Samuel A. 1971. Inflation and stagnation in Brazil. Eco- nomic development and cultural change 19 (January). Munhoz, Dercio. 1978. Inflação: Impulsos de custos reversíveis e não reversíveis e a taxa de inflação de equilíbrio. Brasília: University of Brasília, Department of Economics; processed. Simonsen, Mario Henrique. 1970. Inflação: Gradualismo versus tratamento de choque. Rio de Janeiro: Apec Editora. Simonsen, Mario Henrique, and Roberto Campos. 1974. A nova economia brasileira. Rio de Janeiro: José Olympio. Silva, F. R. 1971. A evolução das funções do governo e a expansão do setor público brasileiro. Pesquisa e planejamento eco- nômico 1(2):235-82. . 1972. Avaliação do setor público na economia brasileira. Relatório de Pesquisa no. 13. Rio de Janeiro: Instituto de Pla- nejamento Econômico e Social/Instituto de Pesquisas (ipEA INPES). Sims, Christopher A. 1972. Money, income, and causality. Amer- ican economic review 62(4):540-52. Singer, Paul. 1977. As contradições do milagre. In A crise do "milagre." Rio de Janeiro: Paz e Terra. Soares, G. 1978. After the miracle. Luso-Brazilian review 15(2). Suzigan, W., and others. 1974. Crescimento industrial no Brasil: Incentivos e desempenho recente. Relatório de Pesquisa no. 26. Rio de Janeiro: Instituto de Planejamento Econômica e Social (IPEA). 48 Post-1964 Economic Growth Tavares, Maria ConceigAo, and Jose Serra. 1973. Beyond stagna- tion: A discussion on the nature of recent development in Bra- zil. In J. Petras, ed. Latin America: From dependence to revo- lution. New York: John Wiley. Taylor, Lance. 1974. Short-term policy in open semi-industrial economies: The narrow limits of the possible.Journal of devel- opment economics 1:85-104. Velloso, J. P. 1975. Atualidade do II PND. Rio de Janeiro: Centro de Servigo Grifico, Instituto Brasileiro de Geografia e Estatistica (IBGE). _. 1977. Brasil: A solugao positiva. Sao Paulo: Abril-Tec Editora. Von Doellinger, Carlos, Hugo Barros de Castro Faria, and Leo- nardo Caserta Cavalcanti. 1974. A politica brasileira de comer- cio exterior e seus efeitos: 1967-73. Relat6rio de pesquisas no. 22. Rio de Janeiro: Instituto de Planejamento Econ6mico e Social /Instituto de Pesquisas (IPEA/INPES). Wells, John. 1976. Underconsumption, market size and expendi- ture patterns in Brazil. Bulletin of the Society for Latin Amer- ican Studies (University of Liverpool) no. 4, pp. 23-58. -. 1977. The standard of living of the SAo Paulo working class during the 1960s. Cambridge, England: Cambridge Uni- versity, Faculty of Economics; processed. Wharton Econometric Forecasting Associates, Inc. Wharton EFA Brazilian econometric model, January 1978 postmeeting fore- cast: Economic developments in 1977 and forecasting review. Philadelphia. 3 Theoretical Framework for Identity-Based Planning Eliana A. Cardoso and Lance Taylor THIS AND THE FOLLOWING TWO CHAPTERS describe a family of macroeconomic models put together to analyze Brazil's recent growth experience and prospects for the future. The models are constructed around the same skeleton of accounting identities that supports the national accounts-hence the generic name given to them. Our approach has been first to exploit the account- ing identities as far as possible in setting up models, only then adding enough additional assumptions to make their solutions determinate. The theoretical basis for identity-based planning is described in this chapter, and in the following two chapters we apply the models empirically to analyze Brazilian growth and distribution during the 1960s and likely tradeoffs during the later 1970s. The identities come close to giving a unique description of the economy, but they must be supplemented by a few main assump- tions. Two particular sets of axioms that can be used to close the model have been widely discussed in recent years. One, usually associated with economists from Cambridge, England, stresses the impact of patterns of demand for investment and consump- tion goods on growth and distribution. The other set of closure assumptions is associated with neoclassical economists from 49 50 Theoretical Framework Cambridge, Massachusetts, and elsewhere, and emphasizes pro- ductivity and thrift. The two sets of assumptions can lead to widely differing interpretations of the same set of data. By and large, the Cambridge, England, version seems to fit better for Brazil. The rest of this chapter is devoted to setting out the identity- based forecasting technique in detail. We use two examples, beginning in the next section with a simple model to define terms and set the scene. In the following section we then describe the more complex set of equations actually used for Brazil. Each set of models is solved in two stages, starting from an initial base year with data satisfying the identities. The first is the calculation of growth rates of a set of endogenous variables from growth rates of exogenous variables, in a differential version of the basic identities and model closure equations. The second stage uses approximations of the values of all variables in the level form of the model (calculated from the growth rates) as the starting point for an algorithm to generate a fully consistent set of macroeco- nomic accounts for some subsequent year. That year in turn serves as a base for another step forward in time. The first stage in practice may be sufficiently accurate to be acceptable as a medium-term forecast by itself; it also helps illustrate the inter- dependency of growth of many variables in one economy. The second stage provides a technique for analyzing long-term growth.I A Simple Model Consider an economy in which only one good ("GNP") is pro- duced. Variables and equations describing the system appear in tables 3-1, 3-2, and 3-3. The first two equations in table 3-2 are the usual national in- come identities for expenditure and factor payments. To provide a basis for forecasting rates of inflation, prices are included ex- 1. Planning techniques similar to the one described here do not figure much in the academic literature, although they are used in practice. At the aggregate level, the best example is given by Cauas (1972). At the disaggregated level, the work of Johansen (1974) is of course relevant. Input-output consistency plans resemble our identity-based projections, although less emphasis is placed on prices. For descriptions, see Lewis (1966), Clark (1975), and the references the latter cites. A SIMPLE MODEL 51 Table 3-1. Symbols in the GNP Model Symbol Definition X Output C Consumption I Investment G Real government expenditure L Employment K Capital stock P Price level w Wage r Rate of profit after taxes t Profit tax rate H* Change in the money base (equals government deficit) aL Labor share aK Capital share 6C, 41 6G Shares of consumption, investment, and government in total spending YL, YK Shares of labor and capital income which are consumed g Growth rate of the capital stock E Rate of technical progress in Hicks-neutral form plicitly in equations in (3.1.1) and (3.1.2). On the quantity side the aggregate capital stock K appears in the definition of factor payments, despite the well-known theoretical demonstrations of its nonexistence. We work with aggregate "capital" in recogni- tion of the extreme difficulty of estimating in any other way the contribution of investment to growth of productive capacity. Equation (3.1.2) also implies that capital gains on the existing stock accrue to recipients of profit incomes, that is, after-tax profits are rPK, and these increase with the price level P, even when the after-tax rate of profit r and capital K are constant. In this illustrative model, the corporate tax rate t is the only fiscal policy instrument. The tax incides on profits in the usual way, (PX - wL) (1 - t) = aKPX (1 - t) = rPK, Table 3-2. Level Form of the GNP Model PX = PC + PI + PG (3.1.1) PX = wL + rPK/(1 - t) (3.1.2) q = wLIPX (3.1.3) aK = rPK/[(1 - t) PX] (3.1.4) PC = ywL + yKrPK (3.1.5) PG = H* + trPK/(1 - t) (3.1.6) I = gK (3.1.7) 52 Theoretical Framework Table 3-3. Log-Change Form of the GNP Model X' = 6CC' + 61I' + 6GG' (3.1.1') X' = aLL' + aKK' + E (3.1.2p') P' = aLW' + aK (r' +P' + t' - (3.1.2c') L' = aL' + P' + X' - w' (3.1.3') K' = a,' +X' +P' - (r'+P'+ t' (3.1.4') P' + C' = (1/fc) [yLaL (w' + L') + yXa1K 0 - t (r' +P' + K')] (3.1.5') trPK/(1 -t) H* P' + G' = [t'/(1 - t) + r' + P' + K'] + H*(H*)' (3.1.6') PG PG I' = g' + K' (3.1.7') and equation (3.1.2) follows directly. In the discussion below of implications of different ways of closing the model, we give examples by varying t and observing how other variables respond. Equations (3.1.3) and (3.1.4) respectively define labor and capital shares in product and will prove useful in setting up neoclassical specifications below. Equation (3.1.5) is based on the usual simplifying assumption that propensities to consume from labor and profit incomes are constant (at levels y, and yK respectively). Equation (3.1.6) sets the government deficit equal to H*. The simplest assumption is that the deficit is financed by borrowing from the Central Bank, adding the amount H* to the money base (or stock of "high-powered" money). For the mo- ment, we assume that the interest rate varies enough to permit monetary emissions to be absorbed.2 Obviously the increase in money stock will have some medium-term feedback on inflation, but its consideration is deferred to the next section. Finally, equation (3.1.7) defines investment as the rate of growth of capital stock. Evidently the capital growth rate g is 2. The actual money stock, however defined, will be related to the money base H through well-known multiplier formulas that possibly respond to the interest rate and that are based on institutional characteristics of the banking system. We are consolidating the banks with the rest of the private sector and thereby avoiding the need to use credit multipliers. This approach follows Christ (1968), who pointed out the importance of constraints like (3.1.6). For an analysis which takes the banking structure into account, see Hansen (1973). A SIMPLE MODEL 53 determined by I and K at any point in time, but as the I/K ratio shifts in response to variations either in investors' anticipations or in available savings, g will shift as well. Behavioral assump- tions underlying the change in g are discussed in some detail below.3 If one eliminates PX from (3.1.1) and (3.1.2) and applies (3.1.5) and (3.1.6), it follows that PI + H* = (1 - y) wL + (1 - yx) (1 - t) aPX (3.1.8) = (1 - yL) wL + (1 - yK) rPK so that private saving (on the right-hand side) equals investment minus government saving, which is equal to -H*. This savings- investment identity represents Walras's law in the GNP model and is consistent with national income accounting conventions. It is also the most basic identity of all, and we rely on it heavily in interpreting our empirical results. Log-Differential Form of the Model Since economists like to think in growth terms, it is useful to throw the table 3-2 equations into logarithmic differential form. So expressed, the model gives growth rates of endogenous vari- ables consistent with the growth of exogenous variables, spec- ified from the outside. The log-differential equation system is of course linear in all growth rates, and it is easy to see how the variables interact in it, as exemplified below. In applications, it also seems to give fairly precise forecasts of endogenous vari- ables in the level form (calculated by applying the consistent set of growth rates from the log-linear system to base-year levels) for a period of two or three years.4 In a later section we discuss how such forecasts can in turn be fed into an equation-solving com- puter algorithm to get a full solution to the identities in the 3. Variables such as g appear in the planning model literature under the name of "stock-flow conversion factors" (see Manne 1963) and are usually assumed to be determined technologically. As mentioned in the text, we prefer to treat the macroeconomic g as behavioral, determined by availability of saving in Cam- bridge, Massachusetts, models and by investors' "animal spirits" in the Cam- bridge, England, variety. 4. This was the period used by Johansen (1974) in one pioneering exercise in forecasting based on log-changes. We found it appropriate in our own work, in the sense that three-year forecasts of variables were still quite close to level solutions of the full model. Accuracy seemed to deteriorate fairly rapidly for periods longer than three years. 54 Theoretical Framework forecast year. For the moment, however, we concentrate on see- ing how the log-differential version of the GNP model operates. Table 3-3 gives the relevant equations, in the form of growth rates of the variables, that is, X' is the rate of growth of X or (dXldt)IX. Equation (3.1.1') is the result of differentiating (3.1.1) after the price P has been canceled out. This equation implies that dX = dC + dl + dG, and this is transformed to (3.1.1'). To get to (3.1.2c') and (3.1.2p') note that the log-differential version of equation (3.1.2) can be written after a bit of manipula- tion as 1t P' - aL'- ak (r +P' + 1 t tt') + 4E =aLL' + aK K - X' + E where the shift variable E can be added to both sides without upsetting the exhaustion-of-product identity. When set to zero, the left side of this equation is a natural expression for a cost- based Laspeyre index of the price level, with cost-decreasing technical progress E included. This is equation (3.1.2c'). But then the right side must also be set to zero to maintain product exhaus- tion, and the resulting equation (3.1.2p') is equivalent, at least in the first order log-differential approximation, to a neoclassical aggregate production function with Hicks-neutral technical change. Double-entry bookkeeping in the price-equals-cost identity (3.1.2) assures that if weighted log-changes in costs are constrained to sum to the log-change in the price level, then weighted log-changes in inputs will just exhaust the output change. Log-differential cost and production functions are born as twins from the same accounting identity. In the rest of table 3-3, (3.1.3') and (3.1.4') are merely the log-differential versions of the corresponding equations in table 3-2. Equation (3.1.5') also follows readily from (3.1.5) when one observes that toL = wLIPX PC PC/PX and similarly for the profit recipients' share in total consumption. The last equations, (3.1.6') and (3.1.7'), follow from their level- form counterparts in straightforward fashion. A SIMPLE MODEL 55 The table 3-3 system comprises eight independent linear equa- tions with fifteen variables. In principle, seven additional restric- tions are required to close the system, either via specification of exogenous variables or incorporation of extra equations. In addi- tion, the model structure itself limits the possible ways it can be closed. For example, with exogenous specification of the rate of technical change E, equation (3.1.2p') shows that only two of the model's three quantity-side log-changes-L', K', andX'- can be fixed independently.5 A similar observation holds regarding cost and price changes and equation (3.1.2c'). Finally, factor shares cannot vary independently, as can be shown by multiplying (3.1.3') and (3.1.4') respectively by a, and aK, summing, and applying (3.1.2c') and (3.1.2p'). This manipulation gives aL (aL)' + aK (aK)' = 0, (3.1.9') that is, the adding-up properties of the model constrain differen- tial changes in the factor shares to sum to zero. In effect, only one factor-share change can be specified exogenously, or at least if both are constrained they must satisfy the adding-up condition. As we will see, neoclassical factor demand equations automat- ically require factor-share changes to satisfy (3.1.9'). To illustrate other characteristics of the model, we must intro- duce more detail. The following subsection is devoted to this task, using log-changes in the profit tax rate t' to "drive" the model under our two sets of rules for closing it.' Closing the Log-Differential Model Assume that the level of investmentI is fixed at any point in time. Then with the historically given level of capital stock and our assumption of full capacity utilization, the capital growth rate K' is predetermined by definition: K' = I/K. For simplicity in medium-term forecasting, we also assume that the rate of growth of employment L' and the technical progress rate E are fixed exogenously (although if desired the identity approach could easily be extended to include endogenous determination of these 5. There is an implicit assumption that neither unintended stock accumula- tion nor changes in capacity utilization are important. In short-term forecasting this could be remedied by adding an explicit variable (perhaps with a correspond- ing behavioral equation) for stock changes, or else by incorporating variations in capacity utilization into the residual E. 6. For a somewhat similar analysis, see Asimakopulos and Burbidge (1974). 56 Theoretical Framework growth rates by appropriate supply functions). From (3.1.2p'), the output growth rate X' then becomes an endogenous variable. To save symbols in the manipulations which follow, we set e equal to zero for the moment, and normalize prices by making P' equal to zero as well. To close the model, three additional restrictions are required. After the Cambridge, England, economists Keynes, Kalecki, and Kaldor, we call a K specification (KS) one in which two of the three government policy variables G', t', and H*' and the log- change in the capital stock growth rate g' are set exogenously. These assumptions boil down to stating that the government respects its budget constraint (3.1.6'), and that at any moment entrepreneurs (or the planner) are modifying the rate at which they accumulate capital in line with changes in expectations and investors' "animal spirits." Together with the predetermined value of K', specification of g' makes the growth rate of invest- ment ' endogenous from (3.1.7'). Also, if we let H*' be endoge- nous, we can insert equation (3.1.5') for the consumption growth rate C' into (3.1.1'), use the production function (3.1.2p'), and come up with an equation in which the "factor price" growth rates w' and r' depend on the already determined variables I', G', L', and K'. The log-differential cost function (3.1.2c') gives another equation for the same two variables. Solving for r' and substituting the result into (3.1.4') finally gives 1 aK I - 1 1 f ± GG' - yKaKtt - [ 1- yaL - yKaK (1 - t)] X'}. This equation shows that an increase in the investment growth rate I' is associated with growth in the profit rate and capital share. Growth in the profit tax rate t' reduces these variables. If, however, the immediate tax proceeds are respent by the govern- ment (so that dG = aKPXtt'), then the capital share will increase with this sort of balanced-budget expansion of government activity. These are standard results for a KS model, arising from its "widow's cruse" specification in which wage recipients save less than those who receive profits, so that the denominator of the fraction outside the braces in the above equation is positive. With the investment growth rate predetermined, even if a tax increase takes income away from capitalists initially, the profit rate then A SIMPLE MODEL 57 has to rise to keep up saving. By the same sort of reasoning, the capital share falls if the output growth rate X' is higher-with more total output becoming available, a smaller share has to be diverted to high savers via high incomes for them. Another fiscal effect is that profits do fall if dG is enough smaller than dt. Contrary to most KS models this can occur be- cause from (3.1.6) decreases in monetary emissions H* must ac- company tax revenue increases if government spending does not change. This government saving (or decrease in growth of private money stocks) can in principle support investment as effectively as private saving. How it is made available to investors is a finan- cial market question, which we do not go into here. The alternative way of closing the model is to adopt the neo- classical N specification (NS), in which factor shares are sup- posed to change in regular, predictable fashion as factor prices shift. Postponing until next section consideration of how tech- nical change affects derived demands for labor and capital, we assume as in cost-minimizing textbook production theory that the labor-output ratio, for example, depends only on the real wage, L/X =f(wlP). In addition, if there is smooth substitution between capital and labor at a predictable rate o-, then the log-change version of this equation becomes L' -X' = -o-(w' -P'), and the tradeoff parameter a- can be interpreted as the elasticity of factor substitution, defined in the usual way. If w' - P' is added to both sides, this expression becomes aL' = (1 - a-) (w' - P'). (3.1.10') The corresponding equation for the capital share is + I tV). (3.1.11') (Note that P' does not appear here, since both output and the existing capital stock are assumed to have the same price.) Adding (3.1.10') and (3.1.11') to the equations of table 3-3 gives an NS model. To save algebra, we adopt the often-applied special case in which a- = 1 and factor shares stay constant-we are saying either that there is an aggregate Cobb-Douglas production function or that we expect factor shares will not be shifting signi- ficantly in any case. In the Cobb-Douglas (or any other neoclassical) case, it is easy to verify from (3.1.10') and (3.1.11') that factor-share growth rates satisfy the adding-up condition (3.1.9'). Hence, only one of the 58 Theoretical Framework two factor demand growth rate equations is really independent. This additional restriction forces some K-specification exogenous variable to become endogenous-the usual candidate is the log- change of the growth rate of capital g'.7 Now we can examine the impact of raising the profit tax and thus set t' > 0. Equation (3.1.4') and the Cobb-Douglas assump- tions show immediately that the profit rate rfalls as the tax rate increases, regardless of whether the government respends the revenue. The balanced-budget theory of an increasing profit share does not hold in an NS model. The reason is that factor payments (and therefore consumption) are fixed by marginal pro- ductivity assumptions, and investment is endogenous. A tax in- crease raises government saving, and this leads in part to an increase in investment, in part to a fall in "required" private saving and thus in the profit rate. The increment in investment is surprising and bears closer study. In the Cobb-Douglas N specification, wage and profit growth rates from (3.1.3') and (3.1.4') are w' = X' - L' and r' = X' - K' - [t/(1 - t)] t'. Substituting these expressions into (3.1.5') gives C' = (1/Cc) -7Ka,tt' + [yLL + y,aK(1 - t)] X'}t so that consumption growth declines as the tax rate increases. Since output growth X' has been determined from the "produc- tion function" (3.1.2p'), the growth rate of investment I' must increase from (3.1.1'). An increase in the profit tax leads to a decrease in the profit rate, a decrease in consumption, and, para- doxically, to higher investment. Such a perverse investment re- sponse is no less disquieting than the widow's cruse in KS models. It is a standard feature of full-employment NS models (see Sen 1965) and illustrates once again how important are hidden implications of seemingly innocuous model-closing assumptions. Finally, note that in both forms of the model discussed so far, output growth X' is fixed in the short run by predetermined variables in equation (3.1.2p'). However, the log-change of capital stock growth g' is exogenous in the K specification but 7. Again we abstract from unintended stock changes. One can have neoclas- sical production specifications, exogenous capital formation, and endogenous unplanned inventory accumulation all together in a model, but the latter is a very small vessel in which to accumulate all the flows resulting from changes in the other variables, especially in the medium run. A SIMPLE MODEL 59 endogenous (determined by "productivity and thrift") in the N specification. Over time, therefore, growth in output can differ between the two models because they determine rates of capital accumulation in different ways. As we shall see in the empirical analysis of Brazil in the following chapters, this potential diver- gence in medium-term output growth rates can prove to be highly significant in practice. Calculation of a New Level-Form Solution The foregoing discussion shows that, at least in its growth rate version, when the identity model is closed in different ways it can produce responses to tax changes of opposite sign, and other surprises. Since neither set of hypotheses on closure is exactly well founded, a certain humility in recommending policy is sug- gested, particularly since we will verify that similar problems occur in level-form solutions of the models as well. We now describe briefly how these can be computed. To get levels of variables in an N specification, some paramet- ric form for an aggregate cost or production function has to be used. The simplest is the standard constant elasticity of substitu- tion (CES) function, derived on the assumption that the parameter we label o- stays constant even for large relative price changes. The level form of the labor demand equation becomes LIX = 6(w/P)-- where ( is a constant of integration (usually called a distribution parameter). There is a similar equation for capital demand, and the corresponding production and cost func- tions themselves are well known. One way to solve the level model is to take the value of to calculated from the log-differential version as an initial guess in an iterative scheme. The above equation for LIX then gives a guess at X and the capital demand equation gives r (both these calculations depend on the exogenously specified growth of labor L and capital K). One can then test if the price-equals-cost identity (3.1.2) is satisfied to an acceptable degree of precision; if not, the guess at w can be modified until the identity holds. For the NS version of the Brazil model discussed in the next section, three-year projections of price and output changes from the log- differential equations of table 3-6 satisfied the price-equals-cost identity in the forecast year to within about 1 percent. In the KS model, price and quantity projections are not tied together by marginal productivity conditions as in the labor de- 60 Theoretical Framework mand equation given above, giving rise to potential inconsis- tency when both (3.1.2p') and (3.1.2c') are applied with arbitrary finite changes of exogenous variables but with factor shares un- changed from their base-year values.' One way to finesse this problem is to treat either the price projection from (3.1.2c') or the output projection from (3.1.2p') as exact and to calculate the other variable from identities holding in the forecast year. We chose to get a forecast of X from the integration of (3.1.2p') over time with constant factor shares. Then a guess at w from the other log- change equations gives a new value for aL from (3.1.3), and aK becomes 1 - aL. We calculate r from (3.1.4) and C from (3.1.5). Investment I comes from (3.1.7), and we can now determine whether the output demand-supply balance is satisfied. If not, w must be modified accordingly. As we see in more detail below, higher investment demand leads to a lower real wage via "forced saving," which is in the spirit of KS models. In the Brazil exer- cise, a few iterations of the type just described were usually necessary to compute new price levels, since the projections over a three-year period from the log-differential model were usually between 10 and 20 percent different from the final values of prices in the level solutions. A Model for Brazil The identities in the preceding section can be extended to more elaborate descriptions of the economy, although bookkeeping complexity and implicit behavioral assumptions increase apace as this is done. In this section we discuss a model put together to make medium-term forecasts for Brazil. The model still has only one production sector, but detail regarding balance of payments and tax structures makes it too cumbersome for algebraic solu- 8. That is, the factor shares turn out to be weights in a first-order approxima- tion to changes in the exhaustion-of-product identity (3.1.2). For finite changes of prices and quantities, this approximation will of course be in error, as the factor shares themselves will vary. One major difference between the KS and NS models is that the latter restricts price and quantity changes in such a way as to make factor-share changes vanish to second order in equations such as (3.1.2c') and (3.1.2p'). (For a straightforward demonstration in a model similar to ours, see Jones 1965.) This goes far toward explaining the difference in accuracy of fore- casts over three years of the two log-differential specifications, as discussed in the text. A MODEL FOR BRAZIL 61 tion. For that reason, numerical simulations are used to analyze its characteristics in the following two chapters. Variables and level-form equations are listed in tables 3-4 and 3-5. We begin by summarizing the most important extensions beyond the GNP model, and later discuss the log-differential ver- sion of the Brazil model and how it is closed. Equation (3.2.1) is analagous to (3.1.1), except that three more demand categories are included: exports net of competitive im- ports,9 government demands for investment goods, and deprecia- tion of the stock of domestically produced capital goods. The coefficient 5x in equation (3.2.1) refers to physical depreciation, but we assume in (3.2.4) that it is also the legal depreciation rate used in calculating the user cost of capital. In general, the two concepts need not be the same. As before, we omit consideration of stock changes for simplicity in medium-term forecasting, but otherwise equation (3.2.1) corresponds to the usual GNP account- ing conventions. Equation (3.2.2) is the balance of payments written in world prices, scaled in terms of the domestic currency by the exchange rate p. Movements in import and export prices can of course affect the model projections through this equation, along with quantum changes. It is convenient to treat Brazilian imports as noncompetitive (see note 9). Because their demand-generation mechanisms are different we distinguish three types- complementary imports of intermediates (M), capital goods for new investment (IF), and consumption goods (CF). Along with a depreciation term for existing imported investment goods, these appear (multiplied by their corresponding world prices) on the right of (3.2.2). The remaining term F represents capital inflows in the simplest possible balance of payments definition, the "foreign savings gap." Both here and below in the money supply equation, we omit exogenous reserve changes, debt service charges, and a host of other financial market complexities. Full treatment of financial flows, although easy to include in the identity framework in principle, would greatly complicate an already messy model. We shunned the additional effort for this reason. 9. In the Brazilian case, competitive importation is in principle illegal be- cause of legislation-summed up as the "law of similars"-intended to prohibit the import of products similar to national ones. In practice, the laws are not followed rigorously, but in what follows we in fact treat all imports as noncom- petitive. 62 Theoretical Framework Table 3-4. Symbols in the Brazil Model Symbol Definition X Domestic production C,X CF Consumption demands for domestic and noncompetitive imported goods Ix, IF Demands for domestic goods and noncompetitive imports for capital formation 8x, 8F Depreciation rates of domestic and imported capital stocks Kx, KF Stocks of domestic and imported capital goods E Exports less competitive imports of domestic goods CG, lG Government purchases of domestic goods for consumption and investment P Domestic price level p Exchange rate M Volume of noncompetitive intermediate imports irE, irM, C, 7rF World prices of E, M, CF, and KF or IF PM, Pc, PK Domestic prices of M, CF, and KF or IF F Capital inflow (balance of payments deficit in world prices) V Value added (net of taxes) per unit output tV, tL, tK Tax rates on value added, payments to labor, and profits w Wage rate r After-tax profit rate OX, 0F, OL User costs of the two types of capital and labor Cr, aL, x, aF Shares of noncompetitive imports, labor, domestic capital goods, and imported capital goods in value added net of taxes L, LG Employment levels in production activity and in government service YL, YK Labor and profit incomes Z Total value of capital stocks DL, DK Direct taxes on labor and capital incomes SMD, SDK Elasticities of direct taxes with respect to labor and capital incomes (L, (K Constant terms in tax functions Q Transfers from government to labor income recipients Sca, Sc Elasticities of consumption with respect to wealth and the "real" interest rate 6c Constant term in the consumption function B Outstanding stock of government price-indexed bonds R Total wealth in base-year prices C Value of total consumption X, 4F Shares of domestic output and noncompetitive imports in consumption YL, YK Propensities to consume from labor and capital incomes gX, gF Rates of growth of domestic and imported capital stocks G Total government expenditure T Government tax revenue H*, B* New emissions of high-powered money and of bonds H Stock of high-powered money (the money base) A MODEL FOR BRAZIL 63 Symbol Definition SHx, SHi Elasticities of money demand with respect to real output and the interest rate Constant term in the money demand function i Rate of interest E Rate of technical progress in Hicks-neutral form A Anticipated rate of inflation Table 3-5. Level Form of the Brazil Model PX = PC, + Ix + P5xKx + PE + PCG + PIG (3.2.1) pirEE + pF = p71wM + P1rCCF + peF (IF + 8,Kp) (3.2.2) OL = w(1 + tL) (3.2.3) Ox = - + 8x P (3.2.4) F 1 tK(3.2.5) VX = P,M + O,L + OxKx + OFK, (3.2.6) P = (1 + tv) V (3.2.7) aWVX = PMM (3.2.8) a,VX = OL (3.2.9) axVX = OxK (3.2.10) aFVX = OpKF (3.2.11) YL = w(L + LG) (3.2.12) Z = PKx + FKK, (3.2.13) Y= rZ + PB (3.2.14) DL = GLS_ (3.2.15) DK YK (3.2.16) R = Kx + K, + (HIP) + (B/i) (3.2.17) C = [y,. (YL -DL +Q) + YK (YK DK)J C (i/A), (3.2.18) AxC = PC. (3.2.19) OF,C = PcC, (3.2.20) Ox + 4F 1 (3.2.21) lX + IG = gxKx (3.2.22) Ip = gFKp (3.2.23) G = (1+tL wLG + FCG + PB + (F = pirE) E + P1G + Q (3.2.24) T t v ,VX + tLtWv(L +LG) + 1- r + TV - P7TM) M + (PC - P7TC) C" + (FK P7TK) (IF. + 8,KF + DL + DK (3.2.25) G - T = H* + (PB*1i) + pF (3.2.26) H 6H~PX S-S(3.2.27) 64 Theoretical Framework Equations (3.2.3) to (3.2.7) are analagous to (3.1.2), setting value added at factor cost equal to the total value of inputs."o The first three equations are simply definitions of user costs of labor and nationally produced and imported capital goods. Labor cost includes a social security tax on wages (responsible for about 20 percent of total tax revenue in Brazil), while capital costs are determined by the rate of profit, the profit tax, and depreciation charges. Along with intermediate imports (valued at a within- country cost Pm, which can differ from the world cost pIr, because of tariffs), these inputs exhaust total value added in (3.2.6). Final output price P is determined by an indirect tax markup on the price of value added V in (3.2.7). Equations (3.2.8) through (3.2.11) define the shares of the in- puts in value added, as do (3.1.3) and (3.1.4) in the simple model. In the N specification of the Brazil model, these equations in log-change form are used to set up derived demand relationships such as (3.1.10'). Details of the specification appear in the sec- tion below on "Price-Sensitive Production Responses in an N Specification." Equations (3.2.12) through (3.2.18) link factor payments to the total value of consumption C. Labor and capital incomes are defined by the three equations beginning with (3.2.12). Capital incomes are assumed to include interest payments on govern- ment bonds, which can be issued along with money to cover fiscal deficits. Because indexation is widely applied in Brazil, the bonds are treated as inflation-corrected consuls, each paying the real equivalent of one cruzeiro a year in base-year prices. Be- cause of the escalator, total income from bonds is PB, that is, the stock of bonds multiplied by the price index. These government interest payments show up in chapter 5 as important determi- nants of saving in the NS Brazil model. Their importance is em- phasized by Blinder and Solow (1974). Equations (3.2.15) and (3.2.16) describe in convenient constant elasticity form the relationship between direct tax revenues and incomes. Here as elsewhere the distinction between capital and labor incomes is supposed to approximate the personal income 10. Our usage is slightly nonstandard, since we include the cost of raw material imports in "value added." This simply reflects reality in an underdeveloped country, where production is strongly constrained by import requirements. Our subsequent assumption of substitutability between imports and other inputs per- haps understates the importance of this constraint. A MODEL FOR BRAZIL 65 distribution-the constant &, is smaller than 6K in a crude repre- sentation of a progressive direct tax- system. The next equation (3.2.17) defines total wealth R (in base-year prices); along with the nominal rate of interest deflated by the anticipated rate of inflation, i/A, this is supposed to have some influence on the value of consumption C in (3.2.18). We assume that the elasticities of C with respect to wealth and the real interest rate are constant." The three equations beginning with (3.2.19) define shares of consumption of national and noncompetitively imported goods in the total budget and require that they sum to one. To complete specification of consumer demand responses, an additional equa- tion is required to determine one of the two budget shares. De- tails of how this can be done in a convenient log-differential specification appear in the section on "Consumer Demand Responses." After the capital growth rate definitions (3.2.22) and (3.2.23), equations (3.2.24) through (3.2.26) give the government accounts. The total value of government expenditure G comprises govern- ment consumption and investment, bond interest payments, the total subsidies paid to exporters (based on the difference be- tween the domestic price P and the domestic value of export receipts p,TE), government transfer payments, and wages and em- ployment taxes for government employees. These are also in- cluded in total tax receipts T, along with value added and profit taxes, direct taxes, and tariffs based on the difference between world and domestic prices of imports.12 Equation (3.2.26) shows that the difference between government expenditures and tax receipts is met by monetary emissions H*, new bond issues with value PB*/i, and capital inflows pF, which in the first round after entry into the country pass through the Central Bank. Equation (3.2.27) closes the model, setting the stock of high- powered money H equal to its demand. As written, this equation 11. The nonstandard specification of interest rate deflation by anticipated in- flation, i/A instead of i - A, allows this to be done easily. 12. In formal terms one can set Pi = p (1 + 7j) 7s, where i indexes traded goods and T is an ad valorem tariff. Unchanging tariffs imply that Pj' = p' + i,', which can be used in specifying price changes in the log-differential model. Such rela- tionships are used informally in the K specification of the present model (since they will turn out to contain only exogenous log-changes), but are treated as explicit restrictions in the N specification (where both p' and the P,' are endogenous). 66 Theoretical Framework makes the demand for "real balances" (HIP) depend in constant elasticity form on real output and the nominal interest rate. In effect, it determines the interest rate, which then feeds back weakly into consumer demand and private saving. This feedback aside, (3.2.27) serves mainly as a check on consistency of mone- tary policy, with the real configuration of the economy deter- mined by other parts of the model. The savings-investment identity in the present model is (YL - DL + Q) + (YK - DK - C = PF + PKIF + H* + (PB*/i), so that private saving (not including depreciation) equals net capital formation plus that part of the current government deficit financed from within the country. Another version of the same identity is PIx + PKIF (YL - DL + Q) + (YK - DK) - C + (T - G) + pF. This sets capital formation equal to the sum of private net saving, government saving, and capital inflows. We rely heavily on this equation in interpreting our numerical results in the next two chapters. The Log-Change Version of the Model and a K Specification Table 3-6 sets out the growth rate version of the Brazil model-the equations at first glance resemble so much hen scratching, but in fact are merely more complicated variants of their counterparts of table 3-3. The dual log-differential approxi- mations to cost and production functions appear as (3.2.6c') and (3.2.6p'). These equations follow from the growth rate version of the exhaustion of product equation (3.2.6), under the same as- sumptions which gave us (3.1.2c') and (3.1.2p'). Overall account- ing consistency requires that differential changes in the factor shares sum to zero, aL aL' + am am' + ax ax' + aF aF' = 0. (3.2.28') Any specification of exogenous variables must satisfy this restric- tion, even if it is not independent of the rest of the model. In table 3-4 there are fifty-seven variables; in the log-change version with its twenty-eight equations (both the dual cost and production functions are included) twenty-nine additional re- strictions are required. For KS closure, these take the form of Table 3-6. Log-Change Form of the Brazil Model PXX' = PCxCx' + PIXIX' + PSXKxKX' + PEE' + PCGGG' + PIGIC' (3.2.V) pIrKE(1,E' + E') + pFF' = pirwM (irm' + M') + prCCF (7rtC + CF) + prFIFIF' + p7riFFKFKF + piTF (I, + 8,K,) -F (3.2.2') OL' =W'+ tL iL' (3.2.3') i + t 1= P, + rM( - tK) (r'+ t, ' (3.2.4') 8, +r/(1 -t) OF' = P' + r'(1 tK ( + tK't, (3.2.5') r + r/(1 - tK) 1 tK V = -E + aMPM' + aLOL' + ax Ox' + aFOr (3.2.6c') X' = E + amM' + a L' + axKx' + aK,' (3.2.6p') p' = V + t, tv' (3.2.7') 1 + tv a= PM' + M' - V' - X' (3.2.8') aL' OL' + LV - - X' (3.2.9') ax' Ox' + Kx' - V' - X' (3.2.10') ai'= OF' + KF' - V' - X' (3.2.11) LL YL' W' + LW + L L' (3.2.12') L +LG L +LG ZZ' = PK, (P' + K,') + PKF (PK' + K,') (3.2.13') YKYK' = rZ (r' + Z') + PB (P' + B') (3.2.14') DL' = SDLYL' (3.2.15') DK' = SDKYK (3.2.16') RR' = KxKx' + KFKF' + (HIP)(H' - P') + (Bli)(B' - i') (3.2.17') C' = [yL (YL - DL + Q) + YK (YK DK)]1 [YLYLYL' - jLDLDL' + yQQ + YKYKYK' - yKDKDi + SKCRR' + Sci (i' •A') (3.2.18') 0,= P' + ci' - C:' (3.2.19') OF' = PC' + C' - C' (3.2.20') Ox(x' + 4,F' = 0 (3.2.21 ) I + X, = gx' + Kx' (3.2.22') Ix+IG ix +G I gF + KF' (3.2.23') GG' (1 + tL)wL( tL+w+ LG' +P(CG + E + B \1 +t,/ + IG)P +'1G'G' +PCGCG' + PBB' + (P - p7rEE)E - pEE (p' + rE) + QQ' (3.2.24') TT' =tvVX(t,' +V' +X') +tLw(L + LG) W' +tL' + L' + LG' t+ tK L LGL t1 1 t4 L+LG L+LG /1 KklK + r' + Z' + (PM - PITM) MM' + PMMPM' - PirMmir', + (PC - P7rC) CFCF' + PCFC' - P1TCCrC' + PK (IF +8FKF) PK + (PK_ 7 F K)IF :,8F r F, F "I+ 8Ä KF' FIF F FJK, TF ~ p[ vryM + ncCF + ir (I, + ?FKF)] p' + D ' + KDK' (3.2.25') GG' = TT'+ H* (*)' + (PB*li)[(B*)' + P' - i'] + pF (p' + F') (3.2.26') H' = P' + S,xX' + Sm,i' (3.2.27') 67 68 Theoretical Framework exogenous specification of variables. It is convenient to group the "obvious" exogenous growth rates as follows: "Naturally exogenous" variables (6): Er,', 7TM' 7c' 7rK', E, A' Tax and expenditure variables (10): IG', CG%> P', C' P', , tW, tK',Q', G' Log-changes determined by stock-flow relationships at the initial time point (4): Kx', K,', H', B'. These make up a total of twenty variables. Addition of either CG', , W, L , Q, LG', TM', TK, TC' Log-changes determined by stock-flow relationships at the initial time point (4): Kx', K,', H', B' Other exogenous variables (6): 4x,' (determined by [3.2.37']), w', F', H*', E', L'. Consumer Demand Responses The next topic is how to determine ox' and F' as prices and the value of total consumption change, for use in both versions of the model. In developing countries one usually has some idea of the shares in the total consumption budget of various classes of goods in the economy, plus some hazy notion of income elastici- ties. Brazil is no exception to the rule. The consequence is that fairly strong hypotheses about consumer behavior must be im- posed in advance to produce usable demand forecasts. Begin by observing that income elasticities in principle de- scribe demand log-changes in response to log-changes in an in- dex of real income, C' - OxP' - OFPc'. Let -x denote the income elasticity of demand for X and -xx and 77F denote compensated price elasticities (with real income constant); the log-change in consumption of good X is Cx' = qX (C' - 4XP' - OFPC') + qxxP' + qXFPC' + AX where 6x is an exogenous shift term. There is a similar equation for CF'. A fair amount of international evidence (summarized by Sato 1972, for example) suggests that quantities such as qx and xx are related as qxx = -ocqx, where oc has a value around one-half. If we impose this relationship and (3.2.21') on the demand equa- tions we find that after a bit of algebra they can be written as ox' == (-qx - 1) C' + (1 - 71x(Ax - -xcC) P' 74 Theoretical Framework + 4 0' C - )c' + 3x, (3.2.37') and similarly for OF'. Implicit are the following restrictions on parameters: OX 7)X + OF 1F = 1 OXOX-3 + OFN3 = 0 _qF= O'F O-C71FIOX WX= OkX OC 711XIOF The first of these conditions is called Engel aggregation in the consumer demand literature; it assures that the budget constraint will be satisfied after income changes. The second condition does the same with regard to taste shifts. The third and fourth conditions define the cross-price elasticities in terms of the pa- rameter oc. Equations (3.2.37') and (3.2.19') through (3.2.21') form a consis- tent demand system. Alternatively, one could use (3.2.37'), the similar equation for OF', and (3.2.19') and (3.2.20'). In either case, the demand specification amounts to making the (,' price depen- dent in a plausible way which also satisfies the adding-up condi- tion (3.2.21'). As always, the basic identities remain in force.17 Solution Techniques for the Level Form Although they are complex algebraically, both versions of the Brazil model were solved in their level form by the methods sketched in the section "Calculation of a New Level-Form Solu- tion." Basically, we used the guess at the price level P coming from the log-change model to generate values of all other vari- ables, and then iterated until the product demand-supply balance (3.2.1) in the K specification or the exhaustion-of-product equa- tion (3.2.6) in the N specification was satisfied. We began these procedures with a completely specified base year (1960 in 17. The method for specifying demand elasticities described here is from Frisch (1959) in its essentials, although he gives an explicit derivation in terms of utility functions. If one assumes that own-price elasticities stay constant over time, then one can update estimates of the Frisch parameter ac as the demand basket changes by letting it satisfy -0c = XXOX + 7rFFOF. Updating income elasticities according to the rule Tj = -c,, then assures that Engel aggregation is satisfied. This procedure can be justified in terms of the direct addilog utility function (Sato 1972). REFERENCES 75 chapter 4 and 1970 in chapter 5), and took three-year time steps to generate forecasts for the subsequent decade. In each of these years, we also generated multiplier matrices showing responses of endogenous growth rates to exogenous ones, which proved quite useful in calibrating the model.' References Asimakopulos A., and J. B. Burbidge. 1974. The short-period incidence of taxation. Economic journal 84:267-88. Blinder, Alan S., and Robert M. Solow. 1974. Analytical founda- tions of fiscal policy. In A. S. Blinder and others. The eco- nomics of public finance. Washington, D. C.: Brookings Insti- tution. Canas, Jorge. 1972. Short-term economic theory and policy: The Chilean case, 1964-70. World Bank, Development Research Center, restricted circulation document. Christ, Carl F. 1968. A simple macroeconomic model with a gov- ernment budget restraint. Journal of political economy 76(1):53-67. 18. Two additional points should perhaps be made here. First, as mentioned above, we set H*', the growth rate of money emissions, exogenously at the beginning of each forecast period. If emissions at the beginning of the period are Ho*, then at the end of the period the level of emissions and money stocks Hr* Is HT* = Ho* exp (H*' T) where exp (H*' T) denotes the exponential function and T is the length of the forecast period. Straightforward integration shows that the money stock HT is given by 1 H= -(H* - Ho*) + H, where H. is the initial money stock. Second in the NS model, we used two-level CES cost, production, and derived demand equations in calculating the level solutions. These are all well known, though it is worth noting explicitly that labor demand is given by L/x = (fxLV/06)" where T is the effective labor force, that is, the employed population multiplied by the increase in labor productivity cumulated over the forecast period, and fL is a scaling factor increasing at a rate EI/L over time from an initial value calculated from the base-year labor share. 76 Theoretical Framework Clark, Peter B. 1975. Intersectoral consistency and macroeco- nomic planning. In Charles R. Blitzer, Peter B. Clark, and Lance Taylor, eds. Economy-wide models and development planning. New York and London: Oxford University Press. Frisch, Ragnar. 1959. A complete scheme for computing all direct and cross demand elasticities in a model with many sectors. Econometrica 27:177-96. Hansen, Bent. 1973. On the effects of fiscal and monetary policy: A taxonomic discussion. American economic review 63: 546-71. Johansen, Leif. 1974.A multi-sectoral study of economic growth. 2nd ed. Amsterdam: North-Holland. Jones, Ronald W. 1965. The structure of simple general equilib- rium models. Journal of political economy 73:557-72. Lewis, W. Arthur. 1966. Development planning: The essentials of economic policy. New York: Harper and Row. Manne, Alan S. 1963. Key sectors of the Mexican economy, 1960-70. In A. S. Manne and H. M. Markowitz, eds. Studies in process analysis. New York: John Wiley. Sato, Kazuo. 1967. A two-level constant elasticity-of-substitution production function. Review of economic studies 34:201-18. -. 1972. Additive utility functions for double-log consumer demand functions.Journal of political economy 80:102-24. Sen, Amartya K. 1965. The money rate of interest in the pure theory of growth. In F. H. Hahn and F. P. R. Brechling, eds. The theory of interest rates. London: Macmillan and St. Martin's Press. 4 Brazilian Growth and Distribution in the 1960s: An Identity-Based Postmortem Eliana A. Cardoso BRAZILIAN ECONOMIC DEVELOPMENT during the 1960s is the sub- ject of a vast literature. Economists agree that this period con- tained three distinct phases: (1) in the years before 1962 indus- trialization based on vigorous import substitution had its finale, accompanied by rampant inflation, (2) from 1963 to 1967 was a period of reduced capital formation and slowdown in industrial expansion with sharply declining inflation rates; and (3) from 1968 onward there was a sustained recovery of output growth accompanied by low inflationary pressure that lasted until 1974. There is, however, strong disagreement on the interpretation of these events. Was growth disequalizing or not? Was worsening in Brazilian income distribution the result of government's wage policy and stabilization? Was the structure of income distribution related to stagnation in the middle 1960s, and was the wage squeeze a necessary condition for the post-1968 economic recovery? Our identity-based model can scarcely put an end to a seem- ingly unending polemic and counterpolemic on these questions, but the results given below suggest that: (1) wage repression may 77 78 Growth and Distribution in the 1960s well have worsened the income distribution; (2) reductions in labor costs may be a very inefficient policy to induce employ- ment growth; and (3) growth of exports may also have been ac- companied by distributional shifts against the poor. It should be stressed that these findings rely both on the spe- cific formulation of the identity model and on the vagaries of the Brazilian national accounts. The latter have very great shortcom- ings not dealt with here but which should be kept in mind. There can be no statistics unless someone has first done the counting. In Brazil, however, often nobody has done that, and in such cases there are only informed estimates or more or less wild guesses. No one can cook a gourmet dinner out of a few rotten potatoes, but if nothing else is available, the cook will have to make do. In 1974 the Vargas Foundation revised its figures for the na- tional accounts at current prices, raising its previous GDP es- timates by 4.4 percent in 1959 and 18.3 percent in 1970. New figures are not available for the intervening years which are the subject of this chapter. In addition, as indicated in chapter 3, there are at least two possible ways of closing national accounting identities to make a complete model. If the Brazilian experience in the 1960s must be forced into one of these alternative Procrustean beds of economic theory, the one built up from Kaldor-Kalecki (KS) hypotheses is far less deforming. Even with a wildly implausible parameteriza- tion, the neoclassical (NS) model cannot reproduce Brazilian ex- perience as recorded by the official statistics. This last point is discussed in appendix B to this chapter, where the actual data are compared with the simulated data obtained from the Kaldorian and neoclassical solutions. Appendix A describes how data were put together for the base year (1960) in a manner which satisfies the particular set of accounting identities used here. The main part of this chapter is organized as follows: in the second section basic KS growth paths are discussed. Effects of wage reduction (third section) and of tax reduction (fourth section) on growth, employment, and labor share are then consid- ered. In the fifth section the relation of exports to recent Brazilian economic performance is briefly analyzed. Conclusions are sum- marized in the last section. Kaldorian Base Growth Paths In this section two Kaldorian base growth paths are presented, and one is chosen for comparison in the following sections. KALDORIAN BASE GROWTH PATHS 79 A complete listing of the assumed values of the exogenous growth rates for the base paths is shown in table 4-6 of appendix B, and table 4-1. in this section contains results for the key vari- ables from the KS model. The first four columns represent base path A, and the first three combined with the fifth represent path B. The only difference between A and B is that the growth rate of money wages (w') in 1966-69 is equal to 0.23 in A and 0.30 in B. To begin with path A, as can be seen in table 4-6, employment growth is nil in the first two periods, which correspond to the phase of slow growth, and in 1966-69 it is set at 0.046 a year. The residual e is set at 0.020 in 1960-63, 0.005 in 1963-66, and 0.012 in 1966-69. Low values for E were chosen because of the existence of excess capacity in 1963-66. The growth rates of domestic and imported capital stock are respectively 0.079 and 0.024 in 1960-63,0.068 and 0.024 in 1963-66, and 0.068 and 0.024 in 1966-69. The elasticity of noncompetitive imports, M, with respect to output is 1.0 and wages grow at 0.45 per year between 1960-63 and 1963-66 and 0.23 per year in 1966-69. Table 4-1. Values of Key Variables in Kaldorian (KS) Base Solutions Base paths A and B Path Path A B KS model 1960 1963 1966 1969 1969 1. Output (X)a 2.675 3.016 3.343 4.063 4.063 2. Output growth rate (X ') _b 0.040 0.034 0.065 0.065 3. Inflation rate (P') b 0.443 0.442 0.259 0.296 4. Profit rate (r) 0.175 0.165 0.153 0.172 0.160 5. Labor share (a,) 0.540 0.489 0.473 0.431 0.476 6. Private savings share 0.104 0.121 0.125 0.141 0.136 7. Government savings share -0.037 -0.046 -0.031 -0.048 -0.049 8. Capital inflow (F) 0.328 0.107 -0.017 0.169 0.362 9. Exchange rate (p) 0.110 0.573 2.209 4.062 4.062 10. Capital inflow/ export value 0.225 0.063 -0.008 0.062 0.143 11. Interest rate (i) 0.060 0.113 0.147 0.126 0.141 12. Government interest payments (PB) 0.004 0.408 1.781 4.369 4.881 Note: The data in the tables in this chapter refer to year-end values in the stated years. a. In billions of cruzeiros. b. The base year is [960 for figuring subsequent rates of output growth and inflation. 80 Growth and Distribution in the 1960s These assumptions generate output growth rates of 0.040 (1960-63), 0.034 (1963-66), and 0.065 (1966-69), as can be seen in the second line of table 4-1. In the third line we see that the inflation rate drops off abruptly from 0.443 a year in the first two periods to 0.259 in the last one. Failure to implement a satisfactory anti-inflationary policy had been haunting Brazilian governments since the early 1950s. In- flation in both 1959 and 1960 was in excess of 30 percent. In 1963 an acceleration of inflation and a decline of output growth oc- curred simultaneously. After 1964 minimum wage changes were maintained at a rhythm markedly inferior to the price level, curb- ing inflation but clearly not rooting out inflationary expectations as had been predicted by policymakers. From 1967 onward, however, inflation did stabilize at around 20 percent a year and output grew rapidly. The fourth line of table 4-1 exhibits a falling rate of profit between 1960 and 1966 and then a recovery in 1969. The labor share declines throughout the period, establishing an impressive difference between the 1969 and the 1960 values. The cause of this development appears in the next line in the shape of a steady increase in the proportion of private saving, not including depre- ciation, in the value of domestic output. (Such proportions are called savings shares in the tables.) As is well known, movements in private saving are mediated in Kaldorian models by shifts in the functional income distribution. The increase in the proportion of private saving corresponds to the official data until 1966. According to Baer (1965), the private sector was the principal source of saving in the economy during the early 1960s. He presents data showing that the proportion of private saving in the product increases from 6.6 percent in 1960 to 12.4 percent in 1963 and 14.1 percent in 1965. From 1965 on, this proportion declines. Private saving increased to compensate for declining capital inflows between 1960 and 1966, as can be seen in line 8, and to compensate for growing government dissaving in 1960-63 and 1966-69. (Similar estimates of government dissaving can be found in Fishlow 1973.) To understand the evolution of govern- ment saving it is necessary to know how tax revenue and public expenditure shifted over time. On the tax side, the elasticity of total revenue with respect to the value of output exceeds unity, so that taxes divided byPX rise KALDORIAN BASE GROWTH PATHS 81 from 0.197 in 1960 to 0.26 in 1969. Once again, this is consistent with the data presented by Baer (1965). According to his es- timates, direct plus indirect taxes divided by gross output rose from 0.20 in 1960 to 0.267 in 1968. This behavior in the tax system comes from growing tax rates on both labor payments and value added. The rate on payments to labor grows from 0.046 in 1960 to 0.113 in 1969 and on value added from 0.157 in 1960 to 0.20 in 1969. Estimates of taxes on labor payments can be found in Bacha and others (1972), and the estimated values for the value added tax are consistent with an appropriate average of Brazilian sales taxes and are related to the government fiscal reforms of the 1960s. The tax increases are also caused by the form of the income tax equations where the elas- ticities of direct taxes with respect to labor and capital income were set equal to 1.1. Not only did the tax share increase during this period but also government activity expanded. Government dissaving increases in 1960-63, falls until 1966, and increases again after 1967. This can be seen in line 7. On the financial side, money emissions H* grow at 0.55 in 1960-63, 0.40 in 1963-66, and 0.30 in 1966-69. Line 11 of table 4-1 shows that this implies a growing interest rate. Government bonds, unimportant in 1960, grow in significance from 1966 on- ward, and nominal payments to bondholders increase, as can be seen in the last line of table 4-1. The increase in the public debt is a response to governrient and foreign trade deficits and results from the interactions among expenditures and monetary and ex- change rate policies. The exchange rate p is assumed to increase at 0.55 in 1960-63, 0.47 in 1963-66, and 0.203 in 1966-69. These trends do not repre- sent real devaluation. On the contrary, p rises much less than the difference between world and domestic prices, so that it really appreciates in the period. Consider now the base Kaldorian path B represented by col- umns 1, 2,3, and 5 in table 4-1. All exogenous rates of growth are the same as before except for the rate of growth of money wages (w') in 1966-69, which is set equal to 0.3. From Bacha and others (1972), the growth rate of money wages in manufacturing was somewhere between 0.23 and 0.30 in 1966-69. The advantage of using a bigger rate in our model is that it generates an increase 82 Growth and Distribution in the 1960s in the average real wage rate, which in fact occurred, as the data indicate. The main difference between the solution labeled B and path A concerns the labor share in GNP. It falls markedly between 1966 and 1969 in path A, but in path B it remains constant after 1966 because total saving adjusts to growth in investment by an in- creased deficit in the balance of payments (line 8), and hence there is less private saving than before. The evidence seems to point to a noticeable rise in the ratio of profits to wages and salaries in the urban sector between 1960 and 1968, consistent with both paths.2 But if one also believes that the gap broadened in 1965-66 as a consequence of stabiliza- tion policies and that thereafter it did not increase markedly, path B should be preferred.3 Effects of Wage Reduction on Growth, Inflation, and Labor Share As shown by Fishlow (1973), after 1964 the Brazilian government prevented the minimum wage from growing at the rate prices and productivity were increasing. To implement wage restriction, an estimate of the expected inflation during the following twelve months was required. This so-called inflationary residual was consistently and deliberately underestimated so that real mini- mum wages fell by 20 percent between 1964 and 1967.4 In addi- tion, between 1960 and 1969 the disparity between managers' and employees' wages increased considerably. Accordingly, the average real wage in path B increased at a rate well below the growth rate of output. Nobody denies that the draconian wage control imposed by the Brazilian government served well the purpose of stabilization. It 1. Again from Bacha and others (1972), the average real wage in manufacturing grew at 0.02 a year from 1966 to 1969. More information on wage behavior during this period can be found in Fishlow (1973), who refers to unpublished data. Data on real wage rates in other sectors appear in chapter 10. 2. Comparisons between macro variables underlying table 4-1 and actual sta- tistics are presented in appendix B to this chapter. 3. This position is advanced by Fishlow, in a revision of his 1973 work, who quotes a study by Wells (1974). 4. A more detailed discussion of the mechanisms of government wage policy that induced a deterioration of the real minimum wage can be found in chapter 10. EFFECTS OF WAGE REDUCTION 83 Figure 4-1. Inflation Rate and the Labor Share P' 0.45 - w' 0.45 0.40 tv 0.357 0.35 1966 L' = 0 ' = 025 1966 L' = 0.025 0.30- 1966 L' = 0.045 1969 L' = 0.045 0.25 - 0.10 w'=0.20 1969 L' = 0.07 0.20 1969 L' = 0.095 - - II 1 1 I, 0.20 0.30 0.40 0.50 0.60 7. Source: Table 4-2. is the effect of such policy on growth, employment, and income distribution that has been the focus of spreading debate. The following exercises suggest that Brazilian wage policy cannot be justified on the ground that increased labor absorption induced by lower real wages helped the underemployed poor, as main- tained by Morley and Williamson (1975). Nor can it be argued that money wage repression served the purposes of accumula- tion, because the deterioration of income distribution in Brazil was not counterbalanced by a greater volume of aggregate na- tional saving.' As suggested by Oliveira (1972), however, it can be maintained that the reduction of wage costs kept profits high and permitted an increase in private saving (this can be seen in line 6, table 4-1). Figure 4-1 demonstrates the tradeoffs between labor share and price stabilization in a K specification. Each curve shows how, for 5. This second point has already been made by Fishlow (1974). Table 4-2. Inflation Rate and Labor Share for Different Employment and Wage Growth Employment growvth Wage 1963 1966 1969 1963 1966 1969 1963 1966 1969 growth L' = 0 L' = 0 L' = 0.045 L' 0.015 L' = 0.025 L' = 0.07 L' = 0.045 L' = 0.045 L' = 0.095 1963 w' = 0.45 P' = 0.443 P' = 0.425 ai = 0.489 a, = 0.536 1966 w' = 0.45 P' = 0.442 P' = 0.422 at = 0.473 at = 0.575 1969 w' = 0.30 P' = 0.296 P' = 0.261 a, = 0.476 at = 0.664 1963 tv' = 0.35 P' = 0.374 P' = 0.356 P' = 0.341 at = 0.446 at = 0.489 at = 0.526 1966 w' = 0.35 P' = 0.390 P' = 0.368 P' = 0.351 at = 0.371 at = 0.456 at = 0.533 1969 w' = 0.20 P' = 0.267 P' = 0.234 P' = 0.201 at = 0.302 a, = 0.427 at = 0.568 1963 tv' = 0.25 P' = 0.309 P' = 0.289 P' = 0.274 at = 0.401 at = 0.443 a, = 0.477 1966 w' = 0.25 P' = 0.359 P' = 0.335 P' = 0.316 at = 0.271 at = 0.336 at = 0.398 1969 w' = 0.10 P' = 0.271 P' = 0.241 P' = 0.210 at = 0.161 at = 0.231 at = 0.312 --Unreasonable results. EFFECTS OF WAGE REDUCTION 85 the same employment growth rate, the variation of the nominal wage growth rate affects the labor share and the inflation rate. The numbers underlying figure 4-1 are presented in table 4-2. They suggest the following: * Reading down the columns, we see that at a constant em- ployment growth rate, slower nominal wage growth leads to declining inflation but also to a fall in real wages and the labor share. Labor loses because the price level does not decrease as rapidly as the money wage. * Reading across the rows, we see that for constant nominal wage growth, the rate of inflation drops as the employment growth rate increases. Inflation is lower because higher rates of employment mean more rapid output growth. With more goods available and the same exogenous changes in invest- ment, exports, and government expenditures, private con- sumption can be more easily supplied, with smaller changes in prices. Lower inflation rates mean higher real wage in- creases. Combined with higher employment growth, these give an increased labor share. * When both the above results are combined for the same year, a decrease in the money wage together with an em- ployment increase produces less inflation and a smaller labor share. More precisely, table 4-3 presents three different paths for the KS model, showing year-end values in three columns under each year. The first corresponds to the already familiar path B where wage growth rate is 0.45 a year in 1960-66 and 0.30 a year in 1966-69. In the second column (C) wage growth rate is set equal to 0.35 a year in 1960-66 and to 0.20 a year in 1966-69. The last hypothesis (path D) combines the decline in the growth of the wage rate with an increase in the employment growth rate to 0.025 in the periods where it was nil and 0.07 in the last period where it was 0.045. This results in 6.089 million more jobs in 1969, a full 25 percent increase in the total labor force estimated by hypothesis B. The immediate impact of the wage decrease is deflationary, as can be verified from the third line of table 4-3. The decline of prices, however, is not enough to compensate the decline of nominal wages, so that real wages decline not only in comparison with the base path, but also along the same path for the whole period, as can be seen in line 6. Table 4-3. Effects of Wage Reduction on Growth, Inflation, and Labor Share 1963 1966 1969 Path B Path C Path D Path B Path C Path D Path B Path C Path D w' = 0.45 w' = 0.35 w' = 0.35 w' = 0.45 w' = 0.35 w = 0.35 w' = 0.30 w' = 0.20 w' = 0.20 KS model L' = 0 L' = 0 L' = 0.025 L = 0 L' = 0 L' =0.025 L' = 0.045 L' = 0.045 U =0.07 1. Output (X)* 3.016 3.016 3.126 3.343 3.360 3.587 4.063 4.082 4.489 2. Output growth rate (X') 0.040 0.040 0.052 0.034 0.036 0.046 0.065 0.065 0.075 3. Inflation rate (P') 0.443 0.374 0.356 0.442 0.390 0.368 0.296 0.267 0.234 4. Profit rate (r) 0.165 0.171 0.159 0.153 0.171 0.148 0.160 0.203 0.167 00 5. Labor share (00) 0.489 0.446 0.489 0.473 0.371 0.456 0.476 0.302 0.427 6. Real wage (w/P)' 0.057 0.052 0.055 0.058 0.046 0.052 0.059 0.038 0.047 7. Real labor income (Y,/P), 1.296 1.180 1.337 1.325 1.045 1.359 1.537 0.979 1.506 8. Private savings share 0.121 0.121 0.109 0.125 0.129 0.104 0.136 0.154 0.116 9. Government savings share -0.046 -0.035 -0.027 -0.031 -0.012 0.006 -0.049 -0.040 -0.010 10. Capital inflow (F) 0.107 -0.055 -0.025 -0.017 -0.316 -0.264 0.362 -0.275 -0.197 11. Employment level in the productive sector (L)b 21.289 21.289 22.947 21.289 21.289 24.734 24.425 24.425 30.514 a. In billions of cruzeiros. b. In millions of jobs. EFFECTS OF WAGE REDUCTION 87 Lines 4 and 5 show that in the KS solution with slow wage growth the rate of profit increases and the wage share decreases in comparison with the base solution. Since real output in the second hypothesis (path C) is maintained roughly constant as compared with the base path, the main impact of wage decreases is on distribution. In the third hypothesis (path D), where a de- crease in the wage is combined with an increase in employment, the real wage declines by 18 percent from 1960 to 1969. Even if this loss is compensated by a huge increase in employment, the gap between the share of profits and wages still increases in an impressive way. By comparing paths B and D for 1969, it can be verified that for an implicit elasticity of employment with respect to real wages equal to 1,6 the labor share decreases from 0.476 in hypothesis B to 0.427 in hypothesis D. In 1966, when a real wage decrease of 15 percent is compensated by an 11 percent increase in employment, the labor share still falls. In 1963 an increase of 7.5 percent in employment against a decrease of 3.5 percent in real wages results in a constant labor share. Even if a wage squeeze results in a huge increase in employment the functional income distribution becomes markedly more unequal. Our results differ from Morley and Williamson's because our model is closed with respect to saving. Like them, we show that an increase in aggregate demand leads to more employment, which may well benefit the poor. The increased demand, however, has to be "financed" by more saving, and in any kind of widow's cruse model this shifts the income distribution toward 6. An employment elasticity with respect to real wages equal to 1 is an overesti- mation of its real value. On this see Bacha and others (1972). The elasticity estimates were obtained as follows: Increase in employment between path B and D (Employment in B + employment in D) /2 Decrease in real wage between paths B and D (Real wage in B + real wage in D) /2 (30.514 - 24.425)/27.4695 (0.047 - 0.059)/0.053 In19 (24.74 - 21.289) / 23.0115 = 1.37 (0.052 - 0.058) / 0.055 In 1963: (22.947 - 21.289) / 22.118 = 2.1. (0.055 - 0.057) / 0.056 88 Growth and Distribution in the 1960s the rich. This effect occurs in the identity model and not in that of Morley-Williamson. Their results are optimistic because they have conveniently left out the nonfavorable distribution mecha- nisms. Wage repression combined with stimulation of effective demand is not the sort of policy that shifts incomes away from rentiers and other high-saving classes. The poor almost always pay for rapid economic growth, and Brazilian policy in the 1960s (hypothesis B) made their burden more onerous. Effects of Labor Tax Reduction on Growth and Labor Share Bacha and others (1972) suggested that a favorable impact on employment would result from reduction of labor taxes. Since they analyzed the effect of a tax reduction in tenns of intersec- toral employment shifts, their conclusions cannot be tested here, but some related macroeconomic results can be studied. Because of the difference in tax rates across sectors, we set our "average" tL equal to 0.046 in 1960 and 0.113 in 1969, instead of the figures 0.18 and 0.34 that apply in Brazilian manufacturing industry and appear in Bacha and others (1972). Table 4-4 supplies the results from three runs. The base KS Table 4-4. Effects of Labor Tax Reductions on Growth and Labor Share 1966 1969 tL' = 0 tL' = 0 KS model Path B t' = 0 L' = 0.01 Path B tW = 0 L' = 0.05 1. Output (X)a 3.343 3.343 3.394 4.063 4.068 4.151 2. Output growth rate (X') 0.034 0.034 0.039 0.065 0.065 0.067 3. Inflation rate (P') 0.442 0.450 0.442 0.296 0.309 0.303 4. Profit rate (r) 0.153 0.163 0.157 0.160 0.183 0.174 5. Labor share (aGL) 0.473 0.450 0.469 0.476 0.418 0.448 6. Real wage (wlP) 0.058 0.057 0.058 0.059 0.055 0.058 7. Real labor income (YLIP), 1.325 1.292 1.365 1.537 1.440 1.569 8. Private savings share 0.125 0.131 0.126 0.136 0.153 0.145 9. Government savings share -0.031 -0.039 -0.036 -0.049 -0.070 -0.064 10. Capital inflow (F) -0.017 0.009 0.031 -0.362 0.494 -0.525 11. Employment level in the productive sector (L)b 21.289 21.289 21.937 29.425 24.425 25.488 a. In billions of cnzeiros. b. In millions of jobs. EXPORTS AND GROWTH 89 solution (B) is in the first column, where the labor tax rate is supposed to have been growing as actually observed. The second column gives results for solutions where labor taxes are supposed not to have changed during the period (tL' = 0). In the third column it is assumed that a reduction in labor taxes could affect employment favorably. As can be seen on line 9, the first impact of tax reduction is to decrease government saving (or increase government dissaving). To maintain real investment, the model operates in the following ways: first there is a more rapid rate of inflation; with rates of employment constant (or even growing) this reduces the labor share. In contrast to the solution without tax reduction, increased profit rates work in the same way as the drop in the labor share to increase private saving. Foreign saving also increases as a counterpart to the increase in imports of consumption goods re- sulting from a fall in their price in relation to home goods. These shifts in saving work against labor income so that the functional distribution in the KS model worsens even when the rate of employment growth is increased. Constant employment results can be predicted on widow's cruse grounds as indicated in chapter 3, but the employment growth results depend on pa- rameters. In column 3 the labor tax rate is assumed to be 83 percent less than in the base path. This implies a reduction in labor costs of 0.067. Real wages also decline by 0.017 percent in the last hypothesis when compared with the base path. If em- ployment responds to the 8.4 percent decrease in costs with an elasticity of 0.5 as suggested by Bacha and others (1972), the labor share still declines as can be seen in line 5 of table 4-4. As is shown in chapter 9, these distributional responses to a tax change carry over in a multisectoral model as well. Exports and Growth One current interpretation of recent Brazilian growth is that it was made possible only by post-1964 trade policy. This point of view is expressed by Von Doellinger, Faria, and Cavalcanti (1974). They also suggest that the incentives and subsidies given to exports may have led to distributional deterioration since those mechanisms benefit the proprietors of scarce resources, espe- cially capital, while trying to achieve a greater supply of such resources in the future. The exercise below suggests that even in Table 4-5. Effects of Export Reduction on Growth and Labor Share 1963 1966 1969 E' =0 E' =0 E' =0 KS model Path B E' = 0 L' = -0.003 Path B E' = 0 L' = -0.002 Path B E' = 0 L' = 0.039 1. Output (X)' 3.016 3.016 3.016 3.343 3.328 3.316 4.063 4.035 3.981 2. Output growth rate (X') 0.040 0.040 0.040 0.034 0.033 0.032 0.065 0.064 0.061 3. Inflation rate (P') 0.443 0.428 0.426 0.442 0.435 0.437 0.296 0.271 0.278 4. Profit rate (r) 0.165 0.155 0.155 0.153 0.138 0.140 0.160 0.127 0.134 5. Labor share (aL) 0.489 0.511 0.509 0.473 0.507 0.501 0.476 0.550 0.527 6. Private sav- ings share 0.121 0.114 0.114 0.125 0.112 0.114 0.136 0.111 0.117 7. Government savings share -0.046 -0.050 -0.050 -0.031 -0.038 -0.039 -0.049 -0.056 -0.060 8. Capital inflow (F) 0.107 0.326 0.321 -0.017 0.404 0.394 0.362 1.334 1.300 a. In billions of cruzeiros. CONCLUSIONS 91 the absence of subsidies, export growth could have led to deterio- ration in the functional income distribution. Table 4-5 presents results of the base path B compared with two different paths. The second column corresponds to the hy- pothesis of export stagnation: export growth rates which were equal to 0.05 in 1960-63, 0.035 in 1963-66, and 0.03 in 1966-69 are set equal to zero in path B for the whole period. In the third column employment response to exports is overestimated, and the elasticity of employment with respect to exports is set equal to the export share in output. When exports stagnate, employ- ment and output decrease: employment growth rate becomes negative in 1960-66 and falls to 0.039 in 1966-69. Even so, the labor share in the third hypothesis still increases when compared with the base path where there are greater exports and employ- ment. These results can be interpreted as follows: lower exports mean more product available- domestically and less need for ex- treme savings efforts via inflation and a fall in real wage to meet investment targets. Conclusions In this chapter the identity-based model developed in chapter 3 was applied to the Brazilian experience in the 1960s. First a base path was developed to track the observed behavior of main macroeconomic variables in the period. Since it was impossible to develop a reasonably close approximation to the observed economic behavior with a neoclassical closure (see appendix B), we chose to simulate the model only for the Kaldorian variant. The main exercise was estimation of the effects of alternative wage paths in the economy. It was concluded that under reason- able assumptions about employment response, wage reductions make income distribution markedly more unequal. Because our model is closed with respect to saving it was possible to show that the burden of economic growth during the 1960s was made even more onerous to the workers by Brazilian wage policy. After an exercise with alternative export paths, we concluded that export growth probably led to deterioration in the functional income distribution by demanding more extreme savings efforts via inflation and a fall in real wages. These results seem to imply that recent Brazilian growth has been disequalizing and that gov- ernment wage policy and export promotion may well have con- tributed to deterioration of the income distribution. 92 Growth and Distribution in the 1960s Appendix A: Data and Sources In this appendix are described the sources of the data used to set up the identities of the model in table 3-5 for the 1960 base year. Equation (3.2.1) is the accounting definition of total product and expenditure in our system. The values of the variables come directly from the national accounts with the minor modifications, including: * The value of X in 1960 cruzeiros (Cr$2.675 billion) differs from the value of GNP in the accounts (Cr$2.756 billion) because there are two differences in definition between X and GNP. First, the value of intermediate imports is included in X but not in GNP, and second, government wage payments are excluded from the value of output PX but are included in GNP. * Since 1960 is our base year, we set price level P to unity. * The value of consumption of goods produced within Brazil (Cx = Cr$1.811 billion) differs from personal consumption in the national accounts (Cr$1.913 billion) because consump- tion imports are excluded from the former. * The value of gross fixed capital formation by firms in the national accounts is Cr$0.336 billion. To find investment in nationally produced goods we subtract Cr$0.056 billion, which corresponds to imports of capital goods. The remain- ing investment is split into two parcels-net capital forma- tion and depreciation of nationally produced capital stock. Since this stock includes all construction, we assume a rela- tively low depreciation rate (5x = 0.02). * Our value for government consumption (CG = Cr$0.287 billion) is lower than the corresponding item in the national accounts (Cr$0.366 billion) because it does not contain gov- ernment wage payments (w[l + t,] LG = Cr$0.079 billion). * The values of E and IG are equal to the corresponding entries in the accounts. Equation (3.2.2) is basically a "trade gap" statement of the balance of payments. It appears in domestic prices since all terms are multiplied by the exchange rate p. We set this equal to an annual average for 1960, with a value of Cr$0.11 to the dollar. The dollar values of intermediate imports .TmM and imports of capital goods n, (IF + 8, + 8,KF) in 1960 were respectively US$0.400 billion and US$0.492 billion according to the Boletin DATA AND SOURCES 93 of the Central Bank of Brazil (vol. 3, no. 5). Our dollar value for imports of consumption goods (US$0.893 billion) differs from that of the Boletim by including nonfinancial services. Finally, the value ofF in equation (3.2.2) was calculated as a residual and represents only a trade gap. Equation (3.2.6) is a breakdown of value added into its compo- nent costs. Value added itself (VX = Cr$2.312 billion) is the estimate of national income at factor cost from the accounts plus depreciation on nationally produced and imported capital goods (5xKx + 5FKF) plus intermediate imports at post-tariff prices (P,M) minus government wage payments. We used depreciation rates Sx = 0.02 and 8F = 0.05 for the two types of capital stock. The total stock itself (Kx + KF = Cr$4.926 billion) was estimated by deflating the values in Langoni (1974). Our employment estimates (L = 21.289 million, LF = 1.362 million) are from Hoffman (1972). The labor tax rate of 0.046 is calculated from estimates of incidence of this tax in manufactur- ing by Bacha and others (1972). The 0.18 percent incidence es- timate from this source was corrected by the share of the labor force (0.25) covered by IMPs, the national insurance scheme. The tax rate on value added (t, = 0.157) was calculated from the difference between value added at factor cost and the value of output. The shares of inputs in equations (3.2.8) through (3.2.11) follow directly from (3.2.6). In equation (3.2.14) capital income includes payments re- ceived for holding government bonds. The total expenditure of the government for bond payments in 1960 was negligible and will become significant only after 1965. Equations (3.2.15) and (3.2.16) define personal income taxes. Total revenue of Cr$62 million comes from the statement of the Brazilian Treasury (Boletim of the Central Bank, 1965). Our breakdown between taxes on labor and on capital incomes was done in view of tax schedules and the overall income distribution. The money base (H = Cr$0.374 billion) appearing in equation (3.2.17) and elsewhere is from the 1965 issue of the Boletin of the Central Bank. Pastore (1973) provides econometric justification for our value of 30 percent for the expected inflation rate in 1960. In equation (3.2.18) private consumption C is set to the corre- sponding value in the national accounts. Transfer payments (Q = Cr$0.119 billion) also come from the accounts. We follow Kaldor 94 Growth and Distribution in the 1960s (1966) and Brazilian evidence in setting yL, the propensity to consume from labor income, equal to 0.99. The propensity to consume from capital income (yK = 0.675) follows as a residual. Elasticities of consumption with respect to wealth (ScR = 0.05) and the deflated interest rate (Sci = -0.05) are from Christ (1968). We set the income elasticity of demand for domestic goods, ix, to 0.98. By Engel aggregation the elasticity for imports, -%, becomes 1.7674. The substitution parameter for consumption, oc, was set equal to 0.5. Equations (3.2.19) through (3.2.23) define consump- tion shares and capital stock growth rates, all of which follow directly from estimates already made. Equations (3.2.24) to (3.2.26) then set out the government accounts. Total tax receipts (T = Cr$0.527 billion) differ from the total of government revenue in the national accounts because of some nontax items (Cr$0.086 billion). The largest receipts come from value added taxes (t,VX = Cr$0.364 billion). Tariffs are other indirect taxes with revenues calculated from equations of the form (Pc - P7rc) C, where all values were previously known. After using independent estimates of income taxes (D, + DK) from the national accounts and of employment taxes (tL w[L + LG ]), we find total profit taxes and the profit tax rate (tK = 0.04) as residuals. In equation (3.2.25) all government expenditure items come from the national accounts as discussed above. The difference between expenditures and receipts is covered by expansion of the money base H* and by new bond emissions PB*/i. Our es- timate of Cr$572.4 million for the money base comes from the Boletim, 1965, and new bond emissions are calculated as a residual. The final equation (3.2.27) enforces equilibrium in the money market. The income elasticity (SHX = 0.7) is from Pastore (1973), and the interest elasticity (S;Hi = 1.0) is from Christ (1968). Appendix B: Failure of the Neoclassical Specification A complete relation of the assumed values of the exogenous growth rates for our base paths is shown in table 4-6. Table 4-7 provides end-of-year comparisons of the state of the Brazilian economy in 1963, 1966, and 1969. The estimates are from the national accounts scorekeepers of the Vargas Founda- FAILURE OF NEOCLASSICAL SPECIFICATION 95 Table 4-6. Values of Exogenous Growth Rates in the Base Solutions: Kaldorian Path A and Neoclassical Path a Kaldor- Neoclas- ian so- sical so- lution lution (KS), (NS), path A 1960-63 1963-66 1966-69 path a 1960-63 1963-66 1966-69 1. L' 0.000 0.000 0.046 L' 0.000 0.000 0.046 2. w' 0.450 0.450 0.230 w' 0.450 0.450 0.230 3. ITE' 0.000 0.030 0.004 VE' 0.000 0.030 0.004 4. 7i.' 0.000 0.019 0.000 7r' 0.000 0.019 0.000 5. 7rc' 0.008 0.012 0.033 1c' 0.008 0.012 0.033 6. rK' 0.000 -0.015 0.031 7r' 0.000 -0.015 0.031 7. e 0.020 0.005 0.012 E 0.020 0.005 0.012 8. A' 0.120 0.000 -0.200 A' 0.120 0.000 -0.200 9. e' 0.550 0.450 0.203 F' -0.500 -5.000 0.000 10. LG' 0.000 0.000 0.040 Lc' 0.000 0.000 0.040 11. I' 0.010 0.030 0.180 IG' 0.010 0.030 0.180 12. Cc' 0.065 0.040 0.070 Cc' 0.065 0.040 0.070 13. PM' 0.550 0.470 0.203 tM' 0.000 0.000 0.000 14. Pc' 0.558 0.462 0.236 tc' 0.000 0.000 0.000 15. PK' 0.550 0.435 0.234 t' 0.000 0.000 0.000 16. tv' 0.000 0.050 0.030 t' 0.000 0.050 0.030 17. t,l' 0.000 0.150 0.150 t,' 0.000 0.150 0.150 18. tK' 0.000 0.000 0.000 tK' 0.000 0.000 0.000 19. Q' 0.499 0.610 0.402 Q' 0.499 0.610 0.402 20. E' 0.050 0.035 0.090 E' 0.050 0.035 0.090 21. (H*)' 0.550 0.400 0.300 (H*)' 0.550 0.400 0.30(0 22. fx 0.000 0.000 0.000 Ar 0.000 0.000 0.000 23. gx' -0.050 0.000 0.050 24. g,' 0.000 0.000 0.100 25. rux 1.000 1.000 1.000 Note: rux = elasticity of imports in relation to output. ,x and rMx are not rates of growth but have been included to complete the values exogenously specified. tion and the base solutions of the KS model and NS model where the elasticity of substitution ox is assumed to be 1.8. The value of output generated by neoclassical solutions and the same exogenous rates of growth used for the Kaldorian path A vary according to different values ascribed to the elasticity of substitution ax. As ax increases, both prices P and real output X increase, providing the following results for 1969: