A STUQV SF THE STRUCTURE OF 7 BRAZIL'S FORETGN TRADE AND AN ANALYSIS OF HER T PROTECTION AS RELATEB TO THE EARNINGS 0F 7 TNDUSTRML LABOR Thesis for the. Degree of Ph. D. MICHTGAN STATE UNWERSITY THOMAS C. LDWTNGER 1970 THhh'S This is to certify that the thesis entitled A’ STUDY OF THE STRUCTURE OF BRAZIL'S FOREIGN TRADE AND AN ANALYSIS OF HER PROTECTION AS RELATED TO THE EARNINGS OF INDUSTRIAL LABOR presented by Thomas C. Lowinger has been accepted towards fulfillment of the requirements for Ph . D . degree in Economics MW% n E [IA/Wipes; Major professor Date May 22, 1970 0-169 ABSTRACT A STUDY OF THE STRUCTURE OF BRAZIL'S FOREIGN TRADE AND AN ANALYSIS OF HER PROTECTION AS RELATED TO THE EARNINGS OF INDUSTRIAL LABOR By Thomas C. Lowinger As the title indicates, this study is divided into two parts. The first part examines empirically two theories widely used to explain trade patterns. According to the Heckscher-Ohlin model, the primary determinants of international trade flows are international differences in countries' relative endowments of capital and labor. Leontief's findings that United States exports are labor intensive compared to an equivalent amount of United States import replacements challenged the validity of this model. Leontief as well as Kenen, Keesing and others suggested that in the empirical estimation of relative capital and labor intensities of countries' exports and imports, differences in labor quality or embodied skill differentials should be added to the value of the "phys- ical" capital. Since both the acquisition of "physical" capital and the possession of superior skills involves Thomas C. Lowinger acts of investments they could theoretically be combined. For this purpose the study uses Hal B. Lary's index of overall factor intensity, value added per man. Chapter III determines the direction and composi- tion of Brazil's commodity trade. Furthermore, the availability of "physical" and human capital is examined relative to Brazil's main trading partners, the indus- trial countries of North America and Western Europe. Chapter IV then tests a few of the "orthodox" theories of international trade. The hypothesis using skills or human capital to explain Brazil's trade flows performs better than the one using "physical" capital alone. Next, a special explanation of Brazil's exports of agricultural products is advanced. From Part One, I conclude that skilled labor is Brazil's most scarce factor of produc- tion. Part Two deals with the structure of Brazilian pro- tection relative to the factor intensities of her manu- facturing industries. The hypothesis tested.is whether Brazil's 1957 Tariff Law raised the earnings of the coun- try's industrial labor. This would be the expected re- sult from the Stolper-Samuelson theorem, which deals with the effect of a tariff on the earnings of the relatively "scarce" factor of production. Chapter VII tests this hypothesis within the framework of a linear model esti- mated by ordinary least squares. The estimated equations Thomas C. Lowinger of the model indicate that the tariff and exchange pre- miums had an adverse protective effect. Instead of raising the earnings of industrial labor, the tariff and total protection in Brazil harmed it. The last chapter examines some policy implications of this study, especially the validity of the claim that Brazil's future growth can be based largely on her exports of manufactured goods. Finally it is noted that to in- crease the international competitiveness of her indus— tries, Brazil has to go further than simply lowering her tariffs. A thorough revision of the country's tariff structure, which considers her actual factor endowments, is necessary. Such a policy would enable the country to compete in world markets for manufactures along the lines suggested by the theory of comparative advantage. A STUDY OF THE STRUCTURE OF BRAZIL'S FOREIGN TRADE AND AN ANALYSIS OF HER PROTECTION AS RELATED TO THE EARNINGS OF INDUSTRIAL LABOR By Thomas C: Lowinger A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Economics 1970 ACKNOWLEDGMENT It is a pleasure to acknowledge my material and intellectual debt to a number of individuals. I thank Professor Mordechai Kreinin, Chairman of my guidance committee, who labored through many drafts of this study and whose contributions are too numerous to be listed in detail. I am especially thankful to Dr. Jose D. Langier for providing me with the initial statistical resources needed for the study and for various suggestions during the course of this work. My colleague, John T. Donnelly, was most helpful because of his extensive knowledge of the Brazilian economic situation. Finally, the most profound gratitude goes to my wife, Carole, who with patience and love bore a great deal of the everyday burden for the past two years. 11 TABLE OF CONTENTS ACKNOWLEDGMENT LIST TABLES LIST OF FIGURES. Chapter I. INTRODUCTION . . . . . . . . II. THE COMMODITY COMPOSITION OF TRADE: THEORY AND EMPIRICAL VERIFICATION . . ITI. COMPOSITION AND DIRECTION OF BRAZIL'S COMMODITY TRADE AND HER FACTOR ENDOW- MENTS RELATIVE TO HER TRADING PARTNERS. IV. THE FACTOR COMPOSITION OF BRAZIL'S TRADE . V. THE EFFECTS OF PROTECTION ON INCOME DIS- TRIBUTION: THEORETICAL ISSUES . . . VI. THE STRUCTURE OF BRAZILIAN PROTECTION . . VII. INTER-INDUSTRY WAGE DIFFERENTIALS AND THEIR RELATION TO PROTECTION I BRAZIL . . . . . . . . . . . VIII. SUMMARY AND CONCLUSIONS. . . . . APPENDIX A: A DETAILED-ANALYSIS OF BRAZIL'S EXPORTS AND IMPORTS . . . . . APPENDIX B: THE AVAILABILITY OF VARIOUS SKILL CATEGORIES IN BRAZIL. . . . APPENDIX C: THE INSTITUTIONAL SETTING OF WAGE DETERMINATION IN BRAZIL. APPENDIX D: DATA APPENDIX . . . . . . . BIBLIOGRAPHY. . . . . . . . . . . . . iii Page ii iv 19 A6 80 101 1A9 188 1914 2141 252 261 271 Table A. .1 \fi .6 .9 .10 LIST OF TABLES Brazil's commodity exports as a proportion of total experts (based on dollar figur’eS) O O I O O O O O O O 0 Geographic distribution of Brazil's exports by major commodities and SITC categories as a proportion of total exports, 196u—65. . . . . . . . . . Brazil's exports by main-trading areas (as a proportion of total exports), selected years . . . . . Commodity composition of Brazil's imports (ratio of each SITC category imports to total imports), 1953-66 . . . . . . Geographic distribution of Brazil's imports based on SITC categories as a proportion of total imports, 1964—65. . . . . . Brazil's imports by main trading areas, as of total imports, 1953-67. . . . . . Levels of human resources for 12 countries by Harbison, Yahr and Correa indexes . . Occupational characteristics of United States' and Brazilian labor forces. Occupational characteristics of the labor force, various countries Estimation of capital stock and gross domes- tic product per capita in industrial countries and Brazil, 196“ (in dollars) Embodied characteristics of Brazil's imports and exports of manufacturers, 1966. . Rank correlation test of Brazil's commodity trade characteristics and the direction of her trade . . . . . iv Page 20 2A 25 28 29 35 37 NO 50 52 Table Page 4.3 - Average monthly wages and salaries paid by United States companies in the Sao Paulo area in June, 1959 . . . . . . 61 4.4 Correlation coefficients of Brazil's trade performance and industry characteristics, 1966. o o o o o o o o o o o o 614 4.5 Rank correlation coefficients of imports and industry characteristics for 20 Brazilian industries, 1966 . . . . . 65 4.6 Brazil's trade with Argentina and LAFTA countries, 1960—65 (in million dollars) . 73 4.7 Nominal and effective tariff rates levied on selected imports. . . . . . . . 74 4.8 Brazil's exports of agricultural commodities, 1952-54 and 1962-64 annual averages (in millions of dollars) . . . . . . . 76 4.9 Total capital investment on coffee farms, 1958 (per cent distribution). . . . . 78 4.10 Labor input required in growing of coffee and other crops (man-days per hectare) . 78 6.1 Exchange rates (cruzeiro per United States dollar), January 1, 1956 . . . . . . 104 6.2 Exchange rates (cruzeiro per dollar) . . . 106 6.3 Range of costs for one dollar's worth of imports into Brazil in March, 1958. . . 112 6.4 Tariff and exchange protection by use classes, 1964 (percentages) . . . . . 113 6.5 Summary statistics for tariff rate of 58 Brazilian industries (in percentages). . 116 6.6 Summary statistics of total protection for 58 Brazilian industries (percentages), 1958 and 1959. . . . . . . . . . 120 6.7 Rank correlation coefficients between pro— tection and wages, Brazil (1958, 1959) and the United States (1962). . 124 Tahle Page 6.8 Correlation coefficient between capital intensity of 20 Brazilian industries and Skill indexes—-l959. . . . . . 133 6.9 Correlation coefficients between tariffs and total protection and skill indexes for 20 Brazilian industry groups, 1959. 134 6.10 Simple arithmetic means of duties and charges applied in three Latin American countries and EEC'S External Tariff (in percentages), 1960 . . . . . . 140 6.11 Average hourly wages of industries classi- fied according to protective classes, 195“ o o o o o o o o o o o o lLl3 6.12 Array of United States industries (1965) and Brazilian industries (1959) based on wage and non—wage value added per employee. . . . . . . . . . . 146 7.1A Cross section regression results, deter- minants of inter-industry labor earnings-—l8 major industries, 1959. . 167 7.1B Cross section regression results, deter- minants of inter-industry labor earnings-—l8 major industries, 1959. . 168 7.2A Cross section regression results, deter- minants of inter—industry labor earnings, 18 major manufacturing industries, 1959 . . . . . . . . 169 7.2H Cross section regression results, deter— minants of inter—industry labor earnings, 18 major manufacturing industries, 1959 . . . . . . . . 170 7.3 Cross section regression results, deter- minants of inter-industry labor earnings, 57 three—digit manufacturing industries, 1959 . . . . . . . . 172 7.4 Cross section regression results, deter- minants of inter—industry labor earnings, 57 three-digit manufacturing industries, 1959 . . . . . . . . 173 vi Table Cross section regress ion results, deter- minants of inter- industry labor earnings, 57 three— digit manufacturing industries, 1959 . . . . . . . Cross section regression results, deter- minants of inter-industry labor earnings, 59 three-digit manufacturing industries, 1959 . . . . . . . . Cross section regression results, deter- minants of inter-industry labor earnings, 55 three-digit manufacturing industries, 1959 . . . . . . . . Cross section regressions of percentage changes in labor earnings of 58 industries, 1958-59 . . . Price and income elasticities of coffee demand and actual and projected coffee consumption (thousand tons) . . . . Elasticity estimates of the demand for COCOa, 1950-61. 0 o o o o o o 0 Annual growth rates (compound) and linear trends of Brazilian experts, based on SITC categories, 1953-66 . . . . . Annual growth rates (compound) and linear trends of Brazilian exports in constant (1953) prices . . . . . . . . . Index numbers of the value of Brazil's exports, import prices and income terms of trade. . . . . . . . Brazil's gross domestic product, exports and the ratio of exports to GDP in constant 1960 prices. . . . Brazil's share of world trade of certain major primary exports (percent); various years . . . . . . . . Import demand function for total imports and two use classes vii Page 174 175 176 177 197 200 212 212 214 215 216 219 Tam Iv Page A.” Import (Ivmzmd elasticities and percentage trends of imports. . . . . . . . 22] A.10 Import coefficients of selected Latin American countries, 1960—66 . . . . 226 A.1l Trend of per capita imports of selected Latin American countries, 1960—66 (dollars per capita). . . . . . . 226 A.12 Composition of category imports as a pro- portion of total Brazilian imports (percentage) . . . . . . . . . 227 A.13 imports as a percentage of total production plus imports for Brazil's manufacturing industries-—l949, 1958, 1961 . . . . 230 B.1 Shortage of technical skills in the metal- working industry--l956 . . . . . . 245 H.2 Primary education enrollment in Brazil and other LAPTA countries, 1962 . . . 249 B.3 Public expenditure on education in Brazil, LAPTA countries and selected industrial Countries—-l960 . . . . . . . . 250 C.l Monthly salaries and wages paid by United States companies in Sao Paulo, first quarter, 1957 (in cruzeiros) . . . . 254 C.2 Average monthly wage and salaries paid by United States companies in Sao Paulo (in cruzeiros), 1951, 1957 and 1959. . 257 C.3 Average monthly wage in Rio de Janeiro, according to activity and employees function, April, 1966 (1,000 cruzeiros) . . . . . . . 260 D.1 Exports of Brazil based on SITC (revised) classification, various years (in millions of dollars). . . . . . . 262 D. F0 imports of Brazil based on SITC (revised) classification, various years (in millions of dollars). . . . . . . 263 viii Table Page D.3 Exports of Brazil by main areas of des- tinations, 1953-67 (millions of d>llars). . . . . . . . . . . 26 U1 v.4 Imports of Brazil by main areas of origin, ‘ 1953-67 (millions of dollars). . . . POL D.5 Brazil's exports, imports and industry characteristics (1959), various Patios 0 o o o o o o o o o o Eff 0.6 Wages, value added, and other industry statistics (in thousand cruzeiros), 20 industries, protection 1958 (in per cent). . . . . . . . . . . . 269 v.7 Wages, value added and other industry statistics (in thousand cruzeiros), protection (in per cent) . . . . . 2o» ix LIST OF FIGURES ifbrure Page I:.1 Tariffs and production worker's average earning: in Brazilian manufacturing industry by major industry groups, 1999 . . . . . . . . . . . . 126 6.? Wage and non—wage value added per employee by major industry groups in Brazil, 1959 . . . . . . . . 129 6.3 Skill index—ll and average nominal tariff rates of 20 industry groups in Brazil, 1959 . . . . . . . . . 135 6.4 Total protection and production worker's average earnings in Brazilian Manufacturing Industry by major industry groups, 1959 . . . . . . 145 CHAPTER I INTRODUCTION The classical doctrine of international Trade dealt largely with the gains accruing to countries engaged in free trade along the lines suggested by the law of com— parative advantage. However, that model was not adequate for a full analysis of the relationship between free trade and income distribution. A more complete discussion of this relationship became possible with the development of the Heckscher-Ohlin model. The Heckscher—Ohlin model predicts that each country will tend to export the commodities that are intensive in its relatively abundant factor. Given the assumptions of the model, an increase in tariffs will have a predictable effect on the functional distribution of income. Stolper and Samuelson demonstrated convincingly that with unchanged terms of trade, a tariff will increase the relative price of the commodity that employs the scarce factor more intensively.1 Thus, a tariff which raises the domestic price of the importable in relation to the domestic price of the exportable good will increase the returns to the "scarce" factor. Metzler noted that Stolper's and Samuelson's conclu- sions have to be modified if one considers the effect of a tariff on the terms of trade and the resulting changes in income distribution.2 Using a criterion developed by Lerner, he showed that a tariff will increase the domestic price of importables relative to exportables if and only if the foreign elasticity of demand pigs the domestic marginal propensity to spend on importables is greater than unity.3 Metzler's insights into the possible effect of tariffs levied by Latin American countries on their domestic income distribution prompted this study. He observed that since world demand for Latin American exports of agriculture and mining products is presumed to be inelastic, a tariff is likely to improve these countries' terms of trade to such an extent so as to injure the region's scarce factor of production.Ll However, the relevant consideration for this study is whether the demand for Brazil's agricultural ex— ports is inelastic and not whether world demand for these commodities is inelastic. (This question will be thoroughly discussed in the following Chapters and Appendix A.) Al— though later writers have expounded and generalized Metzler's theoretical framework, it has not yet been tested empirically. This study attempts to determine empirically the effect of Brazil's 1957 Tariff Law on the country's income -x.) distribution. However, before this problem can be in- vestigated the structure of Brazil's foreign trade must be analyzed. Thus, the first part of the study is a test of the "nee—factor proportions" explanation of the struc- ture of international trade. Specifically, I question whether Brazil's foreign trade conforms to the expecta— tions based on the generalized factor proportions theorem. initially, it is necessary to determine the commodity compositixnl of Brhafi;lhs exports Puui hnports ammitflu3r>f lHlith LHxites (expordxt embodied less capital and more labor than an equivalent amount of competitive imports. Since the United States has more capital per worker than any other major trading country, Leontief’s results appeared to contradict the Heckscher—thin theorem. Subsequentlv, certain writers have criticized the empirical foundations and more impor- tantly, the methodological flaws inherent in Leontief’s 3 study. Perhaps more significantly, as a result of his findings some of the rigid assumptions of the model were challenged and new and interesting avenues for investiga- tion were opened. In a recent study Hufbauer found that the export patterns of 2“ countries at various stages of development can be eXplained by the Heckscher—Ohlin hypothesis, and also by the product cycle, returns to scale and product differentiation hypotheses. Hufbauer concluded: "Export patterns exercise an intriguing kind of selectivity. commodities are favored which contain several character- istics [underlinings mine, T.C.L.] suitable to the nation's economic structure. The composite trading pattern thereby agrees with various theoretical predictions."u It seems reasonable to distinguish between trade conducted among the advanced countries whose industrial base had developed over the last hundred years and trade between industrial and less—developed countries such as Brazil. Since the trade among the less-developed countries is insignificant quantitatively it can be ignored; in 1968 only 10 per cent of the less-developed countries exports of $32.3 billion was accounted for by trade among themselves. This ratio was on the decline since 1960 when it stood at 18 per cent.5 In explaining trade patterns among industrial coun— tries it is appropriate to consider not only the factor proportions account, but also what Hufbauer called the "neo—technological" factors, i.e., scale economies and product age differentiation. For semi-industrialized countries (in Latin America these would include Argentina and Mexico as well as Brazil)—-that also possess a small and unsophisticated industrial sector——the "neo—factor prOportions" account, using physical and human capital, should be sufficient to explain their trade patterns. 10 h :lkills, }hhnan Cayfiinil and tim3(knmnodity Composition of Trade During the last decade several new theoretical ap- proaches have been advanced to explain International Trade patterns. Among the early contributors was Kravis, who argued that the Commodity composition of trade is mainly guided by the domestic "availability" (or the lack of it) of goods.6 In investigating some 330 United States manufacturing industries, he found that industries ex- porting a high proportion of their domestic production pay higher average wages than industries that have a high ratio of imports to domestic production.7 An explanation of Kravis' findings was advanced by Gary Becker in the context of his work on human capital. Becker suggested that the observed earnings of individuals tend to be related in an important way to their investment in them- selves. This investment may be in the form of formal edu- cation, onuthe-job training or informal ways of learning and improving oneself.81 Thus, if the wage differentials observed_by Kravis are the product of skill differences (i.e., investment in human beings) and if it is accepted that the United States has a relatively abundant supply of human capital, then United States export industries "ought" to pay higher wages than the import—competing industries because they employ more skilled workers.9 This approach was further developed by Kenen and 10 some of its implications were spelled out. In Kenen's 11 model, land and labor constitute the country's "natural" endowment. Capital is an indirect factor of production; land and labor are inert until they are combined in pro- duction through the application of capital. Following the formal presentation of his model, Kenen suggested that the relative proportions of "human" and "tangible" capital embodied in commodities traded may be used to predict the structure of United States trade. In fact, given that the United States has a relatively plentiful supply of human capital, one would expect the United States to export products that are intensive in human capital. Approaching the Leontief paradox from the standpoint of human capital, Kenen was able to estimate (roughly) the human capital requirements of United States trade. He assumed that skill differences among labor arise wholly due to the quantity of capital invested in people and that wage differences (ascribed to skill) reflect the gross return to that capital. Based on these assumptions he was able to estimate the capital that is required to con- vert a man-year of crude labor into a man-year of skill.11 Then, using Leontief's own percentage distribution of skill, Kenen found United States export production to be intensive in human capital compared to United States import- competing commodities.12 Keesing, Waehrer and other members of the Inter- national Economics Workshop at Columbia University have l2 centered their attention on the effect of labor skills or human capital on the structure of trade.13 Specifically, they suggested that the category of factors aggregated under the term "labor" should be subdivided; ranging from unskilled to highly skilled (i.e., reflecting varying intensities of human capital). Predictions about trade patterns could then be made on the basis of the relative skill endowments of countries. This approach was adopted in part because of their dissatisfaction with the treat- ment of factors of production in the simplified Heckscher- Ohlin model.1u Their criticism centered on the fact that natural resources were excluded from the "basic" model, as well as different qualities of the same factor of pro- duction, such as various skill categories of labor.15 To partly overcome these shortcomings, Donald Keesing attempted to incorporate into the Heckscher—Ohlin model different qualities of labor as separate factors of pro- duction.16 He then proceeded to show that differences in the skill intensity of products are reflected systemati- cally in the patterns of trade. In effect, he proposed to broaden the framework of the Heckscher—Ohlin model to include various grades of labor according to their skill classification instead of the theoretically convenient but Operationally nebulous factor "labor." Keesing outlined the rationale for the inclusion of varying grades of labor in explaining the location of 13 industry and international trade patterns.17 The first reason is the relative immobility of labor internationally as compared to capital and certain man—made material re- sources. Thus, the initial availability of what is es- sentially an immobile factor (labor) in a certain area is likely to influence the location of industry or pro— duction in that place. Because the movement of capital to the location of labor is usually cheaper than the reverse, one of the most pervasive influences on location of industry will he the skill characteristics of the local labor force. Probably the most important reason is the inherent difficulty in rapidly changing the skill charac- teristics of a lator force, assuming that such a trans— formation is a good substitute for free international movement of skilled labor. Keesing puts it as follows: The general training and experience of a population, together with its attitudes and working habits, re- sist rapid change. Therefore broad classes of skills in any population can only be altered slowly.18 Keesing's study used the ratio of skilled to unskilled labor in a two-factor, many—country model to reflect the differences in qualities of labor.19 Using Leontief's skill calculations, he computed the skill content of 15 "footloose" manufacturing industries in the United States. The United States skill coefficients were in turn used to measure the skill intensity of exports and imports for the United States, seven European countries and Japan. By ordering the countries with respect to skill requirements 1A in eXport production and similarly in import-competing production, Keesing found an almost perfect reversal of the ordering of countries. This led him to conclude that skill availability was a major factor in the determination of the structure of international trade.20 The results of Keesing's studies strongly imply that the United States, being relatively well endowed in skilled labor, tends to export those goods which embody in their production a large proportion of skilled labor (its abundant factor). Corroborative evidence is found in Hufbauer's study relating the skill endowments of 2“ countries to the skill content of their commodity trade.21 Notwithstanding the imperfections inherent in such international comparisons, the study obtained high Spearman Rank correlation coeffi- cients between national skill endowments and skill charac- teristics embodied in traded manufactured goods. Noting the impressive explanatory power of the factor proportions account together with the human—capital theory, Hufbauer remarked that: ". . . a distressingly simple and orthodox formulation woes a long way to explain trade among manu- faxnnrred gyaods."22 The first part of this study will investigate whether "physical" capital and labor skills availability "explains" Brazil's commodity composition of trade. Chap— ter lII deals mainly with the question of "availability" of factors of production concentrating on the skill compo- sition of the labor force and the capital—labor ratio in manufacturing. 15 in Chapter IV various tests are devised to investi— mute the explanatory power of the neo—factor proportions theory of international trade in manufactured goods. The role of Brazil's "natural" resources (defined as climate and soil conditions plus an ample supply of unskilled lahor) is investigated to determine the country's revealed comparative advantage in agricultural commodities. In addition, an ad—hoc explanation of Brazil's manufactured exports is advanced within the context of the country's post World War ii economic development. FOOTNOTES: CHAPTER II 1For a fuller discussion of the theorem and its implications see: J. L. Ford, The Ohlin-Heckscher Theory of the Basis and Effects of Commodity Trade (New York: Asia Publishing House, 1965). For a summary of the present state of the factor endowment theorem see: M. Michaely. ”Factor Proportions in International Trade: Current State of the Theory," Kyklos, XVII (Rasc. U, 196A), 529- H9. 2Wassily Leontief, "Domestic Production and Foreign iride: The American Capital Position Re- examined, " Iroceedings of the American Philosophical Society, XCVII (.ieptemher,1)537,332-39. Further expounded in: w. Leontief, "Factor Pro— portions and the Structure of American Trade: Further Theoretical and Empirical Analysis," Review of Economics and Statistics, XXVIII (November, 1956), 386-HO7. 3See for example: Boris C. Swerling, "Capital Shortage and Labor Surplus in the United States?" Review of Economics and Statistics, XXXVI (August, 1954), 286-89, and M. A. Diab, The United States Capital Position and the Structure of Its Foreign Trade (Amsterdam: North-Holland Publishing Company, 1956). For a succinct discussion of the various criticisms see: R. E. Caves, Trade and Economic Structure: Models and Methods (Cambridge, Mass.: Harvard University Press, 196h), pp. 273—80. “G. C. Hufbauer, "Factor Endowments, National Size, and Changing Technology: Their Impact on the Commodity Composition of Trade in Manufactured Goods" (presented in an NUER Conference on Technology and Competition in Inter- national Trade, October 11-12, 1968, mimeo), p. NO. F" jInternational Monetary Fund and International Bank for Reconstruction and Development, Direction of Trade (June, 1968 and November, 1969). 6i. Kravis, "Wages and Foreign Trade," Review of Economics and Statistics, XXXVIII (February, 1956), iu—30. l6 {HULL , (if). 8For details consult: G. S. Becker, Human Capital: A Theoretical and Empirical Analysis, with Special Refer- ence to Education, National Bureau of Economic Research (New York: [Columbia University Press, 196U), Chapters II L1I1(i .I I Ii. 9 lbid., p. 60. 10P. B. Kenen, "Nature, Capital and Trade," Journal of Political Economy, LXXIII, No. 5 (October, 1965 , 637—60. 11 ibid., “56-57. lgibid., u57, Table 5. 13Donald B. Keesing, "Labor Skills and the Structure of Trade in Manufactures," in The Open Economy: Essayg on International Trade and Finance, ed. by Peter B. Kenen and Roger Lawrence, Columbia Studies in Economics I (New York: Columbia University Press, 1968), pp. 3-18, and Helen Waehrer, "Wage Rates, Labor Skills and United States Foreign Trade," in The Open Economy: Essays on Inter- national Trade and Finance, ed. by Peter B. Kenen and Roger Lawrence, Columbia Studies in Economics I (New York: Columbia University Press, 1968), pp. 19-39. For a complete discussion and in depth criticism of the model and the Leontief study see: Caves, op. cit., 273-82. ET 1”A consideration of the role of natural resources in United States trade is contained in: J. Vanek, Th3 Natural Resource Content of United States Foreign Trade 1870—1955 (Cambridge, Mass.: The M. I. T. Press, 1963). 16Donald B. Keesing, "Labor Skills and International Trade: Evaluating Many Trade Flows With a Single Measuring hevice," Review of Economics and Statistics, XLVII (August, 1965), 287-93, also Donald B. Keesing, "Labor Skills and Comparative Advantage," American Economic Association, Papers and Proceedings, LVI, No. 2 (May, 1966), 2D9-58. l7Keesing, "Labor Skills and Structure . . .," gp. Cite, 5—70 18ibid., 6. l8 ( Keesing, "Labor hkilis and International . . .," op. cit., BUD-"3- “01bid., 291, Tables 1 and 2' ”) “lHufbauer's national skill coefficient includes only the "professional, technical and related workers" category of the ILO classification scheme. This excludes certain highly skilled employees such as managers, supervisory personnel and foremen. See: Hufbauer, op. cit., p. 19. 22Jbid., n. 20. CHAPTER III COMfOSlTJON AND DIRECTION OF BRAZIL'S COMMODITY TRADE AND HER FACTOR ENDOWMENTS RELATIVE TO HER TRADING PARTNERS A. Direction of Brazil's Trade l. Commodity Exports Historically a small number of primary commodities have dominated Brazills export trade. As late as 1964-66 coffee, cocoa, sugar, textile fibers, lumber and iron-ore constituted about 80 per cent of the total value of Brazil's exports. Following World War II the proportion of coffee, cocoa and cotton in total exports has declined steadily. In l966—67, these three commodities accounted for 52 per cent of the value of Brazil's exports, a sharp decline from the 83 per cent existing in l953—5Li.1 The share of all major primary commodity exports declined from about 9A per cent in 1953-55 to about 79 per cent in 1964-66. The eXports of only two commodities, sugar and iron ore, have increased in relative and absolute terms during that period. The most impressive performance among Brazilian eXports was recorded by its manufactures exports (SITC 5, 6, Y, and 8). in 1960 exports of manufactured goods were about $21 million, or less than two per cent of all 19 .AQ xfloccdd oaooe .mm .o .Aoooa .oesofiooeH ooefixootm one “.o .o .eoomeeemozo mesoupmm moose swoopoe< cfipmq .oamchLO .h use .mocmmo .2 .comopsomm .3 .o “ooszom ooo. ooo. moe. mas. eem. moms oeo. ooo. . ooo. . mom. son. Hood ooo. ooo. mao. Hom. moo. mood moo. moo. oao. one. omm. . some omfi. . omH. Hos. oam. mme. oeom moHLoE< Copoq mofispczoo meugomz antsomfimlcoz . no oeHmv sec. cmm. mmo. 0mm. mefl. mam. OOH. moa. assasomczcmz afimmm mas. sac. ems. Hes. sag. i.s. Hmm. mmm. Am oeHmv mHmOHEmno Am oeHmv mam. mam. mmfl. ssfi. Hos. smo. mac. mmfl. mfiwss new mamcmcfiz Am oeHmv mamas mma. mmo. Nae. sag. moo. emu. 33H. ass. samoxm mamacmpmz musso Ao osHmv Hos. Hms. ass. som. ooo. mmo. was. :ms. mHmEHca m>aa new woos mofispcsoo . Hmflsumsncfi moflsos< maoszm mumua moozzu uoxpmx dcwcmo Ico: poppo :Hqu Choummm Hmuos cmomw :memoz coEEoo Ucm .m.: mpLomEH w a o m a m m H - .mwlzmmH .mpLOQEH aspen mo COHQLOQOLQ m mm mmHLowmpmo oeHm :o comma mupodsfi m.afimwsm do COHpanppmHU cacamnwoooll.m.m mqm<9 29 numpmmm oESHo> .mompe mo coapompfio .QmmH .A.o .Q .COQMCHSmmzv wmmfi .mQSh cam mmlmmma .zmuomofi .ocsm mpmumcoz HQCOHmeLwQCH .Axpo» zmzv.mmummma was mmummma Ilmsmmm mEsHo> “mumpe HmQOfimeLmucH mo coapoopfio .QmmH .mzH anOprz UmpHcD "mmopsom Hmo. 30H. Hmo. mom. wmo. mam. sad. mam. smlomma wwo. Hwa. mmo. mmw. omo. mom. mna. com. wwlzwma moo. mza. omo. son. H20. mmm. sma. :mm. mmuowma moo. :NH. mmo. mom. ,smo. Hmm. .m.: 22m. swimmma mmo. mmfi. :mo. smm. mmo. mam. .m.: mom. mmlmmma mmfippcsoo Hmfispmzaca .mmwmmmm .mmmmmm meHH mooszm mmmmmm mmmmmm null: :20: nonpo cfipmq cpmpmmm Hmuoe cwowm :Lmummz coEEoo cam .m.3 mammw m w m m z m m H .smnmmmfi .mpLOQEH HmpOp mo mm .mmmpm wcfivmpu chE mo mpmoqsfi m_HHthmI|.m.m mqm¢e 30 Eastern Europe. The general pattern followed by Brazilian trade is that of many other countries in early stages of economic development-ewhose commodity concentration of eXports is considerably higher than those of imports.7 During the 196U—67 period, only 65 per cent of Brazil's total imports originated in North America and Western Europe, yet 83 per cent of her manufacturing imports came from those sources. B. Brazil's Factor Endowments Relative to Her Trading Partners This part of the study inquires into Brazil's factor endowments in relation to her main trading partners. The identity of those partners has been established in Part A of this chapter. Generalizing from the Heckscher-Ohlin model, one can view the structure of trade as the outcome of the interaction between the production characteristics of commodities (with respect to "factor" intensity) and the countries' availability of these "factors." As was stated in Chapter II, I have chosen to test an "orthodox" version of the Heckscher—Ohlin model, i.e., one based on the national disparities in the availability of human and "physical" capital. 1. Human Capital To approximate the availability of human capital in Brazil, I am presenting three indexes which together reflect Brazil's human resource level in relation to her trading 31 partners. The Harbison index represents one attempt to quantify international differences of investments in skilled manpower. This index is calculated as a sum of two ratios: (a) students currently enrolled in secondary schools as a percentage of school-age population; and (b) students enrolled in institutions of higher educa- tion (colleges) as a percentage of school-age population, multiplied by a factor of five.8 Another proxy for the 9 Yahr stock of human capital was computed by M. E. Yahr. obtained for several countries the mean educational level for males aged 15 and above. To obtain his educational index Yahr classified the male population over 15 by the level of their formal education and then weighed each edu- cational class by the fraction of the male population in that educational group. The last measure was developed by H. Correa and it measures the physical capacity of a person to perform productive tasks.lO Correa related the actual calorie intake of the countries' work force to the calorie requirements of the labor force for 100 per cent working capacity and thereby obtained an index of the ”physical" capacity to work. The evidence presented in Table 3.7 shows that in a sample of 12 countries (both developed and underdeveloped) Brazil ranks the lowest in its endowments of human capital, based on the Harbison and Yahr indexes. Based on the Correa index, Brazil ranks fourth (lowest) out of eleven 32 TABLE 3.7.——Levels of human resources for 12 countries by Harbison, Yahr and Correa indexes. Country Harbison Yahr index Correa index index (school years per worker) Brazil 5 20.9 (1) 2.26 (1) 68.76 (u) Paraguay 22.7 (2) 3.00 (3)' 78.2U (6) Mexico 33.0 (3) 2.96 (2) 60.26 (3) India 35.2 (A) n.a. 27.51 (1) Chile 51.2 (5) u.37 (6) 69.uu (5) Argentina 82.0 (8) “.13 (5) 93.46 (9) Taiwan 53.9 (6) 3.66 (U) n.a. Sweden 79.2 (7) n.a. 86.97 (7) Canada 101.6 (9) 7.35 (7) 92.u5 (8) Japan 111.u (10) 8.57 (9) 59.22 (2) United Kingdom 121.6 (11) 9.U5 (10) 97.50 (11) United States 261.3 (12) 8.U7 (8) 96.23 (10) Notes: The numbers in parentheses indicate each country's ranking with respect to each index ( (1) indicates the lowest, etc.). Source: Merle I. Yahr, ”Human Capital and Factor Substi- tution in the CES Production Function," in The Open Economy Essays on International Trade and Finance, ed. by Kenen and Lawrence (New York: Columbia University Press, 1968), p. 7“, Table l. 33 countries, but this index of "physical" capacity to per- form work is a less appropriate measure of investment in human beings than the first two indexes. in another context Anne Krueger used human capital or rather, three proxy variables closely associated with it, to "explain" over 50 per cent of the differences in income levels between the United States and several under- deveIOped countries.l] With respect to India she was able to conclude that of the actual per capita income differ- ences between the United States and India about two-thirds are explained by the difference in their stock of human capital.12 These estimates led Dr. Krueger to conclude: While many other factors, including the endowments of other resources are undoubtedly very important, the explanatory power of the human capital constel- lation of factors is impressive.13 The recent emphasis on human capital as an important eXplanatory factor of existing trade patterns also pro- vides a useful insight into the sources of income differ- entials between rich and poor nations. The differences among rich and poor countries in their initial stocks of human capital are probably more pronounced than their dis- parities in physical capital or natural resources. In addition, the initial discrepancy among countries in their endowments of human skills may be propagated over time, due to the fact that skilled personnel is needed to train other skilled workers. In fact, the unequal distribution of highly trained personnel may be getting even more skewed 3“ due to the selective migration policies of industrial 1 countries and especially the United States.l‘ 2. The Skill Composition of Brazil's Labor Force Compared to Her Trading Partners I have noted that Brazil ranks low in terms of her human resource development (as measured by the Harbison, Yahr and Correa indexes) not only relative to the indus— trial countries of Western Europe and North America but also compared to other LAFTA countries. Next, further evidence regarding the skill composition of Brazil's labor force will be presented. A comparison with the United States has been attempted because the data on the Brazil- ian labor force are most readily comparable to the occupa- tional characteristics of the United States labor force as reported in the 1960 United States Census of Population. The following labor categories are classified as skilled: (l) Entrepreneurial, managerial and high administrative personnel; (2) Highly skilled professional personnel (especially engineers); (3) Subprofessional technical personnel (engineering assistants, technicians); and (U) Skilled workers (largely foremen and skilled craftsmen). There is a dearth of accurate estimates of the in— dustrial manpower supply in Brazil. I present two recent estimates, one for the country as a whole and the other for the city of Sao Paulo (see Table 3.8). This compares the occupational characteristics of the United States and 35 .amm .aeeaa .aacamm asopma anneapmccmaCH =.aammcm ca pCmEAOHQEm new amZOQcmz: .Nowa< .m .< "mom .mwma ammz on» now coapmozpm mo zapmHCHz one so consumed amzoacme cesamsp cmaaammpm on» mo mmeprm cm :0 comma ma manaamv .omo .amema .mcssv aaaox .3oa>om asoomq amCOapmcpmpCH :.aaNmLm Ca pcmE>oaaEm comps: ..Lh .mmzwwHQ amass: ”com .Looma poaaaxm go zpaaanwaam>m umopmoam esp spas popcoo HmappmSUCH pmmwama n.aapcsoo one so poems msnu ma pH .moa>amm Qacmooabzoaad< HwaaumSUCH awcoapmz map mp :wma .mczh Ca pzo poaasmo zLQmSUCa eo >m>asm Losoacee w Eopm mam Oasmm 0mm pom meme oneamv .Amooa ..o .a .coswcazmmzv asuamv om .ppoqmm Hosam .moapmasopomsmno amcoaumazooc mahogmm pomnesm .omma "ceapmHSQom mo mamcoo .m.: .mSmcoo on» do smogsm .m.: ”anm was memo mopdum owuaca mSBAHV o.cca cco.ooo.m c.cca mam.moa.a o.ooa mmm.ooo.we moses eonaa amaoe . . emaecax cm.ma 1 coo mam m.ma awa.smm s.ma mem.omm.a sea cmgwcom ”mpoxpos omaaaxm mamaoammo amace Imfisoosd .mamw . , a . . . ., (A an oa a . coo mm a a a a a a mas as: u -acaz amccowama “zLOmH>LmQ3m moacoppomam . . . . sq . d . . . one .amoappomam mm oco aa c: sea 3 cs ms: mo: .amacma .amoaemz ”mcmaoaccooe mm. ooc.aa oz. smm.s m.a eaa.oem mammcamcm mosow aozma bosom Loewa meson Lonma ampop do zampOp do aspOp eo ommusmosmd Lonssz esopcmogwd amass: ommpcoosom Lonezz maommpmo Haaxm aammmm mammalmmW amv caeaasm emaaca Ame aa .moohoe Loomfi cwaHaNmLm pcm .woumpm tapas: no mofiumapopompwco ascoHpmqsooOIl.m.m mqm .108 .187 Canada 1961 .106 .183 Austria 1961 .068 .103 Belgium 1961 .080 .106 United Kingdom 1961 .086 .112 Germany 1961 .076 . .107 Italy . 1965 .053 .133 Netherlands . 1960 .092 .123 Sweden 1960 .129 .150 I]. LATIN AMERICA Mexico 3 1960 .036 .OAA Chile 1960 , .OUQ .068 Ecuador 1962 .033 .036 Panama 1960 .0A5 .066 Paraguay 1962 .032 .OAO Peru 1961 .033 .OA8 Brazil* 1963 .037 ‘ .OA8 Notes: Ratio A is the per cent of skilled workers in category 0 (Professional, technical and related workers) as of total employees. Ratio B is the per cent of category 0 and l (administrative, executive and mana- gerial) as of total employees. The data is from: International Labour Office, Yearbook of Labour Statistics 1966 (Geneva: ILO). * = Brazil's ratio was extrapolated from the data on U. S. and Brazil in Table 3.8. 38 scarcity of skilled personnel at all levels, and especially of engineers and technicians. The slow expansion of engi- neering schools combined with a rapidly expanding demand for engineers resulted in very high salaries paid engineers relative to wages paid unskilled workers. (For details, see Appendix B). The scattered evidence available on the supply of managerial and administrative personnel suggests a serious shortage of qualified and well—trained people capable of running modern enterprises. Finally, despite the rapid increases in the number of skilled workers available to industry in the post World War 11 years, serious shortages prevailed as reflected in the high relative wages of such workers (especially foremen). Given the inadequacy of educational opportunities in Brazil and the shortage of facilities for the training of skilled workers, the skill composition of the country's labor force is likely to change very slowly in the near future.16 Thus, relative to the industrial countries, the position of Brazil as a skill deficient country will be maintained in the foreseeable future. 3. Physical Capital Compared to the information I presented on the availability of human resources, international comparisons of capital stocks are meager. Hufbauer estimated the capital stock per manufacturing employee of 22 countries. 39 His crude procedure involved adding up the current expen— ditures for manufacturing investment between 1953 and 1964 and then dividing by total manufacturing employment in 100A. No account was taken of inflation or depreciation of the capital stock. - Using a modified version of Hufbauer's procedure, I estimated the capital stock of Brazil.17 (See Table 3.10). The evidence presented in Table 3.10 would qualify Brazil "scarce" in physical capital relative to her as a country trading partners of North America and Western Europe. However, I have to stress the rough nature of both Hufbauer's and my estimates. To demonstrate how dependent are these types of calculations on the assumptions one chooses to employ, I calculated Brazil's capital-labor ratio in an alternative manner. From Clark and Weisskoff's study I obtained the gross fixed capital formation for the years 1953-64 at constant 1953 prices. This figure was converted into United States' dollars by using the mid-1958 period "general" category exchange rate (lU9.3 CR. per dollar). As a rule of thumb I assumed that 20 per cent of that figure is the value of gross fixed investment in manufac- turing. I then deflated this by the 1964 transformation industries employment. As a result of this procedure the figure obtained for fixed capital per manufacturing em- ployee was $8AO. This figure is not comparable to the U0 TABLE 3.10.-—Estimation of capital stock and gross domestic product per capita in industrial countries and Brazil, 196“ (in dollars). —‘_ Industrial country Fixed capital GDP per capita per employee in manufacturing Canada 8,850 2,002 United States 7,950 3,01U France ”,900 1,53U Belgium U,400 1,U58 Germany U,250 1,591 Netherlands “,750 1,264 Sweden 5,H00 2,032 United Kingdom “,000 1,480 Italy 2,100 916 Brazil 2,100 193 Sources: For industrial countries: G. D. Hufbauer, "Factor Endowments, National Size and Changing Tech- nology: Their Impact on the Commodity Composition of Trade in Manufactured Goods" (National Bureau of Eco- nomic Research, New York, December, 1968, revised verson, mimeographed), Table 6, p. 77. For Brazil: Data on gross fixed investment taken from: United Nations, Yearbook of National Accounts Statistics 1966 (New York: United Nations, 1967). Exchange rates employed are from: International Monetary Fund, Annual Reports on Exchange Restrictions (various years). Note: It was estimated that the share of manufacturing industries in gross capital formation was about 20 per cent. This is somewhat higher than the percentage used in connection with Brazil's three—year plan (1963-65). The figure used in the plan was 18.5 per cent. Thus, our estimate may be an upper limit of the "true" value of Brazil's capital stock. See: ECLA, "Fifteen Years of Economic Policy in Brazil," Economic Bulletin for Latin America (December, 1967), 206, Table 31. U1 calculations reported in Table 10. These estimates are only a rough indication of the domestic availability of Brazil's capital stock. It is possible that Brazil's shortage of "physical" capital is not as acute as certain skills in her labor force. In the first place, the flow of capital inter- nationally is much less inhibited than is the movement of labor across national boundaries. For instance, there was a considerable flow of capital in the form of direct investment from industrial countries into Brazil. Be- tween 1955 and 1961 the flow of direct foreign invest- ment alone into Brazil amounted to $71” million dollars.18 Secondly, given the relative wage differential that exists between Brazil and the United States, the flow of human capital (i.e., skilled personnel) acts to worsen the initial discrepancy in skill availability between the two countries. In this chapter it has been established that Brazil suffers from a great dearth of skilled labor in relation to her cheif trading partners, based on two ratios of skill "availability." Because of the conceptual and statistical problems involved in estimating the country's stock of "physical" capital the shortage of this factor has been less accurately determined. Overall, it has been established that the country is capital (both human U2 and "physical") poor, relative to the industrial nations of North America and Western Europe. Based on these findings, Chapter IV will examine Brazil's commodity trade using certain versions of the factor proportions model. FOOTNOTES: CHAPTER III lComputed from sources in Data Appendix. 2All the data from which these calculations were derived are in the Data Appendix. 3Calculated from: Survey of the Brazilian Economy 1965 (Washington, D. 0.: Brazilian Embassy, 1966). “The imports of SITC 7 commodities averaged about $277 million in 1964-65. The breakdown within that cate- gory was as follows (each three-digit SITC group as a percentage of all SITC 7 category): Power machinery, non—electric (SITC 711) 9.0 per cent Agricultural machinery (SITC 712) 2.0 per cent Office machines (SITC 714) 5.5 per cent Metal—working machinery (SITC 715) 9 6 per cent Textile, leather machinery and machines for special industries (SITC 717 and 718) 12.0 per cent Other non-electric machines (SITC 719) 20.9 per cent Electrically powered machinery and electricity distributing machinery (SITC 722 and 723) 8.7 per cent Other electrical machinery (SITC 724, 725 and 729) 9.6 per cent Railway vehicles (SITC 731) -4.2 per cent Road motor vehicles (SITC 732) 9.4 per cent Aircraft (SITC 734) 4.0 per cent Total of above 95.8 per cent Calculated from: United Nations, Commodity_Trade Statis- tics, 1964 and 1965, Series D., Vols. XIV and XV (New York: United Nations, 1965 and 1966) 5The imports of office machinery were less than $4 million in 1953; in 1966 they amounted to $23 million. See Data Appendix. 6The imports of motor vehicles averaged about $110 million during 1957-59; in 1963-65 they fell to about $28 million per annum. Large scale domestic production of motor vehicles began in 1957. 43 44 7M. Michaely, Concentration in International Trade (Amsterdam: North—Holland Publishing Company, 1962), pp. 31.36. 8F. Harbison and C. A. Myers, Education) Manpower, and Economic Growth: Strategies of Human Resource Devel-v opment (New York: McGraw—Hill, 1964), pp. 31—32. 9Merle I. Yahr, "Human Capital and Factor Substitu- tion in the CES Production Function," in The Open Economy: Essays on International Trade and Finance, ed. by Peter B. Kenen and Roger Lawrence, Columbia Studies in Economics I (New York: Columbia University Press, 1968), pp. 31-32. 10H. Correa, The Economics of Human Resources (Amsterdfim: North-Holland Publishing Company, 1963), pp- 30-3 - 11Anne 0. Krueger, "Factor Endowments and Per Capita Income Differences Among Countries," Economic Journal, LXXVIII (September, 1968), 641-55. 12 Ibid., 650-51, Table II. l31bid., 656. 1“Between 1962 and 1966, 466 Brazilian scientists, engineers and physicians immigrated into the United States. In 1962 Brazilian universities graduated 60 scientists (Ph. D.'s) of which 13 (22 per cent) immigrated to the United States. U. S. Congress, The Brain Drain into the United States of Scientists, Engineers and Physicians (washington, D. C.: A Staff Study for the Research and Technical Programs Subcommittee of the Committee on Government Operations, July, 1967), p. 7, Table VII. 15A study by Nathaniel Leff presents somewhat differ- ent view with respect to skill requirements of capital equipment producing industries. As regards the use of skilled labor in the capital goods sector, Leff found that the lack of formal education (visa via the United States labor force) did not seriously hamper them in carrying out their often technically exacting tasks. Based on interviews conducted by Leff it appears that three or four years of formal schooling was all that was generally re- quired for semi—skilled workers in that sector. In the case of foremen, five to six years of schooling was suf— ficient. In contrast, the median school years completed by foremen in the United States manufacturing industries was 11.8 years, i.e., a high school equivalency. Based on Leff's study we can conclude that at least in certain 45 skilled occupations (especially category 4) there is con- siderable leeway in terms of the formal educational re- quirements of the labor force. See: N. H. Leff, The Brazilian Capital Goods Industry, 1929—1964 (Cambridge, Mass.: Harvard University Press, 1968), pp. 46-49. - 16In Brazil several vocational training activities exist that are administered by the government or by the employers. SENAI (National Industrial Apprenticeship Service) offers a variety of training programs for the young and for adults usually in evening classes. See: W. B. Dale, Brazil/Factors Affecting Foreign Investment (Menlo Park, Calif.: International Industrial Development Center, Stanford Research Institute, 1958). 17 The main difference between Hufbauer's calculation and ours involves the use of exchange rates. Hufbauer used mid-period (i.e., generally 1958) exchange rates to convert local currency units into U. S. dollars. Our procedure used the average of import category exchange rates (since the multiple exchange rates were in effect in Brazil) to convert the annual data on Brazilian capital formation. This procedure is better suited to take ac- count of the rapidly depreciating Brazilian currency resulting from the domestic inflation. For comparison to Hufbauer's method see: Hufbauer, op. cit., pp. 77-80, Table 6, and P. G. Clark and R. Weisskoff, "Import Demands and Import Policies in Brazil," U. S. Agency for International Development, February, 1967 (mimeo), Table 0-2, p. 50 and Table B—SB, p. 45. 18N. H. Leff, Economic Policy-Making and Development in Brazil, 1947—1964 (New York: John Wiley and Sons, 1968), Table 13, p. 61. ‘ CHAPTER IV THE FACTOR COMPOSITION OF BRAZIL'S TRADE A. Exports and Imports of Manufactures In Chapter III it has been established that the bulk of Brazil's trade is conducted with the industrial countries of Western Europe and North America. It has been shown that Brazil is deficient in human capital (skilled labor) and to some extent in "physical" capital relative to her main trading partners. Brazil's most abundant "factor" is the combination of her plentiful natural resources and her unskilled labor.1 In this chapter, I wish to inquire whether Brazil's commodity composition of foreign trade conforms to what would be expected from her factor endowments. For the most part, this chapter will test two "orthodox" theories of the commodity composition of trade. One is the simple factor endowment account, which uses the "physical" capital intensity of manufactured goods. The HeckscheruOhlin model focused attention on inter- national differences in countries' relative endowments of capital and undifferentiated labor as the primary eXplanation of International Trade. The other is the 46 47 human capital or skill—intensity explanation of trade. In certain instances the human and "physical" capital in- tensities have been combined into one "overall" measure of man—made resources. 1. "Direct" Test of the Factor Proportions Theorem and the Human—Skills Hypothesis One possible testing procedure is to use Brazil's Inter—Industry Transaction Table to compute the direct and indirect capital and labor requirements of Brazil's exports and import—replacements. This would constitute a direct test of the original version of the factor pro— portions model along the lines suggested by Leontief's celebrated study, and later applied to United States trade with Japan and India.2 While I understand that an Inter- Industry Transaction Table is presently being constructed for Brazil and will be available for future research, it was not available at this time. In lieu of Leontief's procedure, I used the United States factor composition of traded commodities (based on two or three digit SITC) that were computed by Hufbauer and applied them to a matrix of Brazil's 1966 export and 3 This procedure assumes that the rankings import trade. of commodities produced in the United States, based on capital per employee, skill ratios and wages per worker, can be applied to Brazil, i.e., that no factor intensity reversals occur. 48 Although several writers expressed considerable skepticism as to the empirical significance of the factor intensity reversal phenomenon, I nonetheless tested for this theoretical possibility within the limitations im- posed by my data.5 Specifically I used the Rank correla- tion technique to ascertain whether the pattern of United States industries' factor intensities can be applied to Brazil. In comparing factor intensities across industries, I adepted Lary's concept of "overall" factor intensity based on value added per employee and its breakdown into the wage and salary component and the non-wage value added 6 "residual." Further elaboration of these concepts is contained in part A-2 of this chapter. The highest Spearman (Rank) correlation coefficient of .841 was obtained when United States and Brazilian industries were ranked based on wages and salaries per employee. If it is agreed that industries paying higher than average wages may be characterized as skill intensive, we ought to reject the existence of any "skill reversals" between United States and Brazilian industries. A some- what lower Spearman correlation of .735 resulted when the comparison was based on the industries' "overall" factor intensity. And finally, the lowest coefficient, .694, was obtained using the non-wage value added, i.e., the proxy for "physical" capital intensity across my sample of 20 industries. All three coefficients were easily 4 9 significant at the one per cent level. Thus, this evi- dence leads us to conclude that factor-intensity rever- sals (and especially "skill" reversals) are not of em- pirical importance even in this case. Having disposed of this possible objection, I then proceeded to use the American capital-labor ratios, skill coefficients and wage rates to compute the embodied char- acteristics of Brazil's foreign trade in manufactured goods. My results are presented in Table 4.1. The evidenCe summarized in Table 4.1 conforms to a priori expectations based on Brazil's relative factor availabilities; i.e., Brazil's imports of manufactured goods have more capital and skilled labor embodied in them than her exports of manufactures. This results from the preponderance of above-average capital and skill in- tensive goods in Brazil's imports bill relative to her exports. For instance, Brazil's 1966—67 imports of the highly capital and skill intensive commodity group—— Chemical Elements and Compounds (SITC 51)—-averaged about $113 million or about seven per cent of total imports. Brazil's exports of the said commodity group were only $16 million a year, less than one per cent of Brazil's 1966-67 exports. In contrast, in 1966 Brazil exported $8.5 million worth of leather (SITC 611), which comprised one-half of one per cent of her total exports. At the same time, Brazil imported less than $.2 million of this 50 TABLE 4.1.-~Embodied characteriStics of Brazil's imports and exports of manufacturers, 1966. Capital Skill Wages ‘per man ratio per man Imports of manufacturers $9224.5 .067 $3948.2 ($9457.2) ($4018.9) Exports of manufacturers $1077.2 .0058 $ 439.3 ($1000.5) ($ 444.0) Notes: The characteristics of 1966 exports and imports were based on three-digit SITC trade data figures except in case of the skill ratios where the two-digit SITC category was used. The figures in parentheses were calculated based on the two-digit classification. The embodied characteristics were obtained in each case by multiplying a given commodity character- istic (e.g., skill ratio) by the percentage of Brazil's exports (or imports) of that commodity and summing it up over all three digit (or two digit) commodities. Source: Characteristics of trade goods were obtained from: G. C. Hufbauer, "Factor Endowments, National Size and Changing Technology: Their Impact on the Commodity Composition of Trade in Manufactured Goods" (presented in an NBER Conference on Technology and Competition in International Trade, October 11-12, 1968, mimeographed), Table 3. Brazilian trade data were taken from: United Nations, Commodity Trade Statistics 1966 (New York: United Nations). 51 commodity that is characterized by low skill content and low capital intensity.7 While these findings conform to a priori expectations, given Brazil's factor endowments relative to her trading partners, they do not constitute a "strong" test of the factor prOportions theorem. In the first place, while I rejected the possibility of general factor intensity rever- sals, some factor substitution is likely to take place. Given the serious shortage of human capital (i.e., skilled labor) in Brazil, there will probably be some substitution of unskilled labor (and possibly also "physical" capital) for skilled labor. Thus, the use of American capital and skill coefficients in lieu of Brazilian factor intensities in production of traded commodities may "overstate" my case somewhat. Secondly, while virtually all of Brazil's imports of manufactured goods originated in industrial countries, no such clear—out division exists in the destination of her manufacturers exports.8 In 1964-67 about 42 per cent of these exports were directed to Western Europe, United States and Canada, with somewhat over one-half going to Latin America and other less developed countries. Given that Latin American countries have roughly similar endow- ments of capital and skilled labor, the trade among them in manufactured goods is the likely result of various loca- tional factors as well as tariff concessions granted within the framework of LAFTA.9 52 Based on two—digit SITC categores, I next computed the proportion of Brazil's imports and exports of manufac- Itures that originated in or was destined to the markets of Western Europe and North America.10 Using the rank corre- lation test I then compared these ratios to the character- istics of the traded commodities, with respect to capital per man, skill ratios and wages per man. The results of this test are reported in Table 4.2. TABLE 4.2.-~Rank correlation test of Brazil's commodity trade characteristics and the direction of her trade. Industrial countries' Industrial countries' proportion in Brazil's proportion of Brazil's manufacturers imports total manufacturers (1966) ' exports (1966) Capital per man .450** —,028 (t = 2.523) (t = .143) Skill ratio .500* -.1l9+ (t = 2.883) (t = .600) Wages per man ’ .575* .127 . (t = 3.514) - (t = .022) Note: * Significant at the one per cent level. ** Significant at the two per cent level. i Significant at the 30 per cent level. Source: Data on commodity characteristics are from: G. C. Hufbauer, "Factor Endowments, National Size and Changing Technology: Their Impact on the Commodity Composition of Trade in Manufactured Goods" (presented in an NBER Conference of Technology and Competition in International Trade, October 11-12, 1968). Brazilian trade data are from: United Nations, Commodity Trade Statistics 1966 (New York: United Nations, 1966) U". co Table 4.2 ranks commodities based on their human capital intensity. As is theoretically expected the more skill intensive a given commodity, the higher the pro- portion of that commodity that will be imported from the human capital abundant countries of Western Europe and North America (i.e., one would expect the Spearman Rank to be positive and significant with respect to imports). When analyzing the results of this test, the reader must bear in mind that over 80 per cent of Brazil's manufactures imports originate in the industrial countries of Western Europe, United States and Canada. The tests summarized in Table 4.2 confirm to the hypothesis that Brazil's imports of manufacturers included a high proportion of commodities that are intensive in "capital" conceived in a broad sense to include not only the material means of production but also human skills. Yet based on this test we are unable to discriminate between the "physical" and human capital eXplanations of the structure of Brazilian trade. It is also evident from Table 4.2 that the human and physical capital intensities did not provide a satis- factory explanation of Brazil's patterns of manufactures exports. (In two cases the coefficients are negative as one would expect, but insignificant at the customary level.) This later result is not entirely unexpected since Brazil's exports of manufactured goods were largely a 54 by—product of the evolution of her manufacturing sector, through the process of import substitution. The expansion and growth of the industrial base during the post World War II period was stimulated by means of various incentives and subsidies and in particular, the exchange and tariff privileges granted industries during the 1950's.11 Given the importance of domestic market size in the process of industrialization through import substitution, exports of manufactures have been viewed by Brazilians as an outlet for that part of the output that could not be accommodated domestically. Investment in local production was under— taken without regard to the future feasibility of exporting these goods and with the knowledge that the domestic market was ”secured" through protective poliCies adopted by the government. The textile industry in Brazil is a case in point. For more than half a century the Brazilian cotton manu- facturing industry enjoyed the privileged status of an "infant" shielded by a high protective wall. Brazil's textile industry was one of the earliest established in Latin America and the country has almost achieved self— sufficiency in textiles, wool products and certain man— made fibers. Yet in 1965-66 exports of textiles (SITC 65) averaged only $15 million, just about one per cent of domestic production. In a detailed study of the industry Stein pointed out: "As long as manufacturers could sell 55 production at home and the 'excess' abroad they could Operate successfully without truly competitive conditions arising."12 And the industry's view of exports has been stated by Stein as follows: ". . . manufacturers agreed that exports constituted a method to carry the industry until the level of domestic consumption were raised by 13 measures involving all sections of the economy." 1 Brazil's "exportable surplus' approach to International Trade helps "explain" the lack of any clear-cut results in Table 4.2 concerning exports of manufactures.lu Analysis of Brazil's non-manufacturing exports is taken up separately in the last part of this chapter. 2. Brazil's "Revealed" Comparative Disadvantage and the Industries' Factor Intensities In this section a different set of tests will be devised to analyze the competitive performance of Brazil's industries in international markets. This part differs from Section 1 of this chapter on two important points. In the first place, I deal with industries' performance [with respect to exports and/or imports] rather than with individual commodities. And secondly, a fuller complement of industry characteristics has been devised. I will deal with each of these questions in turn. Numerous measures can be developed in order to gauge the international competitiveness of an industry 56 (e.g., the time trends of exports and imports or compara— tive production costs). In this study, one measure used is the trade balance (i.e., imports less exports of two— digit SIC industries) as a ratio of the value of industry's domestic sales. Presumably, those industries in which Brazil had n2 comparative cost advantage internationally would rank high based on this index of competitiveness and vice versa. As a complementary indicator I computed the ratio of imports to exports for each industry. Both ratios would thus rank Brazil's industries on their ability to compete in foreign markets and on the extent of penetration of other countries' exports into the do- mestic market. Analysis based on the trade balance as a measure of comparative advantage suffers from the known deficiency of being affected by tariffs and quotas in the importing 15 These may understate Brazil's cost advantage countries. in certain labor intensive products, which also may be taking advantage of cheap domestic sources of raw materials. It is of interest to note that almost all of the indus- tries that were net exporters in 1965 were crucially de— pendent in their production on the local availability of cheap and plentiful sources of raw materials. Lumber (net eXports of $62.7 million), leather (net exports of $4.5 million) and textiles (net exports of $13.5 million) are cases in point. 57 To establish the industries' factor intensities the following variables were used: Value added per employee, two measures of the skill content of each industry, and a proxy for the capital-labor ratio. Following the work by Hal B. Lary, value added per employee has been used as a satisfactory measure of the flow of human as well as "physical" capital into the manufacturing process.16 Using the flows of man-made resources into the industries' output, value added per employee is taken as an indicator of the "overall" factor intensity of an industry. The theoretical justification for this approach has been advanced in another context by Peter B. Kenen. Kenen treats the country's natural resources--unimproved land and completely unskilled labor—-as inert inputs that must be "activated" through investment in order to be used in 17 In this context value added per employee production. can be used as a single measure of the aggregate flow of man-made resources into the production activities of an industry. Since one part of the value added can be attributed to human capital, it is imperative that an appropriate measure of industries' "skill content" or human capital intensity be designed. The theoretical underpinning of this approach has been established by Schultz and Becker 18 in their innovative work on human capital. Becker and Schultz maintain that differences in occupational 58 earnings can be attributed to differences in the amount of training required by various occupations. Investment in human beings is usually referred to by the catch-all phrase "training." It includes formal and technical edu- cation as well as on-the-job training. The individual who pays for some or all of the costs involved in im- proving his skill would have increased his lifetime earn- ings stream more than the person with less training.19 Assuming that workers in "skilled" occupations incur greater costs of investment in themselves than "unskilled" workers, the "skilled" worker should receive greater ab— solute earnings than those received by the "unskilled" workers. To test whether there is a relationship between the inter—industry skill composition and the industries' competitive performance, one must first devise a meaning- ful measure of the industries' skill mix; that will be ‘ referred to heretofore as the "skill-index." Mrs. Helen Waehrer constructed such a measure of industry's skill in her study of the wage structure of export and import- competing industries in the United States.20 Her "occupa- tional index" was defined as the percentage of employees in each industry, out of all employees, falling into six occupational groups designated as skilled.21 Mrs. Waehrer found a significant positive correlation between her index of skill and earnings of labor in foreign trade oriented industries. 59 The Waehrer index regrettably suffers from limita- tions as'a measure of skill or as a proxy for human capital. Specifically, it will vary from industry to industry only with differences in the distribution of employees defined by her as "skilled" and those defined as "unskilled." This will conceal wide variations in skill content within each of the broad categories of skilled workers. As an illustration, suppose that two industries rank equally in terms of Waehrer's skill index; but one of the industries has a higher proportion of managers and engineers, while the second industry has an equally large proportion of sales and clerical personnel. Clearly, from the stand- point of human capital intensity, the first industry should rank higher than the second; however, this will not be reflected in the Waehrer skill index, because of her definition of "skilled workers." To avoid these deficiencies I have constructed for Brazil a weighted skill index, where the weights are the earnings of workers in each skill-category. Skill index-I is defined as the percentage of earnings in each industry going to workers classified as skilled, out of total earn- ings of employees in that industry. The classification of workers into "skilled" or "unskilled" categories is some- what similar to that of Mrs. Waehrer. Unfortunately, the Brazilian Industrial Census of 1960 includes a breakdown of earnings in 20 major industries only according to broad 60 occupational classes. The three classes defined in the Brazilian Census are: Class A--professionals and tech- nical workers; Class B—-"other employees," that includes directors and managers, office personnel and other service personnel; Class C——workers directly involved in produc- tion, including operators and kindred workers, as well as foremen and apprentices.22 I defined skilled workers as those belonging to Classes A and B, above. Skill index-I admittedly does not allow for a detailed enough breakdown of occupational classes. Some employees included in Class C and therefore classified as unskilled should really be included in the skilled category. Specifically, foremen whose overall training includes a considerable amount of technical schooling and on-the—job training should be classified as skilled laborers. On the other hand, a number of service workers belonging to Class B and classified there- fore as skilled should really be reclassified in the un- skilled category. Category B includes in part custodial services workers, porters and typists whose training re- quirements are minimal. Table 4.3 shows that these re- servations with regard to skill index—I are well founded. The average monthly salary of a foreman who was classified as "unskilled" is more than twice the salary of a typist or a file clerk that I involuntarily classified as "skil- led," due to the inadequate breakdown in the census classi- fication. 61 TABLE 4.3.—-Average monthly wages and salaries paid by United States companies in the Sao Paulo area in June, 1959. Function or occupational Wages and salaries category ' in cruzeiros Engineer 38,326 Personnel manager 33,905 Accountant 33,222 Salesman 25,118 Billing supervisor 16,995 ‘ Typist (Portutuese only) ' 4 8,101 K File clerk 7,706 Foreman 20,362 Mechanic 11,734 Skilled laborer 9,583 Packer 8,360 Unskilled laborer " 5,900 Source: Gordon and Grommers, United States Manufacturing Investment in Brazil 1946-1960 (Boston: Harvard Uni- versity, 1962), p. 119, Table 18. Based on these considerations I devised a different skill index that resolves some of the aforementioned dif- ficulties. From the Brazilian census data I got the number of employees in each industry group that can be classified 23 as skilled based on their actual earnings. BaSed on Table 4.3, I decided to use the salary of CR 9,000 per month as the cut-off point. That is, employees earning above this salary would be classified as "skilled" and those falling below this as "unskilled." The members of the following occupations were consequently classified as 62 skilled: Engineers, managers and accountants, salesmen, billing supervisors, foremen and "other" skilled workers. I defined skill index-II as the percentage of those employees in each industry that earned in excess of CR 9,000 per month, or fall into the occupational cate- gories listed above. Furthermore, I computed skill index—III: The percentage of employees in each industry that earn more than CR 15,000 per month.214 Skill index-III includes the members of occupational groups previously listed, with the exception of those classified as "skilled" laborers and mechanics. This gave me yet another measure of the human capital content of each major industry group corresponding to the occupations with the highest skill level. Because of the deficiencies in skill index-I as a measure of human capital, I was not surprised to find that the rankings of Brazilian industries based on the three measures of skill do not correspond very closely. The Spearman Rank correlation coefficients between skill index—I and skill indexes-II and III were .651 and .697 respectively; both significant at the one per cent level. The last variable measures the "physical" stock of capital per employee in each industry. Aware of the known pitfalls encountered in measuring the stock of capital, I used as a proxy for this variable the horse-power per employee in each industry. As a test of the extent of linear association between the two sets of variables, the 63 Spearman Rank correlation coefficients were used. Another measure of the degree of conformity is provided by Kendall's Coefficient of Concordance.25 (See Tables 4.4 and 4.5.) The results in Tables 4.4 and 4.5 are in accord with the previous findings that in relation to her trading partners, Brazil's factor deficiency is most pronounced in human capital (i.e., skilled labor). Consequently, Brazil was compensating for its scarcest factor by im- porting from Western Europe and the United States those commodities that contained a high proportion of human capital. My results are the more convincing because they are based on several different ratios of an industry's trade performance, i.e., the industry's exports, imports and the trade balance. The high coefficients of concor- dance with respect to the skill ratios are impressive. Thus it is possible to state that the comparative dig: advantage of Brazil is most pronounced in the manufactured commodities requiring a great deal of skilled labor in their production. Physical capital intensity and the "overall" factor intensity, based both on material and human capital, did not perform as well in "explaining" Brazil's trade patterns. While Brazil's comparative digadvantage has been established, the question of her comparative advantage has not yet been considered. Given the abundance of certain "natural" resources, i.e., climate, soil and 64 TABLE 4.4.-—Correlation coefficients of Brazil's trade psrformance and industry characteristics, 1966. A. Spearman Rank Correlation Coefficients Industry Characteristics(a) Skill Skill Horsepower Value added Index-II Index-III pgr Employee ppr Employee Ratio of . net imports .699* .695* .077 .328 to value of (t=4.l47) (t=4.lO3) (t=.329) (t=1.472) sales Ratio of ' imports .502* .593* .404i .311 to exports (t=2.460) (t=3.126) (t=l.872) (t=1.392) B. Kendall's Coefficient of Concordance Ratio of net imports .726 .758 .347 .558 to value of (z=5.288) (z=5.385) (z=4.120) (z=4.769) sales Ratio of imports .621 .642 .305 .494 to exports (z=4.964) (z=5.028) (z=3.990) (z=4.574) Source: See Data Appendix (Appendix D). (a) Notes: Bach industry's skill index is computed relative to all manufacturing industries; sample of 20 industries used. Skill Index—II-—The percentage of employees in each industry who earned in excess of 9,000 CR per month. Skill Index—III--The percentage of employees in each industry who earned in excess of 15,000 CR per month. * Statistically significant at the one per cent level. ** Statistically significant at the two per cent level. *** Statistically significant at the five per cent level. i Statistically significant at the ten per cent level. 65 TABLE 4.5.~—Rank correlation coefficients of imports and industry characteristics for 20 Brazilian industries, 1966. Industry Characteristics Skill Skill Horsepower Value added Index—II Index—III per Employee pgr Employee Imports by industry as a per- .479*** .533** .458*** .395+ centage of total (t=2.230) (t=2.669) (t=2.186) (t=1.822) imports Imports by industry ’ as a ratio .600* .653* .339 .417+ of irkhls— - try's (t=3.180) (t=3.658) (t=l.528) (t=1.947) sales Source: See Data Appendix (Appendix D). Notes: See Table 4.4. completely unskilled labor, Brazil "ought to" export goods that contain relatively large proportions of this abundant "factor." This includes a handful of agricultural com- modities and especially coffee, cocoa, cotton and sugar. In part B of this chapter further details are spelled out. B. Exports of Agricultural Commodities In Chapter II, mention was made of the fact that as late as 1966—67 Brazil's exports of manufactures comprised only a small proportion of her total exports. Brazil's "revealed" comparative advantage rested with those com- modities that were "nature" intensive, i.e., utilized a high proportion of natural resources and unskilled labor in their production. As I previously observed, a small number of such "nature" intensive commodities, specifi- cally coffee, cocoa, cotton, tobacco and sugar, have dominated in the past and still dominate today Brazil's export structure. Brazil has the climate and the soil suitable for the production of many important tropicalcrOps. These conditions and the availability of cheap labor enabled Brazil to establish herself as the world's largest pro- ducer of coffee. Another natural resource that is plenti— ful in Brazil is forestry. According to FAO's World Forest inventory, Brazil in 1958 had 12.7 per cent of the _ 26 World's forests. At the turn of the century Brazil accounted for over b0 per cent of the world's coffee pro— ., {C} 27 ductlon, though by 1135—59 her share fell to 50 per cent. Sumar, cotton, tobacco, cocoa, oranges and rice are also . y o o p 28 produced and exported in substantial amounts. Only s,attered information exists regarding the production methods and factor content of these agricul- tural commodities. An FAO study of coffee cultivation in irazil enabled me to draw some inferences regarding factor intensities in its production. The study concentrates on (" 3 _ _ o o o 29 sao Iaulo, the leading coffee produ01ng state in Braz1l. The greater part of the State of Sao Paulo belongs to the trapical highland climatic zone where temperatures and 67 the amount of rainfall are suitable for the cultivation of a wide range of agricultural products, including 30 coffee. For the most part the soil of the State of Sao Paulo is exceptionally fertile and suitable for coffee production.31 Labor is by far the most important variable factor 32 The FAO survey indi- used in the production of coffee. cates that on the average, 118 man-hours were needed to produce 100 kilograms of green coffee. About 90 per cent of the labor force was employed in operations related to harvesting coffee beans and weeding (weeding is done several times during a crOp-year). Another five per cent was.used in coffee processing, leaving only five per cent for non-routine operations that may involve more advanced 33 The labor used in the differ- cultivation techniques. ent phases of coffee production is almost entirely com- posed of unskilled workers. It was estimated that in 1958 the average hourly wage paid to workers on coffee farms in the State of Sao Paulo was eight cruzeiros}!4 The monthly average wage of about 1600 cruzeiros paid the coffee farm laborer should be compared to the federal minimum wage of about 6000 cruzeiros per month established .35 in January, 1959 Manufacturing establishments in the Sao Paulo area paid at least this minimum wage to their unskilled workers. Based on both the type of work per- focmed and their earnings, the coffee farm—laborer must he considered largely devoid of any skill at all. 68 Fixed investment in land and coffee trees comprised about 80 per cent of the capital invested in coffee pro— duction. The FAO study discovered that less than 10 per cent of the total investment was in equipment and facili- 36 One of the main ties used in the processing of coffee. findings of the survey was the extremely low level of mechanization and capitalization characteristic of the various phases of coffee production and harvesting in Brazil. The FAQ study concluded: "Coffee is bound to remain a labour-intensive [underlining mine] commodity to be produced only in areas where there is a relatively abundant labour supply that will accept relatively low wages."37 Thus, it is appropriate to characterize Brazil's main commodity export, coffee, as being heavily dependent in its production on unskilled labor as well as 38 natural—-soil and climate-—conditions. No recent studies were available on the production of cocoa and cotton in Brazil, though it stands to reason that largely the same "natural" conditions were responsible for the low produc— 39 tion costs and subsequent exports of these commodities. Conclusions to Chapters III and IV Using "physical" capital labor ratio, the simple factor proportions theory did not fare well in explaining Brazil's trade patterns. The human capital or skill hypo- thesis, however, yielded good results in explaining Brazil's commodity composition of imports. As regards Brazil's 69 exports, "nature" (i.e., natural conditions plus completely unskilled labor) was invoked to explain the country's comparative advantage in agricultural commodities. Unfor- tunately, because of the unavailability of systematic studies and data on the factor intensity of those commo- dities, the conclusions are somewhat tentative. The limited evidence presented in Chapter IV points to the combination of soil, climate and cheap labor as largely accounting for Brazil's export performance. The latest data available show that in the first half of 1969, coffee, cotton and cocoa still constituted 50 per cent of Brazil's exports.l40 In 1060 Brazil exported about $250 million worth of manufactured goods or about 11 per cent of total exports.”1 This figure is impressive since at the beginning of the 1060's eXports of manufactures were only about $25 million. Yet i maintain that these developments are not as encour- aging as they appear because essentially they represent an extension of the post World War II growth based on import—substitution. In a perceptive speech to the 196“ UNCTAD conference, Raul Prebisch noted: Thus, a real vicious circle has been created as regards exports of manufactured goods. These exports encounter great difficulties because internal costs are high, and internal costs are high because, among other reasons, the exports which would enlarge the markets are lacking.“2 70 In effect the exaggerated reliance on import substitution- induced growth may have spread the country's investments over a large number of small and inefficient manufacturing activities. The appearance of negative value added in empirical studies carried out in connection with the ef- fective protection concept corroborated this point.”3 A negative value added measured at world prices signals the existence of domestic production resulting from the preponderance of high tariff protection and that would not exist if free trade would be permitted. Exposing Brazil to a greater degree of foreign com- petition by reducing its excessive levels of protection may be the single most effective step toward a growth of the country's exports of manufactures. The longer run effect of this measure will be to expand Brazil's exports of manufactures along the line suggested by the principle of comparative advantage. FOOTNOTES: CHAPTER IV lirazil has the climate and the soil that is suitable for the production of many important trepical crops. These natural conditions enabled Brazil to establish herself as the world's largest producer of coffee. At the turn of the century Brazil accounted for over 80 per cent of the world coffee production but has been falling since, in 1958-59 Brazil's share fell to 50 per cent. Sugar, cotton, tobacco, cocoa, oranges and rice are also produced in sub- stantial amounts. Another natural resource that is plentiful in Brazil is forestry. According to FAO's World Forest InventoryJ Brazil in 1958 had 12.7 per cent of the world's forests. It must be added that most of these are in the sparsely populated Amazon basin and therefore present more of a future potential than actual reality. Brazil also has very large deposits of several important minerals. The country has almost one—quarter of the world's reserves of iron—ore, second only to the Soviet Union in that respect. Almost all of Brazil's iron-ore deposits are located in the state of Minas Gerais and most of it is high quality ore. For details see: Dale, op. cit. ll 2Leontief, "Domestic Production . . ., Op. cit., aéso; Leontief, "Factor Proportions . . .," op. cit., 3 6—407. For the Japan—United States application see: M. Tatemoto and S. Ichimura, "Factor Proportions and Foreign Trade: the Case of Japan," Review of Economics and Statistics, XLI (November, 1959), HD2-H6. For the United States-Indian case consult: R. Bharadwaj, "Factor Proportions and the Structure of Indo- U.S. Trade," Indian Economics Journal, X (October, 1962), 105—16. 3 ”The theoretical basis and some evidence for the existence of factor intensity reversal internationally is contained in: B. S. Minhas, An International Com- parison of Factor Costs and Factor Use (Amsterdam: North- Holland Publishing Co., 1963). For a critical appraisal of this possibility in light of additional empirical evidence see: H. B. Lary, See: Hufbauer, op. cit., Table 3, pp. 58-70. 71 72 Imports of Manufactures From Less Developed Countries, National Bureau of Economic Research (New York: Columbia University Press, 1968), pp. 51-58. 5The United States data on value added, payroll and employment for 17 industries were obtained from: United States Bureau of the Census, Annual Survey of Manufactures 1959 and 1960 (Washington, D. C.: United States Govern- ment Printing Office, 1962). , 7 For Brazil see Data Appendix (Appendix D). 6The rationale for the use of wage and salary com— ponents as a proxy for human capital and that of the non- wage value added as a proxy for physical capital used in industry, is contained in: Lary, op. cit., 20-22. 7Acknowledging the sensitivity of my result to the choice of weights, I performed the previous set of calcu- lations, using as weights the ratio of the country's two— digit commodities to her manufactures exports and imports (SITC 5, 6, 7, and 8). Capital/man Skill ratio Wages/man imports $15,999 .113 $662M.“ Exports $1U,868 - .072 $6u16.8 While the differences in capital per man and wages per man embodied in Brazil's exports and imports are very small, the large differentials in the skill content of Brazil imports and exports were still preserved. In no case did I get the "perverse" result that Brazil manu- facturers exports could be characterized as containing more capital or skill, than her manufacturers imports. 81h 1964-67 some 8“ per cent of Brazil's manufac- tures imports (SITC 5, 6, 7, and 8) originated in North America and Western Europe. Calculated from Statistics in Data Appendix. 9Linnemann demonstrated that the volume of trade will be systematically related to the distance over which transportation takes place. The resistance to trade is directly related to the distance between the trading na- tions and includes elements such as transport cost, trans- port time and "psychic" distance, the latter refers to the relative familiarity with markets and customs of neighboring countries. These factors no doubt will ef- fect trade among LAFTA members and especially among the two large neighboring countries, Brazil and Argentina. Among LAFTA countries Argentina is by far Brazil's most important trading partner. Both countries have relatively large industrial sectors and their commercial centers are 73 Close. In addition both countries have instituted tariff reforms that were bound to effect their trade volume. For detailed analysis see: H. Linnemann, An Econo- metric Study of International Trade Flows (Amsterdam: North-Holland Publishing Co., 1966), pp. 27-37. TABLE U.6.--Brazi1's trade with Argentina and LAFTA countries, 1960-1965 (in million dollars). Exports Argentina 56.U 67.“ U8.“ ”6.2 90.8 1U0.9 LAFTA 86.A 95.2 75.7 76.0 132.7 197.A Imports Argentina 9U.8 29.8 85.5 87.9 116.3 131.9 LAFTA 108.3 Us.2 128.6 173.9 167.9 190.u Source: Fundacao Getulio Vargas, "Foreign Trade-Brazil in the LAFTA," Conjuntura Economica (January, 1967), 17. 10The industrial countries proportion (Pi) of Brazil's total trade were calculated based on the following formula: Ci + Ui + W1 r1 = Q Where: Ci = dollar shipments from Brazil to Canada of the 1th two-digit commodity. 01 = dollar shipment from Brazil to United States of the ith two-digit commodity. Wu = dollar shipment from Brazil to Western Europe of the ith commodity. Q = Brazil's total imports (or exports) of manu- factgred goods in SITC categories 5, 6, 7, V and . Thus, Pi gives us the proportion of industrial countries' trade in manufactured goods with Brazil in each two-digit SITC category. llThis view was stressed in: A. D. Hirschman, "The Political Economy of Import—Substituting Industrialization in Latin America," Quarterly Journal of Economics, LXXXII (February, 1968), eSpecially 13-17. 7A 1” ‘Stanley J. Stein, The Brazilian Cotton Manufacture: Textile Enterprise in an Underdeveloped Area,~1850-1950 (Cambridge, Mass.: Harvard University Press, 1957), p. 180. 13 Ibid., p. 182. luBased on this approach a country exports only the "surplus" that is left over after the domestic market has been adequately supplied. The idea of concentrating output in specialized production for the export market is shunned. These ideas are developed in: N. H. Leff, "Export Stagnation and Autarckic Development in Brazil, 19u7- 1962," Quarterly Journal of Economics, LXXXI (May, 1967), 286-301. r 1)The nominal and effective tariff rates levied on imports of three labor intensive commodities into the United States and the Common Market are reported by Balassa (see Table U.7). TABLE A.7.—-Homina1 and effective tariff rates levied on selected imports. United States Common Market Nominal Effective Nominal Effective Textile fabrics 2U.1 50.6 17.6 flu.“ Wood products 12.8 26.“ 15.1 28.6 Leather 9.6 25.7 7.3 18.3 Source: B. ialassa, "Tariff Protection in Industrial Countries: An Evaluation," Journal of Political Economy, LXXIII (December, 1965), 573-97, Table 1. 16 For detailed explanation and economic rationale for these measures see: Lary, op. cit., pp. 20-35. l7Kenen, op. cit., u37-U2. 18Among the most notable contributions in this area are: Theodore W. Schultz, "Reflections on Investment in Man," Journal of Political Economy, LXX, Supplement (October, 1962), 1-8, and Becker, Human Capital: . . ., op. cit. For a more recent contribution see: Stanley M. Besen, "Education and Productivity in U. S. Manufacturing: 75 Bomb Cross—Section Evidence," Journal of Political Economy, LXXVI (May/June, 1968), U9U—97. ( 1)For a rigorous presentation see: Gary Becker, Investment in Human Capital: A Theoretical Analysis," Journal of Political Economy, LXX, Part 2 (October, 1962), 9““9 0 20Helen Waehrer,'1nter-industry Skill Differences, Labor Earnings and U. 8. Foreign Trade, 1960" (unpublished Ph. D. dissertation, Columbia University, 1966). 21The six occupational classes are: I--Professional, technical and kindred workers; II--Managers, officials and proprietors; III-~Clerical and kindred workers; IV-- Sales workers; V--Craftsmen, foremen and kindred workers; VI--Service workers. For details see: Waehrer, "Inter—industryy. . .," op. cit., pp. 92—97, Table 3.3. ’3 “2Contained in: IBGE, "Service Nacional de Recenseamento," Censo Industrial de 1960, Brazil, p. 62. 23Ibid., p. 63. 2“The skill indexes of each industry are then divided by the percentage of employees classified as skilled in all transformation industries. Therefore in effect we: obtain a relative skill quotient of each major industry group. ’3 LSFor the difference between these two non-parametric test see: S. Siegel, Nonparametric Statistics for the Behavioral Sciences (New York: McGraw-Hill, 1956), pp- 229-399 26From: Dale, 0 . cit., pp. 1-3, and Survey of the Brazilian Economy 1966 (Washington, D. C.: Brazilian Embassy, 1967)- 27For an interesting and detailed discussion of the role of coffee in Brazil's economic history (especially in the State of Sao Paulo) see: William H. Nicholls, "The Transformations of Agriculture in a Semi-Industrialized Country: The Case of Brazil" (presented to the Conference on the Role of Agriculture in Economic Development, spon- sored by the NBER, New York, December 1—2, 1967), especially DD~ 15-31- ") ‘BSee Table 4.8, p. 76. 76 TABLE A.8.—-Brazil's exports of agricultural commodities, 1952-5“ and 1962-6A annual averages (in millions of dollars). SITC No. Commodity 1952—5“ 1962—64 Exports of each commodity as of total exports 061 Sugar 13.30 48.51 .030 071 Coffee 1029.95 716.88 .530 072 Cocoa 9H.50 “5.88 .030 2a Wood 38.68 u5.11 .030 121 Tobacco 17.8U 25.53 .020 262 Wool 9.27 9.11 .007 263 Cotton 123.5 11H.67 .085 All Exports 1503.3 1350.0 .732* Notes: * = Proportion of all commodity exports listed in the table as of total exports, 1962-64. Source: Statistical Office of the United Nations, Year- book of International Trade Statistics (New York: United Nations, various years). ‘ ') ‘9soonomic Commission for Latin America and the Food and Agricultural Organization of the United Nations, Coffeeyin Latin America, Vol. II: Brazil, State of Sao Paulo (Mexico: United Nations, 1960), pp. 1-111. 3OOther commodities grown in the different regions of the State are: sugar cane (the Southern region), cotton (the Western region), also maize, rice and beans are grown in smaller quantities. Taken from: Ibid., pp. 1U—18. 31One of the main differences between coffee pro- duction in Brazil (chiefly in the State of Sao Paulo) and that in the so—called "mild coffee" producing countries such as Columbia, is the absence of shading of coffee plants in Brazil. In Sao Paulo the coffee plantings are grown in the open field, exposed to direct sunlight. This practice together with the topography of the coffee 77 {cones of .".:m I‘:vulo is responsible for the typical produc- tion pructicvs, which are generally simpler and more labor intensive than those used in Columbia and Central America. For details see: ECLA and FAQ, Vol. II, op. cit., Part I, Chapter III, p. 39. 32This is also true of coffee production in the so- called "mild coffee" producing areas of Latin America (e.g., Columbia). A FAO study of coffee growing in Columbia noted ". . . Columbian coffee--with specific exceptions--is almost exclusively the product of the land and labor factors." Of the total cost involved in coffee production, between 75 and 90 per cent is represented by the work of the farmer and his family or by day labor. ECLA and FAQ, Coffee in Latin America, Vol. I: Columbia and El Salvador (New York: United Nations, 1958), especially Chapter V, pp. u2-u3. For a more thorough discussion of overall produc- tivity trends and wages in Brazilian agriculture consult: William H. Nicholls and R. M. Paiva, "Structures and Pro- ductivity of Brazilian Agriculture," Vanderbilt University and Funcacao Getulio Vargas, undated. (Mimeographed.) 33Weedihg is typically done by hand with the aid of hoes. The picking operation involves thrashing the coffee trees with sticks, and then picking and separating the fallen coffee berries from other materials. Non-routine operations include replanting, fertilization and soil management practices. ECLA and FAO, Vol. II, op. cit., Chapter LV, Table 15, pp. 36—UO. 3ulbid., Part I, Chapter V, p. u8. 35w. iaer, Industrialization and Economic Development in Brazil (Homewood, 111: Richard D. Irwin, 1965), Table 3A-l3ng7, D- 951. 36 See Table “.9, p. 78. 37ECLA and FAO, Vol. II, op. cit., Part I, p. Al. 38One of the more interesting findings of the coffee survey was that regardless of their size, all coffee farms seem to use essentially the same (labor intensive) pro- duction technique. Furthermore, as can be seen from Table A.lO (see page 78) coffee growing can be classified as labor intensive relative to other agricultural com- modities, that were grown on these farms. ' 78 TABLE “.U.—-Total capital investment on coffee farms, 1958 (per cent distribution). Type of investment Per cent per thousand Per cent per producing trees hectare Coffee land 26.3 26.1 Producing trees 51.6 51.3 Housing units 10.5 10.9 Storehouses, other buildings 2.0 1.9 Work animals, equip— ment, vehicles for animals 1.1 1.1 Mechanical power vehicles and equipment 3.8 3.8 Electrical power plants, water pumps, and tools .“ .7 Buildings and equipment for coffee processing “.2 “.2 Total 100.0 100.0 Source: ECLA and FAO, Coffee in Latin America, Vol. II: Brazil, State of Sao Paulo (Mexico: United Nations, 1960), Part I, p. 63, Table 50. TABLE “.lO.——Labor input required in growing of coffee and other crops (man—days per hectare). Crop CrOp , Man-days per hectare Maize 20.9 Rice 39.8 Castor bean 32.8 Sugar cane 65.6 Coffee 73.“ Source: ECLA and FAO, Coffee in Latin America, Vol. II: Brazil, State of Sao Paulo (Mexico: United Nations, 1960), Part II, especially Tables 16, 22, 3, 2“, and 25, pp. 38-“8. ' 79 ( 3)In Brazil the production of cocoa is concentrated in the State of Bahia. As in the case of coffee, specific climatic conditions (those prevailing in the tropical lowlands) and soil conditions are required for cocoa farming. Again the various activities involved in the cultivation of the cocoa trees, such as clearing, shading, pruning, and harvesting, require large proportions of unskilled (low wage) labor. For details see: V. D. Wielcizer, Coffee, Tea and Cocoa (Stanford, Calif.: Stanford University Press, 1951), especially Chapter 13, pp. 282-92. Information as regards cotton production in Brazil was found in: Frank D. Barlow, Jr., Cotton in South America: Production, MarketingJ Consumption, and Devel- gpments in the Textile Industgy (Memphis, Tenn.: National Cotton Council of America, 1952), Chapter II. uolnternational Monetary Fund, International Financial Statistics, XXIII (February, 1970), 58-59. ulMichael Sieniawski, "Favorable Trade Balance Boosts Brazil's Outlook," The Christian Science Monitor, Feb. 26, 1970. uzUnited Nations, Proceedings of the United Nations Conference on Trade and Development, Vol. II, Policy Statement (New York: United Nations, 196“), p. 1“. 73For a full discussion of this concept see: S. E. Guisinger, "Negative Value Added and the Theory of Ef- fective Protection," Quarterly Journal of Economics, LXXXIII (August, 1969), “15—33. C} H) I THE“. ".7 _l .‘ TWIE iiVVtXTVS f)? IWiOWWWTTICWJ URI lNCCNAE 45.;J DISTRIBUTION: THEORETICAL SSUES A. The "Classical" Doctrine and Income Distribution The theory of comparative cost has been associated with the work of the classical economists, primarily David Bicardo and John Stuart .‘iill.l It was further elaborated upon by Marshall, Edgeworth and Mangoldt among others.2 The principle of comparative advantage stresses the gains from free international trade, as a result of letting each country specialize in the production of a commodity it makes relatively more cheaply, or at a lower oppor- tunity cost. Given a fixed amount of productive resources, and a free exchange of goods, the principle of comparative advantage demonstrates that it is possible for each par— ticipant in a voluntary exchange to consume more of at least one commodity, while consuming no less of the other, i.e., increase its real income. The discussion of the effect of trade on income dis— tribution was taken up initially in connection with the existence of non—competing groups or "specific" factors 3 of production. The neoclassical economists generally 80 8i L'(-’(_:U,fo1_l°'zjt?m Umpnwfims webmaocfi coapomuopm Hmpoe + .cofiumcmaqu mom pmummno man» no coapomm wsofi>mpa momIIMMHmmp Umpzwfimz mmmpm>< ++ .Hm>ma ammo you mco on» no pamOfiMchfim * ”mmpoz .AQ xflocmdamQ HmCOHumc Isoch pom zocmw< .m.b =.HHNMLm CH meoHHom pLOQEH cam ocmEmQ pLOQEH: .mmoxmmamz Ucm meHo “mom mmESfiEmsQ mmcmzoxm mafia “Hoanfio .Azmma “nopmzv xH .mbfipms< :fipmq pom capoflasm oHEocoom :amofiLmE< :fiumq CH cofipmwfiamfinpmsccH Ucm EmHQOHpomp loam: aofipmomz "mom mmwgfipmp HmcHEoc ommso>w moosaocfi cofipomuopa Hmpoe .H magma .smH .Asmma .HHLQqu «Naocoom HmOHpHHom so Hmcmsos =.msmcm m.sonmq Ucm mmmfisme m>auommmm mmumpm ompficb: .Hamm Eopm who mofiumfipmpm .m.: one swoopsom mCHLSpommzcmE mmo.+ .m.c .m.: mmm.+ Hom.+ mp copes mSHw> mo chasm .mmwmz U. M . mpmxmoz cofiposo *mmm.n *mmm.| *Hmm.l *mwm.l *mmo.l IOLQ no mmwmz . Hmsccm mwmpm>< wmo%0HQEo *ssu.- *asm.- *am:.- *sms.- *sss.- Ham so mmwmz Hmsccm mmmsm>< Ammmfiv +Ammmav ++Ammmfiv +Ammmfiv Ammmfiv mmfispmsocfi mmfispwsocH mm mofimuwsocfi mm mmampmsocfi om mmfippmsocfi om Hmuummfipmp Incofipompopq Inmgfipmu Has IIQOHpoopopQ IIMMHAmp Hm: m>apommmm .m.: Hobo» Hammsm IHEo: Hammpm Hmpou Hammsm IHEoc Haumsm .Ammmav mmpmpm empacb on» Gem Ammma .wmmHV HHNmsm .momms pew coapomposo cmmzumn mpCmHonmooo coapmHthoo xcmmuu.s.m mamas corresponding Balassa—Ball results are shown. Table 6.7 includes results based on two samples of Brazilian indus- tries. The first sample includes 20 major industry groups while the second sample includes the earning's and protection characteristics of 58 smaller industries obtained from the Brazilian Industrial Census of 1960. The rank correlation coefficients for the two groups of Brazilian industries were calculated with nominal tariffs and total protection rates, instead of effective tariffs used by Ball and Balassa. For the sake of comparison, I included in Table 6.7 the results for the United States as computed by Ball. Figure 6.1 illustrates the apparent inverse rela— ' of production workers tionship between average "wages' and nominal tariffs.“O The inference that can be drawn from Table 6.7 and from reading the scatter diagram (Figure 6.1) is that the Brazilian structure of protec- tion (both tariffs and exchange premiums) accords rela- tively heavier protection to low wage industries. This is contrary to expectations since unskilled labor is the countries' abundant factor. I was unable to find a significant relationship between the share of wages in value added and the extent of protection accorded to the products of 20 industries. I was struck by the similar- ity of the structure of protection of Brazil, an under- dezeloped country, and the United States, a highly Tariff (per cent) manufacturing industry by major industry groups, 1959- Source: Notes: 1. Lumber 2. Furniture 3. Furs & skins U. Textiles 2. Clothing 126.00 ? 111.00 « 96. 81. 66. 51. 36. 21. 00.L 00 «r- 00 .» 004+ 00 56.00 3.6 63.00 126 '* L v 70.00 .13 77.00 i1 19 .18 A A. v 8Q.OO 12 v 91.00 .20 5 A fl 98.00 105.00 Production Workers Earnings/empl (1000's cr.) Figure 6.l.-—Tariffs and production workers' average earnings in Brazilian Food 7. 8. 9. 10. 11. See Data Appendix. Names of the industry groups: Beverages 12. Tobacco 13. Miscellaneous Non-metallic in. minerals 15. Metallurgical 16. Mechanical Paper & paper products Rubber Chemicals Perfumes 17. 18. 13. 20. Plastics Editorial & graphic Electrical & communication equip. Transportation equipment *Average of all manufacturing industries (the 20 included above). 127 industrialized country. This is paradoxical in the light of what must be a generally accepted wide disparity be- tween the two countries' "man-made" factor endowments (i.e., human and "physical" capital). 2. Skill Composition, Capital Intensity and Protection across Brazil's Industries Given the paradoxical results reported in Section D-l, it is necessary to proceed with a more complete analysis of the relationship between industries' factor intensities and the structure of their protection. Specifically, I will examine the relationship between the industries' skill intensity and the protection accorded their product. In Chapter IV, Hal B. Lary's method was used to approximate the flow of human and "physical" capital into the industries' output.“1 In that context value added by manufacture was employed as a reasonable index of the aggregate flow of man-made resources into the production process. Value added for each Brazilian industry was ob- tained by subtracting the value of "materials" (that in- cluded raw materials, supplies, fuel, electric energy consumption, cost of resales and miscellaneous receipts) from the value of shipments. Subseouently, value added in each industry was broken into two components: (1) wages and salaries and (2) the remainder--the non-wage component of value added. Both 1 and 2 were deflated by the number of production workers in each industry. Thus, the industries' value added was ascribed to both human and "physical" capital. The wage-and-salary component of value added was taken as a reasonable proxy for the flow of human capital into the production process, while the non-wage-salary component was taken as a proxy for the contribution of "physical" capital to the production process. Based on these indexes of human and "physical" capital, Lary's procedure was followed in arraying the 20 Brazilian manufacturing industries relative to all transformation industries. To examine their relative factor intensities, I divided the 20 major industries into four groups. Group I industries (south—western corner in Figure 6.2) use unskilled labor intensively. Typically, these industries are below the average in terms of wages and capital per worker. In 1959, Group I included the following industries: textiles, clothing, lumber and wood products, furs and skins, furniture and non- metallic minerals. Group III industries (north-eastern corner) are intensive in skill (human capital) and physical capital relative to the average of all transformation industries. They are characterized by high average wage as well as high non-wage value added per employee. Group III ircluded the following industries: transportation, Total Earnings/Employee (1000's cr.) 182. 166. 150. 13“. 118. 102. 86. 70. 00+» 000 00$ 00.. on 00“ 00 12 18 129 .19 07 ’17 J3 ~20 .16 .15 AA 1U.70 25.30 35.90 06.50 57 10 67 70 78.130 88.00 Non Wage Value Added/Employee (X101) (1000's cr.) Figure 6.2.--Wage and non-wage value added per employee by major industry groups in Brazil, 1959. Source: Notes: “Average for all manufacturing industries. See Date Appendix. For names of industry groups see Figure 6.1. 130 electrical and communications equipment, chemicals, plas- tics, perfumes, rubber, metallurgy, beverages and paper. The remaining industries can be divided roughly into two groups. Group II (south-eastern corner) includes the tobacco and food industries. They have a higher than average ranking in non—wage value added, but rank low on the average wage scale. Finally, Group IV (north-western corner) is characterized by higher than average wages; it includes mechanical, editorial and graphic and miscel— laneous industries. They are somewhat below the average in the non—wage value added per employee. Miscellaneous industry is on the border between Group I and IV and can be described as unskilled—labor intensive. Lary ranked the major United States industries on the basis of the wage and non-wage components of value added for 1965.142 The similarity of the two rankings presumably reflects the absence of factor intensity re- “3 versals between the two countries. Harry Johnson noted that Lary's index of overall factor intensity may reflect, in addition to human and physical capital inputs, also some neo-technological factors, i.e., economies of scale, product age and differentiation; to the extent that those are reflected in the final goods' selling price.1414 Is there any justification for using the non-wage value added instead of the more commonly used capital- labor ratio? Lary argues that it may be preferable to 131 use the "flow" concept of capital whenever dealing with the contribution of a factor input to the industry's pro- duction of goods and services.“5 This will be a more inclusive concept than the conventionally used capital— labor ratio. Some components of the non-wage value added figure are quite closely associated with the physical "stock" of capital; for example, rent, interest payments on borrowed capital and payments to proprietors and part- ners. Other expenditures, such as those related to ad- vertising, legal services and travel, I will assume to be randomly distributed among industries, since no de- tailed information is given on their breakdown. Next, I proceeded to test statistically whether inter—industry differences in non-wage value added accu- rately reflect the "physical" capital intensity, as indi- cated by the industries' capital-labor ratio. The Brazilian Industrial Census provided the data on fixed physical assets for twenty industry groups in 1959. The reliability of the capital assets series is questionable because it combines data on equipment, buildings, and land acquired at different times and at different prices. De- spite these reservations, my results support the asser- tion that a significant association exists between non— wage value added per employee and physical assets per employee (the capital-labor ratio). The correlation co- efficient of these two variables is .677 significant at 132 one per cent. Using a logarithmic transformation of the same variables, the relationship is even more pronounced, yielding a correlation coefficient of .727 significant at one per cent.)46 Finally, I attempted to determine whether a system- atic relationship exists between the skill composition of Brazil's industries and the extent of their protection. In order to achieve such a comparison, I devised in Chapter IV an appropriate measure of the human-capital intensity of an industry and expounded its rationale. By employing skill indexes II and III, I wanted to establish whether the industries that are human capital intensive also use relatively large proportions of "physical" capital. In effect, the inquiry was whether the two "factors" are used in cooperation to produce a high "overall" factor intensity in certain industries. As was noted by Hufbauer, skill—intensive goods are likely to overlap with capital—intensive goods, because the acqui— sition of both human skills and of "physical" capital in— volves acts of investment (and saving). Therefore, I in- quired into the nature of the relationship between inter- industry differences in capital intensity-~measured as a flow or a stock-~and the skill-index of the sample of 20 industries. As reported in Table 6.8, a significant and positive correlation (though not very high) has been ob- tained between the two skill-indexes and the "physical" caoital intensity of Brazilian industries. TABLE 6.8.——Correlation coefficient between capital inten- sity of 20 Brazilian industries and skill indexes--l959. Skill index-II Skill index—III Non-wage value added per .556** .681* production worker (2.836) (3.336) Gross capital assets per .688* .691* production worker (H.026) (H.059) Source: See Data Appendix (Appendix D). Notes: Numbers in parentheses, below correlation coefficient, are the corresponding t-value. * Significant at one per cent. ** Significant at five per cent. Finally, I resumed the empirical examination of the relationship between the human-skill intensity of differ- ent industries and the level of protection accorded the industries' product. Two measures of protection were used--the average nominal tariffs and the measure of total protection, i.e., the combination of tariffs and exchange premiums that have been derived earlier in this chapter. From Table 6.9 and Figure 6.3 the negative relationship between skill indexes—II and III and the industries' level of protection becomes apparent. Based on these findings it is therefore possible to conclude that the Brazilian industries containing a high proportion of skilled labor also have a high capital-labor “7 Patio, measured either in stock or flow terms. Further- more, the structure of Brazilian protection bears a unique l3U TABLE 6.9.--Gorrelation coefficients between tariffs and total protection and skill indexes for 20 Brazilian indus— 1959. try groups, Skill index-II Average nominal —.635* tariffs (3.U91) Total protection -.7OU* (tariffs plus (H.215) exchange premiums) Skill index-III -.665* (3.783) _.751* (H.830) Source: Notes: See Data Appendix (Appendix D).. Numbers in parentheses below correlation coefficient are the co-responding t-value. Significant at one per cent. Significant at five per cent. 126.00w 111.00« 9 96.00% a 81.00» .16 66.00« Tariff (per cent) 51.00# 36.00w 21.00 A a V “2.00 72.00 102.00 1J35 .1“ .20 .ll .17 18 .19 .12 15 A L A A V v v r 132.00 162.00 192.00 222.00 252.00 Skill Index Two Figure 6.3.--Skill indes-II and average nominal tariff rates of 20 industry groups in Brazil, 1959. Source: See Data Appendix. Notes: For names of industry groups see Figure 6.1. “Average for all manufacturing industries. 136 relation to what I defined as the industries' skill quo- tient or skill index. Brazil accords relatively heavy protection to those industries that have a low skill con- tent—-industries characterized by a high proportion of unskilled labor relative to human and physical capital. This must be termed a paradoxical result in light of the extremely unequal distribution of skill or human capital between Brazil and its main trading partners, the United States and Western EurOpe. In Chapters III and IV, I have concluded that Brazil is a human capital "poor" country compared to its main trading partners, and that she reflects this economic reality in her commodity trade. The country is a net importer of skill intensive commod- ities and exports goods that are "nature” and unskilled labor intensive.“8 Hufbauer's and my calculations accounted for only the direct factors going into the final productive pro- cesses. One can question whether indirect inputs should be included, such as those used in producing the materials that go into the final good. First, I found no evidence to suggest that even if it were possible to account for indirect factors, there would be any serious reversals of the major conclusions. Secondly, and more importantly, when considering industries' comparative advantage, materials not possessed by a country can usually be ob- tained in international trade. A labor abundant country, for example, producing and exporting a labor intensive good (such as cotton textiles) can import the needed modern machinery, though that equipment itself may be capital-intensive. In fact, to include indirect factors in such cases does not fit with the very purpose of ex- plaining international specialization and trade. The problem of material inputs may loom important especially in the food processing industry. In these instances the location of the processing industries is determined more by the availability of the "raw" materials (often perish- able) than the relative requirements of capital and labor used in their processing. At least for these types of industries the investigation ought to be pushed back to find out what determines the production characteristics of these material inputs. Brazil, and other skill deficient countries, rely on their commodity trade to supplant their relative scarcity of human capital.149 Thus, for Brazil a theo- retically "correct" structure of protection should accord relatively high protection to those industries that use the country's scarcest factor intensively, i.e., the human capital intensive industries. In fact, one finds that Brazil accords relatively high protection to its unskilled labor intensive industries like the textile, clothing and leather industries. In Chapter VII the effect of the structure of Brazilian protection will be incorporated into a more general discussion of wage determination in Brazilian industry. FOOTNOTES: CHAPTER VI 1For complete analysis of the operation of exchange controls in Brazil consult: Donald Huddle, "Furtado on Exchange Control and Development: An Evaluation and Reinterpretation of the Brazilian Case," Economic Devel— opment and Cultural Change, XV (April, 1967), 269:85. 2The overvalued exchange rate, in effect, subsidizes selected imports. A number of goods considered essential, such as drugs, insecticides and fertilizers could be im- ported freely. Fuels, some foodstuffs, cement and paper and printing equipment received priority in obtaining licenses. Next in the order of priorities came the machinery and equipment needed for modernization of industry. The goods considered non-essential by the government were closely restricted. The incentive feature of the licensing system was limited to assemblers and manufacturers, who had to rely on imports of components and raw materials, although the latter theoretically had priority over finished goods. 3Official Instruction No. 70 of SUMOC (Superinten- dencia da Moeda e Credito) as ratified by laws 21A5 and 2A10 of December, 1953 and January, 1955. “For detailed discussion of the actual operation of the exchange auction system see: Donald Huddle, "Dis— equilibrium Systems, Industrialization and Inflation: The Brazilian Case" (unpublished paper no. 30; Yale Uni— versity, Economic Growth Center, July 17, 1967). 5International Monetary Fund, International Finan— cial Statistics, VIII (February, 1955). 6A special agency was set up to determine whether domestic suppliers could furnish a given product in "sufficient" quantity and acceptable quality. If a positive determination was made, the good was registered as a similar, and its importation was discouraged or totally banned. 7L. Gordon and E. L. Grommers, United States Manu- facturing Investment in Brazil: Impact of Brazilian Government Policies, 1936-1960 (Cambridge, Mass.: Harvard University Press, 1962), p. 23. 138 139 8 M. E. Kreinin, Alternative Commercial Policies: Their Effect on the American Economy, Institute for Inter- national Business and Economics Studies (East Lansing, Mich.: Michigan State University, 1967), pp. 3H-37. 9Among several recent studies see for example: W. M. Corden, "The Structure of a Tariff System and the Effective Protective Rate," Journal of Political Econom , LXXIV (June, 1966), 221—37, also Balassa, op. cit., 573- 9“, also H. Johnson, "The Theory of Tariff Structure with Special Reference to World Trade and Development," Trade and Development (Geneva, 1965), also G. Basevi, "The United States Tariff Structure: Estimates of Effec- tive Rates of Protection of U. S. Industries and Indus- trial Labor," Review of Economics and Statistics, XLVIII (May, 1966), 137-60. For a recent contribution in which some of the restrictive assumptions often made with respect to ef- fective tariffs are relaxed see: J. Clark Keith, "Sub- stitution and Supply Elasticities in Calculating the Effective Protective Tariff," Quarterlngournal of Economics, LXXXII (November, 1968), 588-601, also William P. Travis, "The Effective Rate of Protection and the Question of Labor Protection in the United States," Ege gournal of Political Economy, LXXVI (May/June, 1968), 3— l. lOSantiago Macario, "Protectionism and Industrializa— tion in Latin America," Economic Bulletin for Latin America, IX (March, 196U), 61-101. 11For example, the overall unweighted averages of import duties at the beginning of 1960 were: 151 per cent in Argentina; 93 per cent in Chile and 60 per cent in Brazil. In contrast, the corresponding incidence for France——a high tariff country--was only 18 per cent. See: Ibid., 73, Table A. 12See Table 6.10, p. 139 for details on the inci- dence of the tariffs in Brazil, Argentina, Chile and the E.E.C., based on commodity groups. 13The correspondence between the EEC and United 3 ates Tariff rates is quite close for most commodities. The major difference is that EEC tariffs on food products are higher relative to the United States; while on non- food imports tariffs are somewhat higher in the United States . For detailed study see: Committee for Economic Development, Trade Negotiations for a Better Free World Egpnomy: A Statement on National Poligy by the Research and Poligy Committee (New York: Committee for Economic Development, 196E), Appendix B, pp. 67-78. lUO .mpfimoooo LOHLQ ocp mcfiocmcfim mo pmoo ho xmp upmpcoEmHQQSm on» oozaocfi momsmco mop .zpommumo :Hmfiomom= map CH oofimfimmwfio mppooefi pom omcmnoxo cwflopom mo mafino pom pmoo Locwfln on» ocm Amuosoosa pmoE pom pcoo Loo m psoomv xmu mocmmmoao mEOBmzo map omam HHNmLm no omfiaoom mmmpmco ocm mofipzo 05p on cofipfioom CH mmozaocfi magma mace ”mmpoz .m manme .ms .Azoma .copmsv xH .mOHsme< :Hpmq sou cfipmaasm oHEoCOQm :.w0HLoE< :Hpmq CH cofiummfiamfippmsocH ocm EmficofipoopOLm: .ofimemz .m ”oohsom ma ma ma NH 0H Hr-{m (\J H m H I \J] cum mma mmm mm: mam om as ea \C)r--1L(\ ~J'r—1—TJ V H ,i\ (1‘) l” mHHQQ me 22m 0mm owm wmm ow mza smfi Hammpm HMH moa mmfi wma wow mo mma mm mm mma mm mcfipcomp< Amposoopa mmav ommpo>m Hamso>c Amposooso mav mpozpo .m Ampozoopo :HV mooom oommooosm .H Amposoopa Hmv mLoLSpommscmE LoE3mcoo unoppso ”HHL muomoumo Ampozoopo HHV mooom amazmcoo oaowpso .m Amposoopo mmv mooom oopsuomuzcwEIHEom .H Amposoopq mzv mooow oaomsso Ugo ooLSQOmmscmEIflEom ”AH knowoumo Amposoopc mmv wooom Hmuflamm .m Amposoomo QHV wamflpopme amp HmflppmsocH .m Ampozooso mHV mmmzpmooom commooosocoz .H Amuozoopo Hmv mooom Hmpfiomo ocm moHpHUoEEoo zpmeflpm ”H mpowoumu QSOLQ ocm mpomopmo .cwmfi .Ammmmpcoopmo CHV cmOHpoE¢ :Hpmq mops» CH oofiaoom mommm mm no fismb Hmspopxm m_omm ocm mmfihpQSOQ ecm mmfipse no mamas ossmsnpfism magsflm--.OH.m mamas 1Ul 1“For details consult: Macario, op. cit., 77. On the structure of the European tariff consult: United Nations Commission for Europe, Committee on the Deve10pment of Trade, Report on the Special Meeting_on the Organization and Technigues of Foreign Trade, June 29- July 3, 1959 (Geneva: United Nations, 1959), Annex IV. 15 16Tbid., p. 5. Clark and Weisskoff, op. cit. 17Ibid., p. 39, Table B—2A. 18Ibid., p. U0, Table B-2B. ( 1)The basic data on tariffs are included in: Law no. 3.2M“, Tarifas das Alfandegas, issued August, 1957. 20That study reports the EEC and United States imports for 1959 and 1960. See: K. F. Topping, Com- parative Tariffs and Trade (New York: Committee for Economic Development, March, 1963). 21The Macario study established the existence of similarity in the structure of all of Latin American tariffs (including that of Brazil). His conclusion is that ". . . the structure of import duties and charges in most of the Latin American countries is characterized by its lack of rationality and by the prevalence of excessively high rates, as regards both average levels and those applicable to the vast majority of individual products." See S. Macario, op. cit., 67. 22From: Ibid., Annex II, Table B. ’) L3For a discussion of these summary measures see: W. I. Greenwald, Statistics for Economics (Columbus, 0.: Charles E. Merrill Publishers, 1963), Chapter II. 2“There were two main currencies and several sub— sidiary currencies that were purchased at these periodic auctions. These were the United States dollar and the "ACL dollar," the latter refers to certificates provided for under a multilateral trade and payments agreement between Brazil and certain Western European countries. Allocation of exchange within the system generally made non-dollar imports cheaper and also imports in the ”special" category were much more expensive than those in the "general" category. See: Dale, op. cit., pp. U8-SO. 142 '3!— “JTaken from: IMF, Ninth Annual Report on Exchange Restrictions, 1958 (Washington, D. C.: IMF, May, 1958) pp. 61-63. 26 Clark and Weisskoff, op. cit., p. H5, Table B-SB. 27Ibid., p. 21. 28The classification of each industry's product in the "general" or "special" exchange category was based on: Dale, oo. cit., Appendix A: Brazilian Customs Tariff, pp. 69-73. ( 2)I take the term "nature" to mean relative abun- dance of unskilled labor and certain natural resources. The argument can be legitimately extended to most indus- trial countries with whom Brazil's commodity trade is concentrated. First, the tariff structure of the United States and other industrial countries were found to be quite similar. See: Balassa, op. cit., 586. Second, industrial wage structure among developed countries also exhibits striking similarities. For empirical evidence on this consult: I. B. Kravis, "Availability and Other Influences on the Commodity Composition of Trade," Journal of Political Economy, LXIV (April, 1956), 1&3-55; also: D. S. Ball, "United States Effective Tariffs and Labor's Share," Journal of Political Economy, LXXV (April, 1967), 186. 30Mrs. Vaccara grouped protected industries into five protective classes ranging from "unprotected" (Class I) to "heavily protected" (Classes Ma and Nb). Her classification is based on both the proportion of each industry's output that is protected and also, ac— cording to the average ad valorem tariff applied to the industry's product. The measure of labor use is the number of employees per million dollars of shipment of an industry. Labor cost is measured as the ratio of wages and salaries to the value of shipment. Beatrice N. Vaccara, Employment and Output in Protected Manufacturing Industries (Washington, D. C.: The Brookings Institute, 19607} 31As shown in Table 6.11, p. 142, the most heavily protected classes, Ma and Nb, had average wages about 11 per cent below that of the "unprotected" classes. 1H3 TABLE 6.11.-—Average hourly wages of industries classified accordinCr to protective classes, 1954. Protective classes Average hourly wage rate (dollars) 1 1.81 2 1.81 3 1.86 Ma 1.76 1H) 1.61 All classes 1.68 Source: Beatrice N. Vaccara, Employment and Output in Protected Manufacturing Industries (Washington, h.C.: The Brookings Institute, 1960), pp.6l-62. ’) 3“Basevi objected to use of nominal tariffs and labor intensity, arguing that Vacarra's study failed to take into account the existence of intermediate inputs. See: Basevi, op. cit., 157-59. 33Balassa, Op. cit., 573-9A, L 3‘E—Balassa'smeasure of labor intensity is: The share of wages plus employer-financed social security payments in the value of output. Ibid., 581, Table 2. 351bid., 585. 36Ball, op. cit., 183—87. 37As Ball pointed out, the usefulness of the Balassa index of labor intensity depends crucially on the assump— tion that homogenous labor is employed at equal wages in all industries. Ibid., 18A. 38For the United States, the average wages for the low tariff group, i.e., industries with effective tariffs of less than 20 per cent, were higher than the national average and were about $1,600 higher than the average waxes for the high tariff group. Ibid., 185—85. 391bid., 186. in“ HM . , _ W1”UFC 6.“, p. 1N5, shows the relationship between iJJLLll yn~otxrctiman zinri lfflJOI’ eaiuiirrfs in Iirazxil. Al Lary, op. cit. N2 Ibid., pp. 20—29. 1 43mm: overall similarity of the rankings of industries according to Lary's indexes of skill and capital intensity is apparent from Table 6.12, p. 1A5 for the two countries. The few discrepancies that do exist present an interesting exception to what otherwise is quite a similar pattern. For instance, the rubber industry in the United States is equal to the average United States industry in terms of the wage and salary component of value added and is some- what below the United States average in terms of physical capital intensity. In Brazil the rubber industry is highly intensive in both human skill and physical capital. This may indicate that the production of tires and tubes, which figures prominently among rubber products, is less amenable for substitution of unskilled labor for capital. 0n the other hand, the non-metallic minerals indus- try-—stone, clay, cement and glass oroducts-—may leave more room for substituting unskilled labor for capital in an underdeveloped country like Brazil. In fact, I note that in Brazil this industry uses a relatively high proportion of unskilled labor, while in the United States it is close to the average both in skill content and capital intensity. Another industry group that warrants a comment is the electrical equipment industry. This industry is lower in skill intensity or human capital than the average United States manufacturing industry, but is somewhat above the average in physical capital. Tn irazil the electrical equipment industry is ranked as both skill and physical capital intensive. This may represent the adoption of relatively advanced foreign production techniques that result in a high proportion of capital or skill in value added, and leave small room for substitution of unskilled labor for capital or skilled labor. In spite of the cases discussed above, it is evident that the overall pattern found for the United States concerning value added per employee and its wage and non-wage components is observed for Brazil as well. Ibid., p. 21, Chart 1. 1 “(Harry G. Johnson, "The State of Theory in Relation to the Empirical Analysis," NBER Conference on Technology anl Competition, October, 1968, p. 9. (Mimeographed.) 209. 180. Total Protection (per cent) 65. 37. Source: Notes: 151. 123. 9D. .UO 30v 60+ 900 20+ Bot 80” '10.. 5.60 1115 .16 .20 .114 all l2 19 .15 18 16.10 26.60 37.10 “7.60 58.10 68.60 79.10 Production Workers Earnings/empl (1000's or.) Figure 6.u.--Total protection and production workers' average earnings in brazilian Manufacturing Industry by major industry groups, 1959. See Data Appendix. For names of industry groups see Figure 6.1. *Average for all manufacturing industries. TABLE 6.12.--Array of United States industries 1146 (1965) and Brazilian industries (1959) based on wage and non—wage value added per employee. Group I (low skill and low capital per employee) Group 111 (high skill and high capital per employee) Group 11 (low skill tut high physical capital per employee Group IV (high skill but low physical capital per employee) U.S. manufacturing industries 9% ' major industry groups lfib5* b. (23) Leather and leather products (31) Apparel & related products Lumber and wood products (2U) Textile products (22 Furniture and fixtures (25) Miscellaneous manufacture (30) Rubber and plastics (30) Petroleum and coal products (29) Chemical and allied products (28) Instruments & related products (38) Transportation equipment (37) Primary metal industry (33) (76) Paper and allied products Tobacco manufacturers (21) (20) Stone, clay and glass products (32) Food and kindred products Machinery (except electrical) (35) Printing and publishing (27) Fabricated metals (3A) Electrical machinery (36) Brazilian transformation industries firiicr irrhittriexi 1959 Clothing and related products Furs and skins Lumber and wood Textiles Furniture Non-metallic minerals Rubber Chemicals and pharmaceuticals Plastics Transportation equipment Metallurgicals Paper and paper products Perfumes Electrical and communications eq. Beverages Tobacco Food Mechanical industries Editorial and graphic Miscellaneous products Source: Ranking of U. S. manufacturing industries--Hal B. Lary, Imports of Manu- factures from Less Developed Countries, National Bureau of Economic Research (New York: and Table II. Columbia University Press, 1968), pp. 21-30, Chart I and Ranking of Brazilian transformation industries, see Chart I. Notes: * The numbers in parenthesis are the SIC code for each industry group. Rubber and plastics for U. S. are on the border of group I and IV. Miscellaneous products are on the border of group IV and I. 1&7 r u)Lary also notes other problems when using capital stock figures: "The uncertainty regarding the stock figures is compounded by the familiar vintage problem, i.e., the fact that available data on capital assets in— clude equipment and buildings acquired at various times past and at different price levels and written down ac- cording to depreciation practices varying among indus- tries and influenced by changing tax laws." Lary, op. cit., p. “1. 46 Unfortunately no data on capital assets were re- ported in the Brazilian Industrial Census for the larger sample of 58 industries. I attempted to use the installed power capacity (horse-power) per employee for these various industries as a possible measure of capital intensity. The outcome was that I found no significant association between installed power capacity per employee and non- wage value added per employee. The data were taken from: IBGE--Serviqo Nacional de Recesnseamento, Censo Industrial de 1960, Brazil, p. 6. u7Unfortunately the data needed to compute the skill indexes for my larger sample are not reported in the Brazilian Census; thus, I am unable to test these rela- tionships for the extended sample. u8Many students of economic underdevelopment claim that the most blatant scarcity of underdeveloped countries is in human capital, i.e., in terms of their skill defi- ciencies. Therefore we may see in the future reduced emphasis on merely the amounts of physical capital flowing into the underdeveloped countries. The following remark by Schultz is pertinent in this respect, "This one—sided effort is under way in spite of the fact that the knowledge and skills required to take on and use efficiently the superior techniques of pro- duction, the most valuable resource that we could make available to them, is in very short supply in these under- developed countries. Some growth of course can be had from increase in more conventional capital, but the rate of growth will be seriously limited. It is simply not possible to have the fruits of a modern agriculture and the abundance of modern industry without making large investments in human beings [my italicsE." T. W. Schultz, "Investment in Human Capital," American Economic Review, LI (March, 1961), 16. In recent years, in marked contrast with the 1950's, we can observe an added emphasis on technical assistance, training programs, health programs, educational loans and other contributions to human capital. 1&8 Z49Keesing found that one billion dollars of India's 1962 exports required in their production 72 per cent unskilled labor and only about .7 per cent of highly skilled labor. The United States, on the other hand, has the most skill—intensive exports, and also showed the greatest abundance of hard to acquire skills. [U.S. exports included 5 per cent skilled and “5 per cent un- skilled labor.] Keesing, "Labor Skills and Comparative . . ., op. cit., 25U—55, Table l. H CHAPTER VII INTER—INDUSTRY WAGE DIFFERENTIALS AND THEIR RELATION TO PROTECTION IN BRAZIL A. Introduction This part of the study inquires into the determi- nents of inter—industry wages and wage changes in Brazil's manufacturing sector. More specifically, it deals with changes in the industries' protection brought about by the 1957 Tariff Law, and the effect of these changes on the earnings of the industrial labor force. I will at- tempt to determine empirically whether the changes in protection resulting from the 1957 Tariff Law had the "normal" protective effect as postulated by Stolper and Samuelson. Specifically, I am investigating whether the new Tariff acted to raise the renumeration of Brazil's scarce factor of production--skilled industrial labor. Using the method of multivariate regression analy- sis, I will test for the effects of tariff and total pro- tection changes on the earnings of a cross section of manufacturing industries. Part B of this chapter gives the theoretical rational for the selected independent variables included in the regression equations. Part C 1U9 150 details the various forms of the regression runs. And finally, Part D includes a summary and an economic inter- pretation of the statistical results. B. The Determinants of Inter-Industry Wages: The Theoretical Rationale for Selected Variables Industrial wage structure is the end result of numerous factors, some of which are "market" determined while others are dependent on institutional considera- tions. Wages, as well as returns to other factors, ful- fill the double role of allocating resources and of af- fecting the income distribution. As a response to changes in market demand and supply functions, wage changes, in their allocative role, act to redistribute labor. This results in the bidding up of factor prices, including wages, in the expanding industries while factor payments will generally lag behind in the declining or stable industries. According to the neoclassical price theory, wages will be determined by the value of the marginal product of labor. The fundamental behavioral assumption under- lying the neoclassical model is that the firm's decision makers have a single goal motivating them, i.e., profit maximization. The model also implicitly assumes that the firm possesses perfect information about the nature of its revenue, production and cost functions. Under these assumptions the firm decides how much to produce 151 and the amounts of factors to purchase by equating the various factor price ratios to the relevant factor— product combination (i.e., the value of the marginal product). Thus the demand curves for factors are derived from the relevant production functions on the assumption of profit maximazation. However, the value of the average product (i.e., the value added per production worker man- year in this study) is systematically related to the value of the marginal product over the "usual" range of labor use.l Thus changes in the value of the average product of labor are influenced by both productivity and price changes.2 Several recent studies have shown that earnings of labor will vary with the degree of ccncentration and profitability of industries.3 These findings are in accord with what has been termed the "ability to pay" hypothesis, which suggests that the more profitable in- ‘dustries are more likely to pay higher wages. D. G. Brown, the major proponent of this hypothesis, argued that the more profitable industries will pay higher wages because "liberal" wage policies increase workers' good will, simplify recruitment, reduce costly turnovers and create good will toward the employer in the community.“ However, for the less profitable firms the immediate extra cost of such policies outweighs their possible 5 longer-run advantages. Therefore, he argues wage levels 152 will be generally lower than average in the industries where managers are pessimistic about their ability to earn adequate profits in the future. An alternative hypothesis used to "explain" inter- industry wage differential is the "competitive wage" hypothesis. This hypothesis allows for non-compensating short-run variations in wages that result from differ- ential changes in labor demand. These changes are caused by variations in labor productivity and product-demand among industries and also by the institutional framework within which wages are customarily set. Several writers found these variables to be significant in creating short- run wage differentials. I will now proceed to discuss the relationship between wages and several of the more important indepen- dent variables. 1. Wages and Productivity The relationship between productivity and firms' or industries' wages has been long the cornerstone of the neoclassical theory of the firm. Other things being equal, a rise in productivity permits a proportional wage increase without altering the profit position of an indus- try. Thus, industries experiencing relative gains in productivity may be under short-run pressure from their workers to raise wages. It was noted that the competitive model of wage determination acknowledges the dependence 153 of short-run variations in industry wages on the changes in the demand for labor resulting from changes in product demand as well as those caused by relative productivity changes. Yet most empirical studies found the relation- ship between productivity and wage changes to be rather tenuous.6 For the United States, however, Dunlop and Garbarino obtained a statistically significant rank cor- relation between productivity and wage changes over the 1920—u0 period.7 Reflecting on this contradictory evidence, Perlman argued that one would not expect a close relationship between productivity and wages if competitive conditions do not prevail in labor markets.8 Furthermore, Perlman and others noted that an increase in productivity does not always signal an equivalent increase in the industry's ability to pay higher wages. When a reduction in unit labor costs is caused by an increase in the intensity of capital use, increases in value productivity become a misleading indicator of an industry's "ability to pay" higher wages. This is so because of the role played by prices in the process of wage determination. In one case, wages will tend to increase when the product prices of an industry are rising, even though productivity has changed very little. In another case, wages will tend to decline when an industry's productivity has increased substantially, but product prices have declined by even ( larger proportion.) Thus, in this connection it becomes important to scrutinize the roleof profits in the pro- cess of wage determination.10 2. Profits and Wages On the basis of the "ability-to-pay" hypothesis of wage determination, one would expect differential profits among industries to be closely related to inter-industry wages and wage changes. To quote Brown: "More profitable firms are better able to pay higher wages, and the more profitable firms of today are more likely to be the more profitable firms of tomorrow."11 Studies for a number of industrial countries have related wage structure and the rate of change in wages to the profitability of industries. The OECD study found that for the United States manufacturing industries: "A marked positive association appears between profit- ability and changes in relative earnings through the entire period from 191l8."l2 Similarly a study researched for the Joint Economic Committee of the United States Congress found that within nineteen United States manu- facturing industries, the two most important factors re— lated to wage changes after 1951 were the levels of pro- fits and the degree of competition in the product market.13 On the basis of these studies it seems appropriate to include a "profit" variable in my model of inter- industry wages. Unfortunately, no direct data on 155 profitability of Brazilian industries are available. The closest I could come to approximating the industries' profitability was to obtain the non-wage component of value added for Brazil's manufacturing industries and deflate it by the value of output. Admittedly this variable includes, in addition to profits, also depre- ciation charges, property taxes, insurance, rent and other related items. In lieu of a "real" profit vari— able (e.g., the ratio of after-tax profits to industries assets) I used this proxy for the "real" profit rate. 3. Output and Employment Changes in output may be taken as an index of the strength of demand for the industries' product. It stands to reason that industries facing expanding markets for their products will acquire a greater "ability to pay" higher wages than industries facing stable or con- tracting markets. In the absence of a perfectly elastic supply curve for labor, an increase in demand for the industries' product and the resulting expansion of output would result in higher wages in the expanding industries. To some extent changes in employment among indus- tries reflect inter-industry differences in demand for labor. In the short run the effect of these changes is to raise wages in industries that experience greater growth of output relative to the "stagnant" industries. Yes, the empirical evidence obtained for the United States 156 dues not show a consistent and significant relationship between changes in relative earnings and changes in employment.lu In a study of inter—industry wages in Chile, Peter Gregory found a significant association between employ- ment and wage changes.15 However, in Brazil increased labor earnings in some industries may have been associ- ated with declining or stagnant levels of employment. This could occur because the industries experiencing the fastest rate of growth of output were those that absorbed unskilled labor at a slow pace compared to the "declining" industries.16 Thus, in my regression analysis, I will attempt to test whether any significant relationship exists between changes in employment and output and changes and inter-industry wages. A. The Ratio of Wage Cost to Value Added The ratio of wage cost to value added has been used in the literature as an index of the importance of labor cost to an industry. It has been argued that in- dustries, where labor costs constitute a relatively small part of total expenses, are better able to "afford" higher wages because their profit margins would not be affected in a significant way.17 The ratio of labor cost to total cost also has a bearing on the elasticity of demand for labor. Ceteris paribus; the smaller the 157 fraction of labor cost to total cost (or value added), the more inelastic will be industry's demand for labor, thus making it a prime target for unions' demands for 18 higher wages. 5. Other Relevant Variables A number of studies have considered the industry's degree of concentration as an important cause of above average wage increases. Firms belonging to a monopolistic or oligopolistic industry can usually pass the increased costs of higher wages to the consumer in the form of higher final product prices. In addition some degree of monopoly power helps a firm to guard against possible erosion of existing profits, as a result of the entry of new firms. Furthermore, concentrated industries are very often also subject to greater union coverage or influence, leading to aggressive union demands for higher wages.19 Most of the studies using United States data found a significant, positive association between the degree of industries' concentration and the relative advances in workers' renumeration.2O The concentration index for Brazilian manufacturing industries was provided to me in private correspondence by Dr. J. D. Langier. The Langier measure uses the proportion of total output provided by the 20 largest plants of each two-digit industry. Dr. Langier cautions that the concentration ratios probably provide a biased estimate of the "true" ratios, because of the multiplant operations of many firms.21 The theo- retically correct measure would be based on firms' out- put or sales rather than on plants data. Unfortunately, the Brazilian production statistics are not reported by f1. rms . Finally, several students of labor economics have advanced trade union organization as an important deter— ninant of inter—industry wages in industrial countries.22 in Appendix C several institutional aspects of the Bra- zilian labor market are discussed including the role of labor unions in wage determination. The general conclu- sion is that Brazilian unions were largely ineffective in obtaining higher wages and fringe benefits for their members because of their weak organizational structure, the limited scope for collective bargaining provided by law and the largely passive union membership. This eli- cited the following comment from w. B. Dale: ". . . the trade union (in Brazil) has been less of an influence than in other countries. In Brazil labor asserts itself more through the ballot box than through its unions."23 Therefore, a "union strength" variable was not included in my regression model. C. The Regression Model Based on the foregoing theoretical considerations, I constructed an econometric model of wage determination 159 for Brazilian manufacturing industries. Basically, two versions of the model are tested. The first formulation is based on the "ability to pay" hypothesis. The second version corresponds to the "competitive" hypothesis. The two versions use essentially the same independent vari- ables (i.e., a measure of concentration, capital-labor ratios, percentage change in productivity and the pro- tection variables), but the dependent variable is stated in terms of percentage changes in the "competitive" hypothesis. 1. The Dependent Variable The first decision involved in the construction of the model was to choose the most appropriate measure of "wages"--the dependent variable in the regression analysis. Since the Brazilian Industrial Census does not report wage rates the choice was limited to the use of annual earnings of labor. The Census provides information both on the annual earnings of all employees and of production workers alone. Earnings data were acceptable in this study in place of wage rates because they represent the actual renumeration received by the employees and also comprise the largest proportion of the industry's variable cost. The wage earning variable used in this study in— clude the "basic" wage and salary payments as well as cer- tain legal or customary fringe benefits. Thus, the "wage" variable includes on-the-job earnings plus fringe benefits I60 consisting of: end of year bonuses and gratuities, paid vacations and holidays, subsidized meals, transportation to andfrom work, work clothing, extra medical services, subsidized rental for employees' housing and others.214 The use of annual earnings provides the added advantage of ironing out irregularities that may be present in data reported on a weekly or monthly basis. The independent variables used in the regressions were discussed in Part B of this chapter. 2. The Sample Data and Problems of Estimation Using multiple regression and correlation analysis, I tested for the determinants of wages and wage changes across Brazilian industries. The relationships estimated were based on two samples. The first sample is composed of 18 two-digit manufacturing industries, and the second included 57 three-digit manufacturing industries. Both are based on the system of classifications employed by the Brazilian Industrial Census, 1960.25 In general, the equations estimated were either in linear or log-linear form.26 Initially both samples were tested for the deter- minants of the absolute levels of employees' earnings among industries. For the larger sample of three-digit manufacturing industries, I also tested for the deter— minants of percentage changes in labor earnings across industries. 161 Before stating the various forms of the regression runs, it is necessary to discuss one problem related to the estimation procedure. In principle, a model should specify as many relations as there are endogenous vari- ables. In my regression model a strong presumption exists that one of the explanatory variables--value added per employee man-year--is endogenous. To obtain consistent estimates of the regression coefficients, I have Chosen to employ lagged variables. To this end one must find a lagged variable that is uncorrelated with the regression dis- turbance but is correlated with the "explanatory" endoge- nous variable.27 In my study the lagged value of the pro- ductivity variable was used. A similar approach was adopted in two recent studies that analyzed the behavior of wages in United States manufacturing industries.28 In line with the above discussion the first rela— tionship tested can be expressed in the following linear equation: (I) w = a + blP-l + b2C + b3Z + e where: w = average annual earnings of employees in 18 two-digit manufacturing industries (1959) P = value added per employee man-year, lagged one year (1958) C = index of concentration for each industry (per- centage) (1959) 162 Z = the protection variables; alternatively de- fined as: T = average tariff rate (percentage) TP = average tariff rate plus exchange pre- miums (percentage) 6 = the error term For each regression equation the following statistics were computed: Coefficient of multiple correlation (R), coefficient of determination adjusted for degress of freedom (B2), and the F-statistic. In the next set of regressions the productivity variable was replaced by two measures of the industries' capital—labor intensity, alternating the gross "physical" capital per man agg horse power per employee man-year. A "large" capital—labor ratio may imply that labor costs are a small proportion of total costs, thereby contri- buting to an inelastic demand for labor (see previous discussion on this point). This ratio may be also taken as an indicator of the strength of monopoly forces in the product market; i.e., high capital-labor ratio will often indicate effective barriers to entry of new firms into the industry. “I: *+ ‘+ (2) n a + blK b2C + b3? 6 The added variables not defined with respect to equation (1) are: 163 K* defined alternatively as: K1= "gross" value of capital per employee man— year or K2 = horse power per employee man—year in a few of the regression runs the prOportion of wage cost in value added has been also included (the notation used is WVA). ' Next I employed the extended sample of 57 three- digit manufacturing industries for 1959. With the larger sample a few additional independent variables were intro- duced. In the first place, the ratio of non-wage value added to industries' output was used as a measure of in- dustries' profitability, in lieu of direct data on in- dustries' profit rates. Secondly, more extensive use was made of the protection variable. In the extended sample, I introduced the simple averages as well as the weighted averages of industries' tariff rates. Similarly, the total protection variable was based on the weighted and simple average tariffs, of the 57 industries' pro— ducts. Thus, typically the regression eouation estimated was of the following form: (3) w = a + blP-l + b2WVA + b3Z + e where: w = average annual earnings of production workers in 57 three-digit manufacturing industries (1959) 164 P_1 = value added per production worker man-year, lagged one year (1958) WVA wage cost divided by value added of each industry (1959) Z = the "protection" variable that includes: T = average unweighted tariff, levied on the industry's product T = average weighted tariff levied on the industry's product TP total protection: average unweighted tariff plus exchange premiums TPW = total protection: average weighted tariff plus exchange premiums e = error term In equation (U) the lagged productivity variable was re— placed by a proxy for the capital-labor ratio, i.e., the horse power per employee man-year. = ' 1A (u) w a + blK + b2JVA + b3z + e where all variables were defined with respect to equation (3) with the exception of: K = horse power per production worker man-year, 57 industries (1959) In addition, based on Lary's conceptual framework, I attempted to substitute the non—wage value added per em- ployee man—year for the industries' "stock" of capital (Notation: N_1). Lastly, in this part of the study percentage changes in inter—industry earnings (the dependent variable) were 165 substituted for the absolute levels of these variables used heretofore. On strictly theoretical grounds there is no way to determine the "correct" functional form of these regression equations, i.e., whether the abSolute values or the relative changes in the dependent variable should be employed. The rate of change in productivity as well as the previously discussed protection variables were included among the independent variables in this part of the analysis. I also tested separately for the effect of output and employment growth (percentage change from 1958 to 1959) on labor earnings. The general equation estimated was in the form: (5) W = a + b B +_b l K + b Z + e 2 3 where: w = percentage change in annual earnings of em- ployees in 57 manufacturing industries from 1958 to 1959 ”'30 ll percentage change in productivity (i.e., in value added per employee man-year from 1958 to 1959) ' K = horsepower per production worker man-year (proxy for the capital-labor ratio) (1959) Z = the protection variable; as defined previously with respect to equation (3) D. Summary of Statistical Results and Their Economic Interpretation The results of the regression runs are summarized in Tables 7.1A through 7.8. In spite of the imperfections inherent in this type of study and the inaccuracies in the 166 Brazilian data, the overall statistical results were satisfactory. Tables 7.1A, 7.18 and 7.2A, 7.2B summarize the re- gression results for the sample of 18 "major" manufac- turing industries. First, the computed regression co- efficients for the "protection" variables were negative and statistically significant at the customary level in all of the equations tested. However, the "other" inde- pendent variables--the lagged productivity variable, the ratio of wages to value added, and the alternative formu- lations of the capital-labor ratio-~did not perform as well as expected. In fact, the only non-protection inde- pendent variable that was statistically significant was the index of industries' concentration. Thus one can con- clude that the higher the degree of monopoly in the prod- uct market the higher the industry's wages. The fact that the concentrated industries are often the most ef— fectively unionized is one possible reason for the good showing of this variable. However, the explanation of the other disappointing results probably lies in the ag- gregative nature of the "major" industries' data that conceal important differences within each industry group. This defect has been to a considerable degree rectified by running the larger sample of 57 three-digit SIC industries. In the larger sample the "protection" variable was again negative and statistically significant 1.67 .Ho>oH ucmo poo so» on» no unmeamficmfim zaamOfiumHumpm n + .Hm>ma acme poo o>am map um pcmoHMficwfim haddofipwfiumpm u sea .Hm>oa pcoo poo 03» on» as pesoHMficwfim zaaoofipmfiuwpm n ms .Hm>oa pcmo poo mco one no unonMchHm zaaoofipmfiumpm u * .zum oaooe .mso .o .Amcom one soafiz snow "xmow 3ozv mambo: ofipumeocoom .umfipgo .m Hpmo "mom .mocmo Ifimficmfim mo Hm>oa pcoo poo mco on» no .mpmfixo cofiumaoppoOOpso o>fipfimoo ho mocmofi>o o: .os.H coco mmpmmpw ma oauwfiomum 30 bopmazoamo on» pH .mamsofimop on» :H cofiomaonLOOOpsm unonMfic uwfim mo mocmumfixm on» no ammo m on com: ma oaumfipmpm compmz Cabana one u 3Q .usmHOHmmooo coaumaoppoo oaofioazs u m .Eoooopm mo woopmmo pom oopmSmom one u m .oaooeooom no N on» ma mumxompn mpmsvm on» Ca madman one .pcoaofimmooo :ofimmopwos one no posse opmbcwum one ma mfimonuconmo pesos a“ madman one .psofiofimmmoo cofimmopwoh amassed one ma madman now on» oHanpm> pamUCoooocH zoom pom ”moooz .AQ xfiocoao3 o m e ambasz mpomHLm> osmocmoobcH soapoZUm .mmma .mowsumzocfi LOMmE maulmmcHCLom Lonma appmsocfilsoch mo mpcmcfifimopop .mpadmon cofimmoamop coauoom mmomoln.¢a.s mqm3 0 a o .m hmDESZ moaomfipm> pamoCmmoocH cofipmsmm J .) .mmma .mpappmdbcfi Lemme maanmmcficpmo pooma :Lumspcfilpoucfl mo mpcmcHEmeob .mpHSMoL cosmmopmos coapoom moosoll.ma.m mqmge 169 . manme mom wouoz mom "mopoz .AQ xfipcoao3 0 me -Hx .HOQMWH\H.®PHQMD .m g mmfibmfism> ocoocomoocH coaomSmm gonna zspmsocfinsopcfi mo mocmcfieaooob .mpadmop :on .mmma .mofisumsocfi mcHLSpommscme homes ma . ucficsmo mo mb smog Soapomm mmosonl.3 0 mm .H& Loom4\ampfiomo m pooESZ mmaomfipm> pcmocomoocH soapmSmm momma whom: .mmaa .mofitemooCa weatooooooeos tones ma .mwcficsoo :Humopcfi m0 mpcmcfifimoumm .mpHSmmL cofimmohmms cofipoom mmonunu.mm.s mnmoH onoo goo no» on» be ucoofiuwnmfim zfiaoofiomfioMum n + .Ho>oH ocoo non o>Hu one om pcoofiuficmfim zaaoofipmfipoom u *** .Ho>oH ocoo nod 03» on» no unoOHMficme zaaoofiumfipopm u as .Ho>oH onoo Loo oco onp ow pnoofimfinwfim maaoOHumeoum u * .oocoofiefinmfim no Ho>oa ocoo Loo one on» no .mumaxo coauoHoLLOQ nooso o>Honoo do oocoofi>o mm .oo:.a can» noooonm we oapmfiuopm 30 oouofisoaoo ono MH .maozofimop on» CH coaooaospooOpso unmoflofinwfio no oonooofixo no poop o no pom: mH one owomfiooom nomoozlnfinpso onp ma 3a .Eoooono mo moosmoo Lou oopmSnoo one mm one .oflomfipoomno on» mH mooxoopn oposow CH opswfim one .onofloflumooo noflmmonmon on» no nonno onobnoom on» ow mooxomnn pesos CH opswflm one .onoflofimuooo cofiooonmon Hoflonoo onp ma onsafih do» one oflnofipo> unopnoaoonfi nooo pom “woooz .Aa xflbcooo coapoooopm I mnoflp H m o Io>pomno Lonasz moanofipo> ucoocomoUnH mo Lonesz nofipmsvm J .mmmfi .moflppwzonfi wnHLZpomusnoE ufimfioloopnu em .mdnfinpoo Loan znumsonwlsopnfi mo monocHELoooo .ooHSmop :onmoLwop cofioooo mooaoll.m.s mnm .p u L <>3 H m mCOHp o |m>pomno gonesz moaomwnm> ucoocoooch no Lopezz Coapozwm .dmma .moflnomzonfi mnflnzooomsnoe ofimfibloopnu em .mmcficpoo sonoa znomsonflnnoonfi no monocfisnopoo .opazmop nofimmopwop coauoom moonolu.z.s mnm cofipeouonm <>3 x mnoau o |o>poono Lonezz moanmfino> unobnomobcH mo Lonesz soaposwm onwaouootno em .mmnwcnmo .mmma .mofipumsonfi mCHLSoooosnmE nonma apposenfiunopnfi mo monocfiegoooo .mpHSmop nonmosmoL :ofipoom omoponl.m.s mnmunoe Loans: nofioozpono non woven ozao> ommzncoz u H.2 .m.s oases too ooooz oom "moooz .AQ xaocooo cofioooooto a- a2: an ozone .m lm>hmmDO LQDESZ moflnmfino> unopnomoocH no ponEsz nofiuozvm .mmma .mofinomsonfi mcfinzoomescoe ufimfionoosnp mm .mmcficnoo ponoa apomSUCHInoonfi no mucmnHELooob .muazmop :onmonmoL nofiuoom mmopolu.w.u mnm<9 176 .m.s oHnoE Lou mouoz oom ”mouoz .AQ xfiocooo . Heno.nu3o so.mmunnm.mva fimsno.na ”mooH.oa flmso0.sa oom.umm Amuse.v Asomn.mmv nommo.v mws.n m cows. *mamm.©0H *owma. w:m.wal mm H.> sue me 3% B ANV ofioonto> conooooone HI mcofip x <>3 m uo>pom LonEsz . o :90 mo coHo mofinwflno> pnoocoooch Lopezz losam J .mmma .moHLomsonH mCHLSuoomsnoE pHmHosoosno mm .macficnoo nooofi znpmsocfilnoonfi mo monoCHEnouob .mpHSmop nofimmosmop cofipoom mmopoll.~.~ mnmoa ocoo too coo on» so ocooaoecmao sanoofiooaooom .Ho>oH onoo Loo o>Ho on» no pnoofiufinwfim kHHoOHumfipmom u or .Ho>oH onoo Loo oco onu no unmeamfinmfiw maaoofiomfiuopm n * .oocoofioficmfim no Ho>oH pcoo Loo ono pm pcomopo ma nofipoHoLLOoOpso no oonoofi>o oc xsozmm one zom:.H coozuon maaom ofipop ccoesoz no> ooooHSoHoo on» 9H .maozofimop on» Ca cofiooHoLLOOOpso unmeawficmflm no oocoomfixo ono no poop o no com: me nofinz .cofioop namesoz no> on» ma z> .Eoooopm no moonwoo Lou ooumSnoo ma mm .monoHOHpmooo cofimmonmon on» go nonno bsmbnoom on» ma mHmononopoo CH opsmHm onp monoflofimmooo cofimmonmon Hofionoo on» ma onsvflw goo onp ofinmflnw> pcoonoooonfi nooo Lon .m.s oanoe Loo mooo: one ooo moflnofinm> cofiooooono on» so oceaoficfiaoo pom "moooz .AQ xfionooo onm.ume Amoe.mv Ammm.mmv Aamm.sv noms.ommv emu." q **:oo.w| *om:.mmm +mos.mn mam.oom m.m H.\l_\t\..HHT> . . hnm.umo Amso.mv Amos.aov Anso.sv Asea.oasv 0mm." x **Cmo.sl *Omm.mwm 0:3.Nl wmm.mmma :.® mmmm.au:: , . omn.umm Awmm.ov nomm.oonv Ammo.sv Amne.mmmv mm:.u m :os.aflu *mzc.amm moz.mn oom.Hom m.w omm.anz> men.umm “noo.ov Aoom.noav Asmo.mv Asmo.mmoa ,mH:.u m +moo.cal *szm.zwm mom.sl wom.mwm m.w mmmm.fiu:> smn.umm Amon.moflv Ammo.mv Ames.:osv Gem." m *Hms.oom +m::.wn www.mmau H.m 3am mm 39 E x m mofiooflpo> nofluoooonm . m LonE32 mofinofino> onoonoaoonH coaposmm .mmlmmma .wofinumzocfi mm ho meHCLoo Lonoa CH mowcono oaoonoopoo no mcoflmwonwos coapoom mmopoll.m.s mqm<9 178 substitute I used for want of an industry's profit rate-- the ratio of non-wage value added in gross output—- showed no significant relationship to earnings changes. The final sets of regressions (summarized in Table 7.8) uses the percentage change in the wage and produc— tivity variables. It therefore corresponds to the "com- petitive" wage determination hypothesis. The results were considerably less satisfactory than the previous set of runs. Yet once again the protection variable had a negative sign and was significant in three out of four cases. By running a regression of the percentage change in labor earnings against employment or output changes, I found these variables not to be significantly corre- lated.30 The most significant conclusion from my empirical work is that tariffs and total protection had an impor- tant effect on the earnings of industrial labor in Brazil. It also appears that the negative ("perverse") effect of protection on inter-industry wages is the same regard— less of whether unweighted or weighted tariffs were used in the regressions. The numerical "size" of the regres- sion coefficients is small; thus a "moderate" increase in tariffs or total protection will have a relatively small impact on labor earnings. Yet the direction is unmistakable in all of my re- gression runs. My results show that the Brazilian Tariff 179 did in fact harm Brazil's industrial labor force. These findings are in accord with what would be expected from the Stolper-Samuelson theorem. Namely, the country's scarcest factor would indeed benefit if the tariff was structured to effectively protect the scarce factor, or rather the industries using the scarce factor relatively intensively. Thus a country that has a shortage of skilled labor could increase the earnings of that factor, but only if the tariff actually protected the scarce factor. In that sense the "orthodox" results have been confirmed in this study. There is no presumption that the Metzler case indeed applies to the Brazilian situation. This study is not able to determine the effect of profits on wage rate changes. This indicates the need for better profit information on Brazilian industries. If the omitted variable, i.e., profit rate, is related to some of the independent variables included in the regres— sions—-such as the capital-labor ratio--a specification bias may have been introduced into our estimates of the regression coefficients. Namely, one would be attributing to the variables that have been included, some of the influence due to the omitted variable, the profit rate.31 Thus, if I were able to introduce an adequate profit variable, my estimates would have gained in precision and reliability. 180 As it stands, the foregoing analysis has enabled me to build a model of wage determination for Brazilian manu- facturing industries. I have established the direction and approximate magnitude of the protection variables' effect on the earnings of industrial labor. If it had been possible to obtain a satisfactory profit variable, I believe that the model's explanatory power would have been enhanced considerably. FOOTNOTES: CHAPTER VII 1This results from the Operation of the law of diminishing returns and the assumption of profit maxi- mization. In the so-called stage-II where the cost minimizing firm is likely to operate, both the average and the marginal product of labor will be falling. If the relevant production is of the Cobb-Douglas type-- with neutral technological change, the average and mar- ginal product will differ only by a multiplicative con- stant. For detailed discussion see for example: R. H. Leftwich, The Price System and Resource Allocation (Nth ed.; Hinsdale, Ill.: The Dryden Press, 1970), pp. 117-123. 2In my study I calculated only the value produc- tivity measure for each industry. In addition to the conceptual difficulties noted in employing a measure of physical productivity, in the case of Brazil, price de— flators by industry are not available. Certainly it does not make any economic sense to use a single de- flator, like the wholesale price index for all manufac- turing industries, when we are studying the interindustry structure of wages. 3Excellent review of these findings to which refer- ences will be made throughout this chapter is found in: OECD, Wages and Labour Mobility_(Paris: July, 1965). “David G. Brown, "Expected Ability to Pay and Inter- industry Wage Structure in Manufacturing," Industrial and Labor Relations Review, XVI (October, 1962), US. 5In this connection Brown remarked: "In the less profitable firms teetering at the brink of extinction, managers are more concerned with preserving the firm than with easing their jobs and gaining community prestige; employees are more concerned about unemployment than under-payment." Ibid., A6. 181 182 6"l'he OECD study states: "The data examined by us indicate no significant tendency for above average pro- ductivity gains to be associated with more rapid earnings increase, although due weight must be given to the pos— sibility that the apparent absence of any relationship may be accounted for by the very great difficulties of measuring changes in productivity adequately." OECD, op. cit., pp. 110-12. Several other studies based on United States manu- facturing data arrived essentially at the same conclusion. . The following studies found no significant relation- ship between wage and productivity changes: Harold M. Levinson,"Postwar Movement of Prices and Wages in Manufacturing Industries," Study Paper No. 21, Joint Economic Committee, Congress of the United States, January 30, 1969, also: D. M. Eisemann, "Interindustry Wage Changes 1939- M7," Review of Economics and Statistics, XXXVIII (November, 1956), also: D. G. Brown, op. cit., 51-52, Table 1., also: F. Myers and R. L. Bowlby, "The Interindustry Wage Structure and Productivity," Industrial and Labor Relations Review, VII (October, 1953), also: Donald R. Snodgras, "Wage Changes in 2H Manufac— turing Industries, 19U8-59; A Comparative Analysis," Yale Economic Essays, III (Spring, 1963), 171-222. 7Dunlop suggested that the highest productivity gains are to be expected early in an industry's life when output is expanding rapidly. He attributes the above average wage increases not only to the industry's "ability to pay," but also to its need to attract additional work force. John Dunlop, "Productivity and the Wage Structure," Income, Employment and Public Poligy: Essays in Honor of Alvin H. Hansen (New York: W. W. Norton, 19387, pp. 351—62. Also see: Joseph W. Garbarino, "A Theory of Inter- industry Wage Structure Variation," Quarterly Journal of Economics, LXIV (May, 1950), 282-305. 8Richard Perlman, "Value Productivity and the Inter- industry Wage Structure," Industrial and Labor Relations Review, X (October, 1956), 26-39. 9The relationship between wage, price and physical productivity change has been examined by Clark Kerr. Kerr found several instances during the recent decades when wages, under the influences of price changes, seemed to move independently of productivity changes. It is con- ceivable that productivity increases in one industry may 183 lead to relative price declines rather than to higher wages; while another industry, experiencing favorable demand for its product and consequently, increases in prices and profits, may raise wages without a comparable productivity change. Clark Kerr, "The Short Run Behavior of Physical Productivity and Average Hourly Earnings," Review of Economics and Statistics, XXXI (November, 19U9), 299-309. Perlman found Rank correlation coefficient of .68 and .58 between sales per man—hour and average hourly earnings in 20 United States manufacturing industries for the periods 1939-“? and 1947-53 respectively. See: Perlman, op. cit., 35. 10The neoclassical theory of the firm tells us that a firm will use the combination of resources that equates the marginal physical product per dollar's worth of each and every variable factor used. To maximize profits the firm will produce the level of output that will equate its marginal cost to the product price (or marginal revenue). See: Leftwich, 0p. cit., pp. 282—85. llBrown, op. cit, A7. l20mm, op. cit., p. 105. 13Levinson, op. cit., pp. 3-6, Tables 1, 2 and 3. In another study Bowen found that for six subperiods be- tween January, 19u7 and June, 1959 profits showed a more consistent relationship with wage changes than employment, concentration or labor market organization. For details see: W. G. Bowen, "Interindustry Variations in the Un- employment Wage Relationship," Wage Behavior in the Post- war Period: An Empirical Analysis (Princeton, N. J. Princeton University, 1960). 1“See for example: Melvin Reder, "Wage Differen- tials: Theory and Measurement," Aspects of Labor Eco- nomics, National Bureau for Economic Research Conference, Princeton, N. J., 1960 (Princeton, N. J.: Princeton University Press, 1962), p. 278. 15Gregory did find a positive association between productivity and wage changes. See: Peter Gregory, Industrial Wages in Chile (Ithaca, N. Y.: Cornell Uni- versity, 1967), Chapter V. 16 I am referring here to the evidence that was com— piled by Baer and Kerstenetzky with respect to the dis- tribution of gross value added and employment of labor in Brazil's industry. It was observed that between 1950 18A and 1960 the share in value added of the "traditional" industries, i.e., textile, food products, clothing and wool, has declined. On the other hand, the share of the .industries that were most affected by import substitu— tion, e.g., transport equipment, electrical machinery, chemicals and metal products, have risen impressively. While the "traditional" industries were characterized by less than proportional fall in employment compared to gross value added; the "dynamic" industries saw a greater expansion of gross value added relative to changes in employment. ' For further details see: W. Baer and I. Kerstenetzky, "Import Substitution and Industrialization in Brazil," American Economic Assocation Papers and Proceedings, LIV (May, 196A), D18, Table A. ~17Presumably this assumes that the firms are unable to pass on the increased labor cost to the consumer in the form of higher prices.. See: Brown, op. cit., H8. 18This argument was originally advanced by Marshall and later adopted by Friedman. See: Milton Friedman, Price Theory, A Provisional Test (Chicago: Aldine Publishing Company, 1962), pp. 155-56. 19See: Martin Segal, "The Relation between Union Wage Impact and Market Structure," Quarterly Journal of Economics, LXXVIII (February, 196U), 96-llu. 20Garbarino found the Rank correlation coefficient between the degree of concentration and changes in earning to be .67 and rising to .75 with the elimination of the coke industry. See: Garbarino, op. cit., 299-302. D. G. Brown found the partial correlation coeffi- cient between wages and concentration to be .326. See: Brown, op. cit., 53—5”. However, conflicting results were obtained when the relationship between concentration and wages was tested for other industrial countries. In the United Kingdom the relationship is of limited practical significance, while for Germany an opposite tendency has been observed, i.e., the less concentrated industries experienced larger earning gains. See: OECD, op. cit., p. 1l5. 81The calculations by Langier were based on Bain's method, outlined in: Joe S. Bain, International Com- parison of Industrial Structure (New Haven, Conn.: Yale University Press, 1966), pp. 27—29. 185 “unrhnrino considered increased profits and concen- tration of an industry as providing permissive conditions for wage increases. The oligopolistic market structure may protect the cost decreases and profit increases from the inroads of competition, making them potentially available to wage earners. The compulsion may be in the form of aggressive trade unions which exert upward pres- sure on wages in the light of "excessive" profits or visible increases in labor productivity. Peter Gregory acknowledged the possibility of unions having a consider— able effect on the earnings level of industrial labor force in Chile. He noted that a monopolistic market structure, accompanied by other conditions that give rise to inelastic demand for labor, will enhance the bargaining power of labor unions. This could lead to higher wages in the affected industries. See: Garbarino, op. cit., 299—302. For an analytical treatment of this topic also see: Segal, op. cit., 96-lOA, also see: Gregory, op. cit., p . ()9 . . 23Dale, op. cit., p. 60. The role of trade unions may have been on the ascent during the Goulard era of the early 1960's. During his administration, trade unions received generous government support and some en- couragement in pursuit of higher wages. See: Gregory, op. cit. 1 21Details on the non-wage component of labor earn- ings are found in: Dale, op. cit., pp. 62-66. I. 23Data on the characteristics of industrial activity and labor earnings were obtained from the 1960 Industrial Census that covers the year 1959. See: IBGE, Servigo Nacional de Recenseamento, Censo Industrial de 1960, VII Recenseamento Geral do Brasil. The corresponding data for 1958 were obtained from: IBGE, Conselho Nacional de Estatistica, Produpao Industrial Brasiliera, 1958, Rio de Janeiro, December, 1960. ’3 “6There is no a priori way to determine (based on economic theory) the correct functional form, i.e., linear or logarithmic. Use of the arithmetic form pre- sumes that there is a relation between the absolute dif- ferences in the dependent and independent variables, from industry to industry; while use of the logarithmic form assumes that there is a relation between the per- centage differences from industry to industry in the relevant variables. 186 ,, “(For a theoretical discussion of the use of lagged variables consult: J. Johnston, Econometric Methods (New York: McGraw-Hill, 1963), pp. 165-66. See also: J. D. Sargan, "The Estimation of Eco— nomic Relationships Using Instrumental Variables," Econometrica, XXVI (July, 1958), 393-“15. 28For details consult: E. Kuh, "A Productivity Theory of Wage Levels--An Alternative to the Phillips Curve," The Review of Economic Studies, XXXIV (October, ]()67)a 333‘3600 Also see: R. D. Rippe, "Wages, Prices, and Im- ports in the American Steel Industry," Review of Eco- ngmips and Statistics, LII (February, 1970), especially 3 -3 - 2Q The regression of wages on the "profit" variable gave the following results: w = 63.357 + 25.197? (1u.5u7) (A0.37u) .0831 .0069 l.UQU6 R R2 VN Where: W average annual earnings of production workers in 58 industries. "profit" variable, i.e., non-wage value added divided by industries' gross output. ”:1 II 30, , . The regres31ons of percentage changes of wages on output and employment changes (1958-59): (a) 0 = 32u.589 - .0050 (300.258) (1.506) R = .0000 32: .0000 (U) 0 = 21u.157 + 1.2u85 (317-893) (1.3U5) R2: .1275 a = .0163 WHt'T'L“: W percentage change in annual earnings of all employees in 58 manufacturing industries. percentage change of gross output 1958—59. percentage change in total employment 1958-59. A U __._. o z 0. 54'.‘ 187 ’lwor a full discussion of the effect of a specifi— cation error on least square estimates see: E. Malinvaud, Statistical Methods of Econometrics, trans. by A. Silvey (Chicago: Rand—McNally, 1966), pp. 263-66. CHAPTER VIII SUMMARY AND CONCLUSIONS The first part of this study examined the structure of Brazil's foreign trade with respect to the availability of capital and skilled labor, the country's most scarce re- sources. It has been established that Brazil is well supplied in "nature," a combination of natural resources (such as soil, climate, forestry and deposits of certain ores) and unskilled labor. "Nature" and unskilled labor complement each other in the production of coffee, cocoa, cotton, sugar, tobacco and iron ore and give Brazil her "revealed" comparative advantage in the exports of these commodities. Brazil suffers from an acute shortage of skilled labor at all levels, but most seriously at the highest level, i.e., engineers, technicians, managers and com- petent administrators. Thus, the country imports skill intensive commodities from North America and Western Europe to alleviate her skill shortage. This "explains" why over 80 per cent of Brazil's imports of manufactured goods (SITC 5, 6, 7 and 8) come from the skill abundant coantries; thus the "neofactor" proportions hypothesis, 188 189 based on relative availability of human capital between Brazil and her trading partners, has been substantiated in this study. "Physical" capital did not perform as well as human capital availability in explaining Brazil's trade pat- terns. Furthermore, using Lary's single measure of in- dustries' "overall" factor intensity, I attempted to test whether a combined index of "physical" and human capital could account for Brazil's trade patterns. The results of these tests were inconclusive and it is conceivable that a different measure of "overall" factor intensity may provide a more satisfactory test of this hypothesis. Brazil's exports of manufactures presented a special problem because when examined in relation to the fore- mentioned versions of the factor proportions theorem, they did not show any clear-cut pattern. Based on a priori reasoning one would expect Brazil to export con- siderable amounts of simple, unskilled labor intensive commodities. Yet in 1965, Brazil exported only about $120 million worth of SITC 5, 6, 7 and 8 goods; of which only about A0 per cent were destined to industrial coun- tries. For instance, Brazil's share of less developed countries' exports of manufactured goods into developed co1ntries was less than ppg per cent (.6 per cent to be exact).1 This is an undistinguished record for a country thit built a substantial industrial base in the period 190 after World War I and whose industrialization experienced a "take off" in the post World War II period.2 Thus, it was important to investigate why Brazil's exports of manufactures did not reflect her factor avail- abilities. Largely through import substitution, manu- factures exports came artificially to complement her do— mestic industrialization program. Therefore, the irra- tionalities and inefficiencies imparted to Brazil's pro— duction structure by past import substitution policies inhibited the exports of her manufactured goods. The most blatant example of this is the Brazilian textile industry. One of the earlist industries in Latin America, it is still not making a significant contribution to Hrazil's export earnings (for details see Appendix A). It is widely recognized that Latin American coun- tries "need” to move from an import substitution-based growth into an exports-oriented growth based on manufac— tures rather than primary commodities. My analysis 1Ndi‘ cates that there is little likelihood for Brazil becoming an important exporter of even simple manufactured goods. To bring the study up to date, I would like to report on the recent policies initiated by Brazil that may have important long—run implications for her exports and imports. I am referring to two measures of import liberalization carried out by the Brazilian government between 1965 and 1967. From November, 1966 to March, 1967 191 two major steps were undertaken. One was a thorough re— vision and general reduction in the level of tariffs, and the seCond was a complete elimination of the remaining exchange premiums. Furthermore, the "basic" exchange rate has been devalued several times between 1964 and 1969 to "keep up" with the rapid rate of domestic price increase. This move toward import liberalization constitutes a significant departure from past Brazilian trade policies. In the past the policy has been to provide whatever degree of protection necessary for uninterrupted import substi- tution. The new policy should provide increaSed compe— tition from imports and thus should encourage improved efficiency and lower cost of production in Brazilian manu- facturing industries. This combined with a "continuous" devaluation of the cruzeiro may provide Brazilian firms with a profitable opportunity to expand exports rather than continuing, as in the past, to treat export sales as incidental compared to their domestic business.3 The second part of the study dealt with the impact of Brazil's structure of protection on the earnings of her industrial labor force. It drew on some character- istics of Brazil's foreign trade established in the first part. The point of departure for the second part of the study was the previously established proposition that human capital is Brazil's most scarce factor of produc- tion. Thus, for'the scarce factor to benefit from a 10? change in tariff, those industries using human capital intensively should receive the highest protection; be- cause then the tariff (assuming that there is no change in the country's terms of trade) will tend to raise in- dustrial labor's wages both absolutely and relatively. On the basis of the analysis contained in Appendix A, I concluded that Brazil probably faces an elastic demand for her exports. Therefore one would expect that a cor- rectly structured tariff would raise the earnings of the country's industrial labor force. This proposition was subjected to an empirical test in the last chapter by constructing a wage determination model. The major con- clusion derived from the model was that in Brazil's case the tariff and total protection had a negative effect on the earnings of skilled industrial labor. As was stated in the conclusion to Chapter VII, these results are in agreement with the Stolper-Samuelson theorem and its major implication that import duties pap raise the wage of the scarce factor. But in order for the Brazilian Tariff to have the "normal" protective effect it will have to be restructured so as to accord high protection to those industries that use the country's scarce factor relatively intensively. FOOTNOTES: “CHAPTER VIII lThe statistics for Brazil's exports are included in Data Appendix. For the data on less developed coun- tries' trade see: Lary, op. cit., p. 96, Table 10. T.) “For details consult: Baer, op. cit., Chapters 2 and 6. 3Further details can be found in: Paul G. Clark, "Brazilian Import Liberalization," Research Memorandum Ho. 1“, Center for Development Economics, Williams College, Williamstown, Mass., September, 1967. (Mimeo- wraphed.) 193 APPENDIX A A DETAILED ANALYSIS OF BRAZIL'S EXPORTS AND IMPORTS A. Brazil's Commodity Exports Currently as in the past a relatively small number of primary commodities dominate Brazil's exports. Coffee, cocoa, cotton, sugar, tobacco and more recently iron ore are central to the country's earnings from exports. The growth rate of world trade in many of these commod- ities was highly unfavorable, thus seriously limiting Brazil's export-based capacity to import.1 A more de- tailed analysis of Brazil's exports follows. 1. Coffee Exports Brazil has traditionally been an important exporter of a few primary commodities. For decades it has been the world's largest coffee producer and exporter. During the 1020's coffee usually represented 60 to 70 per cent of the country's exports. In the great depression, world demand for coffee declined markedly and coffee prices feLl sharply. After the Korean War coffee prices rose distinctly, reaching their peak in 195“. But since the 194 105 Inid-I‘VM)H: cukau>;n'ices wcrw\(knylining cnmxérnore, 1%}- bounding to some derree invl96U-65.2 .In the postwar period coffee became increasingly important in the eXports of certain African countries. These countries increased their share of the world market in coffee from 7 to 8 per cent in the 1930's to about 25 3 per cent in the early 1960's. Production of and trade in the Robusta type of coffee, mainly from African sources, more than doubled between the early 1950's and the early 1960's.” This growth took place mainly at the expense of Brazil, whose share of world coffee exports fell from about 53 per cent in the 1934—38 period to about 38 per cent during the 1962-63 period.5 Despite these developments coffee exports are still the single most important source of the country's foreign exchange earnings. During.the l96A-65 period the value of Brazil's coffee exports was about A8 per cent of all commodity exports, by far the largest item in its export bill. Since about 87 per cent of Brazil's coffee exports during this period went to industrial countries the future prospects of coffee exports depend largely on the growth of coffee consumption in North America and Western Europe. Rex F. Daly estimated for the United States the effects of changes in incomes and prices on consump— ticn of coffee, for the period 1922—Al. Daly's empirical results show a low income elasticity of demand for coffee 196 (.25) and a price elasticity of demand of only (-.26), over the period of investigation.6 Daly found the price elasticity of demand for coffee to be stable over time. When estimated for the postwar 19U7-57 years it was about (--25)- Balassa assumed the income elasticity of demand for coffee to be (.20) and by taking into account recent trends in coffee drinking habits he estimated that be- tween 1960 and 1975 per capita coffee consumption in the 7 United States would rise only 12 per cent. According to Balassa's projections, coffee imports of all indus- trial countries would rise by 3U or 37 per cent between 1960 and 1970.8 These estimates appear to me somewhat optimistic given the experience of the 1950's. Most significantly, Cohen's data indicate that coffee imports by the United States, Canada and Western Europe have declined by 12 per cent between 1952—54 and 1962—614.9 It is of interest to compare Daly's and Balassa's estimates with those employed by the FAO in its projec- tions of world coffee consumption (see Table A.1). Based on these studies it is evident that Brazil's coffee exports face a price inelastic world demand. This is collaborated by the government's domestic coffee policy that is a reflection of the tacit acceptance of the notion of the inelasticity of world demand for Brazil's coffee. 197 .momcmco moflha mom oopmSmom coapoomosm Amy ”mopoz .H magma .Hm .a .Asemfi .mCOHpmz amuse: “mscme mess ere meme pom coapomwopmnnmmfipfiooEEoo HmLSpHSoHLm¢ .COHumNHcmeo HmLSpHSoHLw¢ ocm ooom ”mossom m.| mosm sasm mH.I .m.c mmflppcsoo .ppmsocfi HH< m.mH+ mafia Hmofi sm.- .m.c maopsm campmmz Hmpoe m.HH+ mmm mmm m.| m. maopsm esopmoz pocpo m.m+ so» owe m.| w. 0mm m.:| . Hmza Hana H.I m. weapoE< gpmoz puma mwflhMplll coHpQESm Amv coflp ,ov ”xv Icoo oommoo nqezmcoo mommoo mpflowpmmfio mpflofipmmao mops so :nmsmfimmm: HmSpom :mlmmma popothLQ mwma mOHLm oEoocH >chsoo Ame Ase Amv Amv AHV - .AmCOp osmmsocuv coflumazmcoo mmw%Oo pmpoomoso pew Hague; pew pcmsmp mommoo mo mmwufiowpmmam mfioocfi ham moflmdxv.fl.< mumce 198 The government policy was to purchase coffee from planters and store it in years when the world market would not absorb Brazilian coffee exports at the "de- sired" price. This effort reflected the dominant sup- plier position of Brazil in the international coffee market.10 Leff noted that the coffee market could be described by an "open" oligopoly model, which would ex- plain the policy of high prices and the steady loss of Brazil's market share to its competitors.11 Next I will discuss at some length the available vfitimateS of the elasticity of demand as regards Brazil's non-coffee primary exports. 2. Cocoa, Cotton and Other "Minor" Primary Commodity Exports The analysis of world demand for Brazil's coffee exports can apply, with some qualifications, to its cocoa exports. The major difference is that Brazil is not a dominant supplier of cocoa in the world market. In 1960 Ghana and Nigeria together accounted for 50 per cent of world exports, while other African producers supplied another 20 per cent. Brazil's share of the world market in cocoa has been on the decline in recent years. During the period between 19U8 and 1952 Brazil supplied about 1“ per cent of total world exports, but its share 12 fell to about 10 per cent in 1961. Over the postwar years cocoa declined in importance in Brazil's total 199 commodity exports. During 19U5-A9 cocoa exports averaged 4.3 per cent of the total value of exports; in 196U-67 13 its share fell to 3.7 per cent. The main markets for Brazilian cocoa beans are the industrial countries. The largest single buyer is the United States that together with Canada accounted for about “6 per cent of Brazil's cocoa exports. In all, about 70 per cent of Brazil's cocoa exports during the 196A-65 period were marketed in industrial countries.lu Cocoa may be faced with even less "favorable" world demand conditions than coffee. Balassa in his projection of world cocoa consumption used the income elasticity of .35 for the United States, .A0 for Western Europe and .50 for most other European countries.15 Both the Balassa and the FAO studies found the price elasticity of demand of cocoa to be considerably smaller than 933.16 J. Richard Behrman obtained two sets of estimates of demand elasticities of cocoa in the major consuming countries.l7 His major findings are reproduced in Table A.2. His estimates suggest that for the United States and the United Kingdom cocoa is an inferior good. It may be that with higher per capita incomes there is a tendency for cocoa to become an inferior good. This obviously would put the long run trend for world cocoa exports in a very unfavorable position. Thus, the almost continuous decline in the quantity of 2C)O “coowchmmzv mwoalnmEocoom cmflafimmgm we» no amassm .HHH>xx wanes .me .Q .Aewoa .swmmsem cmHHHNmem ”Sosa ooumasoamu 4 ooopowom mocmeoLuoz on» new memenoo ommz .mopMom oouflc: on» .mupooxo moooo :mHHHNMLm prou .Hmsmfi mm. as» ea .fim>ma OH. we» pm .Hmsma me. was as ”.0 .Q .AoESHo> CHV ucoo Loo no ozone Hess mes eo AHV scmoweflcmfim + scmofleacwfim .** semOfleHcmHm *. .Hosofi mooo. one as acmOHoHcmHm * ”mmpoz .H wanes .efis .AmmmH .smev Hgsxa .manOCOUm Esme do Hmebzos :.Hmmauomma .mofispcsoo mcfiesmcoo wcfiommq o>Hm we» CH mofiowofipemfim reason no >o3pw < "moooo= .CMELme omeOHm .h "oopsom mm. +mm.l *O:.I *N©.HI ms. o:.l **HH.I ***mm.H| mmpwpm popes: mm. +om. Ho.l *mm.l mm. ***w©. ***Ho.l *mm.l Eoewcfix oopHCD om. mm. +NH. +ma. cm. as. mH.| +2H.I mocmaponpoz Hm. mo. mo.- see. as. mop. mo. *Hm. semapwe use; an. **mm. +Ho.l +mm. . 2m. +zm.l **:.| +mm. mucosa mesm sea: pmwzm coma zpfiofiuwmae mpfiowpmmao zufioflpmmam >pHoHpmmHo mpfiofiummam zuHOHomme mm ooflpqnmmopo mowed oEoocH mm oofipqummopo mofipm oEoocH zppcsoo mosmsom pmmoa mgmcfiopo moosefipmo ofiemfism> HmocoespumcH Aamawlommfi .moooo Loo sesame on» go wopmefipmo zpaofiummamll.m.< mqmf‘about 65 per cent of all Brazilian exports of textile i'ihers; Eastern Eurone and Japan purchased about 8 per \Hént each. The future expansion of Brazil's exports of Ivztura. f hers depends on the expansion of textile manu- ificturinw in Western Europe and Japan and on the extent (fi‘ substitution of synthetics for natural fibers in the 902 industrial countries. The FAO projections do not chart a very favorable future course for Brazil and other ex- porters of textile fibers. At 1961—63 prices, the volume of world demand for cotton imports in 1975 is projected to be 15 to 20 per cent 19wg£_than in 1961-65. Net im— port requirements in 1975 are projected to be 20 per cent lower in Western Europe and 35 per cent lower in other developed countries.22 Sugar (SITC 061) and tobacco (SITC 121) are two primary commodities whose share in total exports have been essentially unchanged over the last decade. During 196U-67 exports of sugar were about $6M million annually and that of tobacco about $25 million annually. In 1957-59 sugar constituted 3.7 per cent of all exports and tobacco about 1.2 per cent, while during the 196A—67 period the two commodities' share of total exports was 3.9 and 1.5 per cent, respectively. Once more the main markets for both of these commodities were found in the industrial countries. The largest share of Brazilian sugar exports in 1964-65 was directed to the United States (59 per cent), Western Europe being the main secondary market (16 per cent). Regarding tobacco the situation was reversed; Western Europe being the main market (with 73 per cent) and the United States being a distant second (11 per cent). 203 Both the FAO and the Balassa studies forecast only a very small expansion of sugar imports in the industrial 23 countries in the near future. The exports of tobacco face somewhat more favorable prospects in the coming years. In fact, Brazilian exports of tobacco increased from an average of “5,000 tons in 1961-63 to about 57,600 tons in 196u-65, an increase of about 28 per cent.2u The future prospect of exports into Western EurOpe seems especially favorable.25 The limited number of empirical studies available all indicate the demand for tobacco and/or cigarettes to be price inelastic. Sachrin estimated the price elasti— city of demand for cigarettes in the United States to be between —.3 and -.A; and the income elasticity of demand to be about .50.?6 Anna P. Koutsoyannis set out to estimate the demand function for tobacco of 1“ industrial countries. The price elasticity of demand she obtained varied from the high of —.951 (for Austria) to the low of -.1A7 (for Ireland); while the income elasticity of demand varied from the high of .828 (for France) to the low of .115 7 (for Austria).2 All the coefficients of price elasticity were negative, and in no country did tobacco turn out to be an "inferior" good. Thus, if there will be some upward movement in tobacco prices, as projected by the FAO, Brazil and other exporting countries can expect an in— crease in their foreign exchange receipts from tobacco. 20“ There are a few other primary commodities of which considerably smaller quantities were exported in the mid— 1960's. Their average annual exports in l96A-65 were: hides and skins (SITC 21), about $15 million; rubber (SITC 231) about $6 million; and forest products (SITC 2A2 and 243) about $59 million.28 Lastly, we turn to the export of iron ore, which has been growing rapidly in recent years, and soon may replace cotton as Brazil's second most important com- modity export. Exports of iron ore (SITC 281) averaged $92 million during l96u-65 and $102 million in 1966—67. In 1966-67 about A3 per cent of Brazil's iron ore exports were directed to the EEC, about 21 per cent to North America and about 1A per cent to Japan. Cohen's study found that the imports of iron ore by North America and Western Europe grew at the annual rate of 8 per cent between 1952—5“ and 1962-6A. This compares favorably to the growth rate of imports of all primary commodities (3.3 per cent) and even of all commodities (6.9 per cent) 29 during the same period. Future prospects appear quite favorable. Balassa projected that the iron ore exports of underdeveloped countries will rise by 75 or 100 per cent between 1960 and 1970; and 140 or 180 per cent be- tween 1960 and 1975.30 In fact, between 1961 and 1967 the volume of Brazil's iron ore exports increased by 31 about 128 per cent. Given Brazil's very large reserves 205 of iron ore, second only to the Soviet Union, and the deliberate policy of the government to increase produc- tion for exports, the future prospects look favorable indeed. 3. Exports of Manu- factured Goods Exports of manufactured goods (SITC 5, 6, 7 and 8) grew rapidly during the 1960's. For example, in 1962 they were about $37 million, rising to $124 million in 1965; thereby registering a 330 per cent increase over this period. The main markets for Brazil's manufacturing exports in 1964—67 were: Latin America with about 48 per cent, North America with about 28 per cent and Western Europe with about 13 per cent. In all, during this period industrial countries accounted for about 46 per cent of 32 The preferential treat- Brazil's manufactures exports. ment accorded Brazilian industrial products under the Treaty of Montevideo largely accounted for the growth of manufacturing exports into LAFTA countries.33 I divided the overall exports of manufactures into those of the "traditional" industries and those of the "dynamic" or modern industries. The exports of the "traditional" manufacturing industries, many of them using cheap domestic raw materials, were largely directed toward the markets of the industrial countries. About two-thirds of the exports of leather and footwear (SITC 61) 206 were sold in the United States, with the share of all industrial countries being about 76 per cent. The lar- gest importers of veneer, plywood and other wood prod- ucts (SITC 631 and 632) were Western Europe (47 per cent) and North America (38 per cent).3u Textile yarn and cotton fabrics (SITC 651, 652 and 655) exports averaged about $7 million during 1964-65, and almost $10 million during 1966-67. The largest single market for these products during 1964—67 was the United States (with 79 per cent); Western Europe (about 10 per cent) was considerably less important. Brazil's textile industry is one of the earliest in Latin America and the country has largely achieved self sufficiency in textiles and wool products and certain man—made fibers. Yet the country was unable to become a significant exporter of cotton textiles. Let us take the United States market as a special case. In 1965 the United States imported $409 million worth of textiles (SITC 651) from the less developed countries; the share of Brazil of this total was $8.5 million or only about two per cent.35 Among the "dynamic" industries, the case of chemi- cals stands out because it is the only modern industry whose exports were largely directed toward industrial countries. In 1964-65 exports of chemicals (SITC 5) were about $16 million a year, rising to about $28 million a year in 1966—67.36 The main markets for Brazil's chemicals 207 were the United States (48 per cent), Western Europe (23 per cent) and Latin America (13 per cent). ECLA provided an estimate of the average 1959 prices of an extensive group of chemicals produced by six Latin American coun- tries.37 Among the six LAFTA countries, only the prices of Brazil's chemicals were found to be lower than com- parable prices in the United States. This puts Brazil in a favorable position for future expansion of exports of chemicals, within LAFTA and in third country markets. The steel industry in Brazil was established as early as 1925, but the major expansion of productive capacity came with the creation of the Volta Redonda flant in 1946.38 During 1964-66 Brazil's production of steel ingots averaged about 3 1/4 million tons, with a planned expansion of capacity by about one million tons in the near future.39 Exports of iron and steel (SITC 67) during the 1964—67 period averaged about $32 million with the bulk of it going to Latin America (60 per cent) and about 37 per cent going to all industrial countries. Despite the considerable headway made through import substitution in this sector, Brazil was still a net im- porter of steel in 1964-67.“0 However, given the locally available supply of high grade iron ore and a large and expanding domestic capacity, the chances for expanding exports, especially into LAFTA countries, appear quite favorable. 208 Exports of "basic" manufactures (SITC 6) as a whole expanded rapidly. In 195U-55 these exports averaged $8.2 million, rising to $58 million in 1964—65 and finally reaching $73 million in 1966-67. Over the 196U-67 period about “6 per cent of exports in this category were sold in Latin America, with about U9 per cent going to indus- trial countries. ln the category of machinery and transport equip- ment (8 TC 7) two items stand out--exports of non-electric machinery (SITC 71) that averaged about $19 million in 196U—67, and exports of transportation equipment (SITC 73) that averaged about $7.2 million. In the case of both of these goods, between two-thirds and three-fourths of exports were directed to Brazil's Latin American neigh- bors. In 196U exports of motor vehicles were about $2.0 million, rising to about $4.5 million in 1966.“1 While these figures in themselves are not impressive, it must be born in mind that the Brazilian automotive industry came into existence in 1957ou2 In 1957 only about 30,700 motor vehicles (mainly trucks and Jeeps) were produced; ten years hence, in 1966, 225,000 units were produced. Given that Brazil is the lowest cost producer of motor vehicles in South America, it is possible that her exports of automotive vehicles to other LAFTA coun- tries will expand beyond the $U.2 million achieved in 1966.“3 209 Ml tlu' catr nuuflnine {uni trunn:pord;m n L .mez Loo mLmHHoc CH muLooxo Lo ommoLocH mmmLo>w n n ”mmuoz .AQ xfiucodo:.m mm.H: 2:.Hm Hm.mm m:.o: no.0H m:.o m:.mm :m.m mo.mm welmmma oo.m mc.om m~a.| mm.m| :m.:: mm.mw :m.mm mm.mm m:.o mm.m: 50m. mm.m molamma pm.mH Hm.mm as.» m.mma no.0m nm.oa mz.mm mo.m m:.oa mm.om Hs.m n.aoa lesmmfi om.m :m.~m oa.m Ho.Hm ma.mm :m.m mm.om mo.m mw.| mo.ml mm.m H:.Hm wmlmmmfi L o L n L s L o .L o L n mpLoaxe messes m + s + w + m -coc Hesse msLocxm Hmsoe essm s + s eme m eLHo H + o eeHm .mmoHLd Ammefl .L. UCWUMCOO CH huLooxo cwHHHNmLm no mecmLp Lmocfla ucm Aocsoofioov .mamzu HmLOCwEIIm QBHW mmme cosoLm Hmsccm u L .mes me mLmHHon Cw mpLooxm Lo mmmoLoCH mamLo>m n o ”mmpoz .AQ xaeceooon .ooom .omummma .mmHLommLme osHm co poems .mpLoexm :mfiHHNmLm Lo mocmLp mesfia m:m Aocsorcoov wouML cuROLm HmscctPlal output. 66 Leff's hypothesis "explains" the recent Bra2111an recession in terms of factors on the \g_ #_ _ " “A...“ supply side, contrary to the official explanation, that lists deficiency of aggregate demand as the main contri- butory factor.67 7 An ECLA study pointed out that from the macro— economic standpoint further import substitution may even tend to slow down economic growth.68 Given the type of Products that currently constitute the range of substi- tutable imports, further substitution would lead to 232 investment in projects with high capital-output ratios. Thus, to attain the rate of growth experienced in the 1955-61 period, a more intensive rate of investment must be achieved than has been realized so far. Given that Brazil's import coefficient is one of the lowest in the Western world, its maintenance at the present level would imply a substantial increase in the rigidity of the import schedule.69 In fact, Brazil's development process can be seriously hampered by the stagnancy of its imports (see Leff's hypothesis above). These considerations brought forth the following note in the ECLA study: All that has been said so far bears out the argu— ment that the strategic problem confronting the Brazilian economy is how to make the transition from an import substitution model to a self- sustaining growth model.7O In fact, this comment may be applicable to several less developed countries that have reached the same stage in their development. FOOTNOTES: APPENDIX A lln a recent study, B. 1. Cohen calculated the annual growth rates of primary products imports of the major industrial countries from the less developed coun- tries. Cohen computed the growth rates over the period for 1952—59 to 1962-69. The growth rates of the com— modities that are of the major concern to Brazil were: coffee, -l.9 per cent; cocoa, —2.9 per cent; cotton, —2.9 per cent; sugar, +2.7 per cent; and tobacco, +9.2 per cent. See: B. I. Cohen, "The Less Developed Countries' Exports of Primary Products," Economic Journal, LXXVII (June, 1968), 339-93, Table III and IV. 2Since the Korean War, coffee prices have rebounded for several years. In 1959 frost damage in Brazil sent prices of santos no. 9 (in New York) soaring to a 79 cents per pound average for the year. But since 1958 prices of coffee have been generally on the decline, reaching a low of about 39 cents per pound in 1962-63, and rebounding to about 93 cents per pound in 1965-66. 3The four largest African exporters are: Uganda, The Congo, Kenya and Tanzania. See: B. Balassa, Trade Prospects for DevelopingACountries (Homewood, Ill.: Richard D. Irwin, 1969), p. 197. “The lower prices and favorable technical charac- teristics of the Robusta (African) brand relative to Brazil's brand led to their increasing use in the pro- duction of soluble (i.e., "instant") coffee. A study based on United States data for the period 1953-63 found the elasticity of substitution between Robusta's and Brazil's to be -.98. See: G. Lovasy and L. Boissonneault, "The Inter- national Coffee Market," IMF Staff Papers (November, 1969), 378—80. 5Baer, op. cit., p. 39, Table 3-9. Also: ECLA, Economic Survey of Latin America, 1969 (New York: United Nations, 1966), p. 239, Table 2367 6Rex F. Daly, "Coffee Consumption and Prices in the United States," Agricultural Economics Research, X (July, 1958)) 61‘71- 233 239 {Calculated from: Balassa, Trade Prospects . . ., op. cit., p. 201. 8The two different figures arose because of dif- ferent assumptions regarding growth of income in the industrial countries. They also depend on the assumption that export quotas for coffee will be extended within the framework of the International Coffee Agreement, so as to prevent any sharp declines in coffee prices. On May, 1967 the Agreement included 38 exporting countries, accounting for 99.1 per cent of world exports of coffee and 23 importing countries accounting for 96 per cent of total imports. See: Ibid., p. 209. 9 Cohen, 0p. cit., 339, Table IV. 10For detailed exposition of this policy see: Leff, op. cit., pp. 20—32. 11In this model the largest firm may be maximizing profits by setting a higher price than would be consistent with the operations along the long run demand curve for the industry. This policy of high pricing encourages entry of new competitors into the market (re: the growing share of the African producers) and thus reduces the share of the major supplier in the market. See: George Stigler, The Theory of Price (New York: MacMillan, 1952), pp. 232-39. 12These figures refer to exports of cocoa beans as well as cocoa butter and paste. See: Baer, op. cit., p. 39. 13lbid., p. 36. 1“The 1969—65 figures are reported fully in the Data Appendix. The industrial countries include the following: United States, Canada, Western Europe and Japan. 15Balassa, Trade Prospects .'. ., op. cit., pp. 206— 08. The FAO study used the following income elasticities for cocoa-~in the United States, Canada and United King- dom: (.0); EEC countries: (.30); other Western Europe: (.20). Taken from: FAO, Agricultural Commodities——Projec- tion for 1975 and 1985 (Rome: FAO, 1967), p. 52, Table 1.23. ‘( l)I"or example the FAO study employed the following price elasticities of demand for cocoa--North America (-.20); EEC (-.30); other Western Europe (—.U0). Taken 'from: FAQ, on. cit., p. 52, Table I-23. l7Behrman used the standard least squares multiple regression method on the one hand, and a two stage least squares estimating procedure using instrumental variables, on the other hand. See: J. Richard Behrman, "Cocoa: A Study of Demand Elasticities in the Five Leading Consuming Coun- tries, 1950-1961," Journal of Farm Economics, XLVII (May, 1965), A10—17. 18Calculated from the Data Appendix (Appendix D)- l)Balassa, Trade PPOSPeCtS ° ° '3 99’ Cit" pp . 2140-57 . 2OJ. R. Donald, F. Lowenstein and M. S. Simon, "The Demand for Textile Fibers in the U. 8.," U. S. Department of Agriculture, Economic Research Service, Technical Bulletin no. 1301, November, 1963. (Mimeographed.) 21 Ibid., pp. 59—62. ’)') LLFor detailed analysis see: FAO, op. cit., pp. 281-82. 23Balassa projects a two to three per cent increase in sugar imports between 1960 and 1970. See: Balassa, Trade Prospects . . ., op, cit., pp. 183-8“, also: FAO, 0p. cit., p. 190. 2A FAQ, on. cit., p. 217. f.‘ 2”These prospects will depend to a considerable de- gree on the agricultural imports policy of the EEC. See: Balassa, Trade Prospects . . ., pp. 185—90. The FAO study forecast shows that the projected import requirements of the industrial countries in 1975 are likely to increase more than the exports availabili- ties of the developing countries and Eastern Block coun- tries. This may lead to some upward adjustments in world tobacco prices. See: FAO, op. cit., p. 217. 26S. M. Sachrin, "Factors Affecting the Demand for Cigarettes," Agricultural Economic Research, XIV (July, 1962), 81—88. 236 Z/A. P. Koutsoyannis, "Demand Function for Tobacco," The Manchester School of Economics and Social Studies, XXXl (January, 1963), l-20. 28Rides and skins were largely exported to the United States (Ml per cent) and the EEC (34 per cent). Rubber was mainly marketed in the United States (U6 per cent) and Latin America (A8 per cent). The central markets for forest products were Western Europe (A2 per cent) and Latin America (51 per cent). All figures pertain to the l96u-65 period and were calculated from the Data Appendix. G CQCohen, op. cit., 339, Table IV. 30The alternative estimates are based on different projections of the rate of growth of income in industrial countries. See: Balassa, Trade Prospects . . ., op. cit. pp. 291-98. ~ 31From 6,236 thousant metric tons to lu,279 thousant metric tons in 1967. From: Survey of the Brazilian Economy, 1966 (Washington, D. C.: Brazilian Embassy, December, 1967). 32 Calculated from my Data Appendix (Appendix D). 33For details see: Lee, "Brazilian Exports of Manu- factured Goods," Conjuntura Economica, XIII (May, 1966), U5-50. 3 I 1These data refer to the l96U-65 period. t 3')In 1967 the imports of textile products (SITC 651, 652 and 655) were only about $3 million. For U. S. im— ports of textiles (SITC 65) see: Lary, op. cit., p. 117, Table 17. For Brazil's exports of textiles (SITC 65) see: United Nations, Commodity_Trade . . ., op. cit. 36It is of interest to note that the 1966—67 annual exports of Brazilian chemicals alone exceeded the combined export of all_manufactured goods in 1960. See: Lee, op. cit., A6, Table I. 3[The other countries were: Argentine, Chile, Columbia, Mexico and Peru. See: ECLA, The Process of Industrial Development in Latin America (New York: United Nations, 1966), p. 103 237 ’ P O I 3’I).ra'/.L_1_'s consolidated steel industry came to rep- resent-the largest productive capacity in Latin America and was composed of the largest number of integrated plants. See: Ibid., p. 105 39ECLA, Economic Survey of Latin America——1965 (New York: United Nations, 1967), pp. 312—13. 0Imports of iron and steel, averaged about $60 million over the 196H—67 period. I llThese figures include exports of automotive parts and accessories. ' ugPrior to 1957 only assembly and some automotive parts production existed in Brazil. See: Gordon and Grommers, op. cit., Chapter IV. 43A recent study compared the cost of automotive pro- duction amongst three Latin American countries. Munk found that relative to Argentina and Mexico, Brazil was the low cost producer of motor vehicles in Latin America. See: aernard Munk, "The Welfare Costs of Content Protection: The Automotive Industry in Latin America," Journal of Political Economy, LXXVII (January/February, “These primary exports are: fruits and vegetables (SITC 05), sugar (SITC 061), coffee (SITC 071), cocoa (SITC 072), hides and skins (SITC 211), oil seeds, nuts, etc. (SITC 221), lumber (SITC 2&3.2); cotton (SITC 263), iron ore and other nonferrous base metals (SITC 281 and 283). See Data Appendix for further details. 145. Michaely's index is: 1 X. l ..z[_x_11] Where: Xi, = the exports of commodity i, by country j Xi = world exports of commodity i Xj = total exports of goods of country j See: Michaely, Concentration . . ., op. cit., Chapter 3. 238 ubThe annual growth rates of these commodities were computed from sources listed in the Data Appendix. The formula for calculating compound growth rates was obtained from: Daniel B. Suits, Statistics: An Introduction to Quantitative Economic Research (Chicago: {and McNally Co., 1963), Chapter IX, pp. 203—211. ”7 Computed from the Data Appendix. V x P m 8Income terms of trade = where VX is the index number of the value of exports and Pm is the price index of imports. For details see: Gerald M. Meier, The International Economics of Development: Theory and Development (New York: Harper & Row, 1968): Chapter 3, pp. u3-uu. ( u)The rationale for this statement is contained in the following formula: =1. '12, nx A nw A 0 Where: nx = the world demand elasticity for a commodity export, supplied by Brazil. nw = the world demand elasticity for that commodity. EC = the elasticity of export supply for that com- modity from countries competing with Brazil. A = Brazil's share of the world exports of that commodity. B = competitors share of world markets. See: Yeager, op. cit., p. 140, footnote A. U )Ooalculated from Data Appendix. C JlYeager, op. cit., pp. 139-U0. 52Clark and Weisskoff, op. cit. 53 f _r _u Ibid., pp. )2 33, Table C A. r- )ulbid., pp. 11-12. 239 L" L‘ DJMorley included in his estimates the price of Brazilian goods in the relevant category relative to the general price index. The rationale for inclusion of the second relative price ratio was to enable him to test the hypothesis that only relative prices of imports and close substitutes were relevant to import demand. See: Samuel A. Morley,"Import Demand and Import Substitution in Brazil," Brazil Development Assistance Program, University of California, Berkeley, Calif., no date. (Mimeographed.) 561bid., p. 9 C.‘ )7Ibid., p. 19. 58 Hirschman, op. cit., 1—32. f )gharry G. Johnson, "Tariffs and Economic Development: Some Theoretical Issues," Journal of Development Studies, I (October, 196U), 3—30. 60For the "early" import substituting industries expansion will imply movement "downward" along the de- clining section of the industry's average cost curve. Establishment of additional industries through import substitution would lead to a higher average cost of pro- duction, because of the inability of the new industries to engage any significant economies of scale. The devel- opment of the automotive industry may be a demonstration of the latter point. For the connection between market size and econo- mies of scale in motor vehicle production see: Munk, op. cit., pp. 85—98. 61During the period 1953-66 the annual growth (nega- tive) rate of imports (in current dollars) was -.32 per cent; while in the same period in constant 1953 dollars imports decreased by .65 per cent a year. 62United Nations, The Economic Development of Latin America in the Postwar Period (New York: United Nations, 196“), p. llU, Table 113. 63 6uLeff in a detailed study of the Brazilian capital goods industry noted the increase in the share of domes- tic output in the total purchases of producers equipment over time. In the late 19U0's the domestic industry supplied about 60 per cent of the market for capital equipment in Brazil, its share rising to 75 per cent by 1957-59. This increase in the domestic industry's market Calculated from statistics in the Data Appendix. 240 share took place while demand for capital equipment was rising steadily. In 1959 the value of equipment sup- plied by domestic sources was approximately 115 per cent higher than in 1947. See: Leff, The Brazilian Capital Goods . . ., op. cit., Chapter VI. 65Brazil's net domestic product increased at the ' average rate of 6.7 per cent during 1956-62, while rising only at the rate of 2.6 per cent during 1962-66. See: ECLA, Economic Survey . . ., op. cit., pp. 101-02, Table 69. 66Nathaniel H. Leff, "Import Constraints and Devel— opment: Causes of the Recent Decline of Brazilian Economic Growth," The Review of Economics and Statistics, XLIX (November, 1967), D9D-501. 67See for example: J. Bergsman and S. Morley, "Import Constraints and Development: Causes of the Recent Decline of Brazilian Economic Growth: A Comment," The Review of Economics and Statistics, LI (February, 1969), 101-102. 68ECLA, "The Growth and Decline of Import Substitu- tion in Brazil," Economic Bulletin for Latin America, IX (March, 196“), 1-59. 69 Ibid., 26—27. 70Ibid., 57. APPENDIX B THE AVAILABILITY OF VARIOUS SKILL CATEGORIES IN BRAZIL Engineers In Chapter III the overall shortage of skilled per- sonnel at all levels has been noted. This is especially true with regard to the supply of engineers. For example, the Brazilian capital goods producing sector employs 2.9 engineers per 100 employees as compared to 5.7 engineers per 100 employees employed in the United States electri- cal equipment and machinery industries.l Nathaniel Leff found that the salaries paid Brazilian engineers in the capital goods industry were 12 times the wages paid un— skilled labor. This is in contrast to a ratio of “.0 to 5.5 prevalent in several industrial countries.2 This large labor wage differential in Brazil exists because the supply of engineers has not kept pace with the rapidly expanding demand. This demand expansion re- sulted from the rapid growth of output in the manufac- turing sector as a whole and in the heavy equipment in- 3 dustries in particular. One reason that the supply of engineers has not kept pace with the growing demand is 2A1 2A2 the inadequacy of existing educational facilities. In 1958, there were 9,786 candidates for 1,6A5 openings in the engineering schools, i.e., a ratio of 3.3 candidates per opening. While the number of openings rose to 6,556 in 196“, the ratio of candidates per opening in fact in- creased to 3.8 and was as high as A.2 in the non-civil engineering field.“ Thus, because of the slow expansion of engineering schools, excess demand for engineering education has developed; and it was exacerbated by the high quasi—rents paid to members of this profession. Despite the fact that the rate of increase of enrollment in engineering was faster than the overall increase in higher education enrollments, the serious shortages of 5 engineers was not alleviated during the 1960's. Supply of Managerial and Administrative Personnel Almost no evidence exists regarding the supply of these types of highly skilled personnel in Brazil. John Shearer studied at some length the personnel practices of a group of United States' subsidiaries in Mexico and Brazil.6 Based on interviews and questionnaires, Shearer's findings generally confirmed the existing scarcity of highly qualified administrative and managerial personnel in Brazil. He succinctly remarked: "In both countries nationals who are well qualified to handle high- level positions in modern industrial Operations are in ”7 very short supply. He also discovered that United 2A3 States' subsidiaries lack the imagination to tap ade- ‘quately the existing domestic supply of skilled personnel. Most of the companies rely on their own facilities for training and developing managerial personnel, usually consisting of recent college graduates. Shearer's study is of limited coverage because it deals only with United States subsidiaries in Brazil and Mexico. However, the 21 Brazilian subsidiaries included in the study cover most of the major manufacturing ac- tivities and the majority of these industries can be "9 classified as "modern. We can only surmise that con- ditions of employment and training of local management personnel in purely domestic enterprises lag behind those prevalent in the United States' subsidiaries. The Shortage of Skilled Labor Over the post World War II period the number of people with at least primary education (assumed to be the minimum educational input required of skilled labor) in Brazil has increased at an impressive pace.10 As a result, the supply of skilled labor available to industry has expanded as well. It is estimated, for example, that the number of skilled workers in the metal-working industries in the State of Sao Paulo rose from 2A,700 to 75,000 in the 1951-63 period, at a compound rate of 10.A per cent annually.ll Despite these impressive increases in the availability of skilled labor serious shortages prevailed 2AA as reflected in the high relative wage of such workers. The wages of master workmen relative to unskilled workers in Brazil's capital goods industry were between 30 and 90 per cent higher than those paid in several industrial countries.12 The Ministry of Education estimated the 1 country's "needs' for its basic and manufacturing indus- tries at some 2,500 engineers, 5,000 industrial techni- 13 cians and 60,000 skilled workers annually. In com- parison, between 1950 and 1959 it was estimated that the number of new engineers graduating averaged only about 980 a year.lu In the light of past performance it is doubtful whether Brazil's system of education is capable of generating the projected numbers of engineers, tech- nicians and other skilled workers needed to fill indus- tries' demand. A study of the availability of trained technical personnel and skilled workers in the metal- working industry supports this skeptical attitude.15 (See Table B.l for additional information.) The question that comes immediately to mind is whether the productivity of Brazilian labor has been markedly affected because of the lower educational input of these workers, relative to the United States and Western Europe. An ECLA study found that the physical productivity of Brazilian labor in the heavy metal—transforming in— dustry is only about 80 per cent below the labor 2A5 TABLE B.l.——Shortage of technical skills in the metal- working industry—~1956. Total Fulfillment _ Shortage requirements Specialized engineers 240 80 160 Industrial technicians U80 80 “00 Skilled workers 10,000 U,500 5,500 Source: Based on a paper presented at a Latin American meeting of experts in steel-making and trans- forming industries; sponsored by ECLA, Sao Paulo, October 15—28, 1956. Cited in w. B. Dale, Brazil/Factors Affecting Foreign Investment Menlo Park, Calif.: International Industrial Development Center, Stanford Research Institute, September, 1958). productivity in the United States and Germany, when similar capital equipment is used.16 M. E. Kreinin com- pared the labor productivity of American firms producing in the United States and abroad under roughly similar 17 capital and scale conditions. Dr. Kreinin found that in Latin America labor "effectiveness" was some 30 per 18 This difference cent below the United States standard. in labor effectiveness was attributed about half of the time to lower educational inputs and inadequate training and skills.19 A study of the Brazilian textile industry found that in a sample of 204 cotton spinning mills only about 1“ per cent of the variation in productivity can be attributed to obsolescent equipment and to mill size.20 The study 2M6 concluded that the explanation of most of the variations in productivity among Brazilian mills must lie in what it called the "human factors" involved in the production process. While not being able to isolate each one of these factors the study noted: The most important is mill management, including the whole concept of the entrepreneur's or manager's responsibility as regards use of satisfactory raw material, careful machinery maintenance, manpower training and so forth.21 Based on the ECLA study I can conclude that the lack of managerial and supervisory personnel was largely respon- sible for the low productivity in the textile industry and probably in other industries as well. FOOTNOTES: APPENDIX B 1The Brazilian engineering coefficient has been adjusted to reflect employees of comparable output. See: Leff, The Brazilian Capital Goods . . ., 9p. cit., p. H9. 9 LThe ratios of salaries of engineers to semi- skilled workers were: West Germany “.0; Japan 5.0; United Kingdom “.5; and United States 5.5. See: Ibid., p. 61. 3Estimated growth rates of output of mechanical equipment, manufacturing production and GNP, 1960-63. (percent) Year Mechanical equipment Manufacturing GNP output output 1960 1A.o 11.1 6.7 1961 1U.0 8.1 7.3 1962 20.0 - .3 5.“ 1963 2.6 5.1 1.6 Sources: Leff, Ibid., p. 39, Table II-6. ECLA, Economic Survey . . ., op. cit., p. 1&5, Table 95. “John V. D. Saunders, "Education and Modernization in Brazil," in The Shapingiof Modern Brazil, ed. by Eric N. Baklanoff (Baton Rouge, La.: Louisiana State Univer- sity Press, 1969), p. 136, Table 6. 5While enrollments in higher education as a whole increased sixfold between 1940 and 1964, in engineering they increased more than eightfold and about 17.5 times in "economic sciences." In addition, a relative shift in enrollment took place toward professions that may be termed "modern" (non-civil engineering, agronomy, economics, 2A7 248 etc.) from the more "traditional" categories of law, philosophy and civil engineering. No. enrolled Index % of total 1953 1964 1953 1964 1953 1964 Traditional(l) 45,638 88,502 100 194 83.8 67.9 Modern<2> 8,830 41,816 100 474 16.2 32.1 Source: Saunders, op. cit., p. 138, Table VII. (1) Notes: Traditional professions: Philosophy, science, letters, law, civil engineering and dentistry. (2)Modern professions: Non-civil engineering, economics, agronomy, social work, administra- tion and architecture. 6John C. Shearer, High Level Magpower in Overseas Subsidiaries: EXperience in Brazil and Mexico (Princeton, N. J.: Industrial Relations Section, Princeton University, 1960). 7 Ibid., p. 93. 8Very limited facilities for developing and training management personnel exist in Brazil. The most extensive program in business administration on both undergraduate and graduate level is conducted by the School of Business Administration of Sao Paulo. Three times a year this school provides a thirteen-week course in advanced manage— ment as well as certain extension services. See: Shearer, op. cit., pp. 107—08. 9The 21 U. s. subsidiaries in Brazil are distributed as follows: Chemicals 4; Machinery (non-electrical 2; Electrical machinery 3; Transport equipment 2; thus, about one—half are in these four "modern" activities. See: Ibid., Appendix B, p. 147, Table B—2. 10Enrollments in Brazil's primary schools doubled between 1948 and 1961. Yet in 1962 Brazil ranked lowest among LAFTA countries in the number of people enrolled in primary school per 1000 population and as a percentage of primary school age group. See: Table 8.2, p. 249. As an additional comparison we present data on public expenditure on education in Brazil and other Latin American as well as selected industrial countries. See: Table 8.3, p, 250. 249 TABLE B.2.——Primary education enrollment in Brazil and other LAFTA countries, 1962. Country No. enrolled Enrollment Rank Enrollment Rank (thousands) per 1000 of as a % of population age group 7-12 Argentina 3,056 143 5 114 2 Brazil 7,846 106 9 72 9 Chile 1,274 159 2 107 4 Columbia 1,904 116 8 73 8 Ecuador 642 140 6 87 6 Mexico 5,620 146 '4 92 5 Paraguay 323 174 1 120 1 Peru 1,562 147 3 88 Uruguay 343 118 7 112 Source: Sylvian Lourie, "Education for Today or Yester- day," Problems and Strategies of Educational )f—i Planning, ed. by Raymond F. Lyons (Geneva: UNESCO, 1965), pp. 28-44, Table VI. 250 TABLE B.3.——Pub1ic expenditure on education in Brazil, LAFTA countries and selected industrial countries--1960. Country Public ex- Per capita Rank Public ex- Rank penditure public ex- penditure on educa- » penditure on educa- tion _ ($) tion as a (mill. $) % of GNP Argentina 182.6 8.83 2 1.9 5 Brazil 250.5 3.59 7 2.1 Chile 121.9 15.85 1 2.8 1 Colombia 68.0 4.41 6 1.9 6 Ecuador 12.3 2.85 8 1.6 8 Mexico 207.6 5.76 4 1.9 7 Paraguay 3.5 2.00 9 1.5 9 Peru 51.9 5.18 5 2.6 2 Uruguay 24.7 8.73 3 2.3 3 SELECTED INDUSTRIAL COUNTRIES U.S.—1961 23,000 125.0 4.40 Canada-1960 1,670 93.3 4.38 United King— dom—1961 3,267 61.65 4.25 Germany— 1961 2,337 41.59 2.86 France—1960 1,460 31.95 2.42 Source: Sylvian Lourie, "Education for Today or Yester- day," in Problems and Strategies of Educational Planning, ed. by Raymond F. Lyons IGeneva: UNESCO, 1965), pp. 28-44, Table VI. 251 lleff, Brazilian Capital Goods . . ., Op- Cit‘) p. 69- 12Ibid., p. 70- 13A. B. AraoZ, "Manpower and Employment in Brazil," International Labour Review, XCIII (April, 1966), 382. 1“Havighurst and Moreira, Society and Education in Brazil (Pittsburgh: University of Pittsburgh Press, 1965), p. 205, Table 33. C l)Automotive executives in Brazil estimated that within a few years their industry alone will need five or six times the number of technical personnel now traiped annually in Brazil. See: Shearer, op. cit., p. 94. l6ECLA, The Manufacture of Industrial Machinery and Equipment in Latin America: I, Basic Equipment in Brazil (New York: United Nations, 1963), p. 60. 17M. E. Kreinin, "Comparative Labor Effectiveness and the Leontief Scarce—Factor Paradox," American Eco- nomic Review, LV (March, 1965), 131—40. 18ibid., 137, Table 3. 19 2OECLA, The Textile Industry in Latin America: II, Brazil (New York: United Nations, 1963). 21 Ibid., 137, Table 4 Ibid., p. 83. APPENDIX C THE INSTITUTIONAL SETTING OF WAGE DETERMINATION IN BRAZIL Labor Unions Based on the limited evidence available, I con- cluded that labor unions in Brazil occupy a weak bar— gaining position.1 In Brazil, labor unions have a legal status which is bestowed on them by the state upon fulfillment of certain specified conditions. Most matters that are subject to collective bargaining in the United States are regulated by law in Brazil. The tra- dition of paternalism in the relation between the state and the unions was shaped in the 1930's through the policies of the Vargas dictatorship. The end result is that trade union activities are closely regulated by the government, which controls the chartering of Sindicatos and Federations, approves candidates for union elections, interprets their by-laws and audits their income and expenditures.2 Briefly, the organization of the labor unions can be described as follows. At the lowest level of the organiza— tional structure are the Sindicatos, which are formed by workers in a sinGle industry in a given locality. Next in 252 253 the union hierarchy are the Federations, which are com- posed of a minimum of five Sindicatos representing workers in related industries. At the top of the labor union's hierarchy are the Confederations, composed of a minimum of three federations set up on the federal level. One of the four existing Confederations is the National Cdnfederation of Industrial Workers (CNTI), whose member- ship is claimed to be two million but is estimated to be in fact at three-fourths of one million. The unions bargain for "better working conditions," which is interpreted to mean higher wages since fringe benefits are provided for by government legislation. Labor contracts that govern wages are usually negotiated by corresponding workers' and employers' organizations. However, in practice most industrial disputes are settled through an intricate system of labor courts rather than by collective bargaining. This led Alexander to remark that in Brazil: "The labor court system thus largely takes the place of both collective contract negotiations and grievance procedure as practiced in the United ."3 States The threat of striking and the act of strike itself, the most potent weapons available to labor unions in industrial countries, have not been nearly as effective in Brazil.“ Thus, given the structure of Brazilian labor unions, their limited scope for collective bargaining and the largely inactive union membership; their effect on wages cannot be considered critical. 254 Some Institutional Characteristics of Wages in Brazil The Brazilian Constitution guarantees a minimum wage defined as a payment sufficient to meet the "normal needs" of a worker for board, lodging, clothing, hygiene and transportation. Changes in the minimum wage rates are based on cost—of—living studies made in 53 regions ‘0 of the country. As a result, the minimum wage varies from one part of the country to the other. In 1956 the minimum wage ranged from 3,200 cruzeiros per month in Rio de Janeiro and Sao Paulo to 1,300 cruzeiros a month in the cities of the Amazon Valley.5 There are two additional aspects of the minimum wage worth noting. In the first place, wage scales for skilled and semi-skilled workers lie considerably above the minimum although they tend to rise when the minimum wage is being adjusted.6 This can be seen in Table C.l. TABLE C.l.—-Month1y salaries and wages paid by United States companies in Sao Paulo, first quarter, 1957 (in cruzeiros). Low Average High Unskilled worker 3,690 4,230 4,640 Skilled worker 4,490 6,560 9,040 Source: Adapted from W. B. Dale, Brazil/Factors Affecting Foreign Investment (Menlo Park, Calif.: Inter- national Industrial Development Center, Stanford Research Institute, September, 1958), p. 65, Table 60. 255 When reading Table C.1, it is important to bear in mind that the legal minimum wage during the first quarter of 1957 in Sao Paulo was 3,700 cruzeiros a month. Thus, if one may generalize from Table C.1, it appears that wages paid by Brazilian industries run considelably ahead of the legal minimum. More recent evidence, based on a broad segment of economic activity, confirms these findings. Based on it, even the lowest paid industrial workers received earnings above the legal minimum wage of 44,000 cruzeiros per month (in 1966). In fact, the average wages of unskilled workers in all activities were about fifteen per cent above the legal minimum wage.7 The industrial sector of the city of Sao Paulo in 1966 reported that about 15 per cent of all employees earned less than the legal minimum wage. There are con— siderable differences among industries in the ratio of employees who are at or below the legal minimum. In the traditionally low wage industries, like textiles and clothing, the prOportion of workers earning the minimum was higher than the average; about 32 per cent in Sao Paulo and about 38 per cent in the State of Guanabara. in the more "dynamic" industries like transportation equipment, chemicals and machinery, only about 12 per cent of the employees received the legal minimum or below. Thus, it seems that the effect of increases in the mini- mum wage on inter—industry earnings would largely depend 8 on the skill characteristics of the industry involved. 2S6 Hiven the continuous shortage of skilled labor in Brazil, labor market pr ssures force the employer to maintain the real income of his employees by periodical adjustment in their money wages.9 One other important feature of a wage structure is the number and relative size of the components into which total earnings are divided. In Brazil there exists a considerable body of social legislation that provides for such fringe benefits as paid vacations, partly paid sick leave, accident insurance and retirement pensions. In addition to the payment for actual hours worked, em- ployees receive remunerations that are equivalent to a day's pay for the following: Sundays, twenty days annual vacation, average of ten holidays, and fifteen days of annual sick leave. Dale estimated that these non-wage labor costs amount to about U0 per cent of total wages for unskilled workers and about 20 per cent for employees in higher wage brackets.lO Another characteristic of the wage structure of a country is the wage differentials based on the skill re- quirements of a job. The general tendency in Brazil was for percentage differentials between wages of skilled and unskilled labor to narrow (see Table C.2). The reader must be cautioned that Table 0.2 is based on a very limited sample, i.e., the reports of United States com- panies in the Sao Paulo area. The conditions in that area 257 TABLE C.2.——Averagc monthly wage and salaries paid by United States companies in Sao Paulo (in cruzeiros), 1951, 1957 and 1959- Type of labor 1952 -lst quarter June 1957 1959 Skilled worker 2,630 6,560 9,583 Unskilled worker 1,500 “,230 6,894 Differential between skilled and un— skilled wages 1,130 . 2,330 . 6,89“ Percentage by which the skilled wage exceeds the unskilled wage 75 55 39 Source: Adapted from Gordon and Grommers, U. S. Manufac- turing Investment in Brazil (Cambridge, Mass.: Harvard University Press, 1962), p. 119, Table 18. may not be representative of most other parts of the country. Generally it seems reasonable to assume that the supply of skilled labor is considerably tighter in the less developed parts of Brazil (e.g., the Northeast) compared to the Sao Paulo industrial enclave. Furthermore, while the differentials between skilled and unskilled workers' earnings were declining when ex- pressed in percentages, the absolute difference between the two was increasing. An explanation of this discrep- ancy may lie in the theory of investment in human capi- tal.ll Specifically, the widening of the absolute wage differential between skilled and unskilled labor should stimulate investment by workers in their own training, 258 increasing thereby the supply of skilled labor relative to unskilled labor. Assuming that no change took place in the relative demand for labor between the two groups, the upward bias in the supply of skilled labor would have a persistent narrowing influence on the percentage dif- ferential. The evidence for several industrial countries points in the direction of decreasing occupational wage differentials, but for Brazil this conclusion must be qualified in the absence of additional evidence.12 FOOTNOTES: APPENDIX C 1Most of the following discussion of labor unions in Brazil is based on two sources: Robert J. Alexander, Labor Relations in Argentina, Brazil,,and Chile (New York: McGraw Hill Co., 1962); and Dale, op. cit. 2Alexander captured this reality in the following passage: "The freedom of action of both workers' and employers' organizations is seriously impeded. The ex- perience of the corporate state has left the government with a very large degree of control over the contacts between employers and their workers, a control which is exercised through the Ministry of Labor and the labor court system." Alexander, op. cit., p. 87. ’3 JIbid., p. 98. “As Alexander noted: "Strikes have not strength- ened the union's position and the number of collective contracts has been extremely limited. Even in cases where collective agreements are signed, they usually con- cern things which would otherwise be settled by dissidios colectivos——that is, wages and other monetary fringe benefits. There are few contracts of the sort which have become customary in the United States--documents ranging over the whole field of relations between the workers and the employer." Ibid., p. 94. 5Ibid., pp. 125-26, Table 6. Gordon and Grommers, on. cit., p. 118. 7 8Detailed evidence on this point is found in P. Gregory, "Evolution of Industrial Wages and Wage Policy in Brazil 1959-67," USAID/Brazil (September, 1968), Table 6. 9In addition, at times the increases in the minimum wage were illusory. In anticipation of wage increases businessmen would boost prices by more than their normal mark-up. This would have the effect of offsetting par- tially or completely any rise in workers' minimum wages See Table 0.3, p. 260. when they were finally ordered. See: Alexander, op. cit., p. 125. 259 260 TABLE C.3.-—Average monthly wage in Rio de Janeiro, ac- cording to activity and employees function, April, 1966 (1,000 cruzeiros). Branch of Adminis— Office Sales Various Un- activity trators personnel personnel other skilled functions workers Industry 100.3 111.5 79.9. 58.0 51.2 Commerce 205.3 84.5 71.9 53.1 ”6.8 Credit enterprises 305.8 127.3 45.0 66.8 66.“ Land trans- portation 123.3 63.3 n.a. 53.8 “5.0 Education, culture, communica— tions and publicity 3A8.3 90.7 196.7 79.8 U9.2 All activities 198.“ 102.8 73.6 58.3 50.6 Source: Lee, "Brazilian Exports of Manufactured Goods," Conjuctura Economica, XIII (May, 1966), an, Table VI. 10 Dale, op. cit., p. 125. 11See: Becker, op. cit., 66. 12This tendency has been observed by Rothbaum for the United States, France and Italy. See: Melvin Roth— baum, "An Interpretation of Wage Structure Changes in France, Italy and the United States from 1938 to 1952" (unpublished Ph. D. dissertation, Harvard University, 1952). APPENDIX D DATA APPENDIX 261 262 .moma .smmfl ..coauaz oopdca .Ammma .omwu "snow :uzv momfl .awma «ammfl «mama «auduuduaum coupe HacoauaCuoucm no xoonuau» .o:o«umz vouaco “oouzom nAmV . Ao~vg Hm.swo «0.000 mm.mcm o~.mmm «n.anm wm.~mo mo.mmm mm.mam pa.mmm mo.mam a~.~ma am.m~m uh.muo mo.mma p aoouuoo It.» 0 nu Hauoe 3:.aaha m:.mmma ms.m~za ma.wo=a ma.:a~a sm.~oq« om.mo~a hm.ammfl mm.msma Hm.amma mo.~m:a mm.m~=a am.aoma ~m.omma auhoaxm aauop aw.=~a -.amH mm.m> mm.~a mH.~m ms.m: mm.mm mn.mfl wq.oH mfl.mfl mm.Hm ho.o~ Hm.oa mn.oH m + s + o + m om.=o fl>.woa no.5m Fm.mm ma.fi~ om.H~ sq.aH mm.m wo.m am.oa Hq.mfl mm.m mo.m mo.~ h + o wo.mhma ow.MH=H sm.m0ma H:.=~ma wo.maaa ms.aOMH a~.mmaa am.:m~a mm.omHH mm.mmma mo.s~aa o~.H~ma Hm.~mma mm.om=a m + w + A m o mm.a am.oH m~.= ma.m ma.~ oo.~H Ho.~H mm.oa mm.ofl mm.o~ mm.s am.o mo.m om.~ ocux an avocado no: avooo o op.m mo.~ oo.~ Ha.” mm. ad.” a». ma. on. ma. mm. mm. am. mm. avooo oohsuuuuncu: .oual s Hm.mm mm.m~ -.mH mw.oH mo.~H m~.HH oo.m mH.~ mm.a ma.H oH.~ Hm.~ am.a mm. ucoanaawm uhonucdhb .oo:«soax o mo.ow mm.>s -.wm =m.ma ma.m mm.oa n=.~H an.» om.o m~.m Hm.oa aa.o mm.~ m~.~ nopsuoduncux ounce abo.MNo.~so.~ms o..ma mm.mn a~.- =~.m AN.” Hm.m ms.~ am. am. N=.= ”0.9 H:.~ .m.c .u.: done» a :MLH ew.o~ mm.=H mo.~H ow.oH s~.:a oo.o~ mm.m~ mo.m ma.» a=.s mm.» hm.om ms.o 5m.» nuaoaaozo . Hm.mm -.wa ma.oa om.mm os.~m w=.ma om.~m mm.mm mo.mm mm.om so.w~ =H.m~ oo.m~ mm.o~ pan .Hao oaoauomo> Haaacc m oo. Ho. oo.~ ~a.o ma.» om.m~ Hm.mH sm.wm Hm.flm mm.m mo. mo. oo. Ho. .oao.nauam Hague“: w sm.m~. mz.mam om.mmm m~.m~m om.mHm mo.mam mm.hmm om.a~m Am.mm~ mo.wmm mm.amm mm.mmm m=.mam Hm.om~ manna .Huxo n Hfiuho ud‘ Ova-LU ~ mm.m~ sm.wm ma.m~ am.a~ mo.am ofi.sm oo.m~ :o.mH ma.sa os.sa mm.o~ mm.m~ m=.mH mm.o~ oouanob a uomauo>om o o~.H-H mm.mmm mo.om¢ -.moo ms.Has sm.:mo mm.omo mm.mwo mm.aqm Mm.oaoH oo.mm~a mm.=moH Ho.oe~a m~.mamH aaqaac< o>aq a noon m~.mmma mm.momH aa.HomH ms.om- mfl.saofl ms.swfia am.omafl mm.mmo~ oo.oon m~.H=mH =o.mmma mH.on~ sm.=s=a so.~ma~ .HHV...ANV + “How oH.aMH mm.em~ aa.moH mm.mo oc.oo~ =o.mo m4.sm n~.ss mm.ow mm.Hs an.wa oo.om om.mm om.mm new a How flaw om.mm oo.mm om.- om.sm Hm.0m om.am ms.mm :a.mm mH.:m mm.H: sm.ma mo.m mm.m ma.o~ .ocoo .oLo “day: undo anchhvulcoz . ~m~ o~.ooH m¢.~oH =m.om mm.o~ o=.om :H.om am.mm os.m= :H.mm sm.m~ 4H.mm sm.mm mm.- ~m.- nouuuuceocoo .OLO noun ow mo.moa m~.mm~ Hm.msH mm.=mH am.o:fl os.smH mm.as ”H.2m mm.o: mH.H~ h~.maa mm.omd mo.o=~ aa.0m~ vane: nanny a moflauuoa thzuoauzcafllcoz new om.maa no.5o Hm.HHH no.sHH mo.mHH sm.NHH ms.aa mm.mm am.m~ ms.o= sm.om mo.om~ ~H.M- ws.HoH :Maaoo N. am o~.mm am.mm mo.s= mm.mm =m.mm mo.s: m~.m: ma.mm -.Hm mH.:w aw.mm ~a.mm am.mm cm.o= gouacoo coaacm Lanes; wa -.ma ma.- mm. H».@ :a.mH on.» Hm. mo.m mH.s mo.oa Ho.s mo.ma a~.m -.o uaocnoz .uusz .avoom ”do Ham ow.a~ mm.mH sm.oa on.» sm.m no.HH mo.mH mm.ma mw.m mm.m mz.m oa.m mm.ofl mm.ma coanuuucs .acaxm .nouaz ~ho so.fls mo.a= w~.ma mm.om so.fla mm.om ~m.mm om.:m mo.moa a~.flm am.m~ mm.~o~ ha.~=H oo.om amooo H o mm.mn~ mm.sos ao.mm~ m~.man sm.m:m mm.0Hs HF.NHB ao.mms Hm.~mm mm.mam mn.m~oa am.m=m mo.ozm -.mmoH oouuoo Hoe mm.ow ms.mm aa.mm mm.m~ om.mm Hm.mm sm.~m ss.~4 sm.~m sm.ma ow.a Hm.ma mm.~H Ha.- A ammom m umwaw ... o H o. o hm.fia am.mm Ho.m~ sm.mm mm.o~ mo.am m~.om ma.mm mm.Hm mo.mm mm.am 0H.oa -.=m wo.s~ ”canauouo> a adapt uncauaoauumnaao wmma mood awoa mmma mmma Hmma owma mmma mmmfl smma mmma mmma smog mmma osHm so nouuavoasoo .thdHHOU ho oCOdHHHE C.: nhdoz ndOuhd> .Codudoahdnndflo nvcndsihv 09H” :0 vondn Hdfldhfl ho nukOGKmII.H.Q Hams”. 263 mm.a Hm.ma ma.mm~ Ho.o~ mm.=a mm.HH os.~oa mH.oH wo.HmN mm.mom mm.~=s sm.zoa co.mm mo.HN am.m mm.s Ho.m sa.a aw.mmm wo.~m om.ama mm.H oa.ma wH.=~H mm.o~ ow.o~ am.m am.s> on.m~ om.mmm 2m.mma om.mo~ ==.mm~ H:.m= NH.ON Ho.m mm.» Ho.a wa.mom mm.sm oo.==a mm.H mm.ma ma.mza ma.mH m=.m mm.m mm.wH Ha.wmm mm.:HN Ho.:= sm.osH as.=a hH.:H HH.= mm.m m~.ma NN.H a=.wmm oo.mm m:.mHN mm.m mo.om s:.m~a om.m~ om.o~ NN.OH om.mom OH.HMN 35.:m om.opa Om.~o mo.mH mw.oH as.m mH.mH m~.m mm.wzm Ha.mm mm.msa oo.~ mm.wm m=.mma NH.>H mm.HH mm.a m=.mw mm.w oo.mcm mm.0mm mo.om om.=>a os.mw om.MH mm.~ om.mm mm.m:a na.mfi am.OH mm.m mm.mm 2:.m om.smm sw.azm mo.:OH no.5ma ma.ao z~.ma om.a oz.ma mm.am ms.m oz.mmH mo.ma om.oafl mH.a am.wm mm.mma m~.mm mo.oa mo.o mH.mo mm.~ H:.me mm.omm mm.mm~ mw.mHH mm.wo mH.HH om.m N®.NH mm.am om.m mm.mma mm.mH oo.HmH mm.H mm.mm mm.~HH m:.ma ma.o~ mm.o mm.mm NH.m mm.m~m Hm.mmm mm.mmH mw.mmH mm.mm HH.HH mm.H NH.mH mq.mm >:.N Hm.m~fl Hm.nH mm.HaH mH.m mm.om ao.mma m~.mm mw.~om MH.Hmm mm.szfi Nm.mma ma.n: Hm.HH mH.m nm.wfl m:.mH am.m am.mma HN.HH H4.mma on.» H~.m= om.:aH mm.mm mo.omm wo.oom mm.o:H mo.oHH mm.~o 00.:H Om.m mw.:m Nw.OH 3m.m om.wsfi mm.om m~.maa :m.mm am.m:H 3H.mH om.mH om.o Hm.=s wm.m mw.~om ss.mmm 05.5na No.00H mw.wm mm.ma am.m mm.mm 35.: Ha.~ sm.mma mm.ma mm.mma mm.om Hm.oma mH.oH ma.wa om.mm nm.m mz.me om.:mm ma.ssfi No.55 mo.mo mm.HH mo.m mo.mm mm.ma Nm.a vs.aza Hmuom aw.mwa mo.mm Ha.om m>.NNH m:.HH HH.mm Hm.HN pm.ms sm.mH wo.mwm H~.oom mm.wmm m~.m om.ow :H.mH mo.Hm :0.HH so.m mm.~mm mm.Hm Ho.:ma mo nm.ma aposuopm vouma 10m a noHOHuL< antenna .aoau unmm .cpaw oaauxoa «no mm.mm ohmonpoamm a Leann m mm.mm manuaeono Hem mm.m umpzuomuzcm: namnafiauumm Ham HH.~H nauseoum .ouo .HQCaoaco: mm Hw.m .mpmz mcapsoHoo a weaccme .m:«wzo Hm m:.mm mucsoaeoo a mucoEme Hmoaewzo a mo.m gum .aao maneumwu> .HmE«:< m wh.mmm nmvzm Hmpv:«z mm ma.mmm nuancOLm + Esoaouaom mmm om.mmm muoszLm ESMHOLuom Hmm ma. Edeaonovm ocspo m mw.mm mama; .Hoxv mamaneuw: eezpo pm N>.oa mucoum .ovLm a..uom.aaoo .Hoxe meLec«z a vaNHHHuLvm mcsuo mm Hm.m mama: gauge was meaauxoe Oopsuowuzcmslcoz Hmm mo.ma Locum ovum: . gaze Hmm ma. oqumcucmm ovzpo .Lmnnzm H as.m oooanoe a nowauc»vm o mo.omm naasfic< o>aq a goon mo mo.o: moanauomo> a paste mao.w:o.aqo om.mmfl mco«pwuaavum . a waaouoo mwma ammfl mmma mmma meH owmfl mmma mmma nmma mmma mmma amma uncapaoauamnoao mmma oaHm an uoauavoEEoo .Anuuaaoo ho uncaaflus adv undo» unoanu> .couuao«MHnano neomfi>ouv UBHm co nonan Hfiumkm uo nuhanHll.~.Q mqm<9 26L.l .omaa .ncoauaz vegan: "xpox zozv mesa «ammfl .ommfl «mmma .mmmH mafiumfiumpm mumps .Amwma .womfi .mmma .mmma Hmco«umcuoucH Mo xoonpmo» .n:o«umz vou«:D "Quezon H~.omza N:.mmoa m:.mme mm.mma~ mo.m~aa mo.om:a NH.mwaH mz.anma mm.mmma Hm.>om am.mm= wo.mmm bN.>Hm Hm.mom Hm.m wN.Hm om.m~ H=.m mH.om mN.Hmm mm.wa Hm.mm mm.» om.~m mo.om No.mm ma.» mm.Nm =a.mzm m=.Hm hm.NHH wo.~m H=.mH ~m.~oo mo.>~m mo.wma om.waw mz.mom mm.~ mm.mm e=.w o=.o Hw.HH om.==m H~.m as.~m mo.m :m.om mo.HN am.mH mm.m hh.mm mw.mma 0:.mm mm.mm um.sa ma.oa om.zmo oa.=zm mm.~om om.~mm ma.mwm HN.H ma.mm :~.m no.0 wN.:H m~.mom am.=H om.=m No.5H HN.H© mm.mm Hm.=H om.m oa.mm Ho.mmfi as.m~ om.om mm.am hm.oa mo.mmw Hw.w~: mm.omm Hm.wam Hw.mnm Hm.m ma.o: o=.m m:.o No.5H m=.mm= no.o= 25.0m om.Hm co.mm ms.mm mm.wa as.m mm.az oo.m:m Hm.on ao.mm ma.sm mo.aH mz.mom ma.n:m Hm.~mm mm.mmw wa.mmm mm.m Hh.wm mm.o mm.o Hz.wa m~.som om.mm OH.mm Hm.mm am.=m wo.mh mm.wH oo.w mo.sm sm.mma mm.a: m=.mm mm.om mn.ma mm.mmm Hm.mmm om.mmm mm.omm ma.wma mw.m mm.wm Ho.~ ao.m m~.wfi wo.mmm mo.mm mH.mm o~.mm OH.OOH m2.mm mm.mH an.ma mm.mm mm.sam mo.om :m.om mm.am mm.mH mo.mfim Ho.mam =m.mam om.oom mm.mmH ow. mo.mm mm.~ om.= mm.HH mm.Hmm am.mm m>.mm mm.oN mm.mm mm.mm mm.m mm.m ss.mm m~.mmm ms.ma mm.H: Na.mm mm. mm.mmm ma.mmm ao.afim mm.mmm mm.msfl ms. mm.mm m=.s No.2 mm.oa om.mom Hm.ma m~.moa ms.0m om.as mm.ms mo.m mm.m sm.mm HN.>HN mm.om mm.mm om.om mo.H wn.mnm mm.omm oa.mom am.wom mH.mmH an. mm.om m~.m 55.: mm.mH H~.mam :m.om ms.~mH aa.:z om.mm mo.m= NO.HH ao.o mo.mm mH.ooH Hm.mm nm.mm ma.mz mm. mm.smaa sm.~mm :a.mmm om.wam m~.o:m om.msa um.: mw.am 00.0H mm.m NN.NH m~.mmm mm.am NH.=OH mm.~= mo.~w mm.mm mw.m wm.oa mm.mm mm.mmm Hm.mm mm.m= mh.om as. o~.mmma mm.moma aa.mmma w=.mHMH nuhOQEH dance oo.msw mm.sos mm.mHOH m=.oms m + s + m + m am.smm mm.omm am.mmm Hm.mm: a + s mm.omm om.oma mm.zom Ho.mmm a + m + ~ + a + o om.mam mm.som wm.¢om mm.mmm w + a + o mm.HmH mm.maH ww.o=m om.mmm H m o ucax on unauLOO now coauanmaao no: nQOauommcwne a noauacoEEoo m mn.~m Nm.om mo.mm oo.mm noaoauhd consuumuscw: .onu: mam Leave: voucHLm mom m:.= ms.a om.m mm.a nuaaansm adamanmouwsocuo a canawhwouonm Hmm =s.m =m.oa m¢.ma mm.o nauauann< .nucvaahunca s ms.mom sm.oom m~.sma Ho.mo= pcoeqascm upon Insane .wo:«£odl ~:.H m:.H 05.: mm.m mH.HH mo.NH om.NH =m.m =m.m mm.m~ mm.~H Hm.MH uuaww»«< ao.mm os.mz mm.sm Hw.am muaoanu> youoxmmwom mo.om mo.om -.sa om.~m moaoazu> sammwam mm.wm ma.mm om.ms ms.mo moccaaaamw a msuahdnn< apocacoaz Haouuuooam mas oa.~a ma.Hm mm.zm MH.H~ zuocfizou: mauxhozuaauox nap mwcuzou: oouuuo was Ha.» mm.~ m~.o~ mm.m suocficoaz Hausuasouhw< Has oauuooacnco: shenacoa: nosom o ma.mmfl mH.me HH.HHm ms.mo~ mouspoausca: camam mo .w.m.z Have: «0 nousuoausca: mH.m sm.m am.m mo.m oo.mm ow.om hm.mm mm.wm mz.wm mm.Hm mm.~= m~.mm 8 so.o= Ho.Hm mm.so mm.m= mauve: ”sonaouucoz ass.mss.-s.flso ma.m: Hm.sw wm.moa ~m.om Hooum a scam 8 NN.H m~.m hm.ma om.wm m.m.z m0h:podh5:d: Hauocaz oaqqauoaucoz 2653 .Asoummmfl .esH ".o .o .COpmchmmzv soummma woe .mHo> .mpmpe do cofiuomgfla .ucmanam>mQ new coflposppmcoomm Low xcmm Hmcofipmcpmch new peak mmemcoz Hmcofipmcpwch "meadow o.:moa s.zs m.mma o.sm :.mmma H.0m w.zso :.mm: m.mmm swma o.azsa m.mw s.oma H.3QH s.ommfi o.H: :.mas 0.0m: m.:om mesa o.momH ©.mm m.mom 0.0m m.w:mH m.mm m.mso :.mH: H.m:m moma o.mmsa m.me m.oma H.mm m.mzfifi m.sm s.sflm o.msm o.mm: :mma o.mosH H.Ho s.mm m.mo o.m©HH m.Hm m.:mm s.mmm m.mmm mmmfi 0.:Hmfi m.:z s.ms m.ms s.sHoH H.9m H.mm: o.mmm s.oom moma o.mcsfl s.mw o.so :.ms o.omHH w.m: s.omm w.mam m.amm meH 0.0mmfi m.H: 0.0m o.ms m.oe0H n.am s.:m: H.03m m.omm coma o.m®mfi m.oma m.ms m.:m $.0HQH w.om o.omm .m.c :.mom mmma o.m:ma o.ms m.msa m.om m.:mo m.:m :.NH: .m.c o.m:m mmmfi m.HemH m.mm o.m:H o.m: m.HsHH m.sm o.mm: .w.c o.msm smoa w.mm:a w.om m.mofi m.:m n.ammfl m.sm m.:om .m.c o.mms omma m.mm:a H.Hm m.m:a m.mm m.omHH m.om o.mam .m.c s.ofim mmmfi m.HemH o.sm m.msa o.mm 0.0mma m.mm o.mmm .m.c m.mmm :mma m.ommH 0.0m H.0HH 0.2H m.:mma m.oq m.msm .m.c o.oss mmma o m A s m a m m a Asoemcflx pepflca mpmcmo wwwgpcsoo .Hocfiv ApHCSEEoo pcm HmflgpmSUCH mOfiLmE< maogsm odogsm OHEocoom mmumum kuoe Ico: mepo cfiumq spoumwm : + m + H cmomw :memoz cmmaopsm pmpfics mhmmm .Ampmflflop mo mCOHHHHEV moummma .mcoflomcfiomme do weeps :Hwe an Hammpm no mppoquul.m.m mqm .mempe mo coflpompfla .pcmeaoam>ma new cofiposgpmcoomm Lou xcmm Hmcofipmcgmch new pczm mpmpmcoz chofiumCprcH “meadow m.owoa o.HwH 5.0mm $.05 0.05HH m.om m.wmm >.mHm w.omm woma o.mo:H 2.0HH w.wmm H.ms s.zsoa m.z: w.mm: 5.5mm 2.50m coma o.smoa :.mm m.msm m.sm o.ssm m.sm o.aom m.mmH m.mmm moms o.mmma o.oQH m.cmm m.ow H.0mm o.mm m.Hmm w.oam O.Hm: :mma :.mm:a ®.mma :.mwm ©.mm m.waoa ~.Hw H.m~: m.wmm n.am: moma o.mm:a m.mma m.mmm H.mw C.::QH m.om ©.Hom H.mom m.mm: momfi c.0wza m.®za 0.0:H m.os w.mmofi m.os H.ms: m.m©m o.H:m mea c.rwqa w.m:H :.wmm H.mm w.mooa o.mm m.mom m.:®m H.0m: coma m.:~ma ©.mma m.mmm m.o: m.mmm m.wm m.Hm: N.mmm w.mm: omma m.mmMH o.mMH :.>zm w.mm :.z:m m.mm 0.0H: m.:mm H.Hom wmmfi w.mm:H m.HMH :.wmm a.wm w.mmofi m.mm m.mm: w.smm w.Hsm smma w.mmma :.CMH H.>mm m.oc m.Hom 0.0m :.w@m .m.c m.:mm mmma w.mOMH m.:Mfl m.omm m.mm o.Hmw H.mz 5.30: .m.c N.Hmm mmmfl 1.0me m.@:H :.mmm ©.0m m.HmmH m.om n.awm .m.c F.0mm :mma w.mama m.mmH m.:om m.om o.aww H.HH ®.mm: .m.: m.wo: mmma o m s m m z .m m H Asoeacfix UmpHC: mowcmo mmfigpcsoo . .Hocfiv zuHCSEEoo pew Hmfippmspcfi wowgme< moopzm maopzm QHEocoom newspm kuoe Ico: Locpo :Hpmq cgopmmm : + m + H smash :LmumoE cmmoopzm popHCD memo» Amgmfiaop do mcoHHHHSV swimmmfi .chHLo do mecca Cams >9 Hammgm mo mpLOQEHII.:.Q mqm<9 2637 .HHH .Ho> .Hmcodomz wagon .ooma we HmfipumsncH Oncmo .oucmEmrmcmomm on Hmcofiowz Oom>gmm .momH "20pm mum mmma Lou moapmapmuomnmno mapmsvcH .m.o new H.Q manme you mmopsom own muaoqsfi can mppoaxm pom “mohsom om. mo. ooo. mom. m.= m.Hmm mmooom oom, mommma ,noossq oz. om. moo. omo. m.m o.omo omomoom omooo mmammfl ooom oo. om. ooo. Hoo. o. . o.omm . omoooo --- moo wcaouoao oo.. mo. ooo. ooo. m.m o.omm Hooofiom mmom moomm mmaapxoe om. om. mmo. Hoo. m.m o.osm ooosomfl mmflom ommm mompmcfie OAHHmmencoz oo. oo. moo. moo. o.o o.omm . Hofiomm oamm mfimom mmwmgm>mm oo. oo. Hoo. ooo. m.m m.som omomfim mom Hmooa emopomq oo. oo. ooo. ooo. m.o m.ooo oomomo maoom Hom poomm om. oo.H ooo. Hoo. H.m m.mom mmoooo --- Hmm opsoacpsm oo.a oH.H ooo. ooo. m.H H.Hoo ooommm --- mono ooooooe om.H mo. ooo. mmo. m.m m.omo moommo .omom ooHoH mossopmm om.H oo.H moo. mmo. m.H m.on maooom mmmom oooo msoocoaflmomaz oo.H mo.H oHH. HHo. H.o o.omo mmomoom momooo mmooo Hagawpofiamumz om.H mm.H Hoo. moo. m.m H.oom moofioo mmmom oooH moaummam oo.H so.H om. ooo. o.m m.mom omomom mamom --- oscoopw o HmHLOpfiom om.m :N.H mom. oao. H.m m.owm mommmmz ommaom ammo: maMoapzwomEmeq .mHmOfiEmzo mo.a oo.m mam. mmo. o.m o.Hom mommoofi oofimmm oommm Hmoficmcom: om.m oo.H ooo. moo. o.m H.ooo omaooofi Hmooofl momma mcowomOH::EEOO.oofioauoomflm om.m mo.m moo. moo. . o.» o.mmoa oofloom momm oomm umooom oo.m mo.m ooo. ooo. o.m o.ooo ooaomom ooooofi Nomad acmeofioow upcomcmpe _ Azzxo HHH HH MMMMM mm ooooso o» mommwosm mommwo5m Aooofio Aoooflo Aoomao xmucfi xmpCa mppoqefi mpuoaxm umzoq .pwoom mwawm HHme Hafixm mo oaumm mo ofipmm empo: msflm> ho mzflm> muLOQEH mupoaxm asopw anumsUCH Lr ‘ .mOHpmp msompm> .Ammmav momumfinmuowpmno spamsocm 0cm mpuoaefi .muuoaxm n.afimmumuu.m.a mqm<9 268 .oomH .pmnEmomo .opHmcmo on on .mmmH mLHmHHompm HmeomsscH ommdoopm moHpmHumumm we HmcoHomz onHmmcoo .momH .mmmH pom .HHH .HO> nHmCOHOQZ 0““me HOQOH $6 HmHLUMSUCH OowD .OquEmmmcmomm MU HMCOHOGZ 00.2,.th nmme ammmH .Horm "oopsom . . . . .. .. .. . .. . mMHnm m H HH o H oo H m m H mom m c o c w c oH w c H mm mHmempme 0H» HWH WmH mo.H mm. m.m w.mom .m.c .m.: .m.: mo. .m.c H.Ho mmecwo w mamOm .mmesmpmm mHo moH mm.H oo.H o.o o.mso o.moo H.oom mo. oH. H.mom m.mm mowopm>mo oHH om om. oo. m.H H.omH m.oom m.ooH mo. mm. m.mHm m.mm moomcoHHoomHz omH moH om.m mo.H o.m m.Hmo o.mHmH o.mmm oo. mH. m.mmm m.ooH mHmemome QOHuoopoomcmpe soH ooH oH.H os. m.m m.Hmm m.oooH o.mom mm. oH. m.Hmm o.mo mcoHomOHcoe500 o oHHOHpooon om mo om.H ms. o.m m.oom m.ooo m.HoH om. mm. o.mmm m.Ho HmoHcmoooz om on MH.H HH.H H.o o.som m.mom m.ooH mm. om. o.oom m.oo . OHmwsHHmooz os ms oo. oo. m.m m.moH m.moo m.omH mo. mm. H.omH H.om mHowmcHe OHHHmome-coz mo moH oo. om. m.m o.oaH o.mo, m.mmH mm. mm. H.moH m.mo moooompo LmHHEHm a Locomoa muss HHH mo om.H oo.m o.m m.Hmo m.ossH m.Hsm mo. mo. m.oom m.oo gmooom omH oHH om. m. m.H H.o4m m.omo o.uoH om. mm. o.omm m.oo ochotw o HoHLOoHom mo so oH.H om.m m.o o.Hom o.mmoH o. mm mm. mH. H.osm m.om oemooopmo o gmomm mHm ;mH oo. Hm. H.m m oHH m.oso m..< mm. mm. m.ooH m.oo moooHcpom om om mm. mm.H m.o m.ooH H.roo m.moH oo. mm. m.mmH m.mm gmoEoH mHm omm mm. oH. o. o.ms H.Hoo m.mm mm. mm. m.mmH m.mm mmHOHopo coOHo o mgozoooo .mcHsooHo mmm moH Ho. oo. m.m o.mmH m.ooo H.co om. om. o.moH m.mo mmHHoxme oom soH so. mm. m.H H.HHm o.soHH o.moo mo. HH. o.mmo m.Hs coooooe om mo mm.m oH.m H.o A.MHo m.oomH m.ooo .o.c ooH. s.mom H.mo mHoOHoomooELoco o mHmOHEmco mm on om.H oo.H m.m o.mmm o.:ooH o.mom mm. MH. m.HHm m.oo mpoooopo ooom .o.c .m.c oo.H oo.H s.m m.msm H.3o. m.o H om. oH. o.omm s.mm obomoocH coHooeLoomcope omHH ompH omoH omoH omoH omoH omoH mmmH dwv 1V 3 S de. MUD wdA MEN OQN Qua m.HuA 9V JA BA H. M. BJO OIoJ 9J8 00.0 npO 08 Ed? BA 08 J8 I. I. UOJ JJO UO.L JpU QPU H4 UOI J8 1d Trd I .L _D.S M. S ....vn He. .08. AI. _D.n UJ 99 38 .L I “Anna 0108 (Anna 8pm“ RPM 90 Ana TrB 31o tab 9 0 .us 0H0 .u e 1 e .L a 0 ”up zpumsch «+3 8 J J 91d 03 B30 UU due no 918 88 I. E E JIO UPC JIJ wee 88 81. J_D. S 0.4 U 1 1 OM end 0 EJ J MD. .8 UO 0 I. I. U8, UOI. U0 U A A BM 08 U 3 w 0 O J .11 n .08 Up. 0.? JD. U B I. M ..A IE M1 AAJI UT. 0.3 v.1 n I U I I 0.0 BOT. 0d eon In 89 9d 9 B I .1 J8 BU Jn 80.8 18 D. J6 I I I Y1?“ J E val Jn 0 SJ 8 S a 0 O O J 8 JG .4 J S 3 a 3 1 J S Hucoo Loo CH0 mmmH :oHpomuOLQ .mmHmesch om .AmogHomzpo pcmmsoca :HV moHpmHumpm mpumsch Lmzpo pcm .Umppm msHm> .mowmzll.m.o mqmuonmha :uHm m.mmm m.omm .Hm o.oom mm.o om. oH. m.oom m.omm m.mom 5.3m: m.oo 5.m: couscoaa spHan .a.c m.HOH .m.c 0.m mm.H Hm. om. H.Hm m.Hw m.Hmm m.mHH 5.0HH 0.HHH naomno> mo panama a noduoauuncoo H.mmH m.:0H .m5 0.5mH o:.: H:. HH. m.oo5 0.:m: m.m00 5.m0m 0.00 0.05 omccho a ooooHn .mmHu anm> Locos no :oHumoannmm m.HOH m.Hoa .m :.m m5. 5m. H:. 0.05 5.mm 5.mmH m.m0H :.m5 m.om pumLoLHm ho mango; a :oHposhuucoo 0.0ma 0.0mH .00 0.00 m0.m :m. mm. 0.0HH 5.m: :.mwa m.Hoa 5.0m m.H: mamsuca so ooHHoo on 0» ooHoHso> 0.00H 0.00H .omH 0.0mH 5m.m :m. mm. m.mmm m.m5H 0.m5m w.m0m H.5w H.00 mmHOH£o> LOuOE how mwmcom m.oma 5.0:H .00 0.00 mm.m mm. 0H. :.:mm :.5Hm m.mm: 0.0mm m.mm H.50 Hmdhmpms Havahuomam m.omH :.ooH .mo o.oo Hm.m Hm. oH. H.mmm m.Hom m.os: m.osm m.oo :.H5 anomamaoa HmoHuuoon H.mma :.:mH .om 0.mm m:.m mm. mm. :.omH 0.0mH H.mmm 0.00H H.05 o.mm hhumacCH Hahn; a chauasougmw Lou hLoCH20m: o.MMH o.moH .mo o.Ho Hm.oH 5m. oH. m.osm o.ssH m.Ho: m.mom m.m5 m.Ho :oHomawawpq ohmonupmo .Lmnam m.m5H m.mma .00 0.05 .w.: mm. mm. m.m0m 5.0MH m.omm 5.m0m 0.5m 0.0m unmonhoamn mo mmaoaphw Umcmnaoo 0.:0H :.NmH .00 0.H5 m5.m mm. mm. H.5mm H.Hwa 0.0mm H.H:m 0.05 0.0m ohm: Haumhho a opmxwmwaw .mmaH0 0.0ma 0.0ma .:m 0.:m mm.5H mm. 00. m.Hm5 5.0m: m.mmw H.H5m m.05 0.0m ucwsvo m.m0m 0.5mm .50 0.55 ::.m 5:. mm. m.0:H :.:0H m.mHm H.:0H 0.00 0.5: Hwahvpms oaadhoo 5.~5m :.05m .50 0.m0H 05.m m:. Hm. 0.mHm m.5mH 0.00m 0.mmm 5.m0 :.0m mHmuwCHE oHHHmuoslco: ho wcmanCHu a mu0300ha szuo H.00N H.00m .00 0.00 00.H 0m. Hm. 0.Hmm 0.H5H :.MHm m.m:m H.m0 0.0m acmEmo ”soundu a oHumeQ .ucmewo mo mehsu losppm a mucwEdCLo .mmooam m.0ma m.oma .0: 0.0: m0. 0:. :m. 0.00H H.mm m.HmH 0.0mH 0.5m 0.5m oono Eduoamo H.0m H.0m o. o. mm.H 0:. 0H. 5.Hmm m.Hmm m.moo m.mom m.om 5.Hm mnwawamzv: oo wcHocha a mcHuHcm H.0m H.0m .0 0. mm.H m:. 5m. 0.05m m.mmm 0.0:: 0.:0: H.~HH H.00H mawoauofihva 90:00 oo wcHucHua a wcHoHcm :.05 :.m5 .0 0. 0H.H mm. mm. :.00H 0.0m m.m0m m.:0H 0.m5 m.om mawapwuwa ofiumwaozom a Hafiopoesoo mo wcHucaum o.mo m.oo .mo o.oo om. Ho. Hm. m.oom o.moH m.moo o.mom H.mo o.m5 mmoH>pom oHnoupm pmzuo o.mm o.mm .m o.m oH. mo. om. m.moH m.moH m.oom m.Hom H.mo H.om wanna go wcHucHua can wcHuHom h 9m M». m m m.“ m m $2 $2 mm? $3 $3 mm? mm? 3.2 0mg 8 M O J S o J I. J m a U. TI. U. Tron .....d wd H O B N 1. 8 3.1.8 Jwa Jq JI 9J0 JPO 08 mom .5 Jo .... ...mo 0 o: A o 2.2... Nm. 8 a a D. m M m a W PM W. o .3333 coapozo pmvzlcwe 93733 de ad a a 9 3d ddB n o no.8 .Hoa venom coHposcopq 35930 a w m w a m .sm m mea a J usHo> amazncoz pea 0000a wSHw> Hascca owapo>< zuumsucH A m J m J on 0 a A 8 M a.» a.» a a J a n e 0.9 J n 3“ 3 X. U T. 0.3 . m 8 a 3.0 In 8 8 S S J 3 38 D. .. o momoHv cOHooouopm w .Apcoo non 2H0 coHpoououn .HuopHousno ucuusonp 2H0 noHuquwpu apau:0:« nonuo and wound o=Ha> .uowazan.5.0 mqmHoHHuoa HmHmwsoocH.omoooopm «OHuoHoopom co HacoHomz ozHoucoo .momH .omoH pom .HHH .Ho> .Hacououz ofiuom .000H :0 HwauposucH cacao .OpcoEmmucoomm on Hacouowz oom>hom .momH .0000 pom “oopsom H.:0N 0.0NH 0.00m 0.5NH 0.0NH 0.:HH 5.5HH m.mom H.0Hm H.00m :.00H 0.00m 0.50H 0.00m 0.0NH 0.0HH .M.C .N.E .Q.C .Q.C 0.0:N 5.00m 0.55m =.oom m.on 5.00m 5.0om H.:0m m.H:H 0.0mm 0.0:H H.NNH 0.:HH :.:mH 5.00m 0.00m 0.:5m N.50H 0.00m 0.0NH 0.00m 0.0mH 0.NHH 5.00m 5.00m 0.00m 5.00m 0.50m .w.c 0.H5m 0.50m m.00m 5.00m 0.05m 0.00m 5.00m 5.00m 5.000 5.000 H.00m 0.00m .0mH .mm .00 .50 .mm .om .:m .:HH .0m .50H .00 .05 .H0 .0: .00 .HN .M.C .m.C .m.C .M.C .00 .omH .:0H .0:H .0:a .omH .00H .00 .omH .omH .oma .00 .OHH .0HH 0.0NH 0.00 0.00 0.00 0.00 0.0m 0.00 0.0HH 0.00 0.00H 0.H0 0.50 :0. 00.0 50.a 00.H 0N.m m5.H 0N.H 0H.H 00.H 05.0 0:.m oo.m 00.5 0m.0H 00.H :0. H0.m oo.m 55.0 00.m 00.0 00.H 00.0 00.0 H0.m m0.m :0. 00.H 00.5 00.0 00.H 0H. H0. 0m. 0:. 50. :0. H0. 0:. 0:. 00. 00. 0m. 00. 00. 00. :0. N0. 50. :0. :0. 00. 50. :0. mm. :0. Ho. o0. mm. mm. mm. H:. mm. 0H. H0. mm. m.HOH :m. 0.HON mm. 0.0HN H0. 0.00H 5m. 0.00H 0H. 0.50: 5m. 0.00H mm. 0.00m 0m. 0.0:H 0m. :.:0H 0N. N.50H 0H. H.00: NH. N.0:: 00. 0.000 mm. 0.000 5H. 0.00: 00. 0.00H 0H. :.00m 0m. N.00N H0. 0.0HH Hm. 0.0Hm H0. m.00H 0m. m.HmH 0H. 0.000 0H. m.mm0 5H. :.00: 50. 0.00 50. 5.0HH mm. 0.50m 0m. 0.00H mm. :.05H 0H. m.H0m 5d. :.50m 0H. 0.0mm 0.HNH 5.00m N.5Hm m.00H 0.50 0.:0H 0.00H 0.N5H H.05 0.00 0.:0N 0.000 0.0m: o.o0m m.mmm m.om: H.00 m.Hmm 5.05H 5.05 0.00H 0.05 0.00 0.0mm 0.0HH 0.000 m.m0 m.oaa 0.00H 0.H0 0.0HH H.00H 0.00H 0.:om m.m0H m.omm 0.:H0 :.0Hm :.00m H.5:0 0.:m0 m.oam H.mmm 0.00m H.:mm 0.050 0.000 0.0m5 0.5m: 0.mm0 0.05m 0.000 0.500 0.:0H 0.00m :.55H 5.:mm o.mm: 5.00: :.00H 0.000 H.00N 0.00m 0.05m 0.m:m m.mom :.00m 5.00H m.ooH m.oom m.H00 H.000 m.om: H.0m0 H.m00 5.55H o.mmm o.mom m.ooH 0.00m 0.0HH 0.0:H H.00m m.mom :.H5: 5.HHH o.HoH m.omH m.ooH o.0oH 0.mom o.omm 5.500 0.00 :.00 m.H5 0.50 :.05 H.00 m.oo 0.00 0.00 H.00 H.00 m.50 0.05 0.00 5.00 N.N5H 0.00 5.m0 m.50 0.00 H.00 5.:0 0.00 0.m: H.00 0.00 0.00 0.05 0.:0 0.00 0.50 0.00 0.:0 0.0: 0.00 :.00 :.00 0.m0 0.00 0.0m m.00 0.0m :.5: m.5: :.m: 0.0: 0.m0 0.m0 H.00 0.0NH 0.00 :.00 OHN 5502:0100 LOLDH mam :r:Ln-:r5:rm :mm ‘00“! MMCNNO‘M 000 :.00 ooomnoa namuma usoahoulcoc no amasaaauu: mcHnaa a :H» no couuo:00pn .m:«nam»m nvhoooh 0 nucmazhun:« Havana: unammos :oamHouha a on: Hmoacnomu a HmcoaanMOLQ pom uwHoaupm a mucoESLancH ucmsafisoo amazon .oacmaonuho .HmUHMH:m HmHLMuma owzamhwouoca a 00090 coaummhomh a masons Lou moaofiupa a @009 ahamrmn mace .mEOOhn .moznshm mpmxompoohdh .mmnouma .mm>HmoHaxw .mhmclom muosUOLQ maofiuahusc .Hoxm mama .mmxmx .mHHo mwufioomcau a movoofiuommCH .mpCMuooQCHnHQ mmaocwo a madam .mmfishhom HonHu HmHoameLm a wouumwam mocsoasoo 0 mucmemam "meofiEmzo upsuHchsu .omH: anwamaocas ohsuHCLsu Haum: Ammonau .mUOL .Uoozv wus»«:nsm wcme a mpsm mo coauwhwnmhm mpommmppw 000: pwzuo @003 teamHCHh no mohsuozpum a mmovfim mmcHz Hocooaa mo couuwaadumHo mnwwm :0pu00 ho wea>woz H003 no wcd>awz wcHnuoao 00 weapommooo< moonm Lemony .Hoxm mwonm no coproappdm mum: no cofipaoaunmm hamsm paw xnoq a name «0 mw>pomoun .muau mHaEHc< neocoapaccoo .mm> a moqum .moHnuuomv> a nuashh no mo>pwmohm BIBLIOGRAPHY 271 BIBLIOGRAPHY Books Alexander, Robert J. Labor Relations in Argentina, Brazil and Chile. New York: McGraw Hill Company, 1962. Baer, W. Industrial and Economic Develppment in Brazil. Homewood, Ill.: Richard D. Irwin, 1965. Bain, Joe S. International Comparison of Industrial Structure. New Haven: Yale University Press, 1966. Balassa, B. Trade Prospects for Develpping Countries. Homewood, Ill.: Richard D. Irwin, 196“. Harlow, Frank D., Jr. Cotton in South America: Produc- tion, Marketing; Consumption and Development in the Textile Industpy. Memphis, Tenn.: Cotton Council of America, December, 1952. Becker, Gary S. Human Capital: A Theoretical and Empirical Analysis, with Special Reference to Education. National Bureau of Economic Research. New York: Columbia University Press, 1964. Bowen, W. G. "Inter-industry Variations in the Unemploy- ment Wage Relationship." Wage Behavior in the Postwar Period: An Empirical Analysis. Princeton, N. J.: Princeton University, 1960. Brazilian Embassy. Survey of the Brazilian Economy 1966. Washington, D. C.: Brazilian Embassy, December, 1967. Cairnes, J. E. Some Leading Principles of Political Economy Newly Expounded. New York: Harper and Bros.,187u. Caves, R. E. Trade and Economic Structure: Models and Methods. Cambridge, Mass.: Harvard University Press, 1960. 272 273 Committee for Economic Development. Trade Negotiations for a Better Free World Economy: A Statement on National Policy by the Research and Policy Committee. New York: Committee for Economic Development, 196U. Correa, H. The Economies of Human Resources. Amsterdam: North—Holland Publishing Company, 1963. - Dale, W. B. Brazil/Factors Affecting Foreign Investment. Menlo Park, Calif.: International Industrial Development Center, Stanford Research Institute, September, 1958. Diab, M. A. The United States Capital Position and the Structure of Its Foreign Trade. Amsterdam: North- Holland Publishing Company, 1965. DunIOp, John. "Productivity and the Wage Structure." Income; Employment and Public Policy: Essays in Honor of Alvin H. Hansen. Nemeork: W. W. Norton, 19D8. Ford, J. L. The Ohlin-Heckscher Theopy of the Basis and Effects of CommodityiTrade. New York: Asia Pub- lishing House, 1965. Friedman, Milton. Price Theory, A Provisional Text. Chicago: Aldine Publishing Company, 1962. Gordon, L.,and Grommers, E. L. United States Manufac— turing Investment in Brazil: Impact of Brazilian Government Policies l9U6él960. Cambridge, Mass.: Harvard University Press, 1962. lreenwald, W. I. Statistics for Economics. Columbus, Ohio: Charles E. Merrill Publishers, 1963. Gregory, Peter. Industrial Wages in Chile. Ithaca, N.Y.: Cornell University Press. Haberler, Gottfried. Theory of International Trade: With Its Application to Commercial Polipy. London: W. Hodge and Company, 193675 Harbison, F., and Meyers, A. Education, Manpower; and Economic Growth: Strategies of Human Resource Development. New York: McGraw Hill, 1964. Havighurst and Moreira. Society and Education in Brazil. Pittsburgh: University of Pittsburgh Press, 1965. 27“ International Monetary Fund. Ninth Annual Report on Exchange Restrictions, 1958. Washington, D. C.: International Monetary Fund. May, 1958. Johnson, H. "The Theory of Tariff Structure with Special Reference to World Trade and Development." Trade and Development. Geneva: 1965. Johnston, J. Econometric Methods. New York: McGraw- Hill, 1963. ' Keesing, Donald B. "Labor Skills and the Structure of Trade in Manufactures." The Open Economy: Essqyg on International Trade and Finance. Edited by Peter B. Kenen and Roger Lawrence. Columbia Studies in Economics I. New York: Columbia University Press, 1968. Kreinin, M. E. Alternative Commercial Policies: Their Effect on the American Economy, Institute for International Business and Economics Studies. East Lansing, Mich.: Michigan State University, 1967. Lary, H. B. Imports of Manufactures from Less Developed Countries. National Bureau of Economic Research. New York: Columbia University Press, 1968. Leff, N. H. The Brazilian Capital Goods Industry 1929- 196”. Cambridge, Mass.: Harvard University Press, 968. Leff, N. H. Economic Policy-Making and Development in Brazil 19A7-l96h. New York: John Wiley and Sons, 1968. Leftwitch, R. H. The Price system and Resource Alloca- tion. Ath ed. Hinsdale, Ill.: The Dryden Press, 1970. Linnemann, H. An Econometric Study of International Trade Flows. Amsterdam: North-Holland Publishing Company, 1966. Lourie, Sylvian. "Education for Today or Yesterday." Problems and Strategies of Educational Planning. Edited by Raymond F. Lyons. Geneva: UNESCO, 1965. Malinvaud, E. Statistical Methods of Econometrics. Translated by A. Silvey. Chicago: Rand-McNally, 1966. r.) 51 m Meier, G. M. The International Economics of Development: Theory and Development. New York: Harper & Row, Publishers, 1968. Michaely, M. Concentration in International Trade. Amsterdam: North-Holland Publishing Company, 1962. Mill, J. S. Essays on Some Unsettled Questions in Poli- tical Economy. London: University of London, 19A8. Minhas, B. S. An International Comparison of Factor Costs and Factor Use. Amsterdam: North-Holland Publishing Company, 1963. Mundell, R. A. International Economics. New York: The MacMillan Company, 1968. Organization for Economic Cooperation and Development. Wages and Labour Mobility. Paris: OECD, July, 1965. Reder, Melvin. "Wage Differentials: Theory and Measure- ment." Aspects of Labor Economics. National Bureau for Economic Research. Princeton, N. J.: Princeton University Press, 1962. Ricardo, David. The Principles of Political Economy and Taxation. Everyman's Library Edition. New York: E. P. Dutton and Co., 1911. Saunders, John V. D. "Education and Modernization in Brazil." The Shapingiof Modern Brazil. Edited by Eric N. Boklanoff. Baton Rouge, La.: Louisiana State University Press, 1969. Shearer, John C. High-Level Mappower in Overseas Sub- sidiaries: Experience in Brazil and Mexico. Princeton, N. J.: Princeton University Press, 1960. Siegel, S. Nonparametric Statistics for the Behavioral Sciences. New York: McGraw-Hill, 1965. Stein, Stanley J. The Brazilian Cotton Manufacture: Textile Enterprises in an Underdeveloped Area, 1850—1950. Cambridge, Mass.: Harvard University Press, 1957. Stigler, George. The Theopy of Price. New York: MacMillan, 1952. 276 Suits, Daniel B. Statistics: An Introduction to Quanti- tative Economic Research. Chicago: Rand McNally Co., 1963. Taussig, F. W. International Trade. New York: MacMillan Co. TOpping, K. F. Comparative Tariffs and Trade. New York: Committee for Economic Development, March, 1963. United Nations. Commodity Trade Statistics 196“ and 1965. Series D, Vols. XIV and XV. New York: United Nations, 1965 and 1966. United Nations. The Economic Development of Latin America in the Postwar Period. New York: United Nations, 196A. - United Nations. Proceedings of the United Nations Con— ference on Trade and Development. Vol. II. Policy Statement. New York: United Nations, 1964. United Nations Statistical Office. Yearbook of Inter- national Statistics. New York: United Nations, various years. United Nations Commission for Europe, Committee on Devel- opment of Trade. Report on the Special Meeting on the Organization and Techniques of Foreign Trade, June 29-July 3, 1959. Geneva: United Nations, 1959. United Nations Economic Commission for Latin America. Economic Survey of Latin America [Vols.] 196A, 1965, 1966. New York: United Nations, 1966,1967, 1968. Manufacture of Industrial Machinery and Equipment in Latin America. Vol. I: Basic Equip- ment in Brazil. New York: United Nations, 1963. . The Process of Industrial Development in Latin America. New York: United Nations, 1966. The Textile Industry in Latin America. Vol. 11: Brazil. New York: United Nations, 1963. United Nations Economic Commission for Latin America and The Food and Agricultural Organization. Coffee in Latin America. Vol. I: Columbia and El Salvador Vol. II: Brazil, State of Sao Paqu. New York; United Nations, 1958 and 1960 277 United Nations Food and Agricultural Organization. Agricultural Commodities--Projection for 1975 and 1985. Home: FAQ, 1967. United States Bureau of the Census. Annual Survey of Manufactures 1959 and 1960. Washington, D. C.: United States Government Printing Office, 1962. U. 8. Congress. The Brain Drain into the United States of ScientistpLEngineerpJ and Physicians. Washington, D. C.: A Staff Study for the Research and Technical Programs Subcommittee of the Committee on Government Operations, 1967. Vaccara, Beatrice N. Employment and Output in Protected Manufacturing Industries. Washington, D. C.: The Brookings Institute, 1960. Vanek, J. The Natural Resource Content of United States Foreign Trade 1870—1955. Cambridge, Mass.: The M.I.T. Press, 1963. Viner, Jacob. Studies in the Theory of International Trade. New York: Harper and Bros., 1937. Waehrer, Helen. "Wage Rates, Labor Skills and United States Foreign Trade." The Open Economy: Essays on International Trade and Finance. Edited by Peter B. Kenen and Roger Lawrence. Columbia Studies in Economics 1. New York: Columbia Uni- versity Press, 1968. Wielcizer, V. D. Coffee, Tea and Cocoa. Stanford, Calif.: Stanford University Press, 1951. Yahr, Merle 1. "Human Capital and Factor Substitution in the CES Production Function." The Open Economy: Essays on International Trade and Finance. Edited by Peter B. Kenen and Roger Lawrence. Columbia Studies in Economics 1. New York: Columbia Uni- versity Press, 1968. Yeager, L. B. International Monetary Relations: Theopy, History and Policy. New York: Harper and Row Publishers, 1966. 278 Periodicals Araoz, A. B. "Manpower and Employment in Brazil." International Labour Review, XCIII (April, 1966), A18. Baer, W., and Kerstenetsky, 1. "Import Substitution and Industrialization in Brazil." American Economic Association Papers and Proceedingp, LIV (May, 196M), All-25. Balassa, B. "Tariff Protection in Industrial Countries: An Evaluation." Journal of Political Economy, LXXIII (December, 1965), 573—94. Baldwin, Robert E. "The Effect of Tariffs on International and Domestic Prices." Quarterly Journal of Econo- mics, LXXIV (February, 1960) 65-7 . Ball, David S. "United States Effective Tariffs and Labor's Share." Journal of Political Economy, LXXV (April, 19677I183—87. Basevi, G. "The U. S. Tariff Structure: Estimates of Effective Rates of Protection of U. S. Industries and Industrial Labor." Review of Economics and Statistics, XLVII (May, 1966), 1u7-60. Becker, Gary. "Investment in Human Capital: A Theoretical Analysis." Journal of Political Economy, LXX (October, 1962), 9—u9. Behrman, Richard. "Cocoa: A Study of Demand Elasticities in the Five Leading Consuming Countries, 1950-1961." Journal of Farm Economics, XLVII (May, 1965), H10-l7. Bergsman, J., and Morley, S. "Import Constraints and Development: Causes of the Recent Decline of the Brazilian Economic Growth: A Comment." The Review of Economics and Statistics, LI (February, 1969) 101-02. Besen, Stanley M. "Education and Productivity in U. S. Manufacturing: Some Cross-Section Evidence." Journal of Political Economy, LXXVI (May/June, Bhagwati, J., and Johnson, H. G. "A Generalized Theory of the Effects of Tariffs on the Terms of Trade." Oxford Economic Papers, XIII (October, 1961), 225-53- 279 Bharadwaj, R. "Factor Proportions and the Structure of lndo-U.S. Trade." Indian Economics Journal, X (October, 1962), 105-I61 Brown, David G. "Expected Ability to Pay and Inter- industry Wage Structure in Manufacturing." Industrial and Labor Relations Review, XVI (October, 1962), D5—62. Cohen, B. I. "The Less Developed Countries' Exports of Primary Products." Economic Journal, LXXVII (June, 1968), 33A-U3. Corden, W. M. "The Structure of a Tariff System and the Effective Protective Rate." Journal of Political Economy, LXXIV (June, 1966), 221-37. Daly, Rex F. "Coffee Consumption and Prices in the United States," Agricultural Economics Research, X (July, 1958), 61-71. Eisemann, D. M. "Interindustry Wage Changes 1939-47." Review of Economics and Statistics, XXXVIII (November, 1956), HA5. Garbarino, Joseph. "A Theory of Interindustry Wage Structure Variation." Quarterly Journal of Eco- nomics, LXIV (May, 1950), 282-305. Gregory, Peter. "Evolution of Industrial Wages and Wage Policy in Brazil 1959—1967." USAID/Brazil (September, 1968). Hirschman, A. D. "The Political Economy of Import- Substituting Industrialization in Latin America." Quarterly Journal of Economics, LXXXII (February, 1968), 13-17. Huddle, Donald. "Furtado on Exchange Control and Devel- opment: An Evaluation and Reinterpretation of the Brazilian Case." Economic Development and Cultural Change, XV (April, 1967), 269-85. International Monetary Fund. International Financial Statistics, VIII (February, 1955). International Monetary Fund. International Financial Statistics, XXIII (February, 1970). International Monetary Fund and International Bank for . Reconstruction and Development. Direction of Trade (June, 1968 and November, 1969). 280 Johnson, Harry. "Tariffs and Economics Development: Some Theoretical Issues." Journal of Development Studies, I (October, 196“), 3-30. Jones, R. W. "Tariffs and Trade in General Equilibrium: Comment." American Economic Review, LIX (June, 1969), AIS—ifi. Keesing, D. "Labor Skills and International Trade: Evaluating Many Trade Flows with a Single Mea- suring Device." Review of Economics and Statis- tics, XLVII (August, 1965), 287-9“. "Labor Skills and Comparative Advantage." American Economic Association Papers and Proceedings, LVI (May, 1966), 299-58. Kenen, P. B. "Nature, Capital and Trade." Journal of Political Economy, LXXIII (October, 1965), 337-60. Kerr, Clark. "The Short Run Behavior of Physical Pro- ductivity and Average Hourly Earnings." Review of Economics and Statistics, XXXI (November, 19U9Y: 299-309. Koutsoyannis, A. P. "Demand Function for Tobacco." The Manchester School of Economics and Social Studies, XXXI (January, 19637, 1—20. Kravis, I. "Wages and Foreign Trade." Review of Eco— nomics and Statistics, XXXVIII (February, 19567, 15—30. "Availability and Other Influences on the Commodity Composition of Trade." Journal of Poli- tical Economy, LXIV (April, 1956), lU3-55. Kreinin, M. E. "Comparative Labor Effectiveness and the Leontief Scarce—Factor Paradox." American Economic Review, LV (March, 1965), 131-40. Krueger, Anne O. "Factor Endowments and Per Capita In- come Differences Among Countries." Economic Journal, LXXVIII (September, 1968), 6Ul-59. Kuh, E. "A Productivity Theory of Wage Levels—~An Alter- native to the Phillips Curve." The Review of Economic Studies, XXXIV (October, 1967), 333—60. Lancaster, K lvin. "Protection and Real Wages: A Re- statement," Economic Journal, LXVII (June, 1957), 199—210. Lee. 281 "Brazilian Exports of Manufactured Goods." Conjuntura Economica, XIII (May, 1966), 45-50. Leff, N. H. "Export Stagnation and Autarckic Develop- ment in Brazil, 1947-1962." Quarterlprournal of Economics, LXXXI (May, 1967), 286—301. "Import Constraints and Development: Causes of the Recent Decline of Brazilian Economic Growth." Review of Economics and Statistics, XLIX (November, 1967), ugh—501. Leith, J. C. "Substitution and Supply Elasticities in Calculating the Effective Protective Tariff." Quarterly Journal of Economics, LXXXII (November, 1968), 588—601. Leontief, W. "Domestic Production and Foreign Trade: the American Capital Position Re-examined." Proceedings of the American Philospphical SocietyJ XCVII (September, 1953), 9—38. "Factor Proportions and the Structure of American Trade: Further Theoretical and Empirical Analysis." Review of Economics and Statistics, XXXVIII (November, 1956), 386-407. Lerner, A. P. "The Symmetry Between Import and Export Taxes." Economica, III (August, 1936), 306-13. Lovasy, G., and Boissonneault, L. "The International Coffee Market." International Monetary Fund Staff Papers (November, 1963), 378-80. Macario, Santiago. "Protectionism and Industrialization in Latin America." Economic Bulletin for Latin America, IX (March, 19633, 61-101. Metzler, L. A. "Tariffs, the Terms of Trade and the Distribution of National Income." Journal of Political Economy, LVII (February, 1939?, 1-29. Michaely, M. "Factor Proportions in International Trade: Current State of the Theory." Kyklos, XVII (Fasc. A, 196“), 529—“9. Munk, Bernard. "The Welfare Costs of Content Protection: the Automotive Industry in Latin America." Journal of Political Economy, LXXVII (January/February, 1969) 3 85-98. 282 Myers, F., and Bowlby, R. L. "The Interindustry Wage Structure and Productivity." Industrial and Labor Relations Review, VII (October, 1953), 93-102. Perlman, Richard. "Value Productivity and the Inter- industry Wage Structure." Industrial and Labor Relations Review, X (October, 19567, 26-39. Reddaway, W. B. "Some Effects of the Australian Tariff." Economic Record, XIII (June, 1937), 20—30. Rippe, R. D. "Wages, Prices, and Imports in the American Steel Industry." Review of Economics and Statistics, LII (February, 1970), 33—38. Sachrin, S. M. "Factors Affecting the Demand for Cigarettes." Agricultural Economic Research, XIV (July, 1962), 81:88. Sargan, J. D. "The Estimation of Economic Relationships Using Instrumental Variables." Econometrica, XXVI (July, 1958), 393—“15. Schultz, Theodore W. ”Investment in Human Capital." American Economic Review, LI (March, 1961), 16. "Reflections on Investment in Man." Journal of Political Economy, LXX (October, 1962), 1— . Segal, Martin. "The Relation Between Union Wage Impact and Market Structure." Quarterly Journal of Eco— nomics, LXXVIII (February, 1969), 96L113. Snodgras, Donald R. "Wage Changes in 2U Manufacturing Industries, 19U8-1959: A Comparative Analysis." Yale Economic Essays, III (Spring, 1963), 171-222. Sedersten, B., and Vind, K. "Tariffs and Trade in General Equilibrium." American Economic Review, LVIII (June, 1968), 39A-U08. Stolper, W. F., and Samuelson, P. A. "Protection and Real Wages." Review of Economic Studies, IX (November, 1991), 58—73. Swerling, Boris C. "Capital Shortage and Labor Surplus in the United States?" Review of Economics and Statistics, XXXVI (August, 195“), 286—89. Tatemoto, M., and Ichimura, S. "Factor Proportions and Foreign Trade: the Case of Japan." Review of Economics and Statistics, XLI (November, 19597, UH2-U6. 283 Travis, William P. "The Effective Rate of Protection and the Question of Labor Protection in the United States.” The Journal of Political Economy, LXXVI (May/June, 1968), Du3—6l. United Nations Economic Committee for Latin America. "The Growth and Decline of Import Substitution in Brazil.” Economic Bulletin for Latin America, IX (March, 1969), 1-59. Vargas, Fundacao Getulio. "Foreign Trade--Brazi1 in the LAFTA." Conjunctura Economica (January, 1967), 17. Viner, Jacob. "The Australian Tariff." Economic Record, v (November, 1929), 306—315. Unpublished and Miscellaneous Sources Clark, Paul G.v "Brazilian Import Liberalization." Research Memorandum no. 1“, Center for Development Economics,Williams College, Williamstown, Mass., September, 1967. (Mimeographed.) Clark, Paul G., and Weisskoff, R. "Import Demands and Import Policies in Brazil." U. S. Agency for International Development, Washington, D. C., February, 1967. (Mimeographed.) Donald, J. R.; Lowenstein, F., and Simon, M. "The Demand for Textile Fibers in the U. S." U. S. Department of Agriculture Economic Research Service, Technical Bulletin no. 1301, November, 1963. (Mimeographed.) Huddle, Donald. "Disequilibrium Systems, Industrializa- tion and Inflation: The Brazilian Case." Unpub- lished paper no. 30, Yale University, Economic Growth Center, July 17, 1967. Hufbauer, G. D. "Factor Endowments, National Size and Changing Technology: Their Impact on the Commodity Composition of Trade in Manufactured Goods." Paper presented in an NBER Conference on Technology and Competition in International Trade, October 11-12, 1968. (Mimeographed.) Johnson, Harry G. "The State of Theory in Relationship to the Empirical Analysis." NBER Conference on Technology and Competition, October 11—12, 1968. (Mimeographed.) 28A Levinson, Harold M. "Postwar Movement of Prices and Wages in Manufacturing Industries." Study paper no. 21, Joint Economic Committee, Congress of the United S ates, January 30, 1960. Morley, Samuel. "Import Demand and Import Substitution in Brazil. Brazil Development Assistance Program, University of California, Berkely, undated. (Mimeographed.) Nicholls, William. ”The Transformation of Agriculture in a Semi—Industrialized Country: The Case of Brazil." Presented to the Conference on the Role of Agri- culture in Economic Development, sponsored by the NBER, New York, December 1—2, 1967. (Mimeographed.) Nicholls, William, and Paiva, R. "Structures and Produc- tivity of Brazilian Agriculture." Vanderbilt Uni— versity and Funcacao Getulio Vargas, undated. (Mimeographed.) , Rothbaum, Melvin. "An Interpretation of Wage Structure Changes in France, Italy and the United States from 1938—1952." Unpublished Ph. D. dissertation, Harvard University, 1952. Sieniawski, Michael. "Favorable Trade Balance Boosts Brazil's Outlook." The Christian Science Monitor, February, 1970. Superintendencia da Moeda e Credito. Official Instruction no. 70 of SUMOC as ratified by laws 21A5 and 2A1D of December, 1953 and January, 1955. Waehrer, Helen. "Inter—Industry Skill Differences, Labor Earnings and U. S. Foreign Trade, 1960." Unpublished Ph. D. dissertation, Columbia University, 1966. " IIIllllljfllllllfllflfljllflljllII