- ‘YL. fix??? LM 2971 {a}; a «u , a - 7736 h _ ._ _.._ .. ._ ‘ ABSTRACT . THE NORLO SUGAR ECONOMY AN ECONOMETRIC ANALYSIS OF PRODUCTION AND POLICIES By Gordon Gemmill The purpose of this thesis was to estimate supply and demand functions for sugar for each of the major producing and consuming nations of the world and to use these functions to develop a model which would show the impact of alternative trade-policies. The model and its components were designed to give solutions both in long-run equilibrium and in an annual, recursive mode. Special attention was given to developing supply functions for both beet and cane in the U.S.A., taking into account the restrictions on acreage frequently imposed under the Sugar Program. Because the free market for sugar is typified by cycles in supply and price, a function capable of gen- erating these cycles was used in estimating supply from each of the major cane-producing nations. The supply of sugar in the U.S.A. was found to be generally price elastic, long-run elasticities being 0.00 for Puerto Rico, 0.75 for Louisiana, 0.90 for beet in the North and North-East, 0.99 for Hawaii, 2.7l fOr beet in the Nest and North—West and 4.23 for Florida. The supply of beet-sugar in Europe ranged in price elasticity from approxi- nately 0.30 fOr the Communist countries to 1.63 for France. The major cane-producing countries were found to have short-run price elasticity Gordon Gemmill of supply in the 0.l0-0.74 range, while their long-run elasticities were constrained to a maximum of 1.00 during estimation. The demand for sugar, examined for more than 70 countries using both time-series and cross-section data, was found to be generally both price and income inelastic. For the U.S.A. price elasticity was estimated to be approximately -0.03 and income elasticity 0.03. The range in price elasticities across countries was from -l.49 to 0.00 and in income elasticities was from 0.00 to 2.44. In the complete model there were 75 consuming and 68 producing regions, together comprising the whole world. Regions were separated by trade-barriers and transportation costs. Quota agreements were treated as exogenous flows. The model was solved for long-run equilibrium under trade-policies ranging from a most likely set to a set with universal free trade. Using the concepts of producer and consumer surplus, there was found to be a world gain of $330 million from free trade in sugar, the U.S.A. gaining $66 million, the EEC $70 million and the cane-exporting nations $639 million. The losers would be the traditional importers from the free market such as Japan, Canada and many African and Asian countries. Should the U.S.A. continue its current policy of free trade, there was estimated to be a 24 percent reduction in domestic production and a l3 percent drop in domestic price. .Consumers would gain $330 million and producers and government would lose $307 million. Should the EEC begin free trade it would suffer a 23 percent reduction in domestic pro- duction due to a 23 percent reduction in price. Consumers would gain Gordon Gemmill $709 million and producers would lose $525 million. Other policies which were considered included the formation of a cartel to raise prices by the cane-sugar exporting nations. Such a cartel was found to be very ineffective due to the elastic supply of (beet) sugar in the major importing nations. The policy implications of the solutions depend on whether producers and consumers in the developed countries are prepared to face a fluctuating free-market price. Freer trade would reduce the incidence of very low prices on the free market but not affect high prices. Since the international gains from freer trade are large, the Inultilateral reduction of barriers to trade in sugar would be feasible in a new kind of International Sugar Agreement. THE WORLD SUGAR ECONOMY AN ECONOMETRIC ANALYSIS OF PRODUCTION AND POLICIES By ‘q‘i N“ Go rdo nISGemmil l A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Agricultural Economics 1976 Copyright by GORDON GEMMILL 1976 ACKNOWLEDGMENT A large number of persons and institutions contributed to this work. The author is grateful for financial assistance from Michigan State University, the U.S. Department of Agriculture and the U.S. Department of the Treasury. At the University, the author acknowledges the guidance of Carl Eicher throughout his program. Vernon Sorenson, the Supervisor of the research, was always available for discussion and advice. The memberscfiithe Research Committee, William Haley, James Johnson and Mordechai Kreinin, gave very useful comments at various stages of the work. The assistance of Thomas Bates and his colleagues at the San Francisco State University in the preparation of transportation costs and for general discussions is acknowledged. Early in the work, the guidance and encouragement of Bruce Walter and Robert Bohall of the U.S.D.A. was very important. Similarly, Marvin Hayenga of General Foods Corporation was a source of great encouragement. Finally, discussions during l974 with officials (too numerous to list) of the International Sugar Organization, the French Manufacturer's Association, the United Nationslflxxiand Agriculture Organization, the U.S. Department of Agriculture and the Commonwealth Sugar Exporters' Association were very helpful. ii TABLE OF CONTENTS LIST OF TABLES ........................ LIST OF FIGURES ....................... Chapter I INTRODUCTION ................... Context and Objectives ............. The World Sugar Market Since l950 ....... A Conception of Market Behavior ........ II THE MODEL AND ITS SOLUTION UNDER ALTERNATIVE TRADE POLICIES .................. The Model ................... Solving the Model ............... The Model, Trade Policies and Welfare ..... Transfer Costs ............... Trade Policies and Welfare ......... The Context of Previous Research ........ III U.S. DOMESTIC BEET-SUGAR SUPPLY .......... Introduction .................. Procedure ................... Regions and Crops .............. Models ................... Estimation ................. Results .................... Region I (Michigan, Ohio, Minnesota, Iowa, North Dakota, Illinois, Indiana, Wisconsin, New York, Maine) .............. Region II (Colorado, Kansas, Wyoming, Texas, Nebraska, Montana, South Dakota) ...... Region III (Utah, Idaho, Oregon, Washington). Region IV (California, Nevada, Arizona, New Mexico) ................... Region V (Whole U.S.A.) ........... iii 58 59 61 62 63 A Summary and Some Elementary Projections . . . . U.S. DOMESTIC CANE-SUGAR SUPPLY .......... Introduction .................. The Sugar Industry of Each Region ........ Louisiana .................. Florida ................... Hawaii .................... Puerto Rico ................. Procedure .................... General Approach ............... The Cobb-Douglas Function .......... The Translog Function ............ Aggregation, Time and Scale ........... Results from Estimation ............. Production Functions ............. Cobb-Douglas Results ........... Translog Results ............. Auxiliary Time-Series Regressions ........ Projections of Supply .............. Louisiana ............. ' ..... Florida ................... Hawaii .................... Puerto Rico ................. Regional Comparisons and Aggregate Supply Summary and Implications of Chapter IV ..... THE EUROPEAN BEET SUGAR SUPPLY ........... Introduction .................. Policy and Production in the EEC-Six ...... France ....... ~ ............. West Germany ................. Belgium (and Luxemburg) ........... Netherlands ................. Italy .................... The Common Sugar Policy of l968 ......... iv Page 64 69 69 71 71 74 77 78 81 81 87 93 97 102 102 102 107 114 116 121 124 126 128 131 133 135 135 137 138 138 138 139 139 140 Chapter Page France ................... 145 West Germany ................ 146 Belgium .................. 147 Netherlands ................ 148 'Italy ................... 148 Policy and Production in the EEC-Three . . . . 149 Denmark .................. 149 Ireland .................. 150 United Kingdom ............... 150 Denmark ................. 151 Ireland ................. 152 United Kingdom ............. 152 Supply from Other Western European Nations . . 153 Supply from U.S.S.R. and Eastern Europe . . . . 156 Czechoslovakia ............... 159 East Germany ................ 159 Poland ................... 160 U.S.S.R ................... 160 Rest of Eastern Europe ........... 160 Summary .................... 161 VI THE INTERNATIONAL SUPPLY OF CANE SUGAR ...... 162 Introduction ............. '. . . . 162 Theory and Models ............... 162 Data and Results ............... 172 Summary .................... 181 VII THE INTERNATIONAL DEMAND FOR SUGAR ........ 182 Previous Studies ............... 182 Theory, Estimation and Results for Time-Series Data ..................... 183 Theory and Estimation ........... 183 Results of Individual Time Series ..... 189 Theory, Estimation and Results for Pooled Cross- Section and Time-Series Data ......... 197 Results of Pooled Estimation ........ 199 Sumary and Comments ............. 207 Chapter VIII RESULTS FROM THE MODEL UNDER ALTERNATIVE POLICIES . . Introduction ................... Annual Recursive Solution ............ Long-Run Equilibria Under Alternative Policies . . Welfare Implications of Long-Run Equilibria . . . A Summary of Chapter VIII ............ IX POLICY IMPLICATIONS AND SUGGESTIONS FOR FURTHER RESEARCH ...................... U.S. Sugar Policy ................ EEC Policy .................... Less-Developed Countries and Exporters of Cane Sugar .................... International Sugar Agreements .......... Further Research ................. APPENDICES ........................... A A NOTE ON DIRECT VERSUS INDIRECT ESTIMATION OF AGRICULTURAL SUPPLY ................. B A COBWEB MODEL FOR SUGAR .............. C FULL RESULTS OF THE EIGHT MOST IMPORTANT POLICY EXPERIMENTS ..................... D THE TRADE FLOWS UNDER POLICIES I AND II AND THE TARIFFS AND QUOTAS UNDER POLICY I .......... BIBLIOGRAPHY .......................... vi 239 240 241 243 243 260 272 279 DON p 4:- h J5 (A) U) (A) C N d C O 4.6 4.7 SJ 5.2 5.3 LIST OF TABLES World's Ten Largest Producers and Consumers, 1971-1963 and 1971-1973 ..................... World's Ten Largest Exporters and Importers in 1973 . . International Trade in Sugar by Type of Market in 1973. Regions in the Model ................. Gains Under Tariffs and Deficiency Payments Relative to Free Trade for an Importer ............. Beet Sugar Factories and Capacities .......... Crop Systems on Sugarbeet Farms by Region, 1972 . . . . Projected Outputs ................... Results of Klein's Method ............... Results of OLS . . . . . . . . . . . . . . , ..... Results of ILS .................... ZEF Estimated Parameters of the Translog Function . . . Estimated Partial Elasticities of Substitution and Own-Price Elasticities of Demand ........... Results of Auxiliary Time-Series Analysis ....... Projected Supply of U.S. Cane-Sugar in 1985 ...... Production of Sugar, Quotas and Farm Sugar Prices in the EEC-Six, 1968-73 . . .- .............. Summary of Estimated Elasticities for EEC ....... Production, Consumption and Self-Sufficiency in 1974 for Other Western European Countries . . . . . . . . . vii Page 24 35 43 50 66 103 104 105 108 115 113 154 Table 5.4 (1) (”-500 N noooooooopo NOS Production, Consumption and % Self-Sufficiency in 1974 for Communist Europe ................. Cuban Exports to Communist Europe ('000 Metric Tons Raw Value) .................. . . Cane Investment Equations . . . . ........... Yield Equations for Cane ............... Short-run Elasticities of Supply (at an export price of 6 cents per lb.) ........... Time-Series Results .............. Significant (Income and Price) Elasticities From (Ramsey and Semi-Log) Equations ............ Pooled Estimates . .................. Comparison of Time Series and Pooled Estimates Rank Order of Taste for Sugar ............. Price Margins and Income/Population Growth U.S. Domestic Cane Supply in Thousand Metric Tons Raw Value at 1974 Prices ................. Model and Actual Quantities and Prices, 1972—75 . . . . Policy Experiments and Long-Run Equilibria ...... Percent Tariff-Equivalencies of U.S. and EEC Policies . Gains in Welfare Under Alternative Policies ...... ‘ Summary of Gains in Thousands of Dollars ....... Results of Policies I and II ............. Results of Policies III and IV ............ Results of Policies V and VI ' ............. Results of Policies Xc and Xd . . . .......... viii Page 157 158 173 177 180 191 194 201 204 206 211 212 214 218 224 226 Table Page 0.1 Trade Flows Under Policy I .............. 272 0.2 Trade Flows Under Policy II ............. 274 0.3 Tariffs Used for Benchmark Solution ....... . . 276 0.4 Quota Flows Under Benchmark Solution ......... 278 ix Figure 1 45-12: .D wWNNNNN .1 LIST OF FIGURES Boundaries and Areas of Centrifugal Beet and Cane Sugar Production--1961 ................ World Sugar Prices 1950-74 ...... . . . ..... Raw Sugar Prices 1845-1974 (Annual Price to 1900, Monthly Thereafter) .............. World Production Consumption and Stocks of Centrifugal Sugar 1951-75 . . . . ........... Representation of a Tariff ............ Representation of a Variable Levy . . ......... Representation of a Deficiency Payment ...... Tariff v. Deficiency Payment ............. An Export Restriction Scheme U.S. Beet Sugar Production ..... . . . .' ..... Location of Sugar-Beet Production in the U.S.A ..... Quota, Price and Factory Capacity for U.S. Beet Sugar . Supply Schedules in Equilibrium for 1985 . ...... Cane Producing Regions of the Mainland United States Sugar Production in Louisiana ............ The Production of Seed-Cane in Louisiana in Relation to the Price of Sugar . . . s .......... . Sugar Production in Florida . ............. The Production of Seed-Cane in Florida in Relation to the Price of Sugar .................. Sugar Production in Hawaii ....... 14 15 29 31 31 34 36 41 42 47 67 69 73 73 76 76 79 Figure 4.7 4.8 4.9 .11 .12 .13 .14 .15 .16 .17 .18 .19 .20 bbbbhhbhbbb .21 Sugar Production in Puerto Rico ........ TWO-Firm Example of Cost Relationships ........ Inter and Intra-Firm Regressions ........ Elasticity of Substitution and Price Price Weights .................... Projections of Output for Louisiana .......... Projections of Acreage for Louisiana . . ....... Projections of Acreage for Florida .......... Projections of Output for Florida . .......... Projections of Output for Hawaii .......... Projections of Acreage for Hawaii ........... Projections of Acreage for Puerto Rico ........ Projections of Output for Puerto Rico ...... Projected Supply-Schedules for 1985 (in 1974 dollars) . Aggregate Supply-Schedule for U.S. Cane-Sugar (1985) Quotas in EEC Sugar Policy ............. Situation Facing Individual Farm Under EEC Sugar Policy Brazilian Supply and Demand System .......... Asymmetric Investment Function . . . ......... Actual and Estimated Area/Price Ratios for Four Countries ....................... Shapes of Price-Consumption Curves .......... Some Engel Curves ...... ° ............. (a) Price-Consumption Curves; (b) Income-Consumption Curves ........................ xi Page 79 83 92 110 117 122 122 125 125 127 127 130 130 132 132 141 142 165 167 178 187 188 195 I v .‘p ,- u - I .u 4" u ‘a n p CHAPTER I INTRODUCTION Context and Objectives The gyrations of the international market for sugar are now common knowledge to American consumers. During 1974 the retail price of sugar rose from 16.96 cents per pound in January to 62.76 cents per pound in December. Thereafter there was a continuous decline to 32.08 cents per pound at the time of writing, September 1975. Investigations into the causes of this price variation, such as that of the President's Council on Wages and Prices,1 failed to find a culprit more specific than a fluctuating international and domestic supply of sugar. This thesis is an attempt to understand the workings of the international sugar market and to expose through a formal, quantitative approach, the effects of alternative policies on the international distribution of production. The report falls naturally into two parts. The first, comprising Chapters I to VII inclusive, is concerned with the theoretical and em- pirical underpinning of the whole work and gives econometric estimates of supply and demand relationships. Chapter II discusses trade theory, spatial equilibrium and the model being developed in this research. In that chapter some previous research is also discussed in order to empha- size the need for the current work. Chapters III to VI present self- contained studies of U.S. Beet-Sugar Supply, U.S. Cane-Sugar Supply, 1Council on Wage and Price Stability (1975). Staff Report on Sugar Prices. Office of Wage Price Monitoring, May, Washington D.C.: Government Printing Office. European Sugar Supply and International Cane-Sugar Supply. Chapter VII reports on the demand for sugar in more than seventy countries and develops both general and country-specific equations. The second part of the report, comprising Chapters VIII and IX, is concerned with the solution, validation and policy-implications of the complete model of the world sugar economy. . From a methodological viewpoint, the first part Of this research may be considered a "positive" analysis, (in-so-far-as any part of neo- classical economics may be considered positive in nature), while the second part is of a considerably more normative character since certain stronger assumptions are utilized in solving the model and comparing the welfare of different countries under alternative policies. It is not usual to address the "casual” reader of a report, but those wishing to acquaint themselves with the sugar market might usefully omit Chapters III to VII inclusive which have a technical rather than policy-oriented flavor. . The objectives of this research may be listed as follows: 1. To estimate supply and demand functions for sugar for the main producing and consuming nations of the world. 2. To construct a model, using the above functions, capable of simulating the behavior of the market over the last decade. 3. By counter-factual experiment, to examine the effects of the sugar-policies of the U.S.A. and European Economic Community on the price of sugar and the international distribution of income. 4. To project the behavior of the market for sugar under a variety of alternative policies including an examination of the feasibility and effectiveness of a producers' cartel. Summarizing these objectives, the research may be characterized as an attempt, through econometric model-building procedures, to analyze the workings of the market under alternative policies, using somewhat weaker assumptions than have been used in previous research in this area.2 The Introduction continues with a brief review of the interna- tional sugar market and a discussion of the policies of both exporting and importing nations. The World Sugar Market Since 1950 Sugar, or more accurately that substance whice we call sugar and which is actually sucrose, is derived from sugar cane in the tropical and semi-tropical regions of the world and from sugar beet in temperate regions. Because of this duality of sources, sugar is produced in almost every country in the world, as may be seen in Figure 1.1 In 1973 41 percent of all centrifugal sugar was derived from beet, this propor- tion varying but little in recent years with a high of 45 percent in 1964 and a low of 39 percent in 1971.3 Since the so-called "less developed countries" are concentrated in the warmer climates, almost all of their production is from cane. The only "developed" cane pro- ducers of any importance are Australia and the U.S.A. In consequence approximately 53 percent of the world's sugar was produced by less developed countries in 1973. These proportions may be slightly deceptive, as a crude product, called noncentrifugal sugar, is also produced in less developed countries for local consumption, but reliable data on its level of production do not exist. What data there are suggest that it 2Previous research is discussed in Chapter 11. 3Except where otherwise stated, data are from: "International Sugar Organization (Annual). Sugar Yearbook. London: International Sugar Organization, 28 Haymarket, W.C.I." At the time of writing, data for 1974 are not completely available on a world basis. Alternative series are produced by F. O. Licht of Ratzeburg, Germany, the United Nations Food and Agriculture Organization and the U.S. Department of Agriculture. .AmmmFV cornm~mcmmgo gmmzm chovumcgmucm ”mugzom . .3) .D 3:335...” 28 neffm v" you; “mau.< mfluaueu \V \A\\ Fom_--coruu=uoga Lamam acau wee pawn am=m_gu:mu Co mamt< new moveeuesom is an inferior product which is rapidly replaced by centrifugal sugar as income rises. Table 1.1 ranks the most important countries by production and consumption for 1961-63 and 1971-73, as well as giving world totals. In production the U.S.S.R. ranked first in both periods followed by Cuba in 1961-63 and by Brazil in 1971-73. While Cuban and U.S. produc- tion were almost constant over the decade, Brazilian production rose 91 percent, Mexican 66 percent, Australian 58 percent and that of the U.S.S.R. by 45 percent. In consumption the U.S.S.R. and U.S.A. domin- ated both periods but U.S. consumption advanced only 17 percent whereas Soviet consumption gained 38 percent. However, the advances in consump- tion by other large countries, whose per capita consumption in 1961-63 was much lower than that in the U.S.A., were correspondingly greater, being 99 percent for Japan, 90 percent for China, 69 percent for Mexico, 54 percent for India and 49 percent for Brazil. Annual consumption per head in Western Europe, Canada and the U.S.A. has stabilized in recent years in the 40 to 50 kilogram range. For example, in 1963 U.S. per capita consumption was 48.2 kilograms and in 1973 it was 49.8 kilograms. By contrast, Japanese per capita consumption rose from 17.6 to 30.1 kilo- grams in the same period under the influences of a rising real income and low initial consumption. The values in Table 1.1 disguise the relatively small volume of sugar which was traded, 22 million tons in 1973 or 28 percent of total production. Net exports were only 19 million tons in 1973 or 25 per- cent of total production. Table 1.2 shows that, of these exports, Cuba led with 4.80 million tons, followed by Brazil with 2.98, Australia with 2.10 and the Philippines with 1.39. Ten countries accounted for 81 .mwaowumwpmgm wt _m:::< Pwmzumm =.mcmvapcoz m==a>< me ..m.a.m.u .Apm:::< me>Imv cowu353mcou Ammmgm>< me>nmv cowuoauoga Amzpm> 3mm mach uwgpmz ooo_v mmm_-3nm_ new mom_-FNm_ .mcmssmcou new mcmoscoca ummmcmg cop m.cpgo3 .F.~ wFDmH Table 1.2. World's Ten Largest Exporters and Importers in 1973 ('000 Metric Tons Raw Value) Exporters Importers Country Net Exports Country Net Imports Cuba 4,797 U.S.A. 4,830 Brazil 2,975 U.S.S.R. 2,584 Australia 2,103 Japan 2,395 Philippines 1,385 U.K. 1,811 Dominican Republic 1,070 Canada 952 South Africa 913 China 580 Mauritius 738 Iraq 457 Mexico 590 Yugoslavia 380 Taiwan 508 Malaysia 331 Argentina 470 Indonesia 307 Others 3,659 Others 4,754 Total 19,208 19,381 Source: Same as for Table 1.1 except for U.K. for which it was "European Economic Community (1975) Yearbook of Agricultural Statistics, 1974. Luxembourg: E.E.C. Statistical Office." percent of all exports in 1973. Turning attention to imports in 1973, the U.S.A. led with 4.83 million tons, followed by the U.S.S.R. with 2.58, Japan with 2.40 and the U.K. with 1.81. Table 1.2 shows that the top ten countries accounted for 76 percent of sugar imports and the top five alone for 65 percent of imports. The picture of production, con- sumption and trade which emerges is one of widespread production and hence widespread self-sufficiency. However, a few countries, notably the U.S.A., U.S.S.R. and Japan, rely heavily on imports which are also provided by a relatively few countries, dominated by Cuba, Brazil and Australia. Very little trade in refined sugar occurs Since the refined pro- duct is hygroscopic and therefore more expensive to handle. The lack of trade in refined sugar is also the result of higher tariffs on refined than on raw sugar and the development, during colonial times, of the export Of raws for refining in the "mother" country. Raw sugar is the product of the sugar cane mill and a normal rate of conversion is 100 parts of raw to 92 parts Of refined sugar. Sugarbeet is usually proces- sed directly to the refined form Since there are technical economies in this procedure and since beet sugar is generally consumed in its country of origin. The major centers of sugar trade are New York and London for raw sugar and Paris for refined. To talk of a "world sugar market" is somewhat misleading. There are really three kinds of markets. Firstly, there are the domestic markets within producing countries which accounted for 72 percent of all sugar in 1973. Secondly, there are international agreements between some of the largest importers and their suppliers. Such agreements may cover price, quantity, or both price and quantity. Examples are the current or recently expired agreements covering imports to the U.S.A., U.S.S.R. and U.K. Thirdly, there is a residual "free-market” in sugar which has from time to time been regulated by International Sugar Agree- ments. Table 1.3 gives the volumes of sugar traded in 1973 in each of the major markets. The table shows that just over half of traded sugar entered the free-market, however that half represented a mere 14 percent of total production. The U.S. Sugar Act covered 22 percent of trade, Cuban Agreements covered 14 percent of trade and the Commonwealth Sugar .Agreement covered 8 percent of trade. Each of these three institutions Table 1.3. International Trade in Sugar by Type of Market in 1973 Type of Market Exporter Importer '000 Metric Percent of Price in Cents Tons World Exports Per Pound Under U.S. Philippines U.S.A. 1.319 Sugar Act Dominican Republic 676 Brazil 591 (Duty-paid Mexico 577 in New York) Others 1,672 Total 4,835 21.8 10.29 Under Commonwealth W. Indies 3 Guyana U.K. 736 Sugar Agreement Mauritius 386 (f.o.b. in (Negotiated Price Australia 340 Caribbean) Quota) Others 308 Total 1.770 8.0 5.36* Under Bilateral Cuba U.S.S.R. 1,661 Cuban Agreements China 302 (f.o.b in E. Germany 259 Cuba) Bulgaria 213 Others 561 Total 2,996 13.5 6.0-11.0 Free Market Exports Brazil 2.530 Cuba 1,774 Australia 1,497 (f.o.b. in EEC 1,468 Caribbean) South Africa 824 China-Taiwan 428 Poland 422 Dominican Republic 396 Argentina 396 Others 1,930 Total 11.665 52.7 9.61 Free Market Imports Japan 2,395 U.S.S.R. 1.016 Canada 952 Iraq 474 (f.o.b. in Yugoslavia 368 Caribbean) Malaysia 331 Indonesia 307 Iran 302 Others 5,658 Total 11,803 53.2 9.61 Gross Exports in all markets 22,145 100.0 Net Exports in all markets 19,208 Domestically conSumed 55,950 World Production 78,095 *Plus a bonus of 1.18 cents Suppliers. per lb. for West Indies and Guyana and 0.75 cents per 1b. to other "less developed" 10 will now be briefly considered. The U.S. Sugar Act in essence dates from the Jones-Costigan Act of 1934 and lasted until January 1975. The Act was effectively killed by a vote of Congress in June 1974. Under this legislation quOtas for the supply of sugar were allotted to particular foreign suppliers as well as to domestic producers. To encourage compliance with quotas, domestic producers were awarded a bonus conditional upon their remaining within the allotted acreages. The "conditional payments“ were financed from a small duty paid on imported sugar, which was 0.625 cents per pound in 1973. The payments in most years were modest, amounting in 1973 to 1.20 cents per pound of raw sugar to beet growers and 0.56 cents per pound of raw sugar to cane growers. Prior to 1960, Cuba was the main foreign beneficiary of a quota, exporting 3 million tons annually to the U.S. out of total imports of 4.2 million tons. Following the Cuban embargo of 1960, U.S. domestic production was encouraged by an increase in quota and the remaining shortfall was allocated to other foreign suppliers (mainly the Philippines, Mexico, Dominican Republic, Brazil, Peru and Australia). The Act was designed primarily to protect domestic producers, but it also protected the favored foreign suppliers whenever the U.S. price exceeded the free-market price (the usual situation). In 1971 a specific price-objective was written into the Act so that quotas were to be adjusted in order that the domestic sugar price should rise at a rate indexed to the average of the wholesale-price and agricultural input-price indices.4 While, as shall be seen, the Act protected the 4 PSUG Stated formally this is: POB = (PARINDEX + WPI)/2 where: POB = the price Objective for sugar, PSUG = the price of sugar from Sept. lst 1970 to August 3lst 1971. 11 favored suppliers from low prices, (the world free-market price standing at least 2 cents per pound less in most years), it did little to pre- vent prices from rising should there be a relative decline in world supply. A secondary part of the Act legislated minimum wages to be paid laborers in the industry and the probable effect of this on employ- -ment in cane production will be discussed in Chapter IV. The Commonwealth Sugar Agreement, first established in 1951 and still existing, has the objectives of ensuring the supply of sugar to the United Kingdom and maintaining stable prices. Under the agree- ment the Commonwealth exporter is allotted on "overal1-agreement-quota” (OAQ) and a “negotiated-price-quota" (NPQ) (equal to approximately two- thirds of the OAQ). The negotiated price is determined under the terms of the agreement, but is subject to annual adjustments. The U.K. thus obtains, at least theoretically, two-thirds of its imports at a pre- determined price. The remaining one-third is to be purchased by the U.K. at the ruling world price. In a year such as 1974 in which prices rose rapidly in the free market, the negotiated price becomes a fiction and either the Commonwealth price rises to meet the currently existing free-market price or supply in the U.K. is cut. The Cuban Agreements with Communist countries became necessary after the suspension of Cuban exports to the U.S.A. in 1960. New quantities and prices are announced from time to time following nego- tiations on a bilateral basis. Should world sugar prices rise, the previously agreed price is abandoned and a new price contracted. For PARINDEX = the parity index of agricultural input prices (1967 = 100). WPI = the wholesale price index (1967 = 100). 12 example, during 1974 the price to the U.S.S.R. was recontracted from 5 Cuba has not 11.0 cents per 1b. (approximately) to 20.0 cents per lb. been obliged to fulfill its quota undertakings, at least in the case of the U.S.S.R. for which the quota stands at 3 million tons but which level has never been reached. During the 1960's the U.S.S.R. exported large quantities of refined sugar to the free market as well as import- ing Cuban raws, so that the Cuban Agreement did not diminish the volume on that market but simply changed the pattern of trade. Figure 1.2 compares the annual average prices ruling in the three Rain "markets" over the last two decades. "G- 1-2 WORLD 5106“! PRLCES ”50-74 30 CENTS, PER La. as A 20 15 ‘o I m. ,,,,,,,, spun-"ML ...... L..\- ..... J’A“"‘IT'.‘ I" 5 A .‘ 'L _ ~‘- - J, “L .Jr' “\ v " ...‘ ~.——-. ‘ I FREE ARKET/ ‘~~r—---' "0N Atfll 0 1030 1994 1950 1982 1906 1970 1974 YEAR 5New York Times, 26th January, 1975. 13 The U.S. and Commonwealth sugar prices were generally higher than that of the free market. Two exceptions exist, these being for 1963 and for 1974-75. In 1963 a diminished international supply pushed the free- market price temporarily above the U.S. price and also led to an increase in the U.S. price. By contrast, the Commonwealth price, which was pre- determined, was not affected. In the 1974-75 period international supply was again diminished and this pushed up the free-market price which, in turn, took the U.S. price upward at the same rate. The Common- wealth price in 1974-75 also followed the upward movement but with a slight lag caused by the necessity of further price negotiatons, the lag being sufficient to cause actual shortages at retail in the U.K. Until the recent, extended surge in the free-market price, the pattern in that market had been one of fleeting highs followed by extended lows. Such highs can be seen in Figure 1.2 for 1951, 1957, 1963 and 1974-75. A more general view of price fluctuations in the world and U.S. markets may be seen in Figure 1.3 which traces price back to 1845. The figure demonstrates that price fluctuations are not a recently developed phenomenon. For example, in 1864 the spot price fOr raw sugar reached an annual average of 16 cents per 1b. and the monthly average 22 cents per lb. in 1921. These may be compared with the U.S. market's monthly peak of 57.30 cents per 1b. in November 1974. In real terms, the peak of 1974 would not be higher than the previous highs. The prices of the last two decades in Figure 1.2 may be compared vfith worldwide production, consumption and stocks for the same period as in Figure 1.4. It is hardly surprising to find that production and consumption have risen in parallel and that declines in stocks have 14 FIG. I.3 RAW SUCAR PRICES l845-I914 35 P we "11 CENTS PE (ANNUAL PRICE 'I'O I900, MONTHLY THEREAFTERI MAJOR SOURCE: COMMODITY YEARBOON 25 15 I/ U.S. RICE .1- 1850 «no 1920 1930 1900 IOIO W0 LOP IOE/ 1940 YEAR 1955 s,» --- - g, ’---- I960 1975 15 no FIGURE 1.4 woRLo Pagoucnow itwsuupnonl AND STOCKlL/ 1 OF ceme UGAL sue 1951-75 0, I 11000 MET 1c _/ TONS __,’ so 1 Its-5 PRODUCTION ‘>/,/ (I ’d”’ 40 r ’ ._ ,_ p#" cowsuupnow ,./ ,—-¢---- . 20 :"'_"‘~ _.__7..—-.-:r_’_'_‘:‘ . "P“ "---- "’-d-~~~‘—’ " STOCK o 1950 1955 1958 1962 1966 1970 1914 YEAR occurred when consumption has exceeded production, resulting, in turn, in higher prices on all markets. It is a truism to state that free- narket price is strongly (and inversely) related to stocks as a pro- lxntion of consumption.6 Unlike some other agricultural commodities, sugar is expensive to store and stocks are not held for a number of years for speculative purposes. While "supply of storage" theory, such 7 as that developed by Weymar for the Cocoa Market may be useful in explaining very short-term price fluctuations, it has little relevance 6Work such as that of Ingersent'along these lines is of limited value. See Ingersent, K. A. (1975). "Sugar Prices and Stocks,"(British) gggrnal of Agricultural Economics, 26, (2), (May), pp. 227-238. 7Weymar, F. H. (1968), The Dynamics of the World Cocoa Market. Cambridge, Massachusetts: M.I.T. Press. 16 to the annual fluctuations which are the concern of this report. Because the change in stocks is an "identity", being the difference between production and consumption, rather than causally related to market behavior, stocks are ignored in this thesis, especially as their magnitude is not very accurately recorded.8 There remains one further set of institutional arrangements to introduce, the periodic International Sugar Agreements. Such agree- ments existed for 1954-61 and 1968-73 and were made between both major exporting and importing nations. The objective of the agreements was to dampen the movement of price in the free market. At the beginning of the year the International Sugar Council would estimate market requirements and assign initial export quotas pro rata to agreed ton- nages. Should price fall on the free market below an agreed minimum, quotas were reduced and the converse when prices rose above an agreed naximum. The agreements were conceived in years (1953 and 1967) in which free-market prices were very low, there being good reason for collaboration among producers at such times. The low prices were themselves the result of an expansion in output in response to the previous high prices of 1951 and 1963 respectively (see Figure 1.2). The agreement was not extended in 1961 because Cuba claimed an increase in its quota sufficient to offset fully its loss of U.S. sales while at the same time making alternative arrangements with the Com- munist countries outside the terms of the agreement. In 1973 the agreement was not renewed because the price Of sugar was rising rapidly 8During 1974 F. O. Licht, the respected source of sugar-market information, merely calculated stocks as a minimum proportion of con- sumption rather than actually measuring them. 17 on the free market and no quota limitations seemed appropriate. If the International Sugar Agreements had any effect it can only have been minor, since the agreements are seemingly not exogenous to the cycle of free-market prices, i.e., the cycle affected the agreements and not vice versa. Possibilities for stronger producer-dominated cartels will be considered later in this report. A Conception of Market Behavior The objective of this research, a formal analysis of alternative policies, has now been placed in the context of the world's sugar market. The remainder of this Introduction outlines the conception of market behavior on which the model was based. The production of both cane and beet on a worldwide basis is subject to cycles, as has been well documented by Hagelberg.9 The cause gf#the cyclesmis hypothesized in this research to be the invest- ment decisions of the different producing countries. These decisions are, in turn, the result of imperfect knowledge concerning the actions of other countries and they are also influenced by domestic politics, since sugar production is usually regulated by government. The cycles in the supply of beet and cane, together with a relatively smooth ex- pansion of demand through time, induce price cycles. An examination Of the actions of beet and cane producers in the last full cycle, which had its peaks in 1963 and 1974 (see Figure 1.2), may be instructive in explaining the approach to modeling which was taken in this research. 9Hagelberg, G. B. (1975). "Instability of World Centrifugal Sugar Production.” Working Paper, Institut fUr Zuckerindustre, Amrumer Strasse, Berlin. 18 In 1963 the free-market price averaged 8.50 cents per pound in New York, as compared with 2.98 cents per pound in 1962. In 1964 a 22 percent expansion of world beet production occurred and an 8 percent expansion of cane production. Price fell moderately, to an average for the year Of 5.87 cents per pound. In 1965 beet production expanded less than 1 percent more, but cane production expanded an additional 14 percent and price was driven down to 2.12 cents per pound. For the period 1966-68 there was little change in either beet or cane produc- tion, but the price was less than 2 cents per pound in all three years. Thereafter price began climbing slowly, due (it is hypothesized) to the smoothly expanding nature of demand, and price was 3.37 cents per pound in 1969, 3.75 cents in 1970, 4.52 cents in 1971, 7.43 cents in 1972 and 9.61 cents in 1973. Note that in terms of 1963 dollars, the 1973 price was only 6.62 cents per pound, less than the peak of 1963. Dur- ing the 1969-73 period of slowly rising prices, production of beet expanded a mere 3.6 percent but production of cane, being more depen- dent on the free market, expanded 19.1 percent. Suddenly, in 1974, the free-market price rose dramatically to an average of 29.99 cents per pound for the year (equivalent to 18.63 cents per pound in 1963 dollars). The price rose rapidly, it is hypothesized, not only because of the inelastic nature of demand with respect to price in the high-income importing countries, but also because many of the exporting countries did not allow the price to rise in their domestic markets, preferring, 10 instead, to ration exports. By 1975 the price was again falling, under the influence of an expected increase Of 13 percent in beet 10Note that of the top-ten exporters, Brazil, Philippines, South Africa, Mexico and Argentina all have large domestic markets. 19 production but a mere 2 percent in cane production.11 This interpretative description of one price-cycle implies that the following features be included in a model of the world sugar narket. Firstly, beet and cane supplies should be separately included since they are subject to different lags in response to price. A one- year lag for beet and a two-year lag for cane may be expected to be the minima. In addition, cgne supplies should be modeled in such a way that —...-—4- -wnt-‘p.“m.~k, 'WV’D'. 3 «he “P‘s'u high prices induce new investment but low prices do not lead to disin- vestment, due to the fjfit¥-9f assets. It is important to remember that cane is a perennial crop while beet is an annual crop. Secondly, demand should be shifting steadily to the right and of a low price elasticity. Thirdly, the behavior Of exporters should be to ration exports in the interest of low domestic prices; The above statements about modeling are normative, but each will be examined in a more formal and objective manner in the relevant chapter. The nature of responses by beet producers is the subject of Chapter III for the U.S.A. and Chapter V for European countries. Responses by cane producers are examined in Chapter IV for the U.S.A. and Chapter VI for the major exporters. The demand for sugar in more than seventy countries is examined in Chapter VII. The rationing of exports by some countries when prices are high is examined in Chapter VIII, where the results of the whole modeling exercise are presented. In Chapter II, which now follows, the structural equations of the model will be expounded in their simplest form and trade barriers will be discussed as they relate to the model and its solution. The chapter gives an overview of the model in skeletal fOrm which is then "clothed" in Chapters III to VII and solved in Chapter VIII under alternative policies. 11Data from F. O. Licht, Ratzeburg, West Germany. CHAPTER II THE MODEL AND ITS SOLUTION UNDER ALTERNATIVE TRADE POLICIES The Model The model was conceived in both static and dynamic (recursive) forms. For simplicity, the static form is expounded first. The struc- tural relations may be stated in seven equations.1 Let there be m producing and n consuming regions, subscript i always denoting a producing region and subscript j a consuming region. Let Q? = the quantity of raw sugar demanded in the jth region; Q? = the quantity of raw sugar supplied by the ith region; Pj = the wholesale price of raw sugar in the jth region; qij = shipment of raw sugar from region i to region j; G = cost of shipment, including trade barriers, from region 13 i to region j. Then the seven equations are as follows: Demand relations for each consuming region:2 0 D . 2.1 . = . P. , = , ,. . ., ()QJQJ(J) J 12 n Supply relations for each producing region:2 S = S . = (2.2) Qi Qi(Pi)’ 1 1,2,. . .,m Total quantity demanded in equilibrium equals the sum of all shipments: lThe whole of this exposition owes a large debt to the very clear presentation of "Zusman, P. et a1. (1969). Possible Trade and Welfare Effects of EEC Tariff and Reference Price Policy on the European-Medi- terranean Market for Winter Oranges, Giannini Monograph No. 24, Univer- sity of California." 2These relations will be complicated by the inclusion of the other exogenous variables in addition to price later in this chapter. 20 21 (2.3) O'?=zq.., j=12...n J i1J ,’ 9 Total quantity supplied in equilibrium equals the sum of all ship- ments: (2.4) Q? = x q.., 1=1,2,. . .,m 1 . 13 J Shipments cannot be negative: (2.5) qij :0, 1:1,2,. . .,m j = 1,2,. . .,n At equilibrium the prices in any two regions cannot differ by more than transfer cost per unit: (2.6) Pj ' Gij " P1- :0, i =1,2,. o .,m j = 1,2,. . .,n At equilibrium the sum of transfer expenditures is exactly balanced by the sum of price differences times quantities shipped for all regions: (2.7) E §[(Pj - Gij - Pi) qij] = 0 Because of Equation (2.7), should the strict equality hold in Equation (2.5) then the strict inequality holds in Equation (2.6) and the exact converse if the strict inequality in Equation (2.5) holds. Equations (2.3) to (2.7) are nothing more than the well-known transportation model, given the particular unit costs of transportation, Gij' The addition of the demand and supply equations, (2.1) and (2.2), adds to the difficulty of solution but not greatly to the conception. To change the model for recursive solution requires merely that the supply in year t become a function of previous and not current prices; i.e., it is predetermined for the current year. Equation (2.2) becomes: 22 (2.2)' QBR .,P 1 = SR . _ Qi (Pit_],. . it-k)’ 1 - 1,2,. . .,m where Q?R denotes recursively-determined supply in region i. Although it may be shown, under certain conditions, that any quota way be represented by an equivalent tariff, it is simpler in this context to treat quotas separately, Since they are used in such a widespread manner by importing countries. Hence there is an additional identity: (2.8) OgQUOTij g.q .., i 1,2,. . .,m 13 j = 1,2,. . .,n where QUOTij is the quota given by importing-region j to exporting-region 1.3 Should solutions in different time-periods be desired, the supply and demand functions are affected. With respect to demand, population and income change over time so that the fully-specified demand relation- ship becomes: , D D (2.1) Qj (Pit’ POP th = INCjt) j = 1,2,. . .,n it’ where: POP population, INC income, t year. WiUIrespect to supply, technology, the prices of competing crops and input prices change so that the fully-specified supply relationship becomes (in static form): .. S = S (2.2) Qit Qit (Pi, T, PAi, PINi) 3Equation (2.8) implies that actual shipment may exceed the quota. However, if so desired, the shipment may be limited to the quota by imposing a heavy tariff on additional imports; such a procedure was used for the U.S.A. 23 where: T = technology, PA = the price of a competing crop, PIN = the price of inputs. SOme elaboration of the regions and trade relationships used in the model may help to add substance to the bare outline which has now been presented. The world was divided into 66 supplying regions and 75 demanding regions. For simplicity, the 66 were an exact subset of the 75. The regions are listed in Table 2.1. The choice of regions was on the basis of the magnitude of production/consumption in the region, the region's importance in sugar-trade, geographical distribution on a worldwide basis and policy considerations. Thus, the U.S.A. was divided into three regions, Canada into two regions, the U.S.S.R. into two regions and the European Common Market into its constituent countries. Whereas, by contrast, many African and Asian countries were grouped together into single regions. The table also delineates the kind of supply function which was used for each region. In general, beet-producing regions had double- 1ogarithmic functions, cane-producing regions asymmetric functions (i.e., functions for which the response to rising and falling prices nay be dissimilar), and a few regions either simple time-dependent supplies or even totally price-inelastic "point" estimates. The func- tions, their forms and estimates are the subject matter of Chapters III to V1 and will not be elaborated here. Similarly, semi-logarithmic functions were used for demand and these are the subject matter of Chapter VII. 24 Table 2.1. Regions in the Model Cantinent Region Type of Supply Function Log-Linear Asymmetric Time Only Point Hone Europe Austria Belgium r Czech0510vakia Denmark ; Finland France / Germany (West) . Germany (East) ; Greece . Iceland Ireland Italy . Netherlands . Norway Poland , Portugal Spain Sweden Switzerland Turkey { U.S.S.R. (West) A U.S.S.R. (East) U.K. / Eastern Europe: Albania. Bulgaria. Hungary. ROumania. Yugoslavia North America Canada (West) Canada (East) U.S.A. (West) J ‘ U.S.A. (SOuth) .“ U.S.A. (East and North) ; Central America Barbados Cuba Dominican Republic Guatemala Jamaica HGIiCO Nicaragua Puerto Rico Trinidad and Tobago Central America: Bahamas. Belize. Bermuda. Costa Rica, [Cuador. El Salvador. Haita. HOnduras. L Netherland Antilles. Panama, Surinam, Virgin Isles r Sputh America Argentina Bolivia and Chile Brazil Columbia Guyana Paraguay and Uruguay Peru Venezuela Asia China China - Taiwan r Hong Kong India Indonesia Iran Japan Korea (North and Scuth) Pakistan and Bangladesh Philippines ; Saudi-Arabia Singapore Sri Lanka Thailand . Near East: Iraq. Israel. Jordan. Lebannon. Syria Far East: Afghanistan. Burma. Malaysia. Nepal. Vietnam Africa Mauritius 5 South Africa / North Africa: Algeria. Etypt. Libya. Morocco, Tunisia West Africa: CamerOun, C.A.R.. Chad, Dahomey. Equatorial Guinea. Gambia. Ghana. Guinea. Ivory Coast. Liberia..Ma1i, Niger, Nigeria. Senegal, Sierra Leone, Spanish Sahara. Togo. Upier Volta North-East Africa: Ethiopia. Sudan. Somalia ' East Africa: Burundi. Kenya. Rwanda. Tanzania. Uganda. Botswana. Malawi South-Central Africa: Mozambique, Rhodesia. Swaziland. Zambia Sputh-Hest-Central Africa: Angola. Congo, Namibia. Zaire "('Qto“. ‘< O, 3— where: PTH. = the predetermined threshold price in region j, below which imports may not occur. Information concerning tariffs was Obtained from the International 10 Customs Journal. Information on the cost of transportation was pro- vided by Professor Thomas Bates of San Francisco State University. Professor Bates developed a variety of linear cost functions which Tnoved rather complex because of the large number of dependent vari- ables. Given unit cost per mile as a function of distance alone, a nmflinear relationship emerged, implying short-hauls to be more costly per nautical mile than long-hauls. Professor Bates' function was approximated by:11 (2.13) t.. = 0.03 O..°°5 1J is where: tij = the cost in l974 cents per lb. per nautical mile, Dij = the distance between i and j in nautical miles. 10 International Customs Tariff Bureau (various). International Efliggms Journal. Brussels: I.C.T.B. , 1lThe approximation was made informally, but an allowance for 1”Flation to 1974 from data on actual costs for 1971-73 was included. 28 Then simply, (2.14) Tij = tij ° Dij A matrix of distances was drawn up using the U.S. Naval publication entitled "Distances Between Ports."12 Distances within Europe, the U.S.S.R., the U.S.A. and Canada were included on an overland basis where apprOpriate. Overland costs were assumed the same as those by sea. As an example of the shipping costs implied by Equation (2.13), the Cuba-New York route (1,199 nautical miles) is estimated to cost 1.04 cents per 1b., while Australia-New York (9,692 nautical miles) costs 2.95 cents per lb. The distances in the matrix assumed the Suez Canal to be closed. This was a reasonable assumption in simulating 1974 equilibria but would slightly distort prices (particularly in the Near East) in projections. Trade Policies and Welfare A graphical presentation of the trade policies cOnsidered in this research will now be given, together with a discussion Of wel- fare measurement. The objective of this section is to show how trade policies affect international equilibrium. In Chapter VIII, experiments on the model will be conducted which empirically Show how such policies affect equilibrium and welfare for the 75 regions of the world. The presentation assumes two countries, an exporter and an importer, both of which produce sugar under free trade and there are initially no costs of transportation. The first trade barrier to 12United States Naval Oceanographic Office. (1964). "Distances Between Points." Washington, D.C.: U.S. Government Printing Office. u .,r‘ . «I'! T" . fi‘ . «a. s a~ . v., a. . ‘ I I '0. ‘ \». J ”I 29 consider is the tariff. In Figure 2.1 the free-trade equilibrium at price PE (=Pg) implies trade at the level qE. The imposition of the tariff, TAR, reduces quantity traded to qT, reduces quantity supplied but increases quantity demanded in the exporting country i, and increases domestic quantity supplied but reduces quantity demanded in the import- ing country j. Assuming that the marginal utility of money is constant across the two nations. and that the income-effects of change in the price of sugar are negligible, the gains and losses from the tariff nay be summed as follows: Consumers in i gain Pgr u PI. Producers E T in i lose P v w Pi' The government in i gains 1xfg and hkwu in tariff i revenues. It may be shown that the summation yields a "dead-weight" loss equal to 50A in the central, trade graph, this loss being appor- tioned can to the importer and uGA to the exporter. Note also that transfer payments to the government of the importing country j equal nekp. Figure 2.1 Representation of a Tariff Exporter i Trade Importer j n ---- - --- 40 ‘0 D -c 30 The tariff thus analyzed could be either an 2g valorem or a Specific tariff. Similarly, transportation costs are analogous to a tariff in terms of dead-weight loss, although they yield no government revenues. Figure 2.1 could equally well demonstrate the effect of a 92931 imposed by the importer j. Let the quota equal qT and the effects, in terms of price changes from equilibrium, are exactly as for the tariff TAR. The dead-weight loss is also exactly as before, but the distribu- tion of government revenues is different--there are no such revenues. The suppliers in i receive the full sum which previously accrued to the government in j. Should the government in i be the export agent, it receives these revenues. Should the government in j auction the quota qT competitively, then it receives the revenues and the tariff and quota become exactly equivalent. Under the U.S. Sugar Act, quotas were allo- cated to "friendly" foreign suppliers. Since the exporting governments were usually also the export agents, the revenues from possession of a quota passed to them. It is hardly surprising, therefore, to discover that such governments expended large sums of money in lobbying in Washington for the maintenance or expansion of quotas.13 The European Economic Community protects its domestic sugar industry with a variable lery on imports. This is depicted in Figure 2.2. The notation is exactly as in Figure 2.1, except that PTH is threshold price and qL2 is quantity traded under the levy with thres- hold price PTHZ. To explain, the EEC decides upon a minimum import or "threshold" price, PTH. Should this price be less than the free-trade 13For example, Brazil paid $180,000 in 1973 to its agent in Washington. 31 Figure 2.2 Representation of a Variable Levy Exporter 1 Trade Importer 1 Price 1 1 1 1 ' 1 I 1 1 1 l , 1 I i l O a c d ba Figure 2.3 Representation of a Deficiency Payment Exporter i Trade' Price, Pricej E POP Pi * b---‘—- 32 equilibrium price, it has no impact, e.g. PTH1 in the figure. Should this price exceed the equilibrium price Pg, such as PTH2 in the figure. imports are restricted to that quantity entering at this price. The effect may again be interpreted as being exactly equivalent to a tariff of magnitude (PTH2 - Pg). Because the EEC demands competitive bidding on the quantity to be imported at price PTHZ, all of the government revenues accrue to the EEC and none to the exporting country's govern- ment. In the recent past, the threshold price for sugar was fixed by the EEC-SIX at a level such as PTH at which no imports occurred. With 3 Britain's entry to the Community, the situation is somewhat changed, as will later be examined. The next trade barrier to be considered is the direct subsidiza- tion of the industry in the importing country. Such subsidization is often called a "deficiencygpayment." While such payments are not important at present in the sugar-importing countries, they will be considered in Chapter Ix as a feasible alternative to tariffs and quotas. In Figure 2.3, begin again with equilibrium price PE in the exporting E country i, which is the same as Pj in the importing country. The government in j then guarantees producers in j the price PG which leads to an expansion in output from c to d. The consumer in j is not charged PE for sugar, but the new international "equilibrium" price ng and the government pays producers in j the “deficiency" between the guaranteed price and the actual price, i.e. (PG_- P3P), on the d units of output. In the figure the subsidization of producers in j is equivalent to a shift in the supply curve from Sj to Sj' and, in turn, this shifts the import demand curve from IDj to IDj'. The distributional consequences Of a deficiency payment are different from those of a tariff. From the vieWpoint of the exporter, the deficiency payment is preferable to 33 an equivalent14 tariff since both exports and prices are higher. From the viewpoint of the importer, both producers and consumers gain directly from a deficiency payment relative to free trade but the government (and hence indirectly taxpayers in general) loses. It may be demonstrated that the dead-weight loss under a de- ficiency payment is likely to be less than that under a tariff which affords equivalent protection to producers in the importing country. In Figure 2.4 the right-hand graph of Figure 2.3 is redrawn to compare a tariff and a deficiency payment, both of which result in a price to domestic producers of PG (=P})° The free-trade price would be P? and the price which consumers pay under the deficiency_payment is P2P. Under a tariff which resulted in price PG’ the price in the exporting nation necessarily would be lower than P3P (=ng), the exporter's price under the deficiency payment. This is a necessary condition since consumption in j is higher under the deficiency pay- ment than under the tariff, yet, by assumption of equivalent protec- tion, production in j is the same under both policies. Hence PI. DP ,- l in Figure 2.4. Further, since the price to the exporter is higher the exporter's price under the tariff, is marked below ng (=P under the deficiency payment and exports larger, it is necessary that the net effect on the exporter of a deficiency payment is a smaller loss relative to free trade than under an equivalent tariff. Returning to Figure 2.4 and considering now the importer only, the losses and gains relative to free trade may be listed as in Table 2.2. Gains from the deficiency payment in the importing country 14Equivalent in terms of protecting domestic producers in j. A O ‘r‘r I_: .I' I 1:... h 34 Figure 2.4 Tariff v. Deficiency Payment Importer j . 3'1 Price . ,/ x/ 1 ..._ 1\ PJ 1 \ \ “.1 M...” - “ -‘ ‘-‘- QSE QDE Quantity I—w .- “A C. .0- -‘ m "° exceed those from the tariff if (k + 1 + d) exceeds m. Since m is a transfer payment from the exporter, the dead-weight loss under a deficiency payment is shown to be less than under a tariff, but the net effect for the importer depends on the size of m, hence on the elasticity of export supply from i. This digression on deficiency payments versus tariffs helps to explain why some importers use a combination of quotas, tariffs and deficiency payments. An example is the U.S. Sugar Act under which deficiency (conditional) payments were made to producers, imports were subject to quotas and there was a specific tariff. Similarly the EEC 35 Table 2.2. Gains Under Tariffs and Deficiency Payments Relative to Free Trade for an Importer Policy Deficiency Tariff Item Payment Government Revenue -(a+b+e+f+g) +c+h+m Consumer Surplus e+f+g+h+k+l . -(a+b+c+d) Producer Surplus a a Total -b+h+k+1 -b-d+h+m combines its variable levy on imports with a quota on domestic produc- tion and guarantees certain prices through subsidization of exports. Some combination of policies may achieve a given target with a smaller "net loss" relative to free trade than a single such policy.15 The final set of policies to be considered in the context of modeling are forms of export restriction or producer cartels whose objective is to raise the international price of a commodity in order to increase returns to the exporting countries. The International Sugar Agreements were weak forms of cartel. Assuming that exporters have a sufficient community of interest to agree upon, and maintain, restrictions on exports, the gains and losses will be very similar to those from an import tariff or quota, but the distribution of gains or losses is different. Figure 2.5 is completely analogous to the import-tariff diagram, Figure 2.1. The exporter imposes a tax, equal to (n-p) in the central graph, the revenue from which, equal to neAp, passes to the exporting country's government. When there is more than 15See Josling, T. E. (1969). "A Formal Approach to Agricultural Policy," Journal of Agricultural Economics, 20, (2), (May), pp. 175-192. 36 Figure 2.5 An Export Restriction Scheme LL" 0175231.). 1.13.9.9 .LIEPELE‘iL-i Price, 5 (Pan-co-‘I one exporter, agreement upon a uniform export tax is not as likely as upon a minimum export price or a given export supply (quota). Suppose the cartel agrees the minimum export price n or the quota qT on exports, the effect will be exactly as in the case of the tax already discussed. In Figure 2.5 the trade graph also shows the marginal return from import demand, which is labeled MRIDj. The exporter maximizes profit by equating export supply, assumed to be the marginal cost of produc- tion, with the marginal return from exports. In the diagram the tax (or other policy) resulting in exports qT maximizes the exporter's profits. The approach to modeling such a cartel in this research has not been to impose quotas on exports but to place a uniform tax on exports which leads to the same restriction on output as a quota. 37 A word of caution is in order. The gains and losses here described are in a static context. Although time enters the supply and demand functions, through its influence on technology and growth in population/income respectively, gains and losses have been measured at a single point in time. The dynamic gains or losses from trade in sugar lie outside the context of this research but may also be important. The Context of Previous Research The research, which has now been broadly outlined, is an exten- sion and synthesis of several previous works. The spatial-equilibrium approach to modeling the sugar market is based upon the work of Thomas 16 Bates in this area. The controversy in Britain on the desirable size 17 of the domestic sugar industry was an impetus to making some more fbrmal calculations on this subject for the EEC as a whole. The con- 18 clusion of Sanchez, that the U.S. Sugar Act raised rather than lowered the free-market price of sugar, provoked the testing of this hypothesis. ‘9 and R: H. Snape20 on gains Most importantly, the works of Harry Johnson to developed countries from freer trade in sugar and the similar work of D. Gale Johnson21 for the U.S.A. provoked attention in the present 16Bates, T. H. (1965). "The World Sugar Economy and U.S. Supply Policy." Unpublished Ph.D. dissertation, University of California, Berkeley. 17Sturrock, F. G. (1969) "Su ar Beet or Sugar Cane," Journal of Agricultural Economics," 20, (1),?January),p .125-132. 18Sanchez, N. (1972). "The EConomics of Sugar Quotas." Unpub- lished Ph.D. dissertation, University of Southern California. 19Johnson, H. G. (1966). Economic Policies Towards Less Developed Countries. (New York: BrookinQSTInstitutTon). 20Snape, R. H. (1969). "Sugar: Costs of Protection and Taxation," Economica, 36 (141), (February), pp. 29- 41. 21Johnson, D. Gale (1974). The Sugar Program: LargegCosts and Small Benefits. (Washington, D.C.: American Enterprise Institute). 38 research to gains and losses accruing both to exporters and to importers from alternative policies. Finally, the uncertainty concerning new U.S. Sugar Acts, changes in EEC sugar-policy following enlargement of the Community and the current interest in cartels to protect exporters of raw products were all motivating influences. The report now turns to the estimated supply and demand relation- ships, the discussion Of which occupies Chapters III to VII inclusive. Attention will be refocused on the whole model again in Chapter VIII. '1. CHAPTER III U.S. DOMESTIC BEET-SUGAR SUPPLY Introduction In this chapter the objective is to derive supply functions for U.S. beet production and hence determine the kind of price responsive- ness to be expected. In the introduction the structure and recent history of the sector are discussed as a foreword to the development of the econometric procedures of the second section. There follows a third section which gives the results by region and finally there is a brief summary. The production of sugar-beet in the U.S.A. dates from the late 19th century. The first successful factory was constructed in Alvarado, California, in 1870, although four unsuccessful factories had been con- structed between 1838 and 1856.1 About the turn of the century there was a considerable expansion, so that by 1899 production was 82,000 short tons of sugar per annum as compared with less than 2,000 before 1888. By 1906 domestic beet sugar production was 517,000 tons raw value and exceeded domestic cane sugar production for the first time. By 1920, beet sugar represented 48 percent of total domestic sugar produc- tion (including Hawaii, Puerto Rico and the Virgin Islands). However, beet only maintained an approximately 40 percent share of domestic production until the recent expansion of the 1960's and since 1968 has 1Baiiinger, R. A. (1971). "A History Of Sugar Marketing,“ Agricultural Economics Report No. 197, U.S. Department of Agriculture, Washington, D.C. 39 4O regularly been more than half of the total U.S. production of sugar.2 Figure 3.1 demonstrates the relatively steady growth in beet production since 1900. Actual production in 1973 was 3,209,000 raw tons or approx- imately 53 percent of all domestic sugar production and 28 percent of U.S. consumption. The production of beet is widely scattered throughout the country, as may be seen in Figure 3.2. Over time the industry has become more and more concentrated in certain regions, reflecting the climatic and soil differences which exist rather than the possible distributional advantages which are small.3 Production in Maine and New York states existed in the 1960's as shown in the figure, but no longer exists. There were 36 beet Sugar factories operating in 1901 and this number had climbed to 97 by 1920. Because of the economies of scale in processing, the number of factories had declined to 54 by 1974. Table 3.1 presents the distribution of factories by individual state together with the daily slicing capacity of the factories in tons. Daily slicing capacity is only an approximate measure of "capacity" since it is not independent of the number of hours per day that the factory is operated and length of the operating season. As may be seen in Table 3.1, California has the greatest daily Slicing capacity with ten factories, followed by Colorado with its ten factories, Minnesota with its five factories and North Dakota with its three factories. Because the beet cannot be stored for long periods due to respiratory losses of sucrose, the manufacturing season closely follows the harvesting period. In all 2These and other data from U.S.D.A. "Sugar Statistics and Related Data," Volumes I and II, Washington, D.C. 3Walter, B. J. (1972). "The Wholesale Pricing System for Sugar," unpublished Research Report, University of California, Berkeley. 41 ¢ Qua. 2.2 6mm. 0mm. 25. 5mm. ONQ 2m. :02 cacm 23m 623 22.5.32... ¢ is. mzch #823 nzmx >H mmL< HRH mmL< ..u. 43 .u.o mNo—_N mm_mm_ mmmcmp mopmmp mn_mmp mmmmnp mcmoop mo—Nmp emowcp Ommov_ oomva oomva oommvp Nmmocp Nomm¢_ Ommmv_ mkoom_ oomomp oovmmp _uQOF m~-en v~-m~ Om oo-mo Ne-me owumo ¢m-mo moumo oeumm mm-mm mmumm omnmm mmnvm vmnmm mmumm —muom omuae mvnmv mvusc .couocwgmmz .cowumwuomm< mgwxogo Lomsm ummm “mugzom .mkmp mm: :_ :owuuagumcou Love: owns» ¢ comm - come. comm come - mNmo cove oommp - ooom Doom OOONN oomm_ - oowm - cmmem como~ ooppe come Amy ANV AFV A_V A_v “my Amv Acv ANV .Amv Amv A_v AQV Ao_v Ao_v A_v oovm - mmmm_ OmNN come - comm omoo oon - comm comm cemp_ comm, - comm - ooovw CONNN agape oowe ANV A_v ANV Amy “my Rev Ao_v co.“ - m~m0_ ammo comm - Omoo m~_o ooom ooce o_mm ommm oomm_ oooo_ - oocm OO¢~ coomw mNNmN oo_o¢ oo~¢ OONN - mmmofi ammo comm - Omoo ooom ooom oooq o_mo cmum comm, come. oooq owmw ooem cmwow owvmw owmmm oowc _ o. o, oooc - ammo. comm coco - mumo cake oocm oooc comm owow oomm_ oooo_ owow - oovm omom_ oeunw oopoe cam. m _ AP. omqo - coco, mmoo owow - mxmm ooov omow 000W 88V comm comm. oomo_ - ooqw m¢m¢_ ooemm omooe _ _ A. Av omco - coco, m_om com? m~mo oom¢ Como. coma cmvm_ ookm - oovw mmfimm ooemm owomm 3 I. OmNo - Ompm. m_mm oow_ coco oooq cowo_ cmqm oom~_ Gama - comm ccoep cemmw oovgm Amy . ANPV comm oo__ omvm ownm oom_ 00mg Oman comm ammo cmpmp oo_m - oom~ comm. omkom ammom Amy Am_v Ao_v ookm oo__ some oo_m oom_ omoq ommm cowav comm oooop coca - oom_ emmv_ covmm oomom m OONm oo_p oomo 0mm“ cow, Ommm Ommm oom__ comm come. OOMOW - oom_ cmmm_ ooqmm oohmm o oakm oo_. oowo Ommk oom_ omom ommm comp. comm oomo_ cowoW - oow_ ommm. oovmm oonN N oakm com comm 0mm“ com, Omam ommm oom__ ooow 00mm oom_w oomF oowF cmmmw ooemm cem¢~ m m . owsw com comm omkk oo~_ OmNm Ommm oom__ comm Nwmw oome com. oom_ coke. ocemw commm m v a Omcm com com? owe“ OONF omkm ommm oom__ coca comm oooc_ cow. oom— oo~¢_ ooqmw ~o_m~ ANV Amy Aev Aopv Am_v ocwm com CONN OMVW CORP oocm Ommc oom_F oooo_ some cock. oo~_ oom_ oowmw ommom oomNN m m oowm com OONN Om_¢ CON. comm omme oom__ coco. come oom~_ oo~_ oow_ omwQW ammom mNORN N comm coo OONN ompo oo~_ omNM cmmv oom_w coco. come owmxw com. com. mkom_ omNOm mNMcN o m. comm com CONN Om_m co“. 000m ommv 000m. coco, some cmmo_ oo~_ oom_ mfimq_ omwom mmmom Amy A_V A_V “my A_V A_v Avv ARV Amy Amy A¢_v A_v A_v Amy Ao_v A_Pv .»z .?3 .63 .0: .xp am .Lo .50 oz >2 .mz .u: .c: .,z .m: .cx .a~ .u~ .09 .au .~< mmwuwucmeu vcc mwtofiumm Lcdzm umwm _.m 033. 44 states except California and Arizona, planting occurs from February to June, harvesting from September to November and manufacture from October to February. In California there is a spread of seasons, but, in general, planting occurs in the Autumn and harvest in the Spring or early Summer. In Arizona, planting and harvest may be in the Spring or Autumn depending on altitude. The economic structure of the industry is one of l2,000 farmers providing the 54 factories with the beet from which the manufactured sugar is distributed through a smaller rumber of wholesalers. The fac- tories are either owned by relatively large sugar companies, e.g. Spreckels in California or the Great Western Sugar Company in the West and Mid-Nest, or, less commonly, they are owned cooperatively by the farmers themselves. In the latter case the supply of beet may be more price inelastic since the growers are obliged to provide certain quanti— ties of beet for a duration of years rather than having annual contracts. In the more usual case of factories owned by sugar companies, the annual contracts, which are made before planting, cover both quantity to be supplied and the proportion of total returns from the sugar to be pro- vided to the farmer. Thus in 1973, for example, the grower received an average 65 percent of the returns (excluding government payments) and the processor 35 percent. These proportions have been relatively constant since 1950. The size of farm, like that of the factory, has been rising quite rapidly. In 1950, 37,328 farms planted an average of 27.l acres of beet. In 1960, 24,2l9 farms planted an average of 40.4 acres of beet. By l973 only 12,486 farms were supplying beet, but each had an average of 102.3 45 planted acres. The advent of mechanical harvesting prior to l950 and the development of mechanical (circa l950) and chemical (circa 1960) weeding and monogerm beet (circa l960) were the technical advances nec- essary to this increase in farm size. The inducement for such innovation as the increased price of labor which rose from $0.888 per hour in 1950 to $2.455 per hour in 1973.4 Since l934, beginning with the Jones-Costigan Act of that year, quotas for domestic beet production have been fixed annually in relation to total U.S. sugar consumption requirements, and similarly such quotas have been allocated for domestic cane production and for imports. The basic quota for domestic beet production was l,800,000 tons in l948, following the Sugar Act of l947. The quota was slightly increased in a l956 amendment of the Act and was substantially increased in the l962 amendment. Further small changes occurred in the amendments of l965 and 197l. In l973 the basic quota was 3,692,000 short tons raw value. When- ever beet production was deemed likely to exceed the quota, acreage allocations were established by the Secretary of Agriculture on a state- by-state basis, these allocations being some proportion of the average acreage of the last five years. Such acreage limitations were in force in l955-l960, l965-1966 and l970. While such "proportionate shares" did not have the force of law, the inducement for individual farmers to comply was the so-called "conditional payment" or government subsidy which was the surplus obtained from the small import duty on sugar once the costs of the Sugar Program had been deducted. Such a conditional 4See, Hayami, Y. and Ruttan V. w. (l97l). Agricultural Develop: ment: An International Perspective. (Baltimore: Johns Hopkins Press? for a dfiscussion of "induced innovation" in beet production. ... o . w . c .!u 46 payment was $2.00 per ton of beet in 1973 when the direct payments from processors averaged $31.66 per ton. Such a small subsidy is no great inducement and one must conclude that producers and processors saw the price advantage of the Act's protection against imports as an additional incentive for compliance with periodic restrictions. The beet plant is a biennial, but is harvested after one season's growth. Given adequate factory capacity, the delay in response to an increased price or quota for sugar beet is, in consequence, a minimum of nine months and a maximum of 21 months. For example, the amendments to the Act of 1962 occurred in July which resulted in a delay until the following year's planting time (February-June) before farmers in most regions could respond and until the end of 1964 before there was an increase in beet sugar available for consumption. The delay in this case was at least 18 months. Should factory capacity be limiting, the delay may be considerably more, depending on whether the new facilities are an expansion to existing ones or completely new. Since the Jones- Costigan Act of 1934, the development of new processing capacity has followed increases in quota more closely than increases in price, as nay be seen in Figure 3.3 for the period since 1948. Price has been relatively stable, except for a small peak in 1963 and a large rise in 1974. By contrast, quota rose sharply in 1961 and 1962, to be followed by a rise in capacity in the 1962-1966 period. The above introduction has given the background necessary to the development of the econometric models which are introduced in the next section. III‘I-i 47 _m mvm~ P @— Om— omp oar “ me> m“ NR me mm mm on mm «m ooom muHma om \.\ comm \ o ooom . comm x. >H~u 3mm mac» we mucmmzogp cw muozo ummmm om xco> :wz e_ma scan .84 can mucmu cw mu_ca 2mm camam pmmm .m.: com xmwumawu ngpumu use mowed .muozd, m.m mcamwu ooN xco com mco» ncmmsogh cw »u_umamu 48 Procedure Regions and Crops In 1974, sugar beet production in the U.S.A. fell an estimated 9.3 percent while the price of sugar, in response to world conditions, climbed during all but the last month of the year. In the immediately preceding years the price paid to beet growers had risen parallel with the index of farmers' input prices, as intended in the Sugar Act. The decline in beet production in l974 cannot be attributed to a decline in the nominal price of beet but to a decline in the price of beet relative to the prices of products which compete for the same agricul- 5 6 tural resources. The works of Just on California and Storr and Warnken fOr the whole U.S.A. attest to the importance of inter-crop competition in determining sugar beet production. Because beet production is so widely distributed, there is no single crop with which it competes but the situation varies by region. In consequence the U.S.A. has been divided into four regions in this study and a separate supply function has been estimated for each. In addition, a function at the aggregate level has also been estimated. The division into four regions was already shown in Figure 3.2. The states in each region are as follows: I. Michigan, Ohio, Minnesota, Iowa, North Dakota, Illinois, Indiana, Wisconsin, New York, Maine; II. Colorado, Kansas, Wyoming, Texas, Nebraska, Montana, South Dakota; ' 5Just, R. E. (1974). "Econometric Analysis of Production Decisions with Government Intervention: The Case of California Field Craps," Giannini Foundation, Monograph No. 33, University of California, Berkeley. 6Storr N. and Warnken, P. (1974). "The Location of Sugarbeet Production in the U.S.," unpublished manuscript, University of Missouri, Columbia. 49 111. Utah, Idaho, Oregon, Washington; IV. California, Nevada, Arizona, New Mexico. Any such division is rather arbitrary, but the reasons for this particular allocation were mainly geographical location, similarity of competing crops, altitude and presence or absence of irrigation. The Mississippi presents a natural division of the country into a Western portion, where irrigation is practiced, and an Eastern portion, where beet are not irrigated. The states to the East form the first region, notably North Dakota, Minnesota, Michigan and Ohio. There is a central belt of production running from South Montana through Wyoming, Colorado, Nebraska and Kansas. To this belt has been added the small beet-growing area in Northern Texas to give the second region. The third region could be called the Pacific Northwest, namely Washington and Oregon, with the addition of Idaho and Utah. The fourth region comprises California, which is the most important beet-producing state, together with the minor areas of Arizona, New Mexico and Nevada. The four-region division is a simplification of the eight-region division used in the cost surveys conducted under the Sugar Act. In these cost surveys the eight regions were not always consistent with state boundaries, while the four regions and the available time-series data refer to states. Consequently our regions are only approximately consistent with the regions in the cost surveys, but the latter provide very interesting information on the other crops grown by sugar beet producers, as shown in Table 3.2. The Agricultural Stabilization and Conservation Service (ASCS) regions are denoted 1,. . .,8 and our regions I,. . .,IV. In all regions feed and foodgrains dominate the system, with sugar beet representing from 18 to 34 percent of total cropland. Reviewing Table 50 .00000 .0 0n umuw>oca 0~000x wu_>gwm 00000>Lomcou 0:0 :ovum~wpwnmum PmL=u_:u_g0< 0n 00>L00 0000 ”mugzom 0.00_ 0.5005 0.005 0.050 0.005 0._50 0.005 0.000 0.005 0.000 0.005 0.500 0.00_ 5.005 0.005 0.500 campaocu 50000 0.05 5.0 0.0 0.0 5.0 0.0 0.0 0.0 0.0 0.55 0.0 0.- I I I I I I I I I I I I L.050 0.05 0.000 I I I I I I 5.0 0.00 I I I I I I 000000 .000: 5.0 0.00 0.0 0.— I I I I I I I I I I I I 0002 go 000;; 0.0 _.05 0.0 0.0 5.0 0.0 0.0 0.00 5.0 0.50 0.0 0.00 0.0 5.5 I I 5.0 0.0 0.0 0., 0.0 0.0 5.0 0.05 Lmzuo 0.5 0.00 I I I I I I I I I I 5.0 0.0 0.00 0.005 0:000 0..0 000000 0.0 5.00 I I I I I I I I I I I I 0._ 5.0 00000200 .0000 I000> 0.0 0.5 0.0 5.00 0.0 0.00 I I I I 0.0 0.0 0.0 0.00 I I 00000000 0.0_ 0.5_ 0.05 0.05 0.55 0.5 0.0 0.5 . 0.0 0.0 5., 0.0 0.0 5.00 0.5 0.0 5.0 0.00 0.0 0.__ I 0.0 0.0 c0000 00: Ib.mFIkmqm0m 00001:0uwn mw011 0.0m 0.m0 0. . . . . 0.0 0.m0 0.0 0.0 mw_um_< 0.00 0.00 0.50 0.00 5.00 0.00 0.05 5.00 5.0 0.00 _.5 0.0 0.5 5.0 0.0 ~.0_ _.0 5.0 5.5 0.0 5.0 0.00 0.0 0.5 Lwcuo _.0 0.00 - I I - - I EN 32 0.5 0.0 I I I I 520.80 250 5.0 0.005 0.0 N.N_ 0.05 0.00 0.00 0.05 5.0 0.0 0._ 0.0 0.50 0.00_ 0.0 0.5 005L00 week 0.0 0.50_ _.0 0.50 5.00 0.005 0.0 0.0 N.0_ 0.00 0.00 0.05, 0.00 5.000 5.0 0._0 000;: 0 0.0 5.05 0.0 0.00 5.0 0.05 0.5 5.0 0.0 0.00 0.0 0.55 0.0 0.00 000000000 0000 5.0 5.505 0.5_ 0.~5 0.0 0.05 0.05 0.50 0.50 0.00_ 0.00 5.055 0.0 0.0 0.50 0.005 cgou 0.00 N.0_0 5.F0 0.005 0.00 0.F5_ 5.00 0.000 0.05 0.00 0._0 0.00— 0.05 F.000 0.00 0.05 0000;0000 a mwcu< a mwgu< 0 mwgu< x mwgu< 0 Immgu< n mugu< 0 mogu< a mmgu< 0 5 0 0 0 0 N p 0000 >0 HHH HH 51 3.2 by ASCS region, in region 1, Michigan and Ohio, corn and dry beans are the chief alternative crops while in region 2 wheat and barley pre- dominate. In estimation for our region 1, corn and wheat were the con- sidered alternatives. In regions 3 and 4, corn, wheat and sorghum are important while in region 5 the chief alternatives are corn, barley and alfalfa. Since the price of alfalfa is highly related to that of corn, for region II corn and wheat were the considered alternatives. In region 6 the chief alternatives are wheat and barley while in 7 they are corn, alfalfa and potatoes. Wheat, corn and alfalfa were con- sidered for inclusion in estimating our region III. In region 8, which is also our region IV, alfalfa, cotton, wheat and barley are important as well as a whole range of minor crops. In this region alfalfa, cotton and corn (as a proxy for all cereals) were considered in estimation. While certain "alternative crops" have now been delineated, it should be noted that just because a crop is a significant proportion of crop- land on a sugar beet farm does ngt_guarantee that it competes for resources with sugar beet--complementarity is, of course, also possible. Models In specifying the model of supply; the quantity of beet forthcoming nay be expected to be a function of past and present sugar prices, the prices of inputs, the recent prices of competitive products, technology and governmental programs; i.e. (3.1) Qlt = fl(Plt” . "Plt-k; Pit; P2t,. . °’P2t-k; T; Gt) where: 01 = output of beet, u 52 v ll 1 the price of sugar beet Pi = the price level for inputs P2 = the price of a competitive crop T = technology G = government programs, t = a time subscript. Simplifying Equation (3.1) by reasonable assumptions, Pi may be taken as the U.S.D.A. index of input prices which excludes labor since that is generally a family resource. Second, take the price of the alternative product at time t-l, i.e. P2t-l’ to be representative of expectations concerning this variable. The actual alternative will be found by esti- mating with each of the alternatives discussed above. Thirdly, take time as a proxy for technology, since the change in technology has been relatively constant through time. There now remain two problems with respect to the specification of the variables in Equation (3.l), namely the kind of lag to use with respect to the price of sugar beet, P], and how to incorporate government programs, G. These will be addressed in that order. It has already been noted that there are two delays in the response of sugar beet production to higher prices, namely a delay of 9-21 months at the farm level and a possible delay of 2-4 years at the processing level. Both delays may be expected to be of the "inverted v" kind rather than decreasing monotonically through time. However, quantity supplied depends on processing capacity only in years when farmers desire a rapid expansion of output to a new peak level. By observing the tons of beet processed per ton of slicing capacity, such capacity was found to be 53 important in determining output only in 1963-1964 and 1972 out of the last twenty-five years. If 1972 were rated at 100 percent of capacity, 1974 would be only 77 percent of capacity, indicating considerable slack for 1975, especially as several new factories are under construction. Therefbre, because capacity is only crudely measurable, rarely limiting and currently sufficient, it will not be directly included in the supply specification. It will be indirectly included, however, since the data will be utilized to determine the most likely overall lag structure. Turning to the on-farm response, the expectation is that price at t-l will be most important and that current price, Plt’ and prices in the more distant past, P11?2 etc., will be of lesser importance. Using Jorgenson's "rationally distributed lag," which has Plt’ Qlt-l and Qlt-Z on the right-hand side, allows the data to determine the most likely overall lag structure with the possibilities ranging from the geometric lag to the Pascal lag.7 Government programs consisted of periodic enforcement of "pr0por- tionate share acreages,“ but these shares have not always been binding, i.e. they have not always limited production. A comparison of planted acreages with the shares, (which were in existence in the years 1955-1960, 1965-l966 and l970), shows that in only three years (1958-1959 and 1970) were they binding in regiOn I but in most of the years in other regions. Given the two conditions, with and without shares, two models become appropriate. When shares are in force and binding, supply is a function of product price, input prices, technology and share acreage, i.e. 7Griliches, Z. (1967). "Distributed Lags: A Survey," Econometrica, 35. (1). (Jam), pp. l6-49. ‘3 54 (3.2) Qlt = f2 (Plt" . "Plt-k; Pit; T; HAt) where: HA = the proportionate share in hectares. When shares are not binding or not even in force, supply is now a func- tion of the price of competing products as well as the beet and input prices, i.e. (3.3) P = f3 (P T Qit lt” ° "Plt-k; it; ‘ PZt-l) where: P2 = the price of a competing product. The specification of Equations (3.2) and (3.3) may now be made explicit by assuming a logarithmic form for all variables except time (year) and incorporating the rational lag system. The equations become, respectively, a.I + a2 log Plt + a3 log Pit + a4T +.a5 log HAt (3.4) 109011: I a6 I°9 Qlt-l I a7 I°9 Qlt-2 I I°9 Et4 and (3.5) 10901" = b + b 1 2 log Plt + b3 log Pit + b T + b 4 5 ‘°9 PZt-l I b6 '°9 Qlt-l I ”7 1°9 Qlt-2 I 1°9 Et5 where: E124 and Et5 = random disturbances. It is now convenient to combine the two equations into a single equation, since the number of observations on either model alone is too low for 55 reliable estimation. The necessary assumptions are: 1) that the co- efficients of the input variables are not affected by proportionate shares, i.e. a3 = b3; 2) that technology is independent of proportionate shares, i.e., a4 = b4; and 3) that the disturbance terms, log Et4 and 109 Et5’ are independent. Define the binary variable Z, such that Z is 1 when the share is binding and 0 otherwise. This variable is used to a) shift the intercept, b) to change the effect of Plt and c) to introduce HAt and delete P2t-l when shares are binding. Equations (3.4) and (3.5) become together (3-6) 10901t = C1 + CZZ + C3 log Plt + C4Z log Plt + C5 log Pit + C6T + C7 Qlt-l I C10 I°9 Qlt-Z I 1°9 Ect 2 log HAt + C8(l-Z) log P21}1 + C9 log where: logEct = a random disturbance. This is the basic model to be estimated with aggregate data for the four regions for the years 1950-1974. Estimation The estimation of Equation (3.6) is not simple because the distur- tmnce is very likely to be autocorrelated due to the two lagged dependent \mriables on the RHS. OLS estimates would be biased and inconsistent. The alternative methods of estimation are instrumental variables, Gupta's two-stage procedure beginning with instrumental variables8 and maximum 8Gupta, Y. P. (1969). "Least Squares Variant of the Dhrymes Two- Step Estimation Procedure of the Distributed Log Model," International EEEQnomic Review, 19, (1), pp. 112-1l3. 56 likelihood estimation. The simplest approach, instrumental variables, was chosen, with the possibility of extension to Gupta's second stage. Let Plt be an instrument for itself, Flt-2 be an instrument for Qlt-l and P]t_3 be an instrument for Qlt-Z' Equation (3.6) may then be rewritten: (3.7) logQ1t = C1 + C 2 + C log P 2 3 + C42 log P1t + as log Pit it + C6T + C7 Z log HAt + C8 (l—Z) log P2t—l + d9 log Plt-2 + d log P + E 10 lt-3 dt where: Edt m N(0,62). The estimates obtained with instrumental variables are both unbiased and consistent but of unknown efficiency since the degree of efficiency depends on the correlation between the instruments and the variables which they replace. During estimation it became apparent that a second aspect of government policy had been omitted leading to a bias in the estimated residuals. The Sugar Act not only caused the periodic regulation of acreage but also, when amended, changed the overall quota for domestic beet production. The expansion of quota influenced output independently of price, just as was shown with factory—capacity in Figure 3.3. The following variable was therefore defined to represent changes in quota, (3.8) quart = (QUt_1 - QUt_2)/0Ut,2 where: QUOT = the quota variable, 57 QU t quota in thousand short tons a time subscript. The variable QUOTt represents the proportional change in quota in the previous year. When quota does not change, the variable is zero and this is advantageous in making projections in which the quota may be fixed or wholly absent. This quota variable was added to the array of independent variables in the penultimate estimates, as may be seen in the section which follows in which the results of estimation are given. Results Before considering the results for the different regions, a general result concerning lag-structure will be presented which modi- fied further estimation. The result was that for all regions the coefficient of Qlt-Z was less than zero, indicating that the lag was of very short duration and hence that complicated lag structures were unnecessary. In consequence, the rational lag system was rejected in favor of the simple inclusion of three years' prices for beet on the RHS. OLS estimation then became appropriate. Whenever prices at time t and at time t-2 were considered to be of very low importance, judging by the signs and significance of coefficients, they were dropped. The results for each region are now presented. I... 58 Region I (Michigan, Ohio, Minnesota, Iowa, North Dakota, Illinois, Indiana, Wisconsin, New York, Maine) (3.9) longti 8.6475 + 0.5338 109 P + 0.3378 log P (1.658) It (1.132) It‘] - 0.6771 log PCROP,C_1 - 0.8019 log PINt + 0.0627 1 (1.074) (1.641) (6.403) 82 = 0.932 on = 2.040 N = 20 As proportionate shares were binding in only three of the observed years, such years were omitted. The quota variable was not included, as the residuals from Equation (3.9) were in no way related to it and (3.9) is not autoregressive according to the DW statistic. The significance of the coefficients in Equation (3.9) is generally low, only time being significantly different from zero at the 5 percent level. This result is not surprising as, during the 1960's, Minnesota's Red River Valley was an area of great expansion as the result of delib- erate government policy rather than a change in prices: While the separate estimation of Michigan, Ohio and states to the East was consid- ered in order to give more efficient estimates, quotas were binding in +The definition of variables for this and succeeding equations is: Q1 = thousands of metric tons of beet. = price per ton of beet for this region in dollars per short ton. * pcaoi = the USDA crop-price index, l967= 100. PCORN = the U.S. average corn price in dollars per bushel. PALF = the price of alfalfa hay in dollars per short ton in Los Angeles. PIN = the USDA index of input prices excluding labor, 1967 = 100. T = year, i.e. 1966 = 66. Z = a binary variable for share-binding years. HA = proportionate share in thousand hectares. QUOT = proportional change in domestic beet quota from year t-2 to year t-l reckoned on a base of year t-2. '*Note also that P. actually refers to the price received for the cr0p of time t-l since IBayment is spread over the succeeding year. 59 the different states in the same group of years and separate estimation was considered unnecessary. From Equation (3.9) the estimated own-price elasticity of sugar beet over a two-year period is 0.87 and the cross-price elasticity with an index of crop prices is -0.68, i.e. a 1.3 percent rise in the price of other crops will offset the effect of a one percent rise in the price of sugar beet. Input prices are important, having an estimated elasti- city of -O.80 which reflects the relatively intensive use of inputs, which sugar beet requires as compared with other crops. The estimated influence of technology at the 1974 output is +6.5 percent per annum, which appears very high, but reflects both improved yields and changes in farm structure. Region II (Colorado, Kansas, Wyoming, Texas, (Nebraska, Montana, South Dakota) (3.10) 1ogQ.It = 11.3334 + 0.4258 109 P + 0.5117 109 P (1.115) It (1.269) It'1 + 1.3930 2 log P]t_] - 10.8776 2 (1.568) (2.525) + 0.6758 2 log HAt - 0.8331 (1-2) 109 PCROPt-I (1.737) (1 248) -0.7854 109 PINt + 0.0344 T + 0.4628 QUOTt (1.612) (3.290) (1.541) 82 = 0.843 ow = 1.690 N = 23 Equation (3.10) is the full model with both share-binding and nonshare years included. There were seven of the twenty-three years in which the quota was binding, hence, with only seven observations on those variables with a Z dummy, it is not surprising that many coefficients «I; III 60 were not significantly different from zero at the 5 percent level. The alternative product price, as in region I, was an index of crop prices, although the price of corn also gave reasonably satisfactory results. The equation shows very different behavior when shares were binding from that when they were not binding. The intercept term for binding years, Z, has a very large negative value of -10.87 while the slope term for price at t-l in such years has an unexpected positive sign and a large value. In effect the relationship states that farmers in region II are mgrg_price responsive in years when shares are binding, even though they are unable to adjust acreage in such years. According to Starr and Narnken9 the farmers in this region have a diminishing enthusiasm for beet production, especially as the Great Western Sugar Company has been the chief processor in the region. How such diminished enthusiasm relates to price responsiveness in proportionate-share years is not clear, unless the farmers are risk-averse and only grow beet when its price is relatively certain i.e. in such proportionate-share years. The own-price elasticity of beet production in Equation (3.l0) is 0.94 spread over two years, but is 2.33 when proportionate shares exist. The cross price elasticity with other crops is an estimated -0.83, a similar magnitude to that in region I. The elasticity with input prices is -0.79 and the estimated influence of technology at l974 out- put is 3.5 percent per annum. The quota variable, which was included because it reduced autocorrelation, has the expected positive sign and nay be interpreted as indicating that a one percent increase in quotas for U.S. beet in year t-l leads to a 0.46 percent increase in output in year t. 90p. cit. .- ~- 61 Region III (Utah, Idaho, Oregon, Washington) (3.11) 109 Q1t = 5.3717 + 2.6273 log P (2.544) - 3. 3173 2 log P It‘2 (1.841) 1t- 2 + 4. 8769 2 + 0. 8503 2 log HA (1. 306) (2. 263) - 0. 5216 (1- Z) log PCORN - 1.1974 log PIN (1.754) t1 (1.609) t + 0.0240 T + 0.5636 0u01 (1.887) (1.539) t fi2 = 0.896 0w = 1.874 N = 23 Equation (3.ll) has more coefficients significantly different from zero than in previous equations. Proportionate shares were binding in seven of the twenty-three years in this region. The lag structure of the equation is very curious. When the sugar beet prices at t and t-l were included, they were of very low value and of total insignificance; only price at t-2 proved important. This contrasts with our expectation that price at t-l would be the chief influence on output at time t. Further, the estimated effect on price response of the imposition of binding shares is to counteract completely the effect of price and even to make it negative. Clearly the coefficient of Z log Flt-2 is too high, but it may reflect the lagk_of price response in this region in propor- tionate-share-binding years. The reason for Pt-Z rather than Pt-l being most-important is not obvious, but could reflect changes in factory capacity (not seen in the data on such capacity) or problems of crop rotation which do not allow a rpaid change in crop mixture. The estimated price elasticity in Equation (3.ll) is 2.63 which is much higher than in regions I and II and corroborates the findings of 62 Storr and Warnken10 that the farmers in Washington and Oregon are highly price responsive. The estimated cross-price elasticity with corn is -O.52 and the input-price elasticity is (a large) -l.20. Technology, at 1974 output, has a +2.4 percent per annum influence and, also at l974 output, a one percent increase in U.S. beet quota at 5-l would lead to a 0.56 percent increase in output at time t. Region IV (California, Nevada, Arizona, New Mexico) (3.12) 109 Q1t = 7.4779 + 0.4923 109 P (1.508) + 2.2745 log P (4.796) t’] t - 2.3443 2 log Pt_1 + 2.6860 2 + 0.5040 2 log HAt (1.644) (0.915) (1.728) - 0.3300 (1-2) log PALFt_1 - 1.8036 109 PINt (1.028) (3.435) + 0.0494 1 + 0.5413 0001 (6.605) (1.431) 82 = 0.895 0w = 2.197 N = 23 Equation (3.12) is the best of the regional equations from a statistical viewpoint. Several of the coefficients have "reasonable" t-values and all bear the correct signs and have "reasonable" magnitudes. As expected, the lag peaks at time t-l and when price at t-2 was included it bore a coefficient of negative sign. The influence of the proportionate share is to reduce price elasticity to an estimated 0.42, which would seem appropriate for such a yield elasticity. In this region shares were binding in eight of the twenty-three years. Own price elasticity for beet is estimated to be 2.77 and so high 100p. cit. 3 '7. 63 an elasticity is consistent with the opportunities for alternative enter- prises in California. The cross-price elasticity is only -0.33 with the price of alfalfa, which probably understates the influence of alternative product prices. Output is extremely sensitive to input prices, the estimated elasticity being -l.80. When the wage rate as a distinct argument was also included, it bore the expected sign but simply resulted in neither prices of otherinputs nor the wage rate remaining significant influences on output; hence it was not included, although wage-labor may still be important in some parts of California for weeding and thinning, unlike the situation in all other regions. Technology, as measured by time, has a 5.1 percent influence per annum when estimated at 1974 output. A one percent increase in U.S. beet quota has 8 +0.54 percent influence on output, again estimated at 1974 output. Region V (Whole U.S.A.) (3.13) 10901t = 9.9134 +(0.§g§§ 109 Plt +(;.;7;§ 109 Plt-l - 0.9081 2 log P - 3.3851 2 + 0.2960 2 109 HA (1.040) 1t“‘ (1.024) (1.196) t - 0.8585 (1-2) 109 PCROPt_1 - 1.0376 109 PINt (1.569) (2.565) + 0.0345 1 + 0.3661 0001 (1.418) (1 418) 82 = 0.921 OW = 1.863 N = 23 where: 01 = thousand metric tons of sugar in raw value, P1 = the farm price of sugar beet in dollars per ton. q i 64 Equation (3.13), covering all U.S. domestic beet, is a much more "reasonable" result than might have been expected, considering the impor- tance of different crops as alternatives in different regions. All signs and magnitudes of coefficients conform to expectations, although only the coefficients for beet price at t-l, input prices and time are signifi- cantly different from zero at the 5 percent level. The lag places 30 percent of the response to price at time t and the remaining 70 percent at time t-l. Observing individual coefficients, the overall elasticity with respect to the price of beet is 1.66 and in share-binding years this is reduced to 0.75. The elasticity with respect to land in share- binding years is 0.30. Cross elasticity with respect to the price of other crops is -O.86 and with input prices is -l.04. The rate of technological change, estimated at l974 output, is 3.5 percent per annum. A one percent increase in domestic beet quota is estimated, at 1974 levels, to give a 0.37 percent increase in output. A Summary and Some Elementary Projections In summary, the analysis of time-series with a model allowing for government interventions demonstrated that the supply of beet in the U.S.A. is relatively price elastic. An overall elasticity of 1.66 was estimated, but this disguises the high elasticity of more than 2.6 in the West and North-West and the lower elasticity of approximately 0.9 in the other two regions. 80th input;prices and the prices of alterna- tive crops were important influences. Some elementary projections will now be made to demonstrate the expected behavior of the U.S. beet sector under an array of prices. 65 These projections are not central to this report and the reader may pass to Chapter IV without any loss of continuity. Two kinds of projections might be of interest. Firstly, one might project what output w0uld have been in the absence of proportionate shares. Since these shares have not been binding in any region since l970, such a projection is not very important and will not be made. Secondly, one may project supply for each region for 1975, to show the effect of 1974 prices, and for 1985, to give a perspective on the growth of this sector. Projections for 1975 and 1985 have been made under the assumption that only the price of sugar beet changes and that other prices are at their 1974 levels, including the prices of alternative cr0ps. Projections for 1975 have been made under the assumed prices of $40 and $30 per ton for the 1974 beet crop which are equivalent to a 11 raw sugar price of 23.48 and l7.6l cents per lb, respectively. Pro- jections for l985 have been under the assumption of constant prices at the 6, 10, I4 and 18 cents per lb. levels or $l0.22, $l7.03, $23.84 and $30.66 per ton of beet. The results of the projections are given in Table 3.3 and Figure 3.4. For l975 the projected output in regions I and II are somewhat larger than f0r l974 and for region III a very slight increase in output over l974 is forecast. Only in region IV, mainly California, may a dramatic expansion be expected, of the order of +50 percent, and in this region factory capacity will be limiting. Summing the regions, nConversion based on: l) l5.5 percent sucrose in beet, of which 79 percent extracted; 2)’lOO parts of raw sugar equivalent to 93.46 parts refined; and 3) returns divided between processor and farmer on a 35:65 basis. 66 Table 3.3. Projected Outputs Year Price Regional Output Total VI Aggregate Output V t t-l I 11 111 IV 1+11+111+1v ¢/lb. 4/15. ------------ Thousand Metric Tons R.V.------------ 1975 23.481 19.79 829 851 420 11302 3230 37623 17.611 19 79 711 753 420 11302 3014 37623 l985 6 6 500 384 117 226 1208 866 1985 10 10 781 588 448 929 1747 2023 l985 14 14 1047 807 1085 2357 5297 3537 1985 18 18 1304 1021 2100 4725 9150 5368 1974 Actual 663 763 470 750 2646 2644 1Price equivalences are: 23.48 ¢/lb. E $40/ton of beet l7.6l ¢/lb. E $30/ton of beet 2Limited by 210 short tons of beet per short ton of slicing capacity. 3Limited by 150 short tons of beet per short ton of slicing capacity. NOTES: 1, II, III and IV denote regions. VI denotes the sum of the regions. V denotes the result of aggregate U.S. estimation. 67 252 25a 023 002 025 coon oooe coon coca occ— _ _ _ _ _ L _ _ F _ — _ _ _ u _ fi — — _ m=._<> 35— 555:: mzP— nz > >_ _= _ = m=._<> 35— dd \mh2uu 68 (as under VI in the table), at 17.6 cents per pound received during 1975 for the l974 crop, a 13.9 percent increase in output over l974 is fore- cast while at 23.5 cents per pound a 22.l percent increase is forecast. By contrast, the aggregate function V forecasts a 42.3 percent increase in output which must, however, be considered highly suspect even though this output was limited by a capacity constraint. For 1985 a l0 cent per lb. price (in 1974 dollars) would lead to an output approximately at the l974 level in the U.S A. At 14 cents per lb., the output would be much higher than in l974. Figure 3.4 shows how region IV, mainly California, would dominate supply at the higher prices, due to its high price elasticity. The aggregate function V leads to substantially lower estimates than does the sum of the regions, VI, and interpretation of either function at high prices should be very guarded since forecast error is higher the further from the observed mean that one moves. However, at the "reasonable" price of l4 cents (l974 value) a doubling of output over l974 is forecast by summing the regional supplies for 1985 and this is also a 6l percent expansion over the previous record of 1972. The general conclusion is that an expan- sion of domestic beet production is feasible at prices far less than the 29.5¢/lb. of 1974, namely around l4 cents per lb. in 1974 values. CHAPTER IV U.S. DOMESTIC CANE-SUGAR SUPPLY Introduction Sugar cane is produced in three mainland and two offshore regions of the U.S.A., the f0rmer being Louisiana, Florida and Texas and the latter Hawaii and Puerto Rico. The exact mainland locations may be noted from Figure 4.1. Cane may be grown wherever there are relatively high temperatures, plentiful water and a long frost-free season. Con- sequently, on the mainland, cane is located in the extreme South near large water masses which afford the necessary protection from frost. In Hawaii and Puerto Rico the absolute size of the islands limits po- tential sugar production. TEXAS LOUISIANA' FLORIDA O W C ' Figure 4.l. Cane Producing Regions of the Mainland United States. 69 11. 70 Proportionate shares have regularly been enforced to limit produc— tion in Louisiana and Florida, but such measures have not been necessary in Puerto Rico since 1952. In Hawaii shares have never been used and production in Texas is so recent a development of so minor a scale that shares/have not yet been necessary there either. The existence of shares in Louisiana and Florida greatly complicates the Specification and esti— mation of time-series models of supply, particularly as the "free-from- share" supply is of great interest for future projection. Preliminary attempts at time-series estimation resulted in nonsignificant coeffici- ents for all price variables. In consequence a cross-sectional approach was chosen, using the data from the cost surveys of the Sugar Division of the Agricultural Stabilization and Conservation Service of the U.S. Department of Agriculture. The Hawaiian output has been so stable, that time-series analysis was thought unlikely to be of value there also. For Puerto Rico, a previous attempt at time-series estimation by Choudhury1 met with little success except in finding a significant time trend. A cross-sectional approach has therefore also been utilized for both Hawaii and Puerto Rico. The minor production in Texas has been omitted from the formal analysis in this study. The primary objective of this chapter is to derive supply relation- ships for each of the four major cane-growing regions of the U.S. The procedure to be used will be to estimate Cobb-Douglas production func- tion parameters from cross—section data and then to use these parameters with time-series data in synthesizing the aggregate supply response. A 1Choudhury, P. (1967). "An Economic Appraisal of the Aggregate Sugar Supply Response for Selected Major Producing Countries," unpub- lished Ph.D. dissertation, University of Hawaii. Lit 71 secondary objective is to examine the ease of substitution between inputs in the sector, particularly labor and machinery, since unskilled labor is still an important input and minimum wage legislation under the U.S. Sugar Act has had an unknown impact on employment. Ease of input substitution will be found by measuring elasticities of substi- tution and of derived demand, using the Translog production function. The chapter is developed as follows. A brief sketch of the structure and performance of each region's sugar industry is first given, so that the more formal analysis may be set in its context. This is followed by a discussion and explanation of the procedures which have been used. The third section of the chapter gives the results from estimation and the fourth makes some projections from the estimated equations. Finally there is a summary and some implications of the results. The Sugar Industry of Eachgggjgfl_ Louisiana Louisiana has produced sugar cane for sugar since 1795. Produc- tion is confined to the alluvial soils of the Mississippi Valley. Because the watertable is high, there is no need for expensive irri- gation systems. The factor limiting the wider geographical dispersion of cane is neither water nor land but temperature, particularly the temperature in the harvest period of October-December when the cane is liable to frost damage. There were 27 Louisianan weather stations which had less than four days in November on which freezing temperatures occurred in the period 1951-60.2 Of these, six were in urban areas, 2Weather Bureau. (1964). "Climatic Summary of the United States." U.S. Department of Commerce, Washington, D.C. ~- 1" 72 three were on lakes or on the coast, thirteen were in cane-growing counties and five were in counties where cane is not a significant crop.3 Of the five "odd" counties, one grows virtually no crops and the other four have soybeans as the dominant product. In Louisiana the harvest period is October to December and the planting period, August to October of the previous year. Two ratoon crops are normally harvested followed by nine months of fallow, which results in a maximum of three crops per four years on any piece of land. The proportion of cane in its first harvest was 42 percent in 1969-71.4 The structure of the industry has changed dramatically since the 1940's. In 1948, 5,957 farms harvested an average of 49.9 acres of cane each. In 1973, 1,290 farms harvested an average of 264.5 acres of cane each. In 1948, 84 percent of farms had less than 50 acres of cane whereas by 1973 only 22 percent of farms had less than 50 acres of cane. The dramatic shift in farm size reflects economies of scale, particularly in harvesting, and such economies are demonstrated by the cost surveys of the ASCS5 and of Louisiana State University.6 The sugar mills in Louisiana are usually run independently from the primary producer and their number has declined from 59 in 1948 to 39 in 1973. Figure 4.2 presents acreage and output over the last two decades and Figure 4.3 price and production of seed. Output has fluctuated 3Data on production by county according to the 1969 Census of Agriculture. - 4Calculated from ASCS survey data. 5Agricultural Stabilization and Conservation Service, U.S.D.A. (undated). "Returns Costs and Profits Louisiana 1969-71 Crops." U.S. Department of Agriculture, Washington, D.C. 6Campbell, J. (various years). "Returns, Costs and Profits from Sugar Cane Farms." Louisiana State University. 73 .000 SHOflf TONS THOUSANDS OF ACRES RAW VALUE HARVESTED FOR CANE FIGURE 4.2 SUGAR PRODUCTION IN LOUISIANA ’ 800 700 000 500 400 300 200 1950 1955 1960 1955 1970 1975 YEAR 14 °’° OFARE“ nouns 4.3 28:22:?” me Pnooucnow or SEED-CANE IN LOUISIANA IN RELATION TO ‘3 PM“ or we“; THE PRICE or SUGAR 1w new vonx DUTY-PAID CENTS/LB. 10 I PRICE ,' ’I 8 ,.” I \ , ’ \ ,— I” \\ ”’ ' 6 -"-----~“’ut‘ ----------- 4 .0/0 ACRES m seen 2 0 1950 1955 1960 1055 1970 1975 YEAR 74 from a low of 295,000 tons raw value in 1951 to a high of 759,000 tons raw value in 1963. Acreage harvested for sugar was at a low of 203,300 in 1956 and a high of 325,200 in 1964. Acreage was limited by the impo- sition of proportionate share in all_years except 1960-62, 1964, and 1972 to the present time. Figure 4.2 gives some idea of investment behavior in those years . when controls were relaxed. Acreage reflects intended output. Following the ban on Cuban imports in 1960, acreage only climbed slightly in 1961. However, in 1962 a record acreage of sugar was harvested for seed (see Figure 4.3), reflecting intentions to expand output in response, not to price, but to an expansion in the mainland cane quota from 787,000 to 1,072,000 tons. The new cane came into production in 1964. During the 1971-74 period there has been no expansion comparable to that of 1962-64, despite the absence of shares, which indicates that neither prices nor government policy have been conducive to such an expansion. Florida The production of sugar cane in Florida began in 1928 but was of relatively minor importance until the Cuban embargo of 1960 after which there was a rapid expansion. As in Louisiana, neither land nor water is the limiting factor but freedom from frost-damage. In the vicinity of Lake Okeechobee there is relative freedom from frost, but as one moves further from the Lake (to the North) the probability of frost damage rises.7 The higher the expected price of sugar, the more dis- tant from the Lake the margin of cultivation becomes. 7Ballinger, R. A. (1972). "Economic Behavior in the U.S. Sugar Market," California Agricultural Experiment Station Bulletin No. 859. .,... 75 The planting period in Florida is September to October and the har- vesting period is November to May of the following crop year. Two or three ratoon crops would normally be taken before replanting and this is reflected in the 1967-69 average of 25 percent of cane being in its first harvest year.8 Unlike the situation in Louisiana, where smaller farms are continually being consolidated into larger holdings, sugar production in Florida has always been a relatively large-scale under- taking. The average harvested acreage per farm in 1948 was 1,464 and in 1973 was 1,952. However, whereas up to 1960 most of the cane producers were also processors, thereafter the number of "independent" cane pro— 9 The number of ducers increased to be 47 percent of the total by 1969. farms increased from 25 in 1948 to 136 in 1973 and the number of mills increased from three to eight in the same period. Figure 4.4 presents acreage and output for the last two decades. Acreage was relatively constant in the 1950's, as was output, and the rapid expansion came in the period 1961-64. Acreage harvested for sugar was 56,100 in 1961, 114,300 in 1962, 139,900 in 1963 and 219,800 in 1964. Similarly, output rose from 206,000 raw tons in 1961 to 572,000 tons in 1964. Figure 4.5 plots the proportion of acreage held for seed against time and price. Two peaks occurred in the proportion of acreage har- vested for seed. The first, in 1961, was the result, not of price but of an expansion in quota in the 1961 Sugar Act Amendments. The second peak, in 1963, was the result of the high price in that year. Clearly both quota and price have been important in causing new investment to 8Calculated from the ASCS survey data. 9Ascs, USDA (Undated). "Returns, Costs and Profits Florida 1967-69 Crops," USDA, Washington D.C., p. 5. .000 SHORT TONS RAW VALUE 1200 1000 800 400 200 14 12 10 76 THOUSANDS OF ACRES HARVESTED FOR CANE FIGURE 4.4 SUGAR PRODUCTION IN FLORIDA 300 250 ,I”" f I ' I 200 I“ ,' I \ I I \”"-\~ I ’ \‘ ’I l \ I 15° I, \\I’ ACRES I} I ‘00 \I; \ / TONS I so ,,” 1950 1955 1960 1965 1970 1975 YEAR FIGURE 4.5 THE PRODUCTION OF SEED-CANE IN FLORIDA IN RELATION TO THE PRICE OF SUGAR AREA HARVESTED FOR SEED PRICE OF SUGAR IN NEW YORK DUTY-PAID CENTS/ LB. ’1 I ’l / \Pmce ’I\ ”’ ’ \\ ”a’ ’I \-_”"” °/o ACRES /m SEED 1955 1960 1965 1970 1975 YEAR 77 occur. The result of a higher proportion of seed-harvesting in one year is an expanded acreage in the next, as is shown in Figure 4.4 for 1962 and 1964: Just as in Louisiana, the lag in investment is of one to two years' duration, depending on the exact timing of the stimulOs in relation to the crop year. In the period since controls were removed in 1971, there has been a small increase in acreage but nothing of the order of the earlier expansion. Hawaii The production of sugar began in Hawaii early in the 19th century and was encouraged after 1876 by duty-free access to the U.S. domestic market. Production first exceeded one million tons in 1930-31 and has remained at approximately that level ever since.10 The islands involved in sugar production are Hawaii, Oahu, Kauai and Maui. The area of coastal land suitable for cane growing is limited and there is competi- tion for land at the margin for urban, military and recreational uses as well as for livestock production.11 Hawaii is unique among cane-producing regions in having year round planting and harvesting of the crop. The time between planting and harvesting is two years and ratoon crops are also allowed one and a half to two years' growth before harvesting. In consequence of the long growing period, yields per acre in Hawaii are the highest in the world, being over ten tons of raw sugar per acre. It is usual to take several ratoon crops before replanting. In 1973, 393 farms harvested cane as 12 compared with 786 in 1951. However, these farms are mostly not 10Ballinger, "History of Sugar Marketing," 92, 915,, p. 9. 1'Ballinger, "Economic Behavior in the U.S. Sugar Market," 92, git, learlier figures are not comparable. I. -..1 n 5.. A: — on: 78 independent, but owned by the mills. In 1967-69 only 5.8 percent of sugar was produced by independent growers and 94.0 percent was produced by the farms owned by the mills. The remaining 0.2 percent was the product of a few remaining adherent planters and co-producers.13 In 1969 there were 24 plantation companies in operation. The average farm in 1973 harvested 275 acres of cane which may be compared with the average of 139 acres in 1951; i.e. there has been a gradual upward drift in farm size. Figure 4.6 presents acreage harvested and output in the last two decades. Acreage growing is approximately twice the area harvested, ex- cept when a strike (as in 1958) or inclement weather interrupts the harvest. Figure 4.6 is of interest only in demonstrating the remark- able stability of the Hawaiian industry over this period. The chief question concerning the industry is not by how much it could expand its output, but at what price would a decline set in. Since Hawaiian pro- duction has not been restricted by proportionate shares, any change in acreage is a response to prices, either of inputs or of outputs or of other competitive land-using activities. Although the acreage climbed slightly in the late 1960's, there was no dramatic change in acreage or output following the high sugar price of 1963. Whether the current high price of sugar may have any influence will be a matter for dis- cussion later in this chapter. Puerto Rico Production of cane sugar in Puerto Rico dates from the 19th century, but after 1900 there was an influx of U.S. investment which 13ASCA, USDA (Undated).. "Returns, Costs and Profits of Hawaiian Sugar Plantations, 1967 to 1969 Crops." USDA, Washington, D.C. 79 .000 THOUSANDS SHORT TONS OF ACRES RAW VALUE HARVESTED FIGURE 4.6 SUGAR PRODUCTION IN HAWAII ‘1300 1200 130 1100 120 1000 110 ~~---~~ ~~.--.‘ _“ ‘ ACRES 900 100 | I I | 800 90 “ 1950 1955 1960 1965 1970 1975 YEAR .000 THOUSANDS SHORT TONS OF ACRES RAW VALUE HARVESTED 1400 400 ‘s s~‘ 1200 FIGURE 4.7 SUGAR PRODUCTION IN PUERTO n1co ~~~§O"\ \’o \‘ "“ “ ACRES 1000 800 -.--‘ s ‘C . ~~‘ \ aoo \‘ TONS ‘1‘ 600 200 ‘ ‘o"\ \ \ 400 ’ \“s s§~ \ 200 100 O 1950 1955 1960 1965 1970 YEAR 1975 ,...-I ..... .. n- 1 n '1' . \ 'D'n ' vi‘d 80 expanded output from 49,000 tons in 1900 to 994,000 tons by 1939. Puerto Rico has never paid any duty in the U.S. market which gave it a distinct advantage over Cuba in certain years. Production reached a peak of 1,372,000 tons in 1951, but since 1961 has declined in every succeeding year to reach a low of 255,000 tons in 1972. In 1973 there was a slight reversal of the trend, with 291,000 tons produced. The reasons for Puerto Rico's declining production are not well established. Rising wage rates and the slow adoption of mechanization are often cited.14 Wage rates have risen because of the free entry of Puerto Rican labor into the U.S. mainland. Mechanization has been slow because of the rolling topography of some of the traditional cane lands. At the margin, milk production may be more profitable than cane 15 production. Attempts to explain Puerto Rican production with a time- series model16 and to project future output with the mechanistic approach of Markov Chains17 have both been relatively unrewarding. The formershowed only a significant time-trend and the latter failed to predict the substantial recent decline in the industry. Planting occurs in Puerto Rico in both the Autumn and Spring and harvesting of that crop is from December of the following year until July. Two to three ratoon crops would normally be taken on an annual basis before replanting, although more are sometimes taken. In the 14Ballinger, "Economic Behavior in the U.S. Sugar Market," 9p, git, 15Pringle, G. E. (1969). "A Temporal-Spatial Analysis of Sugar Production and Marketing in Puerto Rico," Ph. D. dissertation, University of Wisconsin. 16Choudhury, P. (1967). "An Economic Appraisal of the Aggregate Sugar Supply Response for Selected Major Producing Countries," Ph. D. dissertation, University of Hawaii. ”Pringle, gp_. g3. .. .5, .1 D I. .., '. 'DI .,. up. .'.0 Nu "In. \ ‘i 81 1972-73 crop year 2,954 farms harvested an average of 45 acres of cane which yielded 1.93 tons of sugar per acre. Comparable averages for 1950 were 16,525 farms having 23.7 acres each which yielded 3.16 tons of sugar per acre. In 1950-51 60 percent of all cane farms had less than five acres of cane and by 1972-73 this had only fallen to 45 percent. The structure of the industry, namely one of small independent growers, has not changed much in the last two decades. The number of mills has declined slowly from 36 in 1948 to 12 in 1973. Figure 4.7 traces the declining output and acreage of the industry from 1950 onwards. Note the acceleration in the rate of decline at about 1967, which may be attributed to rapidly rising wages at that time. Restrictions on acreage have not been necessary in Puerto Rico since 1952 and so output is a response to prices rather than quotas. The high price of sugar in 1963 seems to have had little, if any, influence on output in succeeding years. Procedure General Approach The reasons for using a cross-sectional approach have already been mentioned, namely the frequent imposition of proportionate shares, the low variation in output over recent decades and the comparative failure of attempts at time-series estimation by Choudhury for Puerto Rico and by the present author for Louisiana. Under the Sugar Act, cost surveys were conducted every four years in each of the producing regions in order that "fair" prices and wages could be fixed in exten- sions of the Act. The last such surveys were for 47 farms in Louisiana for 1969-71, 29 farms in Florida for 1967-69, 24 farms in Hawaii for "u- an...“ .5... ' 'nn. 4 .f .. . .. .. ‘ .“II '1 ‘1 ..g 5. :3 ‘hnf; . n! / 82 1967-69 and 35 farms in Puerto Rico for 1969-71. The general results of the surveys have been published by the Sugar Division of the Agri- 18 The data consist cultural Stabilization and Conservation Service. of the business accounts together with some physical observations such as tons of cane, area in production and man-hours of labor. The prices of individual inputs, except labor, are not recorded and therefore estimation of cost-functions corrected for input prices is precluded. The estimation of cost as a simplg_function of output would face an additional problem when proportionate shares were in force because short-run costs may rise more steeply for large firms than for small firms resulting in a totally misleading estimate of the shape of the long-run average-cost curve. Figure 4.8 demonstrates such an occurrence in the two-firm case. Each firm is assumed to reduce output by one- third under the proportionate share, from 0* to 0°. A line A* B* would be estimated under profit-maximizing behavior whereas A°B0 would be estimated under the proportionate share. When simple cost functions were estimated for each region, no reduction in unit cost with output was observed in Florida and Louisiana, which would be consistent with the situation in Figure 4.8. Having rejected cost functions, note that production functions do not suffer from the same problems since they relate physical quan- tities, rather than prices or costs, to output. The production function parameters are invariant to governmental policy. Production functions were therefore fitted to the cross-sectional data. The general model nay be written, 18Entitled, e.g. "Returns, Costs and Profits, Louisiana 1969-71 Crops," U.S. Department of Agriculture, Washington, D.C. 83 Figure 4.8 Two-Firm Example of Cost Relationships SRMC, Cost per Unit Output Output 4. . . . . + . ( 1) x01 fl (xli’ ’in) e1 where: X0 = output, X],. .,Xk = inputs, 8 = an error term, (assumed additive), do 1' Following estimation A (4.2) X01 - X01 = where: A The coefficients estimated for Equation (4.1) relate to the the ith firm (i 1,. .,n). one obtains, e. 1 an estimate. - a -3 .1 ‘ 1 . 4 hu- 1‘- .,n I 1 D...‘ .h. ' .‘," 84 average firm in the industry at one point in time. By contrast, the objective of this study is to obtain a relationship suitable for time- projection and for the whole region. Suppose the region is treated as consisting of some multiple of the average firm, one may write, (4.3) X = n X ot oit where: X0t = the regional output, - = , . Xoit the average firm 5 output, t = a time subscript n = the number of firms. A Substituting X from (4.1), one obtains oit (4.4) I =n x at oit Equation (4.4) is estimable only for the sample years and if input qgantities are known. However, assuming profit-maximizing behavior for all inputs expcept land and given the production-function parameters from Equation (4.1), one may write sectoral output as a function of land input per firm and prices of other inputs; i.e. ~ (4.5) Xot = nf2 (LNDt, Px2t" . °’kat) where: Rot = the new estimate of the sectoral output in year t, n = the number of firms, f2 = a known functional relationship, LNDt = average land input per firm, Px2t" . .,kat = prices of var1able inputs 1n year t. ' v 1" I u I . ~ 1; ~‘ . I, . 5'.- I“ “A R - ‘ u ‘I . 4 3;: [I u 9 Id "ra- '- .1 ’1 1 .. 4 I 1 2M .: i ‘ J 1 I.- o “ 85 In the jth year, where j f t, the utilization of Equation (4.5) would yield ~ (4.6) xoj = qu (LNDj, Px2j" . . ij wherezl q = the number of average firms in the jth year which are equivalent to the n of the sampled year. By allocating gll_of the sector's land to the average firm, q may be eliminated and Equation (4.6) rewritten ~ (4 6)‘ X . = f2 (Lij, sz OJ .,P ). j" xkj where: LND = total, sectoral land. Xoj is derived purely from cross-sectional data for the tth year and may be expected to differ from actual, sectoral output in the jth year due to the effects of changing technology (represented by time) and of changes in the scale of output for the whole sector; i.e. (4.7) X . - X . = f3 (T,S) 03 03 where: T = year S = a scale-measure. (4.7) will be called the "auxiliary time-series” relationship. Com— bining Equation (4.6)' and (4.7) leads to the compound relationship for year j (4.8) xoj = f2 (LNDj, Px2j” . ”kaj) + f3 (T,S) 86 where: ioj = estimated output for the sector in year j. The argument may briefly be sumnarized as follows: 1. estimate the production-function parameters from cross-section data; I. l 2. assuming profit-maximizing behavior for all inputs except land, which is exogenous, compute sectoral output as the output of the average firm as if it utilized all of the sector's land; 3. use the time-series of residuals from the difference between cross-sectionally estimated output (2) and actual output to estimate the effects of time and scale; 4. combine (2) and (3) to give estimates of any year's output, given land area and input-prices. Finally, one may impose profit-maximizing behavior with respect to land and synthesize a fully profit—maximizing supply which is free of government intervention. The exact form will be shown later. This approach to "synthesizing" a supply function is complicated, but aggregation bias may not be as great as might be expected since the constant of aggregation is essentially estimated in the auxiliary time- series regression. Only the results may show whether such complexity is justified. Before returning to actual estimates and elaborating on the above, approaches to estimating the Cobb Douglas and Translog pro- duction functions will be described in the next sections. Note that the Cobb-Douglas production function is being estimated for the "syn- thesizing" procedure above, but the Translog production function is being estimated only in order to find elasticities of substitution and of derived demand for inputs. 87 The Cobb-Douglas Function The form of the Cobb-Douglas function when written in logarithms is (4.13),, xoi = a0 + alxli + ,. . ., + akxki + ui (i= 1,. . .,n) where: x0 = log of output, xr = log of the rth input, u = the disturbance term i = a firm subscript. k The function is homogeneous of degree X ar and the partial elasticity r=1 O of substitution19 between pairs of inputs is unitary.2 There are several ways to estimate this function, depending on the assumptions which any be appropriate. As, in the present study, the correct set of assumptions was not entirely apparent, results were obtained and compared from different approaches. 2] is to The simplest method of estimation, attributed to Klein, equate the logarithms of the individual coefficients with the appropri- ate logarithmic cost-shares. That is -__1_" (4.14) log ar-"1E] log [ 19 Defined as Oij = 20See, for example, Henderson, J. and Quandt, R. (1958). Micro- economic Theory. (New York: McGraw-Hill). 2.IKlein, L. R. (1953). A Textbook of Econometrics. (New York: Row. Peterson and Company). 88 where: P _ . th . r - pr1ce of the r input, P0 = output price,, Xr = quantity of rth input, X0 = quantity of output, a firm subscript. “C II The assumptions necessary to this approach are that firms do not differ in the disturbance associated with the production function but only with respect to their success in equating input prices and marginal value products; i.e. it is assumed that all firms are similar and are distributed symmetrically about the profit-maximizing position. Further, since cost-shares sum to sunity, so must the Cobb-Douglas coefficients and so unitary returns to scale are imposed. Given correct assumptions, Klein's estimator is unbiased and maximum likelihood. In the present context it might be useable with the Hawaiian and Puerto Rican data but not with the Louisianan or Floridan data, since in the latter regions proportionate shares precluded profit-maximizing behavior. The second, and most common, estimator is that of ordinary least squares (OLS) on the production function itself, an additive, logarithmic disturbance being assumed. Assuming perfect competition, diminishing returns to scale, profit-maximizing behavior and all inputs variable, each firm in the industry will be exactly the same as every other firm since it faces exactly the same set of prices, and estimation is impossible. However, should one or more inputs be fixed or should firms vary in their ability in maximizing profit, there will be dif- ferences in inputs and outputs between firms. The marginal condition f0r profit-maximization for the rth input may then be written 89 (4.15) x p - pr + ar + x . + v . (i = 1,. . .,n) ri o 01 r1 where: h x = log of the rt input, x = log of output, po = log of output price, Pr = log of rth input price, ar'= log of the rtn coefficient, vr = the rth disturbance do II a firm subscript. The complete set of equations in the model consists of the pro- duction function, Equation (4.13), and the k input equations such as Equation (4.15). The problem which now arises is simultaneity bias when estimating Equation (4.13) alone, since ”i is a compound disturbance and not independent of xri' In this case "least-squares estimates of the production function parameters based on cross-sectional data will be, in general, biased and inconsistent."22 On the other hand, should the prices in Equation (4.15) be "expectations," it may be shown that direct estimation of Equation (4.13) has the desired properties of being un- 23 biased and consistent. Summarizing, OLS estimation of the production 22Kmenta, J. and Joseph, M. E. (1963). "A Monte-Carlo Study of Alternative Estimates of the Cobb-Douglas Production Function," Econo- metrica, 31, (3), (July), pp. 363-385. 23Zellner, A., Kmenta, J. and Dreze, J. (1965). "Specification and Estimation of Cobb-Douglas Production Function Models," Econometrica, .93, pp. 784-795. 90 function allows the relaxation of the assumptions of profit-maximiza- tion and unitary returns to scale, but may introduce simultaneity bias. To overcome simultaneity bias, Theil suggested a method of indirect least squares (ILS) which was further examined by Hoch and Kmenta.24 Suppose the first two inputs in the production function are variable and the others exogenously fixed. Subtract xoi (a1 + a2) from both sides of Equation (4.13) to obtain (4.16) x . (1 - a1 - a2) = a0 + a1 (Xli - 01 ) + a2 (X21 ‘ X °) X . 01 01 + . + + . + ' a3 x31 ,. . ., ak xk1 u1 Divide both sides by (l-a1 - a2) to obtain 30 al a2 (“7) Xm- ‘ W+W (X11 ' X61) + 13:5; “‘21 ' X61) + l-aI-a E arxri + l-auia l 2 r=3 1 2 which may be rewritten (4"8) xoi = bo + b1 (X11 ' X01) + b2 (X21 ‘ X01) + b3X31 + ,. . ., + bkxki + 91’ where: bO = (ao/l-a1 - a2) hr = (ar/l - a1 - a2) The ILS estimates of the production-function parameters are: 24For references see Kmenta, J. (1964). "Some Properties of Alternative Estimates of the Cobb-Douglas Production Function," Econometrica, 32, (1), (January), pp. 183-188. 91 a0 60 (1/1 +61 +6 21. br (1/1 + b] + b2), r = 1,2,. . .,k. DH II r The profit-maximizing conditions for the first two variable or "endogen- ous" inputs may now be written as: I _ = _ 1 (4°15) xri x0i po pr + ar + vri (xri - xoi)’ unlike Xr. in Equation (4.15), is simply a linear function 1 of vri since all other quantities on the RHS are constants. As long as E(uvr) = O , ILS gives consistent estimates of the br and ar coeffi- cients. In relation to the present study, land may be treated as exo- geneously determined when proportionate shares are in force or, in the short-run, due to previous decisions to plant cane. All other inputs may be considered endogenous to the firm and hence may result in simultaneity bias which leads us to utilize ILS. There remains one other problem of OLS or ILS which is not shared by Klein's method, that of cross-sectional bias. In both OLS and ILS the "inter" rather than "intra" firm regression is estimated. Should firms differ with respect either to management or other fixed but unmeasured inputs, (such as quality of land), this will be reflected in input-usage. Figure 4.9 demonstrates the situation in the one-input two-firm case. The inter-firm regression, given the firms at positions A and B, would be AB, whereas the unbiased intra-firm relationships are measured and f 25 by f1 1961, 2. One way to avert the problem, as suggested by Mundlak in is to utilize both cross-section and time-series data in an 25Mundlak, Y. (1961). "Empirical Production Function Free of Management Bias," Journal of Farm Economics, 41, (1), (February), pp. 44-56. .g. I" 92 Figure _4.9 Inter and Intra-Firm Regressions ,1 i , Output I l/////. // f2 a01 / ”// I « ”I 1 Input analysis of covariance framework which essentially gives each firm its own intercept. The model then becomes, in OLS form, (4‘19) xoit = a01' I alxlit + " ' " I akxkit + ”it where: t = year or other period of observation. Note that the intercept, aoi’ is assumed constant for the time-series while inputs and output vary over time. Mundlak's formulation of analy- sis of covariance and Equation (4.19) are essentially the same, as may 26 be seen in Kmenta's textbook. In the present study, since three years of observations were available in each region, Mundlak's 26Kmenta, J. (1971). Elements of Econometrics (New York: Macmillan), pp. 516-517. 93 covariance approach was investigated for both OLS and ILS. Finally, each year was given a dummy variable designed to account for weather variation. In summary, the Cobb-Douglas function was estimated for each region using Klein's method, OLS and ILS. Dummy variables were included to account f0r individual firm effects and the effect of weather variation. The Translog Function The Translog (Transcendental Logarithmic) production function was introduced by Christensen, et al.27 in 1971. It may be viewed as a generalization of the Cobb-Douglas function with terms both linear and quadratic in the logarithms and which approximates constant elasticity of substitution. It may be written 4.20) xoi = a0 + E arxri +-% Z 2 Yrjxrixji (i = 1,. . .,n) r—l r j where: x0 = log of output, xr = log of the rth input, Yrj = a parameter a0, ar = parameters Yrj = er i = a firm subscript. k For homogeneity of degree 2 arf the TEStr'Ctions riinj = 0 and jginj = O are necessary. IE-gddition, for output to increase 27Christensen, L. R., Jorgenson, D. W. and Lau, L. J. (1971). "Conjugate Duality and the Transcendental Logarithmic Production Function," Econometrica, 39, (4), (July), pp. 255-256. 94 monotonically with input it is required that the marginal product of each input be positive at all levels of output. The marginal product of the rth input in logarithmic form may be written in general as k = +2.. (4.21),. Mr ar r=er3xJ where: th M = logarithmic marginal product of the r input. r For the isoquants to be convex to the origin, the corresponding bordered hessian of first and second derivatives must be negative definite.28 The partial elasticity of substitution between inputs r and j may be shown to be: (4-22) Orj = I Grj I / I G I where l G 1 is the determinant of 1A 0 M1 M2 ... Mk M1 Y11+MI’M1 Y12+”1M2 °°'° Ylk+Mle G = M2 Y12+”2M1 Y22+MS'M2 ... Y2k+M2Mk I_Mk Y1k””k”1 Y2k+M2Mk °'° Ykk+ME-Mk‘ and IGVj|1s the cofactor Grj in G. Estimation of the Translog function could be completely analogous to estimation of the Cobb-Douglas function. No report of direct esti- mation of the production function was found in the literature and when 28For this and the following arguments see Berndt, E. R. and Christensen, L. R. (1973). "The Translog Function and the Substitu- tion of Equipment, Structures and Labor in U.S. Manufacturing 1929-1968," Journal of Econometrics, 1, pp. 81-114. 95 such an approach was attempted, using OLS, ILS and analysis of covariance procedures, it met with little success. The monotonicity condition was invariably not fulfilled, possibly because of the high degree of multi- collinearity which existed. 'Klein's method, when applied to the function, leads to the set of k estimating equations. k .= + .. = (4.23) Mr1 ar jg] yme1 + uJ1 (r 1, ,k) where: uji = a disturbance term, i = a subscript for the ith firm Mr = the cost-share of the rth input under profit-maximizing conditions. However, in the present context, the assumption of profit-maximization necessary to use Equation (4.23) is inappropriate. When there is a parametric restraint on profit-maximization, such as a proportionate- share on land or past, erroneous investment decisions; the cost-share on the LHS of Equation (4.23) may be rewritten as: r Rr PO X01 where: Rr = a parametric restraint and O < Rr.i 1.29 Even if only one input is subject to’restraint, the cost-shares of the other inputs will not reflect their logarithmic marginal products.3O 29Following Hoch's reasoning in Hoch, I. (1958). "Simultaneous Equation Bias in the Context of the Cobb-Douglas Production Function,” Econometrica, 29, (3), (October ), pp. 566-578. 30Note that Kmenta and Joseph (1963),gp. c_1_t. deny this in relation 96 Although the values of the R's are not normally observable, they may be estimated as the ratio of the (ILS) estimated Cobb-Douglas coeffi- cients to the cost-shares. The Cobb-Douglas coefficients, when cor- rectly estimated, are equal to the profit-maximizing cost-shares and may therefore be directly compared with the actual cost-shares to give the R values. Hence estimation of Equation (4.23) becomes feasible by first correcting the cost-shares and then using them on the LHS of Equation (4.23). Two restrictions may be imposed on Equation (4.23). Homogeneity requires the restriction .; = O, which results in only (k-l) of the set of k equations, oI-Which Equation (4 23) is a member, being independent. Symmetry requires that Yrj be restricted to equal er' Estimation is then usually accomplished with the Iterative Zellner Efficient (IZEF) method on (k-l) of the equations.31 In the present study, the share of land was deemed the least accurately measured and Zellner Efficient Estimation (ZEF) was used simultaneously on the re- maining three equations to give the desired Translog parameters. The results from estimating the two production functions are given in the third section of this chapter. Before passing to them, the precise relationship between cross-section and time-series for the Cobb-Douglas production function will be elaborated. to Klein's estimator. However.” Xri is not independent of in, where Xoi j is an exogenous input and r is an endogenous input. 3IBerndt and Christensen, gg, git, 97 Aggregation, Time and Scale The approach to aggregation was outlined earlier, but the specific functional f0rm of the "auxiliary time-series" regression, in which the residual is related to time and scale, was not given. The speci- fic forms will now be developed. Note that the procedure was used only with the Cobb-Douglas form since it would have been very complex with the Translog form. In the empirical work, inputs were divided into land, labor, machinery and fertilizer. The production function, once estimated from cross-sectional data, may be written for the whole sector in year t as: (4.25) iot = A0 + alog MCt + 810g [Bt + 810g LNDt + 110g FERT t where: Rot = estimated regional output, MC = machinery 1 LB = labor * for the average farm, LND = land FERT = fertilizer} A0 = a constant, 0,8,3,7 = estimated coefficients. Imposing profit-maximizing conditions, but assuming land to be exogenous, regional output may be made a function of regional land input and the array of input prices, i.e. ~ - II A A A - A (4.26) xot - T:fi-(Co + 5109 LNDt - OI109 PMCt - 8109 PLBt YlogPFt + u 109 USPQt) 98 where: u = G + 6 + 1. LND = regional land area, PMC = the price of machinery, PLB = the price of labor, PF = the price of fertilizer, USPQ = the U.S. sugar price, Co = a new constant t = a time subscript. Subtracting the estimated output in Equation (4.26) from actual, the time-series residual is defined as: - i (4.27) e; = x0t at where: e; = the residual, thus defined. Since C0 in Equation (4.26) is not known, as it is a newly-defined, C . o aggregate constant. it is convenient to add 1:73 to both sides of Equation (4.27) to obtain C C 0 — ~ .11. (4.28) e; +IT:EI - x0t - x0t + I1_UI Choosing a linear form for the relationship between the LHS of Equa- tion (4.28) and time and output, one obtains: Co ‘ (4.29) e; +-T:E = W1 + szot + W3T + et where: X = ot output, 99 W1W2,W3 = parameters e a disturbance term, t a time subscript. Combining Equation (4.29) when estimated and (4.26) leads to the com— pound formulation for time-series projection, (4.30) x0t = 7%uj(3109 LNDt - 6109 PMCt - 8169 PLBt - 9169 PFt + ulogUSPQt) + Q] + 02 x0t + 03 1. Equation (4.30) may be rewritten, with both output-measures on the LHS, where output x0 is now simultaneously determined with its log, x t hence is denoted R ot’ otz .. ...___,_. (4.31) x - W2 Xot - 1'0 (610g LND ot - alog PMCt - Blog PLB t t - vlog PFt + plog USPQt) + W1 + H3 T. To explain the behavior of this "compound" function and to justify the choice of a linear, nonlogarithmic auxiliary equation, Equation (4.29), it is simplest to return to the production function from which Equation (4.31) may be considered to be derived. This function is the generalized Cobb-Douglas of Zellner and Revankar.32 Equation (4.32) demonstrates this function, with time incorporated in the Solow formulation, (4.32) x01; - waot = log A + 0109 MCt + Blog LBt + ylog FERTt + 6109 LNDt + W3 T 32Zellner A. and Revankar, N. S. (1969). "Generalized Production Functions," Review of Economic Studies, §§, (2), (April), pp. 241-250. 100 where: A = a constant, MC = machinery, LB = labor, FERT = fertilizer LND = land. The returns to scale of this function are given by the "returns to scale function" (4.33) T‘(XO) = .I-WZX 0 where: r(XO) = returns to scale, r a constant, (the returns to scale at zero output), xo output. The expectation is that, given r > 0, W2 will be negative, which implies that returns to scale fall from r at Xo equals zero to zero as XO rises to infinity. To derive our "compound" formulation Equation (4.31) from the production function Equation (4.32), the marginal conditions are: w2X0 (4.34) 'Eig.= arxoe 3Xr Xr X ll one of the three variable inputs, - the input coefficient. 01 I Profit maximizing behavior would imply, by comparison, 101 a (4.35) _X_o_ arxo BXr Xr The marginal condition Equation (4.35) may be thought of as "approxi- mate" maximizing behavior, the degree of approximation depending on the magnitude of W2 and X0. In the Zellner and Revankar formulation, all input elasticities vary similarly with scale. For example, the input elasticity for land at output X0 is 7:3—7—-, where 6 is the land coefficient at zero output. 0 2 This leads to the profit-maximizing quantity of land being written, _( 5 * (4.36) LND* WW] Xo Po PLNO where: * = a maximizing value, LND = land X0 = output, P0 = the price of sugar PLND = the price of land. Because X3 depends on LND*, in making projections Equation (4.36) was used iteratively in order to determine the desired quantity of land. Summarizing this section, aggregation, time and scale are all accounted for by the estimation of the auxiliary time-series regression Equation (4.29). Approximate maximizing behavior is assumed for all inputs except land, but in the projections land allocation may also be adjusted to its maximizing level. Returns to scale are expected to decrease continuously with output in the formulation used, from a maximum of the cross-sectional returns to scale. 102 Results From Estimation Production Functions The results from estimating the Cobb-Douglas function under dif- ferent sets of assumptions for the four regions will be presented first. Thereafter the Translog results will be presented and discussed. Cobb-Douglas Results The first approach to estimation was Klein's method in which the geometric mean cost-share of an input is equated with the coefficient of that input. Table 4.1 lists the results by region. It should be noted that the share of land has been measured as that necessary to ful- fill the constraint that the shares sum to unity. A more direct approach was not possible since the cost of land-ownership or rental was very poorly recorded in the surveys. The results are similar in all regions for fertilizer and machinery, at approximately the 0.10 and 0.35 levels, respectively. However, labor as a proportion of total costs rises from 23 percent in Florida to 35 percent in Louisiana, to 47 percent in Puerto Rico and to 56 percent in Hawaii. The share of land, which appears low in all regions due to the bias from the assumption of profit-maximization necessary to this method, follows a converse trend across regions from a high of 28 percent for Florida to a low of 10 percent for Hawaii. Table 4.2, which lists the OLS results, both with, (called "covariance"), and without dummy variables for each firm, makes an interesting comparison. Firstly, the coefficient of land, as expected, is much larger than in Table 4.1 in all instances. It is particularly large for Florida and Louisiana, which reflects the very large MVP 103 Table 4.1. Results of Klein's Method Region Coefficient Land Fertilizer Machinery Labor N Louisiana 0.1899 0.0970 0.3594 0.3537 135 Florida 0.2800 0.0845 0.4059 0.2296 80 Hawaii 0.0088 0.1047 0.3283 0.5582 69 Puerto Rico 0.0641 0.0986 0.3629 0.4744 99 of land when that input is restricted by proportionate shares. Secondly, as the generally low t-values reflect, the estimates are all relatively inefficient and in some cases bear the wrong sign. On this score there is little to choose between the standard and covariance estimates. Two explanations are possible, namely 1) the high degree of collinearity which existed in all cases between land and all other inputs and 2) the simultaneity problem for inputs other than land. Only the covari- ance estimate for Puerto Rico and the standard estimates for Puerto Rico and Hawaii approach acceptability on the basis of having co- efficients which are statistically significantly different from zero and of the expected magnitude and sign. Note also that the covariance model resulted in an increase in returns to scale for Florida and Puerto Rico and a decrease for the other two regions. This dichotomy will appear again later. In all cases the covariance model led to an increase in the land coefficient, suggesting that omitted variables are related to land, e.g., land quality. Table 4.3 presents the results of ILS estimation in both stand- ard and covariance forms. Because ILS involves a nonlinear transforma- tion, the t-values listed in the table refer to the originally estimated 104 .Am_.¢v :owumzcw cm nonwcommc m? Ponce wucmwcm>oo och .uamocmp:_ :3o mum we; Esme comm mocwm .msswe ”Fm sow come one mw quoe ucmvcmpm one .Am_.ev eaveasaw =2 aaaweumae cam» vm>cmmno vacuum Lo» mpnmLcm> xssau we saw» uw>cmmno umcww ace «Famwsm> Azanu Fe m? “capmcoo ache mpmom o» mcczumg u a "mueoz Aaeo.mv Amem.mv Amee.ev Aeem.Pv Amw_.ov Amee.ev ma Paa.o mN~._ Ammo.o- Npm_.o- mmem.o mom_.o .Neoo.o mmmm.o _Nmm.o- apex aetasa . Aeem.cv Aeme.~v A__m.ov Ammo.2v AeoN.ov Aeee.mv me mma.o mem.o Pm_P.o mmmo.o Fmpp.o eo~_.o- Neec.o- e_me.o meae.e Cease: Aomm.ov Acme.ev Ammm.ov AN_A._V Ammw._v Ammm.av om mam.o emc._ memo.o mamm.o- ameo.o Nemm.o Amm~.o- eoom.o emmo.m eatso_a Aeme.ov Amma._v Ampe.ov Amee.mv mm. “ma.o amm.o --- --- memo.o- Ppm_.o- Aomo.o- amme.o mpmo.m aeamma365 «Paco: mocawcm>ou Amoa.mv Aemm.mv Amee.mv A~m~.av AmAA.mv Aee~.av am ema.o wao._ Nee_.o- NmN_.o- 5224.0 eam~.o Kmac.ou aamm.o Noom.o aa_m aeeasa Amoe.ev Aeem.~v A~_m.ov AmAm.mv am aom.o e_mo._ --- --- mmoe.o mamm.o mwmo.o m.om.o AaaN.N- Amaze: Amee.ov Aoem.mv A_Fe._v Aoe~.Pv Acmm.ev om _ea.o Npm.o e~_o.o mm-.o- me_P.o meam.o eefi_.o- omOA.o omme.m eeweo_a Am_e.ov Aaoe.ov Ammo.mv Amma.mv mm, mAm.o mem.o --- --- Ammo.o mmeo.o NANP.c came.o mNeN._ eeaemesas P666: eaaaeeem z 1Nm a mg Fe Loam; ago:_:omz Lm~wpmucmm ecu; ucmumcou ucwmowmwmou commmx mso to ma_=mae .N.e a_aae '105 .noopona :moson nowuoauwm mmnu no» mnemoon one .x.. . ..P n .nn Lo o:_o> ouapomno on» we nu ononz Awe . F\_V no u no on on unaoo mo: nouns -wumo ouewnaonago onp .o>wummon mo; Nn no Pn no>ocon3 .Amn + Pn + F\Fv wn u no mo: nouosmumo one .momoo oEom cw ouownaonaaocm uo>ona Amfi.ov cowuoacm noun: now_noo no>_m nouosmumo mnH wneunoum on» “on“ .ouon omp< .ucaoo one xonu nopnz zopon munoPoFooooo mnH on» on no: one mno_uo:mo uouoepumo NFFonpmpno on» Co monopopmoooo on» cu nomon mo3Po> u o>ono one .ouzaeoo o» upaomvw_u ono mnonno unounouw onu .mnownoaoo noooewumo >_Fo=uom on“ Co mnowomsnoomnonu noonm_:on one mouoswumo mom on» oonmm ”upoz .msonomouco no~_P_unoL xpno N .ucop mo P—oz mo muonomoxo anonmnoozF neNm.Nv Amom.ev Anao.ov Amom.mv nemm.nv Amoe.ev ma oma.o oa_.n mnao.o ~_mn.o mane.o enmn.o meoo.o meme.o .mmm.o mound apnosa nomm.~v ne_o.ov nono.~v Anno.mv nm_c.ov nomm.mv mo nmm.o ene.o memo.o Neoo.o omm_.o mnm_.o aooo.o onmp.o mnmn.n P_asa= neam._v neao.mn Ammo.mv nomm.mv name.mv nnmo.ov ow nmm.c mmo_._ mmmo.o Noo_.o- nem_.o an_m.o mmn_.c cone.o a.om.o _aennapn noem.ov n_mm.mv Amwm._v neo_.~Nv mm_ oam.o Omm.o --- --- Nmo~.o eemo.o ammo.o NNme.o mmoo.~ aea_m_=as povoz ounowno>ou nomn.mv Anno.mv Amno.mv nemm.~v Amom.on nNo~.ov ma aoa.o m_o.n nan_.o enon.o ooom.o e_N~.o a__n.o Nmom.o Femm.o Naa_m annasa n_o~.ev noam.o_v nonm.ov Amea.mv mo Nea.o n_m.o --- --- mnm_.o mmmm.o ooom.o moep.o ommc.P _n_aze= nomo.ov namn.mv Amo_.ov nom_._v noom.ov Anom.nv om mna.o oma.o eemo.o oe_P.o- onoo.o ammc.o o_o~.o nomm.o Anne." .aannopn naoo.ov Amoe.ov nm~o.~v noo_.onv mm, Nnm.o mea.o --- --- N_o_.o omo~.o Nnao.o- maeo.o Nomm._ aeenmwsao _oaaz anaaeaom 2 ma n ma Po nonmn xnocmnooz no~_—_unon anon unoumcou ucowomweoou nowmom mn~ Lo mu_:mom .m.o opnoe 106 coefficients and not the ILS coefficients; their values are, however, still of statistical interest. In some regions all variables other than land have been assumed endogenous whereas in other regions only some of the other variables were thus treated (see footnotes). The results in the table reflect a subjective choice of the specification with respect to which variables should be exogenous and which endogenous. The first noteworthy result is the reduction in size of the land coefficient, as expected. It now lies somewhere between the value from OLS and that from the Klein method. Secondly, the t-values show that ILS led to much more efficient estimates. The increase in efficiency is due to the elimination of simultaneity-bias and also, to some extent, to a reduction in multicollinearity. Thirdly, the results of the covariance model, except for Hawaii, are now more convincing than those for the standard model. The reason for the remarkably low returns to scale in the covariance model for Hawaii is not clear. As with OLS, the covariance approach reduced estimated returns to scale in Louisiana and Hawaii but increased estimated returns to scale in Puerto Rico and Florida. Summarizing the Cobb-Douglas results, as expected the use of ILS with the inclusion of a dummy variable for each firm (covariance procedure) yielded the most credible results in most cases. Klein's method gave a downward bias to the land coefficient and OLS gave an upward bias to that coefficient. The ILS results (with dummy vari- ables except in the case of Hawaii) were used in synthesizing the aggregate response function. Before returning to that "synthesis," the Translog results will be presented and ease of input substitution determined. 107 Translog_Results Table 4.4 presents the results of ZEF estimation with the "corrected" cost-share as dependent variable and the individual input levels as independent variables. The results in Table 4.4 are not by themselves very inter- esting except to note that the parameters were more efficiently esti- mated for Hawaii and Puerto Rico than for Louisiana and Florida. The explanation may lie in the proportionate shares existing in the latter two regions. The fact that the values were often negative is of no consequence: it is not necessary for monotonicity of the function, for example, which depends on the cost share always being positive. The Yrj parameters from Table 4.4 were inverted as shown earlier in Equation (4.22) to give estimates of the elasticity of substitution between pairs of inputs. The results of this inversion are given in Table 4.5. The simplest interpretation of the elasticities of substitu- tion in Table 4.5 is to note that positive elasticity denotes substi- tutes and negative elasticity complements. The most likely substitutes. a priori, are labor and machinery and this is confirmed by the esti- mated value of OCL for all regions. Similarly, though surprisingly, fertilizer and machinery are estimated to be substitutes in all regions. Considering the other elasticities in turn, fertilizer and labor are estimated to be substitutes in Louisiana and Puerto Rico, but comple- ments in Florida and Hawaii; this result could be associated with the higher wage rates in the latter two regions, as demonstrated schema- tically in Figure 4.10. At price ratio P2, which gives X1 a high price relative to X2, X1 and X2 are on the border of being complements, while 108 Table 4.4. ZEF Estimated Parameters of the Translog Function Parameter Region Louisiana Florida Hawaii Puerto Rico 0F -0.0541 -0.0560 -O.l706 0.1384 00 -O.2635 -0.4923 -0.0596 0.0740 aL -1.0681 -0.2530 0.9598 -0.0279 00] --- --- --- --- yFF 0.0336 0.1454 0.1523 0.0656 (41.481) (11.515) (13.941) (18.395) YFC -0.0020 -0.0388 -0.0387 -0.0201 (1.823) (1.334) (5.470) (9.074) YFL -0.0023 -0.0153 -0.0116 -0.0298 (1.176) (1.717) (4.253) (8.676) YFD -0.0293 -0.0913 -0.1019 -0.0157 (10.458) (1.943) (6.572) (2.665) YCO 0.0861 0.2062 0.2185 0.1459 (32.578) (2.672) (25.322) (48.765) YOL -0.0056 -0.0171 -0.0987 -0.0971 (1.195) (0.795) (27.693) (26.881) YCD -0.0784 -O.1504 -0.0810 -0.0287 (11.744) (1.204) (6.104) (5.378) YLL 0.3152 0.1218 0.1227 0.1920 (15.662) (14.707) (61.308) (29.891) YLD -O.3073 -0.9894 -0.0123 -0.0651 (12.089) (2.529) (2.777) (7.208) YDD 0.41492 0.33112 0.19522 . 0.10942 (11.892) (1.598) (5.878) (5.405) Where the simultaneously estimated equations were: "F = 0F + YFF log F + YFC log C + YFL log L + YFD log 0 + 91 MC = ac + YCF log F + YCC log C + YCC log L + YCD log 0 + e2 ML = 0L + YLF log F + YLC log C + YLL log L + YLD log 0 + e3 .. with restrictions Yrj = Yrj and Zyr- = 0 and where: J the cost-share of the it“ input - a constant for the jth input, a parameter, Q: III fertilizer, machinery, labor, an error term. 1Not calculable unless the restriction 2X1 = l is imposed. 1 2This value is based on the assumption of zero covariance for Yroi YM0 a"d YLD' 109 Table 4.5. Estimated Partial Elasticities of Substitution and Own-Price Elasticities of Demand Parameter Region Louisiana Florida Hawaii Puerto Rico oFF -ll4.9 69.1 -6.9 -25.5 oFC 97.6 2.8 9.1 6.8 oFL 53.6 -58.2 -6.6 1.8 oFD 2.2 -10.8 -5.8 2.0 oCC - 96.9 -34.1 -12.4 -51.7 oCL 16.7 51.1 16.3 28.9 000 1.2 7.3 -0.9 -1.0 oLL -6.7 -10.8 -28.8 -20.7 oLD -2.5 -9.6 3.6 4.1 000 1.2 1.8 5.9 -5.1 nFF -4.0 +ll.0 -l.6 -2.8 nCC -9.6 -9.8 -5.1 -11.2 nLL -2.1 -1.3 -5.8 -7.7 nDD +0.6 +0.8 +1.0 -l.5 where: Ukr = the elasticity of substitution between inputs k and r nrr = the own-price elasticity of demand for input r F = fertilizer C = machinery L = labor 0 = land Okr > 0 implies that k and y are substitutes. < 0 implies that k and Y are complements. O x 1 110 at price ratio P1 they are substitutes. Such an interpretation implies either that the production function is different across regions or that it has nonconstant elasticity of substitution. Similar results were found, as shown in Table 4.5, for fertilizer and land, being comple— ments in Florida and Hawaii but substitutes in Louisiana and Puerto Rico. Machinery and land were substitutes in Louisiana and Florida but comple- ments in Hawaii and Puerto Rico, whereas exactly the converse result was found with labor and land; there seems to be no obvious reason for this result, although it may be connected with the proportionate shares in force on land allocation on the mainland. Figure (4.10) Elasticity of Substitution and 06163” AA “*‘"‘""* X2 [isoquant / r a . /’. x " Price-line 1 P2 Price-line P1 111 The ggg elasticities of substitution are more easily interpreted by converting them to own-price elasticities of demand via the identity33 (4.37) nrk = Okr - Mk where: n = price elasticity, o = elasticity of substitution, M = cost-share, r,k = input subscripts. These price elasticities are also given in Table 4.5. The expectation was that all such elasticities, except that for land, would be negative, implying lower quantity demanded at higher prices and this was confirmed by the results except in the case of fertilizer in Florida. The case of Florida is disturbing, but further inquiry produced two possible explan- ations namely 1) the negative correlation between fertilizer use and the quality of land in Florida and 2), the interaction between nitrogen and phosphorus in Florida's soil which induces a phosphorus deficiency, hence lower yields, as more nitrogenous fertilizer is added. Disregard- ing the result for fertilizer in Florida and the land elasticities for the present, the input elasticities are most similar for machinery across regions, ranging from -5.1 in Hawaii to-11.2 in Puerto Rico. The most interesting elasticities are those for labor, which are much lower for Louisiana and Florida, (-2.1 and -l.3, respectivelYI. than for Hawaii 33See Binswanger. H. (1974). "A Cost-Function Approach to the Measurement of Elasticities of Factor Demand and Substitution," American Journal of Agricultural Economics, gt, (2), (May), pp. 377-386. 112’ and Puerto Rico, (-5.8 and -7.7, respectively). This suggests that the demand for labor is more inelastic on the mainland than in the off- shore regions and possible explanations would be the unionization of labor in Hawaii, which has resulted in high wages but lower employment, and in Puerto Rico the importance of labor in the production system (45 percent of costs) for which alternative employment on the mainland was, at the time of the survey, increasingly attractive. The positive own-price elasticities for land in all regions except Puerto Rico were no surprise; they reflect the fixity of land in its present use. More fully, the price of land in the survey years was not determined by its alternative use at the margin but by its endogenously determined MVP in its current use. In considering how much cane to produce, the price of land was not relevant to the producers but only determined as an ex-post identity. In Puerto Rico, by contrast, the price of land appears to be exogenousrand further evidence for this view is rapid decline in Puerto Rico's industry suggesting that the MVP of land in alternative uses is higher than in sugar production. Some final comments on the Translog results are as follows. In general, the estimated price and substitution elasticities are particu- larly large and the standard errors, while not computable, may also be relatively large for Louisiana and Florida since they were large for the production-function parameters in these two regions. However, elasticities measured at the geometric mean of a cross-section of firms nay be expected to be large relative to such elasticities from an aggregate_time-series. A cross-section of firms demonstrates potential elasticities at the micro-level whereas an aggregate time-series demonstrates ex-post 113 elasticities at the macro level. Firms cannot rapidly adjust from one input mix to another, because there are indivisibility problems with different technologies, i.e. a certain scale may be necessary for a certain technology. Family farms, such as many in Louisiana, may be unwilling to bear the risks involved in adjustment to a larger scale. Just as there is an upward bias in aggregated supply response based on individual firms,34 so elasticities measured at the aggregate level may be expected to be smaller than those measured in a cross-section of firms. No simple conclusion emerges from Table 4.5 concerning a division of technology into mechanical, which acts exclusively as a labor sub- stitute, and biological, which acts exclusively as a land substitute, 35 While labor and and this result confirms that of Binswanger. machinery are substitutes, so also are fertilizer and machinery. Similarly fertilizer was in some regions a complement to land and in others a substitute. Binswanger's finding, that for U.S. agriculture fertilizer and labor are complements, is neither confirmed nor denied by this study. When the price of labor is high relative to fertilizer, as in Florida and Hawaii, the two may be complements but at lower wage rates they may be substitutes.36 The two inputs for which the micro elasticities may most closely reflect their macro counterparts are those for fertilizer and labor, 34For a review of sources of aggregation bias, see Egbert, A. C. and Kim, H. M. (1975). "Analysis of Aggregation Errors in L.P. Models, American Journal of Agricultural Economics, g}, (2), (May), pp. 292-301. 35 Ibid. 36This leaves entirely open the question of whether fertilizer and labor are complements or substitutes in less developed countries, c.f. Binswanger. 14“ ff; u a a a. u ‘A '59 u 101' no. I‘d ‘9- II. 114 since indivisibilities pose less of a problem than with machinery and adjustment, except in Louisiana, is likely to be relatively complete at the micro-level. The U.S. Sugar Act prescribed minimum wages for Louisi- ana and Florida, while in Puerto Rico the government and in Hawaii the unions maintain fixed relationships between the price of sugar and the wage rate. Since in all regions machinery and labor are clearly substi- tutes, the result of an increased wage rate is greater mechanization and less employment. This is particularly true for Hawaii and Puerto Rico, where the estimated labor price elasticities are -5.8 and -7.7, respectively. For total wages to rise when the wage rate rises, a price elasticity of more than -1 is required (i.e., nearer zero) and this condition is not fulfilled in any region. Policy in Puerto Rico has been directed to wage supplements to encourage employment and, given the findings of this study, such a policy should be effective. Auxiliary Time-Series Regressions As earlier explained, these were designed to relate a computed aggregate residual to time (technology), output (scale) and a constant of aggregation. The residual was computed from the Cobb-Douglas esti- mates, assuming profit-maximizing behavior for all inputs except land. Twenty-two years of aggregate data were used, from 1950 to 1973, ex- cluding the years 1962 and 1964 in which large adjustments took place and which will later be used as "normalizing" years for projecting investment behavior. Table 4.6 lists the results. The results confirmed a priori expectations for Florida and Puerto Rico, but not for Louisiana and Hawaii. While the residual was posi- tive1y related to time, except in the case of Florida, when regressed Table 4.6. Results of Auxiliary Time-Series Analysis Region Coefficient Constant Time Output 82 DW N Louisiana -5.22599 0.04075 --- .775 .417 22 (8.557) -5.00871 0.02397 0.00153 . .920 .793 22 (6.045) (6.079) Florida 1.96932 -0.00774 --- .023 .865 22 (1.219) 0.21445 0.02696 -0.00101 .246 .622 22 (1.881) (2.628) Hawaii 1.16341 0.05264 --- .904 .587 22 (14.077) 0.76222 0.04576 0.00075 .932 .604 22 (11.784) (3.030) Puerto Rico -7.58525 0.09736 --- .927 .730 22 (16.300) -4.53536 0.05937 -0 00084 .942 .322 22 (3.707) (2.514) on that variable alone, it was only negatively related to output in the cases of Florida and Puerto Rico. A negative output coefficient implies diminishing returns to scale, while a positive coefficient denotes the opposite. With increasing returns to scale there can be no equilibrium in a sector with more than one firm and this is clearly a spurious result in the present context. The cause of this unexpected result was probably the very low variation in output in Louisiana and Hawaii in the last two decades which may be contrasted with the considerable expansion in Florida and the large contraction in Puerto Rico. Given the low variation in output in Louisiana and Hawaii and the collinearity of output with time in these regions, the failure to distinguish 116 diminishing returns to scale is not surprising.35 Because diminishing returns were necessary to the method of projection to be used, such returns were subjectively imposed for Louisiana and Hawaii. Taking the coefficients for time from Table 4.6 the following subjective auxiliary equations were constructed: Louisiana: e* t -4.50 + 0.04075T - 0.0015 X0t Hawaii: 8* t 4.83 + 0.05264T - 0.3330 X0t where: e; the computed residual, year (e.g. 1966 = 66), X 0 II output in 1,000 tons, a time subscript. For Florida and Puerto Rico the results from Table 4.6 were used directly in the projection of supply. Projections of Supply The projections which follow are designed to examine the probable growth or decline of the U.S. cane sugar industry after 1974 under an array of different prices. What would have occurred without the limita- tions imposed on average under the U.S. Sugar Act is also examined. Several further assumptions are necessary before supply may be projected for each region. Those assumptions which are specific to a particular region will be presented later, while assumptions of a general nature are first reviewed. 37From the ow statistics in Table 4.6 one deduces that most of the regressions were autocorrelated. Orcutt transformations on the equations did not greatly change the coefficients, however, and autocorrelation is no problem for "predictive" equations. 117 The first assumptions concern lags and prices, beginning with the lag between the decision to expand the area in cane and the harvesting of the first of that new cane. This lag is likely to be shorter when there is a Sugar Act than when there is none, since the Act reduces inherent risk in investment by guaranteeing at least a minimum price for cane. The biological delay is a minimum of 13 months on the mainland and ranges to 20 months in Hawaii. Because the timing of a price sig- nal in relation to the crop year way be important and there is also a delay in expanding mill-capacity, if necessary, the same lag systems have been used in all regions. Figure 4.11 presents the price weights diagrammatically. Figure (4.llIPrice Weights Price Weight HI The lag is assumed to be of the "inverted-V“ kind with the peak occurring in year t-2. With a Sugar Act the V is assumed to be steeper than without such an Act. Formally the lags may be written: 118 W1th Act: P; = 0.40 Pt-l + 0.60 Pt-2’ ' . * = W1th0ut Act. Pt 0.30 Pt-l + 0.45 Pt“2 + 0.15 Pt_3 + 0.10 Pt-4’ where: P; = expected price at time t, Ptl’Pt-l = actual sugar prices in New York. The above weighting system is only used to determine acreage in any year. The yield_per acre in a particular year is assumed to be a function of the current and last year's prices which are equally weighted. That is: USPQt 0.50 Pt + 0.50 Pt_] where: USPQt the yield-determining expected price. When the Sugar Act was in operation and proportionate share restric- tions were removed in the mainland areas, there was a reluctance to expand acreage because of the fear of the reimposition of shares. This "overhang" effect of the shares has been included in the current model by constraining any expansion of land area to the previous maximum if shares were in force in two out of the last four years. One final price assumption has been made with respect to 1974. In that year the domestic sugar price on the New York market averaged 29.50 cents per 1b., but that average disguises a rise from 12.63 cents Per 1b. in January to 57.30 cents per.lb. in November. Since the peak occurred so late in the year, it could not affect current inputs in the industry in the year 1974. In the model, as explained,yield depends on the average price of the current and past years. For 1974 such an average would be 19.90 cents. Because price rose so rapidly in 1974, 119 producers' expectations were alwyas much lower than the actual price and an expected price of 14.0 cents per lb. has been used in the calculation of yield for this year. The second set of general assumptions concerns contraction, expan- sion and the price of land. 1964 saw a general expansion in cane area and the price of land has been normalized so that its price was equal to its MVP in that year. In Puerto Rico, price was assumed to exceed MVP in that year by an amount which led to the observed rate of contrac- tion of land area in that year, as will be later explained. The prices of land per acre thus found were $140.00 in Puerto Rico, $112.25 in Florida, $62.72 in Louisiana and $49.20 in Hawaii. For years other than 1964, the price of land was inflated parallel to the U.S.D.A. index of input prices except 1) in Hawaii where the abosolute limitation on land areas led to the use of the equation (0.004)(LN0)2 (49.20) PLND = the larger of 40.00 where: PLND = the price of land LND = the thousand acres harvested, and 2) in Puerto Rico where the price of land was inflated parallel to the wage rate in the sugar industry. Given the price of land in any year and no restrictions on acreage, the desired area in cane was that at which the price of land was equal 38 t0 the MVP of land in cane production. Should the desired area exceed ‘— 3 . , - 5 . 8The equation was (4.36). LND* - {7:WEY;; - X*o Po.} /PLND I, .1. i=1! , '5‘. NM 'IV‘ A? v.' 120 the existing area, expansion occurred and, should desired be less than actual, contraction occurred if the ex post MVP of land using this year's 39 price of sugar was less than the price of land. The rate of contrac- tion was constrained, because of the perennial nature of cane, to be LNDt = (3 LND + LND*t)/4 t-1 in all areas except Puerto Rico, and LNDt = (8 LNDt_1 + LND*t)/9 in Puerto Rico, where LND is land area, LNO* is desired land area, and t is a time subscript. The final assumptions concern the starting date and prices for projections. In tracking past production and in projecting what supply would have been without a Sugar Act at the existing prices, the starting year of 1955 was chosen. Projections f0r the future begin with the 1975 crop, although the "actual" output of the 1974 crop is itself a USDA estimate. Projections have been made to 1985 under the assumption that only the price of sugar changes and not the input prices; i.e. projec- tions are made at real 1974 prices. An alternative would have been to project inflation in prices for each input from past experience, but that would be equally subjective. Projections have been made for the 4 cents per lb. to 18 centsper 1b. (domestic, New York) price-range at 1974 prices. The reader is severely cautioned to deflate the sugar price E where: * = a maximizing value. LND = output. po = the price of sugar. PLND = the price of land. 390.9 times the price of land in Hawaii. .-‘-o -\' 71H 14:- "~ : c" - 4p . ._. 2.. .‘2 - d‘ I”) __4 Fri 11 121 of his conception at some future date to the 1974 level in interpreting these results. Louisiana The compound equation utilized may be written: x0t + 0.0015 X0 = -0.1782 + 0.7505 Indt - 0.1401 pmct -0.4700 plbt - 0.0496 pft +0.6597 log (0.5 P0t + 0.5 Pot-l) +0.04075T. where: x0 = log of sugar (thousand short tons raw value), X0 = thousand short sugar tons raw value, 1nd = log of thousand acres, pmc = log of USDA machinery price index, plb = log of hourly sugar wage in dollars, pf = log of USDA fertilizer price index, P0t = New York domestic sugar price in cents per lb., T = year (e.g. 1966 = 66) t = a time subscript. Figures 4.12 and 4.13 present the results for output and acreage harvested respectively. Both in terms of output and acreage the model seems to be a reasonable fit. The large fluctuations in actual output reflect weather variation rather than changes in yield or acreage. When yield equations were fitted directly to the cross-section data, dummy variables for individual years accounted for 29 percent of the total 122 .000 SHORT TONS RAW VALUE . FIGURE 4.12 PROJECTIONS OF OUTPUT FOR LOUISIANA 1200 . 18 CENTS 1000 ,v- “““ 7 I4 CENTS WITHOUT SUGAR ACT {a -_”_,,-' I -""""' 800 I 5 10 CENTS l’\\\ ..- . ’I “ “""- -‘-—----" 000 1”, ‘\_ ' -v ~----- 5 ‘ ~- . a CENTS ---—’-’“ V J --- -~-I ~---- --- 400 ‘ " ’ ACTUAL ESTIMATED 200 o 55 50 6| 64 67 70 73 76 To a: 85 YEAR THOUSANDS OF ACRES 700 FIGURE 4.I3 PROJECTIONS OF ACREAGE FOR LOUISIANA I‘~. 1’ \N 18 CENTS 000 : s. __ ‘ ---- I I I' 500 H...“ I" ‘~- g 14 CENTS WITHOUT { ~ ....... I ‘s‘ 400 SUGAR ACT I, ‘k‘ \ I 1’ \\ \\‘ -- y \ ‘~. 10 CENTS 300 ” ----§~ ~’ \\ ~~~~-- ---------‘--- I II ----- \ \ \ ‘\ 200 ACTUAL \. o CENTS 100 O 55 58 SI 04 ST 70 73 76 79 02 85 YEAR 123 variation for 1969-71.40 Similarly, attempts at direct time-series estimation of yield had weather and time as the only significant influences. Perhaps of greatest interest from a historical viewpoint are the estimated output and acreage in the absence of a Sugar Act. While in the late 1950's there would have been approximately a 10-20 percent increase in output and acreage, from 1961 onward there would have been little difference in output in the absence of a Sugar Act as compared with actual output. This suggests that proportionate shares in Louisiana were I not highly limiting but reflected producers' intentions relatively closely. The projections show an expansion to 760,000 tons and at least 303.000 acres at all prices above 5 cents per lb. in 1975, following the high {vices of 1974. Thereafter, at prices above 10 cents, the higher IEVEISLOf output would be maintained, while at prices less than 10 cents there would be a subsequent decline in acreage which would only be parti- ally compensated by increases in yield over time. Note however that at 14- cents or less the peak output would occur in 1975 while peak acreage would not be found until the following year, 1976. This again reflects the extraordinarily high price in 1974. By 1985 projected outputs are 472,000 tons at 6 cents, 729,000 tons at 10 cents, 925,000 tons at 14 cents and 1,086,000 tons at 18 cents; c.f. estimated 1974 output of 683,000 tons. SimilarTy acreages in 1985 would be 151,000 at 6 cents,287,000 at 10 x ‘40(Xo/LND) = 30.0133 + 0.2009 (F/LND) + 0.0401 (MC/LND) + 0.0195 (3.306) . (1.619) (0.926) (LB/LND) + 5.4061 DUM 69 + 5.3645 DUM 70 - 0.0023 LND - (3.642) (3.579) (3.929) 2 R = 0.359 N =135 there X is tons of cane, MC is machinery expenditure, LND is land acres, 3 1s 181m expenditure, DUM 70 is a dummy for 1970, F is fertilizer exPENditure, DUM 69 is a dummy for 1969. .. "I I‘F' t ":3“ "51 can} ':9 1.2 ‘ 9' ’7' 124 cents, 434,000 at 14 cents and 591,000 at 18 cents in 1985; c.f. esti- mated 1974 acreage of 325,000. In general Louisianan output is projec- ted to be relatively inelastic, as will later be further demonstrated by comparisons with the projections for the other regions. Florida The compound equation for Florida was x0t + 0.001006 X0t = 1.5621 + 1.2794 1ndt - 0.8545 pmct - 0.3620 plbt - 0.4712 pft + 1.6872 109 (0.5 P0t + 0.5 Pot_]) + 0.02696 T, where notation is exactly as for Louisiana. Figures 4.14 and 4.15 present the results on output and acreage harvested. The actual and projected quantities follow a very close pattern which, to some extent, reflects the minor importance of weather in determining Floridan output.“ Had there been no U.S. Sugar Act from '1955 onward, acreage and output would have reached their 1964 levels by 1956. During the 1965-71 period, when proportionate shares were in force,, Floridan output, unlike Louisianan, was limited by these acreage reStrictions. It appears that during 1964, when no restrictions were 1" forx2e, investment in further land preparation occurred so that, when rEStriiztions were removed for 1972, this extra land was brought into “59 for the first time. 'The constant-price projections demonstrate the further potential '0? exDansion which is believed to exist in Florida. .As in Louisiana, \ 41Weather dummies only accounted for 4.8 percent and 1.5 percent of the variation in output in 1968 and 1969 as compared with 1967. 125 THOUSANDS OF ACRES FIGURE 4.14 PROJECTIONS 0F ACREAGE FOR FLORIDA 18 CENTS 000 500 ‘ I I I I'll/.--T‘ ‘00 if \\\ \“‘ '4 CENTS I “ """""" I! 4 \\ 300 WITHOUT ,1 ,I’ ‘. \ SUGAR ACT ’5’ \\ \\ ----------u- --- \ \ 200 ----------- \\ \\\‘ '0 CENTS ---- \‘\ \‘ ------- \\ 100 ACTUAL PROJECTED \x“ ‘~\ S CENTS 0 64 67 70. 73 76 79 82 85 YEAR 55 58 61 J .000 SHORT TONS RAW VALUE I _____________ ”' FIGURE 4J5 PROJECTIONS OF OUTPUT FOR FLORIDA 18 CENTS I800 I4 CENTS \——I—~~‘ --‘ofl §‘ ‘-— --’—' 1500 1200 900 ID CENTS GOO .. ....... 300 B CENTS 82 85 73 61 64 YEAR . 5‘ F" .v 0‘ 01'... L .\ _ oil.- I A“,- H“! *w in ‘ 126 ggtgggg_is projected to peak at prices less than 15 cents in 1976 whereas ggtggt_is projected to peak in 1975. At 14 cents per lb. the expansion would be approximately maintained at about 1,400,000 tons whereas at 10 Cents there would be a subsequent contraction to 600,000 tons by 1985 and at 6 cents to a mere 45,000 tons by 1985. At 18 cents per lb. the projected expansion is very great, reaching 1,604,000 tons IN 1976 and 1,935,000 tons by 1985; c.f. estimated 927,000 tons in l974. Acreage projections are similar to those for output, reaching in 1985. 533,000 at 18 cents, 355,000 at 14 cents, 167,000 at 10 cents and 27,000 at 6 cents; c,f. estimated l974 acreage of 239,000. The supply in Florida may indeed be said to be elastic, as will be shown in the later inter-regional comparisons. Hawaii The compound equation for Hawaii is x0t + 0.003 x0t = 9.6522 + 0.6439 1ndt - 1.6826 pmct - 0.8349 p15t - 0.8838 pft + 3.4014 109 (0.5 P0t + 0.5 P0t_1) + 0.05264 T, where notation is exactly as for Louisiana. Figures 4.16 and 4.17 IWesent the results. The actual and estimated outputs show rather cfifTerent patterns, while acreage has been remarkably stable and hence rot difficult to model. In the cross-section analysis for 1967-69 heather variables were not a significant influence on output and so do "0t explain the underestimated output from 1964 to 1972. However, the °UtPUtS of’l958, 1959 and 1960 were affected by a strike and it is not suI‘lll‘ising that our model does not take this into account. 127 .000 SHORT TONS RAW VALUE 800 FIGURE 4.l6 PROJECTIONS OF OUTPUT FOR HAWAII I ”-.. I8 CENTS ...o 14 CENTS I400 I200 - ‘ ‘ I _ o “--"--“ -- _,_.— 10 CENTS - - --.‘- -., 1000 \ACTUAL ESTIMATED 800 ----.. ___ ___..- .. 6 CENTS 600 .55 58 61 64 67 70 __73 76 79 82 85 YEAR THOUSANDS OF ACRES FIGURE 4.17 PROJECTIONS OF ACREAGE FOR HAWAII 160 I40 ----------------- '8 CENTS I2 "---------------- " CE'JTS o ’k~--‘ I’ ‘:-~ ‘.~ -‘-. ‘ - ---------- -‘O‘ \ ~~‘~ . -------- ‘2 CENTS ‘ ......- IOO ‘, ‘- IO CENTS ACTUAL ESTIMATED \‘ . \\ 80 ‘ . “ S“ 30 s‘ “~ 6 CENTS 55 58 81 64 67 70 73 76 79 82 85 YEAR l‘bc ed a. LA; I A): V‘O '7' A): 128 Output is projected to peak at all prices in 1975, thereafter declining in greater or lesser degree. The 1975 peak ranges from 1,500,000 tons at 6 cents per 1b. to 1,800,000 tons at 18 cents per lb. This may be compared with our estimated 1974 output of 1,275,000 tons and the USDA estimate of 1,040,000 tons. After falling at all prices in 1976, (all the way to 600,000 tons at 6 cents per 1b.), output then rises slowly due to yield-increasing technology at prices of 10 cents or more, whereas at 6 cents per 1b. the output is relatively stable. Acreage shows a much less responsive pattern than output, which demonstrates that the expected response is in yield and not in acreage since the latter issubjectto severe limitation. At all prices acreage peaks in 1976 but drifts downward thereafter at the lower prices. By 1985 acreage is projected to be 56,000 at 6 cents, 101,000 at 10 cents, 129,000 at 14 cents and 132,000 at 18 cents; c.f. our 1974 estimated acreage of 109,000. Output in 1985 is projected to be 608,000 tons at 6 cents, 1,110,000 tons at 10 cents, 1,455,000 tons at 14 cents and 1,695,000 tons at 18 cents; c.f. our estimated l974 output of 1,275,000 tons. Our projections Show an inelastic expan- sionary response with respect to acreage, but a relatively elastic response in output due to changes in yield. These projections result from the low importance of land in the estimated production-function. Puerto Rico The compound equation for projection is xot + 0.00084 X0t = - 7.9779 + 1.7588 1ndt - 0.7642 pmct - 1.8575 plbt - 0.2497 pft + 2.8715 log (0.5 Pot + 0.5 pot-l) + 0.0594 T, 52". ...-J 129 where notation is exactly as used for Louisiana. Figures 4.18 and 4.19 present the results. Actual and estimated outputs follow similar patterns, except that the rate of decline is overestimated up to 1962 and somewhat underestimated in later years. Our model predicts a slight expansion from 1962-64 whereas in actuality such an expansion did not occur. However, the turning point in output in 1972 was shown by the model. Acreage in the model Shows a continuous decline, (as there has been in actuality since 1958), but is subject to the same problems of under and over-estimattion as output.41 Weather variation nay explain a small part of the discrepancy between actual and esti- mated output, since dummy weather variables were significant at the 2.5 percent level in the cross-section analysis, contributing five per- cent and four percent to the explanation of output in 1970 and 1971, respectively. Since at 6 cents per lb. there were difficulties in solving the model, output was assumed to be zero at that price and projections were made only at prices of 8 cents per 1b.or more. It should also be noted that, becuase of its close connection with the price of sugar, the price of labor was linearly related to the price of sugar in making the pro- jections.42 The high price of 1974 is projected to halt the decline in acreage in 1975 and very slightly reverse it in 1976. Thereafter, at prices less than 16 cents per lb., the acreage would continue its pre- vious decline, while at higher prices the acreage would expand to reach _— 41This could be due to the sigmoid relationship between adoption of an innovation (i. e. giving up sugar production) and time. 42Namely PLBE = (USPQ / 7) where PLB is the wage rate in $/hr. and USPQ is the avg rage of rices at t and t-l. THOUSANDS OF ACRES 130 FIGURE 4.18 PROJECTIONS OF ACREAGE FOR PUERTO RICO 350 \ \ \ \ \ ‘\ 300 4‘ ‘\ \\ \\ I8 CENTS 250 \-—-~_,’\ I \\ ”--’ \‘ "‘~ ------ ESTIMATED/ ‘ ,4 20° ‘4 I 16 CENTS \ I \‘ ’ I / x‘ l ,’ ACTUAL \ I, «a 7’ 150 ‘_1_””’¢\‘ :‘ \:___,v T“:‘s ‘~:\ 4 N ‘00 “:3 CE Ts ‘ 10 CENTS 0 55 so 61 64 67 7O 73 ' 76 79 82 05 YEAR .000 SHORT TONS RAW VALUE FIGURE 4.19 PROJECTIONS OF OUTPUT FOR PUERTO RICO 1200 18 CENTS . I!” I 1000 A\\ ”x 5... If \ ’,” ‘Y‘ ,' \ ESTIMATED ." 16 CENTS 600 s I \ I , ‘s I ‘s ’ I ‘x I I ‘ I I l I 7’ “W ACTUAL"""—' ‘. ,..J , -” I I ”‘~-v’ I —~.” ~‘\‘v-~~- ‘00 \ I, \\\ --.--- \ --- \___,’ ---._~_ -. 14 CENTS 200 “ Io CENTS 0 55 58 61 64 67 7O 73 76 79 82 85 YEAR 131 a new plateau in 1977. Output would follow a pattern similar to acre- age except that, with yields increasing over time, output is projected to climb slowly at prices of 16 cents or more. The estimated acreage for 1985 is 82,000 at 10 cents, 95,000 at 14 cents, 178,000 at 16 cents and 248,000 at 18 cents; c.f. our estimated 1974 acreage of 146,000 and an actual 1951 peak of 392,000 acres. Estimated outputs in 1985 are 210,000 tons at 10 cents, 342,000 tons at 14 cents, 808,000 tons at 16 cents and 1,370,000 tons at 18 cents; c.f. our estimated 1974 output of 465,000 tons and a peak of 1,372,000 tons in 1951. In summary, our projections demonstrate a recovery in the Puerto Rican industry only at prices of 16 cents or more. Regional Comparisons and Aggregate Supply Figure 4.20 compares the projected supplies in 1985 in the dif- ferent regions over the price range of 4 to 18 cents per 1b., (prices in 1974 dollars and input prices held constant except for land in Hawaii and labor in Puerto Rico). A11 schedules are Upward sloping at all prices, but those for Florida and Puerto Rico are doubly curved due to returns to scale in these regions approaching unity at low outputs. Considered at 10 cents per 1b. the supply elasticities are (for 1985) 0.00 for Puerto Rico, 0.75 for Louisiana, 0.99 for Hawaii and 4.23 for Florida.43 One should be cautious in interpreting these elasticities, but they do reflect the underlying inelasticity of Puerto Ifican supply, the high elasticity of Floridan supply, and the inter- nediate elasticity of supply in Louisiana and Hawaii. ___. 43Reckoned as % changg_in OUIPUt using the change from 9 cents % change in pr1ce to 11 cents on a 10 cent base. INC! III III DUTY 132 FIGURE 4.20 PROJECTED SUPPLY-SCHEDULES FOR 1965 (IN I974 DOLLARS) PRICE OF RAW SUGAR IN NEW YORK DUTY-PAID CENTS/L8. LOUISIANA PUERTO RICO HAWAII FLORIDA ’ 6 14 12 10 o _l 200 400 600 800 1000 1200 1400 1600 1800 2000 .000 SHORT TONS RAW VALUE 19 FIGURE 4.21 AGGREGATE SUPPLY- SCHEDULE PRICE 0‘ SUGAR FOR U.S. CANE-:SUGAR 119851 IN NEW YORK ,, DUTY-PAID CENTSILB. 14 12 IO 6 0 o 1000 2000 _3000 4000 5000 THOUSANDS OF TONS OF SUGAR RAW VALUE 133 Table 4.7 summarizes the supply-schedules and gives a grand total which is also presented in Figure 4.21. The aggregate supply curve is virtually a straight line with an estimated elasticity at 10 cents per lb. of 1.59 in 1985. Table 4.7. Projected Supply of U.S. Cane-Sugar in 1985 in Thousands of Short Tons, Raw Value Price in Cents per lb. New York Florida Hawaii Louisiana Puerto Rico Total 3 14 121 221 0 356 4 22 280 308 0 610 5 31 449 391 0 871 6 45 608 ° 472 0 1,125 7 63 753 544 0 1,360 8 92 886 611 163 1,752 9 311 1,006 672 203 2,192 10 615 1,112 729 209 2,665 11 831 1,203 781 201 3,016 12 1,013 1,294 835 . 254 3,396 13 1,202 1,380 880 289 3,747 14 1,352 1,455 925 342 4,074 15 1,514 1,525 967 512 4,518 16 1,671 1,586 1,011 808 5,076 17 1,795 1,645 1,049 1,016 5,505 18 1,935 1,695 1,086 1,183 5,899 Summany_and Implications of Chapter IV In this chapter, supply functions for the cane producing regions 0f the U.S.A. were synthesized from both cross-sectional and time-series data. The Cobb-Douglas production function was the basic unit of abstrac- tion. Projections were made to 1985 which showed that in all regions 134 output may be expected to rise considerably in 1975. However, the expansion is likely to be greatest in Florida and least in Puerto Rico. Supply elasticities were estimated to be 0.00 for Puerto Rico, 0.75 fOr Louisiana, 0.99 for Hawaii and 4.23 for Florida. A price of 16 cents per pound was found to be necessary to stem the long-run decline in Puerto Rico's industry, while at such a price total output of the domestic U.S. would approximately double. An examination of the ease of substitution of different inputs, using the Translog production function, showed that labor and machinery were definitely substitutes in all regions, particularly in Hawaii and Puerto Rico. Minimum wage laws, if effective, and union agreementscfi’a similar nature therefore tend to reduce employment, especially in Hawaii and Puerto Rico. CHAPTER V THE EUROPEAN BEET SUGAR SUPPLY Introduction The production of sugar from beet originated in France during the Napoleonic blockade early in the 19th century. Since then, with periodic expansions and contractions, beet has been the major source of sugar for continental Europe. In the United Kingdom, by contrast, imported cane sugar has always been more important than domestically produced beet sugar. While the continental countries protected their domestic producers, the British protected their colonial suppliers (such as the West Indies). The only important producer of beet sugar outside Europe is the U.S.A., although minor quantities of beet are produced in Canada, Chile, China, Iran and Japan. The producing countries of EurOpe may be divided into the EEC (9.24 million metric tons in 1974), the U.S.S.R. (8.53 million metric tons in 1974), Eastern Europe (4.42 million metric tons in 1974) and the remainder of (Western) Europe (2.54 million metric tons in 1974]). As a whole, Europe produced 26.96 million metric tons of sugar in 1973 and consumed 31.57 million tons, i.e., was 85% self- sufficient in sugar. 1Figures from International Sugar Organization are approximate since some 1974 quantities not yet reported. 135 136 This research concentrates particularly on the supply of sugar within the EEC. Just as in the U.S.A., controversy exists on the desirable size of the domestic sugar industry. As earlier explained in Chapter I, the EEC and the U.S.A. are the two major world groups whose policies are liable to change through the democratic process. In this thesis, therefore, particular attention has been paid to modeling supply in these countries so that projections may be made under a variety of policies representing different degrees of protection. The general model of the supply of beet may be written, PALT T ) (5.1) QBt = fQB (PBt_], PLBt, PFERTt, t_], t where 08 = quantity of sugar from beet, P8 = farm price of beet, PLB = rural wage rate, PFERT = the price of fertilizer, PALT = the price of a competing crop, T = technology and t = a time subscript Assuming fQB to be log-linear for all variables except technology and that there is a desired level of output, QB *, to which adjustment is partial, one may write, (5.2) log QBt* = so + a] log PBt_1 +‘32 log PLBt + 33 log PFERTt + 84 log PALTt_1 + 85 Tt + Etl . Specifying the relationship between desired and actual output as, '.‘ b I 137 (5.3) log QBt - log QBt-l = y(log QB; - log QBt-l) + Et2 , where 0_<_y<1. (5.2) may be rewritten as, (5.4) log QBt = 80y + 81y log PBt_1 + 82y log PLBt + 83y log PFERTt + 84y log PALTt_1 + 85y Tt + (l-y) log QBt-l + YEtl + Et2 . If Etl is not autocorrelated,2 (5.4) may be estimated by OLS to give unbiased and efficient estimates of the coefficients. During estimation some minor changes were made to (5.4) as appropriate. Whenever (1:y) was not different from zero at the 5% level of significance, it was dropped from further consideration. Time was used as a proxy for technology, Tt‘ Domestic prices were used for the Common Market countries and international (dollar) prices of sugar were used for the remaining countries, including the U.S.S.R. The presentation of background information, policies and the results of estimation are in the order: EEC-six, EEC-three, remainder of Western Europe and Communist Europe. Policy and Production in the EEC-Six Prior to the introduction of the Common Sugar Policy in 1968, a variety of national policies existed; In general, the countries aimed at internal self-sufficiency which, in the French case, meant 2See Chapter III for a discussion of estimation when Etl is autocorrelated. A. I“ a. a“ .., a. a) *0. CM 138 self-sufficiency for the Franc Zone. Because the statistical analysis begins in 1953, thumbnail sketches of national policies for the 1953-68 period are presented below. France In most years there was a national quota on production and a guaranteed price for this tonnage. Production which exceeded the quota was exported and its price not guaranteed, although the first 300,000 tons received a 30% subsidy on the difference between the domestic and international prices. The quota was usually in the range of 1.50-1.57 million metric tons which may be compared with a basic quota under the EEC policy in 1968 of 2.4 million metric tons. Since both the price and quota were higher under the EEC policy. it is not surprising that the production also rose from 1.5 million metric tons in 1962-63 to 2.7 million metric tons in 1972-73. West Germany An annual plan for the supply of sugar was initiated in 1951. Under the plan, quotas for refineries were fixed for the coming year and the refiners were expected to make contracts with the growers in accordance with their quota requirements. The price of sugar beet was fixed by governmental edict, but other prices were allowed to vary. After 1968 the price of sugar declined to the grower, but production increased due to an increase in the quota. Belgium (and Luxemburg) Since 1951 the industry has been controlled by the growers' association which has fixed quotas and negotiated contracts with the 139 refineries. Sugar in excess of the quota has received only a very low price. Since 1968, price, quota and production have all risen substantially. Netherlands Prior to the formation of the EEC, the industry was the least controlled of the six. Normally there was no quota on production except in years of exceptionally high production, (e.g., 1959-60). Production was indirectly regulated by a controlled retail price for sugar and a minimum price for sugar beet for the farmer. Italy A bewildering mixture of policies was used in the 1953-68 period, some of which were restrictive and some expansionary. Prior to 1956 Italy produced more sugar than could be consumed domestically and for the 1956-59 period there was voluntary restriction of production. However, in 1959 there was a large surplus and the government introduced quotas for the 1960 and 1961 seasons. In 1962 the High Court declared the quotas illegal, but since then there has been a shortfall in production and no necessity for restrictions. In fact, subsidies were instituted and the subsidies were continued under the EEC policy of 1968, but production still falls short of domestic consumption. Once farmers had lost confidence in sugar they diversified into fruit production and hence could no longer be persuaded to switch back to sugar production in the short run.3 3F. Pignalosa, FAO, Rome, personal communication, August 1974. I4 140 The Common Sugar Policy of 1968 Just as for other products, the protection of the EEC sugar industry is achieved by the use of a variable levy (on refined sugar). Each year the Community agrees a target price for white sugar and sets an intervention price of from 5 to 7% less than the target price. The price for refined sugar is maintained at least at the intervention price through purchase by the Community. Imports are subject to a variable levy_which is such that their price, when delivered, is equal to the target price. Since the policy began in 1968, there have effectively been no imports,except those which came from the French Dependencies of Reunion, Martinique, etc. There have also been some imports from Eastern Germany since these cannot be excluded from EEC markets. Unlike the policy for other products of an agricultural nature, sugar production is restricted by an elaborate system of quotas which ' limits guaranteed prices for sugar beet to certain quantities in each country. Each country has a basic or "A" quota equal to approximately 95% of that country's domestic consumption and a second or "B" quota equal to an additional 40% of domestic consumption (45% in 1974). Within each country these quotas are translated into quotas for the processors of sugar beet who, in turn, make contracts with farmers for certain quantities. The processors, and via them the farmers, receive different prices for production in the A and B quotas. Production in the A quota receives a fully guaranteed price. Production in the B quota which exceeds domestic requirements (i.e., almost all of it), is exported and a levy is charged to the processors (and via them the Uh 1._L_ 141 producers of beet) to finance the export, assuming international sugar prices to be less than EEC prices. The levy has an upper limit so that production within the B quota of more than a certain quantity receives, in effect, a minimum guaranteed price. During 1974 production within both the A and B quotas would have been exported, since the international price fOr sugar exceeded the domestic price, and to stabilize the domestic price a tax on exports was instituted. In more normal years, should pro- duction exceed the B quota, it has no price guarantee and all such produc- tion, called "C" quota, has to be exported. 4 The OECD has represented the sugar policy by the following diagram. Figure 5.1. Quotas in EEC Sugar Policy :EN: 111 ‘ " T ' EA l-IJMH-II p — 1r — D fl—J Qm C 8 Total conSumptlon 0 I Baaa quota (”Contingent A") {guaranteed price) 9 0 . Guaranteed quantity (105 “I. 0! consumptr‘on) C Q Om - Maximum quota ”35% 0! base quota) EA = Surplus Subsrdrsed by a producer levy D + EA 8 Guaranteed price -producer levy EM - Surplus wrthOu: any guarantee (World price) L0vv 8 EA X (loss per expmted Ion) l D + EA 1 CONSUMPTION PRODUCTION \— t" 4OECD (1973). "Supply Control in Agriculture." (Paris: Organiza- 7<3r1 for Economic Cooperation and Development). (_r 142 The economic interpretation of the policy for the individual farm is represented in Figure 5.2. Figure 5.2: Situation Facing Individual Farm Under EEC Sugar Policy MC P 1 _, _________________ I PRICE PB .. , I I I I I . MR PW ‘f I J ( I I I I I - I I L I ' I I 0 a 015] 62 QUANTITY The marginal return curve, efghij, has a downward-sloping portion from f to g and a discontinuity from h to i. From zero output to output 3; price PA is guaranteed as this represents the A quota. For production between a and E}, the producer receives PA less 40 percent of the levy necessary to export the given quantity of sugar in excess of a (i.e., less 40 percent of (PA - PW)’ where PW is the world or export price). Note that the other 60 percent of the levy required to export this sugar is paid by the processor. For production between B} and 5', the minimum guaranteed price PB is paid and for production in excess of Bé there is InO guaranteed price, but only the world price of Pw is received. Note 143 that in 1974 the world price, PW’ was actually higher than the domestic A quota price, PA’ (once processing margins were deducted, since there is no trade in raw beet). Given the marginal cost curve MC, the pro- ducer maximizes profit by producing 01 for which the marginal return is P{. Figure 5.2 may equally be interpreted as the aggregate situation facing all of a country's farmers. Given a time series of Pl/Ql coordi- nates for one of the countries, it would be possible to map the supply response. Such a procedure was attempted, but there were certain incon- sistencies in the data5 and only a few years since the policy began, so that such an approach was not fruitful. However, it is known that all countries, with the exception of Italy, have consistently produced some- where within the B quota, which suggests a marginal cost of production in the 10-17 unit of account per 1000 kilograms of beet range, approxi- mately 3.5-5.5 U.S. cents per pound of sugar on farm. Table 5.1 lists the production, prices and quotas for the EEC-Six over the 1968-73 period. Before presenting the results, it should be noted that the estimates will understate supply by a small margin under policies different from that currently existing. The price variable utilized was the average producer return whereas the correct (but unmeasurable) variable would be the marginal return. In Figure 5.2, the average return under the conditions shown would lie between PA and the marginal 50h occasion the average price exceeded the A price, an impossibility. 144 Table 5.1. Production of Sugar, Quotas and Farm Sugar Prices in the EEC-Six, 1968-73 Year Belgium France Germany Netherlands Italy (Production in thousands of metric tons, raw value) 1968 526 2191 1814 661 1186 1969 618 2503 1903 203 1268 I970 551 2479 1890 657 1096 1971 772 2945 2155 771 1153 1972 617 2744 2040 695 1184 1973 718 2916 2253 765 1040 (Quotas in thousands of metric tons, raw value) A Quota 550 2400 1750 550 1230 B Quota 231 1010 737 231 517 Total 718 3410 2487 718 1747 Year All Countries Exc. Italy Italy A guota B guota A guota B guota (Price per 1000 kg. of beet in units of account) 1968 17.00 10.00 18.46 11.46 1969 17.00 10.00 18.46 11.46 1970 17.00 10.00 18.46 11.46 1971 17.00 10.00 18.95 11.95 1972 17.68 10.40 19.63 12.35 1973 17.86 10.50 20.28 12.85 1974 18.84 11.08 21.70 13.95 Source: EEC (1974). Agricultural Markets, No. 8, Brussels: EurOpean Economic Community. For 1975 prices were increased 15% and quotas also expanded. 145 return, P]. The degree of understatement of supply under a dif- ferent policy depends on the magnitude of the difference between marginal and average returns under the current policy, which is believed to not be very large. The results which follow were selected from a larger set of equations which was estimated. The choice was based on the sign and significance of individual coefficients. France (5.5) log oltf = 14.6422 + 1.6392 log P]t_1 - 2.0999 log PFERTt (3.507) (2.396) + 0.00531 (0.331) -2 R = 0.757 on = 2.321 N = 22 In the French equation, quantity of sugar produced annually was made a function of the price of beet, the price of fertilizer and time. Wheat was believed a priori to compete with sugar beet fOr land, but a wheat variable proved to have a coefficient of a very low magnitude and not significantly different from zero. The +The variables for this and the following equations are defined as follows: 01 quantity of sugar in thousand metric tons, raw value P1 domestic on-farm price of sugar beet in national currency P2 the price of a farm product competitive with sugar beet PIN = an index of agricultural input prices PFERT = the domestic price of fertilizer T = time defined as the last two digits of the year, e.g., 1967 = 67 Figures in parentheses are t values. 146 equation above implies a supply elasticity of 1.64 with a large input elasticity for fertilizer of -2.10. West Germany (5.6) log Qlt = 6.8313 + 0.8699 log P - 0.6098 log P (1.492) It" (0.793 2t“ - 0.0968 log PFERTt + 0.02601 (0.219) (2.615) R? = 0.716 on = 2.634 N = 22 In the West German equation, the supply of sugar is a function of the price of beet, the price of wheat (P2), the price of fertilizer and time. Only the coefficient of the time variable is significantly different from zero at the 5% level, but all coefficients have the expected signs and magnitudes and the result shown is corroborated by the equation which was estimated with land as the dependent variable. The equation appears from the DW statistic to be negatively autocorrelated but this will not have biased the coefficients and an Orcutt transfor- mation was deemed unnecessary. The equation gives an own price elasticity of 0.87, somewhat lower than for France. The cross price elasticity with wheat of -O.61 suggests that the production of beet in West Germany is sensitive to the price of wheat. no» I‘ll 147 Belgium (5.7) log 01t = 3.9886 + 0.2978 10g (P]t_]/P1Nt) + 0.2064 DUM (1.932) (1.455) + 0.0385T (4.088) '82 = 0.681 0w = 2.831 N = 22 In the Belgian equation, the supply of sugar was made a function of the on-farm price of sugar, the input price index, a dummy variable for the years in which the EEC policy was in force (DUM), and time. As in the West German equation, only the coefficient for time is signi- ficantly different from zero at the 5% level, but the signs are as expected and the coefficients of expected magnitudes. The production of beet in Belgium was highly responsive to the beginning of the EEC policy and consequently, as the change in policy was radical, a dummy variable for the policy was included. Prior to 1968 the industry was tightly controlled and, following a spurt in the late 1960's, it seems to have reverted to a condition of low price elasticity. The estimated own price elasticity is 0.30 (deflated by input price index). This estimate of low price elasticity was invariant to alternative specifi- cations of the equation, including those in which wheat featured as an alternative product. 148 Netherlands (5.8) log 01t = 20.1777 +(3213g; log P]t_] -(g.§§§§ log 13w1 -3.8655 log PFERTt + 0.04331 (3.481) (4.337) ”82 = 0.831 D“ = 1.943 N = 22 In the Dutch equation, supply of sugar is a function of own price. the price of fertilizer (PFERT), the price of potatoes (P2), and time. All coefficients are significantly different from zero at the 5% level. The supply elasticity with respect to sugar price is relatively high, being 1.14 and the cross elasticity with potato price is -O.29. The supply of sugar is extremely sensitive to the price of fertilizer. the elasticity being estimated at -3.87. Italy (5.9) log 0 = 2.4342 + 0.5741 log P _ - 0.0306 109 P _ It (1.234) It I (0.242) 2t 1 - o_5437 109 PINt + 0.0156T + 0.3222 log Qlt-l (0.914) (0.923) (1.388) 82 = 0.558 N = 22 h = n.a.6 6 n.a. = not available. A test for autocorrelation, when a lagged dependent variable is a regressor, is Durbin's h test. However, in the current context, the value of the coefficient of log Qlt-l 1s such that the test is infeasible. See Durbin, J. (1970). "Testing for Serial Correlation in Least Square Regression When Some of the Regressors are Lagged Dependent Variables," Econometrica, §§_(3), (May), pp. 410-421. The 5 percent level of significance occurs at h 3_l.66. 149 In the Italian equation, supply of sugar is a function of own price, the price of apples (P2), input prices, time and quantity supplied in the previous year. The whole equation shows a significant relation- ship as measured by the F statistic and, while the individual coeffi- cients are of low significance, the signs agree with a priori expecta- tions and the magnitudes are corroborated by the land equations estimated. The equation is of a partical adjustment kind, with adjustment estimated to be 68 percent complete on an annual basis. Short run elasticities are estimated to be 1.23 for own price, -0.03 for the price of apples and -O.55 fOr the price of all inputs. Long-run elasticities are 32 percent higher than these values. Equations and policy considerations for the three new members of the EEC are now given before making some brief inter-country comparisons of supply equations for the whole EEC. Poligy and Production in the EEC-Three Denmark, Ireland and the United Kingdom became members of the EEC in February 1973 and have since that date been adjusting their agricul- tural policies to the Common Agricultural Policy. This adjustment is in stages and due to be complete by 1977. The data used for estimation cover the years 1950-72 and brief sketches of national policies for that period are given below, since they affect specification and estimation. Denmark A quota, representing domestic requirements, was fixed annually and price was guaranteed for this quota. Any excess production, result- ing from variation due to weather conditions, was sold on the world market and the producer received the world market price. Quotas on total 150 production led effectively to quotas on individual farms via the contracts made between the processors and producers. Production regularly exceeded domestic consumption by a small, but fluctuating, margin. Generally, production was of the order of 300,000 tons of raw equivalent per annum. Ireland In the past, the government did not intervene in Irish sugar production except to the extent of limiting the development of new factories via the granting of licenses. Contracts were made annually between the farmers and manufactureres. Production, of approximately 150,000 tons per annum, did not fulfill domestic requirments. It is interesting to note that during the 19605 Ireland possessed a small U.S. sugar quota and both exported and imported sugar simultaneously. United Kingdom Domestic production was limited by a quota on acreage which reserved approximately two-thirds of consumption for imports from Commonwealth countries under the Commonwealth Sugar Agreement, (begun in 1951). Producers received guaranteed prices for their beet and all beet was processed by the state-owned British Sugar Coporation. Under the new EEC regulations, the 1.4 million tons of raw sugar imported from the less- developed Commonwealth countries will continue to be imported. The degree of expansion of domestic beet production under the higher EEC 7 prices and quotas is as yet not clear. Certain authors anticipated a 7Sturrock, F. G. and Thompson. M. C. (1972). "Sugarbeet: A Study of Sugar Production in the U.K. and Feasibility of Expansion,” ..- 04M I D "r f 1 .55 H CI 151 25% increase in production, but in the 1974-75 season an increase in plantings was offset by disease problems which resulted in an actual decline in production from 1973-74. Considering the relative simplicity of the respective policies, no further explanation is attempted except to remark that, because of acreage restrictions, yield equations were estimated for the United Kingdom (i.e., land area in beet was treated as exogenously determined). The estimated equations were as follows: Denmark (5.10) log Qlt = 9.7879 + 1.2972 log PIt-I - 1.6465 log PINt + 0.0093T (2.866) (0.823) (0.187) 82 = 0.206 OH = 2.257 N = 22 The Danish equation is not an especially good statistical fit, partly because variations in production followed world and not Danish prices. However, it is expected that this equation will prove useful in projecting supply under the EEC policy in which domestic and international prices are more independent. The estimated supply elasticity is relativey high at 1.30. No alternative product was included in estimation, but supply is sensitive to input prices, the estimated elasticity being -1.65. Economic Report No. 7, Department of Land Economy, Cambridge. Also Harris, 5. and Smith, I. (1973). "World Sugar Markets in a State of Flux," London: Trade Policy Research Centre. 152 Ireland (5.11) log Qlt = 3.3057 + 0.0250T (6.359) 82 = 0.652 on = 1.759 N= 22 The Irish equation simply relates output to time. More complicated functions were estimated but were all rejected on both a priori and statistical grounds. No alternative product competing with sugar beet for land could be clearly identified. Price, when included, had the opposite sign from that expected. The equation suggests that price had little to do with Irish supply, but that, rather, production was allowed to expand slowly within current factory capacity. United Kingdom (5.12) log (Qlt/th) = 2.5713 - 0.3847 log L + 0.4363 log P (0.311) It (0.874) It - 0.2684 log PINt + 0.0201 (1.339) (2.812) 82 = 0.563 0N = 2.418 N = 22 and where L = land area. The British equation relates yield, the only decision-variable on output which the farmer had, to land area in beet, the price of sugar to farmers. the price of inputs and time. Of these variables on the right- hand side, only the coefficient of time is significantly different from 153 zero at the 5% level. However, the signs and magnitudes of the other coefficients conform to a priori expectations. Diminishing returns to extra land are indicated and the over-all returns to scale are estimated to be 0.78, which suggests that expansion of the industry may be rather limited for physical reasons. On the other hand, the equation implies an MVP for land in the $60-lOO per acre range, which is somewhat higher than current rents and suggests a considerable expansion when land restrictions are removed at EEC sugar beet prices. The actual expansion depends on the MVP of land for other uses and such an "alternative-use" MVP would have to be estimated for meaningful projections. The equation gives a yield elasticity with respect to price of 0.44. Were all inputs variable and all input supplies perfectly elastic at current prices. the supply elasticity would equal (l/l-r), where r is returns to scale. In this case supply elasticity would be 4.60 and this may be treated as an upper bound.8 (See Table 5.2 for a summary of estimated elasticities for EEC). Supply from Other Western European Nations This section and that which follows are briefer than those on the EEC, due both to less data and less interest in policies. Table 5.3 lists the remaining beet producers of Western Europe and gives produc- tion, consumption and self-sufficiency for 1974. Only in Turkey and Spain was production in excess of 500,000 tons in 1974 and Austria was the only net exporter of sugar in that year. The 8In the complete model the following elasticities were assumed for the U.K.: sugar price, 1.00; fertilizer price, —0.50. .— au- ...-— ’— 0 ' II. 6'- 154 Table 5.2. Summary of Estimated Elasticities for EEC Elasticity Country . . Alternative Own Pr1ce Input Pr1ce Prod. Price Belgium 0.30 -O.30 - Denmark 1.30 -l.65 - France 1.63 -2.09# - w. Germany 0.87 -0.10# -0.61 (wheat) Ireland - - - Italy 0.57 -0.55 -0.03 (apples) Netherlands 1.14 -3.87# -0.29 (potatoes) * * United Kingdom 0.44 -O.27 - # fertilizer price rather than index of input prices * for yield only Table 5.3. Production, Consumption and Self-Sufficiency in 1974 for Other Western European Countries Production Consumption Country % Self-Sufficiency ('000 Metric Tons Raw Value) Austria 403 383 105 Finland 82 216 38 Greece 187 240# 78 Portugal 9 270# 3 Spain 667* 1000# 64 Sweden 301 382' 79 Switzerland 72 ' 287 25 Turkey 834 898 93 #estimated * includes 29 thousand tons from cane 155 general aims of policy in these countries have been self-sufficiency (Austria, Turkey), to increase self-sufficiency (Greece, Spain), farm welfare through protection (Sweden),strategic considerations (Finland and Switzerland) and reliance on colonial supplies (Portugal). An attempt at modeling supply was made only in the cases of Spain, Turkey and Greece, supply in the other countries being assumed fixed or made merely time dependent in the complete model. In Spaig_the industry is closely controlled by governmental agencies which regulate tonnage, price and location. The level of beet production has been relatively stable while cane production has not yet proved successful in any part of the country. There is a quota on total production of 92% of domestic demand, but this has never been binding. All estimated equations using the farm sugar price proved to be nonsignificant and had a priori incorrect signs. The only reliable equation was:9 (5.13) log Q.It = 3.2773 + 0.0475 T (6.366) 02 = 0.664 0w = 1.176 N = 21 In Turkey the industry is controlled by the Sugar Corporation which fixes prices at all levels. There is a quota, which is equal to 90.l = thousand metric tons of sugar, raw value. T = year, e.g., 1966 = 66. P1 = the on-farm sugar price. PLAB = the agricultural wage scale. a... F\U 156 110% of domestic consumption, designed to ensure self-sufficiency in most seasons. The only equation which gave "reasonable" results was: (5.14) log Qlt = 1.5617 + 0.0004 log (P]t_]/PLABt) = 0.0343 T (l 026) (1.924) + 0.4090 log 0 (1.891) It“ _2_ .R - 0.788 L = -0.264 N = 20 In Greece the sugar industry was only established in 1961 and has expanded almost every year since. Estimated equations with price in~ cluded had nonsignificant coefficients and the following time relation- ship was deemed more appropriate for projections: (5.15) Qit = -750.0545 + 12.8000 1 (5.189) 82 = 0.722 on = 1.357 N = 11 Supply from U.S.S.R. and Eastern Europe The situation in 1974 for the U.S.S.R. and Eastern Europe is depicted in Table 5.4. There was a relatively low degree of self- sufficiency in these countries, only Czechoslovakia and Poland being (minor) net exporters in 1974. Before the Cuban embargo of 1960, imports to Eastern Europe came largely from Western Europe (e.g., U.K. and Italy), but this situation was completely changed in the post- 1960 period. Table 5.5 shows the importance of Cuban exports to the 157 Table 5.4. Production, Consumption and % Self-Sufficiency in 1974 for Communist Europe Production Consumption Country % Self-Sufficiency . ('000 Metric Tons Raw Value) Albania 19 35 53 Bulgaria 210# 550# 38 Czechoslovakia 750 653 115 East Germany 570 750 76 Hungary 290 500 58 Poland 1600* 1550* 103 Roumania 560# 575# 97 U.S.S.R. 8526 11250 76 Yugoslavia 425# 620# 69 #1973 *estimated region for the years 1964, 1959 and 1974. Such exports always exceeded two million tons, mostly going to the U.S.S.R. However, over the same years there were agreements covering three million tons to the U.S.S.R. alone. The approximate prices for these exports were 4 cents per 1b. in 1961, 6 cents per 1b. 1965-70, between 6 and 11 cents per lb. in 1971-73, 11 cents per lb. in 1973—74 and 20 cents per lb. after mid 1974.10 It appears that Eastern Europe was a residual market for Cuban exports, willing to accept up to approximately four 10Sources: I with Personnel in U. 8.0. (1963). 92, £13,; Personal Communication S.D.A., E.R.S.; New York Times, 26th January, 1975. r" ‘1. '1 '1 .~ 158 Table 5.5. Cuban Exports to Communist Europe ('000 Metric Tons Raw Value) Country 1964 1969 1974 Albania ll 0 13 Bulgaria 87 205 190 Czechoslovakia 52 224 160 East Germany 81 253 276 Hungary 0 17 51 Poland 32 28 28 Roumania O 69 78 U.S.S.R. 1937 1352 1975 Yugoslavia __43_ __67 __50 Total 2243 2215 2821 million tons annually but only importing less than three million tons per annum. The importance of Cuban exports during this period was that they allowed domestic consumption to rise dramatically, e.g.. from 27.7 kg. per head in the U.S.S.R. in 1959, to 44.5 kg. per head in 1974. Market forces are at work in Communist as in other countries with respect to international trade (via the market for foreign exchange) and the response to higher import prices by these countries might be expected to be lower imports and some domestic rationing. However, demand is likely to be very inelastic since the political consequences of rationing so "basic" a commodity as sugar could be considerable. Just as the U.S.S.R. has imported wheat in the 19705, so it has for the past decade imported sugar to maintain per capita consumption. In 1974 the U.S.S.R. imported only from Cuba, but in 1973 it imported 1,208 thousand tons from other sources (including 458,000 tons from Brazil). 159 The above discussion leads to the hypothesis that the Communist countries of Europe are responsive to world sugar prices both in encouraging domestic supply and discouraging domestic demand. Demand considerations are discussed in Chapter VII, but below the results of estimating supply functions for four of these countries are presented as well as a time-dependent equation for the five remaining countries as a group. Czechoslovakia (5.16) log 01t = 5.2340 + 0.0066 log P]t_]+ (0.089) + 0.2159 log 0 _ (0.867) It 1 82 = 0.000 h=n.a N = 21 East Germany (5.17) log 0 = 5.2028 + 0.2710 log 9 _ ‘ It (2.214) It 1 + 0.1419 log 0 _ (0.677) It I 0.232 m II 5' II 3 b 1P] is world free-market price in cents per lb. 160 Poland (5.18) log thT = 1.6528 + 0.0931 log P]t_] + 0.0037 1 (2.455) (1.244) + 0.6679 log L (3.197) It'1 R2 = 0.643 h = 1 071 N = 21 U.S.S.R. (5.19) log 01t = 1.9919 + 0.1273 log P]t_] + 0.0227 1 (1.264) (1.455) + 0.5971 log Q _ (2 514) It I _2_ R — 0.839 h = n.a N = 21 Rest of Eastern Europe (5.20) Qlt = -1984.2778= 51. 4591 T (6- 95 2) R2 = 0.693 DW = 1.340 N = 22 .1- L1 is land area in thousand hectares. 161 Taking the results in order, Czechoslovakia showed negligible price response, its production being stable over the period of observa- tion. East Germany gave a significant (5%) price coefficient, although representing an elasticity of only 0.27 in the short run and 0.32 in the long run. Poland gave nonsignificant results with quantity equations but the land equation (as shown) gave a small but significant short-run price elasticity of 0.09 which would be 0.28 in the long run. The U.S.S.R. gave an elasticity, (not significantly different from zero at 5% level), of 0.13 short run and 0.32 long run. Finally, the output of the remaining five countries was strongly time dependent. In summary, Communist Europe is an important importer of sugar and not just from Cuba. The hypothesis that domestic supply in these countries was related to world sugar prices could not be rejected for (at least) East Germany and Poland. Estimated price elasticities were of the order of 0.3. Summar In this chapter equations were derived for predicting the supply of sugar from the major European producers as a function of the price of sugar, the prices of agricultural inputs, technology and the prices of competitive products. Particular attention was devoted to a discussion of the sugar policy of the EEC, as this affected specification and was relevant to policy runs of the complete model in which free trade was analyzed. Estimated price elasticities of supply were of the order 0.3 - 1.6 for the EEC countries and approximately 0.3 for most of the Communist countries. CHAPTER VI THE INTERNATIONAL SUPPLY OF CANE SUGAR Introduction As was discussed in Chapter I, cane sugar is produced by a large number of countries within certain latitudes and it accounted for 59% of total world sugar production in 1973. However, since beet production is mainly concentrated in Europe and North America and the sugar domestically consumed, raw cane sugar is overwhelming)y dominant in international trade. This chapter reports the results of estimating supply functions for all of the world's major cane producers (excluding the U.S.A. which was covered in Chapter IV). The chapter begins with theory and models, then passes to data, estimation and results and concludes with a discussion and summary. Theory and Models The characteristics of cane supply to be modeled may be described as technical, economic and political, although their exact division into such categories is not possible. Beginning with the technical, the growing period for cane from the young shoot is one to two years and thereafter ratoon crops may be harvested every one to two years until the yield declines so much that replanting becomes necessary (from three years to thirty years after planting). The perennial nature of cane production requires that its supply be modeled in two steps. Firstly, the investment decision may be modeled, i.e.. the decision to 162 163 plant cane, and secondly the decision on quantity to produce per hectare of cane may be modeled. These decisions are taken in different periods and hence may be modeled independently, even though yield per hegtare may be dependent upon the total area in production. (This contraSts with the approach to modeling the production of beet, an annual, and some theoretical considerations on separation of supply into "yield" and "area" decisions are discussed in Appendix A i] If cane production is viewed as mainly involving an investment decision to plant, the key to understanding the cyclical nature of sugar prices may have been found. In Chapter I the six to nine year cycle of price in the world "free market" was discussed. This cycle is the result both of lags in investment decisions and the fixity of assets once invested. Investments in planting cane and in mills are made in response to high prices. The delay between the price signal and new output is of the order of two to five years. Should price fall, once the new investment has been made, the opportunity cost of the cane and mills is very low, hence production will be maintained as long as variable costs are covered by returns. Such an occurrence is sometimes called "asset fixity" and the capital which, before investment, was viewed as "putty," after investment may be termed "clay." An additional inducement to a cyclical supply, or, more correctly, a cyclical expansion of supply, is political in nature. The sugar industry in most countries operates under government regulation. Producers receive the pooled price of sugar from all markets, both domestic and international; i.e., they receive the average rather than the marginal return. Since the average return exceeds the marginal .,o .J-i 164 return, producers would expand production until marginal costs and average return were eqUated unless (as is the case), there were govern- mental restrictions on output. Such restrictions may be on land, on output, or on both. This regulation reduces the uncertainties facing the prOducer but also further delays the response to high international prices and mitigates against contraction when prices fall. As a preliminary to supply estimation for all countries. a relatively detailed study of supply in Brazil was made. Brazil was chosen as it is the largest single producer of cane sugar in the world, accounting for l % of all cane sugar production in 1973. Production in Brazil is regulated by the Institute of Alcohol and Sugar (IAA) which was established in 1933. The IAA has the power to fix prices and allocate quotas and is the sole exporter of sugar. The situation governing supply and demand in Brazil, were national profits to be maximized, is examined from a static viewpoint in Figure 6.1. The figure portrays the situation when Brazil has a U.S. quota. shown as the completely price-inelastic demand Dus. In addition, there is a domestic demand, Ddom, and an expected world "free market" demand of DW*. Should the IAA act to maximize "Brazilian profits" in toto, it would fix the quota on production at 0* where marginal costs and returns are equal and which would result in a domestic price of Pdom, somewhat above the world price. Producers would receive some mix of world, domestic and U.S. prices. As PW*, the expected world price, and MC, the marginal cost curve. cannot be identified, Figure 6.1 does not lead to an estimable system of equations. In addition, for Brazil there is strong evidence that the 165 Figure 6-1 Brazilian Supply and Demand System Dus Price . . \ Pdom----~ ....... .. , MC PW* Qus Qdom ‘ MRdom MRdom + us Quantity Key: Bus is U.S. quota-demand Ddom is domestic demand Ddom + us is domestic plus U.S. demand HRdom is marginal return in domestic market MRdom + us is marginal return in domestic and U.S. markets Pdom is domestic price Qdom is domestic quantity demanded MC is marginal cost of production Dw* is expected free-market demand Pw* is expected free-market price Qus is quota supply to U.S. . ‘5‘ a 3‘! 1" ,. ur:’ a;- ‘ ... .. .0 u} w 2330 .3 LI! (1 a 166 quota has regularly been fixed at an output above 0* -- producers have first planted their cane and then brought political pressure to increase the quota accordingly. This may be contrasted with the more rigid quota system of Australia, where profit maximization may well have been the chief influence in fixing the quota. Because it was not possible to identify the appropriate behavioral assumption for each country, no such assumption could be incorporated in the cane-investment equations. Some exploratory work with a partial adjustment model confirmed the finding of Choudhury1 that such a model was not very suitable for cane supply. Similarly, polynomial lag_ models did not perform much better. Instead, attention was concentrated on developing a simple investment function capable of generating price cycles through the lags in investment and fixity of assets. The general relationship between hectares of cane and a set of prices may be written. (6.1) HAt = fHA (PPt*, me PIN * PALT *) t’ t ’ t where: HA = hectares of cane PP* = expected average or "pool" price from all markets met = maximum value of PP* ever existing PIN* = expected input price PALT* = expected price of a product competing with sugar cane for resources and t = year 1Choudhury, (1967). 92, ci . 167 Note that all prices are in "real" terms. The expected effect of asset fixity is that response to a rising price, whenever price exceeds the previous maximum, will be more elastic than response to a fall in price. Figure 6.2 demonstrates the "ratchet effect" which is the consequence. Figure (6-2) Asymmetric Investment Function Price HA2 HA4 Hectares HA HA 168 Suppose price P1 results in the planting of HA] hectares of cane. Price then rises to P2 which leads to an increase in planting to HA2. Price then falls to P3 but, due to the low opportunity cost of the cane and fixed facilities, the area in cane falls only to HA3 and there is no retUrn along the expansionary part of the function. If price then rises to P4, there is an expansion to HA4 along the path 3..2..4. To make fHA in Equation (6.1) conform to these asymmetric requirements and to make expected prices estimable, the following assumptions were made.2 Dropping PINt* and PALTt* to simplify the exposition, rewrite (6.1) as (6.2) (HAt/PPt*) = 80 + 81 (PPt*/met) which states that the ratio of cane area to expected price is a linear function of the ratio of expected price to previous highest price. Now assume that adjustment to the cane area/price ratio is only partial so that, (6.3) (HAt/PPt*) - (HAt_]/PPt_]) = [(HAt/PPt*)# - (HAt_]/PPt_]*)]. Equation (6.3) is the familiar partial adjustment model, where # denotes a desired value and O §_Y < l. Combining Equations (6.3) and (6.2) leads to, (6.4) (HAt/PPt*) = Boy + Bly(PPt*/met) + (1 -y) (HAt_]/PPt_]*). 2The development here is analogous to that in Griliches, 2. et a1. (1962). "Notes on Estimated Aggregate Quarterly Consumption Functions," Econometrica, 39, (3), pp. 491-500. 169 The next step is to make PP* explicit. Usually the lag between first harvest and planting is two years and it was assumed that price expecta- tions were 75% derived from price at t-2 and 25% from price t-l, i.e., *= (6.5) ,PPt 0.25 PPt-I + 0.75 PPt-Z. Equation (6.5) is equivalent to a particular kind of the inverted-v lag well known in investment studies,3 although it is further modified in the present work by the incorporation of a partial adjustment system as well. Where cane growing rather than harvested was used as the basis for HAt’ the lag was assumed to be of only one year's duration, i.e., (6.5)' PP * = PP t t-I' One way of viewing the combination of Equation (6.5) with the partial adjustment system (6.3) is to consider Equation (6.5) as the biological lag and (6.3) as the technical delay in expanding mill capacity. The ability of Equation (6.4) to generate price cycles under the condition of a continuously expanding demand is discussed in Appendix B , where analogies with a cobweb system are reviewed. The elasticity of area in cane from Equation (6.4) may be found by clearing PPt* to the right- hand side and taking the necessary derivatives. Should PPt* 3_met , i.e., price be rising and greater than or equal to its previous maximum, the derivative of area with respect to price is, 3"Inverted-v" lags were introduced in De Leeuw, F. (1962). "The Demand for Capital Goods by Manufacturers: A Study of Quarterly Time Series," Econometrica, 39, (3), (July), pp. 407-423. 170 .. * (6.6) BHAt/BPPt* - 30y + 317 + (1 - y) (HAt-1/PPt-1)' Should PP*t be less than met , the area elasticity is, (6.6)' BHAt/BPPE = BOY + 281v (PPt/met) + (l - y) (HAt_]/PP¥_]). Since it is expected that 81 < 0, the derivative (6.6)' will be smaller than that in (6.6) and the corresponding elasticity, defined as [(BHAt/BPPE) (PP;/HA)], will be smaller. This is consistent with Figure (6.2). Before estimation, input prices and the price of a competitive product may be reincorporated into Equation (6.4), assuming the same lags for these two prices as for sugar, which leads to, (6.7) (HAt/PPt*) = 80y + 81y (PPt*/met) + 82y (Ppt*/P1Nt*) + 631 (PPt*/PALTt*) + (l - y) (HAt_1/PP;_]) + Et. In Equation (6.7) an error term has also been incorporated, as a ' preliminary to estimation. The properties of this error term are very difficult to ascertain. Nodoubt (HAt/PP*t) is autocorrelated and the 'partial adjustment procedure may or may not have reduced this auto- correlation. As noted in Chapter III, estimation by OLS of an equation such as (6.7) which has an autocorrelated disturbance and which has a lagged dependent variable on the right-hand side will not normally give consistent estimates. Our knowledge of the behavior of Et is such that it has been assumed ggt_to be autocorrelated, in which case OLS gives consistent and asymptotically efficient estimates.4 OLS estimation 4See Kmenta, 0. (1971). 9p.‘git., pp. 487. 171 of Equation (6.7) for each of the countries was the basis of our results on cane investment.5 Yield may be hypothesized to be a function of sugar price, input price, area harvested and technolOQY. i.e., (6.8) YLD = f (PP PIN HA T t YLD t-I’ t-I’ t’ t)' where YLD = yield of sugar per hectare PP = sugar price PIN = input price HA = hectares of cane T = technology t = a time subscript and all prices are in real terms. In Equation (6.8) prices at t-l have been assumed to be those relevant. Assuming fYLD to be linear and adding an error term At, one obtains (6.9) YLDt=01O + alPP + OLZPINt-I +013T +014HA + A . t-1 t t This yield equation was estimated by OLS for each country. Summarizing this section, for each country equations for cane- investment (Equation (6.7)) and yield (Equation (6.9)) were estimated by OLS. As input prices and prices of alternative products proved nonsignificant or had spurious signs in almost all cases, the final system could be considered to be Equations (6.4) and (6.9). 5There is an implicit restriction on the long-run investment elas- ticity in this approach which limits its maximum to a value of unity. This can be shown by multiplying Equation (6.6) by the inverse of (6.4). 172 Data and Results Data on the major variables were obtained from the F.A.0.,6 statistical yearbooks of the various countries and from a variety of national sugar periodicals.7 The data which were collected included price Series for labor, machinery and fertilizer and any available information on planned or current expansion of facilities. However, such data added little to the reliability of the estimates from the statistical viewpoint and are therefore omitted from further con- sideration. Because input prices were omitted from the most reliable equations, it was necessary to deflate all prices into real (l974) U.S. dollars. .In addition, since domestic sugar prices were dif- ficult to obtain at wholeSale, only export prices from the principal markets [U.S.A., Commonwealth and Free Market (f.o.b. New York)] were used to derive the pooled price received by each country for its sugar. For countries which were importers (e.g., Japan), the free-market price was used. For some countries data on land area were not available and in such cases the investment Equation (6.4) was estimated with quantity of sugar in place of hectares of cane and no yield equation was estimated. Table 6.1 presents the results for the 28 countries (or regions) from estimating the cane-investment Equation (6.4). The data related to the period 1948-72, hence 22 observations were normally included after dropping three observations to initialize the lags in response. The proportion of variance in (HAt/PP*t) explained by the equations was 6Production Yearbook. QB, Cit. 7Reviewed at the I.S.O. in London and the French Manufacturers' Organization in Paris. 1.7 3 0 000 000000000000 00 _0>0_ 20 000 0000; 000 0000>00 FoE0oc 00000000 0 00 0 00.0 4 c 00; 0000-_V00-0:o .00_0000 0000. 000 co000_ocoo _0_000 Lo0 0000 0 00 ; 0.0_000o .000: 00000>Loc 0000000; go: new mc_xoLo magauumxm .00_0 0000-00900 0500 0000.000~ .w000000 0000 000000 waa000 00 0000_000 0000000000 _00.0 000.0 000.00 000.0 00 000.. 000.0 00000.0V 0000.0 00000.0 V 00.0.00- 00000.0 V 0000.00 00000.0 0 00000.00 000.0 000.0 000.00 000.0 00 000.0 000.0 00000.0V _000.0 00000.00V 0000.00- 000__.0_V 0000.00 _0000000 0000000 000.0 000.0- 000.0 _00.0 00 000.0- 000.0 00000.0V 0000.0 00000.0 V 0000.0 - 00000.0 V 0000.0. 0.0000000 000.0 000.0 0000.0 000.0 00 000.0 000 0 00000.0V 0000.0 00000.0 V 0000.0 - 00000._ V 0000.0 000000 0 0a0_0_.0 000.0 000.0- 000.00 000.0 _0 000.0 000.0 000__.0V 0000.0 000.0.0 V _000.00- 000_0.0 V 00.0.00 000_0000 0 000.0 000.0- 000.00 000 0 00 000._ 000.0 00000.0V 000_.0 00000.0 V 0000.00- 0.000.0 V 0000.00 ao_e0< 00000 00_.0 000.0- 000.00 000.0 00 000.0- V00.0 00000.0V 0000.0 00000.00V 0.00.00- 00000.00V 0000.00 000000___0a 000.0 000.0 000.0 000.0 00 000._ 000 0 00000.0V 0000.0 00.00._ V 0000.0 - 00000.0 V 0000.0 0.00 000.0 _0_.0- 000.0 000.0 00 000.0 000.0 00000.0V 0000.0 00000.0 V 0000.0 - 00000.0 V _000.0 000000002 000.0 0V0.0 000.00 000.0 00 000.0 _00.0 00000.0V 0000.0 00000.0_V 000_.00- 00000.0_V 0000.00 000002 0_0.0 00_.0 000.__ 000.0 00 000._ 000.0 00000.0V 0000.0 00000.0 V 0000.0_- 00000.0 V 0000.0. 00_0_0002 000.0 000.0 000.00 000.0 00 000.0 000.0 0000_.0V 0000.0 00000.00V 0000.000 00000.00V 0000.00 0._00000 000.0 000.0 00_.0 000.0 00 0.0.0 000.0 00000.0V 0.00.0 00.00.0 V 0000.0 - 00000.0 V 0000.0 . 00.0e00 000.0 000.0 000.00 000.0 _0 000.0 000.0 0_00_.0V 0000.0 00000.00V 0000._0- 00.00.00V 0000.00 .00.. 000.0 00_.0 000.00 000.0 0. 000.0 000.0 00000.0V 0000.0 00_00.0 V 0000.00- 00000.0 V .000.00 a_0000000 000.0 000.0 000.000 000.0 00 000,— 000.0 00000.0V 0000.0 00000.00V 0000.000 00000.00V 0000.000 000000 _00.0 000.0- 000.0 000.0 00 000.0 000 0 00_0_.0V 0000.0 00000.0 V 0000.0 - 00000.0 V 0000.00 000000 000.0 000.0 000.0 000.0 00 000 0 000.0 00_0_.0V 0000.0 00000.0 V 0000.0 - 00000.. V 0000.0 0.0000000 000.0 000.0 000.0 000.0 00 000.0 000.0 0_00_.0V 0000.0 00000.0 V 0.00.0 - 00000.0 V 00.0.0 _0_0 000.0 00_.0- 000.0_ 000.0 00 000.0 000.0 00000.0V 0000.0 0000_.0 V 0000.00- 00000.0 V 0000.00 00.00000 000000000 000.0 000 0 00_.000 000.0 00 000.0 000.0 000.0.0V 0000.0 0000_.00V 0000.000 00000.00V 0000.000 0000 000.0 000.0 000.0 000.0 00 0.0._ 000.0 0000_.0V 0000.0 00000._ V 0000.0 - 00000.0 V 0000.0 00050.00 000.0 000.0 000.0_ 000.0 0. 000.0 000.0 0000_.0V 0000.0 00000.0 V 0000.00- 00000.0 V 0000.00 003000-00000 000.0 000.0 000.000 000.0 00 000.0 000.0 00000.0V 0.00.0 00000.00V 0000.000 0.000.00V 0000.000 __0000 000.0 000.0- 00_.00 000.0 00 000.0 000.0 00000.0V 0000.0 00000.00V 0000.00- 00000.00V 0000.00 0._e__00 0 00.0.00 000 0 000.0 000.0 000.0 00 000.. 000.0 00000.0V 0000.0 00000.0 V 0000.0 - 00000.0 V 0000.0 00000000 000.0 __0.0 00_.00 000.0 00 000.0 000.0 00000.0V 0000.0 0000_.0 V 0000.00- 00000.0 V 0_0_.00 00.0.0000 000.0 000.0 000.00 000.0 00 000.0 000.0 00___.0V 0000.0 00000.0 V 0000.00- 00000.0 V 0000.00 ae_0000a.< 00_o_0wa_0 . 00000000_0 0 0 0 0 10 111 0000000.0_ 000000000. .000 00 0:00.00 :30-ch w. caxntozm V :00: :00: . z a: NW0 .m.m :1: .u.m in .m.m 0,9,... .mfiflomllllrl 0:o_uo:cm 0coeumm>cV 0:00 .p.o 0.000 174 generally favorable and the standard errors of the coefficients (S.E.) were generally small in relation to the coefficients. However, according to the h-statistic, many of the relations were significantly autocorrelated. As expected, Boy was in all cases positive, 817 was in all cases negative and (l - y) was in most cases significantly different from zero (at the 5% level). As 81y was in all cases significantly (5% level) different from zero, asymmetric response to high (PPt* 3_met) as compared with low (PPt* < met) prices may be said to have been universally present. At the right-hand side of Table 6.1 are listed the short-and long-run investment elasticities at their means for the sample period. These elasticities were derived from the derivatives (6.6)' and (6.6) respectively. The short-run elasticities range in value from -O.679 for South Africa to 0.610 for Mexico and the long-run elasticities range from 0.212 for Bolivia and Chile to 1.007 for Argentina. The long-run elasticities seem to be "reasonable" in magnitude, but the seven negative short-run elasticities, implying an increase in supply as price falls (over a limited range), require some explanation. The large negative values for South Africa and Venezuela are probably spurious, although some fixed-asset theories would be . consistent with an expansion of output when prices fall once fixed assets had been committed.8 The other four negative short-run elasticities are close to zero and not therefore of great importance. Indeed, the important features are that the long-run elasticities are of acceptable magnitude and larger than the short-run elasticities. 8Johnson, G. L. and Quance, L. (1972). The Overproduction Trap in U.S. Agriculture. Baltimore: Johns Hopkins Press for Resources for the Future, Inc. 175 Some comparisons may be made with the results from other researchers. Choudhury,9 using OLS estimation of geometric lags, found only two of his nine chosen countries to have significant long-run price elasticities, those being 1.13 for Mexico and 2.29 for Nicaragua. 10 The present results are lower in magnitude. Ilag found an elasticity of 1.09 for the Philippines (c.f., 0.92 here). Fann gave estimated supply elasticities for Taiwan in the range 2.47 - 2.75 (c.f., 0.42 here). Hughes12 projected an unrestricted elasticity of supply of 3.5 for large farmers in Brazil in 1969 (c.f., 0.67 here). All of these results indi- cate, if anything, that the elasticities estimated in this research are conservative. It is outside the scope of this research to report on specific influences on supply other than price and to discuss in detail policy and projected capacity for each country. For such information the reader is directed to the Attaché Reports made available by the Foreign Agriculture Service of the U.S.D.A. and to the.International 13 Sugar Organization's “World Sugar Economy" of which a new edition is 9Choudhury, P.‘ (1967). Op. git. 10Ilag, L. M. (1970). An Econometric Analysis of the Impact of the U.S. Sugar Program on the Philippine Sugar Industry. Unpublished Ph. D. Dissertation, Purdue University. 1lFan, C. L. (1967). Determination of Sugar Supply Functions in Taiwan. Unpublished Ph. D. Dissertation, University of Hawaii. 12Hughes, H. (1971). Analysis Of Sugar Cane Production in Sao Paulo. Unpublished Ph. D. Dissertation. University of Missouri- Columbia. 13International Sugar Organization. (1963). The World Suggr Econo- my: Structure and Policies. London: International Sugar Organization. 176 being compiled. To further justify the estimated investment equations, however, Figure 6.3 plots actual and estimated area/price ratios for four of the world's largest exporters, Australia, Brazil, Cuba and the Philippines. Turning points are well captured and magnitudes are convincing in all four cases. It is interesting to contrast the cyclical nature of HAt/PP*t in the three other countries with the stability of this ratio for the Philippines, this result being due to Philippine access for most of its sugar to the high-priced and relatively stable U.S. market during this period. Table 6.2 presents the estimated yield equations for the 23 countries which had data on both area and yield. The price of fertilizer was omitted from the table since it was never significantly different fron zero at the 5% level and often had a spurious sign: the constant. on, was adjusted accordingly. The influence of price at (t-l) was also mostly of low significance and omitted. The two important effects were a time trend, as a proxy for technology and other omitted influences, and diminishing returns to the cultivation of a larger area of cane. Combining area and yield equations gave estimates of over-all supply, although yield had only a minor influence on responsiveness to alternative prices. Table 6.3 presents short-run elasticities of supply computed by combining yield and area equations at an export price of 6 cents per lb. (in 1974 dollars) and using 1972 as a base year. Long-run elasticities are not listed, since they reach a maximum of unity at high prices. 1'77 .00000 __0 :0 0000 2000 000000000 000:00_000000 00: 003 08 00 002 00000000 000— 00 000.0 000.0 0V00.0V V000.0- 0000.0V 0000.0 0000.00- 0V0000000 00 000.0 000.0 0000._V 0000.0 0000.0V 0000.0 00__.0 000000 0 00000000 _0 000.0 000.0 0000.0V 0000.0 0000.0V 0000.0 0000.0- 00000000 00 000.0 000.0 0000.0V 0000.0- 0000.0V 0000.0 0000.0- 000000 00000 00 000.0 000.0 0000.0V 0000.0- 0000.0V 0000.0 0000._ 00000000000 00 000.0 000.0 0000.0V 0000.0- 0000.0V 0000.0 0000.00 0000 00 000.0 000.0 0000.0V 0000.0- 000.0V 0000.0 0000.0V 0000.0 0000.0- 000000002 0000.0 0000.0- 00000: 00 000.0 000.0 0000.0V 0000.0- 0000.0V 0000.0 0000.00 00000.00: 00 000.0 000.0 0000.0V 0000.0- 0000.0V 0000.0 0000.0V 0000.0 0000.0- 0000000 00 000.0 000.0 0000.0V 0000.0- 0000.0_ 000000000 00 000.0 000.0 0000.0V 0000.0 0000.0- 00000 00 000.0 000.0 0000.0V 0000.0- 0000.00 000000 00 000.0 0000.0 0000.0V 0000.0- 0000.0V 0000.0 0000.00- 0_0000000 0000.0 0000 00 000.0 000.0 0000.0V 0000.0 0000.0 00000000 000000000 00 000._ 000.0 0000.0V 0000.0- 0000.0V 0000.0- 0000.0 0000 00 000.0 000.0 0000.0V 0000.0- 0000.0V 0000.0 0000.00- 00000000 00 000.0 000 0 0000.0V 0000.0- 0000.0 00:000-00000 00 000.0 000.0 0000.0V 0000.0- 0000.0V 0000.0 00000.0 000000 00 000.0 000 0 0000.0V 0000.0- 0000.00 00000.00 00 000.0 000.0 0000.0V _000.0- 0000.0V 0000.0 0000._- 00.000000 00 000.0 000 0 0000.0V 0000.0 0000.0V 0000.0 0000.0- 000000000 m:—m> mspm> m3~m> z 30. 00 0 0a 0 0a 0 _a 00 0000000 0000 000 000000000 0_a_> .0.0 00000 F 178 FIGURE 6.3 ACTUAL AND ESTIMATED AREA/PRICE RATIOS FOR FOUR COUNTRIES HAt/pr AUSTRALIA so 4D ESTIMATED 20 ACTUAL 1955 1960 1965 1970 YEAR C "A" PP‘ BRAZIL so ESTIMATED ' 1955 1960 1965 1970 YEAR 179 FIGURE 6.3 ACTUAL AND ESTIMATED AREA/PRICE RATIOS FOR FOUR COUNTRIES . ”AI/PP: CUBA 60 40 ESTIMATED 20 .“~’ ACTUAL 1955 1960 1965 1970 YEAR “AI/”I" PHILIPPINES so 40 ACTUAL 1955 1960 1965 1970 YEAR 180 Table 6.3. Short-run Elasticities of Supply (at an export price of 6 cents per lb.) Country Elasticity Argentina 0.4909 Australia 0.3705 Barbados 0.5932 Bolivia-Chile . 0.2044 Brazil 0.4880 China-Taiwan 0.2492 Colombia 0.6750 Cuba 0.3416 Dom. Republic 0.2807 Fiji 0.5468 Guatemala 0.6524 Guyana 0.4207 India 0.3190 Indonesia 0.1000# Iran 0.5444 Jamaica 0.6051 Japan 0.4267 Mauritius 0.4536 Mexico 0.7305 Nicaragua 0.5656 Peru 0.6875 Philippines 0.7390 South Africa 0.1000# Thailand 0.1650 Trin.-Tobago 0.4323 Venezuela 0.5060 Central America 0.7621 Parag.-Uruguay 0.4405 #denotes minimum imposed 181 Summary This chapter outlined the cyclical nature of cane supply due to asset fixity and developed asymmetric land investment and linear yield equations which were estimated for the 28 most important cane producers of the world. Significant asymmetry of response to low as compared with high prices was found. Using 1972 as a base year, at a price of 6 cents per lb. for exports the elasticity of supply ranged from less than 0.10 for South Africa and Indonesia to 0.76 for Central America. At prices above the previous maximum (of 10.68 cents per 1b. in most cases), the elasticity of supply was constrained by the estimating procedures to a mazimum of 1.00. In terms of significance of the relationships, the results compare very favorably with those of Choudhury, who found price to be a signifi- cant influence in only two of his nine chosen countries. CHAPTER VII THE INTERNATIONAL DEMAND FOR SUGAR This chapter reports the procedures used in estimating the demand for sugar, both for more thaneighty'individual countries using time- series data and for seventy-three countries using pooled cross-section and time-series data. The discussion begins with a review of previous studies of a similar nature, passes thence to theory, estimation and results for individual countries' time series, then reports estimation and results for pooled cross-section and time-series data and finishes with a summary and some comments on further research. Previous Studies The definitive study of the international demand for sugar was 1 They fitted consump- made by Viton and Pignalosa of the F.A.0. in 1961. tion as.a function of both price and income to international cross sections for the years 1938, 1951 and 1956. They also examined the degree of substitution between sugar and other carbohydrates in the diet and found little evidence of any such substitution. While the F.A 0. has made more recent estimates for individual countries, these have not been published and the F.A 0. projections of 19712 utilized income as 1Viton, A. and F. Pignalosa. (1961). Trends and Forces of World Sugar Consumption, Commodity Bulletin 32, Rome: United Nations Food and Agriculture Organization. 2F.A.0. (1971). Agricultural Commodity Projections. 1970-1980, Rome: United Nations Food and Agriculture Organization. 182 183 the only independent variable. A comparison of the results of Viton and Pignalosa with the work here reported will be made later in this chapter. A multitude of time-series studies exists of the demand by individual countries for sugar, but no attempt was made to gather and classify this information. However, two studies concerning the U.S.A. are of special interest. In a 1969 dissertation, Young3 utilized a cross section of individual households together with an aggregate time series to estimate that the price elasticity of sugar in the U.S.A. was of the range -0.3 to -0.5 and the income elasticity was approximately zero. In a 1967 article, Hayenga4 reported that the cross elasticity of demand between sugar and other sweeteners was relatively high for certain uses, but that federal regulation limited substitution. Theory, Estimation and Results for Time-Series Data Theory and Estimation Ideally the demand for sugar (sucrose) might be considered as a subset of the demand for all sweeteners, these including both caloric sweeteners, e.g., sucrose, corn syrup (of various chemical compositions) and honey, and noncaloric sweeteners, e.g., saccharin, cyclamates and aspartame. Such an approach could be particularly important for the U.S.A., where the raw material of corn syrup is relatively cheap and a 3Young, K. H. (1969). Demand for Sugar in the United States, a Synthesis of Time Series and Cross Section Analyses. Unpublished Ph.D. Dissertation, Columbia University. 4Hayenga, M. "Sweetener Competition and Sugar Policy," Journal of Farm Economics, 49, (4), (Dec.), Pp. 1362-1366. 184 series of Sugar Acts have maintained a relatively high sugar price. By contrast, substitution of sucrose in Japan and the European countries is toward artificial sweeteners rather than toward corn syrup because of the higher price of corn in these countries (specific data are not available to measure this). Following the recent high sugar prices in the U.S.A., it is apparent that output of high-dextrose corn syrup is being rapidly expanded. It was estimated in 1973 that sucrose had a 78.0% share of all sweetener sales in the U.S.A., dextrose and corn syrup a 16.3% share, noncaloric sweeteners a 4.4% share and honey, molasses 5 and other syrups a 1.3% share. The expansion of high—dextrose corn syrup output reflects not just relative prices but also technolo— gical innovation, since this product was only recently developed. While to estimate the demand for all_sweeteners in a full system of simultaneous equations would have been ideal, the data were not readily available and time did not allow the utilization of this approach. -It should therefore be noted that the demand equations here estimated represent upper limits for the rich nations in which the degree of substitution by sweeteners other than sucrose may be expected to increase in the future. It should also be noted that it is the total demand for sugar which is being estimated, rather than retail demand alone. Much of the 5Walter, B. J. (1973). "Sweetener Economics," Paper presented at the 165th National Meeting of the American Chemical Society, Dallas, Texas, April 8-13, 1973. 185 consumption of sugar occurs in the richer nations in the form of manufactured foods and drinks; for example, 64% of sugar consumed in the U.S.A. in the first half of 1975 went to "industrial" uses.6 The derived nature of this demand for sugar (as an input) may in turn partially explain the highly inelastic response of consumption to the price of sugar in such countries. The demand for sugar in an individual country i at time t may be represented as, where Qit consumption per head, Yit = real income per head, Pit = real retail price per unit and Psit = real price of other sweeteners. Some preliminary analyses for the U.S.A. showed that, using the price of corn syrup for Psit’ (aQit/apsit) had a negative sign, implying a positive cross elasticity between sugar and corn syrup. Equation (7.1) clearly simplifies substitution to an excessive degree and hence the price of substitutes was dropped from further analyses, leading to Equation (7.2): 6A full classification for the first half of 1975 is as follows: beverages, 23.2%; bakery products, 13.6%; confectionery products, 8.6%; canned, bottled, frozen foods, jams, jellies, preserves, 6.7%; ice cream and dairy products, 5.6%; other foods. 5.4%; nonfood, 0.9%; retail (including institutions and government), 33.2%; unclassified, 2.8%. Source: U.S. Department of Agriculture. (1975). Sugar Market News, l, (2), (Sept.), p. 15. 186 (7'2) Qit = fi (Yit’ Pit)’ which is a very simple demand function. The form of fi should accord with certain a priori expectations about the shape of the price-consumption and income-consumption (Engel) curves. Because the functional form is especially important in the second part of this analysis, as price and income range very widely across countries, it will be discussed in some detail. The price-con- sumption curve may be expected to be convex to the origin. Ramsey7 suggests that such a curve may ideally be approximated by a linear and an exponential component, thus, 6 P (7.3) Q = MoO + a1 P) + (1 - A) 90 e 2. o 5_x 5_l, where P price of commodity and 0 quantity demanded. In Figure 7.1 G(P) is the exponential component, L(P) is the linear component and F(P) is the actual price-consumption curve. In the empirical work on food consumption which he reported, Ramsey esti- mated the linear and exponential components of (7.3) separately and, on testing the significance of difference between linear and exponential estimates, found none. Consequently the exponential component of (7.3) alone was adopted in this research (although, as will be shown later, this proved unwise in pooled estimation). 7Ramsey, J. B. (1972). "Limiting Functional Forms for Market Demand Curves," Econometrica, 40, (2), pp. 327-341; Ramsey, J. B. (1974). "Limiting Functiohal Forms for Demand Functions: Tests of Some Specific Hypotheses," eview of Economics and Statistics, 56, (4). (Nov.), pp. 468—477. 187 Figure 7.1 Shapes of Price-Consumption Curves Reviewing Engel curves, Aitchison and Brown8 developed a sigmoidal relationship of the following kind, 6 -l (7.4) Q = e 1 Y , where Q = quantity demanded and Y = income Equation (7.4) gives a sigmoidal Engel curve as shown in Figure (7.2) and which passes through the origin and has an asymptotic upper bound, all of which are desirable characteristics. Countries with low incomes may be expected to lie around the lower inflection point on this curve while high-income countries may lie near the upper asymptote. Aitchison and Brown noted that, by comparison, semi-logarithmic and double logarithmic functions (see Figure 7.2) did not give sufficient curvature at high incomes and had no inflection point at low incomes. 8Aitchison, J. and J. A. C. Brown. (1954). "A Synthesis of Engel Curve Theory," Review of Economic Studies, 22, (1), pp. 35-46. 188 Figure 7.2: Some Engel Curves double- . . - logarithmic stemi—logarithmic \ -5 M 4 Q /// Sigmoidal Y Combining Equation (7.4) with the exponential price-consumption component of Equation (7.3) leads to (for country i in year t), -1 = e[ elyit + eZPit] + (7'5) Qit Uit . where 61 and 62 are strictly negative, and Uitd:N(O, o. 2 It I t)‘ To estimate Equation (7.5), it may first be approximated in the logarithms9 as, _ -1 (7.6) log Qit - a + elYit + 62 Pit + log Eit , where 6 = a constant. 9The approximation lies in changing an additive to a multiplica- tive error tenn. See Ramsey, J. B. (1973). ”Classical Model Selection through Specification Error Tests," Frontiers in Econometrics, Chapter 1, ed. P. Zarembka, New York: Academic Press. 189 Estimation depends on the behavior of Eit‘ For each country, Eit was assumed to follow a first-order autoregressive scheme and Orcutt Transformations were utilized to remove the autocorrelation.10 OLS estimation was then used on the transformed equations. Similar estimating procedures were also followed for the more usual semi- logarithmic and double logarithmic functions (7.7) and (7.8): (7.7) Qit = (10 + a] log Yit + a2 log Pit + log )‘it; (7.8) log Qit = 80 + 8] log Yit + 32 log Pit + log “it . The derived income and price elasticities are -81Y 1 and 92p for the "Ramsey" equation, oqQ'] and aZQ-l for the semi-logarithmic equation, and simply B] and 82 for the double logarithmic equation. Results of Individual Time Series Before presenting the results, the sources of data will be noted. Data on gross domestic product, consumer price index and popula— 11 tion were taken from the I.M.F. Data on consumption came from the 12 I.S.0. and C.E.F.S. and on retail prices came from statistical yearbooks of individual countries and from the annual I.L.0 survey of 10A very standard procedure: see, for example, Kmenta, J. (1971). Ibid., pp. 287-288. HInternational Monetary Fund (various). International Financial Statistics. Washington, D. C.: International Monetary Fund. 12International Sugar Organization (various). Sugar Yearbook. London: International Sugar Organization, 28 Haymarket, w. C. 1; Cmnfié Européen des Fabricants de Sucre (various). Recueil Annuel de Statistiques. Paris: Comité Européen des Fabricants de Sucre, 45' Avenue Montaigne. 190 retail prices.13 For the Communist countries, data came from statistical yearbooks. The data related mainly to the period 1950-72, but in some cases (e.g., U.S.A.) to the period 1950-74. Although data exist for years prior to 1950, rationing was widespread prior to this date and so such data were not utilized. For the Communist countries, since rationing by means other than price is the rule, the results should be viewed as exploratory. Table 7.1 presents the complete set of results for the three kinds of equations. Derived income and price elasticities (nY and np) are also listed in the table for the "Ramsey" and semi-logarithmic equations, while the price and income coefficients of the double logarithmic equation are themselves also the respective elasticities. Testing with the t-values (in brackets), many of the coefficients were not significantly different from zero at the 5% level. Table 7.2 lists those income and price elasticities which were significantly different from zero for both the "Ramsey" and semi-logarithmic equations. A comparison of the estimates from the twoequations reveals substantial differences. 0f the 38 income elasticities which were significant with both functional forms, the "Ramsey" equation gave lower values for 32. By contrast, the "Ramsey" equation gave larger (negative) values for the price elasticity in 15 of the 18 significant cases. The "Ramsey" equation therefore attributed more of the variation in consumption to price and less to income than did the semi-logarithmic equation. A full examination of which was the correct specification was 13International Labor Office (various). Bulletin of Labor Statistics. Geneva: United Nations International Labor Office. 191 wm_.— mo~.o. —vv.o- mmN.o n¢~.o- moo.o vmc.o. mnm.o- —F—.o. www.c- Nmo.o- moo.c m~m.o. Nmo.o moo.a m~o.o. own.— owo.o _ao.o. vwm.o —-.o nw_.o mom.c ch.a emo.o. m_N.o- moo.p o_N.o _,<.o —N~.o vo—.o _oo.o VN~._ vwv.o NON.— F~w.o 50—.0. Non.o mmo.c. _nm.c _No.m mm¢.o «cm.— c~o.~ _mN._ ana.— m-.~ m~m.. com.— «FN.~ cvc.— NNO.» wmm.w va.— moo.~ «No.~ mom.— Non.— oa~.~ _Na.— m—o.m oF~.N cam.— vNo. _ Nae 0 000.0 com.o o~v.o mmw.c _ow.o 39o N36 .3... 37° omm.o ~¢o.c w.w.o —uN.o —a~.o mn~.o ooo.o o—c.o omo.c mv~.o mmo.o mNm.o coo.o an.o 3:... : 2:. : .amo.oV omoo.~. Ammm.NV ooee.m_. “mac. _V exam.” :2. CV om~o.m. o_m—.o~. :3. .8 momw.m. :8. 0V Va_o,mV VmNo.NV .moq.eV h ._ONu._V .emm.nV Noon.on Vacs.mme m_N.N mNm.o omm~.N NNmo._ _omN.Nn . oNQ.N ane.o _N vo—.N mam.o new...nm_ "Voo.a- mama N ..cso..=u ” mucosa VmNo.oV . . .o_N. oV meo.oV _ ANNN.DV V_o_.oV 33.0 33 N. N8 _ coed mEN. a. 3:6. fiend. h BN6. wood. 2 m8.— oood 9399:. 58.9 mmmNN 2.25 Vm3.3 VSmNV VfiNdV V ASENV .SnéV «noo.F. -mo.mn. mmo.~ -m.o Nom~.o. “_No.oV mam... w Nmm.o- a_c.o. a. noo._ xmm.o mm~n.on. o_oe.o mmNo.n Ne.::.a .1 _FN.o . N—m.o NN~.o .-. «NNN.o m_e~.. . ... ... mV .-. .-. ... .-- ... Ne.e:ou .w VNmN.mV Rom—.oV mem.NV . AQQN.o V Vam~.NV exam . amNo ”— coo._ 0Nq.o cmm_.o ”we..F ooVN.n . who a. mmn.o _V mem.. m...o Nexo.o ocmo.o. wNmn.N cacao Ve_e. «V Vnmm NV AN....V Va__.NV Vemm.mV oSNm 2:5 :9. ”$5 338.c2No cafN 27? "85 S at. mgé Naffi. :85. _$<. BE: VmNm.mV VNom VV Vomk.nnV . Acmo._V Vac». mV cam—.0 mme_.~a «am._ ama.e «Nee o .ooN.o a.-.e Foo.c om_.o m_ eo~._ owo.o m¢~n.m mN_o. o. _mmo.n ue._“.. ..Vc Aqm_.NV V Van. oV VONN.NV ..o..oV .m_o.NV Nmmm.o~ mmpn.oo. .FN.N NNo.c Ne.o. o. mmao. V memN.c omo.o. NNV._ a .m~.~ NNo.a meo.~. ammo.o. m.nn.. canesou .Akem.oV AVNN.oV flea” oV Vm_o.cV Ammm.oV mcoc.N o~_m.oN. «MN._ vom.o more o. ammo.o ,NMN.c on..o. emN.o VN cco._ _oo.o seam.en~- mooo.o- com... a»: e.o.e_:aa .mom._V VVoN.~V Vam~._V AOHQ._V Aaem..V «Vac... szN.o~ nmm._ _mm.o wmc_.ow memo.aw cokn.n mN—.o. Nmo.o. mp m.m.. mpm.o _mmo.m. oooo.a _moo.e stages: .mo.. m...N m_mu.o. ._m«.Om. NmV.N ook.o _m_k.o. Namw.c aka... -.. .-. o— --. .-. A... V .-. .-. ..a.>o_aaeoo~u v_~.m Homw.o . 38 NN . V amN.o . x aoa coNe. mNV I muOV. ..m. . ecu aasuV Akce..V Voo_.oV V.V_.eV N. VaeN. oV Ane_. .V N_mm._m nm_m.m~h. Nom._ «_m.o Nemo.o. chN.. and. o- ..o.o. ~._.o «V on._ "_m.o oeoo.o- Noon..- ammo.m .o,= .uaeu VNoa.~V Amon.VV Avoo.vV Aemo.oV me~._NV Nam..mp mmo~.mm NVo.F “mo.o _m__.o _Nmm.o c_mo.o _oo.o VmN.o a. m__.p oVo.o Nemm.o akoo.c- nNm~.n ._aso_eu Amoc.mV Aem~.NV ANNN.NV .Nma.NV Amoa.cV ccpm.o mVVw.o~ .o..~ mNN.o mmmN.o. ,m_m.a mmmc.m mmN.o. om_.o m. man.~ cm~.o wcso.n.V .NNo.c. some." =.x..~..=.gu _Ne mm scam.o . 5 ON . a mmm o . foam. - cao>V onV. no. . _Nem. MMNx. . ecu ~e.;uV Ammo. NV Amoo. V Heme. FV Nu ANNN. aV ANNN. _V _Nem m .mmF.N_ °c«._ -_.o Nose. m~m~.o n.a».N e.g.: aao.o _N mNm._ mN_.o mo_a.c oweN._. coke." o__gu VnmN.~ Amkm. V A.ON.NV VaNo.oV Acme.~V moON.N mmma._a “mm._ Ne~.o Nch. - new—.0 m.m—.~ .mmo.o. e.g.: «N on._ N.N.c n.ao._N. ..oo.o. mocc.. cvueau 22.: 3:. V 2%.: 83.8 $3.: mmom.V mowm.~ Noo.~ VNN.c ewes. mmow.o Famo._ mme.o NMN.o a Nom._ mo~.c «Neo.c mpo_.o. ".80.. esocnseu Vokm.oV Vere. V Am~o.oV Vans..V Amma._V meme.” omm~.ofi amm._ coo.o memo. mam.n_ _onm.__ ch.o oom.c o mmm.. moo.o moo—.0. m_oo.o. NmNa.N guts. Amme.eV Venn. V Amem.nV VN_M.NV Anew .V NOON.0N No_o.VN o.N._ omm.o QONN. . comm.o mNeN.n o.~.o. mo~.o .N NaN.. .oo.o omom.o. .NN..o. coo... _*~eca AmNm.__V Acme. V Ammm.o_V figme.aV Ana..__V oNeN.on ova—.om ewq.N o_o.o OOVN. - Neme._ moNN.o. .oN.e- ....N a. ove.~ u_m.o ~__o.a. mean.~e. moko.- ..>__an RN.N.NV V_N_. V .emo.NV Home.oV Ammo.NV NNN¢.N_ mo_o..e VmNo.. _ee.o mm_o. weeo.c mmm~.m occ.o m_N.o m. .a_.N nkk.o owNe.c movN.c- o_mo.. .s.a_om V_mo.oV m Amok. V Vooo._V NQNN m. News _N .ooo.~ Nae.o NNNa. - mo__ 8. _Nom.n .-- -.. a, .-- - - ..- - --. moo-ac.» VmM~.oV V AmNN. V AomN._V “can.cV “we“.pV meeo.n, NNma.oo “oeo._ nmm.o MVQV. macm.o momm.. no. a. me~.o a. -N.N mmc.o moo—.n Nm__.a- _Nmm.m ..c.a=< Vme._V V Ame. V aNno.oV AND—.9V .cNm. _V exam «. oos~.am “Noo._ swo.o mace. awxo.c. .N.~.m N_o.= moc.o. _N cko._ “no.0 ..am.~ ”_oo. o «Noa.m .,_.c.ms< flame.NV H Ream. V VmNe NV V_ck.oV “Nam NV woma.m. NNNn.mn ”mmo.. can.o coke. - ”aka 0 Nwmm.m omc.a. ”_m.o up mma.o ..n.a on_m.o. swan. o. m.~o.o .e.a=umc< Ame°o:_V . . u u x . U ” “no.5mV Ausauc—V M _ ..caumcou an Neaumcou _ . m p n :n m N: N m _. mueuuucou _ ..-. -11.- . . V . a . mu.=»uu m~_.om-i-__ ...N o-n- 192 ...... V 3...: m . ...... V ...: V :2. V 3.... V .555. .8... 25.. 5...... .3... . 3...... 85.... 3.. . mom... ...... . ...... 23.. .85. 8.5 a 2.... .....5 5.5.. . :3. . 33.... E... ...... V 38.: . ...... V 3...: 3... V ....N. V ...... 2.... .53.. 53... .....N. 3...: 35.55 5... . 53... z... ..3. 22.8 3.5 8... 5. ...... .35 .3... . 5.5... . 5.2.. .52.... 38. V :5: . 3... V .2... V 3... V ...... V .....5 .....N ...... 85..., . .35.. N53... 5.... .3 .. .....N. . ...m... 5...... .35. 25... ... ....N 5...... mm... .. .N... . 53.... .25.. :8. V 3...: 3.... V 3:. V 3...... V .52. V 3:. ...5 .8. ..N... ...... . 5...... .8... 35.. 3.5 8%. . 82. ...»... 8.5. .2... .. 5...... 5...... 3.... . 5.5... . .25.. :35... .55.. V 3.5.: .58. V .3... V 8.... V 38. V ....5- .555- 855 5:... 53... . .525- 3.5... m3... ...... 53.. . 3.... . SN... ...... 3...... 5 555... ...... 33.... 3.... ...... 5 .35... .5 9.3:. ......V 39. V .3: V .3... V 3...: :2. V 3.5. .2... 5.5.. $55 3...... 2...... . 55.5. 25.. NS... ...... . 8:. 3...... 3.5. 5:... ... 3.... 535 5...... ~.. 5.5... . .52.. 55...... 5.. 3...: 3.... V .5... V :3. V .8... V 3... V .25. :55 85.. ...... 22..- 53... $3.: 58.. N3... 552. . 25... 5...... 52.... ~55 ... ...... 8.... 8.... .. 3.5... . 8.5.. $5.55. 3...: 3.5. V 33. V .2... V ...... V .5... V 3:. .25. z... 55.... 55.... . .3... . 8.55.- 5.. N35 ...... . 3:. . 3...... SN... .55.... 5. .3. .3... 3.5. .. ...8 53... 5885.. ...... V 3.5. .V ...N. V .53. .V :3. V .25. .V N85. 53.0 «RN 35.0 R... . .25. . 3mm... .....N $5... 8:. . «mom. 2:... m8... 3N... NN woNN Gad N.N.. - :8. . n3... 8.5: ...: V 3.... V ...... V :5... V ...... V ...... V S5. 3...... 3.5 .85. 2.... . ...... . .35.... a... .85 ....N. . 83... . N3: 8.5. 5555 a. 3.5.... :55 5.2.... 85... N35... $3.3! . .55.. V .53. V 3... V :3. V 3.... V .58. V 3.5 3.5 .....N .8... 5...... 53.... .3... 3.5 5...... 55:. 3.... .....m as... ...... ... 2... 58... .3... 3... . .555... .8355! .55... V 3...: 3.5. V 3.... V . 3.5. V ...... V .35. ...... .....N ...... 3... . .2... 835 3... 3.... ..Nn. . ...... .....N .35. .3... . 3.5 555... 2.... - N5... . 3.... 23... :3. V 35...... 33.: ...... V 3:. V ...... V SN... .2... 3N. Nam... Mom... .V wow... V 58.6 .8.N of... 32. - Nmon. 3:... 23.5. Emé o. mmm.. now... “So. w .32. w NNomN .533 3.8. N... ...... m... 5... 3:... 33.0 03.. Nam... 33. . 3m... _ 0.8.3 53.. ..m... SON. - EN: _N.N.m N.N.o. 2.... 2 2m..— .mmd SNN. T :8. - 8:5 1.5.. 3:. V 3... V 3...: .2... V :3. V ...... V 53.... «N... 89. Rm... 3.... . ...... .Nmm... .Nm; 2w... #3. - $3. SEN .36- .86 .N. .3”.— .Ed .3”. . EB. - men...” :32. :3. V 3...... :3. V 3.... V .3... V .m... V 5.5.... 535 ....N 3.5 3...... 3.5.3 28.3. x: 55m... 5...... . NN... 58.5 355. m8... : 3.5 3.... 2.... m- .58. . 33.. 3.1.... .25.; 3...: ...... V :3. V .58. V ...... V .35. 2.... 8... .35 ....N... . .53.... 33.... 53.. .35 2... . ...... 5...... .35. N25 ... a... .3... ...... - 85.. . 33.. .38 to... .53. V ......V ...... V 35... V ...N... V ...... V .35 3...... ......N 5...... .35. . .....2 33.... 5...... 2.5 #5.. 5m... 38.. ...... 2.5 ... 3.... 3.5 53.. 55.... . 5.8.. ...: 33. V 325V .5... V 33. V :3. V ...... V ...... ...... 5.5.. 3.5 .35. N 3.5.... 85...: .55.. £55 55.... 53.... 5.5.... E... .85 .. 3... NS... 5.55.... 5.... . 35... 3.5. :3. V 335V :3. V 3.... V ...... V .3... V 9...... 5%... 3... NE... ...... .. £3... ....5... 53.. ...... R... . .....m. 5...... 5...... EN... ... 5.... 5...... ...... 2- 85... . 3.3.. 5.3.»... ...: V .....NV .80. V .5... V .25. ~35 .3. RN... ...... . NS... .35.... ...... ...N... 3.... . 8.. NE... .25. ...... : .... 3.5 ...... - ...... . 53... a... :3. V ......V ...... V ....5. V :3. V :3. V 55.5. 3.... .....N 8.5 .35. . :3... $3... mmmN ...... ...... . .N... 3.5.. 8.5. .25 n. 3.5 ...... 85... - 8.... . 855.. ... .53... 82.2 .3: V 33.: 3... V 8...... V 53... 535. a... 5:... 3.5.. 3.....- N3... ...N. 2.... 8.... 5:... . 235 m8... 855. 5 N... .2... 3.5. .8... 3.... .328... 32.: 355.8 :5... V ...... V .58. V ...... V .535. 5R... 2... 5...... .23”. $3... :2... .55.. 855 .53. . 5.... 83.. 5...... 2...... o. 35.. $5... .35... N8... . 5.5... :5... ......3. 9.3.5.. .53.: :.N...V .53.: 3...: 35...: ..."...V 525- 355 5...... 85... £35. 2.5... 335. 585 5...... .....5. 53.... ...: 3...... ~85 5. .....N 85... 58.....- 558..- 5:: 5:3. 23.. mob-6:3. 33.: ......V ......V :8: _ .58.: 33.: .85. 8.5 ......N 3.5 5...... .535 8...; 535 ...5 5.555. 5...... 38... _ :55. .35 ... 5...... 2.... 2.3.3- .855. .53.. ...-.5 3.5.3:: _ m _ _ ~ 35...: 363:: V... 3:333 IV _ a: c N. _ a: . ... _ ...—Jug“. W ... .... p _ I. N: Nr .0 222.8 . _ . . V .....-.. ...... 5.5.2.53 .... an _ a . .. 223:3 .... o3 . 193 55..5 ..5. .5 .55.5-. 55..5 ..5. .5 .5..5-. ..5..5V .5...5V V .5...5V ..55.5V _ .....5. ..5..5V ..5.5 ....5 5.5.5 555.5 .555.. 55.5.. 55.5... V55..5 555.5 5555.5- ...5.5 5..5.5 555.5- 555.5 5. 5.5.5 555 5 5555..5- 5555.5- 55.5.. ..5M5. u_>e—mo 3> .m.5.NV .555.5V . ..5..~V ..5..5V . .55..NV .....5V 55..5- 55..5 .55.5 55..5 .5...5- .....5 55.5... V.55.. 55..5 .555.5- ..55.5 5..5.. 55...- 5.5.5. .. 555.. .5..5 5.55.5- ..55.5 5.5... .5.=55V 5.5.5.5 ...5 .V .55..5V V .5...5V .55..5V V .5...5V .5...5V .....- ..5.5 555 . .55.5 ......5. 55 5 . 5555..5N- ...5.. 555.5 555.... 5555.5 55.5..- . 55...- 5.5.5 .5 555.. 555.5 5.55.5..- 5.55.5- .5.~.5 5.55.5555 .555 .V .555... V ...5 .V .555..V V .5.5 .V .555..V .5..5- 555.5 5.... 55..5 5.55..- .5.5 .5 ~55...5 V555.. 55..5 .m5..5- 5..5.5 5555.. V 5...5- .55.5 .. 555.. ....5 5.55..- .555.5- .55... 55555.5 . ..55..V .5.. .V . ..55..V .....NV . ..5...V .555.NV 555.5- ..5.5 ..5.. 5...5 555. .- ..5. . 555... V5.5.V 555.5 .555.5- ..55.5 5.55.. V 5.5.5- 55..5 5 555.. .55.5 555..5- .55..5- .555.. ...5> .5555 .5.5... ..55..V V ......V ..5 5. .V V . ......V ..55..V ..5.5- 5...5 5.5.. 555.5 5.....- 555. 5 5555.55 ...... 555.5 55.5.5- 55.. 5 5555.. .5.5.5. 555.5 .N 555.. 5.5.5 555....- 5.55.5- 5555.. ...5.5 ..55.5V ......V V ..55.5. .555._V V .5.5.5V ..55..V 555.5- 55..5- .55.. ....5 55.5.5- 5 5..5- 55.5.5. ..55.. 55..5 5555.5. .55. 5- 5555.5 555.5 .5. 5- 5. .55.. 5.5.5 5.55.. ..55 5 .555.5 ...5 .5.5.5V .555..V V ..55.5. .5....V .5..5- 55..5 .5... 55..5 5.55.55- .m.5_5. ..55..5. ".55. . 55..5 5555..- 5555.5 5..N.. 5. .5.5.5.5 .555.5.V .5...5.V V .......V .55....V ..55.5V ......V 555.5- 5.5.5 555.. 555.5 5.55.5.- 5555.5. 55.5.5. V..5 . .55.5 .....5- .5...5 55.5.. 55..5- ....5 5. ..... .5..5 5....55- 5555.5- 55.5.5 .5..5. .55..5V ..55.5V V ..5..5V ......V ..5. 5V ..55.5V 555.5- 555.5 555.. 5.5.5 .5555..- 5_.5.5. 55.5... V 5.... 555.5 55...5- .55... 5.55.. 555.5- 55..5 N. 55... .5..5 .555.5- 5555.5- 5555.. ....ca. .55..5V .5.5..V V .5...5V ...5..V .5...5V .....5V 555.5 55..5 555.5 555.5 .55... 55.5.5. 55.5.5.. .555. 5 5.5 5 5555.5 5.55.5 555... 555.5 555.5 5. 555.. 55..5 5....5. .555.5- 55.5 . 55.55.55.5V5... .5.5 .V .5.5... n .555 .V ...5..V .5.5..V .5.5. .V ..5.5- 555.. 555.. 5.5.5 .5...- ..55.5 5555 5 V 555.5 555.5 5555.5- 5555.. 5555.. 55..5- 555.5 5 555.5 5.5.5 .5.5.5- 5555 5- 5.5..5 555. .55..NV ...5 5V V .555... ..5..5V .55N.NV .5....V . 55..5- .5..5 .5... .5..5 55.5..- 5... 5 5.5..55 .555 . 55..5 .....5- 5.55.. ...... 5.5.5- .55.5 5. 555.. 55..5 ..5..5.- 55.5.5- 555... 55...... ......V ...5 5V V ...5.5V .5...5V .5.5.NV .555.5V . ....5- 5.5.. .55.. 55..5 .555..- .555 .. 5.55.55 _5 . 555.5 5... .- 5555.5 5555.. 55...- .55.5 N. 555.. 5.5.5 555...5- 5555.5. 5555.5 .....5.. .5.5.NV .55.... . .5.5.~V .555..V ..5..NV .55..~V 55..5- .5..5 .55.. 555.5 .....5- 5...... .5.5... V..5 N .5..5 5... 5- ..55 5 .555.5 555.5- 555.5 5. 5.5.. N...5 5.55.5.- 5.55.5- .555.5 ...55 .555..V .555... V .5.5... ...5..V ......V .555..V 5.5.5- 55..5 ..5.. ....5 55.....- 5555.5 ...5.5 V5.5 . 5.5.5 .5...5- 55...5 .555.5 555.5- 555.5 5. 5.5.. 555.5 .5.....- .555.5- 5555.. 555......35 .....5V .5....V .55..5V .555..V ..55.5V .555..V 555.5- .55.5- ..5.. .55.5 ..55.5- .555... .55.... V5.5.N 5.5.5 55.5.5- ..55.5- ~55..5 555.5- .55.5- 5. .55.. 5.5 5 55.5.5- 5..5.5 555..5 555535 .55...V .....5V V .5..... .555.5V .5.5..V .555-5V 555.5- 55..5 5.5.. 5.5.5 .5.5 5.- 5.5... .5...5. V5.... 5.5.5 555..5- .55..5 5555.5 .55.5. .55.5 5. 55... 5.5.5 5555.5- .555.5- .55..5 5.555 .555..V .55...V V .55...V ......V .5....V .5.5.5V 55..5 5.5.. 555.. .5..5 ..55.. 55.5... .555.5.. ...5. . 555.5 55... .55... ~55..5. 555.5 ....5 5. 555.. 555.5 .555... 5.55.5- 5.5..5 ..55. ..5 .5.5.NV .55N.5V V .555..V .55..5V .5...5V ..5..5V .5..5- ..5.5 .55.. .55.5 .....5- 5....5. .55~.55 V5.5 5 ..5.5 5.55.5- 5.5..5 .5...5 555.5- 555.5 5. .5... 555.5 ..55.5- 5.5..5- ...... 5..55 V ..55.5V ..55.5V .55..5V ...5.~V 5.5.5- 55..5 5...5 555.5 5555.5.- 5555.5 ...5.5.- .5...5 555.5 5.55.5- ..55.5 5555.5 .55.5- ..5.5 .5 555.5 555.5 ...5.55.. 5555.5- 5555.. ...... 5.555 .55..5V .5.5.5V V .555.5. .555.5. .555.5. ...5... ....5- 555.5 555.. 555.5 555..5- 5.55.5. .5.5... .5.... 555.5 5.5..5- .55..5 ...5.5 ....5- .55.5 N. ..... 555.5 55.5.5- ..55.5- ..55.. 5.55.55.5 .555.5V .55..5V V ...5.5V .555.5V .5.5.5V ..55.5. 55..5- 555.5 .55.5 555.5 .555..- 5555.5 .555... V..5.. 555.5 .55..5- 5555.5 55.... .5..5- 555.5- 5 5.5.5 555.5 .5.5.5- 5555 5 ..5..N 5555. .....5 .555..V .55...V V .5.. .V ..55..V .555..V .55...V 555.5 555.5- 555.5 555.5 5.55... 55...5.- 5..5... 555 5 555.5 5555 5 55.5.5- ~55... 555.5 ~55.5- .. 5.5.. 555.5 .55... .....5 5.55.. ..55555 5.5... .53.5 .5....V .5...5V V .555._V ..55 .V .55.... .55..5V .5..5- 55..5 .55.. 5.5.5 5.....- 5.55... 55.5.5. .... . 5.5.5 .55..5- 5.5.. 5 .555.5 55..5- .55.5 5. 555.. 555.5 5.55.5- 5555.5- .555.. ..55..55 .55..NV .5.5.5V . .55...V .5.5.~V V 5...5- 555.5 55... 555.5 55.5.55- 555.... 55.5... ...5._ ..5.5 ..5..5- 5.55.5 5555.5 5. 55..5. .5.~.5V .5....V V .....5V .....5V V ..5..5V .5.5.5. 555 5 5...5 5.5.. 5.5.5 5555.5 555.... .55..55 .....5 5.5.5 55...5 55.5.5 5..5. . V 55. 5 5.5.5 .5 -..5.. .55.5 5555.55 ..55.5- 5555.5 555.55....5 .l‘ v. ... I II I]: III ..|u|-t'll.v|«l .. JIV’oO..In'.J.III.|u-l‘u. -I0iJ-|l||II|II|-J.uAJuogil‘i- ..... .1 -kl-Io- nl!.I»D--.I'In «vb-11.- .l!.l|ll...'b|rlo.[ll€ l.!.|r--.V4.It-uo-III'I.ID.DII:I-IIA .tlEl-JI'JIIIIIIIP V . V V V Ami-...: V 355...: V . V V V V V V 32...: 3.52.: a. ..,. _ 10 m N... V m. V ... V3328 V 3a V ma V a. . N. V 5.. ¢ __ V 2.35:8 . a: a: V h V N« m... .., 5:323 ¥ .1.- ..... -.-. .--.|-.. ”-12-IV. .. 141--391»! .1 5 -- it. -. ... .------.-_-----i-I-.«.-.-----.-.zo: - a. .iL-. 35 ----3-31.5! - ...... 32.-1:51-; -11------.--------------5.:._.....Cow-o..-.Ewnzii-ii-I----3-1---- ---- ....... ----5.552503 5.2.8----21-1: .. -------: .......... ---------1- ----- x5553.-- ....... ----------I-I-In--- .- -I|| vl.....- - I . ...b‘\ I‘- ..III .....I .v- . . .Illl".II-.lll-Dilnl'lI.-II'\-‘§|p E. u. .-I blIO-I . >‘l. ..... Iii-I'll'll'- 555:..co5 .... 5.5.. 194 Table 7.2. Significant (Income and Price) Elasticities From (Ramsey and Semi-Log) Equations Country Income Elasticity Country Price Elasticity Ramsey Semi-Log Ramsey Semi-Log Bolivia 2.444 1.202 Venezuela -1.494 -l.242 Pakistan 1.400 2.072 Tanzania -1.200 -0.774 Madagascar 1.318 2.169 South Vietnam -1.102 -0.723 Ecuador 1.172 0.892 Thailand -0.675 -0.358 Upper Volta 0.702 0.543 Ivory Coast -0.591 -0.551 Thailand 0.667 0.734 Dom. Republic -0.436 -0.548 Libya 0.624 0.371 Portugal -0.430 n.s. South Korea 0.570 0.737 South Korea -0.389 -0.256 Tanzania 0.503 1.522 South Africa -0.381 -0.370 Tunisia 0.463 0.868 N. Germany -0.352 -0.441 Spain 0.389 0.547 Japan -0.344 -0.260 Sri Lanka 0.374 1.325 France -0.330 n.s. Belgium 0.374 0.226 Syria -0.326 -0.339 Peru 0.366 0.568 Norway -0.275 -0.141 Gabon 0.355 0.824 China-Taiwan -0.258 -0.228 Guatemala 0.346 1.320 Brazil -0.246 -0.224 Argentina 0.315 0.341 Morocco -0.206 -0.238 Iran 0.301 0.305 Bolivia -0.204 -0.158 Mexico 0.253 0.499 Spain -0.203 -0.181 Cameroun 0.238 0.691 New Zealand -0.159 -0.157 Ivory Coast 0.222 0.826 Belgium 0.215 0.421 Philippines 0.213 0.718 Ireland 0.212 0.289 Israel 0.207 0.745 Brazil 0.205 0.484 Portugal 0.201 0.465 China-Taiwan 0.186 0.411 Finland 0.180 0.183 Iraq 0.144 0.422 Costa Rica 0.142 1.002 Turkey 0.114 0.512 Syria 0.098 0.451 France 0.094 0.221 South Africa 0.041 0.199 Canada 0.040 -0.164 U.S.A. 0.030 0.119 Japan 0.014 0.254 Trinidad and Tobago -0.030 n.s 195 outside the scope of the present research, but some comments on the shapes of the price-consumption and income-consumption curves may be instructive at this point. Since the "Ramsey" equation yields higher price elasticities, its price-consumption slope must generally be greater than for the semi-logarithmic equation, as depicted in Figure 7.3a. Similarly, since the "Ramsey" equation gives generally lower income elasticities, its shape would appear to be as in Figure 7.3b, having the lower point of inflection very near the origin. Figure 7.3 (a) Price-Consumption Curves (b) Income-Consumption Curves Consumption Consumption Semi-logarithmic per per head head "Ramsey" emi-logarithmic "Ramsey" / Price per unit Income per head Using the semi-logarithmic results as a standard, the range of price elasticities for 1972 is from ~l.242 for Venezuela to -0.l4l for Norway. The fact that many price elasticities were not significantly different from zero indicates unreliable data in some cases and a very low price elasticity in others (e.g., Indonesia and the U.S.A. respec- tively). The income elasticities ranged from a high of 2.169 for 196 Madagascar to a low of 0.119 for the U.S.A. Many of the European countries and Australia gave negative income elasticities which were, however, not significantly different from zero but do indicate that their consumption is at saturation level and sugar may border on the classification of an "inferior good." Looking more losely at North America, the results for the U.S.A. and Canada gave significant income elasticities of 0.119 and 0.164 respectively, using the semi-logarithmic equation. However, it is probable that the estimates of 0.030 and 0.040 respectively from the "Ramsey" equation are a truer representation. Similarly, price elasti- cities for these two "highest income” countries were -0.044 and -0.051 respectively with the semi—logarithmic specification, and -0.028 and -0.055 respectively with the "Ramsey" specification. In neither case was the price elasticity significantly different from zero. The above elasticities are for 1972 prices, but at the higher retail price of 1974, the Ramsey specification would give a higher price elasticity of -0.054 for the U.S.A. and -0.124 for Canada, since the formula for the price elasticity is price dependent, namely, Up = BZP . From 1973 to 1974 there was a 92.78% rise in price and a 2.15 % fall in consumption in the U.S.A., indicating a crude elasticity of -0.023. However, the meteoric rise in price occurred late in 1974 so that full adjustment was not captured by the 1974 data on consumption and the "Ramsey" estimate of -0.054 for price elasticity would appear of the correct magnitude (c.f., semi-logarithmic estimate of -0.044 at all price 14 levels). Finally, the estimates of Young of -0.3 to -0.5 for price elasticity for the U.S.A. appear unreasonably large. Further ”Ibid. 197 comments and comparisons follow the pooled demand estimation which is next reported. Theory, Estimation and Results for Pooled Cross-Section and Time-Series Data ‘The advantage of pooling the individual countries' time-series data lies in the increased degrees of freedom thus obtained. In estimating individual countries' relationships from time-series it was found that three major problems arose from the data, namely: (i) countries with relatively low incomes have only short series of data which are also often unreliable; (ii) at high levels of consumption the variation in consumption is very low, hence elasticities are inefficiently estimated; and (iii) "any countries have maintained relatively stable retail prices for long periods thus leading to inefficient estimates of their price elasticities. Pooling the data may help increase the efficiency of estimation. 0n the other hand, the elasticities derived from a pooled analysis may be ex- pected to differ from those of simple time series. It is sometimes suggested that cross-section results be considered lgng_run_in nature and time series §h933_§gg, This argument considers the pattern of tastes to change as a country's income rises so that, in the long run, elasticities may be higher than those estimated from a times series (of fifteen to twenty years). The cross-section analysis of Viton and Pignalosa,15 for example, may be considered to give such long-run elasticities. In a pooled study, the derived elasticities will lie somewhere between the short- and long-run extremes, probably being nearer the long-run kind since there may be less variation in the time-series than in the cross- section data. If the "taste" variable could be identified, its 15Ibid. 198 inclusion at a given level for each country would lead to pooled esti- mates exactly equivalent to time-series estimates for the individual countries. In this research, pooled estimates were made in which each country's "taste" was assumed to differ, this difference being accommodated by the inclusion of (N-l) dummy variables, where N is the number of countries. Writing the model in its simplest form, (7.9) Qit = f (Yit’ Pit’ Ni). t = l. 2, . . . , T; i = 1, 2, . . . , N where i = country i t = year and N. = dummy variable for this country. 1 In the "Ramsey” functional form the equation becomes, N-l 1 (7.10) log Qit = .2 wj + 5 + e Yit + 92p + log E1 J=1 1 it t which differs from Equation (7.6) only in the inclusion of the dummy variables. The behavior of Eit is assumed to be as follows: 2) = 0.2 , i.e., heteroscedasticity in cross section; (i) E(E 1 it (ii) E(E E = 0, i.e., cross-sectional independence; it’ jt) (iii) Eit = pi Ei,t-l + Uit’ i.e., autoregre551on in time series; . 2 . (W) ”11:” M0. om- ). (V) E10” M0. 0111 ); _ 2 pi . ) (vi) E(E = 0, for all i, j. i,t-l’ UJt 199 From the above it may be deduced that Generalized Least Squares (GLS) is the appropriate procedure for estimation, the matrix of disturbances being corrected for autocorrelation and each country having its own correction for heteroscedasticity. GLS was accomplished in two stages, in the first of which each country's data were transformed to remove autocorrelation. and OLS was conducted. The residuals were then used to correct for heteroscedasticity and OLS was again conducted to give the final estimates which should be unbiased and asymptotically efficient.16 Similar procedures were also used in estimating the semi-logarithmic and double logarithmic functions. Using exactly the same approach, but dropping the dummy variables, a set of "longer-run" pooled estimates was also obtained for each of the three functional forms. It should be noted that, unlike the situa- tion for the individual countries where domestic currencies and deflators were used, in the pooled estimation all prices and incomes were converted to constant (1974) U.S. dollars, using the U.S. consumer price index as a deflator and end-of-period exchange rates given by the I.M F. Results of Pooled Estimation Seventy-three countries were included in the pooled estimation, the range in consumption being from less than 2.to more than 50 kilograms per head per year, in price from less than 6 cents per kilo to more than $1.00 per kilo and in income from more than $6,000 per head to less than 17 $70 per head. The mean values of the sample were 27.8 kilograms per head per year, 37.9 cents per kilo and $373 per head of income per 16For a full derivation and discussion , see Kmenta, J. (1971). Ibid., pp. 508-517. 17See the 1972 values on the right-hand side of Table 7.3. 200 year. Data were for the years 1950 to 1972 and, in some cases, to 1974. The 73 countries were selected on the basis of available data, particular- ly the availability of an exchange rate. . Equations (7.11), (7.12) and (7.13) present the results of the GLS estimation for the Ramsey, double logarithmic and semi-logarithmic forms respectively. Note that the coefficients of the dummy variables are not included with the equations, but are listed in Table 7.3. N-l (7.11) log 01t = z w. + 11.9322 - 0.1299 Y. 11 -1 - 0. 7348 P. 11+ e1t j=l (19. 819) (21 228)‘ 82 = 0.999 NT = 1196 N-l (7 12) log 01.t = z w. + 10 6456 + 0. 2812 log v.11 - 0. 3083 log P. 11 i=1 3 (24. 510) (22.106) + eit 82 = 0.998 = 1196 N-l (7.13) Q1t = z w. + 8. 6892 + 6 4861 log v.11 - 5. 4827 log p.11+ e1t j= 1 3 (23.830) (19.573) 82 = 0.986 NT = 1196 The double-logarithmic Equation (7.12) is the simplest to inter- pret, giving constant price elasticity of -0.308 and income elasticity of 0.281. The corresponding income and price elasticities for each of the 73 countries for the Ramsey and semi-logarithmic equations are listed in Table 7.3. Beginning with the Ramsey results, income elasticity ranged from a low of 0.020 for the U.S.A. to a high of 1.428 for Burma. 2(11 8.88 888 8.8 888.8- 888.8 88888.88 8888.88- 88888. 88 8888.8- 888.8- 888.8 88888.88 8888.8- 8888888 88888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8- 88888. 88 8888.8 888.8- 888 8 88888.88 8888.8 88888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.8- 88888. 88 8888.8 888.8- 888.8 88888.88 8888.8 88888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8- 88888. 88 8888.8 888.8- 888.8 88888.88 8888.8 8888888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8- 88888. 88 8888.8 888.8- 888.8 88888.88 8888.8 88888 88888 8.88 8888 8.88 888.8- 888 8 88888.88 8888.8 88888. 88 8888.8 888.8- 888.8 88888.88 8888.8 88888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88- 88888. 88 8888.8 888.8- 888.8 88888.88 8888.8 888888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88 88888. 88 8888.8 888.8- 888.8 88888.88 8888.8 8888888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8- 888888. 88 8888.8 888.8- 888.8 88888.88 8888.8 8888 8.88 888 8.88 888.8- 888 8 88888.88 8888.88 88888 88 8888.8 888.8- 888.8 88888.88 8888.8 8888 8.88 888 8.8 888.8- 888.8 88888.88 8888.88- 88888. 88 8888.8- 888.8- 888.8 88888.88 8888.8- 888888888 8.88 888 8.8 888.8- 888.8 88888.88 8888.8- 88888. 88 8888.8- 888.8- 888.8 88888.88 8888.8- 88888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88- 2 8888. 88 8888.8 888.8- 888.8 88888.88 8888.8 8888 8888 8.88 888 8.88 888.8- 888.8 88888.88 8888.88 88888. 88 8888.8 888.8- 888 8 88888.88 8888.8 888888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8- 88888 88 8888.8 888.8- 888.8 88888.88 8888.8 888888888 8.88 888 8.8 888.8- 888.8 88888.88 8888.8- 88888. 88 8888.8- 888.8- 888.8 88888.88 8888.8- 88888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.8 88888. 88 8888.8 888.8- 888.8 88888.88 8888.8 8888888 .3 8.88 88 8.88 888.8- 888.8 88888.88 8888.8- 88888. 88 8 88. 8- 888.8- 888.8 88888.88 8888.8 88888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88 88888. 88 8888. 8 888.8- 888.8 88888.88 8888.8 888888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88 88888. 88 8888.8 888.8- 888.8 88888.88 8888.8 8888888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8 8888. 88 8888.8 888.8- 888.8 88888.88 8888.8 8888888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8 8888. 88 8888.8 888.8- 888.8 88888.88 8888.8 88888888 .888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88 88888. 88 8888.8 888.8- 888.8 88888.88 8888.8 8888888 8.88 888 8.88 888.8- 888.8 88888.88 8888.88- 88889 88 8888.8 888.8- 888.8 88888.88 8888.8 8888 88888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8- 88888. 88 8888.8 888.8- 888.8 88888.88 8888.8 88888888 8.88 888 8.88 888 8- 888 8 88888.88 8888.8 88888. 88 8888.8 888.8- 888.8 88888.88 8888.8 888888-88888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88- 88888. 88 8888.8 888.8- 888.8 88888.88 8888.8 88888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.8 88888. 88 8888.8 888.8- 888 8 88888.88 8888.8 888888 8.88 888 8.8 888.8- 888.8 88888.88 8888.8- 88888. 88 8888.8 888 8- 888.8 88888.88 8888.8 88888888 8.88 88 8.8 888.8- 888.8 88888.88 8888.8- 88888. 88 8888.8- 888.8- 888.8 88888.88 8888.8- 88888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8- 88888. 88 8888.8 888.8- 888.8 88888.88 8888.. 888888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8 88888. 88 8888.8 888.8- 888 8 88888.88 8888.8 88>8888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88 8888. 88 8888.8 888.8- 888.8 88888.88 8888.8 8888888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.8 M8888. 88 8888.8 888.8- 888.8 88888.88 8888.8 8888888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88 88888. 88 8888.8 888.8- 888.8 88888.88 8888.8 888888888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88 88888 88 8888.8 888.8- 888.8 88888.88 8888.8 888888888 111-11? 8888888 8888 888888 :88 - ----- - 8.888888888878888 -------- - 88.. 82:8 -- .......... 8888.88 ------------- 8888W8 8 888888 88888888888888 888888 . 8888 8888 888888 888888 8888 8888888 8888 .88 .888 8 - .8.8 -8888 .8.8 -888> .8.8 -8888 .3\.8 _ .8- :i 3w a: 88: 88888828 88888:: a: 88: 885885 888888888 888888 .8.8 88888 2(32 8.88 888 8.88 888.8- 888.8 88888.88 8888.88- 88888.88 8888.8- 888.8- 888.8 88888.88 8888.8- 888888 8.88 888 8.88 888 8- 888 8 88888.88 8888.88- 88888.88 8888.8- 888.8- 888.8 88888.88 8888.8- 8888888 8888888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88- 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 888888888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8- 88888.88 8888.8 888.8- 888.8 88888.88 8888.. 8888888 8.88 88 8.8 888.8- 888.8 88888.88 8888.8 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 88888 88888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 .8.8.8 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 .8.8 8.88 888 8.88 888.8- 888.8 88888.88 8888.8 88888.88 8888.8 888 8- 888.8 88888.88 8888.8 888888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8- 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 8888888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.8- 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 888888 8 88888888 8.88 888 8.8 888.8- 888.8 88888.88 8888.8- 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 8888 8.88 888 8.88 888.8- 888 8 88888.88 8888.8- 88888.88 8888.8- 888.8- 888.8 88888.88 8888.8 88888888 8.88 888. 8.8 888.8- 888.8 88888.88 8888.8 88888.88 8888.8- 888.8- 888.8 88888.88 8888.8 88888888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8- 88888.88 8888.8 888.8- 888 8 88888.88 8888.8 88888 8.88 8888 8.88 88_.8- 888.8 88888.88 8888.8 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 88888888888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.8 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 888888 8.88 888 8.88 888.8- 888.8 88888.88 8888.. 88888.88 8888.8 888 8- 888.8 88888.88 8888.8 88888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 88888 888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.8- 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 88888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.8 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 888888 88888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88¢.88888.88 8888.8- 888.8- 888.8 88888.88 8888.8- 888888888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8- 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 88888 888888 8.88 888 8.88 888 8- 888.8 88888.88 8888.88- 88888.88 8888.8- 888.8- 888 8 88888.88 8888.8- . 8888888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.8- 88888.88 8888.8 888.8- 888.8 88888.88 888... 88888888 8.88 888 8.88 888 8- 888.8 88888.88 8888.8- 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 88888888888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 8888 8.88 888 8.88 888.8- 888.8 88888.88 8888.8 88888.88 8888.8- 888.8- 888.8 88888.88 8888.8- 88888888888888888888 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88 88888.88 8888.8 888 8- 888.8 88888.88 8888.8 883882 8.88 888 8.8 888.8- 888.8 88888.88 8888.8- 88888.88 8888.8- 888.8- 888.8 88888.88 8888.8- 8888882 8.88 888 8.8 888.8- 888.8 - 8888.8 - 8888.8 888.8- 888.8 - 8888.8 88882 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 8888888 :82 8.88 8888 8.88 888.8- 888.8 88888.88 8888.88 88888.88 8888.8 888.8- 888.8 88888888 8888.8 8888888888: 8.88 888 8.88 888.8- 888.8 88888.88 8888.8 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 888888: 8.88 888 8.88 888.8- 888.8 88888.88 8888.8 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 888882 8.8 888 8.88 888.8- 888.8 88888.88 8888.8- 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 88888888: 8.88 888 8.8 888.8- 888.8 88888.88 8888.8 88888.88 8888.8 888.8- 888.8 88888.88 8888.8 888888888: 8.88 8888 8.88 888.8- 888.8 88888.88 8888.8 88888.88 8888.8 888.8- 888 8 88888.88 8888.8 88888 -..8888—oo 8888 m:_8> 38m- --------- 88588888888-8588 ------- --- 888 m_naoo - -------- ..... - 888588 ------------- 8888Mv 88mo88 888888 888888 888888 8888 8888 888888 888888 8888 8888888 8888 .88 .888 . . .8.8 -8888 . . -8888 . . -888 .33 m .9: 3 a: 88: 88:.an u m 88888828. ac 88: m m 888:5” 888888888 .8.8 88888 203 In general these income elasticities accord well with the time- series estimates previously noted. The price elasticities for the Ramsey equation ranged from -0.067 for Mauritius to -0.529 in the Sudan. These elasticities show a much narrower range than those from time series, a point which will be examined shortly. For the semi-logarithmic Equation (7.13), income elasticity ranged from 0.094 for Ireland to 2.703 for Nigeria and price elasticities ranged from -0.079 for Ireland to -l.566 for Burma. The income elasticities from this equation are large at high incomes (e.g., the U.S.A.) by comparison with the time-series results, while the price elasticities are relatively consistent with the results fron time series. A comparison between the pooled and time-series results may be made by observing the results for those countries for which both price and income coefficients were significant in time series. This is done in Table 7.4. In general, the table shows a "reasonable" correspondence between the pooled and time-series estimates. In 7 of the ll cases. the Ramsey income elasticity (ny), was larger in the pool than in time series, while in only one of the ll was the semi-logarithmic value larger, implying that the Ramsey pooled estimate of income elasticity was a closer approximation to time series. Turning to price elasticity (tip), in 7 of the ll cases the pooled estimate was larger with the Ramsey equation and in 5 of the ll with the semi-logarithmic equation. However. the semi-logarithmic pooled price elasticity may be the more accurate because the Ramsey pooled price elasticity is independent of the level of consumption and depends only on price. This leads to a priori unacceptable estimates. For example, since the price in Thailand is lower than that in the U.S.A., the price elasticity in the U.S.A. is 204 ~m~.o- mom.o Nom.o- Npm.o ~¢N.o- m¢~.o mm¢.o- e~_.o xuxczh me.o- mom.o mmm.o- vmn.o ~m_.OJ Npm.o mmo.o- ncm.o uanPmcp mom.o- oc~.o mmm.o- _me.o o~—.o- mom.o mmm.o- mao.o mwgam mmp.o- mpm.o FmF.o- Nem.o F¢~.o- _wo.o mom.o- mmm.o :wmam o~_.o- m¢~.o oxm.o- mm~.o ~o~.o- nmp.o —mm.o- Fwo.o mowgw< gasom _mm.o- mmm.o omm.o- ~m~.o mom.o- mom.o mwm.o- omm.o Agusomv wagox omp.o- MFN.o oom.o- emm.o mme.o- mmo.o cam.o- epo.o . :mama mv¢.o- num.o ”mm.o- omm.o mm~.o- ow~.o _mm.o- NNN.o “mmou xuo>H omm.o- mmm.o mNN.o- _Pe.o ~m~.o- mmm.o mm~.o- omp.o cmzemh-m=_gu _m_.o- mmp.o ¢~N.o- eme.o mm—.o- o-.o mem.o- mo~.o Fw~mgm mpm.o- nm~.o mmp.ou ~o~.~ ~P~.o- mme.o com.o- ¢¢¢.N mw>wpom m: mu a: x: a: a: a: xr um_ooa mmmcmm ms_h umpooa mmwgmm mew» xgucaou awasuwcmmob-wsmm xmmsom mmumswumm uwpooa ucm mmwgmm me_p mo comwgmasou .¢.N opnap 205 estimated to be higher than that for Thailand. In summary, the income elasticity from the Ramsey equation and the price elasticity from the semi-logarithmic equation best approximate the time-series estimates. In the pooled estimation each country had its own intercept through the use of dummy variables, as listed in Table 7.3. The coefficients of the dummy variables may be considered as measures of all excluded influences, once price and income effects are removed. These influences might be called "taste" for sugar and countries with similar "tastes" should have similar magnitudes for the coefficients of their dummy variables. Table 7.5 lists this "taste" variable from the Ramsey equation in rank order, the values ranging from +5.3844 for Belgium to -3.8l00 for South Korea. Should one wish to test for significance of difference in taste between countries, the standard errors in Table 7.3 may be used. In general a difference in taste of one unit is significant since the standard errors are small. An approximate grouping of countries for "taste for sugar" beginning at the highest level would.be: (i) w. Europe, Canada, U.S.A., Australia, New Zealand; (ii) South and Central America; (iii) N. Africa and Middle East; (iv) Sub-Saharan Africa (excl. South Africa); (v) South and Southeast Asia (excl. Taiwan and Hong Kong). The "taste" variable may be interpreted as indicating, for example, that at the same price and income level consumption of sugar per head in Western Europe would be considerably higher than in Southeast Asia. Since the variable is logarithmic, a simpler interpretation (in terms of kilograms per head for example), is not readily forthcoming. As well as estimating the pooled relationships including individual-country dummies, equations were estimated with these variables Table 7.5. Rank Order of Taste for Sugar 2()6 _A 1 Country Rank Insirfigpt Country Rank 1 In5§rfigpt Belgium l 5.3844 Israel 38 1.4275 R. Germany 2 5.0619 Spain 39 1.4160 Guyana 3 5.0419 T090 40 1.4077 Norway 4 4.8680 Iraq 41 1.3209 U.K. 5 4.6277 Jamaica 42 1.2876 France 6 4.5041 Hong Kong 43 1.2694 Sweden 7 4.2949 Colombia 44 1.1766 Finland 8 4.0794 Portugal 45 1.1236 Australia 9 3.9325 Sudan 46 1.1055 Denmark 10 3.8880 Madagascar 47 0.6895 Bolivia 11 3.7281 Mauritius 48 0.6486 Ireland 12 3.4381 Kenya 49 0.5786 Dom. Rep. 13 3.4323 Syria 50 0.5440 Libya 14 3.3989 Tunisia 51 0.5257 U.S.A. 15 3.3648 Cameroun 52 0.4240 Argentina 16 3.3317 Costa Rica 53 0.4045 Austria 17 3.2577 Ivory Coast 54 0.3503 ' Iran 18 3.2109 Trinidad & Tobago 55 0.3260 New Zealand 19 3.1768 Sierra Leone 56 0.3168 Netherlands 20 2.8039 Guatemala 57 0.2825 Switzerland 21 2.6718 Tanzania 58 0.2762 Italy 22 2.6430 Japan 59 0.2600 Ecuador 23 2.6229 Gabon 60 0.1592 Peru 24 2.3691 Philippines 61 0.0410 Turkey 25 2.3461 Niger 62 0.0000 Mexico 26 2.1522 Senegal 63 -0 0229 Canada 27 2.0027 Nigeria 64 ~0.3887 Chile 28 1.9486 Ghana 65 —0.4350 Upper Volta 29 1.7887 Singapore 66 '0-5199 Thailand 30 1.7831 India 67 -0 6201 Brazil 31 1.7533 Zambia 68 -0 9559 South Africa 32 1.7051 Pakistan 69 '1-0124 Uruguay 33 1.6267 Indonesia 70 -2.5712 Venezuela 34 1.5952 Burma 71 -2 6038 Morocco 35 1.5421 S. Vietnam 72 ~2.9022 China-Taiwan 36 1.5313 S. Korea 73 -3 8100 Sri Lanka 37 1.4934 207 excluded. The equations were corrected for autocorrelation but not for heteroscedasticity, since the latter correction led to a "blowing-up" of the estimates. Equations (7.14), (7.15) and (7.16) list the Ramsey, double-logarithmic and semi-logarithmic estimates respectively. (7.14) log 0it = 1.7254 -(041250) Yit'] -(?}1;g§)pit + eit ‘EZ = 0.258 NT = 1196 (7.15) log oit = 1.0169 +(gé7iig)log Yit -(;é8gzl)log Pit + eit Re = 0.726 NT = 1196 (7.16) Q.t = 8.5478 + 12.8248 log Y. - 20.8910 log P. ‘ (51.697) ‘t (41.450) ‘t fi2 = 0.773 NT = 1196 The equations give both much higher income and much higher price elasticities than the earlier equations which had dummy variables. The derived elasticities bear little resemblance to the individual time- series estimates or to the purely cross-sectional estimates of Viton and 18 Pignalosa. The conclusion is that "taste" differences may not be omitted from consideration without severely biasing the results. Summary and Comments In this chapter the elasticity of demand for sugar was estimated for a large number of countries, both from time series and from pooled time-series/cross-section data. Alternative functional forms were 181616. 208 discussed and, for individual time series, the ”Ramsey" form was favored over semi- and double-logarithmic alternatives. In pooled estimation, no single form was superior, but the inclusion of a dummy variable to cover "taste" variation was found to be important. An unique taste scale was devised which demonstrated that. at the same price and income, Western countries consume more sugar per head than South American, African and Asian countries (in that general order). Income elasticity was found to range from 0.02 for the U S.A. to more than 1.40 for the poorest countries. Price elasticity ranged from approximately -0.04 for the U.S.A. to -1.50 in the poorest countries. Except in the poorest countries, sugar may be said to be both price and income inelastic. By way of comment, three areas for further investigation are apparent. Firstly,the cross elasticity between sucrose and other sweeteners was not estimated and a separate study of this substitution for the U.S.A. would be valuable, updating the work of.Hayenga.19 Secondly, further work on functional forms for demand equations would be of value since no single form proved completely satisfactory for pooled estimation. Estimating both linear and exponential price components of the Ramsey equation20 could be fruitful in this respect and a full set of specification error tests might assist in choosing the most appropriate functional form. Finally, the identification of a taste variable presents the interesting possibility of defining more clearly what comprises taste and how it varies in both space and time. 19Hayenga. (1967). 9p, git. 20Rather than just the exponential. CHAPTER VIII RESULTS FROM THE MODEL UNDER ALTERNATIVE POLICIES Introduction This chapter presents the results from simulating the world sugar economy under a variety of policies. The model was solved in two distinct modes which may be termed "recursive" and "equilibrium." The objectives of the recursive solutions were to validate the model, by demonstrating its ability to track historic market behavior, particularly the price-fluctuations of 1973-75, and to obtain some indication about price variability under alternative policies. The objective of the equilibrium solutions was to appraise the long-run distribution of losses and gains to be expected under an array of alternative policies. Further discussion of the policy and welfare implications of the findings will be addressed in Chapter IX. Before presenting the "recursive" results, a recapitulation of the relation- ships used in the model and a few further assumptions necessary to the solution of the model will be given--these occupy the remainder of this introductory section. Considering first, demand, semi-logarithmic functions were used throughout for the 75 regions. Whenever individual equations from the time-series of Chapter VII were deemed reliable, these were used. Otherwise a general equation from the pooled time-series/ cross-section analysis was used. Each constant term was normalized so that agtggl_consumption in 1972 was accurately predicted, or, where possible, equations in domestic currencies were converted to 209 210 a 1974 dollar basis using estimated consumption in 1972 as the base.1 Because the demand equations related to retail prices for refined sugar and the supply equations to wholesale prices of raw sugar, it was necessary to add a retail/wholesale margin and a refining margin in the model. A refining margin of 3 cents per 1b. was assumed for all countries. Retail/wholesale margins were observed for 1972 and are listed in Table 8.1. Also in Table 8.1 the assumptions on growth of population and income may be found. Turning to supply,2 of the 67 equations which were used 28 were asymmetric (cane producers), 14 were log-linear (beet producers), 18 were simple functions of time, 4 were assumed totally price inelastic and 3 were derived from interpolation with a table of values previously calculated (U.S. cane areas). Elaborating slightly on treatment of U.S. supply, the four beet regions of Chapter III were combined into two for which the ports of San Francisco and New York were utilized. The equations were, with terms other than those shown at 1974 levels: New York (Areas 1 8 II): log OS = 5.1892 + 0.9056 log PS - 0.7934 log PINt + 0.0481 T t-l .San Francisco (Areas III & IV): log 05 = 5.5120 + 2.7135 log PSt-l - 1.5719 log PINt + 0.0397 T where: 08 = thousand metric tons of raw sugar, 1For the Communist countries the following incomes per capita were used for purposes of normalization: U.S.S.R. and Eastern Europe $1000 per head; East Germany $1500 per head; China $300 per head. For Cuba no assumption was necessary since income elasticity of demand was assumed zero. 2For a list of countries and functional forms see Table 2.1 of Chapter II. 21 l . . . . . .. . - .000.002 000.000220. :02 .000. .00.00.0000 00000000 .000.002 20 2000200. .0.0.0 000.002 000.00 a . < m 2. u o coucc.:mcz .muco newcp new magma zuzogo ”uuauOLq Poco.ucz 00020 ..000.. ucmsao.m>mo .mco.00:2muc. 2o. zucmo< .m.: .000.002 000.00 0020» 302 .xoon2mw> 0.000200500 .z.: ..0.0.. 000.002 000.00 ...u.3 .00020200: ..o.m.. 0:00:00 .xoon2mw> 2mmzm ..0so.20>v 00.000.00020 2000m .0:o.00:2m0c. 00002som 0.0 0.0 ..0 00.200 00000 ... 0.. 0.0. 0002 0002 ... 0.0 0.0 0200000.0 0.. 0.0 0.0 000000 0.0 0.0 0.0 00.2.0 .020000 .2 .0 0.0 0.0. 0.0 0.0020-.0000 0.0 0.0 ..0 0.0000000 0.0 0.0 0.0 00.2.0 .020000 .0 0.0 0.0 0.0 .0000200 0.. 0.0 0.0 000020 ..0 0.0 0.0 00.2.0 0000 0.0 0.0 0.0. 000.00 ..0 0.0 0.. 0000200 0002 0.0 0.0 .0.0 00.2.0 .0.2 0.0 0.0 0.. 000.00...00 0.0 0.0 0.0. 0000200 0000 ..0 0.0 0.0 00.200 0002 0.0 0.0 0.. 0200 0.0 ..0 0.0 000020 ..0 0.0 ..0 00.220 00202 0.0 0.0 0.0 .0020 0 .00200 0.0 0.0 0.0. 000.0.0 0.0 0.0 0.0 0000 200 0.0 0.0 0.0 .000 0 0000.200 0.. - 0.0 .0.0 0.0 0.0 0.0 0000 2002 ..0 0.0 0.0 000202 0.0 0.0. 0.0 .000 .000 0.0 0.0 0.0 00.2000 .020000 0.. 0.0 0.0 0000200.2 0.0 ..0 0.0 0200000 m.o Q6 0.0.. mQOLam cgmumwm m._. _..¢ _..m “0:3.me :52 Q6 o.c DA: m..xm>o—mo;um~u 0.0 0.0 0.0 0.0000002 0.. ..0 0.0 0000.200002 0.. 0.0 0.0 0000 0.0 0.0 0.0 00.0 .0-.<.0.0 0.0 0.0 0.. 00.x0: 0.0 0.0 0.0 0.000.00 0.0 0.0 0.0 .0.0.0 0.. 0.0 0.0 00.0.2002 ..0 0... 0.0 000.0.-00.00 0.0 0.0 ..0 .2.0 0.0 .... ... 00202 ... 0.0 0.0. 00.00 0.. ... 0.0. .0.0.0.0 0.. 0.0 ..0. 00000 0.. 0.0 0.0 000000 0.0 ... 0.0 00020. 0.. ..0 0.0 00.0000 0.0 0.0. 0.0 ..0020 0.. 0.0 0.. 00000. 0 000.0.2. 0.0 0.0 0.0 0.00. 0.0 0.0 ... 0..00 0 0.0..00 0.0 0.0 0.. 000..00. 0.0 0.0 ..0 000.02. 0.0 ..0 0.0. 00.0.00 0.. 0.0 0... 000.2000.00 0.0 0.0. 0.. 002. 0.0 - 0.0 00000200 0.0 0.0 0.0. 000000 0.0 ..0 0.0 0.000000. 0.0 0.0 0.0 0.20000 0.. 0.0 0.. 02000 .20 ..0 0.0 0.0 0.00. 0.. 0.0 0.0 0..020000 ... 0.0 0.0 , 0.000 0.. 0.0 ..0 000.00. 0.. 0.0 0.0 00.000020 .0.00.0000 000 .0 .02\0 0.00 00.00.0000 000 .0 .02\0 0000 00.00.0000 000 .0 .02\0 0.0. 0o :03o2u 202020 c. 0.020: 00 £0302o 503020 c. c.o20z mo cuzo2o zuzo2o :. 0.020: .0000< 0 .0000< 0 ..003\..0000 0200000 .00000 0 .0000< 0 ..00z\..0000 0200000 .0000< 0 .00000 0 ..002\..0000 0200000 cpzo2o co.00.3a00Nquuc. can 0:.m20z mo.20 ..w m_noh PS PIN T t year (1966 = 66), a year subscript. 212 an index of input prices (1972 = 100), wholesale price of raw sugar in 1974 cents per lb., The mainland cane regions, Louisiana, Florida and Texas, were combined into a single region located at the port of New Orleans. computational burden in making projections as in Chapter IV would Because the have been large, the 1985 projections listed there were recalculated for 1974 to give the fo1lowing estimates: Table 8.2. U.S. Domestic Cane Supply in Thousand Metric Tons Raw Value at 1974 Prices Price Region Total ¢/1b. 1 Texas Florida Louisiana Hawaii Puerto Rico 3 - 11 178 98 - 287 4 - 18 249 226 - 493 5 - 25 316 362 - 703 6 - 36 381 491 -’ 908 7 - 51 439 608 - 1,098 8 18 84 494 715 132 1.443 9 36 251 543 812 164 1.806 10 54 496 589 898 169 2,206 11 64 671 630 972 162 2.499 12 73 818 674 1,045 205 2.815 13 82 970 710 1,114 333 3,109 14 91 1,091 746 1,175 276 3,379 15 91 1,244 780 1,231 414 3,760 16 91 1,349 816 1,281 652 4,189 17 91 1,449 847 1,328 820 4,535 18 91 1.562 876 1,369 955 4,853 1Assumed and not estimated. 213 Continuing the list of assumptions and approximations, the model was initialized with all input prices and the prices of competing crops at l974 levels. Quota agreements by the U.S.A., U.K. and Cuba were included at their 1973 levels, unless otherwise stated.3 The asymmetric supply functions were initialized for 1972 on the basis of 1970 and l97l export prices. Finally, the model was solved to an accuracy of 0.2 cents per lb., or approximately a 2 percent level of error, the endogenously determined variables being price, production and consump- tion in each region and trade flows between each pair of regions. Annual Recursive Solution Because the computational burden proved greater than expected, simulation of the market was restricted to the period l972-75. The results are therefore of an interim rather than final nature; a fuller set of results awaits the development of a more efficient algorithm. A summary of the results for the major regions and a comparison with actual quantities and prices is given in Table 8.3. In the table two kinds of solution are shown, one called "basic" and the other "taxed." Discussion begins with the "basic" solution, which assumes that export- ing countries allow their domestic prices to rise to the international equilibrium level when supply declines relative to demand. Simulated world supply, (on the right hand of the table), in l972 is 75,308 thousand tons and rises in l973 to 77,859 thousand tons. Actual supplies in these two years were similar to these estimates. In l974, however, simulated supply falls drastically to 3An exception was the elimination of Australia's quota to the U.K., which was not included as it eXpired in l974. 214 .mumswumm uzuwb .o .d o“ _mp_acaa N . :mmnnwgmu :_ woven acqm upco3_ woom._m mkm_ Nmmo.NN mmm.~ omm.m _mo.mm Nmm.¢ _mm.o .m.: omm.~_ mmm.m .m.: moo._P mmm.m om.mm mmm.op mom.m c~m~ _o_.wk mmm.m mmm.m Fpo.m oom.q “mm.m .m.c oom._F ooo.m .m.: o_~.__ Nn_.o_ mm.o_ mmo.o_ mmm.m mxm. _Nm.mm mom.m mwo.q _mv.N mm_.¢ _m_.o .m.c om~.o_ v~o.m .m.: mN¢.o_ omm.m mo.m m_o.o— QNN.m Nxmp 4 um mmuocmv m “mmuoz *mmONN nmmm omop womm mm_m om.mm mo..m . + - + + m Fommm comm mmm_ Pmmc «mam mm.w ¢~.~F _o.PF ow.m . + + + + m “OF _~> ommuu mama mwm_ m_oc Nome em.m mm.m_ mm.m m~.m - + + n + . ~> moms“ _mVN “mmm w¢_m w_mm om.o_ _m.o~ m_.__ ¢~.__ - + + + - a > “moss oeFN wpom _mmm Nmme mp.w Km.m mm.__ -.m u + + + u + >~ m_m- momm mwm_ acme ammo m..m om.~_ mm.o_ m~.o— u + + + + - n am— msmnu ~__w womm o_~m «mom Fm.m mm.PP N~.o_ mm.o_ u n - u u u Hm mmmmfi momm mmm_ swam Name -.o m~.N_ mm.__ o~.m u + + + + + H :owuq53mcou co_uu:uoga mucanH cowuuauoca mugoaem u_umweoo Amuccgmv xco> zmz xco> 3oz xmh mwemgm» mouozc mmuoao A>m4 uu< Lomsm consaz. use ovummeoa u_ummeoo canau gum u_umm50a “exec: ugoaxm Locuo :u_mm3 cmnzu gum .m.= ucwswcoaxm 55269:. .m .2 new: -coEou urge: .u.w.u .<.m.: mwu_ca »u__om 7 uwcaw__:mm camumcog use mucoewgmnxm »w_~om ¢.m mpnmh 219 different policies the volume of world production is relatively con— stant. This results not only from the low magnitudes of changes in price which are induced by the alternative policies, but also from the ease with which beet production may be substituted for cane pro- duction. Even under a huge export tax of 20 cents per pound (Xd), the volume of world production is not greatly curtailed but its geo- graphical distribution merely changed. Secondly, it is worth noting that priCes are considerably lower than those existing in 1974, even under the imposition of a large export tax by cane exporters. The results suggest that a return to lower average prices is very likely for future years. Turning to specific questions of the effects of alternative policies, the first comparison to be made is between the most likely set of policies (I) and completely free trade (II). Surprisingly, world production would decline by 960,000 tons under free trade. The underlying cause is the increased free-market price, from 7.76 to 10.85 cents per pound, and the associated increase in sugar prices in exporting countries. Since the price elasticity of demand is higher in the exporting nations as a group than in the importing nations, high prices reduce consumption by more in the exporting countries than they increase consumption in the importing countries. The net effect in equilibrium between production and consumption is a small decline in world production (and consumption). -The effect of free trade on U.S. and EEC prices and production is less than might have been expected. In both regions some domestic production is replaced by imports and the domestic price falls to meet the free-market price (which has risen). Imports to the U.S.A. increase by 792,000 tons or 16 percent 220 and to the EEC by 1,505,000 tons or 103 percent. For a full listing of national prices, production and consumption under these and the six most interesting other policies, the reader is directed to Appendix C. The second comparison to be made is between the most likely set of policies (I) and a set in which the U.S. Sugar Act is ended (III). World production and consumption decline slightly by 717,000 tons. U.S. and free-market prices become synonymous, but the free-market price rises much more (+2.47 cents) than the U.S. domestic price falls (-l.56 cents). The hypothesis of Sanchez5 that the U.S. Sugar Act raised free-market prices is rejected by this experiment--the converse is true. Because the U.S. domestic price falls more than it would under free trade, imports rise correspondingly more. As compared with the benchmark I, imports rise 1,498,000 tons or 31 percent and domestic production declines 1,443,000 tons or 24 percent. The third comparison is between the benchmark (I) and the uni:_ lateral end of its protective levy by the EEC (IV). dust as with the abolition of U.S. protection in III, the free-market price is raised (by 1.41 cents), but in this case the EEC price (as measured in France) falls (by 2.92 cents) even more, indicating that the U.S.A. influences world price more than does the EEC. The consequent decline in EEC production is quite large, being 2,163,000 tons or 23 percent, while imports expand correspondingly by 2,535,000 tons or 183 percent. World production remains remarkably constant under this as under each of the other policies. 5Sanchez, N. (1972). "The Economics of Sugar Quotas," Unpub- lished Ph.D. dissertation, University of Southern California. 221 The fourth comparison is between the benchmark (I) and the simultaneous abolition of trade-barriers by the U.S.A. and EEC (V). The major effect is to raise the free-market price even more than under free trade. U.S. prices are slightly higher than under free trade, but EEC prices fall and there is a corresponding decline in EEC pro- duction.6 The implication of experiment V, as compared with III and IV, is that orchestrated reduction of trade-barriers by the U.S.A. and EEC would lead to smaller problems of domestic adjustment than the unilateral reduction of trade barriers by either region alone. The fifth comparison is between the unilateral ending of the U.S. Sugar Act (III) and the simultaneous ending of the U.S. Sugar Act and Cuba's Quota Agreements (VI). Cuban sugar may now enter the U.S.A. in larger amounts and the free-market price, (as measured at New York) and the U.S. domestic price are both lower under VI than under III. U.S. domestic production suffers its severest decline, by 1,974,000 tons (33 percent) as compared with the benchmark (I). U.S. imports rise similarly by 2,020,000 tons (41 percent) as compared with I. The sixth comparison is between III, the policy set with no U.S. Sugar Act, and VII, a policy set in which the U.S. imposes a 10 percent ad valorem tariff on sugar. The effect is very slight. There is a small decline in free-market price, a small rise in U.S. domestic price and a correspondingly small replacement of imports by domestic production in the U.S.A. 6While New York U.S. prices rise from II to V, U.S. production does not rise due to slightly lower prices in the other U.S. regions. 222 The seventh comparison is between free trade (II) and the 292:. tinuation alone of the Commonwealth quotas and other countries' tariffs (VIII). All prices fall in this set relative to free-trade and consequently U.S. and EEC domestic production also fall, but the magnitude is small. The eighth comparison is between the benchmark (I) and the endingiof the Commonwealth Sugar Agreement (IX). Free-market price declines slightly, but the EEC price rises somewhat and the latter region becomes almost self-sufficient in sugar, importing a mere 366,000 tons. The importance attached by the U.K. to continuing the Commonwealth Sugar Agreement in the interest of Commonwealth exporters is seemingly justified by this experiment. The final comparisons are between the most likely policy set (I) and sets with similar policies except for the addition of an export tax of varying mggnjtude by the cane exporting countries (X a, b, c, d). Export taxes of 2, 6, 10 and 20 cents per pound were considered. At taxes of 10 and 20 cents the U.S. Sugar Act and Common- wealth Sugar Agreements were no longer functional (c and d). Taxes of 2 or 6 cents per pound would merely be impositions on importers from the free-market such as Canada and Japan, somewhat similar therefore Unthe weak International Sugar Agreements of the past. The free-market price would not rise to the level of the U.S. or EEC prices, thus avoiding disruptions in those markets. However, an export tax of 10 cents per pound, if also levied on the U.S.A. and EEC, would raise prices in these two regions and encourage domestic production. A tax of 20 cents per pound would result in the EEC becoming a net exporter and the U.S.A. importing a mere 1,443,000 223 tons (a reduction of 70 percent). _The relatively elastic supply of domestic sugar in the U.S.A. and EEC and the inelastic supply of the cane exporters result together in the easy substitution of domestic fbr imported sugar and only a small reduction in output worldwide when exporters conspire to raise prices. Under a 10 cent tax several traditional exporters of cane cease to export, for example Argentina, Brazil, Colombia, Guatemala, Mexico, Nicaragua, Peru, South Africa, Thailand, Venezuela and Central America. Under a 20 cent tax Australia, Barbados, China-Taiwan, Dominican Republic, Fiji, Guyana, Jamaica, Philippines and Trinidad and Tobago are added to the list of nonexporters. By contrast, the U.S.S.R. and Eastern Europe greatly expand output when the cane producers tax their exports.7 The results in tenms of price, supply and demand are now complete, but it is interesting to convert the protection given by the U.S. Sugar Act and by the EEC's variable levy into the tariff-equivalencies necessary to achieve the same price under the alternative polices. This is done in Table 8.5. Comparison with free—trade gives tariff equivalencies in the 10-20 percent range for the two regions. However, approximately a 20 percent tariff is required to achieve equivalent pro- tection in the U.S.A. should the U.S. Act continue to be defunct and the EEC would need approximately a 30 percent tariff to equal its levy should that cease. No U.S. Sugar Act and no EEC levy together imply somewhat lower tariff-equivalencies-due to the higher free-market price which results as compared with unilateral action. 7Refer to Appendix C. 224 Table 8.5. Percent Tariff-Equivalencies of U.S. and EEC Policies Region Policy Free No U.S. No EEC No U.S. Act Trade Act Levy or EEC Levy EEC Belgium 12.4 --- 31.5 26.5 Denmark 13.4 --- 26.4 19.0 France 12.5 --- 29.5 24.1 W. Germany 14.7 --- 27.0 20.6 Ireland 13.3 —-- 30.2 24.7 Italy 16.0 --- 32.8 23.4 Netherlands 15.0 --- 31.3 24.4 U.K. 11.9 --- 28.5 22.5 U.S.A. Hawaii 6.9 20.0 --- 8.6 Puerto Rico 16.0 25.1 --- 17.2 Mainland Cane 1 19.5 28.4 --- 18.2 Mainland Beet (l;1 10.9 15.2 --- 6.9 Mainland Beet (2 13 7 16.0 --- 7.3 1Area (2) comprises the West and Northwest (beet areas III and IV). Area (1) comprises all other regions (beet areas I and II). Before proceeding to welfare comparisons, it is interesting to note the saving in worldwide transportation costs which occurs under free trade (II) as compared with a fully protected market (I). The costs are 819 and 1,059 million dollars, respectively, the saving from free-trade being therefore 240 million dollars or 23 percent. The actual direction of trade is not of major interest to this study, but trade-flows under policies I (full distortions) and II (free trade) are listed in Appendix D. The largest single change implied by free trade is, not unexpectedly, the redinection of Cuban sugar to fill almost all of the U.S. import requirements. 225 Welfare Implications of LonggRun Equilibria In Table 8.6+comparisons are made, in terms of producers' and consumers' surplus and government revenue, between the benchmark solu- tion I, (full distortions), and the four most important alternatives: Free Trade (II), no U.S. Act (III), no EEC protection (IV), and a 10 cent export tax by cane exporters (X c). tive to the benchmark or most likely solution, I. All calculations are rela- In order to simplify the calculations, the supply and demand functions were assumed linear over the appropriate ranges and the small changes in tariff-revenue which accrue to importers under III, IV and Xc were assumed negligible relative to I. The calculations are summarized in Table 8.7. Table 8.7. Summary of Gains in Thousands of Dollars Policy Free No U.S. No EEC No U.S. 10 Cent Trade Protect. Protect. or EEC Tax Xc II III IV Protect. Region _V U.S.A. + 66,406 + 33,125 0 __26,004 -245,142 EEC + 70,059 0 +184,331 +140,848 - 32,873 LDC's +328,557 + 71,951 +123,022 +147,277 -482,635 DC's + 1,402 - 90,572 + 49,089 - 40,468 -440,934 Cuba +392,285 +313,664 +179,238 +44l,811 +551,795 Exporters of Cane +638,622 +17l,703 +405,622 +432,133 -454,574 Total +329,959 - 20,053 +172,1ll +107,380 -923,569 Free-Market Price] 10.85 1 9.17 11.24 13.77 10.23 1 + See pp. 226-27. Cents per pound f.o.b. New York. 226 .2...- .3. £0. a... 8... :3. R;- _ m8. 0..... o ...... S:- 03. :- ...... .2: ...... :3- ..3 .335; o 3.0.. 33. o o _ 0..:- o o 8:. 8.... o So.- 3... o o 8%. 32.... $3- .3... :3. 3.8. m3:- . .33 93.: o 2.2- BS .391 .3... .m..o- ..... 0...... ...:- 3.3..- .23. 38. $3 $8.- 35. 2...... . 8.... a... 0......- x..- 83 ..m. 3.. 8..- ..E- .3: 3....- 5....- O we...- 0..-.... o . 2......- o . 2m...- 83- o . .03- 33.- .. mm...- :2 $03 .28.. :22. 2:. . 82... ..o.. n 3......- 837 2.3 .8..- ..£o.- ......»- mvm.... 3...... .2... 8. ..3 $8. mm... 3...- . 3...- ..8. Q... m...- .2... ...... ..3 .3: .3..- 83- .3.. 38. 53. ......- . w 8... 3.... o 3.. 3.3. o a... o o o o W o a 3% 8m- m... 3.....- o ...... 9...... mos:- EST :3. . 38.. ...... 88 :2- 387 5%. .....n. 3.2 .3... m3.- 8:... ....T .38. :3.- . ......- ooma. .... ....3 3..- g... .3...- 33. ...... $2...- 0 ...:- .3. o . . o o ...- om..- o o ..5- o .3.- S ..- o o . o o a? $3- .. a $3. 2....- ..m. m... :2. n.a.: ......- 3; .2..- ..o.- 33. 3.: .....- .3.- o... 3.... ......- .8. 0.8.- 0.3- .... .3 23. ...S- .3. .3.- 8m- ...... 9.2- a... $0 3.: .....- ......- 83. .8... $2. 8.. 83 o8... :2. 3m. 33. ....m- mm... 3:.- 8: ...... ...o. w... 88. .58. .2...- o..- 3%. ms... ......- 33. S..- .88. mm... ......- .S.- ...... .2. :8. .9..- .2. 2:..- o.8. 8;. o 28. 8.2- 2.. so. 3... ...... 8.... . .3: .....- a... on... ...m o... ... . 3... $8. 28.. mm..- .32 .3..- z... .20.- 0..;- 3.... a o :8. .3...- o £8. ...... 3....- 2?. .55.- m..v.- ...: a o a... ...:- o mam. €5.35... .0... 5.3. 32- 3.27 3;..- .E: a o ......»- mzz. c 32. ...-E... .... ...... «.3.- ..m: Em...- 33..- SE: 0 o .8. ......- a 28. .8... mm:- S...- 9...... :87 ..3. o o ......- 8...- o ...: ..m..- ...... S... 8.... ...... 3;- ...S no... 3%.- S... 3:- 3...- 2:- ...8 .8..- ..mo- .... 3.; £2 .23.. 1....- Em. o o .o. .3... o 33 S... a... .3. 8.... 8.... :3.- ..m ...m. m3.- ..3 3%.. o .38- 82... 8.8. 2...: :5: £8: :32 ...:- 282. ...... :8..- 9...... ......q a ......- .£:. £2- 3... ...... m3... .2. ...... 3...- a... 08.... £3. 3..- .3. as... 3:- ......- .mo... 2... 3.: £8 2... ...;- cmom. 8.3- o...- as 3;- .....- em... 3:. a..- 2.2. ...... 3......- .o... 83. ..a. :3.- 3? no..- ...... 3.5.. 8...: a .38.- ..m... .....- ...... ... .... 8....- n...- ... 3m.- ..3..- o... :3- ..5- o.$. 3....- 3.... :9. 8.. .....- z...- 82 ......- m.....- 8.. as..- .3...- c 28.- .2... o o .8..- 3.3.. o 3?.- 8m..- 0 .8:- .8? 0...... :3- 38: :2. :3... :3- 3.8. 3%... 23- 8.... :3- 882- $3. 88: :3..- .83.. 3...: .. .- 3.:- mg... ...... 3... ...... 1m... 2... 5... .3- ...:- w..2- 8.... x..- .....- 5:5 . ...:8 .3..- 8. z..- ..m- .80. ......- ..... .3- .mm. as... o... .2. ...-....- o..- 9.; ”.3. . 4... 52.3.5. 3...: as... 3 a... o ...: 3...- ...- mo: ...:- 8. $3. o ......- :52... .5...- 3.3. :8... . S... .3... 23.- .a... as... o..:. 5.. .2.- .. $3.. .3... :3.- ..88- 5:33. .8..- .mom- .3... h 3.... 8.3 .mmm- . .32 ...: 85.. 28- 23.. mom. 3.... ...... z... .3... 28..- 3.258.. ...... . .mq . .... m .... m ...... . .mq . :2 . .w... H ...... H .md .... u a... n ...u. .4. 5.. fingJ ...... — .md . .1 ..,-Til-L. . ... ... . 93.58 a. z. :8; :3 o. . .3258. U: 3 a... e. _ .. ...-.... ...: oz .: 3.. S 9. : no...» out 3.528 2.22:3: .325 9.53: E “...-a .06 03c.— 227 ..mvmq- mmemmv- mommmo- .... ..xn o o...- mwoc. omrem a _ .cw- ~.Mm- .mumm .mq- mom..- .omw.- wmxcm c m.s..- .mcec- .mmo o ...om- oo..m- chcx. 0 «mm...- vomom.- mm... o ..Jc..- memo.- .xoo. o Nome.- omom- mm.m. e mm...- ...mm- .....- Om...- ma..o ...mc- oer... o .rmmvm- oamm- o.m..- co..- x..¢. mm.m ...m o wwmm- mmmmom- ...cm o m.m..m- own m.mom o ammo.- m....- .mm.. o ....o- .mmo. .rcmm c m...- mo..©- .osmv o m..:.- m.m..m- o o m.m..m- m.m.mm mm.mo._ 0 ....mm- momm o.moo o .m..@- m.mm- .mmo.- ..mm .39. mo..- ..mq- ..m.- .mm. vao.- N..m o co...- m....- ocmx. o woo..- .mo..- mmooo 0 mo»...- ..mmm- om.oo.- mmmm.- oommm. o.m..- o o 3....- .o.... o c .c.o.- m.omm- .... o o.mo.. .mm.o N.om.. o o...m.- Nmmom_- nxovmm- m_nwv- mnmwm. mmmmh- cm.om- cccwm- __mmo .....- we... a .....- m.oo.- o o .....- ..z “ .m.. ...o w .m.u ._f..--.-. o ...OM ¢m¢o. nr.cm m.vm .max. 0... m... Mxo. mnmym- ..rx.. mr.:v acme- .wmm_- _qmmm- c_m.m- n.a..- mrc_m- o oocvmv ... .T¢.. n.a.. ..mm m.... C)C‘C7C>C C’C’C5C‘GJC‘C‘: .oM.n. “mm..- 9.0%- _mde_ Um...» .Omm.m ....m _nnn.- .o omNNN mmmm .o 3.mm. mo... .0 .mmm- ..mm- OVTEI mcom- “mm..- ”.nmo. “momm- "arwo- Hq¢x.- min: ..m Mocyv mam- “066:: ...OD- .-- . ..,. C I CD A me.3;, C5C‘C‘CFC5C>C>C‘C>C2C)C)C>C)C‘CfC3N~CBFF A ..z . >v|t > -o-« >~>- , >_ .a «.u ... oz 0 c : maov_ um®_- on..- m 2.0 c w...m- m~_e_ o wm~.m- mmmm o umom_- cxcnv C CMTrx- Moo. 9 3mcm_- .cw. o ..rn.- ~©Om new: Smoc- mmxnm- cmmmm- c.wcn mrmvc c _mmmx- mvvmm ve_m- M\_vx- m_.m- 0 came mmw_vu mouse- 06.2.— xfmvxn C Ommm_— cmoom- o .mc.c_ wmmmn- a mom. 0 o o c c m__c:_ emoOON e mm_.~. m.m o m:- mov—— ove- xm_m- 3.20 «_m_- acce- 9co_ c crvx- «Tim C _~$C_- ammo. a omcnx- cmcv_ xcoN- o~_x- o c mom- 0 c amoe- _xm o v_©o- moxco o mc¢\v- _Nxom m:..:_- ommo—- \mcc- cccmn- wax» _wmo c ...:_- o—wM- .m.. ...u M .m . ... ..d m: .fi ....... . . . .mmn.. . 3m0¢mm _ N . 1Jx< ...m- ” x..om m.xqm- Nfimmmn v\_v:- imTflfi- ~mmwx- “mnem- ~TPQI ...om Immo- m3xem mxcx_ C@~—_ NTTM, p: mmxmm 0¢mmom Nq¢ owe .¢¢N- c..o.- mmva. .....- were. Ne.m- .m:..- m.rm_- mnemm m..oo. _ .9...- . w... n3.m_- hm: o_xvrm- .— -—. mm _wxcn &_¢Pp ..en__- «xv- Mywoc_- vw :xi ”V39 - o.:.c- v.._m an; —IM- mc.c- mmc_m- _v.oc- ...:m- rvmv—u ¢¢._m- c _m.mm~ omrom xx"...— rmm.m mo.m «our. renew x».m. c c ~_m- rx_m\. $n_mqm ray.— m:..m- : on- n «. .wm - h c... _ ©.cc;. H m.:. _ cameo. . ... w fimeN —I _ .o.. _ mo... . m...w-. .....- M m:. " m..mm._ cox.x . .0... _ n.a . o..ox * c.-.o.-w .xxzom- mficmm- o...- ..c..- cho- .....- 20%- .....- _oer .03..- mmzm mom...- o....- or..- mMnoc_ “ 4:. . . No.32 Jinx: Ftuo. cs....< .u.3.m .e....< .u.m .14—Lvd .w «a...¥< ...z ca..L.q .3 c5u...< .z ..muw .a. ..wn. Lazz .AQRVPS m xiiiiea ISUwLis< —GLMCQU $30.3. .u c_z:~0:9> au.x o..o:a 1.5» 302 - .m.: mcwo_.e 3:2 - .m- cUm.Ucu.. cam - .m. ..oIQI - .m. .x. 3 32:23 ..u... .m.m.m.: nuwonv .a.w.m.: xoxga. .ooeae. . .uu_c... .vcc_.oc~ ace—Lw~..rm cccmxm c.1Qm au...< cuzom .vgoauvc.m a_oa.<-.n3em .ea:..0a ch_0g ..~:.aa.._;. csgaa ccwmun_occm a cvum..aa caac..=ou .o.m ¢_co. 228 Considering firstly free trade, II, the net gain would be $329,959,000 of which $328,557,000 would go to less developed countries. The largest single beneficiary would be Cuba, the largest exporter, which would gain $392,285,000. Since Cuba would gain more than the overall gain to less developed countries, it follows that on the average other such countries would lose under this policy. This is not surprising if transfers to LDC's under the U.S. Act and Commonwealth Sugar Agreement are considered. Under these policies the premia given to exporters are estimated to be worth $377,367,000 and $l25,83l,000, respectively, a total of $503,198,000, most of which goes to LDC's. In consequence, Cuba's gain is offset by the losses of many other countries which were protected by the U.S.A. and U.K. Looking at the net situation in the developed countries of the West, the EEC would gain $70,059,000, most of which would accrue to the U.K. ($55,255,000) and Italy ($20,674,000). Producers in the EEC would lose $299,700,000 and consumers would gain $369,759,000. The U.S.A. as a whole would gain $66,406,000 resulting from a gain to consumers of $273,304,000, a loss to producers of $l39,630,000 and a loss of tariff-revenue of $67,268,000. Considering, secondly, unilateral action by the U.S. in ending its Sugar Act (experiment III), there would be an overall world loss of $20,553,000. In this experiment Cuban sugar was allowed access to the U.S.A. and hence Cuba had a large gain of $313,664,000 which was offset by the $377,367,000 which was the premium previously paid by the U.S.A. to quota-holding countries. For example, Australia, Brazil, Dominican Republic and the Philippines lose under this policy due to the end of U.S. quotas. Importers from the free market, such as Canada 229 and Japan, also lose due to the higher world price. Turning to the U.S.A., the net gain from an end of the Sugar Act is estimated to be $33,125,000, resulting from gains to consumers of $330,214,000 and losses to government of $67,268,000 and to producers of $229,821,000. The EEC is unaffected by this particular change in policy. The third welfare comparison concerns experiment IV, the uni- lateral end of‘protection by the EEC. There is an estimated inter- national gain of $172,111,000, divided $123,022,000 to the LDC's and $49,089,000 to the developed countries. As before, the gainer of greatest magnitude is Cuba, gaining an estimated $179,238,000. Because of the increase in world free-market price, exporters of sugar gain and importers lose. However, because there is no longer any Common- wealth premium since all countries receive the same price from the EEC (although the experiment maintained 1,383,000 tons of Commonwealth imports to the U.K.), Commonwealth countries such as Barbados, Guyana, Jamaica, Mauritius and Trinidad and Tobago suffer small losses. The EEC as a whole gains $184,331,000, chiefly due to the consumer gain of $709,424,000, while the producers' loss is $525,093,000. The gains would particularly accrue to the importers in the EEC, that is, to the U.K. ($119,910,000), Italy ($40,568,000) and Ireland ($2,600,000). The U.S.A. is unaffected by this change in EEC policy. The fourth set of welfare measurements was made with respect to experiment V in which both U.S. and EEC protection ceases. The inter— national gain of $107,380,000 is less than under unilateral EEC action because there was previously an international gain from the U.S. Sugar Act. However, gains to LDC's of $147,277,000 exceed those under uni- lateral action by the EEC or U.S.A., mainly because the free-market 230 price is raised more by this bilateral action. Cuba is again the chief beneficiary, to the extent of $441,811,000. So large a gain by Cuba implies that other LDC's lose under this policy. The EEC as a whole gains an estimated $140,848,000 under this policy and the U.S.A a mere $26,004,000. The final welfare measurements of this kind were made for_policy Xc, a 10 cent per 1b. export tax being imposed by all cane-sugar exporters. It has already been noted that many exporters would simply become producers for their domestic markets under this policy and, as a group, they are estimated to lose $454,574,000 under this policy. As under other policies, however, Cuba is a large gainer, this time to the extent of $551,795,000. Other substantial gainers are China ($107,082,000) and the large beet producers of Eastern Europe, namely East Germany ($154,470,000), Czechoslovakia ($152,262) and Poland ($91,532,000). The total world loss would be a huge $923,569,000, this loss resulting both from the cessation of exports by certain countries such as the Philippines, (loss of $156,237,000), and from the higher free-market price to be paid by all importers. Under this policy LDC's as a group would lose $482,635,000 and DC's $440,934,000. The U.S.A. would lose $245,142,000 due to the high cost of 3,443,000 tons of imports, but the EEC would lose only $32,873,000 due to its low dependence on imported sugar, (only 322,000 tons). A Summary of Chapter VIII The results of this chapter are the product of combining many of the relationships which were developed in the previous chapters. The model was first solved in a recursive mode for the years 1972-75, 231 but the results were of an interim rather than final nature due to un- resolved computational problems. However, restriction of exports by cane-producing nations in 1974, in order to satisfy their domestic markets, was shown to be a possible influence on price in that year. 1974 represented the zenith of a price-cycle and steadily falling prices for 1975 and 1976 may be expected as a new cycle develops. Further solutions in the recursive mode await the development of a more efficient algorithm. The solution of the model in a long-run equilibrium mode showed that the developed countries, which are importers or protect their domestic sugar industries, have less to fear from freer trade than might have been expected. The absence of U.S. protection raised free- market price by 2.47 cents and lowered U.S. price by 1.56 cents. The absence of EEC protection, by contrast, raised free-market price by 1.41 cents and lowered EEC domestic price by 2.92 cents. The change in domestic production under these two movements to freer trade, considered separately, would be 24 percent (U.S.) and 23 percent (EEC) less domestic production. Should both the U S.A. and EEC switch together to free trade, the effects would be even less, since free- market price is raised even further, the U.S. decline in production being 14 percent and the EEC decline 20 percent. World production was remarkably constant under all policies, since beet production may easily substitute for cane production. This was particularly the case when the cane exporters were hypothesized to impose a uniform export tax which resulted, (at least at a level of 10 cents or more) in a deterioration in their earnings except in the case of Cuba. 232 From a welfare viewpoint, Cuba would gain under almost any policy-set relative to the set existing in 1973. Under free trade almost every country improves its welfare, except the traditional importers from the free market. Free trade by the U.S. alone would have little global effect except to redistribute gains to Cuba which previously accrued to other quota holders. The end of EEC protection would bring large benefits to domestic consumers and foreign producers. At the aggregate "world level” of abstraction, free trade leads to a $330 million dollar gain, a large but hardly remarkable amount. A 10 cent export tax by cane exporters would lead to a world loss of $924 million, much of which would be borne by the exporters themselves. Strong cartels would therefore seem a very dim possibility. CHAPTER IX POLICY IMPLICATIONS AND SUGGESTIONS FOR FURTHER RESEARCH This chapter attempts to draw together the implications of the results, not only of Chapter VIII but of the whole work, for the poli- cies of the U.S.A., EEC, less-developed countries and cane sugar ex- porters. Value judgments concerning different groups in society are implicit in alternative policies, but such judgments have as far as possible been kept to a minimum. The chapter concludes with suggestions for further research. U.S. Sugar Policy The U.S. Sugar Acts, which ran continuously from 1934 to 1975, spoke of a "fair division of benefits" from protection.' However, the division clearly favored producers at the expense of consumers. The former were protected from low prices through quotas on imports, but the latter were not protected from high prices should there be a coin- cidence of reduced domestic and international production (as there was in 1974). It was a coalition of consumers and industrial users that brought about the downfall of the Act in l974--they had nothing to lose. The equilibrium solutions of this research suggested that without protection, mainland cane production would contract 34 percent, Hawaiian cane production 12 percent, Puerto Rican cane production 8 percent and mainland beet production 23 percent. The major adjustments would be 233 234 in California and the North-Nest with respect to beet and in Florida with respect to cane. Only in Florida might the fixity of assets and the lack of suitable alternative products pose a problem for producers' incomes. The end of U.S. protection was estimated to hurt overseas sup- pliers to this country, who previously held quotas, to the extent of $153 million. However, most of this loss was the result of assuming that Cuban sugar could enter freely. If Cuban sugar remained excluded, the losses of these countries (mainly the Philippines, Dominican Republic, Brazil, Mexico and Australia) would be minimal. The "aid" aspect of the Sugar Program is therefore not very important. From a static-equilibrium viewpoint, such as that of D. Gale Johnson, the case against any form of protection is clear cut. Protec- tion is a priori inefficient. Johnson estimated in 1974 that the Sugar Act cost consumers and taxpayers approximately $616 million to transfer $100 million to domestic producers and $198 million to overseas suppliers. His estimates implied a "deadweight loss" of $318 from protection. The estimates in this report, on a slightly different basis, are of a total cost to consumers of $330 million to transfer $230 million to domestic producers and $67 million to government, a "deadweight loss" of a mere $33 million. This implies that Johnson probably overstated the cost of resource misallocation under the Sugar Act. Whether the Act is in "the public interest" or not therefore hinges not so much upon the misallocation of resources as upon the de- sirability of transferring income from consumers to producers in this manner. The answer lies in the political arena and outside the context of this report. 235 If protection is deemed in the public interest, a deficiency payment is likely to lead to a smaller "deadweight loss" than a tariff (or quota).1 To give protection equivalent to that of the Sugar Act would require a subsidy of at least 1.56 cents per pound. This subsidy would be paid on all domestic production, giving a minimum total cost of $206 million. To offset this there would be the gain to consumers rela- tive to the Sugar Act of at least $330 million. The problem with a deficiency payment would be that it is based upon the willingness of the populace to be taxed an additional $206 million in order to subsidize sugar producers and maintain a low price of sugar. An implicit tax, through a tariff or quota, may be politically more acceptable although economically less efficient. Thus far only static-equilibrium considerations have been re- viewed in relation to protection. In a more dynamic context, there may be losses of producers' and consumers' welfare due to price fluctuations. A Sugar Program which led to a more constant price might appeal to . both producers and consumers. If it is assumed that domestic producers should be protected from low prices and consumers from high prices, an array of alternative policies may be reviewed with respect to these objectives. The policies include tariffs, quotas, variable levies and deficiency payments, either separately or in some combination. Before 1934, tariffs were used by the U.S. to protect producers, but they were rejected in favor of quotas in the Jones-Costigan Act because they did not give sufficient protection from low prices. 1See Chapter II for a discussion. 236 The combination of quotas, tariffs and deficiency payments in successive Sugar Acts was eminently successful in avoiding low prices for producers. Of these policies, however, only the deficiency (conditional) payments may have had any (extremely marginal) influence in reducing consumer prices.' To protect consumers, policies of a different nature are required. Some possibilities are: 1) buffer stocks built up when the price of sugar is low and released under high prices; 2) a direct subsidy to consumers on sugar when the price is high; and 3) the pooling of a regulated domestic and a fluctuating international price so that consumers pay the pooled, average price. Buffer stocks, the first alternative, are not very attractive for sugar because storage is expensive, although they would achieve the desired smoothing of price. For buffer stocks to work, the U.S. market would have to be separated from the free market by a quota system so that the release of sugar could have an impact in the U.S. alone. Since the price elasticity for sugar in the U.S. is of the order of -0.03, a 1 percent change in quantity could induce a change in price of 33 per- cent, hence the stocks would not have to be very large in relation to total consumption. Considering, secondly, a direct subsidy of consumer prices, it is not likely to be acceptable in the U.S.A., although it has been used in other countries for staple products, e.g., bread and milk in the U.K. Thirdly, the pooling of domestic and international prices has been used in the U.S.A. in relation to oil, but is the kind of regulation which is politically not very acceptable (and would further induce the free-market price to rise). 237 The above discussion on price variability and policies has not included free trade by the U.S. as an alternative. Protection under the U.S. Sugar Act was estimated to reduce free-market price by 2.47 cents per pound. Should protection be reinstituted, the proponents will undoubtedly claim that the new protection saved the U.S. from the ensuing, low, free-market prices. They forget that U.S. protection actually contributes to that low price. There is, therefore, reason to believe that the free market with the U.S. included would be less prone to low prices. However, free trade by the U.S. would not limit price rises. To conclude the discussion of U.S. policy, should a policy of no protection be continued(as in l975)the major effect will be a redis- tribution of income from producers to consumers. Should the estimated 24 percent reduction in domestic production be politically unacceptable, a tariff could be imposed or quotas reinstated but a deficiency payment to producers would be more efficient. The level of protection required would vary from year to year, but would average approximately 20 percent tariff-equivalence. Should it be desirable to avoid price rises of the kind that occurred in 1974, a program of buffer stocks could be institu- ted or domestic prices could be regulated and the wholesale price of sugar be made the pooled average of domestic and international prices. EEC Policy Unilateral action in ceasing protection by the EEC would reduce domestic prices an estimated 23 percent but also reduce domestic pro- duction by 23 percent. The burden of adjustment would fall most heavily 238 on France, where production would contract an estimated 30 percent (c.f. 26 percent in Netherlands, 20 percent in Italy, 15 percent in U.K., 12 percent in West Germany and 8 percent in Belgium). The end of protection would give consumers a huge gain of $709 million, while producers would lose $525 million, giving a net gain to the EEC of $184 million. With so large a net gain, compensation to displaced producers could be arranged-~there are many such precedents in relation to milk production under the Common Agricultural Policy. In 1966-67, 45.6 percent of sugar beet in France was grown in lots of 100 hectares or more per farm2 and a similar, though less extreme, pattern existed in the other countries. If protection of sugar producers is based upon the social necessity of transferring income to the impoverished rural sector, such a policy is sadly mis- judged since beet production is concentrated on the larger, more wealthy farms. The Common Sugar Policy differs from that for most other agricul- tural products in two ways. Firstly, production is to some extent controlled by the complex system of quotas.3 Secondly, surpluses are not stored, but they are jettisoned onto the free market and the pro- ducers receive only the return from that market. Hence the misalloca- tion of resources under the Sugar Policy is likely to be less than that for other products. Turning to the more dynamic aspects of the market, the EEC avoided so meteoric a rise in price as occurred in the U.S.A. in 1974 by 2EEC Statistical Office (undated). Enquete sur la Structure des Exploitations Agricoles, 1967/67. Luxembourg: EEC Statistical Office. 3See Chapter V. 239 taxing exports from the community and subsidizing imports from the Common- wealth producers and from the free market. This action was somewhat equivalent to the pooling of domestic and international prices suggested as a way of combatting price rises earlier in this chapter. However, such a policy reduces the total export supply and leads to greater peaks in price on the residual free market.. Future EEC policy seems aimed at internal self-sufficiency, with imports from Commonwealth countries being balanced by exports. Such a policy implies even greater protection than the (approximately) 30 percent tariff-equivalence at present. The Sugar Policy is very successful in terms of raising the price to producers and has been adjusted in years such as 1974 to protect consumers from very high prices. These gains must be weighted against the cost to consumers of $709 million and the "deadweight loss" of $184 million. The conclusion must be that resource misallocation under the EEC policy by far exceeds that under the U.S. Sugar Act, although the volumes of sugar consumed in the two regions are similar. Less-Developed Countries and Exporters of Cane Sugar Under all policies of free trade, the total gains of all LDCs are less than the gains accruing to Cuba alone. Freer trade would, therefore, lead to a loss of welfare (as measured here) for the majority of LDCs. However, LDC exporters (not including Cuba) would gain approxi- mately $146 million under free trade, despite the loss of revenues from preferential markets. Conversely, LDC importers would lose approximately $120 million under free trade, due to the higher free-market price. 240 The world's exporters of cane sugar comprise the LDC exporters plus Australia, South Africa and Venezuela.4 Their joint gain from free trade would be $639 million or from the end of EEC protection would be $406 million. As before, Cuba is the main beneficiary. The end of U.S. protection is beneficial to this group as a whole, but the gains to Cuba exceed the gains to other countries as a group. Should the cane exporters consider forming a cartel, their effectiveness is likely to be very low. The elastic international supply of beet sugar ensures that the restriction, at least at the level equivalent to a 10 cent per pound export tax, hurts the exporters (except Cuba) as much as the importers. A minor restriction of exports, such as that accomplished under the International Sugar Agreements, might, however, raise the free-market price while not affecting the U.S. and EEC prices (assuming the latter to have protective policies). There is, therefore, little likelihood of a strong cartel develOping--the motivation of widespread gains in income is lacking. International Sugar Agreements International Sugar Agreements have in the past brought together both exporters and importers of sugar with the objectives of smoothing price fluctuations and avoiding disastrously low prices (such as the average 1.86 cents per poundcn1the free market in 1968). The small degree to which price may be raised by export restriction has already been discussed. 4Not considered a LDC here because of oil revenues. 241 The question then arises as to whether an international sugar agreement to remove, rather than impose, barriers to trade might be possible. The U.S.A. and EEC would both gain from freer trade, as would the exporters as a group. The losers would be, as D. Gale Johnson’has pointed out, the traditional importers from the free market, notably Canada, Japan and a group of Asian and African countries. The protection of their domestic producers by the U.S.A. and EEC has caused other countries, which are both producers and importers of sugar, to raise similar barriers to avoid the entry of low-priced sugar from the free market. There could well be many mutual gains from the orchestrated removal of such barriers to trade, implying an adjusted pattern of production in which comparative advantage would be the paramount influence. One function which the International Sugar Organization now ful- fills,in addition to administering International Sugar Agreements, is the collection and dissemination of sugar statistics. Since imperfect knowledge concerning the actions of other producers is a major source of the instability in the sugar market, the 1.5.0. could usefully increase its services both by improving the quality of its data and by extending the coverage to include investments being undertaken in the sugar industry throughout the world. Further Research The suggestions for further research will be listed: 1. One objective of this research, to simulate the dynamics of the world sugar economy, was only partially fulfilled. The development of a more efficient algorithm f0r solving the model is essential in this respect. 242 2. Supply was most adequately modeled for the U.S.A. Further work on European beet sugar supply and the international supply of cane sugar is needed. Further work of a theoretical and empirical nature on cane investment decisions would help in accepting or rejecting the asymmetric model developed in Chapter VI. 3. In analyzing demand, the relationships between alternative sweeteners were ignored. A study of substitution between sucrose, corn syrup, saccharin, etc. in the Western countries would be very useful. 4. A larger number of alternative policies could be evaluated with the model. This would be particularly interesting with respect to U.S. policy in 1976 and the development of International Sugar Agreements. APPENDICES APPENDIX A A NOTE ON DIRECT VERSUS INDIRECT ESTIMATION OF AGRICULTURAL SUPPLY APPENDIX A A NOTE ON DIRECT VERSUS INDIRECT ESTIMATION OF AGRICULTURAL SUPPLY It is a common practice in estimating agricultural supply to separately estimate the derived demand for the predominant input, (e.g., land area, number of animals), and the output per unit of that input (e.g., tons per acre, pounds of carcase weight per animal). The purpose of this note is to clarify the pros and cons of such separate or "indirect" estimation of supply as contrasted with the "direct" estima- tion of the equation in which quantity supplied is the dependent variable. The procedure to be followed will be to state each of the arguments in favor of indirect estimation which have been found in the literature and to critically examine their validity. Henceforth it will be assumed that the supply of a crop is being estimated, although the argument is equally valid for animal products. Perhaps the most influential study of farmers' responses has been that of Nerlove in the 19505.1 However, Nerlove was careful to point out that he had only estimated the derived demand for land and not quanti- ty of product supplied. His justification was that the area which was planted was a better guide to farmers' intended responses than was the quantity of product, since the latter was subject to variation due to weather, pests and diseases. If one were interested only in the derived demand for land, there could be no questioning of Nerlove's approach. 1Nerlove, M. (1958). The Dynamics of Sgpply: Estimation of farmers' Response to Price. (Baltimore: Johns Hopkins Pressfl‘ 243 244 However, if the elasticity of land as an input is taken as the elasticity of supply, this implies that fanmers do not respond to economic variables in determining yield per unit of land, which is highly questionable. As an estimate of the elasticity of supply, the elasticity of land is likely to be both biased and inefficient. Several authors, among them Hee,2 recognized that yield per unit of land was a response to economic variables within the farmer's control and established the tradition of separately estimating the derived demand for land and the yield per unit of land. Quoting Hee, Acreage response and yield response are two separate and distinct functions; a considerable quantity of information in regard to farmers' behavior may be lost when only a single supply function is considered. The second part of the statement is not in dispute, but the first part is questionable. Hee and other researchers made yield a function of a different set of variables from that used to estimate the demand for land. Yet, intuitively, in deciding on the area to plant the farmer must surely have some expectation concerning yield, hence yield and area cannot be completely independent. Yield is not so much a response to different variables, as to the same variables in a later time period at which time their values may have changed. That is, expectations about prices may have changed from the time of planting to the time at which other variable inputs are applied. The following, more formal, presentation may help to clarify the arguments. Suppose there are two inputs, land and fertilizer. Assuming 2Hee, 0. (1958). "The Effect of Price on Acreage and Yield of Potatoes," Agricultural Economics Research, 19, (4), (0ct.). pp. 131-141. 245 a production function to exist, it may be written (1) Q = fQ (L, F), where O = output, L = land and F = fertilizer. Profit may be defined as (2) n = P0 - Q - PL - L - PF . F, where H profit, U ll Q product price, .0 II price of land and PF price of fertilizer. Taking the first derivatives and setting them equal to zero, the marginal conditions for profit maximization may be written, (3) P = flL - PQ , and L (4) PF = le - PQ , where f1L - -€fi%— and -f]F = -§§%— . Utilizing the implicit-function theorem,4 the existence of the following reduced form equations may be established 4For a discussion of implicit functions. see, for example. Thomas, G. B. (1968). Calculus and Analytic Geometry, Reading, Mass.: Addison Wesley, p. 79. 246 A U1 v r— I _ 9L (P09 PL, PF): (6) F = gF (PQ, PL, PF), and /\ \J \J O I — gQ (PO, PL’ PF)‘ Equation (5) is the derived demand for land, (6) is the derived demand for fertilizer and (7) is the supply function. The analogous yield equation is (8) Y = 9y (P09 PL, PF) L), where Y = yield=(Q/L). Should decisions on land and fertilizer be taken at different times, Equations(5) and (8) may be rewritten as respectively, (5) L = 9L (PQ*, PL*, PF*), and (8) Y = 9Y (PQ#. PL#, PF#, L), where * = an expectation and # = a different expectation. This formal presentation has the following implications: (i) the derived demand for land and the yield per unit of land are functions of the same set of variables, except that yield is additionally a function of land area and that the values of the variables may be different due to a change in expec- tations during the gestation of the product; (ii) the longer the delay between planting and the application of other variable inputs, the more probable it is that expecta- tions about prices will have changed so that indirect estima- tion of supply becomes the more promising approach: e.g., for perennial crops indirect estimation would be preferable to di- rect estimation or for cr0ps whose prices are subject to ex- treme variability. 247 Comparing these implications with the procedure of Hee, the crop in question was an annual, the potato, and different variables were included in the land and yield equations. The formal presentation is useful in condemning another approach sometimes found in the literature,5 the use of two-stage least squares (ZSLS) on Equations (5) and (8). The two equations may be collapsed into Equation (7), the supply function, and ordinary least squares (OLS) on (7) is bound to be at least as efficient as ZSLS on (5) and (8). The use of ZSLS implies that decisions on yield and area are simultaneous, which in turn implies that estimation of (7) is the correct procedure. The most common self-deception in the indirect approach to estimating agricultural supply appears to have arisen from an overly 6 Without a well specified model, the pragmatic approach to research. researcher estimates the supply Equation (7) and finds coefficients which are not significantly different from zero. He attributes the lack of significance to "too many variables" and then estimates the demand for land as a function of one subset of the variables and the yield per unit land as a function of a different subset of the variables. More happy, perhaps, with the "significance" of the newly estimated coeffi- cients, the researcher "estimates" total supply by multiplying the yield and area responses together. The chief deception lies in supposing that the supply thus estimated is more significantly related to the 5For example, Ilag, L. M. (1970). An Econometric Analysis of the Impact of the United States Sugar Program on the Philippine Sugar Industry. Unpublished Ph.D. Dissertation, Purdue University. 6 .‘ See, for example, Oury, B. (1966). A Model for Wheat and Feed- grains 1n France. Amsterdam: North Holland Publishing Co. 248 variables than was that directly estimated. Were the researcher able to compute the standard errors of the "compound" coefficients in his new supply function, they would bear values at least as great as in the directly estimated function. Also, by omitting relevant variables in the land and yield equations, the researcher's indirect results are biased. Finally, there is a situation in which indirect estimation is clearly superior. When the demand for land is exogenously determined, such as by governmental control, the land equation becomes a function of governmental decisions and the yield equation a function of the array of prices facing the farmer. Note, however, that when the governmental program is the restriction of output rather than land, the farmer's behavior may be cost minimization for the given output but the variables and argument are exactly as before when profit maximization was assumed. The argument of this note may be summarized as follows. To base estimates of agricultural supply on the derived demand for the most important input alone will lead to biased and inefficient estimates. In general, the separate estimation of the derived demand for the most important input and the output per unit of that input (i.e., indirect estimation), will at best be as efficient as directly estimating supply. Should variables be omitted from either of the equations in the indirect approach, bias will result. In two cases the indirect approach is preferable to the direct. The first involves the supply of products (e.g., perennials), whose gestation is sufficiently long for price 249 expectations to differ at the time of initiating production from those at the time of applying certain variable inputs. The second involves the exogeneity of the major input from the farmer's viewpoint due to governmental or other controls. APPENDIX B A COBWEB MODEL FOR SUGAR APPENDIX B A COBWEB MODEL FOR SUGAR] It has been observed that there is a six to nine year cycle in the price of sugar in the world "free market." The purpose of this note is to demonstrate that such a phenomenon is consistent with two conditions, (i) a demand curve which is shifting steadily to the right over time and (ii) a supply curve which is asymmetric with respect to price and for which quantity supplied is a function of both previous high price and previous (rather than current) prices. The exposition is in two stages, a graphical and an algebraic. The discussion borrows liberally from Waugh,2 who in turn has borrowed from the writings of Ezekiel, Leontief and Nerlove, among others. Graphical Presentation Although the "free market" has many sellers and buyers, assume that the aggregate behavior may be approximated as if there were only one supplier. Also assume that the sugar is all derived from sugar cane, (whereas in actuality only two-thirds is from that source). Figure 1 depicts the supply and demand system, which is envisaged, for an eight year period. abcd is the elastic, long-run supply curve 1This note was written in response to a question raised by Professor J. C. H. Fei of Yale University. , 2waugh, F. v. (1964). "Cobweb Models," J. Farm. Econ 39, (4), (Nov.), pp. 732-750. 250 Pn'cq. fWicc. 251 Figure 8.1. Price and Quantity _ 1c 9: __- - -.- - - " ‘ l __6' A. --‘: - ‘ fly. 1 ' I G - .. ' I l I RH" - -- ‘1 I l P. ----- - -':. . g _. _. _ .— fi --““‘1 "L ' 1 . h 1 .3 1') l. 1 1.: I \ ) fi.'.-'.-:.-§- 'I —- — - L: ‘*bL‘M 1 " :31 a. ‘o. 9 1 Ida—.— a1 “‘0’ ‘ Guatty Figure 8.2. Price and Year 252 and xb and yc are the inelastic, short-run supply curves. A series of demand curves DO . . . D7 is shown, the subscripts denoting the applicable years. Assume that there has been a recent expansion in capacity to point b, which has driven price down to PO. Along the short-run curve there is a single year's delay in production response and hence supply in year 1 is 0], resulting in price P1 which clears the market. In the next year 02 is supplied which results in price P2. The normal action continues until year 5 in which price P5 exceeds the previous high price and there is in consequence a movement along the long-run supply curve from b to c in the next year. The large quantity Q6 depresses the price to P6 and a new cycle is initiated. Figure 2 plots the price from Figure 1 over time. Price rises steadily from P0 to P5, falls heavily to P6 and then begins climbing once again. The length of the cycle depends on the slopes of the short- and long-run supply functions, on the slope of the demand function and on the rapidity with which p0pulation and income shift demand to the right. The market behavior of Figures 1 and 2 assumed that there was a lag of a single year between price and supply for both the short and long runs. Short-run response is an adjustment within current capacity, hence is accomplished in a one year period. This response is highly inelastic because producers resist any restriction on utilization of their investments. Long-run response is an adjustment of capacity, involving investment in extra cane and often in extra factory equipment. This investment lag might be expected to be at least two years in duration and during this delay price will climb to a greater level than 253 shown in Figures 1 and 2, since demand will shift two "steps“ to the right rather than one. Figure 3 demonstrates the action in a market which is typified by such annual short-run adjustments and two-year long-run adjustments. Figure 4 plots price from Figure 3 over time and shows that a two-year investment delay leads to much larger cycles in price. Algebraic Presentation An algebraic presentation will now be given, beginning with a simple cobweb model which is then adapted to the conditions of a shifting demand and an asymmetric supply. The standard cobweb model has price as a function of current quantity, (1) Pt = - aQt and quantity supplied next year as a function of current price, (2) Qt+1 ‘ w1b1 Pt’ where Q = quantity supplied P = price w], b1 and a = parameters and t = a year subscript. By successive substitution from Equation (1) and (2) one may obtain - - = 2 0 o o o Qt+2 ‘ "1b1Pt+i ‘ ‘WibiaQt,1 (w1b1a) 0t. and . (3) Qt+2k = ("1b1a)2k0t” PW'C 0.. 254 Figure 8.3. Price and Quantity f‘:---“"““ 1 1 1 I f} -.. - ‘- - - ’__ __ __ __ __, __ a [I 1 \ I“" ,3 , __ ._ 1 ,H I P‘ --- -‘- ‘ 1 r. ——-- ---*1 1 '1 f. -—--—- '1' "" "l 1 i‘ 1‘1. I1.Rl I'. r..---—-¥ :1» i ’1 -- --4_§£{_-)—° ‘h 1.0.“ ‘1 Figure 8.4. Price and Year 255 Waugh shows that (w1b1a)2 Qfl implies system divergence. The demand function will now be complicated by the inclusion of income and population. It may be stated as (4) Pt = aIQt + azYt + 33Nt9 where Y income t and Nt population. The supply function will be complicated in two stages, in the first of which quantity supplied is made a distributed lag of previous prices and in the second of which the property of asymmetry is intro- duced. Firstly, then, let short-run supply become a function of a geometrically distributed lag of prices,3 SR _ 2 3 (5) Qt+1 ' wlblpt + ("1b1) Pt-l + (”1b1) Pt-2 I ' ' - : SR . . where Q = short-run quantity supplied and w] = a parameter. A Koyck transformation on Equation (5) leads to SR _ SR (5) Qt+l - Wllet + "lblpt' Secondly, let long-run supply be defined as a function of a second distributed lag of prices, beginning one year further back, 3The geometric lag was chosen for simplicity: other lags could have been used. 256 LR _ 2 3 (7) Qt+i “ w2b2Pt-l + ("2b2) Pt-2 T ("2b2) Pt-3 + ‘ ° ° , LR _ . . where Q - long-run quantity supplied, and w2 = a parameter. A Koyck transformation on Equation (7) leads to LR _ LR (8) Qt+1 ’ wzbzot-i + "zbzpt-i ' In the graphical exposition, the long-run supply function was activated whenever price in the current year exceeded the previous maximum price. Because there is assumed to be a two year delay in long-run response, the long-run supply function only operates in year t when price in year t-2 was higher than the previous maximum in that year. Equations (6) and (8) may be combined with this "trigger" mechanism by making w] = 1, when P ;: Pg_2, 0 otherwise, and t 0 . > Pt-2’ 0 otherwise, w2 1, when Pt where Pg_2 = the highest price which has existed up to and including that at time t-2. The combination of the two equations then becomes, (9) 0t = w1b1Qt-1 I WZbZQt-Z T wlblPt-l + w2b2Pt-2’ in which short-run and long-run superscripts are no longer necessary. The system of equations so far consist of a demand function (4) and an asymmetric supply function (9). To complete the algebraic equivalence with the graphical presentation, income and population may be allowed to grow in the following manner, 257 (10) Y (l+gy)Yt , and t+l (1]) Nt+1 (1+gn)Nt 9 where gy and gn are the respective growth rates for income and population. The complete model of the free market in sugar row comprises Equations (4), (9), (10) and (11). Equations (4) and (9) may be combined to give ('2) Qt = w1b1 (1+al) Qt-l T ”zbz (1+al) Qt-Z + w1'31 (aZYt-l l a3Nt-l) T wzbz (aZYt-Z + a3Nt-2)' Equation (12), given (10) and (11), may be solved for any desired year to give the recursively determined supply. The generalization of (12) is cumbersome and will not be elaborated. Because population and income continually shift the demand function, the system is always in dis- equilibrium and one cannot define it as divergent or convergent. However, were population and income to be held constant, the condition for convergence would collapse to (13) (Wlbl)2 (1+a])2 T (W2b2)2 (1+a1)2 < 1 9 for which the first term on the LHS is zero in the long run and the second term on the LHS is zero in the short run. Since b2>>b], by definition, the long-run function, when operating, will lead to greater divergence than the short-run function: A cobweb model for sugar has now been specified, its most impor- tant feature being the asymmetric supply function (9). However, 258 multicollinearity and problems of autocorrelation would make that equation very difficult to estimate. Instead of estimating (9) one may define another equation which possesses the same asymmetry for the short and long run but, by utilizing variables in ratio form, avoids multicollinearity. The function is analogous to Duesenberry's "ratchet" consumption function as adapted by Griliches et al.4 Let Xt = Qt / Pt‘2 _ 0 and Zt - Pt-Z / Pt-Z . The model posits (14) x: = a + szt, and * ('5) xt ' Xt-i = Y(Xt ‘ xt-l)’ where y an adjustment coefficient and * a desired value. Combining Equations (14) and (15) leads to . (16) xt = my + 812t + (I-Y)Xt_] Equation (16) gives asymmetric responses. Its estimation will depend upon the nature of the error term associated with it. Given a "well- behaved“ error, OLS would be appropriate. One may question the use of the same adjustment coefficient, y, for both the short and long run components. Since the sugar cycle is 4Griliches, Z. et a1. (1962). "Notes on Estimated Aggregate Quarterly Consumption Functions," Econometrica, 30, (3), pp. 491-500. 259 of six to nine years duration, it has occurred only three times in the post-war time series and it is not feasible to estimate different adjustment processes for the short and long runs. Summary and Implications This note has shown how an asymmetric supply and smoothly shift- ing demand may produce price cycles of several years' duration. An algebraic formalization led to four equations concerned with supply, demand and the growth of income and population. The system never attains equilibrium, but the conditions under which convergence occurs, given a static income and population, were derived. An empirical approach to modeling the asymmetric supply was described. The world sugar economy consists of many suppliers and demanders rather than the extreme case of a single supplier and a single demander of this note. Further, the sugar entering the market is derived both from the annual, sugar beet, and the perennial, sugar cane. The price cycles of this note are therefore an extreme case and in reality the cycles, as may be observed, are much more complex. Underlying the cycles which have been described is the assumption that imperfect knowledge leads to "irrational" investment behavior and the imperfec- tions may be the result of politically-induced price distortions in the major exporting nations. The continued existence of hog cycles, after many years of economic analysis of their cause, demonstrates that know- ledge among producers within a single country is far from perfect. At the international level knowledge about the action of others is even less, as is the case with sugar. APPENDIX C FULL RESULTS OF THE EIGHT MOST IMPORTANT POLICY EXPERIMENTS N.B. All prices in (l974) cents per lb. and all quantities in thousands of metric tons in this and following appendices. Table C.1. COUNTRY Argentina Australia Austria Barbados Bol/Chile Brazil Canada W. Canada E. Sri Lanka China Taiwan Colombia Cuba Czechosl. Denmark Dom. Rep. Belgium France W. Germany Netherlands Italy Fiji Finland E. Germany Greece Guatemala Guyana Hong Kong Iceland India Indonesia Iran Ireland DEMAND 1,059 712 373 12 558 4,721 360 719 244 2,629 306 649 457 696 275 158 378 2,235 2,299 592 1,831 24 252 758 232 160 35 89 10 4,039 965 961 209 Benchmark I PRICE 7 «cowomooxi 1 O \IO‘NCDCD .0740 .7975 .0705 .9965 .0052 .2470 .4448 .9997 .5894 .7023 .3699 .0601 .7231 .5681 .5021 .9007 .5925 .7892 .3383 .8651 .5965 .7363 .6561 .5607 .9531 .8523 .0310 .9529 .4752 .0344 .6382 13. 3109 13.0729 SUPPLY 260 Results of Policies I and II 1,493 2,909 370 136 284 6,242 O 130 7 3,150 736 856 6,240 734 275 1,021 768 2,426 2,679 413 1,463 379 80 695 145 221 332 O 0 3,360 423 961 173 DEMAND 1,037 708 360 12 551 4,408 358 716 237 2,064 286 605 455 665 279 148 383 2,269 2,346 ’ 655 1,875 24 244 725 235 148 34 85 9 3,895 1,000 1,002 213 Free Trade II 8.81.91: 9. 10. .0341 .8058 11 9 10. 9. .6905 .2914 .7562 10. 10. .9006 .8051 .9610 .0209 9. .1991 .3633 10. .1919 .6306 10. 10. 10. .7691 .8749 11 11 11 9 9 10 11 11 11 11 11 11 9 9. .5241 .4518 .0253 .0308 .4991 .5410 11 11 12 12 11 11 2150 1011 9418 7224 7442 6991 7400 7547 0141 5822 6162 5787 SUPPLY 1,604 2,919 370 153 218 6,218 O 130 7 3,150 710 893 6,220 734 236 912 741 1,971 2,361 356 1,289 410 80 778 145 237 332 O 0 3,637 403 830 173 261 Table C.l. Continued Benchmark I Free Trade III COUNTRY DEMAND ERI§§_ SUPPLY DEMAND PRICE. SUPPLY Jamaica 102 6,9115 492 98 9.7900 611 Japan 3,342 15.5291 970 3,526 11.8957 751 Korea, 398 12.2273 0 400 11.5045 0 Mauritius 333 8.2056 620 31 10.5058 623 Mexico 2,047 7.0329 3,059 2,008 9.937 3,103 New Zealand 174 7.7851 0 167 10.0078 0 Nicaragua 84 6.9586 163 81 10.1486 202 Norway 197 8.7353 0 191 10.8853 0 Pak/Bangla. 712 17.0327 427 954 11.5613 427 Peru 459 9.1101 830 453 9.7611 828 Philippines 763 8.5754 2,160 719 11.1327 2,152 Poland 1,658 7.2122 2,236 1,585 10.5252 2,477 Portugal 261 12.3116 11 266 11.4365 11 Saudi Arabia 157 8.5810 0 149 11.8478 0 Singapore 123 23.4180 0 129 11.6348 0 S. Africa 1,092 7.8738 1,885 1.063 9.8336 1,482 Spain 1,143 8.9063 858 1,120 11.2570 858 Sweden 401 7.6861 260 385 11.8778 260 Switzerland 339 10.2646 77 330 ' 11.5402 77 Thailand 475 9.7249 431 438 12.0854 385 Trin/Tobago 49 6.9549 248 47 9.7989 275 Turkey 910 10.6064 910 855 11.3213 921 USSR W. 5,833 7.3874 10,426 5,586 10.4128 11,575 USSR E. 5,715 8.6816 0 5,482 11.8753 0 U.K. 2,865 12.8185 1,105 2,898 11.4545 982 U.S.A. Hawaii 34 10.8107 944 35 10.1085 914 West 2.829 11.9999 1,920 2,832 11.4759 1,705 South 1,837 12.6177 '1,657 1,849 10.5552 1,260 East 6,033 11.8943 1,281 6,053 10.7233 1,170 P. Rico 134 11.5446 185 136 9.9563 165 Venezuela 509 6.8877 578 478 9.8563 587 E. Europe 2,368 10.6636 1,826 2,351 11.1886 1,826 C. America 603 11.8442 913 620 9.5093 782 262 Table 0.1. Continued Benchmark I Free Trade II COUNTRY DEMAND .PRIQE SUPPLY DEMAND ERIC§_ SUPPLY Parag/Urug. 193 9.5621 148 198 9.2849 147 Near East 921 11.0257 103 913 11.9069 103 Far East 1,327 12.1381 119 1,319 12.1548 119 N. Africa 1,407 17.5210 819 1,622 11.6340 819 W. Africa 506 11.3285 61 536 10.9763 61 N.E. Africa 374 31.0393 271 575 11.8557 271 E. Africa 433 11.1098 319 429 11.3106 319 S. C. Africa 485 7.9887 915 454 10.3083 915 S.W.C. Africa 172 9.7188 172 164 9.9734 172 TOTAL 78,535 78,535 77,575 77,575 263 Table C.2. Results of Policies III and IV No U.S. Sugar Act III No Protection By E.E.C. IV COUNTRY DEMAND ERI§§_ SUPPLY DEMAND PRICE SUPPLY Argentina 1040 7.8906 1570 1045 8.3750 1583 Australia 712 7.9874 2889 710 9.2435 2910 Austria 368 9.0195 370 366 9.5074 370 Barbados 12 8.9774 152 12 8.3467 148 Bol/Chile 552 9.9542 274 548 10.4425 239 Brazil 4577 8.2782 6268 4552 8.4724 6251 Canada W. 357 12.4139 0 357 11.9714 0 Canada E. 714 12.1490 130 716 11.4674 130 Sri Lanka 240 11.3175 7 235 12.1971 7 China 2951 8.6561 3150 2998 9.9333 3150 Taiwan 296 8.7822 736 296 9.8443 736 Colombia 614 9 0293 895 609 8.3764 885 Cuba 456 9.1887 6214 457 8.1302 6230 Czechosl. 686 8.5428 734 680 9.1815 734 Denmark 275 12.6072 277 279 9.9883 208 Dom. Rep. 148 8.8499 956 165 8.2708 1011 Belgium 378 12.5888 768 385 9.5779 707 France 2240 12.5612 2430 2288 9.8741 1588 W. Germany 2299 12.3334 2675 2354 9.7152 2134 Netherlands 597 12.7357 408 721 9.8013 304 Italy 1826 13.4384 1465 1892 10.2386 1164 Fiji 24 8.3212 405 24 9.0424 410 Finland 250 8.4105 80 247 9.3020 80 E. Germany 746 8.5290 723 738 9.2491 744 Greece 230 13.7556 145 228 14.3169 145 Guatemala 151 8.9395 233 149 8.3375 232 Guyana 34 8.9741 335 34 8.3551 335 Hong Kong 92 9.2978 '0 89 10.3796 0 Iceland 10 9.6025 0 10 9.9127 0 India 3941 11.6896 3546 3855 12.3310 3732 Indonesia 932 13.2952 448 897 14.6434 491 Iran 961 13.3103 961 961 13.3206 961 Ireland 209 13.1197 173 215 10.0430 173 264 Table C.2. Continued No U.S. Sugar Act III No Protection by E.E.C. IV COUNTRY DEMAND PRICE SUPPLY DEMAND ERI§§_ SUPPLY Jamaica 98 9.1545 592 100 8.3471 552 Japan _ 3333 x 15.8293 989 3293 16.9998 1069 Korea 397 12.7175 0 390 14.1611 0 Mauritius 33 8.7600 620 32 10.2070 459 Mexico 2034 9.1937 3110 2029 8.5720 3166 New Zeal. 168 8.5920 0 170 9.0972 0 Nicaragua 81 8.9969 192 83 8.4646 184 Norway 195 9.5476 0 193 10.2753 0 Pak/Bangla. 686 17.7297 427 656 18.5203 427 Peru 463 8.7276 813 452 9.0366 822 Philipp. 752 9.2059 2138 738 10.1806 2147 Poland 1625 8.5352 2341 1614 9.0578 2382 Portugal 256 13.4700 11 257 13.7524 11 Saudi Ar. 156 9.9214 0 154 10.4660 0 Singapore 122 23.5294 0 122 24.7888 0 S. Africa 1036 8.2143 1860 1060 9.0155 1702 Spain 1144 9.9629 858 1126 10.1593 858 Sweden 394 8.9235 260 391 ' 9.5619 260 Switzerland 328 11.4130 77 327 12.0092 77 Thailand 458 10.1593 413 448 11.1888 354 Trin/Tobago 47 8.9755 267 48 8.3344 262 Turkey 909 10.0221 909 909 10.0288 909 USSR W. 5757 8.1933 10764 5692 8.9702 11058 USSR E. 5578 9.5339 0 5588 10.3743 0 U.K. 2862 12.9568 1105 2922 9.9723 860 U.S.A. Hawa. 35 9.0064 827 35 10.5634 938 West 2842 10.3423 1311 2830 11.9222 1926 South 1855 9.8271 1093 1839 12.5878 1671 East 6057 10.3271 1142 6035 11.8243 1278 P.Ric. 134 9.2292 171 133 11.4955 176 265 Table C.2. Continued No. U.S. Sugar Act III No Protection by E.E.C. IV COUNTRY DEMAND ERI§§_ SUPPLY DEMAND PRI§E_ SUPPLY Venezuela 480 9.1006 587 496 8.3280 591. E. Europe 2319 12.2605 1826 2303 13.1344 1826 C. America 641 8.8900 751 603 11.8442 913 Parag/Urug. 194 10.5021 148 194 11.1572 155 Near East 915 11.7600 103 903 12.4695 103 Far East 1293 12.7491 119 1215 14.1661 119 N. Africa 1316 20.0507 819 1318 20.3653 819 w. Africa 405 12.9882 61 402 12.9707 61 N.E. Africa 354 33.7464 271 338 36.2026 271 E. Africa 418 11.9651 319 409 12.6141 319 S.C. Africa 473 8.7172 915 465 9.4470 915 S.W.C. Africa 172 9.7188 172 172 9.7188 172 TOTAL 77818 77818 77626 77626 Table C.3. COUNTRY Argentina. Australia Austria Barbados Bol/Chile Brazil Canada w. Canada E. Sri Lanka China Taiwan Colombia Cuba Czechosl. Denmark Dom. Rep. Belgium France N. Germany Netherlands Italy Fiji Finland E. Germany Greece Guatemala Guyana Hong Kong Iceland India Indonesia Iran Ireland 266 Results of Policies V and VI No U.S. Sugar Act or C.A.P. V DEMAND 331g§_ SUPPLY 1042 8.7292 1589 709 9.0123 2907 363 10.1792 370 12 10.1302 154 552 10.9118 225 4476 9.1089 6253 355 13.1984 0 711 13.0052 130 235 12.1928 7 3158 9.7172 3150 297 9.7398 738 604 9.6953 903 455 10.2011 6226 677 9.5382 734 278 10.5043 217 139 10.1484 899 388 9.9544 720 2297 10.3060 1686 2363 10.2301 2215 694 10.3410 322 1883 11.0147 1228 24 9.1557 410 248 9.7055 80 732 8.8321 761 223 15.1591 146 148 9.7957 238 34 9.5793 '330 89 10.5339 0 10 10.7298 0 3828 12.4175 3752 874 14.5876 486 961 13.3104 961 216 10.4807 173 No U.S. Sugar Act and No Cuban Quotas VI DEMAND PRICE SUPPLY 1045 8.3750 1583 710 9.2435 2910 366 9.5074 370 12 8.3467 148 548 10.4425 239 4552 8.4724 6251 357 11.9714 0 716 11.4674 130 235 12.1971 7 2998 9.9333 3150 296 9.8443 736 609 8.3764 885 457 8.1302 6230 680 9.1815 734 279 9.9883 208 165 8.2708 1011 385 9.5779 707 2288 9.8741 1588 . 2354 9.7152 2134 721 9.8013 304 1892 10.2386 1164 24 9.0424 410 247 9.3020 80 738 9.2491 744 228 14.3169 145 149 8.3375 232 34 8.3551 335 89 10.3796 0 10 9.9127 0 3855 12.3310 3732 897 14.6434 491 961 13.3206 961 215 10.0430 173 Table C.3. COUNTRY Jamaica Japan Korea Mauritius Mexico New Zealand Nicaragua Norway Pak/Bangladesh Peru Philippines Poland Portugal Saudi Arabia Singapore S. Africa Spain Sweden Switzerland Thailand Trin/Tobago Turkey USSR w. USSR E. U.K. H.S.A. Hawa. West South East P. Rico Continued No U.S. Sugar Act 267 or C.A.P. V Ff DEMAND PBI§§_ SUPPLY 98 10.0380 620 3335 16.8545 1060 389 14.0941 0 32 9.5967 621 2007 9.9827 3096 169 9.4741 0 81 9.7412 199 192 10.6739 0 657 18.5322 427 455 9.5585 830 751 10.0051 2202 1606 99.4353 2412 251 14.6126 11 152 10.8829 0 122 24.5853 0 1103 8.9608 1712 1125 10.9602 858 390 9.8572 260 323 12.5808 77 451 11.1828 356 46 10.0160 275 895 10.3458 914 5657 9.4153 11255 5558 10.8345 0 2913 10.4643 885 35 9.9507 892 2836 11.1855 1590 1842 10.6709 1290 6016 11.1298 1209 136 9.8528 166 No U.S. Sugar Act and No Cuban Quotas VI DEMAND PRI§§_ SUPPLY 100 8.3471 552 3293 16.9998 1069 390 14.1611 0 32 10.2070 459 2029 8.5720 3166 170 9.0972 0 83 8.4646 184 193 10.2753 0 656 18.5203 427 452 9.0366 822 738 10.1806 2147 1614 9.0578 2382 257 13.7524 11 154 10.4660 0 122 24.7888 0 1060 9.0155 1702 1126 10.1593 858 391 9.5619 260 327 12.0092 77 448 11.1888 354 48 8.3344 262 909 10.0288 909 5692 8.9703 11058 5588 10.3743 0 2922 9.9723 860 35 10.5634 938 2830 11.9222 1926 1839 12.5878 1671 6035 11.8243 1278 133 11.4955 176 268 Table C.3. Continued No U.S. Sugar Act No U.S. Sugar Act and or C.A.P. V No Cuban Quotas VI COUNTRY DEMAND 331g§_ SUPPLY DEMAND 3319§_ SUPPLY Venezuela 484 10.0495 586 495 8.3280 591 E. Europe 2283 13.6148 1826 2303 13.1344 1825 6. America 625 10.0245 796 503 11.8442 913 Parag/Urug. 193 11.6443 158 194 11.1572 155 Near East 894 13.0054 103 903 12.4695 103 Far East 1209 13.7811 119 1215 14.1551 119 N. Africa 1239 21.8941 819 1318 20.3553 819 N. Africa 351 13.8578 51 402 12.9707 51 N.E. Africa 340 35.9241 271 338 36.2025 271 E. Africa 409 12.5293 319 409 12.5141 319 S.C. Africa 454 9.4815 915 455 9.4470 915 S.W.C. Africa 172 9.7188 172 172 9.7188 172 TOTAL 77393 77393 77626 77626 269 Table C.4. Results of Policies Xc and Xd Benchmark Plus 10 Cent Export-Tax Xc Benchmark Plus 20 Cent Export-Tax Xd COUNTRY DEMAND P_RI_C_E_ SUPPLY DEMAND m SUPPLY Argentina 1070 4.1751 1070 1099 4.3955 1099 Australia 729 2.2226 1525 741 .9281 741 Austria 353 13.0293 370 348 14.4841 370 Barbados 13 2.5943 67 13: .4977 13 801/Chile 530 15.2738 305 522 17.4247 350 Brazil 5366 4.1815 5366 5365 4.1868 5365 Canada w. 355 15.4706 0 352 17.5911 0 Canada E. 708 15.9054 130 705 17.6280 130 Sri Lanka 221 14.7304 7 213 17.7823 7 China 1552 13.2889 3150 1215 15.0793 3150 Taiwan 382 2.8729 632 421 1.7921 421 Colombia 713 4.6646 713 714 4.6451 714 Cuba 467 2.7330 5035 473 1.3560 3218 Czechosl. 655 12.4664 734 645 14.1216 734 Denmark 273 13.4967 308 271 14.6822 323 Dom. Rep. 326 2.9323 635 449 1.7992 449 Belgium 374 13.7141 794 376 14.5369 801 France 2214 13.7782 2807 2198 14.5928 2996 N. Germany 2269 13.4452 2877 2247 14.3703 3002 Netherlands 564 13.7743 458 573 14.7595 495 Italy 1800 14.5128 1553 1804 15.1525 1636 Fiji 25 2.3834 178 26 .1970 26 Finland 240 12.2814 80 238 14.0802 80 E. Germany 712 12.2615 817 698 14.1893 853 Greece 215 17.8165 146 211 19.3723 147 Guatemala 176 4.4789 176 176 4.4789 176 Guyana 36 2.5865 177 42 .6576 42 Hong Kong 83 13.4554 0 79 15.9297 0 Iceland 9 13.4063 0 9 15.1247 0 India 3798 12.6013 3798 3798 12.6013 3798 Indonesia 774 18.3788 622. 686 20.6314 686 Iran 961 13.3185 961 961 13.3194 961 Ireland 207 14.5130 173 205 14.8578 173 270 Tab1e C.4. Continued Benchmark Plus 10 Cent Export-Tax Xc Benchmark Plus 20 Cent Export-Tax Xd COUNTRY DEMAND £8125. SUPPLY DEMAND 231g§_ SUPPLY Jamaica 107 2.8370 233 122 1 5140 122 Japan 3194 19.9425 1257 3142 21.8485 1377 Korea 371 18.4229 0 . 353 20.9048 0 Mauritius 41 1.9574 585 53 1255 53 Mexico 2175 3.4451 2175 2175 3.4342 2176 New Zeal. 150 13.2131 0 157 15.8784 0 Nicaragua 95 3.4059 95 94 3.4514 94 Norway 186 13.3156 0 183 15.1505 0 Pak/Bangla. 571 21 1400 427 498 23.5522 427 Peru 518 3.3175 518 548 3.6528 548 Philipp. 1159 2.5915 1398 1194 2.3597 1194 Poland 1555 12 3602 2592 1535 13.8205 2572 Portugal 241 17 1390 11 235 18.9195 1 Saudi Ar. 145 13.4943 0 138 15.1472 0 Singapore 121 27 6264 0 120 30.4040 0 S. Africa 1737 2.8771 1737 1711 2.9912 1711 Spain 1082 13.7139 858 1062 15.2854 858 Sweden 375 12.5349 250 375 14.4104 260 Switzerland 310 15.1083 77 312 16.8345 77 Thailand 484 9.2956 484 484 9.2955 484 Tin/Tobago 54 3.0236 129 50 .1 4047 50 Turkey 790 13.2759 935 747 14.8415 945 USSR w. 5481 12.0977 12083 5384 13.9381 12531 USSR E. 5411 13.3902 0 5384 15.2550 0 U.K. 2840 14.5553 1248 2835 14.7818 1285 U.S.A. Hawa. 34 12.2581 1052 34 14.1520 1192 Nest 2819 13.4820 2562 2801 15.1531 3779 South 1835 13.5511 1883 1822 15.4299 2201 East 5995 14.4780 1535 5975 15.2398 1711 P.Ric. 132 13.3527 239 130 15.1037 436 Tab1e C.4. Continued 271 Benchmark Plus 10 Cent Export-Tax Xc COUNTRY Venezuela E. Europe C. America Parag/Urug. Near East Far East N. Africa w. Africa N.w. Africa E. Africa S.C. Africa S.W.C. Africa TOTAL DEMAND 526 2199 666 188 847 1074 1117 191 289 374 436 106 76244 PRICE 5.4206 17.4914 7.3973 13.6762 15.4664 17.3452 27.5233 18.6760 44.5782 15.2131 11.9960 13.2675 SUPPLY 526 1826 666 188 103 119 819 61 271 319 915 172 76244 Benchmark Plus 20 Cent Export—Tax Xd DEMAND 528 2219 666 188 842 922 1067 125 271 344 411 100 75281 PRICE 5.2866 19.3885 7.3894 13.6764 17.3010 20.7206 30.3567 20.0908 48.6692 17.6662 14.4814 14.1337 SUPPLY 528 1826 666 188 183 119 819 61 271 319 915 172 72536 APPENDIX D THE TRADE FLOWS UNDER POLICIES I AND II AND THE TARIFFS AND QUOTAS UNDER POLICY I 272 Table 0.1. Trade Flows Under Policy I TRADE FLOWS UNDER BENCHMARK SOLUTION TRADE FLOWS UNDER BENCHMARK SOLUTION Quantity Quantity §22£E§ Destination Shipped Spppge Destination Shipped Argentina Argentina 1059.27 Cuba North Africa 489-95 Argentina 80]. Chile 131.38 Czechoslavakia Czechoslavakia 546.64 Argentina New Zealand 174.55 Czechoslavakia Switzerland 187.36 Argentina Par. Uruguay 51.67 Denmark Denmark 275-34 Australia Australia 712.47 Dom. Rep. Dom. Rep. 158 60 Australia Indonesia 542.70 Dom. Rep. POCtUQal 93-96 Australia Japan 81.33 Dom. Rep. Spain 98.00 Australia Singapore 123.00 BEIQIUW BEIQIUW 378 55 Australia Far East 1106.59 Belgium Netherlands 13.17 Austria Austria 352.29 Belgium U.K. 376.74 Austria Greece 7.71 France France 2235.68 Barbados Barbados 12.85 France Italy 154-39 Barbados Portugal 19.12 France ICEIBOd 36 23 801. Chile 801. Chile 278.33 N- Germany W. Germany 2299 01 Brazil Brazil 4721.32 W. Germany Netherlands 166.12 Brazil India 350.23 W. Germany Italy 213.88 Brazil Spain 128.12 Netherlands Netherlands 413.41 Brazil w. Africa 445.40 Italy Italy l463 10 Canada M Canada E 130.00 F131 Fljl 24-73 Sri Lanka Sri Lanka 7.00 Fiji Japan 189.36 China China 2329.44 Finland .Finland 80.00 China Japan 541.63 E- Germany E. Germany 508.57 China Korea 278.93 E. Germany Norway 186.74 Taiwan Taiwan 325.50 Greece Greece 145.57 Taiwan Hong Kong 89.97 Guatemala 801. Chile 4.92 Taiwan Japan 260.29 Guatemala Guatemala 160.72 Colombia 801. Chile 137.70 Guyana Guyana 35.27 Colombia Colombia 549.48 Guyana Spain 59 23 Cuba B01. Chile 5.57 India India 3262.51 Cuba Canada E 589.08 Indonesia Indonesia 423.15 Cuba Cuba 1 457.19 Iran Iran 961.55 Cuba Greece 79"” ' Ireland Ireland 173.00 Cuba Iceland 10.17 Jamaica Jamaica 102.84 Cuba Japan 1218.76 Jamaica Portugal 118.17 Cuba Portugal 19.54 Japan Japan 970.39 Cuba Near East 524.53 Mauritius Sri Lanka 160.01 Table 0.1. Continued TRADE FLOWS UNDER BENCHMARK SOLUTION Quantity Shipped Mauritius Mexico Mecico Mexico Nicaragua Nicaragua Pak. Bangladesh Peru Philippines Philippines Philippines Poland Poland Poland Poland Poland Poland Portugal S. Africa S. Africa S. Africa Spain Sweden Switzerland Thailand Trin/Tob. Trin/Tob. Turkey Destination Mauritius Canada W Japan Mexico Canada W Nicaragua Pak. Bangladesh Peru Philippines Thailand Far East Austria Poland Saudi Arabia Sweden Switzerland EurOpe Portugal India Pak. Bangladesh S. Africa Spain Sweden Switzerland Thailand Trin/Tob. N. Africa Turkey A11 Prod. A11 Mkts 33. 352. 80. .55 .35 85. 2047 7 427 61 27 11 157 141 74 217 11 77 69525.01 273 39 74 79 88 .00 460. 763. .68 .20 .23 1633. .89 .73 .80 .50 .00 324. 285. 1092. 858. 260. 00 43 56 99 33 09 00 00 .00 414. 49. 56. 910. 26 43 75 25 TRADE FLOWS UNDER BENCHMARK SOLUTION Source USSR RI USSR RI USSR RI USSR RI USSR RI U.K. Hawaii Hawaii San Francisco New Orleans New York New York Puerto Rico Puerto Rico Venezuela Venezuela E. Europe C.Nmrka Par/Uruguay Near East Far East N. Africa W. Africa I N.E. Africe . Africa .C. Africa .C. Africa .C. Africa .C..Africa .C. .W. Africa C. Africa Destination Finland Norway USSR RI USSR VL Near East U.K. USA HAW USA SFR USA SFR USA NOR USA NOP USA NY USA NOR Puerto Rico Venezuela N. Africa E. Europe C. America Par/Uruguay Near East Far East N. Africa W. Africa N.E. Africa E. Africa Sri Lanka India N.E. Africa E. Africa S.C. Africa S.W.C. Africa Quantity Sh'pped 172 194 35 1657 129. .76 .41 134. 589. .34 1826. 603. 142. 103. 119. 819. .00 .00 312. 1151 51 41 61 271 77 101 103. .88 485. .00 121 172 .61 10. 4333. 5715. 99 37 73 .01 1105. .00 909. 1920. .04 77 09 38 54 10 60 00 98 11 00 00 00 00 30 .80 45 56 274 Table 0.2. Trade Flows Under Policy 11 TRADE FLOWS UNDER FREE TRADE TRADE FLOWS UNDER FREE TRADE Quantity Quantity Sgggge Destination Shipped Source Destination Shipped Australia Indonesia 416.88 Belgium Belgium 383.71 Australia Japan 1171.52 Belgium U.K. 358.21 Australia Singapore 129.69 France France 1971.85 Australia USSR VL 493.92 W. Germany W. Germany 2346.26 Austria Austria 360.17 W. Germany Netherlands 14.76 Austria Near East 9.83 Netherlands Netherlands 356.02 Barbados Barbados 12.77 Italy Italy 1289.67 Barbados Portugal 140.88 Fiji Fiji 24.65 801. Chile 801. Chile 218.90 Fiji Japan 386.15 Brazil Brazil 4488.87 Finland Finland 80.00 Brazil Italy 419.82 E. Germany E. Germany 725.58 Brazil Portugal 114.62 E. Germany Switzerland 52.81 Brazil Switzerland 132.24 Greece Greece 145.11 Brazil N. Africa 675.82 Guatemala Guatemala 148.32 Brazil W. Africa 467.56 Guatemala USA SFR 89.32 Canada E Canada E 130.00 Guyana Guyana 34.46 Sri Lanka Sri Lanka 7.00 Guyana Spain 262.43 China China 2064.85 Guyana U.K. 35.60 China Korea 400.67 India India 3637.34 China USSR VL 684.48 Indonesia Indonesia 483.77 Taiwan Taiwan 286.13 Iran Iran 830.02 Taiwan Hong Kong 85.39 Ireland Ireland 173.00 Taiwan Japan 338.51 Jamaica Jamaica 98.82 Colombia Colombia 685.49 Jamaica U.K. 512.53 Colombia USA SFR 288.45 Japan Japan 751.39 Cuba Cuba 455.56 Mauritius Sri Lanka 230.36 Cuba USA NOR 899.14 Mauritius India 258.08 Cuba USA NY 4865.61 Mauritius Mauritius 31.88 Czechoslovakia Czechoslov. 655.79 Mauritius Saudi Arabia 103.55 Czechoslovakia Switzerland 68.21 Mexico Canada W 358.05 Denmark Denmark ' 236.37 Mexico Canada E 95.88 Dom. Rep. Canada E 490.53 Mexico Mexico 2008.48 Dom. Rep. Dom. Rep. 148.22 Mexico USA NOR 641.30 Dom. Rep. Ireland 40.12 Nicaragua Nicaragua 81.40 Dom. Rep. U.K. 233.83 Nicaragua USA SFR 121.20 2755 Table 0.2. Continued TRADE FLOWS UNDER FREE TRADE TRADE FLOWS UNDER FREE TRADE Quantity Spppgg Destination Shipped §EEEE§ Destination Pak. Bangl. Pak. Bangl. 427.00 USSR RI Near East Peru 801. Chile 332.87 U.K. U.K. Peru Peru 453.89 Hawaii Japan Peru USA SFR 41.44 Hawaii USA HAW Philippines Indonesia 180.46 San Francisco USA SFR Philippines Philippines 719.46 New Orleans Pak. Bang. Phi1ippines Thailand 52.32 New Orleans USA SFR Philippines Far East 1200.37 New Orleans USA NOR Poland Netherlands 240.82 New York USA NY Poland Poland 1585.99 Puerto Rico Puerto Rico Poland Sweden 125.65 Puerto Rico N. Africa Poland E. Europe 525.50 Venezuela Iceland Portugal Portugal 11.00 Venezuela Venezuela S. Africa S. Africa 1063.44 Venezuela N. Africa S. Africa Near East 418.83 E. Europe E. Europe Spain Spain 858.00 C. America USA SFR Sweden Sweden 260.00 C. America C. America Switzerland Switzerland 77.00 Par/Uruguay Italy Thailand Thailand 385.72 Par/Uruguay Par/Uruguay Trin/Tob. Trin/Tob. 47.31 Near East Near East Trin/Tob. U.K. 228.43 Far East Far East Turkey Italy 65.32 N. Africa N. Africa Turkey Turkey 855.76 W. Africa W. Africa USSR RI Denmark 42.79 N.E. Africa N.E. Africa USSR RI France 207.58 E. Africa E. Africa USSR RI Netherlands 43.87 S.C. Africa Saudi Arabia USSR RI Iran 172.68 S.C. Africa N.E. Africa USSR RI Norway 191.80 S.C. Africa E. Africa USSR RI USSR RI 5586.06 S.C. Africa S.C. Africa USSR RI USSR VL 4303.73 S.W. C. Africa W. Africa USSR RI U.K. 547.47 'S.w.C. Africa S.W.C. Africa All Products, All Markets 77575.2706 Quantity Shipped 225.89 982.19 879.01 35.01 1706.00 527.97 424.07 308.71 1170.04 136.56 28.97 9.95 478.64 98.71 1826.00 162.35 620.47 100.96 46.99 103.00 119.00 818.00 61.00 271.00 319.00 46.21 304.18 110.60 454.02 7.49 164.51 COUNTRY Argentina Austrailia Austria Barbados B01. Chile Brazil Canada W. Canada N.E. Srilanka China Taiwan Colombia Cuba Czechoslovakia Denmark Dominican Republic Belgium France W. Germany Netherlands Italy Fiji Finland E. Germany Greece Guatemala Guyana Hong Kong Iceland India Indonesia Iran Ireland Jamaica Japan Korea Mauritius Mexico New Zealand Nicaragua Norway Pak/Banglasesh Peru Philippines Poland Tab TARIFFS USED FOR BENCHMARK SOLUTION 276 1e 0.3 SPECIFIC TARIFF (CENTS OO—‘VOOOOOOU'IOONO-‘OCOO-#0000OOOOOOOOOOOOd-‘OOOOOO PER POUND) .00 .00 .00 .00 .00 .00 .77 .77 .00 .00 .00 .00 .00 .00 .00 .00 AD VALOREM TARIFF (PERCENT) 140 01 NN —‘ N 00 01ONOOOOOOUTOOOOUTOOOOOOUWOOOOOOOOOUTOOOOOOOOUWOOO -'N N NKD COUNTRY Portugal Saudi Arabia Singapore S. Africa Spain Sweden Switzerland Thailand Trin/Tob Turkey USSR W. USSR E. U.K. USA Hawaii USA West USA South USA East Puerto Rico Venezuela E. Europe C. America Par/Urug Near East Far East N. Africa W. Africa NE Africa E. Africa SC Africa SWC Africa 277 Table 0.3. Continued TARIFFS USED FOR BENCHMARK SOLUTION SPECIFIC TARIFF (CENTS OONOOOONOOO-‘OOOOOOOOOOONOONDOW PER POUND) .69 AD VALOREM TARIFF (PERCENT) 0 OOOOOOOWU‘IOOOOO-‘OOO \INUWN NNN OOU‘IOOOUWOUW NM 0101 278 Table 0.4. Quota Flows Under Benchmark Solution Source Destination Quantity Shipped Argentina USA New York 77.0000 Australia USA NY 343.0000 Barbados U.K. 103.0000 Barbados ' USA NY 2.0000 Bolchile USA NY 6.0000 Brazil USA NY 597.0000 Taiwan USA NY 80.0000 Colombia USA NY 69.0000 Cuba China 300.0000 Cuba Czechos 150.0000 Cuba Egerman 250.0000 Cuba Korea 120.0000 Cuba Poland 25.0000 Cuba USSR 1500.0000 Cuba E Europe 325.0000 Cuba Far East 75.0000 Dom. Rep. USA NY 671.0000 Fiji U.K. 142.0000 Fiji USA NY 41.0000 Guatemal USA NY 56.0000 Guyana U.K. 209.0000 Guyana USA NY 29.0000 India U.K. 25.0000 India USA ' NY 73.0000 Jamaica U.K. 271.0000 Mauritiu U.K. 386.0000 Mauritiu USA NY 41.0000 Mexico USA NY 578.0000 Nicaragu USA NY 70.0000 Peru USA NY 370.0000 Philippi USA NY 1308.0000 S. Africa U.K. 86.0000 5. Africa USA NY 97.0000 Thailand USA NY 17.0000 Trin/Tob U.K. 133.0000 Trin/Tob USA NY 9.0000 Venezuel USA NY 28.0000 C. Americ U.K. . 21.0000 C. Americ USA NY 289.0000 Par/Urug USA NY 6.0000 E. Africa U.K. 7.0000 S.C. Africa USA NY 25.0000 BIBLIOGRAPHY BIBLIOGRAPHY Agricultural Stabilization and Conservation Service. (undated). "Returns, Costs and Profits, Florida 1967-69 Crops." U.S. Department of Agriculture, Washington, D.C. (undated). "Returns, Costs and Profits, Hawaiian Sugar Plantations, 1967-69 Crops." U.S. Department of Agriculture, Washington, D.C. ' . (undated). "Returns, Costs and Profits, Louisiana 1969-71 Crops." U.S. Department of Agriculture, Washington, D.C. Aitchison, J. and Brown, J. A. C. (1954). "A Synthesis of Engel Curve Theory." Review of Economic Studies, 22, (1), pp. 34-46. Ballinger, R. A. (1971). "A History of Sugar Marketing." Agricultural Economics Report No. 197, U.S. Department of Agriculture, Washington, D.C. (1972). "Economic Behavior in the U.S. Sugar Market." California Agricultural Experiment Station, Bulletin No. 859. Bates, T. H. (1965). The World Sugar Economy and U.S. Supply Policy. Unpublished Ph.D. Dissertation, University of California, Berkeley. Berndt, E. R. and Christensen, L. R. (1973). "The Translog Function and the Substitution of Equipment and Labor in U.S. Manufacturing, 1929-1968." Journal of Econometrics, l, (1), pp. 81-114. Binswanger. H. (1974). "A Cost-Function Approach to the Measurement of Factor Demand and Substitution." American Journal of Agricultural Economics, 56, (2),(May), pp. 377-386. Campbell, J. (various). "Returns, Costs and Profits from Sugar Cane Farms." Louisiana State University, Department of Agricultural Economics. Choudhury, P. (1967). An Economic Appraisal of the Aggregate Sugar Supply Response for Selected Major Producing Countries. Unpublished Ph.D. Dissertation, University of Hawaii. Christensen, L. R., Jorgenson, D. W. and Lau, L. J. (1971). "Conjugate Duality and the Transcendental Logarithmic Production Function." Econometrica, 32, (4), (July), pp. 255-256. 279 280 Comité Européen des Fabricants de Sucre. (annual). Recueil Annuel de Statistiques. Paris: Comité Européen deS FaBricants 3e SUcre, 45 Avenue Montaigne. Commodity Research Bureau Inc. (1974). Commodity Yearbook. New York: Commodity Research Bureau, Inc. Compagnie Financiére Sucres et Denrées. (1974). The Rise of Sugar Prices: Its Effect on Consumption. Paris: Compagnie Financiére Sucres et Denrées. Council on Wage and Price Stability. (1975). Staff Report on Sugar Prices. Office of Wage Price Monitoring, Washington, D.C.: Government Printing Office. De Leeuw, F. (1962). "The Demand for Capital Goods by Manufacturers: A Study of Quarterly Time Series." Econometrica, 39, (3), (July), pp. 407-423. Durbin, J. (1970). "Testing for Serial Correlation in Least- Squares Regression when Some of the Regressors are Lagged Dependent Variables." Econometrica, 33, (3), (May), pp. 410-421. Egbert, A. C. and Kim, H. M. (1975). "Analysis of Aggregation Errors in L. P. Models." American Journal of Agricultural Economics. 3], (2), (May), pp. 292-301. European Economic Community. (1974). Agricultural Markets, No. 8, Brussels: E.E.C. European Economic Community Statistical Office. (undated). Enguete sur la Structure des Exploitations Agricoles, 1966/67. Luxemburg: E.E.C. Statistical Office. Fan, C. L. (1967). Determination of Sugar Supply Functions for Taiwan. Unpublished Ph.D. Dissertation, University of Hawaii. Griliches, Z., Maddala, R., Lucas, R. and Wallace, N. (1962). "Notes on Estimate Aggregate Quarterly Consumption Functions." Econometrica, 39, (3), pp. 491-500. Griliches, Z. (1967). "Distributed Lags: A Survey." Econometrica, _3__5_g (1), (11311.), pp. 16-49. Gupta, Y. P. (1969). "Least Squares Variant of the Dhrymes Two-Step EstTmation Procedure of the Distributed Lag Model." International Economic Review, 13, (1), pp. 112-113. 281 Harris, S. and Smith, I. (1973). World Sugar Markets in a State of Flux. London: Trade Policy Research Centre. Hayami, Y. and Ruttan, V. W. (1971). Agricultural Development: An International Perspective. Baltimore: Johns Hopkins Press. Hayenga, M. (1967). "Sweetener Competition and Sugar Policy." ’Journal of Farm Economics, 53, (4), (Dec.), pp. 1362-1366. Hee, O. (1958). "The Effect of Price on Acreage and Yield of Potatoes." Agricultural Economics Research, 19, (4), (0ct.) pp. 131-141. Henderson, J. and Quandt, R. (1958). Microeconomic Theory. New York: McGraw-Hill. Hoch, I. (1958). "Simultaneous Equation Bias in the Context of the Cobb-Douglas Production Function." Econometrica, 33, (3), (0ct.), pp. 566-578. Hughes, H. (1971). Analysis of Sugar Cane Production in Sao Paulo. Unpublished Ph.D. Dissertation, University of Missouri, Columbia. Ilag, L. M. (1970). An Econometric Analysis of the Impact of the U.S. Sugar Program on the Philippine Sugar Industry. Unpublished Ph.D. Dissertation, Purdue University. Ingersent, K. A. (1975). "Sugar Prices and Stocks." Journal of Agricultural Economics, 33, (2), (May), pp. 227-238. International Customs Tariff Bureau. (various). International Customs Journal. Brussels: I.C.T.B. ' International Labor Office. (various). Bulletin of Labor Statistics. Geneva: United Nations International Labor Office. International Monetary Fund. (various). International Financial Statistics. Washington, D.C.: International Monetary Fund. International SUgar Council. (1963). The World Sugar Economy: Structure and Policies. London: International Sugar Council (now Organization), 28 Haymarket, S.W.l International Sugar Organization. (annual). Sugar Yearbook. London: International Sugar Organization, 28 Haymarket, S.W.l. Johnson, D. Gale. (1974). The Sugar Program: Large Costs and Small Benefits. Washington, D.C.: American Enterprise Institute. Johnson, G. L. and Quance, L. (1972). The Overproduction Trap_jn U.S. Agriculture. Baltimore: Johns Hopkins Press for Resources for the Future, Inc. 282 Johnson, H. G. (1966). Economic Policies Towards Less Developed Countries. New York: Brookings Institution. Josling, T. E. (1969). "A Formal Approach to Agricultural Policy." Journal of Agricultural Economics, 30, (2), (May), pp. 175-192. Just, R. E. (1974). Econometric Analysis of Production Decisions with Government Intervention: The Case of California Field ’CrOpS. Monograph No. 33, Giannini Foundation, California Agricultural Experiment Station. King, R. A. and Ho, F. S. (1972). Reactive Programming: A Market Simulatipg Spatial Eguilibrium Algorithm. Economics Research Report No. 21, Department of Economics, North Carolina State University. Klein, L. R. (1953). A Textbook of Econometrics. New York: Row, Peterson and Company. Kmenta, J. and Joseph, M. E. (1963). "A Monte-Carlo Study of Alternative Estimates of the Cobb-Douglas Production Function." Econometrica, 31, (3), (July), pp. 363-385. Kmenta, J. (1964). "Some Properties of Alternative Estimates of the Cobb-Douglas Production Function." Econometrica, 33, (1), (Jan ), pp. 183-188. (1971). Elements of Econometrics. New York: Macmillan. Licht, F. 0. (various). International Sugar Newsletter. Ratzeburg, West Germany: F.0. Licht. Mundlak, Y. (1961). "Empirical Production Function Free of Management Bias." Journal of Farm Economics, 31, (1), (Feb.), pp. 45-56. Nerlove, M. (1958). The Dynamics of Supply: Estimation of Farmers' Response to Price. Baltimore: Johns Hopkins Press. Organization for Economic Cooperation and Development. (1973). Supply Control in Agriculture. Paris: O.E.C.D. Oury, B. (1966). A Model for Wheat and Feedgrains in France. Amsterdam: North Holland Publishing Company. Pignalosa, F. (1974). Personal Communication. F.A.O.: Rome. Pringle, G. E. ’(1969). A Temporal-Spatial Analysis of Sugar Production and Marketing in Puerto Rico. Unpublished Ph.D. Thesis, University of Wisconsin. Ramsey, J. B. (1972). "Limiting Functional Forms for Market Demand Curves." Econometrica, 59, (2), pp. 327-341. 283 (1973). "Classical Model Selection through Specification Error Tests." In Zarembka, P. ed. Frontiers in Econometrics. New York: Academic Press. . (l974). "Limiting Functional Forms for Demand Functions: Tests of Some Specific Hypotheses." Review of Economics and Statistics, 99, (4), (Nov.), pp. 468-477. SamuelSOn, P. A. (1952). "Spatial Equilibrium and Linear Programming." American Economic Review, 99, (3), (June), pp. 283-303. Sanchez, N. (1972). The Economics of Sugar Quotas. Unpublished Ph.D. Dissertation, University of Southern California. Smith, V. L. (1963). "Minimization of Economic Rent in Spatial Price Equilibrium." Review of Economic Studies, 39, (1), (Feb.), pp. 24-31. Snape, R. H. (1969). "Sugar: Costs of Protection and Taxation." Economica, 39, (141), (Feb.), pp. 29-41. Storr, N. and Warnken, P. (1974). The Location of Sugarbeet Production in the U.S. Unpublished manuscript, University of Missouri, Columbia. Sturrock, F. G. (1969). "Sugar Beet or Sugar Cane." Journal of Agricultural Economics, 99, (1), (Jan.), pp. 125-132. Storrock, F. G. and Thompson, M. C. (1972). "Sugarbeet: A Study of Sugar Production in the U.K. and Feasibility of Expansion." Economic Repprt No. 7, Department of Land Economy, Cambridge Univeristy. Takayama, T. and Judge, G. G. (1964). "Spatial Equilibrium and Quadra- tic Programming." Journal of Farm Economics, 99, (1), (Feb.), pp. 67-93. Thomas, G. B. (1968). Calculus and Analytic Geometry. Reading, Mass.: Addison Wesley. Tramel, T. E. and Seale, A. D. (1963). "Reactive Programming Techniques. In Interregional Competition Research Methods Workshop. North Carolina State University, January 10-11. United Nations. (1974). Demographic Yearbook. New York: United Nations. (1975). Yearbook of National Accounts Statistics, 1973. New York: United Nations. United Nations Food and Agriculture Organization. (1971). Agricultural Commodity Projections, 1970-1980. Rome: U.N.F.A.O. 284 United Nations Food and Agriculture Organization. (annual). Production Yearbook. Rome: U.N.F.A.O. U.S. Agency for International Development. (1974). Gross National Product: Growth Rates and Trend Data. Washington, D.C.: U.S.A.I.D. U.S. Department of Agriculture. (various). Sugar Statistics and _ Related Data. Vol. 1 and 2. Washington, D.C.: U.S.D.A. United States Naval Oceanographic Office. (1964). Distances Between Ports. Washington, D.C.: U.S. Government Printing Office. Viton, A. and Pignalosa, F. (1961). Trends and Forces of World Sugar Consumption. Commodity Bulletin No. 32, United Nations Food and Agriculture Organization, Rome. Walter, B. J. (1972). The Wholesale Pricing System for Sugar. Unpublished Research Report, University of California, Berkeley. (1973). "Sweetener Economics." Paper presented at the 165th National Meeting of the American Chemical Society, Dallas, Texas, April 8-13, 1973. Waugh, F. V. (1964). "Cobweb Models." Journal of Farm Economics, 49, (4), (Nov.), pp. 732-750. Weather Bureau. (1964). Climatic Summary of the United States. Washington, D.C.: U.S. Department of Commerce. Weymar, F. H. (1968). The Dynamics of the World Cocoa Market. Cambridge, Massachusetts: M.I.T. Press. Young, K. H. (1969). Demand for Sugar in the United States: A Synthesis of Time Series and Cross Section Analysis. Unpublished Ph.D. Dissertation, Columbia University. Zellner, A., Kmenta, J. and Dreze, J. (1965). "Specification and Estimation of Cobb-Douglas Production Function Models." ‘ Econometrica, 33, (4), pp. 784-795. Zellner, A. and Revankar, N. S. (1969). "Generalized Production Functions." Review of Economic Studies, 39, (2), (April), pp. 241-250. Zusman, P., Melamed, A., and Katzir, I. (1969). Possible Trade and Welfare Effects of EEC Tariff and Reference Price Policy on the Eupgpean-Mediterranean Market for Winter Orapges.v Giannini Foundation Monograph No. 24, California Agricultural Experiment Station.