V. _ . . n...” y s\. —\‘.. ,-V‘_._‘.flv,> sh ( ' ‘ / “ 'V“ '__‘ #16»- —2a 1* '1 CHIGANSTAT men x MMMMMMMMM M M MMMM M MM MMMMMMM “M M Michigan State Jnivenityj- Egg 51v LIBRARIES This is to certify that the dissertation entitled AN ANALYSIS OF INSTABILITY IN INTERNATIONAL WHEAT AND RICE MARKETS & EFFECTS OF PRICE UNCERTAINTY ON PRICE LEVELS AND MARKETING MARGINS IN preISNTElébLATIONAL WHEAT TRADE Thomas S. Jayne has been accepted towards fulfillment of the requirements for Ph.D. degree in Agricultural Economics flaWt‘Q 873%“ Major professor Date—111134282— MS U is an Affirmative Action/Equal Opportunity Institution 0-12771 PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. DATE DUE DATE DUE DATE DUE MSU Is An Affirmative Action/Equal Opportunity Inflitution u-p sf ‘n - f. -.'1'.' ‘\ I "Irlue'SI' VJ l PART I ., AN ANALYSIS or INSTABILITY IN INTERNATIONAL . HHEAT AND RICE MARKETS and PART II EFFECTS OF PRICE UNCERTAINTY 0N PRICE LEVELS AND MARKETING NARGINS IN INTERNATIONAL WHEAT TRADE by Thomas S. Jayne A DISSERTATION Submitted to Michigan State University "Irin partial fui iiieent of the requirements for the degree of 1 '« gar. <- nocma or PHILOSOPHY Department of Agricultural Econoeics 1989 ‘ I ‘ ”I v .. . . It V .‘ \ . r T-g 5.. . M‘ w l '._ ‘. . I,. 1‘ \4‘4— u \ u '1’” A. 0'. 1‘ 5 .5? (004 0044. ABSTRACT AN ANALYSIS OF INSTABILITY IN INTERNATIONAL HEAT AND RICE MARKETS and EFFECTS OF PRICE UNCERTAINTY 0N PRICE LEVELS AND MARKETING MARGINS IN INTERNATIONAL WHEAT TRADE By Thomas S. Jayne Government trade and commodity policies are normally assumed to destabilize world prices. An implication is that movements toward free trade in grain would appreciably stabilize world prices. Part I identifies the conditions required for government insulation policies to destabilize world markets and econometrically examines whether these conditions apply to the trade behavior of seven major rice trading countries over the period 1960-87. The analysis indicates that while government stabilization policies may exacerbate world price instability in some cases, movements toward freer trade will not address fundamental sources-- mainly structural and chronic--contributing to instability in the rice market: (1) the geographic concentration of world production in an area of unstable weather; (2) very low consumption responsiveness to domestic prices, which would therefore prevent national trade Thomas S. Jayne sensitivity to world prices from rising much under free trade; and (3) a persistently thin and fragmented market where price information is difficult to obtain, and where the development of coordination mechanisms to reduce trade uncertainties are thwarted by major traders’ tendency to sporadically float in and out of the market. The analysis suggests that research and policy work designed to mitigate instability in international grain markets Should be broadened to include structural and organizational considerations. The objective of Part II is to determine the effect of price uncertainty on the level of prices and marketing margins in major international grain trading routes. Uncertainty in U.S. and Japanese wheat port prices are measured by a GARCH—M process, and are contrasted with results obtained using alternative models of risk devised in previous studies. Results indicate only moderate correlation between measures of price variability used previously and price uncertainty as measured by the GARCH-M model. The GARCH results indicate that price uncertainty is significantly associated with lower U.S. export price levels and to a lesser extent, Japanese import prices. The average loss to the U.S. wheat sector has exceeded $7.6 million per year due to the adverse effects of price uncertainty in this trade route alone. These results have implications for U.S. trade policy because they suggest that important benefits arise from stabilized international wheat prices. - A ACKNOWLEDGEMENTS I have benefitted greatly from the assistance of many people in the course of this study and during my graduate program in general. In particular, I owe a sincere debt to Dr. James Shaffer, thesis advisor, and Dr. Michael Neber, whose direction, advice, and conceptual insights over the last four years have been invaluable. Their counsel and help have extended far beyond academic concerns, and have been deeply appreciated. I also owe a great deal to Dr. Robert Myers. His guidance and assistance in this research, as well as in my academic development, has been considerable. Thanks also to Drs. John Staatz and Vernon Sorenson for their insights, advice, and careful readings of an earlier draft. This research has also benefitted from the varied assistance of Fred Surls, Julia McKay, Vince Peterson, and Kelley Harrison. Funding for this research was provided by the Michigan Agricultural Experiment Station Projects 1502 and 1566, in connection with North Central Regional Research Project 194. Finally, I am very grateful to my wife Connie Currier-Jayne for her unwavering support throughout my doctoral program. She is perhaps the only one more happy than me that it is finished. iv ‘. TABLE OF CONTENTS Page LIST OF TABLES ....................... I. . . LIST OF FIGURES .......................... PART I AN ANALYSIS OF INSTABILITY IN INTERNATIONAL WHEAT AND RICE MARKETS CHAPTER 1 INTRODUCTION ..................... l l 1 Problem Statement and Research Objectives ....... l 1.2 Research Organization and Methods ........... 4 1.3 The Importance of Instability in International Grain Markets ..................... 6 l 4 Point of Departure from Existing Literature ...... 10 CHAPTER 2 THE ORGANIZATION OF THE WORLD "HEAT AND RICE MARKETS ..................... 13 2.1 Trade by Type of Wheat and Rice ............ 13 2.2 Major Actors in the Wheat and Rice Markets ...... 17 2.3 Coordination Mechanisms and Marketing Institutions . . 32 2.4 National Trade and Stockholding Patterns ....... 43 CHAPTER 3: GOVERNMENT INSULATION POLICIES AND HORLD MARKET INSTABILITY .................. 49 3.] Theoretical Rationale ................. 49 3.2 Past Studies ..................... 51 3.3 The Model ....................... 55 3.4 Interpretation of Hheat Results ............ 64 3.5 Interpretation of Rice Results ............ 75 3.6 Conditions Necessary for Government Insulation to Affect Horld Price Instability ........... 82 3.7 Conclusions ...................... 100 BEER CHAPTER 4: ORGANIZATIONAL SOURCES OF INSTABILITY IN THE WORLD HHEAT AND RICE MARKETS ............. 102 4.1 The Reliability of Foreign Markets and Transaction Costs ................... 102 4.2 Geographical Concentration of World Rice Production ...................... 107 4.3 Green Revolution Technology and Self-Sufficiency Policies ....................... 111 4.4 Petroleum Prices and Rice Import Demand ........ 114 4.5 Wheat-Rice Interrelationships ............. 115 CHAPTER 5: CONCLUSIONS AND POLICY IMPLICATIONS .......... 117 5.1 Summary ........................ 117 5.2 Instability and Market Organization .......... 126 5.3 Policy Implications and Alternative Mechanisms to Reduce World Market Instability .......... 130 5.4 Areas for Further Research .............. 135 REFERENCES ............................ 138 APPENDICES Appendix 1 ......................... 149 Appendix 2 ......................... 152 Appendix 3 ......................... 163 Appendix 4 ......................... 166 Appendix 5 ......................... 171 Appendix 6 ......................... 173 PART II EFFECTS OF PRICE UNCERTAINTY 0N PRICE LEVELS AND MARKETING MARGINS IN INTERNATIONAL WHEAT TRADE CHAPTER 1: INTRODUCTION ..................... 177 CHAPTER 2: CONCEPTUAL FRAMEWORK: DETERMINANTS OF MARKETING MARGINS IN INTERNATIONAL GRAIN TRADE ROUTES ...... 182 2.1 Actors in the International Grain Marketing System . .182 2.2 Factors Affecting Prices and Marketing Margins in International Grain Trade ............... 189 2.3 Model Specification .................. 195 vi CHAPTER 3: ALTERNATIVE MODELS OF PRICE UNCERTAINTY ........ 203 3.1 Uncertainty Measures Derived from Variations in Past Prices .................... 203 3.2 The GARCH-M Model ................... 206 CHAPTER 4: ESTIMATION AND RESULTS: THE CASE OF U.S.-~JAPAN WHEAT TRADE ...................... 211 4.1 GARCH-M Estimation Results .............. 211 4.2 Comparison of Results from Alternative Indicators of Price Uncertainty ................. 225 CHAPTER 5: CONCLUSIONS AND POLICY IMPLICATIONS .......... 233 5.1 Empirical Conclusions and Policy Implications ..... 233 5.2 Methodological Conclusions .............. 237 5.3 Areas for Further Research .............. 238 REFERENCES ............................ 240 vii I-Z. I-3. I-4. 1-6. I-7. I-8. I-9. I-IO. LIST OF TABLES PART I The Magnitude of Instability in World Wheat and Rice Markets .......................... 2 Rice Import Price Per Metric Ton Among Major Importers . . . 15 Major Traders by Type of Rice ................ 15 Matrix of Alternative Trading Actors and Their Prevalence in the World Wheat Market, 1982/83-1985/86 ......... 21 State Trading and Other Forms of Administrative Coordination in World Wheat and Rice Markets ........ 24 Long Run Bilateral Wheat Contracts, 1981-87 ......... 36 The Five Largest Importers and Exporters in the World Wheat and Rice Markets: 1960-87 ........... 44 Annual Average Stockholdings of Wheat and Rice: 1980-87 . . 47 Variable Definition and Data Sources ............ 62 A Comparison of Price Elasticities of Net Export Supply in World Wheat Trade ................. 65 . Potential Trade Instability Resulting From National Production Variations: Wheat (1960-87) ........... 68 . Sources of Production Instability in Major Wheat Trading Areas, 1960-87 ................... 73 . Price Elasticity of Net Export Supply Estimates in the World Rice Trade ................... 76 . Potential Trade Instability Resulting From Domestic Production Variations: Rice (1960-87) ........... 78 . Sources of Production Variability in Major Rice Trading Countries ................... 80 . Estimates of World-to-Domestic Price Transmission Elasticities for Selected Rice Trading Countries ...... 87 viii II-3. II-4. II-S. II-6. II-7. II-8. . Estimates of World Price Elasticity of Net Export Supply: Actual and Free Trade-Implied .......... . Estimates of Domestic Price Elasticity of Consumption and Endstocks for Selected Rice Trading Countries ..... . Correlation Coefficients of National Rice Production Deviations from Trend in Monsoon Asia ........... PART II . Ocean Freight Rates and Marketing Spreads in Three international wheat trading routes: 1970-86 Averages. . . . Correlation Coefficients of First Differences in Monthly Wheat and Rice Prices in Various International Markets, 1972-1986 .................... F-Test Results for Presence of Seasonality in Major World Grain Markets .................... Alternative Indicators of Price Uncertainity ....... Maximum Likelihood Estimations Results for Equations 13 ....................... Maximum Likelihood Estimation Results for Equations (14) . Maximum Likelihood Estimates of Equations (15) ...... Correlation Coefficients of Alternative Measures of Uncertainty Applied to Monthly Japanese Wheat Import Prices, 1970-1986 ................. . OLS Estimates of Equations (16) .............. . 89 . 92 .108 .191 .194 .201 .204 .213 .217 .222 .227 .230 II-3. LIST OF FIGURES PART I Theoretical Effect of Government Insulation Policy on World Price Instability ................. 50 Transmission Effect under Free Trade for Small-Country Case ..................... 94 Transmission Effect under Free Trade for Large-Country Case ..................... 100 Distribution of National Consumption to Production Ratios, Wheat: 1980-87 ................... 104 Distribution of National Consumption to Production Ratios, Rice: 1980-87 ................... 105 Heuristic Model Representing the Effects of Correlated Production in Asian Rice Importing and Exporting Countries on Price Instability ............... 110 PART II . Marketing Margins in Selected International Grain Trade Routes: 1978-86 ................... 192 . Time Varying Conditional Variances and Covariances of U.S. Export and Japanese Import Price Forecast Errors from GARCH-M Model (13) ............... 214 Time Varying Conditional Variances and Covariances of U.S. Export and Japanese Import Price Forecast Errors from GARCH-M Model (15) ............... 221 PART I AN ANALYSIS OF INSTABILITY IN INTERNATIONAL WHEAT AND RICE MARKETS Chapter 1 INTRODUCTION 1.1 PROBLEM STATEMENT AND RESEARCH OBJECTIVES Heightened instability is one of the most conspicuous features distinguishing current world grain markets from those of the 1950-605. Because rice and wheat are the two most important food grains in world consumption, increased instability in these markets may have severe effects on hunger in many regions of the world. The 1986 Ministerial Declaration of the Uruguay GATT Round emphasized the "urgent need to bring more discipline and predictability to world agricultural trade by correcting and preventing restrictions and distortions...so as to reduce the uncertainty, imbalances, and instability in world agricultural markets.” This research is motivated by the need to better understand the sources of instability in world wheat and rice markets and their effects on the organization, behavior, and performance of these markets. Since 1960, the magnitude of price instability in the world rice market has far exceeded that in the world wheat market. This is true in both quarterly and annual terms, using alternative instability measures for often-quoted price indicators in each market (Table I-l). Moreover, instability during the 19705 and 19805 appears greater in both markets than in the 19605. Conventional wisdom posits that much of the heightened volatility in prices is due to government insulation TABLE I-l THE MAGNITUDE 0F INSTABILITY IN WORLD WHEAT AND RICE MARKETS VARIANCE OF COEFFICIENT or VARIATION ANNUAL PERCENTAGE CHANGES * or THE REGRESSION ** PERIBD ' WHEAT PRICE RICE PRICE WHEAT PRICE RICE PRICE QUARTERLY 1 1955-71: 20.5 54.0 .05 .21 1972-79: 204.4 185.1 .28 .35 1980-87: 30.5 55.5 .07 .15 1965-87: 95.4 114.0 .27 .39 ANNUAL 1950-71: 78.1 201.9 .05 .17 1972-79: 938.4 1800.5 .23 .37 1980-87: 86.4 511.5 .05 .18 1950-87: 358.8 748.4 .29 .41 Wheat Price: SUS/mt, United States No. 2 Hard Winter, f.o.b. Gulf Ports Rice Price : SUS/mt, Thailand milled 5% brokens, f.o.b Bangkok Source: International Wheat Council (various issues), International Monetary Fund (1986). n * Computed as: VAR [ Z (ln(pt)-ln(pt-1))*100)] t-l ** Computed as: 0 / p , or, the standard deviation of the dependent variagle (computed by regressing price on a linear time trend and taking the standard deviation of the residuals), divided by the mean price. 3 policies that stabilize domestic markets at the expense of international price instability, achieved by severing the link between domestic and world prices (Johnson, 1975; Bale and Lutz, 1979; Shei and Thompson, 1978; Falcon and Monke, 1979-80; Bigman, 1985). Since rice is among the most heavily regulated commodities in world trade, its degree of instability is also often attributed to such insulation policies. For example, in their analysis of the rice market Falcon and Monke conclude that: The preference of governments for quantitative trade controls rather than tariffs isolates domestic price movements from world price movements, and fluctuations in world prices become largely irrelevant to the short-run production and consumption adjustment mechanisms within each country. The structure of trade causes the relationship between domestic and world price movements to be a function mainly of government policy, because the quantities of imports and exports, or at least the variations in these quantities over time, are largely determined by policy makers. Given this institutional context, it is understandable why past econometric studies of rice trade could not link production, consumption, and trade with world prices in a statistically significant manner (1979-80, p. 288). The objectives of this study are to reassess the validity of the above-stated conventional wisdom and to identify other potentially important sources of instability and uncertainty in the wheat and rice markets. In addition, the study attempts to account for the great degree of price instability in the world rice market relative to the wheat market, and to examine whether this has produced different influences on the organization and performance of these markets. Specifically, the study is guided by the following research questions: 4 1. Have government "insulation" policie5--i.e., commodity and trade policies designed to stabilize domestic markets by venting domestic variations onto the world market-- significantly increased instability and uncertainty in world rice and wheat markets? What are the conditions necessary for such policies to destabilize the world market? 2. Are there other potentially more important sources of instability in these markets? 3. What factors account for the relatively high degree of instability in the world rice market compared to the wheat market? Have these sources of instability influenced the organizational structure, behavior, and performance of these markets? Readers interested only in a summary of main research findings and policy implications may consult the concluding sections 5.1-5.3. 1.2 RESEARCH ORGANIZATION AND METHODS This study is divided into five chapters. The remainder of this chapter briefly discusses the importance of instability in international grain markets, placing the study within the literature on this subject. Chapter 2 examines the organization and behavior of the wheat and rice markets, and their influence on risk and instability. The chapter begins with a description of the major public and private actors in the markets, their objectives and behavior, and the institutional mechanisms through which trade is coordinated. The remainder of the chapter examines the trade, stock, production, and consumption patterns of the national actors in the wheat and rice markets. Chapter 3 examines the relationship between government insulation policies and international market instability. A theoretical model is developed to demonstrate the mechanism by which government insulation 5 may destabilize international markets, according to the "conventional wisdom." This theory is examined empirically in a partial equilibrium econometric model. In particular, the model examines (1) the degree to which importing and exporting countries transmit production instability onto world markets (the "transmission effect"), and (2) the extent to which countries stabilize the market by adjusting annual trade volumes in response to world price signals (the “absorption effect"). The chapter concludes by identifying the conditions required for government insulation policies to affect world market stability, and empirically examines whether these conditions apply to the trade behavior of seven major rice trading countries over the period 1960-87. This is done by comparing the estimated transmission and absorption effects (under actual insulation conditions) with those implied under free trade, by incorporating several behavioral restrictions drawn from standard trade theory. Sensitivity analysis is then used to discern how robust the results are to Changes in domestic and international price elasticities. Since domestic wheat price data for many countries was unavailable, a similar analysis of the wheat market was not performed. Chapter 4 identifies a set of alternative factors-~largely related to the inherent organization of the rice trade--that generate a high degree of instability and uncertainty in the world rice market relative to the wheat market. The final chapter is divided into four sections. The first section synthesizes the results of Chapters 2-4 with respect to the main sources of instability in the rice and wheat markets. The second section summarizes the ways in which such instability has in turn 6 affected the structure and organization of these markets. These observations are more conjectural due to the complexity and multifaceted influences exerted on world commodity markets, and are presented as hypotheses subject to further analysis. The third section discusses the policy implications of these results, and examines the advantages and drawbacks of several alternative policy mechanisms to reduce the level of instability in wheat and rice markets. Areas for further research are discussed in the final section. 1.3 THE IMPORTANCE OF INSTABILITY IN INTERNATIONAL GRAIN MARKETS Instability is not "bad“ per se; it is the uncertainty often associated with instability that can reduce social welfare. Uncertainty is an ex-ante concept, and can be defined as the variance of a forecast error (Engle, 1982; Engle et al., 1987; Bollerslev et al., 1988). Instability is more often measured as the ex-post deviation in an outcome from some combination of past outcomes, such as a mean, moving average, or historical trend. While it is crucial to retain this distinction between instability and uncertainty, there may be a high degree of correlation between the two (Brennan, 1982). Until recently, the literature relating instability and uncertainty to the behavior of market actors and welfare amounted to the study of instability under certainty (Just, 1987). The often-cited works by Waugh (1944), Di (1961), and Massell (1969) attempted to measure the welfare effects of price instability under the assumption that either there is no time lapse between resource allocation .decisions and the sale of output, or that future prices, while 7 unstable, are nevertheless known and that agents can costlessly adjust their known marginal costs to these known future prices. Yet in the real world of distributional and biological production lags, agents must often decide the amount they expect to produce (procure) without knowing the price at which they can sell (buy) it. Technically, they can respond only to the parameters of a subjective price distribution rather than to the future price itself. Abundant empirical evidence has shown that producers, due to risk aversion, often decrease acreage or production in response to increasing price uncertainty (Behrman, 1968; Just, 1974; Traill, 1978; Brorsen, Chavas and Grant, 1987), indicating that the supply curve may Shift with changes in the variance of prices. The more recent literature examining economic efficiency under uncertainty, information costs, and incomplete forward and contingency markets has decoupled the link between competitive equilibrium and pareto optimality (Hart, 1975; Newbery and Stiglitz, 1981). The general conclusion from this literature is that when markets are incomplete, instability and risk can lead to suboptimal resource allocation in agriculture, and government stabilization policies can theoretically improve economic efficiency (Myers and Oehmke, 1987). Whether they actually possess the resources or political will to do so is an empirical question. While very valuable, this literature has confined itself to a particular aspect of instability: its effect on efficiency, holding market structure constant. There are at least two important paths by which further study can augment this literature and contribute 8 meaningfully to policy discussion. The first is to consider explicitly a fuller range of market performance criteria beyond efficiency. Many governments in the world formulate their grain market policy goals in terms of food security, political and economic stability, growth, and equity in addition to efficiency (Sorenson and Rossmiller, 1983). Thus, policies that influence the level of instability in grain markets must be evaluated with respect to this broader range of goals. The second addition to the instability literature is to examine its feedback effects on market organization and policy responses, and the corresponding changes in the level and distribution of risk in the system. Changes in technology, institutional arrangements, and general economic interdependence since the 19605 have clearly created new levels of uncertainty, transactions costs, and external effects in world grain production and trade. These changes have in turn created incentives for new modes of coordination, new reactive steps by private and public actors to reduce exposure or vulnerability to risk and transactions costs, creating new patterns of investments and external effects (Shaffer, 1969). Therefore, institutional organization is influenced by instability and risk, just as instability and risk are influenced by institutional organization. The above issues are related to ongoing research under North central Regional Research Project 194 (Conner et al., 1988). This study attempts to contribute to this area of research by examining how instability influences market organization, government behavior, and trade patterns. 9 The reverse question -- the effect of government behavior on market instability —- has been the focus of a different set of recent studies (Zwart and Meilke, 1979; Blandford and Schwartz, 1983; Blandford, 1983; Siamwalla and Haykin, 1983; Wilde et al., 1985). These studies tend to substantiate the contention that government insulation policies have produced low import and export responsiveness to world prices. Blandford’s study of the wheat market, for example, finds that only one country/region out of 12 (the United States) varies its net exports in response to Changes in world wheat price at the 10 percent significance level. Less than one third of the 55 country groups in the Siamwalla and Haykin rice study Show any short-run responsiveness to world rice prices. It has become somewhat of a truism that such low trade elasticities are attributable to government price and trade policies that insulate domestic markets from world price fluctuations. However, low net export elasticities do not necessarily mean that government insulation is the cause. Nor does government insulation require that increased world price instability is the result. This study identifies the conditions necessary for government insulation policies to affect the stability of world rice and wheat markets. 1.4 POINT OF DEPARTURE FROM EXISTING LITERATURE Previous studies linking government insulation policies to world market instability have focused on two effects: the extent to which production variations have been transmitted abroad in the form of trade variations (transmission effect), and the degree to which annual trade 10 volumes of large trading actors are sensitive to world price signals (absorption effect).1 If the transmission effect is high and the absorption effect is low, it has commonly been concluded that government insulation policies are the cause, and that such policies have thus exacerbated world market instability (Blandford, 1983; Schwartz and Blandford, 1983; Falcon and Monke, 1979-80; Siamwalla and Haykin, 1983; Bigman, 1985). However, this approach is unwarranted in many cases because no comparison is made with the magnitude of these effects under non-insulation. Such methods can lead to inappropriate conclusions regarding the sources of instability in world grain markets. This paper identifies the theoretical conditions necessary for government trade/commodity policies to affect world market stability and then estimates the extent to which they have done so for seven major rice trading countries during the period 1960-87. This is done by comparing the estimated transmission and absorption effects (under actual insulation conditions) with those implied under free trade, by incorporating several behavioral restrictions drawn from standard trade theory. Moreover, past conclusions of low absorption effects appear sensitive to model specification and definition of world price. In the rice trade for example, the national trade elasticities of many countries become much more elastic when exchange rate and domestic infTation data are included in the definition of world price, when a non-linear relationship between trade volume and world price is 1 The transmission and absorption terms are attributable to Blandford (1983), and have since been used frequently in the literature on the subject. 11 allowed, and when a broader range of variables affecting trade enter into the analysis. The study provides empirical estimates of the transmission and absorption effects for 18 countries and regional blocs in the world wheat and rice markets over the period 1960-87, and analyzes their implications for world market instability. The relationship between government insulation and world market instability is not unidirectional. National food security, economic efficiency, and political stability are greatly influenced by the performance of domestic grain markets, which are in turn affected by the stability of international grain markets. It is perhaps understandable why governments seldom allow international market forces to freely affect their constituents. Efforts to improve the performance of domestic wheat and rice markets require a better understanding of the cause-effect relationships between government insulation policies and instability in world markets. Inappropriate conclusions regarding the causes of instability may result in strategies that fail to address the roots of the problem and further entrench political opposition to multilateral market coordination stressing comparative advantage and freer interplay between domestic and world prices. The study attempts to document the dynamic feedback effects of world price instability on market organization and the behavior of public and private actors in the system. Finally, the study analyzes alternative strategies to reduce the instability of the world rice and wheat markets and their probable consequences. CHAPTER 2 THE ORGANIZATION OF THE WORLD WHEAT AND RICE MARKETS A myriad of public and private actors and institutions interact to coordinate grain trade across national borders. If instability is related to the organization of a market, analysis must begin with a description of the actors and attributes that affect it. 2.1 TRADE BY TYPE OF WHEAT AND RICE While there are many varieties of wheat (most commonly winter, spring, and durum) and some specialized uses, the substitutability between these products is generally high. This is evident from the high price correlations between different types of wheat (International Wheat Council). Grades and standards are well-defined and internationally recognized. Prices for each type of wheat can be readily quoted in f.o.b. or c.i.f terms in a number of international wheat trading ports. Indicators of future prices by type are quoted daily on the Chicago futures market. Rice, by contrast, is comprised of four distinct varieties that are of limited substitutability in major rice consuming areas. Eighty- five percent of rice traded is indiga, a longer-grain variety that is f1uffy and non-sticky when cooked. The remainder is jangniga_(a shorter-grain that is semi-sticky and moist when cooked), aromatic (a 12 13 scented long-grain variety), and glutinous (waxy and gelatinous when cooked).2 Demand for indica rice is divided into milled and parboiled types, which are both further divided into different qualities. Major quality criteria are percentage of broken grains, chalkiness, translucency of the grains, and odor when cooked. The differences in quality among internationally-traded rice is evident from the extreme variation in unit import prices among different countries (Table I-2). For example, unit prices in a particular year may vary 200 percent or more. Many countries have distinct preferences for one type of rice, although not all of them appear willing or able to pay a high premium for that type (Table I-3). Saudi Arabia and Nigeria prefer high quality parboiled rice and are willing to pay for it. In Thailand, however, parboiled rice sells at a discount to regular milled rice. The demand for high quality, U.S. and Thai milled rice is concentrated in Iran, Iraq, Western Europe, and the United States. While low- quality (more than 20 percent broken) milled long grain rice is sold mainly to poorer countries, some countries (e.g., Senegal) actually prefer this type of grain. Because of distinct preferences, there may be a surplus in the market for one type of rice and a shortage in another at the same time. As a result, prices for indica and japonica rice may move somewhat independently of one another (Rastegari-Henneberry, 1985; Petzel and Monke, 1980). However, among the medium- and long-grain indica markets 2 Another variety, glaberimma, is found in West Africa, but is of minor importance in world rice trade. 14 TABLE I-Z RICE IMPORT PRICE PER METRIC TON AMONG MAJOR IMPORTERS 1983 1984 1985 COUNTRY S U.S. / metric ton SAUDI ARABIA 564 540 477 NIGERIA 438 454 426 IRA 420 411 400 IRAN 324 420 400 HONG KONG 374 346 303 RAZIL 360 487 224 INDONESIA 329 319 280 MALAYSIA 283 270 260 SENEGAL 224 212 214 BANGLADESH 137 229 115 Source: computed from FAO (1985). TABLE 1-3 MAJOR TRADERS BY TYPE OF RICE TYPE QUALITY MAJOR EXPORTERS MAJOR IMPORTERS Brokens Thailand, Burma Senegal, Madagascar South Vietnam, Gambia Low Thailand, Pakistan, Indonesia, most of INDICA, China, Burma West Africa MILLED Medium United States, Brazil, Hong Kong, Thailand, Pakistan Indonesia, U.S.S.R. High United States, Western Europe, Iran, Thailand Iraq, Malaysia INDICA , Low Burma, Thailand Bangladesh, Sri Lanka PARBOILED MILLED High United States, Nigeria, Saudi Arabia, Thailand Western Europe, Canada IJAPONICA --- Japan, China, Australia Indonesia, South Korea I BROWN Parboiled --- W. Europe, South Africa Source: Rastegari-Henneberry (1985) 15 which account for the bulk of internationally traded rice, prices are highly integrated across countries, qualities and time (Petzel and Monke, 1980). On this basis, Falcon and Monke argue that "the price of any widely traded variety, such as Thai 5 percent brokens can serve as a reasonable indicator of movements in all world markets" (p. 284). This conclusion is supportable when the time frame of analysis is annual or even semi-annual; in the Short run, however, prices across countries and varieties may move independently for a month or two at a time. Japonica markets especially may not be well integrated with indica markets such as the Thai export market (Petzel and Monke, 1979- 80). In this respect, the rice market is more fragmented and price information less readily available than in the wheat market. Another major distinction between international rice and wheat trade is that the former has no universally used grades or standards. Also in contrast to the wheat market, no reliable, internationally accepted Spot or futures prices can be quoted for each type or quality of rice (Rastegari-Henneberry, 1985). Several past attempts to establish rice futures markets on various U.S. exchange boards have ended in failure. While the reasons for this are discussed in Chapter 5.2, the important point here is that matching buyers with sellers of a particular type and quality of rice is more difficult than in the wheat market because (1) grades are inadequately standardized, (2) reliable price information about a particular market (specifying variety, quality, time and location) may be difficult and costly to obtain, especially in the short run, and (3) the rice market lacks an established trading forum in which buyers and sellers can interact with 16 low search costs. This conclusion is supported by the existence of numerous rice brokerage houses earning very high fees to match potential sellers with buyers (see Chapter 2.2). Because of wide differences in variety, quality, and consumption preferences, it may be more accurate to regard the world rice trade as a set of distinct markets with moderately to highly substitutable products. 2.2 MAJOR ACTORS IN THE WHEAT AND RICE MARKETS While much attention has been given to government policies as a source of grain market volitility, relatively little research has focused on instability induced by the structure and coordination of trade within the grain marketing system. An understanding of price movements in international markets requires a better understanding of how government and private actions interact (McCalla, 1979). This section develops a framework for analyzing this issue. Grain trade across borders requires coordination between at least three broad types of marketing actors: exporters, importers, and an international middleman who links them together. Exporters are the set of firms that collect grain from inland markets and transport it to export ports. These firms include exporters who could sell the product either to international middlemen or directly to importers. A similar set of importing firms procure grain for domestic storage, processing and/or resale in the importing country. The third group of actors are the specialized national or multinational grain trading firms that link domestic markets together by buying from exporters and selling to importers. Such international middlemen exist as distinct 17 from exporters because of "the complexities of international marketing which require specialized knowledge in exchange rate conversion, ocean shipping, international legal issues, and particular information about the international market" (McCalla, 1979, p.210). Frequently, the role of exporter, middleman, or importer may be combined by a single marketing actor that vertically integrates operations either backward or forward. Integration may occur to reduce transaction costs and uncertainty across stages (Williamson, 1975), to provide economies of scale, or to increase market power (Schrader et al., 1986). Market power, and hence pricing outcomes, may also hinge on the number of firms fulfilling the roles of exporter, importer, and international middleman (McCalla 1979). The actors who fill these three roles vary greatly from country to country. Exporters and importers are typically state trading agencies, farmer cooperatives, flour milling firms, multinational grain trading firms, and smaller private firms. Of these, only the large multinational traders operate extensively as middleman in the world wheat market. In the rice trade, cooperatives and millers are relatively more important. However, there has been an increasing trend in both the wheat and rice market toward direct government-to- government contracts, especially when the countries involved possess market power. An example of this would be the Canadian Wheat Board negotiating to sell wheat to the Japanese Food Agency. In such cases, international middlemen do not assume ownership of the grain, although they still may handle the grain in a logistical capacity. This is because state trading agencies seldom become directly involved in 18 logistics; they sell principally on an f.o.b. basis and buy on a c.i.f. basis (Schmitz, 1986). Therefore, trade across borders, even under government-to-government contracts, usually involves a middleman, normally a multinational grain trading corporation. A critical factor affecting the coordination of trade between exporter, middleman, and importer is the domestic policy environment of the trading countries. A plethora of policy combinations are possible, each of which fit into one of four broad categories. The first category is that of complete transmission of world price movements to the domestic markets of both the importing and exporting country. This would be the case in a free trade environment, or one where government policy does not directly alter the relationship between domestic and world price (e.g., deficiency payments or a simple tariff).3 Category 2 is that of complete price transmission on the import side, but where the exporting government isolates its domestic price from the world price, through either price supports, variable export taxes/subsidies, or state marketing boards. The third and opposite category is that of complete world-to-domestic market price transmission on the export side, but insular policies on the import side. Category 4 represents insulation policies in both the importing and exporting country. Table 1-4 summarizes the possible combinations of market actors and policy environments, and their prevalence in total wheat trade between 3 I use the term I'free trade“ with the recognition that it is largely an abstraction, since all market exchange--free or administered--takes place within a set of legal and institutional constraints imposed by governments (Shaffer, 1987). In this study, free trade refers to a condition under which world and dometic prices are equal, except for transport costs. 19 12 major trading regions (United States, Canada, Argentina, Australia, European Community-12, Japan, China, India, Soviet Union, Egypt, the 17-state Mideast bloc, and Brazil). Table 1-4 indicates that, for the 12 regions examined (accounting for 57 percent of total annual wheat trade over the 1982/83-1985/86 period), no transactions occurred in which prices were unregulated in both the exporting and importing countries. In fact, no importing country has allowed world market forces to freely affect its domestic wheat price. Of the major exporters, only the United States (since 1974) has allowed domestic and world market forces to freely interact, although the government greatly affects both markets through its grain policies. Table I-4 also indicates that 98 percent of wheat trade between these countries involves a state trading agency on one or both sides of the market. At the same time, 95 percent of all US exports was handled by one of six large private multinational firms (Gilmore, 1982). This is the reason for the large percentage of trade in Policy Category 3. Aside from the United States, only Argentina and the EC let the multinationals assume ownership of their grain for export. For rice, no precise data are available to determine the proportion of trade carried out under each policy category. However, Slayton (1984) reports that in 1983, state trading agencies arranged 46 percent of all rice exports (categories 2 and 4), while state importers arranged 60 percent of all rice imports (categories 3 and 4). Falcon and Monke (1979-80) estimate that 93 percent of rice imports and 76 percent of exports were affected by insulation policies. At least 30 20 TABLE I-4 MATRIX 0F ALTERNATIVE TRADING ACTORS AND THEIR PREVALENCE IN THE WORLD WHEAT MARKET, 1982/83-1985/86 Role in World Trade 1 Policy Example Total Environment Exporter International Importer Trade Middleman ‘ l px,mt px,mt,pm mt,pm - O 2 px,sx,mt px,sx,mt,pm mt,pm - 0 px,mt mt mt,pm USA-EC 2 3 .................................................... px,mt mt sm USA-USSR 42 sx mt sm CANADA-JAPAN 4O 4 sx mt,pm mt,pm CANADA-EC 3 px,mt px,mt sm EC-USSR 13 Sources: compiled from data in USDA (1986), International Wheat Council (1987), and Jabara (1982). Key px: private domestic exporting firm/cooperative/miller sx: state exporter mt: multinational trading corporation sm: state importer pm: private domestic importing firm/cooperative/miller Policy Environment Border pricing on both import and export side Border pricing on import side; insulation policies on export side Border pricing on export side; insulation policies on import side Insulation policies on both import and export side AUNH O. O. O. O. 21 percent of total rice trade is coordinated by government-to-government contraCts. The objectives and behavior of the five actors in Table I-4 are now examined in more detail. Stete Trading Agenejes Grain policy almost everywhere is fundamentally driven by the societal perception, in place for several millennia, that government is at least partially responsible for supplying the population with adequate food at affordable prices.4 Since governments base their grain policy decisions on a variety of objectives in addition to efficiency, the type of production, stockholding, and trade patterns suggested by domestic resource cost/comparative advantage analyses may be incompatible with existing political and social realities, regardless of their potential long-run benefits. Theory and empirical evidence indicate that many wheat and rice producing countries have stabilized domestic prices through price and trade policies that sever the link between domestic and world prices (Zwart and Meilke, 1979; Falcon and Monke, 1979-80; Bigman, 1985). The removal of such policies may promote international price stability at the expense of domestic price instability, unless large and expensive grain stocks are held (Zwart and Meilke, 1979; Paarlberg, 1988). In this regard, it may be 4 This conception was clearly in place by the time of ancient Rome, where “bread and circuses' were a common response of Roman emporers to civil unrest. Even before this, the interrelationship between politics and food was evident from Socrates’ caution that ''no man qualifies as a statesman who is entirely ignorant of the problems of wheat" (in Morgan, 1979). A 22 difficult to stabilize world markets through GATT-type negotiations whereby governments forfeit the right to use trade and price policies to stabilize domestic markets (Jabara, 1982).5 Recent food riots in the Sudan, Tunesia, Poland, Turkey and Morocco serve as constant reminders to other governments that they may no longer be the government if food prices exceed affordable levels. A list of countries relying primarily on state marketing agencies to fulfill the role of importer and exporter is provided in Table 1-5. The objectives and behavior of these state actors may differ according to the country’s position in the market. For example, Japan is a consistent importer; reliability and continuity of supply are important objectives. Therefore, it is not suprising that the Japan Food Agency traditionally locks-in about half of its wheat imports through contracts with exporters. India, conversely, enters the market sporadically, usually in response to domestic production shortfalls or surpluses. Countries in this situation rarely rely on long-run contracts, preferring other modes of coordination that do not commit the state agency to a specified volume of imports before the domestic production situation is known. McCalla (1979) hypothesizes that the objective of India’s state trading agency is to minimize the cost of domestic grain shortfalls. 5Others, of course, disagree. D.G. Johnson, for example, contends that 'a low cost means of increasing price stability in international markets is by liberalizing trade'l (p. 741, 1984). It is unclear, however, whether many governments would consider increased domestic price instability a "low cost' to pay. 23 TABLE 1-5 STATE TRADING AND OTHER FORMS OF ADMINISTRATIVE COORDINATION IN WORLD WHEAT AND RICE MARKETS State Tariffs, Quotas, Variable Trading Taxes, Restrictive Levies Agencies Subsidies Licensing Exporters: Argentina Australiac Canada EC x x Pakistand Thailand Burma United States x XX” XU'X Importers: Soviet Union China Egypt Japan Brazil S. Korea Nigeria Iran Iraq Senegald Indonesia XXXXXX X0 XI‘DXX (D a: The National Grains Board purchases small quantities of grain at minimum prices. It is active in trade only in the case of government-to-government sales. b: The majority of rice trade is through private exporters. Exporters are required to register with recognized trade associations and comply with Department of Foreign Trade regulations. C: Wheat only d: Rice only e: Parastatal monopoly on rice imports abolished effective December 86. Sources: Jabara (1982) and USDA (1986). 24 lt‘ i n ' ' i The United States exports about 40 percent of the world’s traded wheat; of this, about 95 percent is handled by the largest Six multinational grain traders (MTs: Cargill, Continental, Louis Dreyfus, Bunge, Mitsui/Cook, and Garnac). In addition, the HTS deal heavily with state trading agencies such as the Canadian and Australian Wheat Boards. As mentioned above, this is sometimes confined to the execution of logistical functions after state importers and exporters have negotiated a deal. However, the MTS often buy wheat directly from the wheat boards without having made prior sales arrangements. In this capacity as speculator or arbitrager, the HTS may stabilize prices both temporally and Spatially, assuming that they act competitively. It may be hypothesized that the objectives of the major MTS include increasing market volume, market share, and profits. The market environment that facilitates these objectives is more likely one of price instability rather than stability. Profits of the large private grain traders seem to move over time not with total volume traded, but rather with the degree of disruption and instability in the year’s trading patterns (Caves, 1979; Gilmore, 1982). Such an environment enables multinationals to put their superior market intelligence to work and capture the rents resulting from market disturbances. However, large volumes are not unimportant; they promote size economies in two ways. First, the MTs make only a few cents on each ton of grain handled; therefore, profits in the industry require large volumes. Second, the MTs’ potential to maximize their objectives depends on having large, diverse positions to draw from quickly to 25 exploit temporary arbitrage opportunities (Peterson, 1988). For these reasons, the grain traders appear to resist programs that reduce traded volume, such as U.S. acreage reduction programs. The six largest MTs have the following general characteristics: they are diversified, have vast global market intelligence systems, and cross-subsidize operations (Schmitz et al., 1981). In addition, all of the major MTs have a maze of affiliates and subsidiaries. Inter-affiliate sales, which appear to account for a large portion of MT sales (40 percent for Cargill, Inc. from 1970-75), provide several informational advantages. For example, the Commodity Futures Trading Commission reporting rules do not require foreign firms to fully identify or report their futures positions, as they do of U.S. firms (Gilmore, 1982). This allows a MT with foreign subsidiaries to temporarily conceal important market information.6 Also, because importing subsidiaries generally serve as agents and do not take legal possession of the grain, neither the ultimate purchaser’s identity nor the final destination of the shipment need be declared. This provides a camouflage for trade flows and conceals the identity of the ultimate buyer. While this may not greatly affect price instability ex post, it certainly introduces uncertainty into the market. In 1974 for example, President Ford, concerned that the USSR might destabilize the wheat market by buying on a scale comparable to 1972, invoked a moratorium 5 Export sales reports are released to the public by the USDA with a lag of 7 to 14 days after an MT grain sale occurs. Studies by Conklin (1982) and the GAO (1985) have concluded that '...the full adjustment of prices to export sales information does not occur until the sales report is released...‘' (GAO, p. vii). Therefore, a large trader may possess information that the market as a whole does not learn about until one to two weeks later. 26 because it could not be established with certainty what Soviet buying intentions or existing contracts were, since foreign subsidiaries served as intermediaries (Gilmore, 1982). MTs may have a cost-procurement advantage over state exporters or farmer cooperatives in the international marketing of grain since they can buy grain from many sources of supply to meet export commitments (multiple sourcing). For example, an MT can buy grain not only from U.S. or Argentine exporters but also from the Canadian or Australian Wheat Boards. The MTs generally do not operate on the basis of individual sales, but rather on net positions, reflecting their expectations of future market conditions (Peterson, 1988). They are constantly acquiring physical supplies of grains and making sales, thus the operation resembles a pipeline (Conklin, 1982). This has given rise to the use of "optional-origin contracts,’' which allow the seller to contract with an importer without specifying the grain’s origin. The MT may also procure grain from various sources in excess of actual requirements, and then cancel the less attractive contracts before the time of final sale. This allows the trader to juggle different purchases in several countries to fill an identical order (Gilmore, 1982). Peterson (1988) states that the flexibility of these contracts often allows the MT to pass along cost-savings to the foreign buyer. Aside from price-hedging opportunities, optional-origin contracts protect the MT and buyer from restrictive national agricultural policies such as embargoes, which might otherwise cut the foreign buyer off from a supply of grain already under contract. 27 Schmitz et al. (1981) argue that "when the large companies shop around for the best deal themselves, their actions do not necessarily always benefit the producers in the country in which the parent company is located (or in any particular country for that matter)" (p. 41). However, such behavior would tend to improve market integration, and bring lower prices to importers, as long as the industry is competitive. If it is not, the MT may extract rents from both producers and importing consumers, especially if export and import elasticities are low. Emerging evidence suggests that world grain market instability creates higher marketing margins between f.o.b and c.i.f. ports (Binkley, 1983). The major functions performed by the HTS-— transport, storage, processing, and handling--typically involve high fixed investments. This suggests that adjustments to varying trade volumes are made with difficulty and higher per unit costs. Seme sources contend that this has created second-round effects on market structure. Bulk Systems International (1980) reports that because of the “erratic nature of the grain trade," the development of efficient, low-cost port facilities "has proceeded at a far Slower pace than with other large volume major bulk commodities." Moreover, large transport ivessels, which provide lower operating costs, are less prevalent in the grain trade than in commodities such as coal and iron ore, whose trade volumes are more stable than grain (Binkley, 1983). 28 0 've While four regional grain cooperatives play a prominent role as exporter in the U.S. wheat market, they rarely function as international middlemen. Inland cooperatives account for roughly 40 percent of off-farm grain sales, but as grain moves through the marketing channel into export position, it is increasingly controlled by the large multinational grain traders. The cooperatives control less than 20 percent of the grain moved to U.S. export ports, and account for only 7.5 percent of U.S. sales to foreign buyers. Thus, although cooperatives are active near the beginning of the marketing chain, they turn over about 70 percent of their potential export volume to other grain traders (Schmitz et al., 1981). While Farmers Export, the export arm of 12 large farm cooperatives, was originally designed to reverse this trend, it is now defunct due to mismanagement and conflict of interest by member co-ops that competed with it (Rowan, 1981). In countries dominated by state exporting agencies, such as Canada and Australia, cooperatives have an even smaller role in the international marketing of wheat. The situation with rice is somewhat different. Several large U.S. milling cooperatives procure rough rice from their producer members, mill it, and retain a relatively high proportion of the milled product for direct sale to importers. The remaining export rice is usually sold to international middlemen. The reasons why farmer cooperatives do not play a large role in the international marketing of wheat are numerous. Thurston et al. (1976) provide the following reasons: (1) lack of experience and expertise; 29 (2) lack of access to cooperatively-owned export facilities; (3) less risk in indirect sales to multinationals; (4) economies of size favoring the large multinational traders; (5) lack of multiple sourcing and commodity diversification; (6) relatively inferior market intelligence; and (7) fear of the unknown. Cooperatives’ small role in international grain marketing is in part due to the rapid change in the structure of the U.S. grain marketing system. Traditionally, when the bulk of harvested grain was consumed domestically, the cooperatives were geared toward handling, distribution, and marketing at the local level. Over the past thirty years, however, the end use composition of US grain has changed Significantly. U.S. wheat and rice exports have risen dramatically, especially Since the early 19705. Cooperatives do not appear to have positioned themselves well for the increased internationalization of grain marketing (Gilmore, 1982). Major risks, uncertainties, and information costs appear to be better absorbed by the size of the large multinational traders. For example, Caves (1977-78) argues that the characteristics of information as an input create scale economies in coordination and risk-bearing. Information procurement has a fixed cost that can be spread over numerous transactions, and information about trading locations is subject to increasing returns in the trading possibilities that it reveals. There are also economies of scale in storage and transshipment facilities at particular locations, and in pooling the risks associated with price and exchange rate variability, uncertain government policies, and legal idiosyncracies of importing countries. Large, diversified investment portfolios allow the larger firms to 30 appreciably reduce the variance of returns. Moreover, the perishability of information creates scale economies in the maintenance of a continuous and extensive trading presence to exploit transitory trading opportunities (Caves, 1977-78). Not suprisingly, therefore, the ratio of multinational to cooperative grain volume traded is larger in international trade--where information and risk-bearing costs are high-~than in intra-national grain trade. Finally, most cooperatives are national organizations and cannot draw supplies from different countries to capture the cheapest possible source. Even domestically, the regional cooperatives view their function as selling their members’ grain rather than exploiting profitable trading opportunities wherever they arise (Caves, 1977-78). A notable exception has been the Farmers Export Company, which had speculated aggressively and extensively on grain markets during the early 19805. It unfortunately incurred sizable financial losses in the process (Rowan, 1981). Gilmore (1982) notes the potential of cooperatives to overcome these traditional barriers to participation in international grain markets by combining operations with multinational- type firms. The joint venture between Toepfer and European and American cooperatives has provided the latter with an assured market outlet, expanded export marketing capacity, a valuable market information network, and overseas administrative and marketing infrastructure with which to conduct trade. In summary, a diverse set of actors are involved in the movement of grain within domestic boundaries, from farm to export ports, and from import ports to consumers. Whether these are primarily public or 31 private actors varies by country. However, the international exchange of grain is dominated by state trading agencies and multinational trading firms. The physical transfer of wheat (as opposed to ownership) is almost always carried out by a small group of large private traders. The degree of market power exercised by this group is hotly contested (Conklin, 1982; GAO, 1985; Caves, 1977-78; Schmitz, 1986; Gilmore, 1982; Morgan, 1979). While numerous studies have pointed out the theoretically important effects of market power on instability and trade patterns (Binkley, 1983; Schmitz et al., 1981; McCalla, 1979), relatively little information is available to draw reliable conclusions concerning whether market power indeed exists.7 2.3 COORDINATION MECHANISMS AND MARKETING INSTITUTIONS The relationship between a marketing firm’s inputs and outputs and more generally the productivity and stability of the international grain system, is a function of the institutional organization of the system. This section examines the methods of coordination between actors in the international wheat and rice markets, and how risks and uncertainties in the system are shaped by alternative methods of coordination. 7 However, the market power hypothesis must explain several questions, such as why the c.i.f. Rotterdam price of U.S. wheat over the past three decades has averaged only about 10 cents per ton higher than the f.o.b. U.S. Gulf price, instead of the oligopoly outcome, which would be that grain traders raise the c.i.f. price to the EC threshold price and capture the variable levy. 32 o ' i n International grain transactions between importers and exporters generally takes place in one of three types of markets: the open market, the public tender market, and the private tender market (Schmitz et al., 1981). Rotterdam, Hamburg, and London are examples of open markets, in which there is a continuous two-way flow of offers and bids, often through brokers. As mentioned previously, brokerage fees appear much higher in rice trade than for wheat, because of the greater search costs involved in matching sellers with buyers wanting particular varieties and qualities that are poorly standardized. In public tender markets, the buyer, often a state importer, issues tenders ahead of the requested offering date. Private tenders are less formal: the buyer, either a public or private actor, invites a few selected exporters to make offers. The terms of grain export sales are specified in legal contracts that include the quantity, quality, shipping period, destination, delivery terms (generally f.o.b. or c.i.f.), payment terms, and price. In both wheat and rice contracts, price is almost always denominated in U.S. dollars, no matter whether the United States is involved in the deal or not (Peterson, 1988). Price is usually determined by flat price or basis price contracts. Flat pricing fixes the price at which the future delivery will take place. Buyers under flat pricing are often state trading agencies that are more concerned with locking in a supply of grain at a known price, and less concerned with potentially paying a higher price than would be necessary, as would be the case if cash prices fell prior to delivery. Basis pricing specifies some 33 agreed upon amount over a particular futures price, and allows the importer to choose, sometime prior to the delivery date, the date for the calculation of the relevant futures price. Basis pricing is used by more sophisticated private and state buyers, such as the Japan Food Agency (Harrison, 1988). However, basis pricing requires reference to a futures price. Because rice futures trading is poorly developed, very little rice trade is basis-priced. Other types of formula pricing are possible based on a spot price modified by inflation and exchange rate movements. Ver ' l n rati Vertical integration is pervasive in the international wheat and rice marketing systems. Cargill, for example, is a principle buyer of grain from farmers, a 'fobber' (i.e., a grain handler responsible for moving the crop from production region to port), an importer and exporter, a processor, a livestock and poultry farmer, a transporter, a speculator, hedger and Futures Commission Merchant on the commodity exchanges, and a borrower and lender of commercial credit (Gilmore, 1982). Vertical integration in the international marketing of grain has developed, at least in part, to reduce the potentially great transactions costs and uncertainties of trading across national boundaries with different currencies, languages, Units of measure, commercial codes of ethics, etc. (Peterson, 1988). While vertical integration is not unique to the multinational traders (cooperatives and state traders also combine vertical stages within their operations), MT vertical integration is of a unique sort. 34 While MTs have operations at each stage in the marketing chain, each stage tends to pursue its most advantageous opportuhities whether or not that entails internal coordination within the firm’s various stages (Peterson, 1988). For example, an MT’s shipping fleet may find it more profitable at a particular point in time to haul iron ore or some other commodity than transport its own wheat to fulfill a contract. This is quite different from most cooperatives or state traders, which, as stated above, tend to provide services mainly for other stages of their firms’ operations. Leng Run Agreements Grain contracts lasting for more than a year in duration are categorized as long term. They are often of an ongoing nature, such as the multi-year wheat trade understandings between government agencies of the U.S. and those of the USSR, Poland, and Israel. Perhaps suprisingly, long run agreements are relatively more important in the world wheat market than in the rice market. Long run wheat agreements comprised about 25 percent of total wheat trade in the 19805 (Table I- 6). By contrast, only 6-10 percent of the rice trade was coordinated through long run contracts (Slayton, 1984). The rationale behind long-run agreements is clear. From the perspective of grain-dependent importers, a continuous pipeline of wheat or rice, especially during tight world market conditions, may be crucial for political stability. Governments in this positions prefer a secure, long-term source of supply from reliable exporters rather 35 TABLE 1-6 Long Run Bilateral Wheat Contracts, 1981-87 1981 1983 1984 1986 1987 1981 1983 1984 1986 1987 million metric tons % of total country trade Importers: Soviet Union 8.5 6.5 8.2 8.0 11.3 43 32 30 52 54 China 13.4 12.4 9.3 -- -- 97 95 97 0 0 Japan 2.2 2.2 2.2 2.1 2.1 41 39 39 39 42 Egypt 1.0 1.0 1.0 2.8 2.5 17 17 16 44 35 Brazil 0.5 4.0 1.2 1.3 2.1 13 100 29 59 75 Iraq 0.3 1.3 1.7 1.2 1.4 23 45 61 60 70 Exporters: Canada 9.9 14.7 11.9 8.0 9.4 54 68 68 39 40 Argentina 1.3 1.4 1.9 1.7 4.3 37 18 20 38 77 Australia 4.5 5.4 6.2 4.4 4.7 41 41 42 28 45 United States 12.1 8.0 8.0 4.5 4.0 25 21 21 17 9 ----- % of world trade----« World 27.7 29.5 28.1 18.6 23.7 26 27 25 18 21 Source: Computed from International Wheat Council (various issues). Note : Figures denote approximate volumes specified by bilateral agreements, not necessarily actual volumes transferred. than taking advantage of low prices offered in particular years by non— traditional exporters (Tolley et al., 1984). Many exporters, likewise, prefer to lock in stable trade volumes, which aid in the orderly structuring of domestic supply and price targets. Predictable import demand and price stability were a major imputus behind the U.S.- Soviet Union grain agreements Since 1976. Prior to that time, large but sporadic entry of the Soviet Union into the market caused considerable price instability, since its import plans were not 36 revealed early enough to permit adjustment of production plans in exporting countries. These concerns highlight the fact that price, while important, is not the sole factor that links grain buyers with sellers. Other factors such as security and stability reduce the importance of price in the coordination task. The pricing arrangement in long-term contracts is often incidental to the contract; price is usually fixed according to prevailing spot market prices at the date of delivery. It is important to note that although long run contracting bypasses the open market in the allocation of grain, it nonetheless requires the existence of the open market to determine the pricing terms of the contract. Although long-term agreements represent an institutional innovation by market actors to reduce their vulnerability to instability, uncertainty, and transactions costs, such agreements create side effects that increase uncertainty for others. As long-run contracting and conventional vertical integration become more important, the volume traded on the open market invariably declines, making it less able to absorb quantity variations in world trade. Consequently, the open market becomes more vulnerable to wide price swings caused by sporadic market entrants. This is often referred to as a thin market problem. If the open markets have been sufficiently thinned, the potential arises for poor resource allocation and volatile price movements that do not reflect underlying supply and demand conditions. Unfortunately, this would not only affect price discovery on the open markets, but all other forms of coordination that depend on open markets for price 37 determination. This may be a problem particularly in the rice market because traded volume is low to begin with (only 4 percent of world rice production is traded internationally), grades are poorly standardized, prices in various markets and for various varieties may move independently of one another in the short-run, and because of the unstable trade volumes of several large rice traders. In addition, while long-run contracts facilitate the security goals of consistent importing and exporting countries, they increase risks and transaction costs borne by others. Countries with occasional rice surpluses or deficits frequently have difficulty breaking into the market because major import and export centers are already locked into bilateral relationships. This raises search costs and uncertainty for these actors, and provides incentives for them to develop alternative forms of coordination that thin the market even further (NC-117, 1979). See; end thgres Markets and Eriee Diseeveny Price discovery is the process through which buyers and sellers "discover’ the competitive prices that best represent the consensus of what traders think commodity prices ought to be in the future based on information available today. Broad dissemination and publication of exchange-generated prices can foster competition in establishing cash prices for commodities in localized markets as well as in related services such as storage, trasportation and processing“ (GAO, 1985, p. 5). The reliability of the price actually "discovered" may be 38 sensitive to the legal, informational, and broader institutional organization of the market. A critical distinction between the organization of the rice and wheat market is that there are no reliable, widely used spot or futures markets for price discovery for the various types of rice. Although the f.o.b. Bangkok rice price quotes are often referred to as the world price for rice, it is important to note that these are posted prices set by the Thai Board of Trade, and are supposed to be the minimum price that private exporters must demand from their buyers, otherwise no export license will be issued (Siamwalla and Haykin, 1983). Actual f.o.b. prices may diverge substantially from these weekly-posted prices, as exporters will illicitly raise or lower prices depending on prevailing supply and demand conditions. Therefore, the Bangkok posted prices may be of limited value for price discovery. On an annual basis, however, they are more representative of world supply and demand conditions (Falcon and Monke, 1979-80). In the other major rice exporting center, the United States, price discovery is also hampered by Spot and futures market deficiencies. Reported spot prices are less specific than those for wheat: there is no daily or even weekly rough rice spot market price for a specific variety and grade (Gordon, 1984). Weekly cash prices are reported for milled rice, but they are more indicative of private treaty arrangements dominated by several large rice millers on the buying side rather than spot prices discovered in a vibrant competitive market. In addition, these prices are usually expressed as a range, and often may not vary for several weeks at a time. In short, U.S. rice price 39 determination is far from transparent, occurs mainly at the mill level, and, for several submarkets, is dominated by several large mills and exporters that suppress or distort price information (Business Week, 1978; Caves, 1979). These spot market characteristics have also affected the viability of rice futures markets. For example, because there are no published spot market prices for a specific deliverable variety and grade of rough rice, no 'basis' (i.e. the difference between cash and futures price) can be determined on which to negotiate contracts.. Basis analysis of the milled rice market (Gordon, 1984) indicated that the basis often moved in the wrong direction, that the correlation between cash and basis prices were relatively low compared with more heavily traded crops, and that rice futures prices were often biased predictors of their delivery month values. These attributes reduce the ability of farmers to hedge rice on the futures market, and, not suprisingly, the milled rice futures market on the New Orleans Commodity Exchange closed in 1984. . By contrast, the U.S. wheat futures markets on the Chicago Board of Trade is used extensively both by U.S. and foreign actors, including a number of state traders. U.S. spot prices can be quoted daily for a deliverable variety and grade. It can be argued that the U.S. spot and futures markets form the central pricing point in the international wheat markets. In this sense, the information generated from these futures market are public goods; the additional use of the information, once discovered, adds little or nothing to cost, nor does added use interfere with its use by others (Schrader et al., 1986). For example, 40 Canada and Australia peg their minimum foreign sales prices to the Chicago Wheat futures price. The EC variable levy Could not function without a spot or futures price discovery process. In this way, those who do not participate in open market arrangements can benefit from the information generated from them without having to pay the costs. Risk shifting is performed through hedging on the futures market, where actors who are unwilling to bear price risk can shift risk to those who are willing to assume it in return for potential profit.8 However, the risk Shifting potential of the futures market is primarily short-run in nature. Futures price quotes do not normally extend more than a year into the future, and thus the market cannot provide safeguards against severe inter-annual instability. Many investments (storage terminals, production equipment, ships, etc.) span across many years, and the potential for low or negative returns cannot be altered by the short-run hedging potential of futures markets. However, the intra-annual risk management afforded by futures markets is a very important asset to the wheat market. Timely information is central to efficient marketing. The absence of viable Spot and futures markets for rice to synchronize prices in other sub-markets has probably hampered allocational efficiency as well as making other forms of coordination (that rely on spot price information to establish a transaction price, such as contracting) more difficult, and subject to higher transactions costs. Traders are also exposed to larger profit/loss variability, which increases trading and 8 There is a longstanding debate, however, whether futures markets can provide efficient price discovery and insurance functions at the same time (see Houthakker, 1959; Dusak, 1973). 41 storage risks, risk premia, and marketing margins, unless other viable contingency markets exist. The accompanying paper has documented the much higher marketing margins between f.o.b and c.i.f. ports in rice trade relative to wheat, which do not appear explainable on the basis of tranSport costs alone. Transaction costs are higher in rice markets than in wheat markets because of need to search for buyers, sellers, and reliable price information in the absence of a centralized market (Siamwalla and Haykin, 1983). Brokerage fees appear to be about 5 to 10 percent of sales value, which is much higher than for wheat. The Chicago wheat futures market may give importers like the USSR and China an advantage over exporters, since it probably reflects information about current and impending market conditions in the U.S., Canada, and other open economies more so than in closed economies. For example, to the extent that information on the USSR’S wheat harvest or policy-induced demand changes is suppressed, it cannot be reflected in futures prices. This would give the USSR an information advantage in international grain markets.9 The Chicago Mercantile Exchange has played an increasing role in international grain trade Since the transition to a flexible exchange rate system in 1973. Exchange rate risk creates the need to engage in currency arbitrage on the CME, otherwise the dollar gain from a contract or hedge involving foreign currency could evaporate into losses if the dollar depreciated over the duration of the contract. 9 This argument is weakened by the fact that USDA estimates of Russian wheat harvests have often been more accurate than Soviet estimates (Shaffer, 1989). 42 While MTs use currency arbitrage extensively to guard against exchange rate risk, most farmers do not. But to the extent that there is adequate competition in grain marketing, farmers should benefit from MTs’ use of currency futures markets. 2.4 NATIONAL TRADE AND STOCKHOLDING PATTERNS Trade patterns in the rice market have been rapidly changing since the 19605. Of the five major rice importers in the 19605, only Indonesia remains among this set in the 19805 (Table 1-7). Major importing centers have shifted from south and east Asia to the Mideast and Africa. Prior to World War 11, Asia accounted for 93 percent of rice exports and 75 percent of imports. This dropped to 60 and 39 percent by 1980 (Barker, Herdt, and Rose, 1985). This is because of successful self-sufficiency drives in Asia after the experience of the early 19705, changing tastes and preferences in Africa, and the stimulus of oil revenue in the Mideast on rice imports. Four of the five largest rice importers are now OPEC nations, suggesting a possible relationship between the price of oil and rice demand. There has been more continuity among rice exporters: Thailand, the United States, Pakistan, the PRC, and Burma have been the five major exporters since the 19605. However, the rapid change in rice import patterns indicate a lack of deeply entrenched, durable trade relationships among the major actors. By contrast, wheat trade patterns have remained relatively stable over the past three decades (Table 1-7). Over this period, 85 percent 43 TABLE 1-7 The Five Largest Importers and Exporters in the World Wheat and Rice Markets: 1960-87 1960-71 1972-79 1980-88 Wheat Exporters USA (19117) USA (31131) USA (37716) Canada (11485) Canada (13560) Canada (19260) Australia (6937) Australia (8876) France (15030) France (3345) France (8328) Australia (12084) Argentina (2685) Argentina (3217) Argentina (5882) Wheat Importers India (4563) China (5942) USSR (18655) China (4353) USSR (5801) China (11176) Japan (3752) Japan (5546) Egypt (6133) Brazil (2164) Egypt (3970) Japan (5360) Egypt (2119) Brazil (3285) Brazil (3242) Rice Exporters Thailand (1490) USA (2090) Thailand (3846) USA (1440) Thailand (1822) USA (2328) China (1068) China (1437) Pakistan (1106) Burma (1030) Pakistan (790) China (622) Pakistan (132) Burma (458) Burma (569) Rice Importers Indonesia (716) Indonesia (1398) Iran (642) Sri Lanka (471) Vietnam (475) Iraq (501) India (444) Iran (354) Indonesia (487) Vietnam (411) Sri Lanka (330) Nigeria (467) Malaysia (366) Bangladesh (299) Senegal (340) Numbers in parentheses are trade volumes in thousand metric tons. USDA, 1988. Source: 44 of the world’s wheat exports have consistently been supplied by the United States, Canada, Australia, France and Argentina. 0f the five largest wheat importers in the 19605, four of them (the PRC, Egypt, Japan, and Brazil) remain among this set in the 19805. The most important change in wheat trading patterns has been the transition of the Soviet Union to a major importer during the 19705. This reflected a major shift in Soviet policy around 1970 to stimulate domestic livestock production and consumption, which consequently caused large surges in wheat and coarse grain imports. The central plan economies, namely China and the USSR, now account for 30 percent of the world’s wheat imports. The Mideast and North African countries, while individually importing only moderate amounts, as a group now account for about 20 percent of all wheat imports. Stoek Levels A major distinction between the world rice and wheat markets is the amount of stocks available to stabilize the market in response to production or demand Shocks. From 1980-87, the average ratio of worldstocks to world production was 29 percent in the wheat market compared with only 15 in the rice market. The major rice exporters especially hold much smaller stock levels relative to production than in the wheat market. High short run instability and the absence of viable risk-shifting institutions such as futures markets may partially account for the low level of stockholding in the rice market. Causality may not be unidirectional however. Larger stocks, accessible for trading in the world market without being subject to government 45 restrictions, could reduce the responsiveness of price to crop forecasts. For instance, if stocks are low, impending production shortfalls in key areas may cause prices to be bid up higher in order to ration available supplies. But if stocks are large and "overhang the market”, then world production variations may produce less price movement, because actors know that there are large stocks available to dampen any price rise (Stein and Smith, 1977). . The dominant wheat stockpilers during the 19805 have been the United States, China, the Soviet Union, and the EC-12 (Table 1-8). In the rice market, China has held over half the world’s rice stocks Since 1980. In average terms, this has been twice the volume of annual world rice trade, leaving China in a strong position to either stabilize or destabilize the market. However, given the budgetary costs of shielding such a huge domestic market from world price fluctuations (over a third of the world’s rice production is consumed in China), the government would apparently have powerful incentives to stabilize the world market. Sh -Run rade Vo m n tabilit The international wheat and rice markets are not very large relative to total world consumption or even compared to the domestic markets of several large actors.lo Since 1980, the average annual volume traded in the wheat and rice markets is 21 and 4 percent of 1° Despite a 70 percent increase since the mid-19605, annual wheat trade is still smaller than the combined production of the United States and India. Annual rice trade has grown by 50 percent over the same period, but it is smaller than annual production in Bangladesh. 45 TABLE 1-8 ANNUAL AVERAGE STOCKHOLDINGS OF WHEAT AND RICE: 1980-87 WHEAT RICE (000 mt) x of world (000 mt) % of world stocks stocks United States 36,516 25.7 1,561 3.3 China 29,366 20.7 25,466 54.3 Soviet Union 16,712 11.8 na na EC-IZ 13,094 9.2 -- -- India 10,388 7.3 6,722 14.3 Canada 9,087 6.4 na na Indonesia na na 1,990 4.2 Japan -- -- 1,699 3.6 Australia 4,590 3.2 -- -- Turkey 4,444 3.1 -- -- x of world % of world production production WORLD 141,848 28.9 45,885 15.4 Source: USDA, I988 na: information not available less than 3 percent; all other countries holding less than 3 percent of world stocks on average are not reported. 47 world consumption. Because international rice trade is a relatively small fraction of production, small percentage changes in domestic production or consumption can translate into large perCentage changes in trade volumes and world prices. Regardless of whether government insulation policies contribute to the problem, a high level of instability in the world rice market can be deduced without reference to them. In fact, the trade volumes of major actors in the world rice market have been much more unstable than those in the wheat market, as measured by the coefficient of variation of country net export regressions.ll This is true when comparing instability between wheat and rice exporters, and especially when comparing wheat and rice importers. However, these results are not good indicators of the relative level of uncertainty and risk in these markets. For example, the implications for world price instability would be very different if net export fluctuations of major actors were caused by changes in domestic production rather than by changes in stock releases onto the world market reacting to world price movements. It is often stated that government insulation policies which treat the world market as a residual market have been a major cause of fluctuations in net exports and hence world price instability. The following chapter assesses the validity of these claims. 11 This is measured as STD(nx) / nx, or, the standard deviation of the dependent variable (computed by regressing net exports on a linear time trend and taking the standard deviation of the residuals), divided by the mean net export level). CHAPTER 3 GOVERNMENT INSULATION POLICIES AND WORLD MARKET INSTABILITY 3.1 THEORETICAL RATIONALE A plethora of recent studies have concluded that mounting government insulation in agriculture is the major factor responsible for the growing instability in world prices (Johnson, 1975; Shei and Thompson, 1977; Zwart and Meilke, 1979; Falcon and Monke, 1979-80; Sarris and Freebairn, 1983; Blandford and Schwartz, 1983; Bigman, 1985). The basic theoretical rationale behind this argument is presented in a two-region partial-equilibrium framework. Figure I-1a shows domestic supply and demand curves for a commodity exporter (Sxo, on) and an importer (Smo: Dmo)- The middle panel Shows the resulting import demand and export supply curves under free trade, which are the sum of the domestic supply and demand slopes. Supply in both countries is determined by domestic price (which is equal to world price under free trade), weather, and other factors. Now let the importing government introduce a producer price stabilization scheme (such as a variable levy or price support program), while allowing consumer prices to still be determined by world conditions. The domestic supply function (5"1) becomes insensitive to world prices and is now affected by the policy- influenced domestic price (Figure I-lb). As domestic supply becomes vertical relative to world price, the import demand function rotates to 48 49 Figure 1a PW P" P" 0x0 5X0 . -_-_fs.°..----. 0 Figure lb PvM\ PW PW \ l 5141 MDMI \ \l M‘ / \ L I \\ I00 102 ID} 0 Figure 1c PW PWM 0 FIGURE I-I-1.Theoretical Effect of Government Insulation Policy on World Price Instability 50 ID}. If both the producer and consumer price were fixed, then both the domestic supply and demand curves would be vertical (5M1: 0M1), and the import demand curve facing the exporter would be completely inelastic (102). The importing government has made its import volume completely unrelated to the world price, implying a world price elasticity of import demand equal to zero. Now introduce a weather-induced supply shock in the exporting country (Figure I-Ic). Domestic supply is reduced to 5x14 which reduces the supply of exports to E51. The supply shortfall in the exporting country raises world price to'PWz. In the absence of the price stabilization scheme in the importing country, a comparable supply shortfall would have raised the world price only to PWl. This is because quantity supplied and demanded in the importing country would have been allowed to adjust to the changed world market conditions rather than being totally insulated from them. The more inelastic the import demand curves and export supply curves, the greater are changes in world price resulting from shocks in domestic markets. This is the general basis for the argument that government insulation policies increase world price instability. 3.2 PAST STUDIES A number of empirical studies have found support for the above theoretical framework. The general line of inquiry is to measure how national production and world price variations affect annual trade flows. Blandford (1983) defines these interactions between national and world markets as follows: 51 1. Transmission effect: ”the degree to which short-term variability in supply and demand in a given national market is passed on to the world market through variations in the country’s volume of trade.“ 2. Absorption effect: ”the degree to which short-term variability originating in other countries, and reflected in the variation in world prices, is absorbed through quantitative adjustment in the national market and its volume of trade" (p. 383- -4). Blandford estimated these two effects on a national/regional basis for wheat and coarse grain using the following equation: (NM - NM*)t - a0 + a1(PROD - PROO*)t + a2(PW - PW*)t where NM is net imports of country i, PROD is domestic production of country i, PW is world price, and NM*, PROD*, and PW* are linear trend values of these variables. The extent to which domestic production instability is transmitted to the world market is measured by al. This transmission coefficient would have a value of -1 if all production instability were transmitted to the world market and a value of zero if none were. The absorption coefficient, a2, measures the effect of world price deviations from trend on a country’s net imports. A negative value would indicate a normal type of demand relationship, which would be stabilizing; the more negative (elastic) the response, the more the country absorbs world market instability. A zero value would indicate complete insulation with no import response to world price changes. This is considered destabilizing because a lack of trade adjustment forces prices to adjust comparatively more in order to reach an equilibrium. 52 Similar variants of this model have been used by Abbott (1979), Wilde et al. (1985), and Mitchell and Duncan (1987) to examine the wheat market, and by Siamwalla and Haykin (1983) and Falcon and Monke (1979-80) to examine the rice market. By and large, these studies have found that the absorption effect for most countries is quite low and often statistically insignificant. Only one country/region in Blandford’s wheat study, the United States, showed a significant relationship between net exports and world price at the 10 percent level. Siamwalla and Haykin note that less than one-third of the 55 countries examined showed any trade responsiveness to world rice price. These results have often been used as evidence of government insulation policies that prevent trade flows from responding to world price, and that therefore represent a major cause of the growing instability in world prices (Bigman, 1985; Falcon and Monke, 1979-80; Blandford and Schwartz, 1983). Reserving until later a discussion of the logic behind these conclusions, let us examine why econometric estimation of import demand and export supply elasticities may be so low or statistically insignificant. First, the Blandford, Siamwalla and Haykin, and Falcon and Monke studies estimated net export responsiveness to world prices for all countries using the nominal US dollar value of wheat and rice (No. 2 dark northern spring, f.o.b. U.S. gulf; and 5% broken, f.o.b. Bangkok), without accounting for exchange rate fluctuations between the dollar and a country’s currency, or fer the country’s rate of inflation. According to standard trade theory, a country’s net export responsiveness to world prices would be affected by these variables, 53 and the observed lack of responsiveness to world prices may be partially due to the exclusion of these relevant variables from the models rather than from insulation policies per se. In addition, these studies have estimated net exports as a linear function of world price. Evidence from other sources indicate that trade volumes may be related to prices in a non-linear fashion (Hillman, Johnson and Gray, 1975; Peck and Gray, 1980; Abbott, 1988). For example, import demand may be relatively inelastic when prices are high (due to pipeline demand) and relatively elastic when prices are low (due to increased use of the commodity as a feed grain). Misspecified functional form may in some cases account for the observed lack of trade responsiveness to world prices. A third potential complication concerns matching the annual world price to the relevant marketing period for each particular country. Harvest and marketing calandars vary greatly from country to country. Use of one annual price series, such as an average of January to December monthly prices, in all of the country equations will be out of sync with many countries’ marketing periods and will therefore be inappropriate price data to estimate some countries' trade elasticities. Annual price data generated from monthly data that matches with a particular country’s marketing year is more appropriate. Fourth, net exports in any year are simply annual production minus consumption minus the change in national endstocks. Therefore, a model of trade instability should include those factors that cause instability in domestic consumption, production, and stocks. 54 While it may indeed be true that government insulation policies do not contribute to the stability of world grain markets, these four factors may in part account for the low and/or statistically insignificant trade elasticities observed in world rice and wheat markets. The remainder of this chapter incorporates these considerations into a related model, and examines the implications of government insulation policies on the transmission and absorption of instability onto/from these markets. 3.3 THE MODEL It is identically true that in any given year, (I) NXt - (PRODt) - (coust) - (ENDt - ENDt-1) where NX, PROD, CONS, and END are net exports, domestic production, consumption, and endstocks, in per capita terms. The time relationship between these annual variables is as follows: Plantingt Harvestt Plantingt+1 Harvestt+1 *-——=m_*======== : "' {productiont} A { ------------ marketing yeart ------------------ )B This study regards the relevant annual period as the marketing year, which begins at point A on the time path and ends at point 8. Within marketing yeart occurs consumptiont, stock changest, and net exportst. At point A, or just prior to it, production has already occurred and is hence exogenous. While production is obviously affected by some set of 55 explanatory variables, it is reasonable here to view production as exogenous since the purpose of the analysis is to measure how annual fluctuations in production (which occur before the start of the marketing period and are hence known) and world price affect instability in national trade volumes. Identity (1) highlights the fact that variations in production in each country must be counteracted by an equal adjustment in consumption, stocks, net exports or some combination of the three. The magnitude of the transmission and absorption effect for a particular country depend on how such adjustments are made. Next, national consumption and endstock equations are formulated in per capita terms: (2) CONt - a0 + 81(PDt) + 82(GNPt) (3) ENDt - 50 + b1(PDt) + b2(PD*t+1) + b3(Rt) where PD is domestic price, PD*t+1 is expected domestic price in the next marketing year, GNP is real per capita income, and R is nominal interest rate. Equations (2) and (3) represent standard consumption and stock demand equations, and can be traced to consumer utility maximization and theories of temporal arbitrage. Under the proposition that price expectations are formed adaptively through the process (4) PO*t+1 - P0*t + 0(PDt - PD*t) 56 equation (3) reduces to (3b) ENDt - co + c1(PDt) + c2(PDt-1) + C3(Rt) + C4(Rt-1) + c5(ENDt-1) where co - abo; c1 - b1 + 520; c2 - b1(0-l); C3 - b3; C4 - b3(0-1); c5 - 1-0.12 Domestic and world prices may be linked by a government price determination equation such as (5) PDt - do + d1(PWt) This equation is Similar to "price linkage" equations in Abbott (1979), Zwart and Meilke (1979), Skold and Meyers (1987), and Bolling (1988) that consider domestic price to be some policy-induced function of world price. The parameters do and d1 can be viewed as the net impact of a combination of instruments which are used to establish a domestic price that reflects the pricing objective of the government. The model may represent many alternative types of policies. For example, if do - 0 and d1 - 1, government allows variations in.world price to freely affect domestic prices. Conversely, if 01 - 0, domestic prices are totally insulated from world price movements. However, there are both theoretical and empirical grounds for believing that domestic and world prices may be related in a non-linear fashion (Lattimore, 1974; Skold and Meyers, 1987). For example, 12For derivation, see Appendix 5. 57 governments’ ability to moderate consumer price increases may be weakened as world prices climb progressively higher. Governments’ capacity to affect PD is also influenced by beginning stocks and production. A low level of beginning stocks may impede governments’ ability to maintain prices at desired levels if world prices rise. Production levels may also affect PD, especially where import and export parity prices diverge substantially due to geographical location or poor transport infrastructure. In such cases, local production, beginning Stocks, and government policy determine PD in the range between import and export parity prices. Therefore, a more reasonable variant of equation (5) might be13‘ (5b) PDt - do + d1(PWt) + d2(PW2t) + d3(PRODt) + d4(ENDt-1) Substituting equation (5b) into (2) and (3b) yields (5) COM - a0 + a1[do + 01m.) 4 d2(PW2t) + d3(PRODt) + d4(ENDt-1)1 + azIGNPL) 13 Several potential explanatory variables, such as foreign exchange earnings, are not used here. While it is frequently held that foreign exchange limits the ability of poorer countries to import grain, this argument breaks down in many cases. Adequate grain imports are often so important for political reasons that when foreign exchange earnings drop, governments may reduce imports of other goods (raw materials, spare parts, etc.) to maintain grain imports (Scobie, 1983). As a result, foreign exchange as an explanatory variable cannot avoid measuring the substitutability of grain for other imports. This is largely supported by Wilde et al. (1985), who showed that net wheat imports were sensitive to foreign exchange earnings in only 2 of 19 low- or middle-income countries at the 5 percent level. 58 (7) ENDt - co + c1[do + d1(PWt) + d2(PW2t) + d3(PRODt) + d4(ENDt-1)] + c2[do + d1(PWt-1) + d2(PW2t-1) + d3(PRODt-1) + d4(ENDt-Z)] + c3(Rt) + C4(Rt-l) + c5(ENDt-1) These equations Show that coefficient d1 (i.e., the degree to which government policies allow world price variations to affect domestic price) will influence the responsiveness of consumption and stocks to changes in world price. The closer d1 is to unity, the closer consumption and endstock responsiveness to world price will resemble that under free trade. Moreover, the closer d1 is to unity, the closer the net effect of production changes on consumption and endstocks will be to that under free trade. Substituting equations (6) and (7) into (1) gives the following reduced-form equation: (8) NXt I £0 + fil(PWt) + fl2(PWt-l) + 33(PRODt) + 54(PRODt-1) + 35(GNPt) + fl5(Rt) + 57(Rt_1) + 53(ENDt-1) + 59(EN0t-2) + RIOIPHZt) + 511(PW2t-1) where Bo - - (co + cldo + czdo + aldo + ao) 31 - (€191 + aIdl) £2 - - c291 #3 - 1 - cIda - alda A4 - - Czda 55 - - a2 59 35 ' - c3 37 = - c4 58 - 1 - a1d4 - cs - c1d4 B9 - - c2d4 410- - (Cldz + a1d2) 311- - Czdz Equation (8) is the reduced-form equation to be estimated. It consists of six component parts: H O A net export identity (equation 1) 2 Domestic consumption (equation 2) 3 Domestic endstock demand (equation 3) 4. Adaptive price expectations process (equation 4) 5 Exogenous annual domestic production and beginning stocks 6 World-to-domestic price transmission equation (equation 5b) The reduced-form specification represents the effects of exogenous variables on trade through their influence on consumption and changes in endstocks. Past as well as current changes in production and world price alter the time path of endstocks and hence net exports in a dynamic process. It is clear from the structural definitions that the transmission and absorption effects (i.e. the coefficients on B3 and 81, respectively) are influenced by both domestic price elasticities and the degree to which government policies allow variations in world price to freely affect domestic prices. ~ 60 Equation (8) was estimated for 18 countries and regions in the wheat market, and for 22 countries/regions in the rice market. In each case, a restricted model without the quadratic price terms was also estimated. The non-linearity hypothesis was then subjected to an F- test. PWZt and PWZt-1 are included or excluded from the model based on this criterion. The ”world price“ used for wheat is No. 2, Hard Red Winter, f.o.b. U.S. Gulf; for rice, it is 5‘% broken, milled f.o.b Bangkok.14 The derivation and sources of all data are presented in Table 1-9. Several problems with the model must be considered. The structural parameters in equations (2) and (3b) cannot be recovered from estimation of (8) due to underidentification. However, where domestic price data are available, equation (5b) can be estimated to draw inferences about the degree to which national absorption and transmission effects might change in a transition to free trade. This is explained further in Chapter 3.6. Second, the government policy coefficients in equation (5b) may not be stable. Whenever the underlying policy changes, the coefficients will change as well (Abbott, 1979). The appropriate sample period is therefore a period under which government policy is 14 There may be some question whether the world rice price is better approximated by the Thai price, or the price of a prominant U.S. mill port, due to the potential effect of U.S. rice programs on the world market. However, rice price analysis by Brorsen, Grant and Chavas (1983) indicates that the Thai price appears to have a stronger influence on U.S. prices than the other way around. 61 TABLE [-9 Variable Definition and Data Sources Variable Source NX : per capita net exports (kg); A,K (1960-87) values are negative for importers PROD : per capita production (kg) A,K (1960-87) END : per capita endstocks (kg) A,K (1960-87) pw (wheat) : f.o.b. U.S. Gulf No. 2, Hard Red 8 (1960-62) Winter C (1963-86) 0 (1987) pw (rice) : f.o.b. Bangkok, milled 5% brokens 8 (1960-62) C (1963-86) E (1987) ER : national currency units per U.S. F (1960-84 A dollar, annual average G (1985-87) DFL : national deflator index C (1960-85) (1980-100) G (1986-87) PW : (pw*ER/DFL)*100 derived GNP : real per capita gross national H,K (1960-83) income (1980-100) G,K (1984-87) R : interest rates (3 month discount I (1960-87) window borrowings, U.S. commercial paper) . _ - PD : real domestic retail price of J (1960-82) wheat/rice (domestic currency per mt Sauces: (A) U.S. Department of Agriculture, 1988. Prochction, Stocks, and Disappearance Database. Wity Economics Division, Economic Research Service, USDA, Washington, DC. (9) Food and Agriculture Organization of the lhitsd Nations (various issues). “FAD Bulletin of Monthly Statistics," Rome. (C) International Monetary Fund, (1986). “intermtional Financial Statistics, Supplement on Price Statistics,“ Washington, D.G. (D) U.S. Department of Agriculttre (various issues). Meet Situation mid Outlook Report,“ CED/ERS, Washington, D.G. (E) U.S. Department of Aariculttte (veriotm issue). “Rice Sitution and Outlook Report,“ cal/6R3, Washington, D.G. (F) International Monetary Fwd, (1%5). “international Financial Statistics, Swpluient on Emchsrae Rates,“ Washington, D.G. (G) International Morietsry Find, (veriotm issues). “international Financial Statistics," Weshimton, D.G. (ll) lntsmtionel Monetary Fad, (1954). “international Financial Statistics, mums-mt on Output Statistics,“ Washiraton, D.C. (1) Board of Governors of the Federal Reserve System (various issues). “Federal Reserve Bulletin,“ Washington, D.G. (.1) International Rice Research Institute, (1%). “World Rice Statistics,“ Menille, Philippines. (K) Orb-i, F. srd P. Rose, (1%8). ”World Population by Comtry and Region, 1950-86, and Projection to 2050,“ Agriculture and Trade Analysis Division, U.S. Department of Agriculture. Staff Rqaort AGES 880308, Weshimton, D.G. 62 relatively consistent. This problem is, of course, not unique to trade issues. A third potential problem arises when a country’s net exports are large relative to the world market. In such a case, the world price may no longer be considered exogenous, and estimation of equation (8) would present an identification problem. However, a dominant firm oligopoly model of world grain trade (Mitchell and Duncan, 1987) reports that only the United States has exerted any price leadership in the rice market, and only the United States and Canada have done so in the wheat market. The report concludes that “the remaining major exporters in these markets have behaved in a manner consistent with a small-country exporter model in which their market demand is perceived to be perfectly elastic at the world price set by the dominant exporter" (p.20). To account for potential Simultaneity between net trade and world price, this study uses an instrumental variables procedure for the United States, Canada, and the Soviet Union in the wheat market, and for the United States, China, and Thailand in the rice market (Appendix 2). Finally, the data used in estimation may be subject to measurement errors. Production and population data, for example, are in some cases little more than time trends altered by limited information (Abbott, 1979). Parameter estimates must be interpreted accordingly. To reiterate, the purpose of the regressions is threefold: to provide insights into the interactions between national and international markets, to allow inferences to be made concerning the extent to which particular countries transmit or absorb quantity 63 instability onto/from world markets, and to evaluate the contention that government insulation policies contribute greatly to world market instability. 3.4 INTERPRETATION OF WHEAT RESULTS The results of regression (8) are presented in several ways. The complete regression results are contained in Appendix 2. The more immediate focus of this study concerns the estimated transmission and absorption coefficients, which are presented and compared with those from other studies in Table 1-10. These estimates are not directly comparable because of differences in model specification and sample period. There is little consistency across studies for most countries, which may partially indicate the sensitivity of results to model specification. Nevertheless, the results support the conclusions of other studies that national wheat trade is generally unresponsive to changes in world price. This indicates an unwillingness of most countries to promote international price stability by reducing net imports (raising net exports) as prices rise and vice versa. What is perhaps suprising is that all of the major exporters except the United States are in this category. At the average world price level over the period 1960-87, U.S. net export supply elasticity for wheat was roughly .60. The other major exporters were very unresponsive to world prices. Among wheat importers, only Brazil, the Peoples’ Republic of China, Turkey, Nigeria, and Sub-Saharan Africa appeared sensitive to 64 TABLE 1-10 A Comparison of Price Elasticities of Net Export Supply in World Wheat Trade This Coffin Abbott Blandford Wilde Research et al. Exporters: ‘ United States .59** -.16 .48* Canada .11 .56 (a) Australia .01 9.61 (a) Argentina .15 -.83* (a) M France .08 Importers: EC-12 .23 .07 (a) Soviet Union 1.16 -4.23 Brazil .53*** -.20 —2.48 .08 China .77* (a) Japan -.02 .69 India .10 Pakistan -.19 Turkey 5.30 Egypt .09 1.17** .46 Nigeria .33* .21 East Europe -.57 Mideast .38 North Africa -6.30**f Sub-Saharan Note: the elasticities are d(NX)/d(PW) * PW/NY, with PW and NX defined * in Table I-9. Negative signs indicate perverse results. significantly different from zero (two-tailed test) at the 90 percent level. significant at the 95 percent level significant at tbs 99 percsnt level F-stotistic on PW t and PW t-I significant at the 95 percent eve . coefficient not reported by author of original report because of statistical insignificance. 65 world price. China and Brazil are currently the second and fifth largest wheat importers. Their prominence consequently exerts a stabilizing influence on the world market. Responsiveness to world price may in some cases reflect an inability to afford the high cost of domestic stabilization policies. In such cases, low-income consumers bear a large part of world market adjustments. In other cases, such as China and most of Africa, other grains are important in consumption, thus providing alternative and possibly cheaper sources of food to fill import requirements when world wheat prices rise. Chinn (1981) and Timmer and Jones (1986) have noted the ”calorie arbitrage” undertaken by China since the mid 19705. When the wheat/rice price ratio rises, the government tends to reduce imports of wheat and compensate by rerouting potential exports of rice back into domestic markets. When the wheat/rice ratio falls, both rice exports and wheat imports are increased. In this way, China--with its heavily administered system-- exerts a stabilizing effect on both the world wheat and rice markets. However, most major importers adjusted trade volumes very little in response to changes in world market conditions, most notably Japan, Eastern Europe and the EC-12. The reluctance of these large countries to vary trade volumes in response to world market conditions causes more extreme price fluctuations and forces adjustment to occur elsewhere, predominantly in low-income countries. As the major wheat importer, the Soviet Union’s import behavior critically affects the stability of the world market. The results indicate that Soviet wheat imports are quite sensitive to world price, but the elasticity estimate of 1.16 is Significant only at the 85 66 percent level. Josling and Barichello (1984) contend that Soviet stock movements have been very sensitive to world price. Updating their data, over 20 million tons of wheat were accumulated during the low price years of 1975/76, 1976/77, and all years since 1981/82. During the tight market conditions of 1972/73, 1973/74, and 1977/78--1980/81, the Soviets cumulatively released 24 million tons from their stockholdings. However, Soviet wheat import sensitivity to world price appears to have been mitigated by food policy changes undertaken in the early 19705. Since that time, agricultural planners have sought to raise per capita grain and meat consumption. Toward this end, the Soviet Union started to rely heavily on wheat and corn imports, which prior to 1970, were kept to a minimum because reliance on nonsocialist countries for food was viewed as both risky and politically embarassing (Koopman, 1986). As a result of this change in policy, per capita wheat consumption rose from 270 kilograms in the early 19605, to 380 kilograms in the 19705 and 19805. Almost all of this increase was due to increased livestock production; wheat fed to animals increased fivefold from the early 19605 to the early 19805. This sustained commitment to wheat consumption goals during the 19705 occurred during wide swings in world pricés, mitigating the Soviet Union’s price responsiveness. Many of the countries examined vent over half of their production instability onto world markets. This is evident by examining the national transmission effects reported in column (a) of Table 1-11. The statistical strength of this coefficient for most countries Potential Trade Instability Resulting from National Production Variations: 67 TABLE I-ll Wheat (1960-87) Transmission Standard Deviation (c) as % 4“ Coefficient: of Production of Average 8 in Variations from Annual World Country EquaIion (8) Trend (a*b) Trade (000 tons) (000 tons) (a) (b) (C) (d) Exporters: United States .59* 6345 3744 4.99 Argentina .93*** 2294 2133 2.80 Australia .56*** 3108 1740 2.32 France .58*** 2594 1504 2.02 Canada .36 3871 1384 1.86 Importers: Soviet Union .30*** 15461 4638 6.21 . EC-IZ .43*** 5607 2411 3.18 ; East Europe .35*** 2908 1018 1.42 India .26* 2756 717 1.03 Brazil .89*** 779 693 .92 Mideast .39*** 1517 592 .78 ; North Africa .74*** 770 570 .75 . China .08 6825 546 .71 3 Japan 1.13*** 367 415 .60 5 Turkey .34*** 1092 371 .49 Pakistan 44*** 657 289 .42 Sub-Saharan ; Africa .89** 154 137 .24 ; Egypt .27 137 37 .03 *** percent level. ** Significant at the 95 percent level. * Significant at the 90 percent level. Significantly different from zero (two-tailed test) at the 99 68 examined indicates that short-run production variability is an important source of variation in national trade volumes. However, the degree to which this affects world market instability depends on the magnitude of national production fluctuations. For example, although the Soviet Union exports less than half of its production variation onto the world market, the sheer size of its production fluctuations can cause destabilizing swings in import demand. The standard deviation of production deviations from trend, i.e. (PRODt - PROD*t) (where PROD*t is the linear trend value in year t) is reported for each country over the sample period in column (b). If normally distributed, then the deviation from trend production would be less than its standard deviation approximately two out of three years. Ninety-five percent of the production fluctuations would be within two standard deviations. Even though (PROD - PROD*) may not be normally distributed in all cases, this figure is a first approximation of the extent of expected production variability in a given year. The product of columns (a) and (b) thus provides a measure of potential trade instability resulting from production variability in national markets. With the normality caveat mentioned above, the figures in column (c) may be interpreted as the expected fluctuation in net eXports resulting, ceteris paribus, from a domestic production variation of one standard deviation. Not suprisingly, the largest producers tend to have large production variability. However, the degree to which such variability is absorbed by consumption and endstock adjustments varies greatly, and profoundly affects the extent to which production changes destabilize 69 the international wheat market. Over the 1960-88 period, world wheat trade averaged 75 million tons per year. Thus in a normal year, not even the Soviet Union’s trade fluctuations resulting from production variation account for more than 5-6 percent of average world trade, and most countries contribute much less than this (Column 0). In extreme years however, four or five countries, including the United States, EC-12, and the Soviet Union may vent production instability onto the world market to a degree that affects world price. For example, a production shortfall in the United States of two standard deviations from trend, which might occur with a probability of around 5 percent, would be expected to reduce net exports by about 7.5 million tons, or 10 percent of annual world wheat trade, ceteria paribus. By contrast, China’s production fluctuations have not greatly affected world trade volume (even though it is the second largest producer and importer in the world). This is because it tends to transmit very little of its production variability into trade variability. A production fluctuation of two standard deviations from trend would be expected to change Chinese imports by less than 1.5 million tons, or 1 percent of annual trade in a normal year. WWW While production variability coupled with a high transmission coefficient represent a major source of trade instability on world markets, what is the major source of the production variability? Production variability can be broken down into area and yield components. Vield fluctuations from trend in most cases reflect 70 variable weather patterns, and are largely unrelated to policy shifts in the short run, although this is not always the case. On the other hand, year-to-year variation in area cultivated is more often attributable to changes in government policies that alter crop and input prices or place restrictions on resource allocation. The sources of production instability are determined by decomposing the variance of annual percentage changes in production into area and yield components. Starting with the identity that production (PROD) equals the product of area (A) and yield (Y), which may reasonably be assumed functionally unrelated, the total derivative of production is: d(PROD) - Y*d(A) + A*d(Y) Dividing both sides of the equation by PROD yields: d(PROD)/(PROD) - d(A)/A + d(Y)/Y Taking the variance of both sides gives: Variance [d(PROO)/PROD] - Variance [d(A)/A] + Variance [d(Y)/Y] + 2*Covariance [d(A)/A, d(Y)/Y] This formula is used to analyze the production, area, and yield variability of the major national actors in the world wheat trade. The method implicitly accounts for trend movements in the data. Hence, 71 steadily rising yields or acreage are not counted as instability by this indicator (see Appendix 1 for further discussion of the properties of this measure). The results indicate that yield instability accounts for the bulk of production instability among 9 of 12 major trading regions (Table 1- 12). This is especially true for the Soviet Union, which accounts for more annual variation in wheat production than any other country. A substantial portion of arable land in the Soviet Union is Situated in what are called “high risk farming areas” subject to extreme weather variation (International Wheat Council, 1976). In the long run, the ability to predict USSR weather and crop supplies better (through satellite photography or exchange of information with Soviet authorities) would contribute greatly to predicting world wheat market outcomes. By contrast, area instability is the major source of production instability in the United States. This is not suprising, Since the profitability of US wheat production and agriculture in general has been buffeted by many forces since the mid 19705. Easy world credit, growing world population, and a low dollar stimulated the demand for US wheat exports and accounted for the rapid expansion of acreage in the 19705 and early 19805. When tight credit and the dollar appreciation eroded world demand several years later, US government policy actively strove to reduce wheat acreage. Thus, the instability in U.S. wheat production at least partially reflects deliberate adjustments to world market conditions, albeit at substantial cost and hardship to the U.S. farm sector. 72 TABLE 1-12 Sources of Production Instability in Major Wheat Trading Areas 1960-87 % of variance in production from: Var Var Cov (AREA) (Yield) (AREA, YIELD) Exporters: United States 104 39 -41 Canada 32 62 6 Australia 10 78 12 Argentina 45 35 20 France 17 49 34 Importers: Soviet Union 5 89 6 EC-12 16 56 28 China 5 85 10 Egypt 57 27 16 Japan 40 66 -6 Brazil 44 68 -12 India 19 53 28 World: 9 83 8 Cogelusiens Only several of the many actors in the world wheat market are large enough to significantly absorb potential world price shocks. These major players include the United States, Canada, the Soviet Union, China, the EC-12, and possibly Australia. All of these countries have internally absorbed their production variations to some extent through stock and consumption changes. Although the Soviet Union has frequently been identified as a major destabilizer in the world wheat market, the truth in this statement owes mainly to its sheer Size, being the largest producer and importer in the world, and 73 the fact that Soviet yield instability is high, not because the Soviets transmit a majority of their production instability onto the world market. The Soviet Union has actually transmitted a smaller proportion of its wheat production variability onto the world market than each of the major exporters including the United States. The Soviet Union’s transmission coefficient of .30 means that 70 percent of its production variations are absorbed by domestic consumption and/or stock adjustments. Unless it is prepared to hold much greater stocks (the USSR has been the third largest wheat stockholder since 1980), or permit much greater consumption variability, the sheer size of its production variations will cause marked fluctuations in world import demand. Production instability of most countries, including the Soviet Union, is associated mainly with yield instability rather than area instability. Unless technological changes such as irrigation or more weather-robust cultivars are deve10ped, this component of world market instability is not likely to be reduced. Among major actors, only the United States, PRC, and Brazil appear to vary trade volumes in a stabilizing manner when world prices fluctuate. These countries have acted as shock absorbers for the world market by increasing stocks and/or consumption when world prices are low, and vice versa. Most of the major actors in the wheat market have been unresponsive to changes in world market conditions. This presents serious risks for poorer countries that are dependent on wheat imports and unable to afford the high costs of domestic price stabilization. 74 3.5 INTERPRETATION OF RICE RESULTS Earlier.studies (Falcon and Monke, 1979-80; Siamwalla and Haykin, 1983) have found a general lack of responsiveness between the net exports of major actors and world price. However, as mentioned earlier, these studies have used the US dollar price of Thai 5% broken rice as the world price and have assumed a linear relationship between price and net exports. The results of this study indicate that when this price is adjusted by national exchange rate and domestic price indicators, and is allowed to affect trade béhavior in a non-linear manner, many countries appear very price responsive (Table I-13). Two points follow from this. First, as suggested by the purchasing power parity theory, the relevant price series facing rice market actors is clearly the import or export price in the national currency relative to the domestic price level. The domestic price level is important because the rate of inflation may counteract or exacerbate exchange rate fluctuations in real domestic currency terms. Second, the results indicate that when world prices are low, countries such as Iran, Iraq, Pakistan, Vietnam, Senegal and the EC-12 may be much more price responsive than in tight world market conditions, when minimum domestic rice supplies and pipeline stocks are critical. As the major rice exporter, Thailand’s lack of price responsiveness represents a potentially destabilizing element in the rice market. The lack of response is not suprising given the government’s historical reliance on variable export taxes on rice. 75 TABLE I-13 Price Elasticity of Net Export Supply Estimates in the World Rice Trade Exporters: Thailand -.32 United States .32** China .74** Pakistan ; 1.07**f H Burma 5 .03 Importers: g Iran - 1.47**f Iraq .95**f Indonesia .11 Nigeria .01 Brazil I 2.43 Senegal .03***f EC-12 1.69***f Korea ' .35 India . .34 Japan .10 Malaysia : .01 Vietnam .55**f Bangladesh .02 Soviet Union .91* Mideast .19f Sub- Saharan Africa .31** Note: the elasticities are d(NX)/d(PW) * PW/NY, with PW and NX defined M: ** : t f in Table 1-9. Negative signs indicate perverse results. Significant at the 99% level. Significant at the 95% level. Significant at ths 90% levgl F- statistic on PWt and PW5 significant at the 95 percent level, in all cases tsignifying that the trade elasticities become more elastic as PWt falls. 76 When world prices rose, the government tended to raise the export tax as well, with the objective of diverting potential eXports to the domestic market in order to stabilize prices. The correlation between the f.o.b. Bangkok 5% broken price and the export tax was +.32 over the available data period (1967-82). Hence, the Thai government’s variable export tax served to cut off potential export volume that otherwise could have moderated world price rises.15 Fortunately, other major exporters -- China, Pakistan, and the United States -- responded significantly to world price, especially during surplus periods, despite heavy government involvement in their domestic markets. Table I-14 presents the transmission coefficients from equation (8) and the standard deviation of rice production variations from trend, as described earlier for the wheat market. As indicated in column (a), the five major exporters normally absorbed 50-100 percent of their production variations internally through stock and consumption adjustments. The remainder was transmitted onto the world market. Among_importers the transmission of national production instability varies widely. China and India have the greatest production variability of all countries (column b), yet very little of this variability is vented through trade. By contrast, the southeast Asian countries of Indonesia, Burma, Vietnam, and Thailand represent the four greatest transmitters of production instability onto the world market. 15Since 1986, Thai rice policy has undergone major reforms, including the suspension of the export tax. This may be expected to increase the future responsiveness of Thai rice exports to world price movements. 77 TABLE I-l4 Potential Trade Instability Resulting from National Production Variations: Rice (1960-87) Transmission Standard Deviation (c) as % Coefficient: of Production of Average , Variations from Annual World 5 Country EquaTion (8) Trend (a*b) Trade 5 (000btons) (000 tons) d P-—--—-— _-,_-__________ (a) ( (C) ( ) IExporters: j Burma .49*** 776 380 4.13 5 Thailand .36*** 722 260 2.78 , United States .38 553 210 2.32 I China .01 4507 41 .41 M Pakistan .00 235 0 -- MImporters: I Indonesia .45** 1196 538 5.79 n i Vietnam .46*** 770 354 3.82 M Korea .28 445 124 1.36 ; EC-12 1.25*** 100 125 1.36 9 Japan .13* 875 114 1.24 ' Nigeria -1.04* 112 -116 1.18 ; Mideast .92** 101 93 1.01 9 Bangladesh -.15 462 -69 .75 5 Malaysia .34** 154 52 57 i India -.01 4197 -42 46 , Brazil .07 505 35 38 M M Iraq 52** 52 32 35 ' Sub-Saharan M M Africa .19 121 23 .25 7 Iran .24 68 16 .18 Soviet Union -.09 114 -10 .11 1 . Senegal .23 21 5 .05 .M *** SignificaNtly different from zero (two- -tailed test) at the 99 percent level. ** Significant at the 95 percent level. * Significant at the 90 percent level. 78 Sources of Rjee Ereductien Instability As with wheat production, the major source of production instability in the rice market is unstable yields (Table I-15). This is especially true throughout south and east Asia, where the rice crop is greatly affected by the Asian monsoon. Parts of Thailand, Bdrma, China and southern India have highly variable rainfall and are prone to drought. Improved and more extensive irrigation will be required to appreciably reduce production instability in these areas (Barker et al., 1985). In other parts of China, Japan, Korea, Bangladesh, and southeast Asia, flooding and typhoons also create great yield variability. Area instability has dominated in several countries, most notably Indonesia (the third largest importer) and the United States (the second largest exporter). Area instability in both of these countries appears highly related to policy-induced changes in production incentives facing farmers. In Indonesia, this has taken the form of wide annual variations in real input prices and production subsidies (International Rice Research Institute, 1986). In the U.S., acreage reduction programs, fluctuating support prices, and recently, the use of commodity certificates to release stocks onto the market--all responses to changing world market conditions--have significantly affected the acreage devoted to U.S. rice production over the years. Changes in the variability of U.S. producer and milled rice prices have also created variability in rice area cultivated (Brorson, Chavas and Grant, 1987; Grant et al., 1984). Therefore, policy-induced uncertainty, which manifests itself through annual domestic price 79 TABLE 1-15 Sources of Production Variability in Major Rice Trading Countries % of Variance in Production from: Var Var Cov (AREA) (YIELD) (AREA, YIELD) Exporters: Thailand 39 45 16 United States 89 11 0 China 52 61 -13 Pakistan 30 45 25 Burma 21 58 21 Importers: India 4 71 25 Indonesia 53 27 19 Iran 44 93 -37 EC-12 15 89 - 4 Nigeria 48 72 -30 Senegal 23 72 5 Bangladesh 21 56 23 Vietnam 21 72 7 Japan 19 71 10 Korea 4 91 5 Brazil 52 45 3 Soviet Union 51 30 19 World: 24 46 30 variability, may exacerbate the instability of production and annual trade volume. An econometric simulation analysis by Mahama and Meyers (1985) indicates that the removal of trade restrictions may result in less world price stability than expected, because such policy changes would normally increase the variability of domestic prices, which would in turn increase production and trade instability. 80 n 1 ion Despite the higher level of world price instability in the world 'rice market than in the wheat market, major actors in the rice market appear more price responsive than in the wheat market. This is even more suprising considering that the rice trade is one of the most heavily affected by government regulation and state trading. Many analysts have concluded that such government activities account for a large part of the instability in world rice prices. However, judging simply by the magnitude of the transmission and absorption effects of national actors in the wheat and rice market, one would conclude that instability is higher in the world wheat market. Yet this is not the case. Many governments, while devising policies to stabilize domestic rice prices, have done so in ways that permit private actors or parastatal agencies to respond in a statistically significant manner to world price conditions. Many of these countries (Pakistan, Iran, Iraq, Senegal, Vietnam, and the EC-12) are more price responsive during surplus conditions than during tight market conditions. Most of the trade responsiveness to world price has been through stock changes, as is suggested by econometric evidence in Section 3.6. It must be recognized that national transmission of production instability cannot be blamed solely on government insulation policies, as is sometimes implied. Under free trade, variations in production are also vented abroad as goods migrate from surplus regions to deficit regions. The important question is how much greater is the extent of venting under government insulation policies. 81 3.6 CONDITIONS NECESSARY FOR GOVERNMENT INSULATION TO AFFECT WORLD PRICE INSTABILITY The conventional wisdom that government insulation policies exacerbate price instability in world markets are based on several combinations of the following assumptions: 1. The insulating country is large relative to the size of the international market, and therefore faces an upward (downward) sloping export supply (import demand) curve, rather than the perfectly elastic function faced by small trading actors. 2. If the insulating country in question is small, then its’ production deviations from trend must be positively correlated with world production deviations from trend. 3. The pr0portion of domestic production variation vented onto world markets (i.e. the transmission coefficient) is higher under governmnet insulation than under non-insulation. 4. .The net export supply elasticity under free trade (related to the sum of domestic supply and demand elasticities) must be appreciably greater than the country’s net export supply elasticity under insulation, as estimated in equation (8). These assumptions are examined in turn. The first assumption requires that national trade variations caused by insulation policies are large enough to actually affect the world market. This immediately rules out all but a handful of countries in both the wheat and rice markets. Even for four of the five major importers (Iran, Iraq, Nigeria, and Senegal), transmission of national production instability would appear to have a negligible effect on the world market. This is because national production deviations are small in these countries relative to the volume of world rice trade. For none of these countries would a national production deviation from trend of one standard deviation (column b in Table I- 14), when multiplied by the proportion of production variation 82 transmitted abroad (column a), amount to more than 2.5 percent of average annual world trade over the period 1960-87 (column d). However, production fluctuations in Indonesia, the third largest importer, are large relative to the world market. About half of Indonesia’s production fluctuations are transmitted onto the world market (as Shown by the transmission coefficient for Indonesia in column a). A one standard deviation change in production from trend in Indonesia would be expected to amount to 5.7 percent of world rice trade over the period 1960-87. Among exporters, expected trade variations resulting, ceteris paribus, from a domestic production variation of one standard deviation from trend are highest for Burma and Thailand (column c). In neither case, however, would such fluctuations amount to more than 4 percent of annual world rice trade over the period 1960-87. A production deviation of two standard deviations from trend would be expected to cause an 8 percent change in annual world trade, but such extreme fluctuations would be expected to occur with a probability of five percent, assuming that Thai and Burmese rice productibn around trend approximates a normal distribution. Overall, the results suggest that in all but four countries examined in the rice market (Thailand, Burma, Vietnam, and Indonesia), production variations of two standard deviation from trend create less than a 5 percent change in mean world trade volume over the period 1960-87. The bulk of rice importers therefore face a rather elastic export supply function. The more elastic the export supply (import 83 demand) curve facing an individual importer (exporter), the less a random fluctuation in its trade volume will affect the world price. With such a modification to the theoretical framework presented earlier in Figure 1-1 (representing the mechanisms by which government insulation is claimed to destabilize world prices), it becomes apparent that little world price instability can be attributed to the insulation policies of most rice trading countries. I The same points hold for the wheat market, except that a larger number of actors appear able to affect mean world trade volume through their production changes. The results suggest that production variations of two standard deviations from trend in the United States, Soviet Union, Argentina, France, and Australia would each create a 4 percent or greater change in mean world trade volume. For most countries, however, national production and/or trade variations are too small relative to world trade volume to greatly affect world price, even if the country completely prevents net trade from responding to changes in world price. The validity of assumption 2--i.e. that trade fluctuations of small countries, when aggregated, destabilize world prices-~is examined in Chapter 4. Assumptions 3 and 4 are relevant only for the handful of large countries remaining. The extent to which government insulation policies of these countries destabilize the world market depends on the difference between the absorption and transmission effects under insulation compared with those under no trade restrictions. 84 Reconstructing the functional relationships in equations (1), (2), (3b), and (5b) gives: (9) Nxt - PRODt - CONt [GNPt, PDt(PWt, PWZt, PRODt, ENDt-1)] - ENDt [Rt, ENDt-1, Rt-1 PDt(PWt, Pvzt, PRODt, ENDt_1), Pot-IIPMI-I: PVZt-I» PRODt-I: ENDt-Ill + ENDt-1 Differentiating (9) with respect to PWt yields: (10) d(NXt)/d(PWt) = -[d(CONt)/d(PDt) + d(ENDt)/d(PDt)] * d(PDt)/d(PWt) This is the absorption effect under both insulation and non-insulation. The relationship has already been estimated under insulation, and is equivalent to 51 + 2*filo(PW) in equation (8). Differentiating (9) with respect to PRODt gives the transmission effect under both insulation and non-insulation: (II) d(NXt)/d(PRODt) - I - [d(CONt)/d(PDt) + d(ENDt)/d(PDt)] * d(PDt)/d(PRODt) This relationship has already been estimated under insulation, and is equivalent to B3 in equation (8). This issue now is to Compute the transmission and absorption effects implied by free trade, and evaluate whether they are larger or smaller than those under insulation. 85 or ' n ct As is apparent from equation (10), the responsiveness of net exports to world prices can be broken into two effects: the responsiveness of domestic consumption and stock demand to domestic price, and the responsiveness of domestic price to world price movements. Concerning the second effect, recall that under free trade (and no transport costs), the coefficients in equation (5b) PDt - do + d1(PWt) + .... would reduce to do-O and d1-1, as PDt would in fact equal PWt. Therefore, under free trade, equation (10) reduces to d(NXt)/d(PWt)FT - - [d(CONt)/d(PDt) + d(ENDtl/d(PDt)] Assuming for the moment that the responsiveness of consumption and endstocks to domestic prices are the same under both insulation and free trade, the difference between net export price responsiveness under insulation and free trade hinges on the degree to which government dampens the transmission of world price movements to domestic markets (i.e. the degree to which d(PDt)/d(PWt) is less than one). Thus, the free trade-implied trade elasticity may be calculated by dividing equation (10) under insulation, which was previously estimated for each country in equation (8) and reported in elasticity 86 form in Tables 10 and 13, by the price transmission coefficient d(PDt)/d(PWt) under insulation. This coefficient was estimated using OLS in equation (5b) for seven rice trading countries for which domestic price data were available. The resulting price transmission elasticities are presented in Table I-16. Complete regression results are given in Appendix 3. The degree to which governments allowed world price fluctuations to affect domestic prices varied widely. Domestic prices in Burma and the United States were quite responsive to world price movements. In TABLE 1-16 Estimates of World-tO-Domestic Price Transmission Elasticities for Selected Rice Trading Countries Exporters: Thailand .25::: United States .83** Pakistan .45“f Burma .87 Importers: * Bangladesh .68 Indonesia .07 Malaysia .19*** Note: Elasticities are d(PDt)/d(PWt)*PW/PD. Source: Domestic price data obtained from International Rice Research Institute (1986). All other data sources are listed in Table 1-9. ***: significantly different from zero at the .99 level significant at the .95 level significant at thg .90 level F-statistic on PW t significant at the .95 level ::.:.i 87 the United States, a 10 percent increase in world price generated a 8.3 percent increase in the domestic consumer price, calculated at the respective price means over the estimated period. The relationship between domestic and international rice prices was non-linear in the case of Pakistan. The government allowed domestic rice prices to rise by smaller amounts as world prices rose progressively higher, apparently to protect consumers from transitory food insecurity during tight world market conditions. Other countries, notably Indonesia, Malaysia, and Thailand, heavily insulated their domestic markets from world rice price movements. In Indonesia, for example, a 10 percent increase in the world price produced less than a one percent increase in domestic consumer prices on average. Estimates of the absorption effect under free trade can now be computed by dividing both Sides of equation (10) by the price transmission coefficients (represented in elasticity form in Table I- 16) for each country. This forces d(PDt)/d(PWt) to be unity, as is implied under free trade. The free trade-implied elasticities are reported in Table I-17, under three alternative estimates of consumption and stock responsiveness to domestic price. This is necessary because there are no data under free trade for any of the countries examined. Thus, it is unknown how a transition to free trade would affect consumption and stock demand responsiveness to domestic price (i.e. d(CONt)/d(PDt) + d(ENDt)/d(PDt) in equation 10). The three alternative scenarios are: (1) no change in domestic price responsiveness under free trade and insulation (this is equal to [81 + 2*Blo(PWt)]*d(PWt)/d(PDt) as estimated from equation 8); (2) domestic 88 TABLE I-l7 Estimates of World Price Elasticity of Net Export Supply in World Rice Trade: Actual and Free-Trade Implied ----INSULATION --------------- FREE TRADE ---------- Actual Estimates Sensitivity Analysis Scenario: Country from Equation (8) (1) (2) (3) i Exporters: Thailand -.32 -1.60 -2.00 -2.40 United States .32 .33 .41 .50 Pakistan 1.07 2.82 3.52 4.23 Burma .03 .06 .07 .09 Importers: Bangladesh .02 .03 .03 .05 Indonesia .11 2.20 2.75 3.30 Malaysia .01 .05 .06 .08 Sources: Domestic price data obtained from International Rice Research Institute (1986). All other data sources are listed in Table I-9. Sensitivity Analysis Scenarios: (1) (2) (3) assumes that consumption and endstock response to domestic price is equal to that implied from estimation of equation 8, after adjusting for the effect of incomplete price transmission under insulation. Thus, the world price elasticity estimates in this column are derived by dividing each country’s net export supply curve under insulation, estimated in equation 8, by the price transmission coefficient, d(PDt)/d(PWt), as estimated in equation 5b. This effectively adjusts for the depressing effect of government insulation policies on a country’s trade responsiveness to world price. assumes that consumption and endstock response to domestic price is 25 percent higher than that implied from equation 8. assumes that consumption and endstock response to domestic price is 50 percent higher than that implied from equation 8. 89 price responsiveness under free trade is 25 percent higher than that measured under insulation (i.e. [81 + 2*fllo(PWt)]*d(PWt)/d(PDt)*1.25); (3) domestic price responsiveness under free trade is 50 percent higher than that measured under insulation (i.e. [81 + 2*Blo(PWt)] *d(PWt)/d(PDt)*1.50).15 While the free trade-implied elasticities are often much larger in relative terms than those estimated under insulation, they are generally not much higher in absolute terms. The results indicate that only in Indonesia and Pakistan would the volume of trade be significantly more responsive to world prices under free trade than under current insulation policies. This is for two reasons. First, in several of the countries examined, government allowed over half of annual world price fluctuations to affect domestic market prices. This implies that either the government policies were unsuccessful in meeting their objectives, or that they were not designed to remove all price variation, but rather to mitigate price movements outside of certain explicit or implicit ranges. Second, and more importantly, even if domestic prices were quite divorced from world price movements, as in Malaysia, Bangladesh, and Thailand for example, there is no indication that the transition to freer trade would substantially increase net trade responsiveness 16While each of these scenarios assume that the responsiveness of consumption and/or stocks to domestic price under free trade will be greater or equal to that under insulation, this is not necessarily the case. Government insulation policies designed to stabilize domestic prices must, to be successful, make stocks more responsive to price than if private actors alone performed this function. Therefore, the assumption that the domestic elasticities are equal or greater under free trade may upwardly bias the free trade-implied elasticities. 90 (Table I-17). This is because domestic consumption and stock responsiveness to domestic price may be quite low as well. Even assuming a 50 percent increase in the responsiveness of consumption and endstocks to domestic prices, these responses are too low to substantially increase the trade elasticities in most of the countries examined. In such cases, the effect of government insulation policies on net export responsiveness, although potentially large in proportional terms, may be quite small in absolute terms. This was examined further by estimating the responsiveness of consumption and endstocks to domestic retail rice prices for the same seven rice trading countries for which domestic price data were available. The procedure involved two-stage least squares estimation using equations (2), (3b) and (5b). In the first stage, equation (5b) is estimated; fitted values for domestic price are then used to derive domestic consumption and endstock demand elasticities in equations (2) and (3b). These estimates are compared with those of a similar study by Ito, Wailes, and Grant (1985), and are presented in Table I-18. Complete regression results are contained in Appendix 4. The evidence indicates that rice price elasticities of consumption are very low. All of the estimates in both studies were below .50. This is not surprising Since rice is a critical staple food with limited substitutes in most Asian countries.17 These results are consistent 17 Even in areas where other staple grains offer consumption substitutes for rice, rice demand elasticities were still found to be very low in many cases. In parts of Sub-Saharan Africa for example, this appears to be the case because of lower fuel costs and ease of preparation for rice compared with potential coarse grain substitutes (Rogers and Lowdermilk, 1988; Reardon, 1988). 91 TABLE I-18 Estimates of Domestic Price Elasticity of Consumption and Endstocle For Selected Rice Trading Countries DOMESTIC PRICE ELASTICITY OF: ----- CONSUMPTION-------- -----STOCKS----- This Ito, Wailes, and This Study Grant (1985) Study fl Exporters: | Thailand +.14 .00 -.71 United States -.09 -.18** -5.60** Pakistan -.28 -.14 -1.54*** Burma -.08* -.02 na Importers: India -- .00 -- Bangladesh -.01 -.03 -1.94 Indonesia -.48** -.05 -3.21 Japan -- .00 -- S. Korea -- -.17 -- Malaysia +.03 na -.23 Note: Consumption elasticity: d(CON)/d(PD)*PDZ§OE__ Endstock elasticity: d(END)/d(PD) * PD/END Sources: Domestic price data Obtained from IRRI (1986). data sources are listed in Table I-9. ***: Significant at the .99 level ** : Significant at the .95 level * : Significant at the .90 level na : Stock data not available All other 92 with the survey of grain studies compiled by Scandizzo and Bruce (1980), which reveals that 14 of 16 estimates of rice and wheat demand elasticities are less than .50. Ongoing research suggests that low domestic supply and demand elasticities are frequently the result of multivaried production and marketing constraints, urbanization, weak processing techniques, and other factors common in developing areas, rather than an unwillingness to respond to price per se (Weber et al., 1988; Delgado, 1988; Martin and Crawford, 1988; Rogers and Lowdermilk, I988; Streeten, 1987). Evidence of limited consumption responsiveness to domestic price suggests that even if most governments allowed domestic prices to move freely with world prices, national trade responsiveness to world prices would not greatly increase in absolute terms. Stock demand estimates vary greatly among countries. Although they are quite high in some cases, reference to Table I-B reveals that most of the countries examined (the United States and Indonesia excluded) maintain small stock levels compared with the volume of world trade. Thus, most rice trading countries have a limited ability to stabilize the world market through stock adjustments. W Equation (11) shows that the magnitude of the transmission effect under both insulation and free trade depends on (a) the responsiveness of consumption and endstocks to domestic price, and (b) the effect of production on domestic price. In the absence of trade restrictions or domestic price policies, world and domestic prices are equated 93 (abstracting from transport costs). In the small-country case (i.e. small in terms of production; the country may still be a major trader), changes in production cannot affect world price, and therefore equation (11) reduces to: d(NXt)/d(PRODt) - 1 The logic behind this result is demonstrated in Figure I-2. The case of an exporter is presented, but the results apply equally to a large or small importer (in terms of trade volume); the main condition is' that the country’s production fluctuations are small relative to the world market. Because the time frame of the analysis is short-run, supply is considered unresponsive to current price, but again the results do not depend on this. Supply and demand (So, Do) define the excess supply curve (ESo). In the small-country case, the import demand curve is horizontal, and is equal to world price (IDoaPW). Under free trade, the excess supply curve in conjunction with import s s 00 0 1 E50 E51 Put \\\\h\\\\\ ,zzzz”’r//’ 1” 100 l <;;;;: ------ M | ‘ ----- M I l L I X0 X1 X2 ' X0 X1 X2 FIGURE 1-2. Transmission Effect under Free Trade for Small-Country Case 94 demand determine the volume of exports (Xl-Xo). Now introduce a weather-induced increase in supply (51). The excess supply curve shifts to E51, inducing exports of Xz-Xo. Because world price is not altered by the change in domestic production, the entire increase in production, Xz-Xl, is exported via trade, inferring d(NXt)/d(PRODt)-1. Note that this same logic extends to large actors in the world market, as long as they produce insufficient quantities domestically to alter world price through the transmission of production variability. This would include four of the largest five rice importers (Iran, Iraq, Senegal, and Nigeria) and three of the largest five wheat importers (Japan, Brazil,and Egypt). Tables I-11 and 1-14 indicate that these Countries’ potential trade instability resulting from a production shock of two standard deviations from trend would in no case amount to more than 2.6 percent of average annual world rice trade or 2 percent of world wheat trade over the 1960-87 period. To reiterate, the above theoretical model indicates that under free trade, national production fluctuations would be totally transmitted onto the world market (i.e. d(NXt)/d(PRODt) . 1), given the condition that national production fluctuations do not affect world price for the country in question. The venting of production disturbances through trade is consistent with the classic gains from trade argument that goods migrate from surplus to deficit areas until the marginal revenue (cost) of selling (buying) an additional unit of grain abroad are the same as in the domestic market. Perusal of Tables 11 and 14 reveals that the vast majority of countries transmit less than all of their production variations onto 95 world markets under present and past insulation policies. Therefore, the transition to freer trade by small and large trading nations may actually increase the-transmission of instability onto world markets. This reveals the questionable logic underlying assertions often made that government insulation policies of major traders destabilize world markets through the venting of production variations. The analysis thus far has not addressed the transmission effect for the small number of countries in which production variations may plausibly affect world price. This list would include Thailand, Indonesia, Burma, Vietnam, China and the United States in the rice market, and the Soviet Union, United States, France, Argentina, and Canada in the wheat market. For exposition, the cases of Thailand, China, and the United States are analyzed for the rice market. The task is to discern the magnitude of d(PWt)/d(PRODt) to draw inferences about the transmission effects of these large producers under free trade. To do this, I employ the same model as before, consisting of: I 1. A net export identity (equation 1) 2. Domestic consumption (equation 2) 3. Domestic endstock demand (equation 3) 4. Adaptive price expectations process (equation 4) 5. Exogenous annual domestic production and beginning stocks 6. World-to-domestic price transmission equation (5b, but modified to reflect the free trade case that PD-PW) to arrive at a free trade-implied net export equation for exporter i. 96 (12) int = PRooit ' 310 - aillPVt) - ai2(GNPit) - cio - CillPflt) - CiziPVt-Il - CialRt) - Ci4lRt-1) - CISIENDII-Ii + ENDit-I Equation (12) is identical to reduced form equation (8) except for the modification in (Sb) that PDt-PWt under free trade. Then, a simple rest-Of-world net import equation, representing the aggregation of equation (12) across all other rice trading countries yields the reduced-form equation (13) N"RON,t = - PRODRov,t + Ao + A1(PWt) + A2(GNPR0W,t) + Co + C1(PWt) + C2(PWt-1) + C3(Rt) + C4(Rt-1) + c5(ENDRON,t-1) - ENDRov,t-1 where each A and C coefficient represents the aggregate of the respective a1 and c1 coefficients across all other countries. Equation (13) is equivalent to the horizontal summation of all importer excess demand functions minus the summation of excess supplies of competing exporters. The equation can be simplified by letting the coefficients on PWt, A1 + C1, equal a reduced-form coefficient 61, and letting all other variables and their coefficients be represented in vector form by 52(2) (14) N"R0N,t ' 51(PVt) + 52(1) Equating equations (12) and (14) and solving for PWt gives 97 (15) PWt - [PRODMt - ao - a2(GNP1t) ' co - c2(PWt-1) - C3(Rt) - C4(Rt-1) - C5(ENDt-1) + 62(2)]/(a1 + C] + 61) Differentiating PWt with respect to PRODit yields: (16) «Pun/dunno“) - I/IaI + c: + 61) < o where a1 . d(CONit)/d(PWt) < 0 c1 - d(ENDit)/d(PWt) < 0 61 I d(NMROWtI/dipwt) < 0 It is clear that the more price responsive are domestic consumption, endstocks, and rest-Of-world imports to world price under free trade, the closer equation (16) is to zero, and thus the closer the transmission coefficient in equation (11) is to 1. It is often stated by proponents of trade liberalization that the transition to free trade would substantially increase the magnitude of world import and export elasticities. This analysis implies that such a transition would increase the transmission of domestic production variability of large- country actors in the world market. To examine this further, I use empirical estimates of parameters 61 borrowed from the commodity trade literature. Most estimates of the short run elasticity of demand for US, Thai, and Chinese rice exports are in the range of -4 to -10 under existing insulation policies of importing countries (see Tyers and Anderson, 1988). Under free trade and assumed price transmission elastiticies of unity, import demand 98 elasticities facing these countries have been estimated to rise to the range of -27 to over -40 (Tyers and Anderson, 1988).18 For arguments sake however, let us choose the lowest rice import demand elasticity estimate found in the literature: -0.08 as estimated by Siamwalla and Haykin (1983). While this estimate applies to the situation of insulation policies in effect during the early 19805, let us assume, again for arguments sake, that this elasticity would not increase under a transition to freer trade. Using the range of coefficients for d(CONit)/d(PWt) and d(ENDit)/d(PWt) as derived in Table I-17, the free trade-implied transmission effects are calculated for Thailand, China, and the United States (Appendix 6). In no case were these transmission coefficients lower than those estimated from equation (8) for these countries under insulation policies. This result is more robust considering that an implausibly low import demand elasticity estimate was used in the calculation. The logic behind this result is presented in Figure I-3. It is the same as Figure 1-2, except a more inelastic import demand curve is drawn, reflecting the case of a large country exporter. Domestic supply and demand curves (So, Do) determine excess supply curve ESo, which in conjunction with import demand curve 100’, determine the volume of net exports (XI-Xo). Now let a production Shock shift supply to S1 and by qz-ql, exports increase by x3-x1 only (i.e. (qz-q3)-(q1- qo) in the left hand side panel), implying that d(NXt)/d(PRODt) < l. 13 Adjusting this range of elasticities into their respective coefficients (using mean world price and volume data over the 1960-87 period) to obtain 61 in equation (16), and using estimates of a1 + Cl from Table 17, the value of d(PWt)/d(PR001t) in equatiOn (16) becomes effectively zero, implying a transmission effect Of 1. 99 90 93 91 92 X0 X1 X3 X2 FIGURE I-3. Transmission Effect under Free Trade for Large-Country Case If the import demand curve becomes more elastic as an outcome of freer trade (e.g., IDo), a greater proportion of variations in production would be vented through trade, approaching one as the import demand curve approaches negative infinity (as in Figure 1-2). 3.7 CONCLUSIONS While it is Often stated that government price stabilization policies that directly influence trade patterns introduce Significant instability in the world market, the results of this analysis question such conventional wisdom. It is certainly true that many countries with such policies transmit part of their production instability onto world markets. It is also apparent that net trade elasticities of these countries are quite low in some cases. However, this is not in itself evidence that government insulation policies exacerbate world price instability. The issue is whether instability would be greater or lower with a transition to freer trade. The transmission and 100 absorption effects are two prominent criteria for examining this issue. Given the assumptions underlying the model in this chapter, freer trade would increase countries’ trade responsiveness to world price and thus their ability to absorb quantity Shocks on the world market. However, in many cases the magnitude of such improvements would be small in absolute terms. This is because most rice trading countries examined do not exhibit enough responsiveness to domestic prices even without ,trade restrictions to greatly affect these countries’ responsiveness to world prices once such trade restrictions are eliminated. Therefore, low national trade elasticities can be deduced without reference to government insulation policies. The net trade curves facing individual trading countries may become Significantly more elastic if all countries liberalized trade. However, this would bring mixed blessings. 0n the one hand, such a transition would increase the ability of the world market to absorb quantity variations from an individual trading nation. 0n the other hand, the magnitude of such quantity variations would appear to be much greater under free trade. The model suggests that as the import demand (export supply) function facing an individual exporter (importer) becomes more elastic, a greater proportion of its production instability will be transmitted abroad. Moreover, the variability of production itself may increase, because domestic prices-~the expected values of which affect production decisions--may themselves become more variable under free trade. This would occur as world market shocks are increasingly allowed to affect domestic prices. CHAPTER 4 ORGANIZATIONAL SOURCES OF INSTABILITY IN THE WORLD WHEAT AND RICE MARKETS Instability and uncertainty are critically influenced by organizational aspects of a market, distinct from, or in addition to government insulation policies per se. In the case of wheat and particularly rice, these factors include (1) the lack of durable, reliable import markets for rice; (2) the geographical concentration of world production; (3) self-sufficiency policies of traditional rice importers and the role of green revolution technology in achieving these ends; (4) the relationship between petroleum prices and rice import demand; and (5) the evolving substitutability between wheat and rice among major consuming countries. 4.1 THE RELIABILITY OF FOREIGN MARKETS AND TRANSACTION COSTS A striking difference between the organization of the world rice and wheat market concerns the reliability of surplus and deficit areas. This is a crucial factor for all countries that depend on the world market. For exporters, reliable import markets represent a source of national revenue and a vent for surplus production that would otherwise need to be absorbed domestically, with potentially high internal disruption. For importers, reliable surplus-producing areas are necessary to meet domestic food requirements, broader national policy goals and, in some cases, political stability. 101 102 Figures 1-4 and I-5 contrast the fragility of trade patterns in the rice market relative to the wheat market. Production to consumption ratios averaged over the 1980-87 period are measured along the horizontal axis. The further a given country’s ratio is from unity, the more likely it consistently appears on one side of the market, either as an exporter (if the ratio is greater than one) or an importer (if the ratio is less than one). The vertical axis shows each country’s importance in the world market. The plot of wheat trading countries shows a.relatively wide dispersion of most countries from one, but many actors in the rice market are concentrated near one, indicating that annual production variations may cause them to appear on either side of the market in any given year. This is indeed borne out by examining the trade data of a number of rice market actors. Korea, Vietnam, Indonesia, India, Brazil, and Japan have sporadically switched from importer to exporter (or vice versa) several times since the mid 19705. Of these, Indonesia and Vietnam were identified in the previous section as potentially able to destabilize world market prices due to the magnitude of their domestic production variability and their willingness to transmit large portions of it onto world markets. In the wheat market, India and the Eastern European bloc (once) have entered the market as both importer and exporter since 1970. Several points fellow from this. First, the potential for several large import markets to evaporate in any given year presents great uncertainty for exporters, in terms of venting their surpluses at a remunerative price. Because many importers have distinct preferences 103 c s 35 s % of s world I trade 30 s W s 25 l s s 20 s b s j s s s 15 s s s as s s as a e is k 10 I! l W 8! 8 as s I as s o s! Id s as s s p 5 s! as s g as s l m s eq Is I! s f ah s: I s s n I ssr Is as s s as a! s s a s I see .25 .50 .75 1.0 1.3 2.0 4.0 20 Ratio of consumption to production Key: (a) Australia; (b) Canada; (c) United States; (d) Argentina; (e) EC-12; (f) India; (9) Eastern Europe bloc; (h) Pakistan; (i) China; (j) Soviet Union; (k) Middle East bloc; (l) Algeria; (m) Brazil; (n) Iran; (0) Egypt; (p) Japan; (q) Korea; (r) Nigeria. Source: USDA, 1988. Figure 4. Distribution of National Consumption to Production Ratios, Wheat: 1980-87 104 % of world d trade 30 s s s 25 s I s 20 b s i i I I 15 s I I I I s j m 10 I c I k n I s I h l o s I I e gi I s r 5 a s a s s as s s p qt t I s s I sfss s I s as s! u s s s s 1st! I I a a! me I .25 .50 .75 1.0 1.3 2.0 4.0 20 Ratio of consumption to production Key: (a) Australia; (b) United States; (c) Pakistan; (d) Thailand; (e) Burma; (f) India; (9) China; (h) Japan; (i) Indonesia; (j) Korea; (k) Bangladesh; (l) Vietnam; (m) Soviet Union; (n) Brazil; (0) EC-IZ; (p) Malaysia; (q) Nigeria; (r) Iran; (5) Senegal; (t) Iraq; (u) Hong Kong. Source: USDA, 1988. Figure 5. Distribution of National Consumption to Production Ratios, Rice: 1980-87 105 for certain varieties or qualities of rice, the drying up of an import market may make it difficult for an exporter with a certain kind of rice to find another buyer without accepting a large discount. Second, the combination of large, unreliable markets, distinct tastes and preferences within these markets, and no clear price discovery process (as in the wheat market) creates high search costs for both buyers and sellers. This is exacerbated by the lack of internationally accepted grades and standards in the rice market. Consequently, it is not suprising that the rice market supports a number of large brokerage houses located in the United States, Europe, Singapore and Hong Kong, charging fees of 5 to 10 percent of sale. Such rates are ”almost inconceivable in the wheat trade” (Siamwalla and Haykin, 1983). While such agents perform needed services, they impose costs on market participants that would not have to be incurred if trade patterns were more reliable and if more widely accepted grades and price discovery mechanisms were successfully instituted. Countries producing sporadic rice surpluses are even worse off. They frequently lack the milling facilities to ensure standardization and have insufficient exposure to the market to have acquired a reputation regarding quality (Siamwalla and Haykin, 1983). Furthermore, a sizable proportion of rice import markets may already be committed in the form of government-to-government and long run contracts. Hence, search costs for sporadic exporters are generally high. Siamwalla and Haykin conclude that this situation characterizes the rice trade in general: 106 The main problem with the rice market in our view is not instability in the sense generally understood, i.e., exhibiting large fluctuations in prices, nor particularly that these fluctuations are the consequences of 'thinness”:...Rather the main problem lies in the fact that the transaction costs involved are very large. When a country enters the world market (either as exporter or importer), or even when it is staying put and buys or sells the same volume as before, it has to search for trading partners. There is no rice “supermarket“ as there is in the United States for' wheat (quoted in Barker, Herdt, with Rose, 1985, p. 192). 4.2 GEOGRAPHICAL CONCENTRATION OF HORLD RICE PRODUCTION Another aspect of the rice market exacerbates the above problem. In Section 3, it was concluded that the domestic markets of only a handful of countries, ceteris paribus, are large enough to potentially disrupt the world market through venting their production instability abroad. However, the geographical concentration of rice production adds a new dimension to this issue. Eighty percent of the world’s rice is produced in China, Thailand, Burma, Bangladesh, India, Indonesia, and Vietnam. Production in these countries is highly correlated due to the Asian monsoons. 'There are in fact three distinct monsoon patterns, but only two are relevant here: (I) the Indian monsoon, which controls air movements over India, Bangladesh, and parts of Burma and Thailand; and (2) the Malayan monsoon, which affects Indonesia, Burma, Thailand, Vietnam, and most of China. Correlation coefficients of production variations from trend from 1960-87 are strikingly high, as shown in Table I-l9.19 Moreover, 19 It is to be expected that production levels themselves will be correlated between countries, as acreage and yields rise through time. More relevant for the examination of instability are unexpected variations in production, approximated here by the deviation from a linear time trend (PRODt-PROD t)- 107 TABLE I-19 Correlation Coefficients of National Rice Production Deviations from Trend in Monsoon Asia INDIAN MONSOON REGION India Bangladesh Burma Thailand India -- Bangladesh .68 ~- Burma .13 .42 -- Thailand .59 .40 .45 -- MALAYAN MONSOON REGION Thailand Burma Indonesia China Vietnam Thailand -- Burma .50 -- Indonesia .47 .34 -- China .27 .05 .60 -- Vietnam .34 .36 .67 .51 -- Source: USDA, 1988. the correlation coefficient of production deviations from trend between the Asian exporter bloc (Thailand, China, and Burma) and importer bloc (Indonesia, Bangladesh, Vietnam and Malaysia) is .66. These results suggest that, within monsoon Asia, when a major exporter generates a surplus, there is a good probability that other major exporters and importers will have done the same. This puts pressure on domestic markets to absorb the production variation through 108 consumption and stock adjustments, but as the econometric results in Chapter 3 indicate, part of the production variability is invariably transmitted onto the world market. As a result, net trade variations tend to occur in the same direction among these countries, causing wider price swings than would otherwise occur if world production were more geograpically dispersed. This phenomenon is shown in Figure I-6. Asian exporters' (e.g. Thailand, Burma, China) aggregate supply and demand are represented in panel (a), while importers’ aggregate supply and demand (e.g. Indonesia, Bangladesh, Vietnam, and Malaysia) are shown in panel (c). Their export supply (E50) and import demand curves (100) are shown in panel (b). All other countries’ net import demand is assumed fixed, which when added to 100, gives world import demand (ID(u)o). If the monsoons bring favorable weather throughout south and southeast Asia, supply in both the importing and exporting regions may shift to SM] and 5X1 (panels d and f), causing movements in the Asian ES and ID curves (and thus ID(H)) that reinforce rather than offset movements in world price away from its normal level (PRO-PHI in panel e). In the wheat market, production is more geographically dispersed. The five major exporters (USA, Canada, France, Australia, and Argentina) are dispersed over four continents, and the largest wheat importers (Soviet Union, China, Egypt, Japan, and Brazil) are equally dispersed geographically. The correlation coefficient between production deviations from trend among these wheat importer and exporter blocs was -0.34 between 1970 and 1987. This means that when 109 Panel (a) Panel (b) Panel (c) PH 0x1 5X0 . E50 IDo 0M0 5M0 0 Panel (d) Panel (f) 5M0 W 5M1 _¥ x. £51 5’ mum) 10ml 5x0 5x1 I 0 FIGURE I-6. Heuristic Model Representing the Effects of Correlated Production in Asian Rice Importing and Exporting Countries on Price Instability 110 the major importers as a group experienced production shortfalls and a corresponding increase in import demand, it was probable that the major wheat exporters had produced a better-than-average harvest to compensate, and vice versa. Therefore, the argument that government insulation policies have exacerbated world wheat price instability when their effects are aggregated across countries is very tenuous, because the transmission effects of exporters and importers have tended to wash out in any given year. 4.3 GREEN REVOLUTION TECHNOLOGY AND SELF-SUFFICIENCY POLICIES Available evidence suggests that, on the whole, the introduction of new rice technology has increased year-to-year fluctuations in production (Barker, Gabler, and Hinkelmann, 1981; Mehra, 1981; Hazell, 1982). Much has been written on this, and it is not examined here. Instead, the following examines a much greater source of market instability generated indirectly by the advent of new rice production technology in the past several decades. I Horld food shortages in the early 19705 raised the specter of mass starvation in many developing countries that hitherto had relied greatly on the world market to relieve domestic production shortfalls. These events set in motion self-sufficiency policies among many developing countries. However, such policies, and the green revolution technology that operationalized them, created very different outcomes for world rice and wheat trade. In the case of wheat, the new technologies were widely applicable to the climates of traditional exporters as well as importers, and therefore did little to change the 111 direction or stability of wheat trade. However, advances in rice production technology generally favored the major rice importing regions. This was because the new technology performed best in irrigated areas with good water control rather than in the major river deltas, the traditional source for export rice (Barker, Herdt, with Rose, 1985). As a result, the major beneficiaries have been traditional import markets such as Indonesia, Korea, Malaysia, Sri Lanka, and India, as well as several exporters (China and Pakistan). Yet the longstanding export centers of Thailand and Burma produce rice largely under recessional flooding with limited water control, and thus had been unable to use the new technology to its full potential. This technology bias toward traditional importers has produced both direct and indirect effects on the volume and stability of rice trading patterns: Hhen the technology proved successful, the importing countries instigated rice programs designed to promote self- sufficiency, the ultimate security against an unstable market. The exporters, on the other hand, saw little to be gained from promoting the new technology (which initially performed relatively poorly in their environment) or increasing production, since many of the traditional importers appeared to be moving toward self-sufficiency. Thailand, for example, was more concerned with maintaining stable domestic prices than with promoting production for exports (Barker, Herdt, with Rose, 1985, p. 195). Hence, the Asian beneficiaries of the new rice technology have inadvertently generated much uncertainty in the rice market, especially in the long-term. In the short run, large countries such as Indonesia and Vietnam have transmitted a moderate to high proportion of their production variability into trade variability. These countries are 112 also among the group that move from importer to sporadic exporter and back in an unpredictable way. The growing self-sufficiency among once- large importers such as India, Indonesia, Korea, and Sri Lanka has, ceteris paribus, depressed world import demand and prices over the long run. Hhile this has been largely offset by soaring demand in the Middle East and Africa, self-sufficiency policies have been undertaken in many of these countries as well. It is uncertain where and what the world demand for rice will be IO-15 years in the future, and this uncertainty may well affect the rate of production investments in traditional exporting countries. The long run uncertainty associated with the combination of new rice production technology (which has altered the traditional structure of comparative advantage) and self- sufficiency policies may produce very regrettable consequences should world prices rise sharply again in the future. This could arise if incomes rise rapidly (particularly in India and Chinazo), while potentially productive investments in exporting countries are shunned today because of the risks and costs of surplus production under uncertain future world demand. 20 Because China and India consume one-third and one-sixth of the world’s rice production, respectively, shifts in their demand relative to supply can create substantial reverberations on the world food system. Timmer and Jones (1986) note that national income levels in China are at a point where food and feed grain demand typically rise very sharply with additional increases in income. Since China’s per capita meat consumption is among the lowest in the world, rising incomes may cause very large increases in world feed grain demand, and this would probably have spillover effects in the rice and wheat markets. 113 4.4 PETROLEUM PRICES AND RICE IMPORT DEMAND Since the late 1970s, OPEC countries have increasingly dominated rice import demand. In fact the leading four importers during the 19805 are Iran, Iraq, Indonesia, and Nigeria, with Malaysia, Saudi Arabia, and the Soviet Union not far behind. Rice consumption in these countries may be affected by petroleum revenue because it commonly accounts for over 90 percent of national export revenue. This suggests a possible relationship between the world demand for rice and the price of petroleum.21 To examine this relationship, the real price of petroleum was added to the list of regressors in structural equation (3) in the previous section. The variable thus appears in the estimated trade equation (8), and was subjected to an F-test. Countries examined were Iran, Iraq, Indonesia, Nigeria, and the Middle East region. In the case of Nigeria, the petroleum price variable was multiplied by a dummy variable equaling zero until 1975, to account for the period in which Nigeria obtained little or no oil revenue. The F-tests showed a significant relationship between the real price of petroleum and net imports at the 5 percent level in the cases of Iran, Nigeria, and the Middle East region. This is not suprising since both Iran and Nigeria, which earn over 90 percent of their export revenue from oil, have engaged in extensive countertrade agreements involving the exchange of oil for rice and other grains. Rising oil prices, ceteris paribus, enable them to afford greater food imports. The real petroleum price 21 Ideally, petroleum revenue would be better, but this data was unavailable; petroleum price (f.o.b. Ras Tunuras, Saudi light crude) was used as a proxy. 114 elasticities of rice import demand were calculated as .87, 1.93, and .41 for Iran, Nigeria, and the Middle East region. Collectively, these countries have accounted for 22 percent of annual rice imports since 1980. These results suggest that rice market instability is affected not only by production fluctuations, government policy, and their interaction, but also by important demand-side shocks originating from outside the rice market. This phenomenon is not as pronounced in the wheat market because world wheat import demand is not dominated by oil exporting countries as it is in the rice market (see Table I-7). 4.5 HHEAT-RICE INTERRELATIONSHIPSZZ Hheat imports have increased substantially in almost all rice- importing countries of Asia. Among five countries that have implemented policies designed to achieve rice self-sufficiency (Indonesia, Korea, Malaysia, Sri Lanka, and the Philippines), the ratio of rice imports to wheat imports has switched from 2/1 to 1/2 over the past two decades. Nheat production has also increased dramatically in China, India, Pakistan, and Bangladesh. Factors driving this surge in Asian wheat imports include (1) urbanization; (2) market incentives and policies of developed country wheat exporters; and (3) the long run decline in wheat prices relative to rice prices since Horld Har II (Cimmyt, 1983; Barker, Herdt, with Rose, 1985). 22 This section draws mainly from Barker, Herdt, with Rose (1985). 115 There are at least three major reasons for the decline of wheat- rice price ratios. First, world wheat production has been growing faster than rice production (approximately 3.0 vs. 2.5 percent per year). Hheat production has increased rapidly especially in China, India, Bangladesh (where rice is the primary staple food in most areas), and Pakistan. These increases are largely the result of production technology advances that occurred earlier in the higher- income areas. The second reason for the decline of wheat-rice price ratios is that population in the main rice consuming areas of the world has been growing more rapidly than in the main wheat-consuming areas (2.2 vs 1.4 percent per year). Third, the income elasticites among rice consumers appear higher than among wheat consumers (Barker, Herdt, with Rose, 1985). ' The increasing substitutability between wheat and rice in food import requirements may promote stability in both world markets. For example, China, which has been observed to engage in a “wheat-rice calorie arbitrage" can increase rice exports and scale back wheat imports when world wheat prices rise relative to rice, and vice versa (Chinn, 1981; Timmer and Jones, 1986). This behavior, which promotes trade responsiveness to world prices and promotes world price stability, is made possible by some degree of substitutability in consumption between rice and wheat, at least at the national level. Hith wheat making rapid inroads into consumption patterns in a number of traditional rice-consuming countries, price instability originating in one market is more likely to be moderated by substitution behavior in the other market. CHAPTER 5 CONCLUSIONS AND POLICY IMPLICATIONS This concluding section is divided into four parts. First, the major findings of the study are summarized. Next, the feedback effects of instability on the organization and performance of the rice and wheat markets are discussed. The third section discusses the policy implications of these results, and examines the advantages and drawbacks of several alternative mechanisms to reduce the level of instability in world wheat and rice markets. Areas for further research are discussed in the final section. 5.1 SUMMARY Horld rice prices since 1960 have been significantly more unstable than those of other traded agricultural commodities such as wheat, corn, soybeans, or even petroleum. This is true in both annual and quarterly terms, using several alternative measures of instability. Prices in both the world wheat and rice markets have become substantially more volatile since the early 1970s, although they have stabilized somewhat during the last five years. Hhile much of the increased world price instability since the 19705 has been attributed to government insulation policies that sever the link between domestic and international prices, the logic and empirical evidence behind these conclusions are tenuous. It is not 116 117 valid to examine net export responsiveness to production and world price variations, and simply conclude that a high transmission effect and/or a low absorption effect are evidence that government insulation policies aggrevate world price instability. Rather, such conclusions require the following conditions: (1) the country in question must be large enough for its production and trade variations to affect the world market (alternatively, such variations must be positively correlated with aggregate world production variations); (2) the transmission effect under insulation must be greater than it would be under free trade; and (3) the absorption effect under free trade must be larger than it is under insulation. The following summarizes the empirical results of this study and assesses whether these conditions are met in the world rice and wheat markets. Conditign_(11: Only a handful of countries in the rice and wheat markets exhibit sufficient production variability and a propensity to transmit this instability abroad to appreciably destabilize the world market. Given the empirical history of national production deviations from trend, and the proportion of such deviations that are transmitted abroad, only several countries (the United States, Soviet Union, EC-lZ, and Argentina in the wheat market; Burma, Thailand, Indonesia, and Vietnam in the rice market) behave in ways that would alter world trade volumes by more than 5 percent, even during extreme years when production varies by two standard deviations from trend. However, the effects of government insulation policies may be magnified if national production fluctuations are correlated. This is because such production variations aggregated across countries may be 118 large enough to seriously affect both world import demand and export supply. This was found to be true in the rice market because of the geographical concentration of world rice production. More than 80 percent of the world’s rice is produced in a narrow swath of south and southeast Asia. The correlation coefficient of production deviations from trend between the 3 largest Asian rice exporters (Thailand, Burma, and China) and 4 major Asian importers (Indonesia, Bangladesh, Vietnam, and Malaysia) was .66 between 1970 and 1987. This means that poor harvests among these blocs (which account for over 60 percent of world rice production) tend to occur at the same time, putting greater pressures on the world market to cope with simultaneously rising import demand and falling export supplies (and vice versa in years of surplus). The geographical concentration of rice production appears to be a major source of price instability in the world rice market, and partially explains why world rice prices are more unstable than other traded agricultural commodities. Insulation policies may exacerbate this source of instability. . In the world wheat market, national production variations of major actors are much less correlated. The five major exporters are dispersed over four continents, and the largest wheat importers are equally dispersed geographically. The correlation coefficient between production deviations from trend in the wheat importer and exporter blocs was -0.34 between 1970 and 1987. This means that when the major importers as a group experienced production shortfalls and a corresponding increase in import demand, it was probable that the major wheat exporters had produced a better than average harvest, and vice 119 versa. Therefore, the argument that government insulation polices exacerbate world wheat price instability when their effects are aggregated across countries is very tenuous, because the transmission effects of exporters and importers tend to wash out in any given year. In marked contrast to the rice market, the negative covariance between production deviations of wheat importers and exporters has had a stabilizing influence on world wheat prices. Conditiog_(11: Fluctuations in trade volumes resulting from production variability do not infer that government insulation policies are the cause. Under ”free trade”, production variations are partially and invariably transmitted onto foreign markets as goods migrate from surplus to deficit areas. As noted by Bigman (1985), production instability itself influences trade flows under free trade in addition to the classical elements of comparative advantage. Therefore, the instability-related issue is whether the transmission of production instability is greater under government insulation than under free trade. The analysis in Chapter 3.6 suggests that, contrary to the conventional wisdom, a greater proportion of national production variations would be transmitted onto foreign markets under free trade than under insulation. This is because freer trade, as argued by its proponents, would make the net trade functions facing large trading countries more elastic (a result of liberalized trade by the rest of the world). Hhile such a transition would increase the ability of the world market to absorb quantity variations from an individual trading nation, it would also increase the magnitude of such quantity 120 variations. The model suggests that as the import demand (export supply) function facing an individual exporter (importer) becomes more elastic, a greater proportion of its production instability will be transmitted abroad. Moreover, the variability of production itself may increase, because domestic prices-~the expected values of which affect production decisions--may themselves become more variable under free trade. This would occur as world market shocks are increasingly allowed to affect domestic prices. Therefore, more elastic trade functions--often used as the basis for the argument that freer trade would stabilize world prices-~would have countervailing effects on price stability. andjtign_(§l: Low trade responsiveness to world price exhibited by many countries does not imply that government insulation policies are the cause, or that the removal of such policies would significantly improve such responsiveness. This is because the_short- run responsiveness of national production and demand to domestic price -- which would determine trade elasticities under free trade -- are often very low in any case. In many Asian countries, there are few substitutes in consumption for rice. Moreover, interrelated production, marketing, and information constraints impede producers’ ability to respond to current price signals, especially where physical and institutional infrastructure are weak. Therefore, government insulation policies, by partially severing the link between domestic and international prices, may greatly dampen world price responsiveness in proportional terms, but potentially very little in absolute terms. For example, although Malaysia’s price responsiveness 121 would be five times higher under free trade than under present insulation policies (from .01 to .05), the change in absolute terms is still very low. If government insulation policies have been overemphasized as a source of instability in the world rice market, an explanation is still required for the high degree of volatility in this market. The major factors include (1) the relationship between world rice demand and petroleum prices; (2) the lack of durable, reliable import markets in the rice trade; (3) self-sufficiency policies of traditional rice importers and the role of green revolution technology in achieving these ends; and (4) the geographical concentration of world rice production. Rice import demand appears highly linked to petroleum prices. This is because four of the major five rice importers are oil exporting countries. Rice imports of Nigeria, Iran, and the Middle East bloc are quite sensitive to oil prices. Therefore, rice market instability is affected not just by production fluctuations, government policy, and their interaction, but also by important demand-side shocks originating from outside the rice market. A major distinction between the world rice and wheat markets is the amount of stocks available to stabilize the market in response to production and demand shocks. Between 1980-87, the average ratio of world stocks to world production was 29 percent in the wheat market compared with only 15 percent in the rice market. The difference is partially attributable to the diametrically opposing price policies of major wheat and rice producing countries. Stock accumulation has been 122 a direct side effect of government programs in the United States and the EC, where producer prices in the 19805 have been higher than commonly used world price indicators. By contrast, major rice producing countries such as Thailand, Burma, Pakistan, Indonesia, and India have consistently taxed rice farmers and subsidized consumers; such a policy environment is not conducive to the accumulation of large stocks. None of these countries have held stocks which amount to more than 12 percent of production during the 19805. The greater proportion of wheat held in storage may mitigate upward price swings during shortage conditions better than in the rice market. Evidence elsewhere suggests that the responsiveness of prices to production forecasts are influenced by the size of available stocks (Gardner, 1975). If stocks are low, impending production shortfalls in key areas may cause prices to be bid up higher in order to ration available supplies. But if stocks are large and ”overhang the market", then world production variations may produce less price movement, because actors know that there are large stocks available to dampen any price rise (Stein and Smith, 1977). As mentioned above, the geographical concentration of production is a paramount cause of price instability in the world rice market, as it leads to positively correlated production variations in major rice importing and exporting countries. Yield variation appears to be more important than area variation as a source of rice production fluctuations in most producing countries. This is epecially true in most of the southern and south-east Asian countries affected by the Asian monsoons. 123 In the United States, on the other hand, area variability has been more important than yield variability as a source of production instability, for both rice and wheat. This appears due to periodic changes in U.S. commodity programs that greatly influence domestic production incentives. Hhile an important purpose of these policy changes has been to realign domestic prices with world market conditions, their success has been limited when examining annual U.S. production and world price movements. Econometric evidence presented elsewhere indicates that U.S. rice and wheat production have responded to lagged prices, but not current prices (Tyers and Anderson, 1988; Zwart and Meilke, 1979). Therefore, production changes in year t may have a destabilizing influence if current world market conditions are very different from the preceding year. U.S. rice production and annual f.o.b. Bangkok rice prices moved in the opposite direction five times during the volatile 1970s, and three times during the 1980s. U.S. wheat production and annual world wheat prices moved in opposite directions six times during the 19705. This has put added pressure on U.S. stocks to adjust in order to reduce the transmission of production variations onto the world market. The organization of the rice market has a profound effect on the level of world price uncertainty. First, the world rice market is very fragmented, and borders on being a misnomer. In the short-term, prices of indica and japonica varieties in various submarkets frequently move independently of one another, because of limited substitutability in consumption among the different varieties and qualities. Matching buyers with sellers of a particular type of rice is more difficult than 124 in the wheat market because (1) grades are inadequately standardized, (2) reliable price information about a particular market may be difficult and costly to obtain, especially in the short run, and (3) the rice market lacks an established trading forum in which buyers and sellers can interact with low search costs. This has created high transactions costs of international rice trading. Partial evidence of this includes the high brokerage fees charged by trading houses, as mentioned earlier. Transactions costs are also raised by unreliable and unstable trade patterns that characterize the rice trade. A number of large actors, including Indonesia, Korea, India, Brazil, and Japan, may appear as either importer or exporter in any given year. This is because such countries are now near or at self-sufficiency. Therefore, relatively small changes in domestic supply or demand can cause these countries to enter the world market as either importer or exporter. This has impeded the development of reliable long-run trade patterns, contributing to the uncertainty faced by major rice traders. Marked shifts in import markets are due to increased emphasis on self-sufficiency, new production technologies that generally favored traditional importing regions, a rising preference for rice as a staple foodgrain in African countries, the rise of petroleum revenues in large food-deficit countries to finance rice imports, and increased inroads made by wheat in consumption patterns in major rice consuming countries. Hhile it is clear that these factors related to market organization have influenced the level of instability in the rice 125 market, the causal flow is not one-way. The following section examines potential effects of instability on the institutional organization and performance of the rice market. 5.2 INSTABILITY AND MARKET ORGANIZATION The evolving structure of the rice trade is both a cause and an outcome of world market instability. Since at least Norld War 11, world rice trade as a proportion of world production has been extremely low compared with the wheat market. Comparable variations in the volume of trade therefore cause larger shocks in the rice market. This is particularly significant considering the geographical concentration of rice production in monsoon Asia. Therefore, a strong prima facie case can be made that the rise of govenment insulation policies for rice are a response to, as much as a cause of, inherent instability in the world rice market. Insulation policies may be very rational and understandable responses to an unreliable world market, although they may generate secondary effects which even further destabilize the world market. Horld market instability has also been a major motivation behind the self-sufficiency drives of numerous importers since the early 1970s. Operationalized by the green revolution rice production technologies, countries such as India, Indonesia, Korea, and Sri Lanka have approached self-sufficiency (i.e., production levels that match consumption in an average year), making the world market even more thinly traded. Due to the success of these poliCies in many countries, a structure of the world rice market has emerged in which numerous 126 countries float in and out of the world market as trade flows are dominated by random and temporary aberrations in domestic supply and demand. This situation has created high search costs, greater price variability and market thinness, each of which reinforce the others. Thin markets are often viewed as transitory phenomena, in which various factors cause participants to shift exchange from the spot market to alternative forms of coordination such as vertical integration or long run contracts (NC-117, 1978). However, Siamwalla and Haykin (1983) argue that thinness in the world rice market is a steady state. Because of many countries’ unreliable and sporadic participation in the rice trade, long-term agreements are not viable alternatives. Because trade channels are not established ruts, and widely-quoted spot prices are unavailable, transactions costs are high (in terms of both time spent identifying and negotiating a contract, and brokerage fees or marketing margins). These factors tend to further steer governments away from reliance on the world market. Siamwalla and Haykin (1983) conclude: The participants in the world rice market--importers and exporters alike-~are in an n-country version of the prisoner's dilema game. It surely is in the interests of everyone--importer or exporter--to have a well-functioning international rice market, yet each country has found it to be in its own best interests to avoid relying too much on it. This pursuit of individual interest has consequently led the rice market to become a residual market, and therefore an imperfectly functioning one. Hithout a collective and binding agreement, a movement to a more active world rice market appears to be impossible. If conditions remain unchanged, the prisoner’s dilenmIa game implies that unilateral action on the part of the exporters, say, to improve the functioning of the market, will bring no benefits to them. Hence they are disinclined to undertake such an action (p. 63). 127 In the long run, this gridlock may be relieved somewhat. A new set of actors in the Middle East and Africa may, for the foreseeable future, provide the stability on the import side of the market that is required for the development of long-run bilateral agreements with major exporters such as Thailand, Pakistan and the United States. This could reduce search costs substantially and provide a more stable trading environment for regular participants. Yet such a development would be hazardous to sporadic traders, who may have great difficulty obtaining or selling grain in an evaporated spot market. Another consequence of the unstable, thin, and fragmented nature of the rice market may be the failure of a viable internationally- accepted rice futures market to develop. Viable futures markets have always depended on sufficient volume to avoid the risk of price manipulation by large actors (Houthakker, 1959). But this presents a critical trade-off in the fragmented rice market. On the one hand, greater volume to lubricate the market may be achieved by standardizing price differentials between different varieties and grades of rice, and by increasing the number of delivery points. This would make more contracts fungible and make the futures market more attractive to speculators. However, the market’s value to hedgers depends on the correlation between the futures price and the spot price realized when the futures contract matures, and that correlation is reduced as less substitutable contracts are made interchangeable (Caves, 1978). As mentioned earlier, monthly spot prices among various sub-markets and varieties do not always move together (Falcon and Monke, 1979-80). The unpredictable and perverse basis movements in rice futures observed on 128 the New Orleans Commodity Exchange during the early 19805 (Gordon, 1984) may partially be due to contract standardization problems, which was an outgrowth of market thinness, fragmentation and uncertainty. The failure of a viable rice futures markets to develop has also been linked with (l) U.S. domestic prices being set above the world price for prolonged periods which dampened price variability needed to foster hedging and speculation; (2) the lack of “offshore” rice stocks that are immediately accessible for trading in the world market without the possibility of intervention by any government (Siamwalla and Haykin, 1983); and (3) Thailand’s historic lack of interest in the development of a rice futures market on its soil. Thailand, because of its major exporter position, has frequently been identified as a promising location for a futures market. Rice is cheaper closer to its origin and therefore requires less investment. Also, as rice exports are more concentrated than imports, major stockholdings in a prominent export market would make it easier to rechannel them to various countries as shortfalls develop (Siamwalla and Haykin, 1983). Although Thailand's rice policies have historically been geared toward domestic market stability, major changes since 1986 may make Thailand a more attractive candidate for an active rice spot and futures market. The abolition of the Thai variable export tax on rice in 1986, the reduction in direct government rice procurement, and greater incentives for private storage (Konjing, 1988) may make trading on a Thai market more attractive to foreign traders. Problems with grade and contract standardization would still need to be overcome. The recent changes in Thai rice policy may also create greater demand for risk-hedging and 129 transaction cost-reducing mechanisms. Such conditions in the major trade center of southeast Asia may foster the development of a major world spot and futures market, similar to the roles that Chicago and Kansas City play in the international wheat trade. However, the policies of other major actors in the rice market, primarily the United States, may influence the development of such a central market. The proclivity of the U.S. to heavily subsidize its rice exports in order to relieve domestic surpluses has already affected Thai perceptions concerning the profitability of sustained future investments in rice production, exporting, and marketing infrastructure (Siamwalla, 1989). The prospects of engaging in a export market-share war with the United States and its greater treasury resources has already prompted a reallocation of private and public investment toward other higher- valued agricultural exports. Such trends would be deleterious to world rice market stability if they hampered the development of forward markets and risk-transfer mechanisms to facilitate coordination in world rice trade. 5.3 POLICY IMPLICATIONS AND ALTERNATIVE MECHANISMS TO REDUCE HORLD MARKET INSTABILITY This analysis suggests a shift in the focus of policy work on instability in international grain markets away from government price stabilization policies per se, and toward other structural and organizational features of these markets that exacerbate instability. A belief underlying many analyses of market performance suggests that much of the instability and lack of coordination within a market are due to government policies that interfere with the smooth 130 functioning of private market forces. This analysis indicates that, while such policies may have negative side effects in some cases, there are other inherent features of the rice market, unrelated to government policy per se, that generate serious coordination problems. Efforts to stabilize the market by moving toward freer trade will not address the fundamental sources--mainly structural and chronic--of the instability in the rice market: (1) the geographic concentration of world production in an area of unstable weather; (2) very low consumption responsiveness to domestic prices, which would therefore prevent national trade sensitivity to world prices from rising much under free trade; (3) the fragmented nature of the rice market, characterized by numerous, poorly standardized varieties and qualities of limited substitutability in consumption; and (4) a persistently thin market where reliable price information is difficult to obtain and costly, and where the development of alternative coordination mechanisms to reduce trade uncertainties are thwarted due to many actors’ tendency to sporadically and unpredictably float in and out of the market. The institutional development of the world wheat trade contributes to the stability of this market relative to rice. Various grades and varieties of wheat are well standardized. Price information is readily available for both current and futures quotes. These prices appear to be fairly accurate indicators of prevailing supply and demand conditions in the short run. In the long run, endemic temporal coordination problems persist. Production decisions are made with incomplete knowledge of the aggregate impact of the simultaneous decisions of producers in other regions. There are no private 131 institutions by which producers, arbitragers, consumers, and governments can assess the aggregate decisions being made by others and then modify their decisions through a series of interations that would equilibrate expected and actual prices over the long run (Marion and Rhodes, 1986). Governments have circumvented (but not solved) this problem by instituting support prices, consumer floor prices, variable levies, etc. These policies appear to be more a reSponse to existing instability than the root cause of it. Unfortunately, successive moves to self-sufficiency and domestic market insulation may produce second- round effects on the organization of the market that further destabilize it in a vicious cycle. In light of the interactive relationship between instability and government policy responses, perhaps more progress can be made toward the stabilization and coordination of world grain markets by addressing several key root causes. Given the geographic concentration of world rice production, an internationally-financed reserve with trigger prices at which to purchase and sell would moderate the extreme fluctuations that characterize the rice market. Such a scheme may largely pay for itself given a sufficiently wide price band to cover transport and storage costs, and given a pricing structure that anticipates shifts in medium- to long-term supply and demand conditions rather than basing the price band on past price trends. Such an international reserve could also be used to keep sporadic importers and exporters out of the market; by providing incentives for these countries to unload sporadic surpluses to the reserve, or purchase occasional imports from the reserve, potential dislocations to the 132 world market could be avoided or at least reduced. The transactions costs involved would also be reduced since the reserve would represent a ready buyer and seller for sporadic market participants when a great deal of trade between major importers and exporters is already tied up in bilateral contracts. In addition, efforts to standardize grades and varieties across national markets may reduce the fragmentation and transaction costs currently associated with rice trading. Lack of reliable price and trade information raises transaction costs, and is a symptom of weak spot markets that impedes price discovery. In the U.S., these problems are exacerbated by the concentration of the rice marketing industry, which is composed of several large cooperatives/millers and multinationals that have strong incentives not to disseminate price information. Policy changes to promote greater entry and participation in grain marketing-~both in the United States and globally--may facilitate the generation and dissemination of much needed price information in the rice market. Policies designed to reduce the instability of international grain markets must confront the fact that national production variations (which are inevitable for the foreseeable future) must be counteracted by some combination of adjustments in consumption, stockholdings, or trade. These three forms of adjustment connote three broad types of alternative policies. Greater adjustment in consumption and/or stocks would result from movement toward free trade (reduction in trade barriers). This is the approach being stressed by U.S. proponents of the GATT. To the extent 133 that large stockholdings are discouraged in developing countries (as they frequently are by international donors for reasons related to structural adjustment and debt reduction), production variations would increasingly be borne by consumption variations--politically risky in many developing countries. Proponents of freer trade must also consider the incidence of consumption adjustments to world price changes. It is widely acknowledged that growing affluence reduces consumers’ short-run elasticity of grain demand, in the form of both grain and livestock products (Blandford, 1983). It is highly likely that a freer-trade solution would force low-income consumers in developing countries to bear the adjustment burden, because they lack the purchasing power to sustain normal consumption levels during adverse fluctuations in national and international supply and demand. Policy analyses that focus on efficiency--usually in a static context-- often ignore the potential dislocations and losses in productivity that result from political instability. The disruptions caused by food policy reforms in Jordan (1989), Zambia (1987), the Sudan (1985), and Tunisia (1983), demonstrate that trade liberalization is not always ”low-cost". At the same time, it is clear that many resource-strapped governments in developing countries have been incapable of controlling food prices. Their attempts to do so have often resulted in a different but equally severe set of disruptions to market participants. Greater adjustments in trade (in response to production shortfalls) characterize approaches such as the International Monetary Fund’s “food financing facility“ scheme. This approach provides countries in need with additional funds to purchase imports in response 134 to domestic production shortfalls. The approach thus increases the transmission of production instability onto world market through trade. This would cause little effect on the world market if the country were small, but represents a potentially destabilizing policy if implemented in countries with large populations. In such a case, the IMF insurance approach would force consumption variations to take place in other, probably poor, countries that cannot receive IMF-type funds or cannot afford to insulate their consumers from world price shocks. This approach also cannot ensure that import deliveries arrive in time to alleviate supply shortfalls in many inland countries with poor transport networks. These countries would have to carry larger working stocks to last until imports could arrive. Greater stock adjustment in response to production variations are characterized by most buffer stock schemes and the world wheat stockholding agreements considered during the 1970s. Hhile this approach can stabilize domestic markets without transmitting production instability abroad, it is expensive and beyond the capabilities of many resource-strapped low-income countries. 5.4 AREAS FOR FURTHER RESEARCH This analysis is largely short run in scope, in that it considers the effect of government insulation on year-to-year fluctuations in prices and quantities. A longer-term time frame may provide additional and potentially contradictory insights into this issue. For example, a transition to free trade could, over the long run, bring such non- marginal changes in the incidence of world production, that the whole 135 world wheat or rice system could be organized quite differently than it is now. Japan and Korea, for instance, would be large importers of rice; the EC would be a much larger wheat import market; and wheat stocks in the United States would likely be lower than current levels, perhaps altering its role as the stabilizer of the world wheat market. The effect of such non-marginal changes was beyond the scope of this research, but is clearly relevant for longer-run coordination and planning within an interdependent world grain system. However, such a proposed analysis would entail far more than deducing the structure of comparative advantage (and implications for stability) after accounting for current levels of national protection. Present patterns of comparative advantage are the product of past policy decisons (which directly influence factor productivity, demand, resource endowments, exchange rates, and other factors commonly taken as given by standard trade theory). Therefore, the particular distribution of world wheat or rice production based on efficiency criteria is policy- and time-dependent. So, then, is the resulting level of instability in the world system. The appropriate pattern of investment in world grain production depends on how static or dynamic one wants to view comparative advantage. A potentially more tractable extension of this research would be to make world price endogenous to the analysis of absorption and transmission effects. A shortcoming of this analysis is its partial equilibrium orientation; world prices and stock levels, consumption, and trade, aggregated across countries, were not simultaneously determined. Hhile this is not crucial for the majority of actors in 136 the world wheat and rice markets (that are 'small'), it certainly is important for the major players, whose transmission and absorption effects may be determined simultaneously with world price; A third extension of this research pertains to the behavioral interactions between state trading agencies and the large international grain traders. The reliability and stability of prices "discovered“ in prominent spot and futures markets are greatly influenced by the coordinating mechanisms of the large private traders. Information on these firms’ operating practices is very scarce, and the limited information that is available may have researchers looking at the "wrong questions“, considering that the direction of inquiry is influenced by constraints on available data. There may be very high payoffs to additional descriptive research that further clarifies the operating procedures of the international grain traders, their responses to various types of government insulation policies, and their influence on the organization and stability of world grain markets. REFERENCES Abbott, P. 1988. "Estimating U.S. Agricultural Export Demand Elasticities: Econometric and Economic Issues.“ In Elasticitjgs ‘ Tr . C. Carter and H. Gardiner (eds.). Hestview Press, Boulder. Abbott, P. 1979. “The Role of Government Interferente in International Commodity Trade Models.” Amgr1g3p_gpppppl_pf MW Vol 61. No. 1. pp. 135-40- Abbott, P. 1979. "Modelling International Grain Trade with Government Controlled Markets.” r r' ral Epppppjgs. Vol 61, No. I, pp. 22-31. Bale, M. and E. Lutz. 1979. "The Effects of Trade Intervention on International Price Instability." Amerispp_fipprppl_pf Agrjgultgral Ecpnomiss. Vol. 61, No. 3, pp. 513-16. Barker, R., E. Gabler, and D. Hinkelmann. 1981. “Long-Term Consequences of Technological Change on crop Yield Stability: The Case of Cereal Grain.“ In Eppd Sgsprjty fpr Qevglppipg Cpuntries. A. Valdes (ed.). Hestview Press, Boulder, Colorado. Barker, R., R. Herdt, with 8. Rose. 1985. Ihe_31gs_£gpppmy_pf_As1p. Resources for the Future, Hashington, D.G. Behrman, J. 1968. Supply Bgspppse in Updgrdevelppgg Agrjgpltupp. North-Holland Publishing Co., Amsterdam. Bigman, D. 1985. o P Foo it under abili . Lexington Books, Lexington. Binkley, J. “Marketing Costs and Instability in the International Grain Trade.“ WW. Vol. 65, No. 1. pp. 57-63. Blandford, D. 1983. 'Instability in World Grain Markets.” Jpppp31_pf 59W Vol. 34. No. 3. pp. 379-95- Blandford, D. and N. Schwartz. 1983. 'Is the Variability of Horld Nheat Prices Increasing?“ Eppd_£p11§1. November, pp. 305-312. Bolling, C. 1988. ' Price and Exchange Rate Transmission Revisited: The Latin American Case.‘ In E1ast151L1gs_1p_1ptgppp§1pppl C. Carter and H. Gardiner (eds.). Nestview Press, Boulder. 137 138 Board of Governors of the Federal Reserve System. Various Issues. “Federal Reserve Bulletin.” Hashington, D.C. Bollerslev, T., R. Engle, and J. Hooldridge. 1988. "A Capital Asset Pricing Model with Time-Varying Covariances." Journal of Bpljtjeal Ecpnpmx. Vol 96, No. 1, pp. 116-31. Brennan, J. 1982. ”The Representation of Risk in Econometric Models of Supply: Some Observations." Australian qurpal pf MW. August. pp. 151-56- Brorsen, B.H., J-P. Chavas, and H. Grant. 1987. “A Market Equilibrium Analysis of the Impact of Risk on the U.S. Rice Industry.” r n urnal l nomic . Vol. 69, No. 4, pp. 733-39. Brorsen, B.H., J. Chavas, and H. Grant. 1985. "A Market Equilibrium Analysis of the Impact of Risk on the U.S. Rice Industry." Ameriean Joanna! pf Agrjeulpural Eepnomics. Vol. 69, No. 4, pp. 733-39. Brorsen, B.H., H. Grant, and J. Chavas. 1983. "Dynamic Relationships of Rice Prices.“ Rise Qutlpok and Situatiop. RS-42. United States Department of Agriculture/Economic Research Service. Bulk Systems International. 1980. "Grain and Derivatives-—Trade with Uncertain Constraints." Bulk Systems Ipternatiopal. CS Publications, Surrey, England, April issue. Business Meek. 1978. “The Rice Market Splits over a Futures Market." Business Heek, Feb. 13, pp 88-90. Caves, R. 1979. "Industrial Organization and the Problem of Thin Markets.“ in “Pricing Problems in the Food Industry (Hith Emphasis on Thin Markets).' M. Hayenga (ed.). N.C. Project 117 Monograph 7. Research Division, College of Agricultural and Life Sciences, University of Hisconsin, Madison. Caves, R. 1977-78. “Organization, Scale, and Performance of the Grain Egade.' Food Research Institute Studies. Vol. 16, No. 3, pp. 107- Chinn, D. 1981. 'A Calorie-Arbitrage Model of Chinese Grain Trade." t . Vol 17, pp. 357-370. CIMMYT. 1983. “World Hheat Facts and Trends.“ Report No. 2. The International Maize and Nheat Improvement Center (CIMMYT), Londres, Mexico. , 139 Conklin, N. 1982. ”An Economic Analysis of the Pricing Efficiency and Market Organization of the U. S. Grain Export System. Report # CED- 82- 61. United States General Accounting Office, Gaithersburg, Md. Conner J., 0. Henderson, P. Abbott, H. Meyers, L. Myers, J. Shaffer, and K. Schneeberger. 1988. ”The Organization and Performance of World Food Systems: Implications for U.S. Policies: Regional Research Project Statement.“ North Central Regional Research Project NC-194 mimeo. Delgado, C. 1988. “The Scope for Price Policy to Slow Down the Consumption of Imported Rice and Hheat in Nest Africa." CLD Price Policy Conference Brief. International Food Policy Research Institute, Hashington, D.C. Dalziell, I. 1985. “Sources of Agricultural Market Instability.” Unpublished Ph.D. dissertation, Michigan State University, East Lansing. Dusak, K. 1973. Futures Trading and Investor Returns: An Investigation of Commodity Market Risk Premiums. gpgr pa! of Epljtisal Econpmy. Vol. 81, November/December, pp. 1407- 29. Engle,R . 1982. 'Autoregressive Conditional Heteroscedasticity with Estimates of the Variance of United Kingdom Inflation. Eggngmelglga. Vol 50, No. 4, pp. 987- 1007. Engle, R., D. Lilien, and R. Robins. 1987. "Estimating Time Varying Risk Premia in the Term Structure: The ARCH-M Model.” Eeppppeppiea. Vol. 55, No. 2, pp. 391-407. Falcon, H. and E. Monke. 1979-80. “International Trade in Rice.” Food Research Institute Studies. Vol 17, No. 3. pp 279-306. FAO (Food and Agriculture Organization of the United Nations). Various issues. 'FAO Bulletin of Monthly Statistics.” Rome. GAO (General Accounting Office). 1985. “Controls over Export Sales Reporting and Futures Trading Help Ensure Fairness, Integrity, and Pricing Efficiency in the U.S. Grain Marketing System.“ Report # RCED-85-20. U.S. General Accounting Office, Gaithersburg, Md. Gardner, 8. 1977. 'Instability in U.S. Agriculture: Discussion.“ WW. Vol. 59. pp. 185-87. Gardner, 8. 1975. 'Price Response of Grain and Livestock Futures to Crop Production Estimates.‘I private memorandum to A. Greenspan and P. MacAvoy dated 9/29/75. Gilmore, R. 1982. ° f P s an ' r in . Longman Books, New York. 140 Gordon, D. 1984. ”Performance of Thin Futures Markets: Rice and Sunflower Seed Futures." Agrjepitupa! Ecoppmjes Researeh. Vol. 36, No. 4. Economic Research Service, United States Department of Agriculture, Nashington D.C. Grant, N., J. Richardson, B.H. Brorsen, and M. Rister. 1984. “Economic Impacts of Increased Rice Variability: A Case Study with Rice.“ W311. Vol. 36. pp- 17-27- Harrison, K. 1988. President, Harrison and Associates, personal communcation, August, 1988. Hart, 0. 1975. 'On the Optimality of Equilibrium when the Market Structure is Incomplete." Qppppa1_pf_£epppm1e_1neppy. Vol. 11, pp. 418-43. Hazell, P. 1982. 'Instability in Indian Foodgrain Production." Research Report No. 39. International Food Policy Research Institute, Hashington, D.C. Hillman, J., D.G. Johnson, and R. Gray. 1975. “Food Reserve Policies for Horld Food Security: A Consultant Study on Alternative Approaches.“ Food and Agriculture Organization of the United Nations. ESC:CPS/75/2. Houthakker, H. 1959. "The Scope and Limits of Futures Trading. In - ‘llo . . if 01o u '- . . . =rnard fpapeis_fla1ey. M. Abramowitz (ed. ). Stanford University Press, Stanford. pp 134- 59. International Monetary Fund. Various issues. "International Financial Statistics.” International Monetary Fund, Hashington D.C. International Monetary Fund. 1986. ”International Financial Statistics, Supplement on Price Statistics.” Hashington, D.C. International Monetary Fund. 1985. "International Financial Statistics, Supplement on Exchange Rates.“ Hashington, D.C. International Monetary Fund. 1984. 'International Financial Statistics, Supplement on Output Statistics.“ Hashington, D.C. International Rice Research Institute (IRRI). 1986. prlg_31ee Statisties. International Rice Research Institute, Manila. International Hheat Council. Various issues, 1960-1986. Hh a Statistjes. International Hheat Council, London. International Hheat Council. 1976. “The Grain Economy of the Union of the Soviet Socialist Republics.’ in Beyiew at the prld Grain Signatipp. International Hheat Council, London. pp. 77-85. 141 International Hheat Council. Various issues. Review of the Hprld firaip_§itpatipp. International Hheat Council, London. Ito, S., E. Hailes, and H. Grant. 1984. “Asian Rice Elasticities.” in Bite Qgtlppt ang Sjtgation Report RS-QS. USDA/ERS, Washington, D.C. Jabara, C. 1982. "National Grain Policies." F P c . February pp. 39-45. Johnson, D.G. 1984. ”Domestic Agricultural Policy in an International Environment: Effects of Other Countries’ Policies on the United States.” an I of ic tural n mi . Vol. 66, No. 5, pp. 735-44. Johnson, D.G. 1975. 'Horld Agriculture, Commodity Policy, and Price Variablility.‘ Amerjeap Jpgrpal pr Agricultpral Eepnpmjes. Vol 57, p. 823-28. Josling, T. and R. Barichello. 1984. "International Trade and Horld -Food Security.“ Epod Epliey. November, pp. 317-330. Just, R. 1987. "Economic Efficiency and Helfare Measurement in a Dynamic, Uncertain, Multimarket World." In Eeonomic Effieiency in ur n 00 Ma k n . R. Kilmer and N. Armbruster W (eds). Iowa State University Press. Just, R. 1975. ”Risk Response Models and Their Use in Agricultural Policy Evaluation.” Amerjeap Jpppnal of Agriepltpral Eeppomjes. Vol 57, No. 5, pp. 837-43. Just, R. 1974. 'An Investigation of the Importance of Risk in Farmers’ Decisions." Amepieap_Jppppal of Agricultural Eeppomies. Vol 56, No. 1, pp. 14-25. Konjing, C. 1988. “The Rice Policy of Thailand." Unpublished working paper. Department of Agricultural Economics, Kasetsaart University, Bangkok. Koopman, R. 1986. Soviet Union Analysis in "Global Review of Agricultural Policies.“ Staff Report AGE5880304. USDA/ERS. Lattimore, R. 1982. 'Norld-Grain-Market Instability and Food Security: An Exporter’s Perspective.“ In Eppd Seeupity; Ipepry, Epliey, I E II E E i I I! E 'E' B 1 im. A. Chisholm and R. Tyers (eds). Lexington Books, Lexington. Lattimore, R. 1974. 'An Econometric Model of the Brazilian Beef Ecznomy.‘ Ph.D. Thesis. Purdue University, H. Lafayette, n iana. 142 Mahama, R., and W. Meyers. 1985. "World Wheat Market Instability: An Analysis of EC Price Insulation Effects.“ In ri tur ’ d ono . A. Maunder and U. Renborg (eds.). Gower Publishing Co., University of Oxford, Oxford. Marion, 8., and V.J. Rhodes. 1986. "Summary and Policy Implications.” In W. 8. Marion (ed.). Lexington Books, Lexington, Ma. Martin, F. and E. Crawford. 1988. ”Analysis of Alternative Producer Price Policies in Senegal using a Micro-Macro Modelling Approach." Staff Paper No. 88-18. Department of Agricultural Economics, Michigan State University, East Lansing. Massel, B. 1969. “Price Stabilization and Welfare." Qparterly qurnal fp Eeonpmjes. Vol. 38, pp. 285-97. McCalla, A. 1979. ”Strategies in International Agricultural Marketing.“ In 1 e A It : Theorysapd Prestige. J. Hillman and A. Schmitz (eds). Westview Press, Boulder. McCalla, A. and T. Josling. 1985. Agrjc cultpr al Eplieies and Wprlg Markets. MacMillan Publishing Co. , New York. Mehra, S. 1981. 'Instability in Indian Agriculture in the Context of New Technology.” Research Report No. 25. International Food Policy Research Institute, Washington, D.C. Miller, T., J. Sharples, R. House, and C. Moore. 1985. "Increasing World Grain Market Fluctuations: Implications for U.S. Agriculture.“ r u l omi Re ort No. 541. UDSA/ERS, Washington, D.C. Mitchell, D. and R. Duncan. 1987. "Market Behavior of Grain Exporters." Researeh Observer. The World Bank, Washington, D.C. Vol 2, No. 1, pp. 3-22. Morgan, 0. 1979. Merehapts_pf_§ra1p. Viking Press, New York. Myers, R. and J. Oehmke. 1987. 'Instability and Risk as Rationales for Government Intervention in Agriculture.” draft mimeo. Michigan State University, E. Lansing. NC-117. 1979. l'Pricing Problems in the Food Industry (with Emphasis on Thin Markets).' M. Hayenga (ed.). North Central Regional Research Publication 261. College of Agriculture and Life Sciences, University of Wisconsin, Madison. Newbery, D. and J. Stiglitz. 1981. Ihe_Iheprx_p£_§pmmpdit1_£r1ee mmwmmmmmsk Clarendon Press, Oxford: 143 Oi, W. 1961. “The Desirablility of Price Instability under Perfect Competition.” Econometrica. Vol. 29, pp. 58-64. Paarlberg, P. 1988. "Price Intervention by Exporting Countries and the Effects on Price Instability.” r r Agrjegltural Ecoppmjes. Vol. 10, No. 1, pp. 63-74. Peck, A. and R. Gray. 1980. “Grain Reserves: Some Unresolved Issues.‘I food Polity. Vol. 5, No. 1. Peterson, V. Vice President, U.S. Wheat Associates, personal communication, July, 1988. Petzel, T. and E. Monke. 1979-80. ”The Integration of the International Rice Market." Eppp_Besearep_1pstjtpte_§tpgjes, Vol. 17, No. 3. Pindyck R. and D. Rubinstein. 1981. nom t ic Mo and onomi Epreeasts. Second edition, McGraw-Hill Book Company, New York. Rastegari-Henneberry, S. 1985. “The World Rice Market." Giannini Foundation Information Series No. 85-2. University of California, Davis. Reardon, T. 1988. "Analysis of Grain Consumption in Burkina Faso," draft mimeo. International Food Policy Research Institute, Washington, D.C. Rogers, 8. and M. Lowdermilk. 1988. Food Prices and Food Consumption in Urban Mali.” Presented at "Beyond Mindelo: Food Security Policy and Competitiveness of Agriculture in the Sahel. International Food Policy Research Institute, October 1988, Washington D.C. Rowan, R. 1981. '...Hands of High Rollers." fortune. April 20, pp. 148-59. Sarris, A. and J. Freebairn. 1983. “Endogenous Price Policies and International Wheat Prices.“ n ur l f r t r Egpppmigs. Vol 65, No. 2, pp. 214-24. Scandizzo. P. and C. Bruce. 1980. “Methodologies for Measuring Agricultural Price Intervention Effects.“ Wpr1d_fiapt_§tafifi W Washington. D.C. Schmitz, A. 1986. I'Marketing Institutions in International Commodity Markets.“ h o r tu l n mi . July. pp. 41-48. Schmitz, A., A. McCalla, D. Mitchell, and C. Carter. 1981. Graln Market tartels. Ballanger Books, Cambridge. 144 Schrader, L., M. Hayenga, D. Henderson, R. Leuthold, and M. Newman. 1986. "Pricing and Vertical Coordination in the Food System." in The Qrgapjzatioh app Eerfprmange pf the U.S. Fpod System. 8. Marion (ed). Lexington Books, Lexington. Scobie, G. 1983. "Food Subsidies in Egypt: Thier Impact on Foreign Exchange and Trade.“ d l Re ut MW- Washington. D.C. Shaffer, James D. 1989. Professor, Agricultural Economics, Michigan State University, personal communication. Shaffer, James O. 1987. “Does the Concept of Economic Efficiency Meet the Standards for Truth in Labeling When Used as a Norm in Policy Analysis." Eephppje Effjejepey jh Agrieultgre and Eggs Marketjhg. R. Kilmer and W. Armbruster (eds). Iowa State University, Ames. Shaffer, James D. 1969. 'On Institutional Obsolescence and Innovation--Background for Professional Dialogue on Public Policy.“ Amerjean qurpal pf Agrieultpral Eeonomics. Vol. 51, No. 2, pp.245-267. Shei, S. and R. Thompson. 1977. ”The Impact of Trade Restrictions on Price Stability in the World Wheat Market." Aperiean Jpprnal pf Agrjegltgral Eepnomjes. Vol 59, No. 4, pp 629-38. Siamwalla, A. 1989. Personal communication to Dr. M.T. Weber. Siamwalla, A. and S. Haykin. 1983. “The World Rice Market: Structure, Conduct and Performance." International Food Policy Research Institute Research Report #39. Washington, D.C. Skold, K. and W. Meyers. 1987. “Forecasting Government Responses to Changing Economic Conditions in Regional Commodity Trade Models." paper presented at the NCR Conference on Applied Commodity Price Analysis, Forecasting, and Market Risk Management, Chicago, April, 98 . Slayton, T. 1984. ”Some Pieces of the World Rice Puzzle.“ . RS-43, March 1984. USDA/ERS, Washington, D.C. Sorenson, V. and E. Rossmiller. 1983. ”Future Options for U.S. Agricultural Trade Policy.“ WW Etphpmies. Vol. 65, No.5, pp. 893-900. Stein, J. and R. Smith. 1977. “The Economics of U.S. Grain Stockpiling.“ Report Prepared of the Council on International Economic Policy. R-1861. Rand Corporation, Santa Monica. 145 Streeten, P. 1987. What Prjee Fpog: Agricultural Price Policies in W n l . St. Martin’ s Press, New York. Thurston, S., K. Phillips, M. Haskell, E. James, and D. Volkin. 1976. ”Improving the Export Capability of Grain Cooperatives.“ FCS-34. USDA Farmer Cooperative Service. GAO Printing Office, Washington, D.C. Timmer, C. P. and B. Jones, 1986. "China: An Enigma in the World Grain Trade. in East J. Jones (ed). Westview Press, Boulder. Tolley, G., V. Thomas, and C. Wong. 1983. Agrjepltpra1_£rjee_£plje1es apd_the_Qeyelppjhg_§pghtrjes. Johns Hopkins University Press, Baltimore. Traill, B. 1978. ”Risk Variables in Econometric Supply Response Models." Jpgrnal of Agrjeultgral Egpnpmies. Vol 28, No. 1, pp. 53-61. Tyers, R. and K. Anderson. 1988. ”Imperfect Price Transmission and Implied Trade Elasticities in a Multi-commodity World." In in nt r l ric ur Tr . C. Carter and W. Gardiner (eds. ). Westview Press, Boulder. U.S. Department of Agriculture, 1988. Production, Stocks, and Disappearance Database. Commodity Economics Division, Economic Research Service, USDA, Washington, DC. U.S. Department of Agriculture). 1986. Global Review of Agricutural Policies.” Staff Report AGES880304. USDA/ERS, Washington, D.C. Waugh, F. 1944. ”Does the Consumer Benefit from Price Instability?" r r urn f n m . Vol. 58, pp. 604-14. Webb, A. and B. Jones. 1986. "The Effects of the U.S.-Soviet Bilateral Trade Agreement on the World Market: Implications for United States Policy." in East-West Agrjegltural Trade. J. Jones (ed). Westview Press, Boulder. Weber, M., J. Staatz, E. Crawford, J. Holtzman, and R. Bernsten. 1988. “Informing Food Security Decisions in Africa: Empirical Analysis and Policy Dialogue.“ Staff Paper No. 88-58. Department of Agricultural Economics, Michigan State University. Wilde, L., L. Cornell, V. Sorenson, and R. Black. 1985. ”World Grain Trade: An Evaluation of Factors Affecting Net Import Demand for Wheat and Coarse Grains by Selected Countries. Agrieultural R . No. 487, Department of Agricultural Economics, Michigan State University, East Lansing. 146 Williamson, 0. 1975. Markets and Hierarchies: Analysis and Antitrust Inplieations. Free Press, New York. World Wheat Statistics. Various Issues. International Wheat Council. London. Zwart, A. and K. Meilke. 1979. ”The Influence of Domestic Pricing Policies and Buffer Stocks on Price Stability in the World Wheat Industry.“ ri l n . Vol 61, No. 3, pp. 434-49. APPENDICES APPENDIX 1 THE VARIANCE OF ANNUAL PERCENTAGE CHANGE MEASURE OF INSTABILITY APPENDIX 1 THE VARIANCE OF ANNUAL PERCENTAGE CHANGE MEASURE 0F INSTABILITY Because each indicator of instability is sensitive to particular movements in a data set and insensitive to others, a brief description is given of the measure commonly used in this study to measure instability: the variance of annual percentage changes. Computationally, the indicator is n VAR { 2 [(ln(xt) - ln(xt-1))*100]}. t-l where x is the price or quantity variable in question. The indicator is dimensionless, and so can be used to compare data from different commodities. The variance of annual percentage changes effectively detrends the data series in question. Thus, if the series increased by a constant percentage each year, then there would be a zero variance. The assumption behind this measure is that market participants can readily adjust to constant percentage changes each year, but they will have difficulties if period to period changes are highly variable (Dalziell, 1985). This indicator may or may not accurately represent uncertainty (this is examined further in Part II). This is because there is no a priori information to assess whether market actors anticipate the trend or other variations in a particular data series. 149 Among the indicator’s drawbacks are: (1) it gives excessive weight to outliers and hence to data errors; (2) the detrending process gives greater weight to the endpoints in the data series; (3) it assumes that agents know the long term trend in the data (Dalziell, 1985). 150 APPENDIX 2 INSTRUMENTAL VARIABLE REGRESSION RESULTS FOR HORLD WHEAT AND RICE PRICE IN DOMESTIC CURRENCY FOR USE IN EQUATION (8) FOR "LARGE COUNTRY" CASES and OLS REGRESSION RESULTS OF EQUATION (8): NATIONAL WHEAT AND RICE NET EXPORTS APPENDIX 2 INSTRUMENTAL VARIABLE REGRESSION RESULTS FOR DEFLATED WORLD WHEAT PRICE IN DOMESTIC CURRENCY FOR USE IN EQUATION (8) FOR ”LARGE COUNTRY” CASES List of Instruments for Estimation of Deflated World Wheat Price in Domestic Currency (PW) WORLD WORLD Country Constant PWt-1 PWt-2 PWt-3 ENOt-1 PROOt-1 UNITED STATES -568.9 .18 .17 -.34 -9.54 -5.91 (1963-87) (-2.71) (0.64) (0.63) (-1.47) (-3.09) (-2.88) R2 - .82 DN - 2.21 F - 6.47 SOVIET UNION 6.09 .66 -.25 -.2e -.05 -.03 (1963-87) (2.74) (2.72) (-0.93) (-1.20) (-1 51) (-2.47) R2 - .74 DN - 1.96 F - 10.59 CANADA 254.1 .71 -.32 -.30 -l.85 -.47 (1963-86) (2.50) (2.80) (-1.08) (-1.22) (-1.04) (-0 67) R2 - .62 Du - 1.93 F - 5.45 Parentheses contain t-statistics For definition and derivation of variables, see Table I-9. For Soviet Union, PW is approximated by f.o.b. U.S. gulf ports, hard red winter #2 divided by global price deflator. For discussion of instrumental variables estimation and its properties, see Pindyck and Rubinfeld (1981), pp. 178-80. 152 OLS REGRESSION RESULTS FOR EQUATION (8): WHEAT e en V iab : Net Wh t x o t kilo rams UNITED STATES CANADA SOVIET UNION FRANCE CHINA Country (1963-87) (1963-86) (1963-87) (1962-86) (1974-87) Constant -157.1 443.4 -169.4 ~231.3 -23.7l . (-2.71) (0.80) (-4.55) (-9.00) (-3.26) PRODt .59 .36 .31 .57 .08 (2.01) (1.52) (3.71) (8.66) (0.82) PRODt-1 -.28 .05 .03 .00 -.42 (-O.96) (0.18) (0.42) (0.01) (-2.94) PWt .47 -.39 16.53 1.29 2.95 (2.51) (-O.33) (1.51) (0.64) (1.50) PWt-1 .13 -.58 5.55 .54 5.63 (0.68) (-O.46) (0.47) (0.26) (2.95) GNPt 41.97 19.01 na 87.41 84.16 (0.29) (0.47) (3.92) (4.33) Rt 2.70 -7.29 -3.62 .73 -.90 (0.93) (-O.44) (-1.61) (0.35) (-2.51) Rt-1 0.64 13.12 -1.00 -.54 -.37 (0.21) (0.82) (-0.45) (-0.28) (-1.48) ENDt-1 0.69 -.15 .23 .75 .35 (2.20) (-O.59) (1.69) (2.13) (2.21) ENDt-2 -o.06 .13 .13 -.01 -.15 (-0 26) (0 42) (0.92) (-0 04) (-1 38) PW2t -- -- -- -- -- PW2t-1 -- -- -- -- -- R2 81 .38 85 98 95 OH 2.03 1.82 1.80 1.77 2.57 F 7.52 1.08 11.83 100.6 9.03 Parentheses contain t-statistics For variable definition and derivation, see Table 1-9. 153 Wh t x rts kilo ram TURKEY JAPAN PAKISTAN KOREA NORTH AFRICA Country (1971-84) (1962-87) (1962-87) (1962-86) (1962-87) Constant -98.11 -54.25 -66.02 41.99 -6.77 (-2.00) (-10.28) (-5.42) (1.81) (-0.52) PRODt .34 1.13 .44 .77 .74 (3.68) (6.18) (4.51) (0.39) (6.45) PRODt-1 -.04 .21 -.00 -5.99 .36 (-0.18) (1.36) (-0.01) (-2.72) (3.33) PWt .0001 -.002 -.19 -.0004 -15.59 (2.05) (-0.40) (-O.93) (-0.10) (-2.19) PWt_1 .001 .004 .15 -.008 -17.35 (0.87) (0.85) (0.82) (-1.74) (-1.96) GNPt -16.46 -1.08 4.78 -31.25 -3.18 (-1.95) (-0.17) (0.30) (-5.66) (-8.50) Rt -1.29 .03 .37 .42 .48 (-0.88) (0.16) (0.53) (0.42) (1.16) Rt-1 -1.20 -.14 -.41 1.00 .78 (-1.02) (-0.60) (-0.63) (1.07) (2.08) ENDt-1 .57 .36 .05 .32 .46 (2.20) (1.14) (0.22) (0.36) (1.20) ENDt-2 -.08 -.24 -.04 -1.28 -.25 (-0.45) (-O.80) (-0.17) (-l.42) (-0.82) PW2t -- -- -- -- 2.35 (1.83) PW2t-l -- -- -- -- 2.70 (1.75) R2 .84 96 85 .89 ..99 OH 1.95 2.19 2.12 1.65 2.14 F 8.69 35.17 10.00 15.82 119.48 Parentheses contain t-statistics For variable definition and derivation, see Table 1-9. GNP for North Africa approximated by time trend na: data not available 154 Dependent Variable; Per tapita net wheat exports (kilograms) ARGENTINA AUSTRALIA EC-IZ BRAZIL EAST EUROPE Country (1971-85) (1962-85) (1962-87) (1962-86) (1962-87) Constant -426.8 ' -428.5 -113.7 -44.90 -78.46 (-2.l7) (-4.l9) (-l4.35) (-5.89) (-3.29) PRODt 0.93 .56 .43 .89 .35 (15.37) (4.98) (7.36) (7.59) (4.05) PRODt-1 .17 .39 -.18 -.34 .13 (2.20) (1.91) (-1.56) (-2.64) (1.37) PWt .37 .03 .01 .001 -8.50 _ (1.73) (0.03) (0.47) (3.95) (-1.71) PWt-1 -.22 .60 -.009 -.00007 -1.65 (-1.38) (0.70) (-0.38) (-O.18) (-O.35) GNPt 35.07 27.41 128.4 -12.03 -2.48 (0.71) (0.72) (3.07) (-1.19) (-2.21) Rt .05 -15.25 .60 -.89 .19 (0.02) (-1.22) (1.00) (-2.51) (0.15) Rt-1 6.63 2.98 1.14 -.01 .20 A(1.93) (0.72) (1.61) (-0.02) (0.16) ENDt-1 .45 -.11 .96 2.12 -.34 (1.13) (-0.41) (3.04) (4.38) (-O.65) ENDt-2 1.12 .43 .06 -.09 -1.24 (2.00) (1.71) (0.33) (-O.22) (-1.94) PW2t -- -- -- -- -- PW2t-1 -- -- -- -- -- R2 99 .82 98 90 80 OH 2.54 1.40 1 72 1.83 1.62 F 41.46 6.98 88.07 12.33 7.14 Parentheses contain t-statistics . For variable definition and derivation, see Table 1-9. GNP for East Europe approximated by time trend. GNP for EC-IZ approximated as unweighted average of GNP of Germany, France, Italy, and the U.K. na: data not available 155 erendent_!ariable; Per capita net wheat exoorts (kilograms) SUB-SAHARAN AFRICA MIDEAST EGYPT INDIA Country (1962-87) (1962-87) (1962-87) (1962-87) Constant -7.28 -73.21 -127.0 -17.03 (-1.61) (-l.84) (-2.00) (-4.58) PRODt .88 .38 .27 .26 (2.42) (2.37) (0.30) (1.76) PRODt-1 -.09 .04 .98 -.11 (-0.29) (0.20) (0.91) (-0.39) PWt .64 6.79 8.58 .04 (2.88) (1.48) (0.47) (0.15) PWt-1 .03 -7.13 3.76 -.10 (0.18) (-1.51) (0.22) (-0.44) GNPt na -1.82 na -18.82 . (-3.20) (-0 58) Rt -.24 -1.54 -1.61 .65 (-3.39) (-1.17) (-0.51) (1.60) Rt-1 .12 .85 -3.29 -.28 (1.35) (0.73) (-1.33) (-0.55) ENDt-1 3.60 .10 -.93 .48 (4.04) (0.28) (-I.22) (1.09) ENDt-2 -2.60 .05 -.33 .10 (-2.50) (0.17) (-0.37) (0.49) PW2t -- -- -- -- PW2t-1 -- -- -- -- R2 98 .82 76 83 OH 1.87 1.86 0.68 1.18 F 95.35 8.14 6.87 8.20 Parentheses contain t-statistics For variable definition and derivation, see Table 1-9. GNP data for Mideast region approximated by a time trend na: data not available 156 APPENDIX 2 INSTRUMENTAL VARIABLE REGRESSION RESULTS FOR WORLD WHEAT AND RICE PRICE IN DOMESTIC CURRENCY FOR USE IN EQUATION (8) FOR “LARGE COUNTRY” CASES and OLS REGRESSION RESULTS OF EQUATION (8): NATIONAL WHEAT AND RICE NET EXPORTS APPENDIX 2 INSTRUMENTAL VARIABLE REGRESSION RESULTS FOR DEFLATED WORLD RICE PRICE IN DOMESTIC CURRENCY FOR USE IN EQUATION (8) FOR "LARGE COUNTRY” CASES List of Instruments for Estimation of Deflated World Price in Domestic Currency (PW) WORLD WORLD Country Constant PWt-1 PWt-2 PWt-3 ENDt-1 PRODt_1 UNITED STATES 1529 .75 -.50 .09 25.2 -25.6 (1961-87) (3.33) (3.80) (-2.12) (0.43) (1.74) (-2.93) R2 - .67 OW -1.81 r - 7.6 THAILAND 21615 .99 -.74 .26 254.6 -336.9 (1961-87) (2.58) (4.82) (-2 94) (1.24) (0.95) (-2.14) R2 - .70 OH -2.05 F - 8.7 CHINA 4.66 .82 -.54 .18 .21 -.06 (1961-86) (0.78) (3.44) (-l.98) (0.69) (0.96) (-0.45) R2 ' .49 OH -2.07 F - 3.3 Parentheses contain t-statistics For definition and derivation of variables, see Table I-9. 157 Constant PRODt PRODt-1 PWt OLS REGRESSION RESULTS FOR EQUATION (8): Dependent Variable: United States (1963-87) .97 (0.29) .38 (2.81) .03 (0.23) .008 (3.50) -.000 (-.02) -9.03 (-O.81) 0.34 (2.59) -O.4O (-2.89) 0.55 (2.42) -0.02 00.18) .81 1.96 7.05 Thailand (1963-86) -29.4 (-O.58) .36 (3.16) .17 (1.45) -.002 (-1.62) -.002 (-1.52) 4.01 (0.47) -.83 (-0.62) 2.08 (1.49) -.35 (-0.97) -.30 (-O.92) 1.90 11.43 China (1963-85) -2.17 (-1.21) .01 (0.46) .04 (1.57) .19 (2.04) .05 (0.51) 1.87 (0.12) .12 (2.12) -.17 (-2.38) -.10 (-2.51) -.02 (-0.55) .77 2.31 4.88 Per capita net rice:exnorts (kilograms) Bangladesh (1974-86) 57.2 (0.61) -.15 (-0.66) -.34 (-1.45) .0008 (1.11) .0006 (0.90) 7.88 (0.23) -.70 (~1.18) .16 (0.27) 1.14 (1.76) -.54 (-1.16) RICE Pakistan (1962-87) -37.8 (-1.93) -.01 (-0.02) .42 (1.60) .005 (2.92) .001 (0.56) 20.2 (2.09) -.86 (-1.81) .38 (0.93) .05 (0.09) -.25 (-0.44) -.0000003 (~2.64) -.0000001 (-1.18) .83 1.78 6.03 Parentheses contain t-statistics For variable definition and derivation, see Table 1-9. 158 Dependent Variaple: Per eapjta net rice exports (kilograms) Indonesia EC-12 Brazil Soviet Union Nigeria Country (1971-84) (1962-87) (1964-86) (1962-86) (1962-87) Constant -.24 -1.78 -4.98 -.63 6.56 (-1.41) (-1.58) (-0.45) (-0.61) (1.97) PRODt .45 1.25 .07 -.09 -1.04 (2.68) (6.82) (0.56) (-0.19) (-3.20) PRODt-1 .02 -.27 .02 .01 .61 (0.08) (-1.60) (0.14) (0.01) (1.13) PWt .0003 .008 .004 .002 .001 (0.22) (2.28) (0.36) (1.79) (0.33) PWt-1 -.001 -.01 .01 -.001 .0003 (-O.99) (-3.83) (1.04) (-l.13) (0.08) GNPt -62.56 -15.5 -8.26 na -.11 (-1.56) (-4.08) (-1.18) (-0.93) Rt -.41 .03 .08 -.29 -.33 (-0.54) (0.66) (0.19) (-2.55) (-O.80) Rt-1 0.52 0.06 -.20 .02 .30 (0.53) (1.50) (-0.50) (0.18) (0.86) ENDt-1 .60 1.06 .43 na -.03 (0.65) (2.75) (1.62) (-0.05) ENDt-2 -.04 1.15 .20 na .88 (-0.07) (2.64) (0.85) (1.51) PW2t -- -.OOOOO7 -- -- -- (-2.20) PW2t-1 -- .000009 -- -- -- (4.25) R2 93 .86 59 51 77 DW 2.72 .92 1.51 1.53 2.01 F 6.21 8.13 1.89 2.51 5.15 Parentheses contain t-statistics For variable definition and derivation, see Table I-9. GNP for EC-IZ, approximated by unweighted average of GNPs of Germany, France, Italy, and the U.K. na: data not available 159 na: data not available 160 en r'abl ' P r a ° n t r' e x r il r Japan India Korea Burma Senegal Country (1962-87) (1962-87) (1962-86) (1962-86) (1962-87) Constant -55.0 -8.86 19.62 19.08 -118.9 (-2.75) (-4.19) (0.49) (0.49) (-3.36) PRODt 0.13 -.02 .28 .49 .23 (1.77) (-O.89) (1.55) (3.24) (0.48) PROOt-1 .19 .02 -.30 .10 .25 (1.34) (0.97) (-1.57) (0.54) (0.64) .PWt .00004 .02 .001 .04 -.03 (0.02) (0.75) (0.45) (0.30) (-3.31) PWt-1 -.0001 .01 -.001 -.38 -.01 (-0.07) (1.15) (-0.43) (-0.50) (-1.00) GNPt 23.05 52.6 -10.20 -30.76 4.24 (3.50) (3.55) (-1.32) (-2.37) (2.70) Rt .12 .07 -2.34 -1.89 -.62 (0.36) (0.90) (-1.77) (-0.89) (-0.83) Rt-1 .11 -.12 -1.25 -.86 1.00 (0.32) (-1.34) (-0.86) (-0.45) (1.25) ENDt-1 -.04 -.01 .51 na na (-0.38) (-O.21) (1.50) ENDt-2 .04 .11 .70 na na (2.19) (2.36) (1.64) PW2t -- -- -- -- .00002 (4.78) PW2t-1 -- -- -- -- .000005 (1.41) R2 .85 .82 .73 .64 .77 DW 2.04 1.60 2.33 .96 2.01 F 9.99 7.84 4.45 4.41 9.15 Parentheses contain t-statistics For variable definition and derivation, see Table 9. Dependent Variable; Country Constant PRODt PRODt-1 PWt PWt-1 GNPt Iran Per ca ita Iraq ' e x ort Malaysia Mideast ki o rams Sub-Saharan Africa Vietnam. (1962-84) (1962-87) (1962-86) (1962-86) (1962-87) (1962-86) -3.47 (-1.61) .24 (0.56) .33 (1.33) .07 (3.01) -.05 (—1.69) -0.97 (-0.54) -.47 (~1.28) -.32 (-0.97) -.08 (-0.23) -.35 (-0.83) .85 2.51 413.85 -43.77 (-5.33) .62 (2.23) .35 (1.32) .08 (2.31) .03 (1.01) na -2.60 ('3.33) 16.02 -85.50 (-4.43) .34 (1.85) .18 (1.11) .001 (0.07) -.007 (-0.40) na 1.10 (1.00) -O.62 (-0.48) -.06 (-0.22) 1.56 4.30 -10.03 (-2.68) .92 (2.24) .56 (2.14) .008 (0.92) -.02 (-2.77) -.50 (-2.86) -.11 (-0.57)' .06 (0.42) 1.08 (2.30) -.87 (-1.84) -.000006 (-0.82) 50.33 -2.41 (-O.48) .19 (0.40) .25 (0.64) .002 (2.18) -.01 (-1.45) na .84 1.83 13.82 -149.4 (-5.21) .45 (2.98) .17 (1.22) .09 (2.20) .0006 (0.05) na -.18 (-O.18) 2.22 (2.36) na na Parentheses contain t-statistics For variable definition and derivation, see Table 9. GNP data for Mideast approximated by a time trend data not available na: 161 APPENDIX 3 REGRESSION RESULTS OF DOMESTIC PRICE EQUATION (5b) FOR SEVEN MAJOR RICE TRADING COUNTRIES Dependent Variable: APPENDIX 3 REGRESSION RESULTS OF DOMESTIC PRICE EQUATION (5b) FOR SEVEN MAJOR RICE TRADING COUNTRIES Deflated domestic consumer rice price (PD) Country (1963-87) 159.4 (1.86) .96 (9.21) -s.04 (1.27) 2.14 (0.35) .89 1.13 50.07 United States Thailand (1962-86) 3411 (1.44) .20 (3.58) 5.19 (0.61) .33 (-1.62) .41 1.31 4.82 Bangladesh (1974-82) -7495 (-0.52) .73 (2.01) 58.27 (0.68) 158.8 (0.60) .47 1.98 1.45 Pakistan (1974-86) 2211 (1.42) .64 (2.49) 14.8 (0.56) -l69.5 (-1.99) -.00004 (-2.03) .68 1.42 9.19 Parentheses contain t-statistics For variable definition and derivation, see Table 9. 163 Dependent Variable: Deflated domestic consumer rice price (PD) Country Constant Indonesia (1971-82) 3613 (3.50) .05 (0.94) -12.31 (-1.26) -29.05‘ (-O.80) Malaysia (1971-82) 414.4 (1.94) .21 (3.20) 6.68 (3.58) -6.99 (-2.11) 22.39 Burma (1961-82) 3334 (2.54) .53 (2.19) -16.08 (-2.67) na .39 1.77 6.00 Parentheses contain t-statistics- For variable definition and derivation, see Table 9. 164 APPENDIX 4 REGRESSION RESULTS OF NATIONAL CONSUMPTION AND ENDSTOCK EQUATIONS (2) AND (3b) FOR SEVEN MAJOR RICE TRADING COUNTRIES APPENDIX 4 REGRESSION RESULTS OF NATIONAL CONSUMPTION EQUATION (2) FOR SEVEN MAJOR RICE TRADING COUNTRIES Dependent Variable: National per capita consumption in kilograms (CON) United States Thailand Bangladesh Pakistan Country (1963-86) (1962-86) (1974-82) (1974-86) -Constant -1.27 174.9 -7495 2211 (-0.72) (3.83) (-0.52) (1.42) PDt -.001 .004 .73 .64 (-1.11) (0.65) (2.01) (2.49) GNPt 22.4 -11.77 58.27 14.8 (6.15) (-1.57) (0.68) (0.56) R2 .72 .15 .47 .68 OW 1.38 1.70 1.98 1.42 F 26.14 1.83 1.45 9.19 Parentheses contain t-statistics For variable definition and derivation, see Table 9. 166 Dependent Variable: National per capita consumption (kilograms) Indonesia Malaysia’ Burma” Country (1971-82) (1971-82) (1961-82) Constant 163.7 173.9 6.70 (4.72) (5.18) (0.12) PDt -.08 .003 -.01 (-2 69) (0.18) (-1.81) .GNPt 98.97 -3.40 71.86 (6.63) (-3.78) (3.85) PDZt .00002 -- -- (2.53) R2 .97 .86 .82 OH 2.39 2.02 1.92 F 82.23 16.13 28.00 arentheses contain t-Statistics Cochrane-Orcutt correction for serial correlation For variable definition and derivation, See Table 9. 167 REGRESSION RESULTS OF NATIONAL ENDSTOCK DEMAND EQUATION (3b) FOR SEVEN MAJOR RICE TRADING COUNTRIES Dependent variable: National per capita endstocks in kilograms (END) United States Thailand Bangladesh Pakistan Country (1963-87) (1962-86) (1974-82) (1974-86) Constant .66 -26.63 10.91 17.45 (0.13) (-1.35) (1.04) (3.66) -PDt -.05 -.003 - 001 -.003 (-2.96) (-1.06) (-1.35) (-2.94) PDt-1 .05 .009 -.0001 .0001 (2.73) (3.16) (-O.16) (0.17) Rt -.13 .63 .17 .12 (-0.64) (0.71) (0.52) (0 71) Rt-1 .28 -.65 .24 -.26 (1.30) (-0.77) (0.66) (-1 47) ENDt-1 .63 .50 ~-.52 -.18 (3.29) (3.13) (-1.14) (-0.62) PDZt .00004 -- -- -- (2.78) PDZt-1 -.00003 -- -- -- (-2.55) R2 .86 .53 .57 .72 OW 1.89 1.76 2.61 2.22 F 12.48 4.22 0.79 8.17 Parentheses contain t-statistics For variable definition and derivation, see Table 9. 168 Dependent variable: National per capita endstocks in kilograms (END) Indonesia Malaysia Country (1971-82) (1971-82) Constant 22.96 4.68 (1.77) (0.16) PDt -.01 -.003 (-1.83) (-0.20) PDt-1 .001 .004 (0.48) (0.19) Rt .64 1.35 (3.14) (1.32) Rt-1 .42 -.64 (1.21) (-O.64) ENDt-1 -.83 .36 (-1.99) (0.99) PDZt -- -- Pth-I " " R2 .91 .48 OH 2.40 2.01 F 12.81 1.12 Parentheses contain t-statistics For variable definition and derivation, see Table 9. 169 APPENDIX 5 DERIVATION 0F EQUATION (38) FROM EQUATIONS (3) AND (4) APPENDIX 5 DERIVATION or EQUATION (38) FROM EQUATIONS (3) AND (4) Starting with: (3) END. - b. + PIU’Dt) + 92(Po*t+1) + ham) and (4) PD*t+1 - PD*t + 0(PDt - PD*t) arrange (3) in first-difference form: (A1) ENOt - ENDt-1 - b1(PDt - PDt_1) + b2(PD*t+1 - 90*t) + b3(Rt - Rt-l) Next, subtract PD* from both sides of equation (4) and substitute into equation (A1), giVTng (A2) ENDt - ENDt-1 - b1(PDt - PDt-1) + b2[9(PDt - 90*.)] + b3(Rt - Rt-1) Rearranging equation (3) and lagging all variables one year yields (A3) 20*, - [ENDt-1 - b0 - b1(PDt-1) - b3(Rt-1)] / b2 ’ By substituting (A3) into (A2) and solving for ENDt, one Obtains the reduced-form equation (3b) used in the study: (3b) ENDt . co + c1(PDt) + c2(PDt-1) + C3(Rt) + C4(Rt-1) + c5(ENDt_1) where the reduced-form c; coefficients are defined in terms of structural parameters on page 57. 171 APPENDIX 6 FREE TRADE-IMPLIED TRANSMISSION COEFFICIENTS OF MAJOR RICE PRODUCING COUNTRIES UNDER ALTERNATIVE ESTIMATES OF IMPORT DEMAND ELASTICITIES APPENDIX 6 FREE TRADE-IMPLIED TRANSMISSION COEFFICIENTS OF MAJOR RICE PRODUCING COUNTRIES UNDER ALTERNATIVE ESTIMATES OF IMPORT DEMAND ELASTICITIES The purpose of this appendix is to Show the method and calculations underlying the assertion made in the main body of this report that the free trade-implied transmission coefficients for .several large-Country rice traders (the United States, Thailand, and China) may be much larger than those estimated under insulation via equation (8). Calculations are performed assuming the most inelastic estimates of world rice demand elasticity found by this author in the literature (-0.08, in Siamwalla and Haykin, 1983). Furthermore, I do not consider that the transition to free trade may increase the import demand elasticity facing exporter i. AS shown in the main body, more elastic import demand figures would imply higher transmission coefficients. As indicated by equation (11), the transmission effect under free trade is (All) d(Nx()/d(PRODit) - 1 - [d(CONit)/d(PWt) + d(ENoit)/d(PWt)] * d(PWt)/d(PRODit) where (as shown in equation 13), (A13) d(PWt)/d(PROD;t) - 1 / [ d(CONit)/d(PWt) + d(ENDit)/d(PWt) + d(NMrow’t)/d(PWt)] 173 Using mean values of annual rice trade and deflated world rice price in SUS, the import demand estimate of -0.08 implies that a one dollar increase in the world price would reduce world import demand by 2,554 metric tons. Using mean exchange rate and relative deflator information, the change in world import demand resulting from a 1 Thai bhat and 1 Chinese yuan Change in the world price is 151 and 1,737 metric tons respectively. These values are used for the coefficient 61 -in equation (13). The procedure for estimating net export responsiveness to a change in local currency (coefficients a1 plus CI in equation (12)) are described in Table 17. The same range of estimates are used for sensitivity analysis (i.e. scenario (1) uses the national consumption and endstock responsiveness to domestic prices as estimated from equation (8); scenarios (2) and (3) raise these estimates by 25 and 50 percent).23 These estimates are then changed from per capita to total volume terms. For the United States, the three alternative estimates for the increase in export supply resulting from a one dollar rise in the world rice price are 1,605, 2,006, and 2,408 metric tons). Similarly, a one bhat rise in world price is estimated to increase Thai exports by 0, 123, and 246 metric tons for the three alternative scenarios. For China, these figures are 1,666, 2,105, and 2,543 metric tons. Using these figures presented above, the transmission effects can be calculated using equations (A11) and (A13). The transmission 23 Thailand’s trade elasticity as estimated in equation (8) was negative. The range of net export elasticities used here (under the assumption of free trade are zero, +.50, and +1.00. 174 coefficients are presented for each country under each alternative scenario below: Country Alternative Net Export Supply Elasticity Scenarios . (1) (2) (3) given e - .33 given e - .41 given e - .50 United States nx nx nx dNX/dPROD - .61 dNX/dPROD - .56 dNX/dPROD - .51 given enx - .00 given enx - .50 given enx - 1.0 Thailand dNX/dPROD - 1.0 dNX/dPROD - .55 dNX/dPROD - .38 given enx - .74 given enx - .95 given enx - 1.15 China dNX/dPROD - .51 dNX/dPROD - .45 dNX/dPROD . .40 As can be seen by comparing these estimates with those in Table 14, the free trade-implied transmission effects (given the range of net export and import elasticities examined) are in all cases larger than those measured under insulation, for each of the countries analyzed. 17S PART II EFFECTS OF PRICE UNCERTAINTY ON PRICE LEVELS AND MARKETING MARGINS IN INTERNATIONAL WHEAT TRADE Chapter I INTRODUCTION Farmers, consumers, and governments throughout the world are increasingly affected by uncertainty in international grain markets. Despite ongoing interest in the effects of government and private behavior on price uncertainty, little analysis has focused on the reverse question: how world price uncertainty affects the behavior of international marketing actors and resulting price levels. One would expect this effect to be important since private and state agencies involved in international grain trade typically operate on small margins, maintain large inventories, and incur a high proportion of fixed-cost investments such as transport, storage and handling facilities. If increased uncertainty iS accompanied by a risk premium (i.e. compensation required by risk averse agents for holding risky assets), then uncertainty may directly affect import and export price. levels, with broader implications for farm income, food security, and government budget outlays.l This research is motivated by the need to 1This is the international analog of national market models that analyze how Changes in marketing costs affect consumer and producer welfare. In the international case, consumers include low-income importers where national food security is progressively dependent on low-cost imported grain. Similarly, producers correspond to grain exporting countries in the international analog. Government budget outlays are also affected by changes in international marketing margins to the extent that trade and commodity programs attempt to stabilize internal prices in the face of international market fluctuations. 177 178 understand better how price uncertainty affects these actors and the resulting effect on international grain prices and marketing margins. A large body of empirical evidence demonstrates that grain producers tend to decrease output in response to greater price uncertainty. But what is the net effect of uncertainty on price levels? While the reduction in supply would seemingly raise average producer prices, this may be negated by uncertainty-induced increases in marketing margins that depress producer prices. Brorsen et al., (1987) for example, found that increased risk in the 0.5. rice industry resulted in reduced acreage, lower producer prices, and higher marketing margins. This example stresses the importance of considering the effect of uncertainty on both producers, consumers, and marketing firms. A better understanding of who bears the cost of price uncertainty in international grain markets may have important implications for the evaluation of U.S. trade policies. A plethora of alternative measures of price risk appearing in recent literature suggests that there is little consensus as to how price uncertainty should be formulated in econometric models.2 This may be of more than academic concern if empirical evidence on the magnitude and importance of price uncertainty is sensitive to the procedure used to measure it. Effective policy responses to risk , depend on an empirical knowledge base concerning its effects; such information may be obscured by inconsistent approaches to measuring it. 2 See for example, Behrman (1968), Just (1974), Lin (1977), Traill (1978), Hurt and Garcia (1982), Brennan (1982), Grant et al. (1984), Brorsen et al. (1985, 1987), Aradhyula and Holt (1988), and Anderson (1989). 179 Price uncertainty may be defined as the variance of a forecast error (Engle, 1982; Bollerslev et al., 1988).. This is analogous to E(Pt - P*t)z, where P*t is the expected price. Unfortunately, expected prices are unobservable. This highlights the importance of discerning how price expectations are formed. Numerous studies have sidestepped this problem by measuring uncertainty as some function of the variation in past prices. Yet not all price variations are unpredictable. Ideally, uncertainty should be measured in a way that sorts out anticipated from unanticipated price movements. The objective of this study is to determine the effect of price uncertainty on the level of prices and marketing margins in major international grain trading routes. Wheat trade between U.S. Pacific ports and Japanese ports is used as a case study. The objective is addressed by deveIOping a GARCH-M model (generalized autoregressive conditional heteroscedasticity in the mean; see Bollerslev, 1986; Bollerslev, et al., 1988) which measures uncertainty as the variance of price forecast errors (conditioned by available information), and is thus consistent with the definition of uncertainty above. The results of this model are then contrasted with results obtained using alternative models of uncertainty devised in previous studies. Chapter 2 develops a brief conceptual framework of the organization and behavior of major international grain marketing channels. This conceptual framework is then used to specify an empirical model of U.S.-Japanese wheat marketing margins. Chapter 3 surveys a variety of alternative methods of modelling price risk that have appeared in recent agricultural economics reports: (1) the moving 180 average of past percentage Changes (Anderson, 1989); (2) a weighted moving average of past absolute price changes (Brorsen et al., 1985); (3) a moving coefficient of variation of past price changes (Grant et al., 1984); (4) a moving variance of past percentage price changes (Dalziell, 1985); and (5) a GARCH-M process. Chapter 4 provides empirical evidence on the influence of price uncertainty on f.o.b. U.S. and c.i.f. Japan wheat prices and the marketing spread between them, as well as methodological implications concerning the sensitivity of results to the type of risk measure used. There appears to be little correlation between (a) various measures of price variability in the selling price, as consistent with previous marketing margin studies, and (b) measures of price uncertainty in both acquisition and selling price from a GARCH-M process. Because actors involved in the sale, marketing, and procurement of U.S. wheat to Japan face uncertainty in both the import and export price, the analysis of results is based on the latter type of model. Results indicate that price uncertainty is significantly associated with lower U.S. export price levels, and to a lesser extent, Japanese import prices. Price uncertainty is positively correlated with the marketing spread between the U.S. and Japanese ports, but the effect was not statistically significant at the 5 percent level. The results suggest that an increase in wheat price uncertainty from current levels to those experienced between 1973-75 would reduce U.S. Pacific port prices by 10.7 percent. The average loss to the U.S. wheat sector has been upwards of $7.6 million per year due to the 181 adverse effects of price uncertainty in the U.S. export and Japanese import prices. These results may have important implications for 0.5. trade policy because they suggest that benefits arise from stabilized international wheat prices. More generally, the results support the conclusion of Binkley (1983) that behavior within the international marketing system should not be ignored in questions dealing with price risk and instability in the world grain trade. Chapter 2 CONCEPTUAL FRAMEWORK: DETERMINANTS OF MARKETING MARGINS 1N INTERNATIONAL GRAIN TRADE ROUTES This chapter is comprised of three sections. The first section describes the objectives and behavior of major participants in the international marketing of grain, and the environment in which they operate. Particular attention is given to the U.S.-Japanese wheat trade. The second section examines the determinants of marketing spreads between major trading centers in world grain markets. On the basis of this discussion, the third section develops an empirical model for examining the effect of price uncertainty on the level of export and import prices and the margin between them. 2.1 ACTORS IN THE INTERNATIONAL GRAIN MARKETING SYSTEM The actors involved in temporal and spatial arbitrage of grain between major trading centers are part of a much broader marketing chain, from input distribution, production, and bulking in exporting countries to processing, distribution and consumption in importing countries, each stage heavily affected by government policies. Changes in costs and risks borne by actors at the various stages in the system may have countervailing effects on price levels and the marketing margins between them. Ninety percent of the United States wheat exports are handled by six multinational grain traders: Cargill, Continental, Louis Dreyfus, 182 183 Bunge, Mitsui/Cook, and Garnac. The multinational traders also handle a sizable proportion of U.S. rice exports. However, Conklin (1982) presents evidence that the U.S. grain export system is becoming less concentrated. Japanese trading houses such as Mitsui, Marubeni, Mitsubishi, and C-Ito have rapidly forged inroads into U.S. wheat procurement, and have assumed a greater role in the export of U.S. wheat to Japan and other countries. The Japanese import market is also highly concentrated. The Japanese Food Agency strictly controls national wheat and rice prices and circumscribes the activities of private traders operating within Japan. The Japanese government determines domestic wheat prices and then subsidizes or taxes consumers by the difference between this fixed price and the fluctuating c.i.f. import price. Approximately 60 percent of Japan’s wheat imports since 1970 have come from the United States (International Wheat Council, various years). Suppliers of ocean transport and other specialized handling services also play a large role in the determination of international grain marketing margins. These margins are highly correlated with freight rates that shipowners charge to transport grain between ports. The correlation coefficients between monthly wheat margins and freight rates were .60, .59, and .68 for the U.S. Gulf-Rotterdam, U.S. Gulf- Tilbury, U.K., and U.S. Pacific ports-Japan routes. The dry bulk shipping industry is highly competitive (International Wheat Council, 1986; Binkley, 1983). Entry is unrestricted, rates are set by competitive bargaining between shippers and shipowners, and virtually no regulations are imposed by 184 governments. Adjustments in the supply of shipping occur with a significant lag because it takes at least two years to construct a ship. In the short run, shipowners can respond to increased demand by operating at greater speeds. However, as speed increases, fuel consumption increases more than proportionally, causing freight rates to rise (Binkley, 1983). When demand is slack, shipowners may reduce speeds to save fuel (a short-run measure) or remove vessels from service (a long-run measure). Some governments also provide Shipping services. The Soviet Union has its own fleet for grain transport, enlarged considerably since the 1980 grain embargo (International Wheat Council, 1986). Multinational grain traders also have sizeable fleets to transport grain. However, they are not vertically integrated in the traditional sense of simply providing shipping services for internal coordination within a firm’s various stages. For example, a multinational trader’s shipping fleet may find it more profitable at a particular point in time to haul iron ore or some other commodity than transport its own wheat to fulfill a contract (Peterson, 1988). This is quite different from most cooperatives or state traders, which tend to provide services mainly for other stages of their firms’ Operations. Wm It may be hypothesized that the objectives of the multinational grain traders (MTs) include increasing market volume, market share, and profits. The market environment that facilitates these objectives is more likely one of price instability rather than stability. Profits of 185 the large private grain traders seem to move over time not with total volume traded, but rather with the degree of disruption and instability in the year’s trading patterns (Caves, 1979; Gilmore, 1982). Such an environment enables multinationals to put their superior global market intelligence to work and capture the rents resulting from market disturbances. However, large volumes are not unimportant; they promote size economies in two ways. First, the HTS make only a few cents on each ton of grain handled. Therefore, profits in the industry require large volumes. Second, the MTs’ potential to maximize their objectives depends on having large, diverse positions to draw from quickly to exploit temporary arbitrage opportunities (Peterson, 1988). For these reasons, the grain traders lobby against programs that reduce traded volume, such as U.S. acreage reduction programs. Some studies (e.g. Brorsen et al., 1985) contend that marketing firms are risk averse and that they would benefit from price stabilization policies. This is certainly contrary to the pronouncements of mOSt major private grain traders (Harrison, 1988; Peterson, 1988). Whether price risk is good or bad depends on the size and resources of the firm in question. As mentioned above, price instability affords greater profit opportunities from arbitrage for those with superior market information networks. The Characteristics of information as an input create scale economies in coordination and risk-bearing. Information procurement has a fixed cost that can be spread over numerous transactions, and information about trading locations is subject to increasing returns in the trading possibilities that it reveals. There are also economies of scale in pooling the 186 risks associated with price and exchange rate variability, uncertain government policies, and legal idiosyncracies of importing countries. Large, diversified investment portfolios allow the larger firms to appreciably reduce the variance of returns (Caves, 1979). Furthermore, the large multinationals appear to prefer an environment of incomplete contingency and forward markets, which poses risk-related barriers to entry. Prior to the introduction of the now- defunct rice futures market on the New Orleans Commodity Exchange in 1981, smaller rice milling and exporting firms had no way to hedge except by maintaining large inventories, which increased operating costs and other risks. Investment by small or new firms had been dampened because banks were understandably less willing to lend when Operations are unhedged and subject to instability-related losses (Business Week, 1978). The larger firms also have deeper pockets and are thus better positioned to survive speculative losses than smaller competitors. An unstable market environment may thus be conducive to increased market concentration. It is not surprising that the New Orleans rice futures market was welcomed by many smaller rice milling and exporting firms while disparaged by larger rice traders (Business Week, 1978). This discussion suggests that attitudes toward risk depend not only on a firm’s aversion to short-run profit variability, but also depend on longer-run dynamic effects on market structure and market share. A firm could be DARA (decreasing absolute risk averse, see Robison and Barry, 1986) and still prefer an environment of high market uncertainty. 187 Multinationals’ operating procedures (and their effect on market structure) appear to influence the degree of price risk they incur. For example, the practice of multiple sourcing enables international traders to fulfill contracts with importers without specifying the origin of the grain to be delivered. The ability to shop around for the best deals in producing countries therefore reduces the risk associated with unforeseen price fluctuations in a particular export market. Importers usually buy grain on either flat or basis pricing terms, or they may vertically integrate forward to procure grain in surplus countries, where this is permitted. Flat price contracts, which characterize the buying terms of many state importing parastatals, fix the price at which future delivery will take place. This prevents them from altering the pricing terms should more favorable market conditions occur prior to delivery. Yet more sophisticated importers, such as Japan’s Food Agency, frequently use basis-priced contracts, which specifies some agreed-upon amount over a particular futures price, and allows the importer to choose, sometime prior to the delivery date, the date for the calculation of the relevant futures price. This allows the importer to hedge against adverse price fluctuations. However, many state importers, especially those in low-income countries, do not participate in futures trading. The incidence of a rise in risk-related costs is traditionally thought to depend on export supply and import demand elasticities (Tomek and Robinson, 1972; Fisher, 1981). The more inelastic the demand (supply), the greater the share of the marketing margin borne by 188 the buyer (seller). This framework would suggest that importers with very low trade elasticities would bear the major brunt of increased international marketing costs.3 Using findings from Part I of this study, one would expect uncertainty-induced risk costs borne by international marketing firms to be built into c.i.f. Japan prices, without much reduction in volume traded. However, this logic ignores the possibility that risk also affects ' internal suppliers in importing countries and marketing firms in exporting countries--i.e. the stages on either side of the export port- to-import port marketing function. If these firms are risk averse, then they too may require compensation for engaging in risky activities. For example, internal suppliers in the importing country may be concerned primarily with c.iIf. price risk. If risk averse, they may respond to increased uncertainty by raising their margin, either by raising consumer prices or reducing the c.i.f. purchase price. Analogously, grain export firms in the United States are most concerned with output price risk. The international grain traders may face uncertainty in both the export and import price. Therefore, they are concerned with uncertainty in the marketing margin. Therefore, under the assumptions that (a) a variety of actors in the international grain marketing system face price uncertainty, and (b) risk presents costs that firms attempt to recover by increasing their margins, we are left with the conclusion that price uncertainty 3 Part I of this study indicates that Japan’s wheat import demand elasticity and the U.S. export supply elasticity have been :pprgximately zero and +.60 at mean price levels over the past several eca es. 189 may thus exert countervailing and theoretically ambiguous influences on c.i.f. and f.o.b prices. 2.2 FACTORS AFFECTING PRICES AND MARKETING MARGINS IN INTERNATIONAL GRAIN TRADE Marketing margins in this study are defined as the difference between the c.i.f. and f.o.b. price of a particular type and grade of grain. These prices represent supply and demand conditions, market structure, and the effects of government policies at particular sub- markets within the larger global market. Movements in the margins reflect changes in the cost of marketing services (which are represented by the cost functions of marketing and shipping firms) and the integration of the world market system. If the world market is poorly integrated, changes in f.o.b. port prices may not necessarily be fully transmitted to c.i.f port prices and vice versa. This would ' cause instability in the margins. Therefore, international marketing margins are not just a function Of the technology of the particular firms involved in transporting grain internationally, but also the institutional organization of the world grain marketing system. Prices in major trading ports (and the margins between them) are simultaneously determined by each other’s movements, as well as their past values, with the strength of the relationships depending on the extent of spatial and temporal integration between the respective submarkets. The differential between these sub-market prices are thus affected by the availability and cost of market information, market power, transactions costs, and the technical elements of marketing firms’ cost functions. 190 T hni et rminants Ocean freight rates since 1970 have accounted for 80, 63, and 86 percent of the monthly wheat price differentials between the U.S. Pacific ports-Japan, U.S. Gulf ports-Tilbury, and U.S. Gulf ports- Rotterdam (Table II-l).4 This indicates that variable costs incurred by suppliers of ocean transport may affect port prices and thus marketing margins. As mentioned above, petroleum is a key input in the supply of ocean transport. Interest rates also affect the variable costs of large-scale investments such as vessels and port facilities, as well as grain storage costs borne by marketing firms. The demand for shipping services is partially a derived demand for internationally-traded grain. However, the correlation between monthly seaborne grain volume and freight rates is not strong. This is because grain is one of many commodities affecting the demand for shipping services. Coal, iron-ore, fertilizers, and even petroleum compete with grain for dry-bulk shipping services.5 Data in Binkley (1983) indicate that world seaborne grain trade was less than 10 percent of world dry bulk trade during the mid-19705. Unless this proportion has changed 4 Freight rates as a proportion of monthly marketing spreads are lower in the U.S.--Tilbury route because for part of the period under study, England lacked adequate port facilities to handle larger vessels that became common in international grain trade in the past several decades. Consequently, grain delivered to England was frequently sent first to Rotterdam, unloaded, and then reloaded on smaller vessels bound for Tilbury. The greater handling costs reduced the proportion of the margin accounted for by ocean freight rates. 5 Since the early 19605, innovations in cleaning techniques have facilitated the ability of tankers to switch from oil- to ore- to grain-carrying. They can therefore be used in the most lucrative market at a given time (International Wheat Council, 1977). 191 TABLE II-l Ocean Freight Rates and Marketing Spreads in Three International Wheat Trading Routes: 1970-86 Averages MARKETING (1) AS A TRADE ROUTE FREIGHT RATES SPREAD* ‘% OF (2) U.S. Pacific 18.3 22.8 Ports--Japan U.S. Gulf-- 9.2 10.6 Rotterdam U.S. Gulf-- 11.6 18.4 Tilbury ,5 area defined as c.i.f - f.o.b. price Note: The type of wheat used in this table are No. 2 Western White (U.S. Pacific ports--Japan) and NO. 2 Dark Northern Spring, 14% (U.S. Gulf ports--Rotterdam, and U.S. Gulf ports--Tilbury). Source: International Wheat Council, various years. greatly in the past decade, variations in international grain shipments would have only a marginal effect on freight rates. The relationship is further decoupled by time chartering (shipping services chartered for specific periods of time at fixed rates). MW Rice marketing margins have consistently exceeded wheat margins for five major trade routes examined (Figure 11-1). In spite of identical shipping distance, rice margins were six times larger than 192 SUB/NI I.“ ‘\ ’5‘ .0 r" -- I ‘5 ‘1 I \ (I) ‘4! ‘4.“ {I {I 1‘ 16‘ a" '75 ' \ g I I '7 ‘3’. '7‘. g I I I ‘5; \‘—-\ J I (I . \ so (4) "' ‘- 1978 1979 1988 1981 1982 1983 1984 1985 1986 Average Margin Level 1978-86 (SUS/Inetric ton) 1. Houston-Rotterdam rice margin, 71 .6 No.2 milled, 4% bagged 2. U.S. Gulf Ports-Rotterdam wheat margin, 11.8 No.2 Dark Northern Spring, 14% 3. U.S. Pacific Ports-Japan wheat margin, 32.3 No.2 Western White 4. U.S. Gulf Ports-Tilbury, U.K. wheat margin, 22.6 No.2 Dark Northern Spring, 14% 5. Bangkok-Rotterdam rice margin, Thai SWR 58.8 100%, milled (not shown) Figure II-1. Marketing Margins in Selected International Grain Trade Routes: 1978-86 ' 193 wheat margins between the U.S. gulf and Rotterdam since 1970. This may be due to (1) the less transparent nature of the world rice market; (2) higher search costs and brokerage fees involved in the rice trade; and (3) the marked volatility in world rice prices as compared with wheat (both in short-run and long-run terms), which may increase risks and marketing costs. Also, fragmented spot and futures markets with inadequate standardization across grades and varieties of rice impede the dissemination of price information and contribute to poor market integration (see Part I of this study). The vast difference between U.S.-Rotterdam wheat and rice margins clearly indicates the importance of organizational and institutional factors in the determination of international marketing costs. These same factors may influence the degree of instability in international grain margins. Table II-2 compares correlation coefficients between first differences in selected monthly wheat and rice prices.6 Low price correlation between two trading ports means that the marketing margin between them is subject to high instability. Actors in the rice trade may therefore need to take higher margins to protect themselves from adverse price fluctuations. 5 First differences in prices are analyzed rather than the price levels themselves to remove certain trend components from the data. This allows a more accurate assessment of the degree to which random shocks in one market are reflected in the prices of another market. Although correlation coefficients have traditionally been used to measure the degree of integration between markets, the potential for spurious correlation due to omitted third variables is possible. Therefore, correlation coefficients are only a rough indicator of market integration. 194 TABLE II-Z Correlation Coefficients of First Differences in Monthly Wheat and Rice Prices in Various International Markets, 1972-1986 WHEAT U.S. Tilbury, U.S. MARKET Gulf Rotterdam U.K. Pacific Japan F U.S. Gulf 1.00 Rotterdam .85 1.00 Tilbury, U.K. .75 .78 1.00 U.S. Pacific .72 .72 .72 1.00 Japan .54 .54 .57 .68 1.00 RICE Rotterdam Rotterdam MARKET Houston (U.S. No.2) Bangkok (Thai, Grade A) Houston 1.00 Rotterdam .68 1.00 (U.S. No.2) Bangkok .28 .29 1.00 Rotterdam .21 .16 .78 1.00 (Thai, Grade A) Note: The type of wheat used in this table and throughout the report are No. 2 Western White (U.S. Pacific ports--Japan) and No. 2 Dark Northern Spring, 14% (0.5. Gulf ports--Rotterdam, and U.S. Gulf ports-- Tilbury). For rice: No. 2 milled, 4% bagged (Houston--Rotterdam) and Thai SWR 100% (Bangkok--Rotterdam). 195 Analogously, temporal changes in uncertainty within a given trade route and commodity may affect price levels and marketing margins through time. The major international grain traders seldom operate from totally hedged positions, but rather on net positions, reflecting their expectations of future market conditions. This is especially true for rice, which lacks a viable futures market. Therefore, an implicit risk premium may be incorporated into marketing margins to guard against adverse price fluctuations. The ability to take risks, and cover them by raising the margin, would appear to be more feasible and profitable in a highly concentrated market (i.e. several large traders exerting market power). However, if these firms sell to (buy from) equally concentrated importers (exporters), the outcome may be less clear. For example, Japanese wheat imports and prices are strictly controlled by the Japanese Food Agency. As mentioned previously, Japanese trading companies have their own access to grain supplies in the United States and other grain exporting countries, which may mitigate the ability of other international traders to pass on costs to the Japanese government. Facing highly concentrated importers, risk and other costs incurred by the HTS may be more easily passed upstream to farmers and first handlers, where the market may be less concentrated. These issues are explored further in Chapter 4. 2.3 MODEL SPECIFICATION There are two general approaches to econometric model specification. The first is to posit structural relationships derived from some specific theory regarding the behavior of each set of actors 196 in the system under analysis. From these, reduced-form equations can be specified containing structural restrictions based on the particular theory used. The second approach is to construct a model using minimal structural restrictions in order to summarize the historical patterns in the data, the results of which are not conditional upon any particular behavioral theory. This approach is common in time series models; specification may be based on variables common to alternative theories without imposing restrictions particular to any one of them (see for example Sims, 1986). This study is based on the second approach. This approach has both advantages and disadvantages. The major drawback is that a reduced-form statistical model (which does not arise out of any particular economic theory) and results cannot be used to derive structural information about the behavior of particular economic agents. Rather, the results may be viewed as measuring the net effect of all actors’ behavior in the system on the endogenous variable. Also, such models have been validly criticized as having no explicit theoretical foundation. However, in many cases, this drawback may also be an advantage. Unless there is definitive evidence and agreement on the "correct” theory, a structural model imposes restrictions on the data such that the results become conditional on that theory being the “correct” one. For example, mean-variance, rational expectations, and adaptive expectations frameworks each impose certain restrictions regarding functional form and the relationship between variables. All results and implications are thus predicated on the restrictions of each 197 particular model being the correct ones. In the absence of such consensus, it may be advantageous, especially with complex lag effects and Simultaneity, to construct a model containing basic common elements of alternative theories in conjunction with statistical criteria. This requires a brief survey of previous marketing margins studies. Brorsen et al., (1985), used a mean-variance framework to estimate the effect of output price risk on U.S. wheat millers and retailers. Their model specified annual margins as a linear function of input costs incurred by marketing firms, quantity of wheat consumed domestically, and output price risk. Grant et al. (1984) estimated an identical linear model for the farm-to-mill rice marketing margins in the U.S. A subsequent analysis of risk in the U.S. rice industry (Brorsen et al., 1987) specifies annual farm-to-mill and mill-to-retail rice margins as a linear function of simply milling and marketing costs, and output price risk. Even though several studies questioned the competitiveness in these markets, a competitive market framework was used based on the assumption of market contestability. Moving from annual to monthly data, prices may be subject to stronger lagged effects and adjustment costs. It may be appropriate, therefore, to include lagged prices in own and other submarkets as predictors of current port prices. The strength of the lagged relationships would depend on the degree of spatial and temporal integration between prices at the various markets. Therefore, one-month ahead forecasts of grain port prices and the margin between them may be influenced by: 198 (a) the expected values of international traders’ and transporters’ variable marketing costs in the following month. Prominent among these are petroleum price (Poil) and interest rates (rr); (b) recent port prices at own and other submarkets (with the strength of these relationships depending on spatial and temporal correlation between prices at the various markets); (c) potential seasonality in port prices; and, if marketing actors are risk averse, (d) some measure of the uncertainty of prices, which may cause unforeseen fluctuations in exporters, traders, and importers’ profit margins. Addressing only issues (a) and (b) for the moment, a set of equations predicting international port prices from one month to the next can be proposed of the form: n n * (l) Pfobt - 2 A11(Pfobt-i) + X 811(Pcift-i) + C11(rr t) + 1-1 1-1 C12(Poil*t) + Flt n n * (Z) Pcift - X A2;(Pfobt-i) + X Bz;(Pcift-i) + C21(rr t) + 1-1 l-I C22(Poll*t) + ‘2t where * denotes expected values. Subtracting (1) from (2) generates the marketing margin equation n n * (3) MMt -izl A3;(Pfobt-1) + ‘21 831(Pclft-1) + C31(rr t) + C32(POil*t) + €3t 199 "here A31 ' Azi - A11; 331 = 321 - 311; 531 = 021 - CIT; afld €3t = 62t ‘ 51t- Equations (1)-(3) may be regarded as reduced-form equations which specify the historical relationships between current port prices and past prices and input costs. No attempt is made to derive structural interpretations concerning the behavior of any particular actors in the system; rather, the model represents the net effect of all actors influencing the system. The expected monthly values for rrt and Poilt are approximated as last month’s value. The criterion for Tag length of port prices was the minimum number required to purge the error terms of autocorrelation. Using January, 1972 to February 1986 data, OLS estimation of (1) and (2) indicated that two lags were required using a Q-test to accept the null hypothesis of no autocorrelation.7 Based on the foregoing, (1) and (2) may be reformulated as (4) Pfobt - 310 + 811(Pfob_1) + 312(Pcif-1) + 513(Pf0b-z) + 614(Pcif-2) + 815(rr-1) + 315(P011-1) + ‘It and (5) Pcift - £20 + 821(Pfob-1) + 622(Pcif-1) + 823(Pfob-2) + fiz4(Ple-2) + 525(rr-1) + 825(P011-1) + ‘2t Seasonality was examined for five prominent grain trade routes by comparing the explanatory power of equations (4) and (5) with an unrestricted model containing monthly dummy terms: 7 The O-statistics for equations (1) and (2) with 12 monthly lags were 13.68 and 8.64 against a Chi-square critical value of 21.03 at the .05 significance level. 200 (6) Pit = 510 + Bil(PfOb-1) + 512(Pcif_l) + fii3(PfOb-2) + fii4(Pcif-2) + fli5(rr-1) + 515(Poil-l) + ci1(JAN) + ciz(FEB) + ... + C111(NOV) + Fit where i - 1 for the export price and 2 for the import price. The proposition that cl-c2-...-c11-O was examined using an F-test. Results indicate that seasonality in grain prices was statistically insignificant at the 5 percent level in all trading ports examined except the U.S. Gulf wheat price (Table II-3). It is possible that seasonality may be present in the conditional variance, but this was not analyzed further. Mpdelling of Price Dneertainty Economic theory frequently suggests that market agents respond to the unpredictability as well as the level of prices. If increased uncertainty is accompanied by a risk premium that becomes built into price levels, then a more plausible formulation of equations (4) and (5) would be: (7) Pfobt . £10 + 611(Pf0b-1) + 512(Pclf-1) + 813(Pfob-2) + 614(Pcif- 2) + 315(rr-1) + 615(Poil-1) + 617(RISKt) + Flt and (8) Pcift - £20 + 621(Pfob-l) + fizz(Pcif-1) + fiz3(PfOb-2) + 624(Pc1f- 2) + 825(rr-1) + fiz5(POll-1) + 327(R15Kt) + (Zt Subtracting equation (7) from equation (8) generates the marketing margin equation 201 TABLE II-3 F-Test Results for Presence of Seasonality in Major World Grain Markets 1 Market Comodity Rho-value" 11, 173 U.S. Gulf Ports (1972-86) wheat 0.01 U.S. Gulf Ports (1979-86) rice 0.52 Rotterdam (1972-86) wheat 0.11 Rotterdam (1979-86) rice 0.71 U.S. Pacific Ports (1972-86) wheat 0.63 Japan Ports (1972-86) wheat 0.08 Tilbury, U.K. (1972-86) rice 0.19 Bangkok (1979-86) rice 0.45 * The rho value is the level of significance of the included monthly dummy variables. Note: monthly rice data prior to 1979 not available in data source .used (USDA, various issues). (9) MMt . Pcift - Pfobt - £30 + 631*(Pf0b-1) + fi32*(PCif-1) + fi33*(PfOb-2) + 834*(Pcif-z) + 335(rr-1) + 535*(Poil-1) + 837(RISKt) + €3t where fi3j - £23 - filj, and 63t - EZt - Elt. This specification permits one to examine the effects of price uncertainty on the levels of the f.o.b. and c.i.f prices as well as the margin. This is important for measuring the incidence of price risk, which may be affected by structural and organizational aspects of the market. Alternatively, price uncertainty may affect both port prices 202 to a similar degree. If such were the case, estimating a margin equation alone would indicate that price uncertainty has no detected effect (i.e. b2] - b17 - b3; - 0), even though it could markedly affect both c.i.f. and f.o.b. prices in a parallel way. Estimation of equations (7-8) as a system provide more information than models based on equation (9) alone concerning how price uncertainty affects international wheat marketing margins, i.e., through the import port price, the export port price, or both. Furthermore, if price uncertainty arises from more than one source, it would be appropriate to distinguish between these sources. This is important for discerning how sensitive the margin and port prices are to particular sources of price uncertainty. For example, a particular margin may be more sensitive to changes in c.i.f. price uncertainty than f.o.b. uncertainty if firms’ organization or practices are designed to cope with one source of uncertainty better than another. Sensitivity to various sources of price risk may be affected by factors such as multiple sourcing, vertical integration, and market concentration at particular vertical stages in the channel. Chapter 3 ALTERNATIVE MODELS OF PRICE UNCERTAINTY Price uncertainty may be defined as the variance of a forecast error, conditional on available information, or (10) ht - E(EZt/available information) where ft is mean-zero and modeled as (11) ct - EIPt - P*t) .... N ( 0. ht) However, P*t and thus 6t are unobservable. Most studies have sidestepped this problem by approximating price uncertainty as some function Of the variation in past prices (e.g. Anderson, 1989; Brorsen et al., 1987, 1985; Grant et al., 1984; Brennan, 1982). These types of uncertainty indicators are briefly reviewed. 3.1 UNCERTAINTY MEASURES DERIVED FROM VARIATIONS IN PAST PRICES Table II-4 presents the formulas for four alternative measures Of price uncertainty appearing in recent agricultural economics literature: (a) moving average of percentage changes in price (Anderson, 1989); (b) weighted moving average of absolute price changes (Brorsen et al., 1985); (c) coefficient of variation of past monthly prices (Grant et al., 1984); (d) variance of percentage changes (Dalziell, 1985). 203 204 TABLE 11-4 Alternative Indicators of Price Uncertainty INDICATOR FORMULA a. moving average of past - (P-1-P-2)/P-2 + (P- -P-3)/P-3 twelve month’s percentage + ..... + (P-12-P-13I P-13 changes b. weighted moving average - [ 1.2 * P-1-P-2 of past absolute price +1.1 * P-1-P 2 deviations, divided by + ... + 0.1 * l P-12-P-13 I mean price : Pmean where Pmean - average Of past twelve months’ prices 12 c. moving coefficient of variation . [ 2 (Pt-i-Pmean)2/12 ]°-5 of price Changes over past i-1 twelve months Pmean where Pmean - average of past twelve months’ prices O. 12 moving variance of monthly . [ 2 ( dPt-;-deean)2/12 ] percentage Changes in price i-l over past twelve months where dP - monthly percentage Change in price deean - average of past twelve months’ percentage changes in price Authors used varying lengths of past prices to calculate respective measures of risk, based on data sampling frequency and the nature of the commodity under study. Twelve months are used here to standardize the empirical ‘analysis. 205 These indicators have a number of similar properties: Each of them is based on some function Of past price movements rather than on the notion of a forecast error. Such a specification misses the fact that part of the price change may be anticipated, and thus would not count as uncertainty. Clearly, not all fluctuations create uncertainty.8 Second, each indicator presumes that there is some time frame within which price Observations matter, and that Observations beyond this chosen lag length are irrelevant for the measurement of risk or uncertainty. Third, indicators (a), (b) and (c) count all observations within the chosen time frame equally; indicator (b) assumes that the most recent Observations count more in the measurement of uncertainty, and that economic agents have a declining memory of past observations until the chosen lag length, at which time observations are considered irrelevant. Fourth, indicators (c) and (d) measure squared deviations from the mean price over some period as an appropriate gauge of uncertainty, as compared with indicators (a) and (b) which look at absolute deviations. The former indicators give greater weight to outliers, and thus assume that agents consider extreme fluctuations to be proportionally more important than smaller fluctuations. Fifth, indicator (d) removes the trend component from the measurement of uncertainty, while (a), (b), and (c) do not. It may be reasonable in some cases to treat a trend as not contributing meaningfully to instability because the trend can be readily adapted to. But this would not apply in other cases. Where structural 8 Examples may include situations of steady inflation or where other data is available to help agents anticipate future price movements. 206 conditions change frequently, the assumption that market actors can and do anticipate price trends--even without knowledge of the future observations needed to establish the trend--is somewhat tenuous. From an econometric standpoint, finally, each of these measures are internally inconsistent. Either the errors in equation (11) are heteroscedastic or they are homoscedastic, but they cannot be both. If homoscedastic, this implies that the forecast errors are invariant over time, and thus the effect of uncertainty would be subsumed in the constant term. If the errors are heteroscedastic, then their past values provide relevant information with which to measure future price uncertainty, since they represent the discrepancy between actual and predicted prices. If agents use information about past errors to help predict future price, then it is inconsistent that these errors are drawn from forecast models that treat them as constants (i.e., one cannot simply derive forecast errors from a model without explicitly accounting for the effects of uncertainty on the forecasted price). Rather, uncertainty and its potential effects on price level must be accounted for in the process Of generating the forecast errors. This implies that the level and variance of prices are simultaneously determined. 3.2 THE GARCH-M MODEL In contrast to these measures, another class of models attempts to relate uncertainty to the forecast variances Observed in recent periods, in a manner consistent with equations (10) and (11). The relationship between current unpredictability and past unpredictability 207 has been observed by econometric forecasters for some time. Mandlebroit (1963) documented that "large changes [in forecast errors] tend to be followed by large changeS--of either sign--and small changes tend to be followed by small changes...” McNees (1979) observed that 'the inherent uncertainty or randomness associated with different forecast periods seems to vary widely over time,” and that "large and small errors tend to Cluster together (in contiguous time periods)." This suggests that the conditional variance of forecast errors (i.e. the variance of ft in (10) conditioned on past information) may be heteroscedastic through time, as the market environment changes. The GARCH-M model expresses the conditional forecast variance as a linear function of past squared prediction errors. The model used in this study takes the form n (12a) conditional mean: Pt - ¢ + 2 11(Pt-1) + ¢1(rrt_1) + i-l ¢2(Poilt-1) + 5(Ht) + 6t (12b) conditional variance: Ht - a + b(6t-1*6't_1) + c(Ht_1) et/past information ~ N (O,Ht) where Pt is a (2x1) vector of endogenous port prices, a is a (2x1) vector of constants, 1; and 8 are (2x2) coefficient matrices, pl, ¢2 and 6 are (2x1) coefficient vectors, and a, b and c are symmetric coefficient matrices. This Ht matrix is derived from the lagged (2x1) vector of forecast errors from the conditional mean equation (6t-1), and lagged H values. Therefore, the degree of price uncertainty experienced in month t (Ht) 208 is related to the forecast errors in previous periods. With only one lag on the squared error matrix and one lag on the H matrix, this specification is a particular form of a more general GARCH-M model with p and q lags respectively. However, letting p and q equal 1 is a reasonable simplification since all prior error variances are contained in Ht-1 in a Nerlovian-type process with geometrically declining weights.9 Thus, uncertainty is measured with an infinite but declining memory of past forecast errors. Estimation of this model is occasionally problematic because the estimation process may generate an Ht matrix that is not positive definite, meaning that the likelihood function sometimes cannot be evaluated. In light of this difficulty, Kroner (1988) proposed another parameterization of the GARCH equation that is guaranteed to be positive definite: (12c) Ht - c'c + A’(Et_16’t-1)A + B’(Ht-1)8 9 To see this, equation (12b) can be lagged several periods to obtain: (12b’) Ht-1 - c + a (12b”) Ht-2 - c + a By substituting (12b”) into (12b’) and then into (12b), it is clear that error variances lagged two and three periods influence Ht. This recursive substitution can be carried out indefinitely to show that all prior forecast variances are allowed to affect the current level of uncertainty, with weights determined from the estimation process rather than being arbitrarily chosen by the analyst. 209 where C is a (2x2) symmetric matrix of constants, and A and B are (2x2) coefficient matrices. This parameterization has the drawback of imposing certain restrictions on the model in order to achieve positive definite covariance matrices. However, such restrictions appear reasonable since it is clear that all such matrices must be positive definite. Estimation of (12c) also reduces the number of parameters to be estimated, which may spare considerable computational cost in maximum likelihood estimation. For these reasons, the positive definite parameterization is used in this study. The major advantage of the GARCH-M model is that it presents an internally consistent framework for examining changing expectations and uncertainty over time. By determining the conditional means and variances simultaneously, it allows study of their effects under a unified framework. In addition, the GARCH model expresses uncertainty as a function of past forecast errors and is thus consistent with the definition of uncertainty in equations (10) and (11). The method attempts to disaggregate that portion of price movements which is unpredictable from that which may be anticipated from knowledge of available information. While the GARCH-M model has clear theoretical advantages over the previously mentioned measures of uncertainty, it also is not without limitations. Similar to other models of ex-ante forecasting, parameter estimates of the GARCH model are derived using data from the entire sample period. Therefore, it is assumed that market actors know the estimated parameters Of the model even without knowledge of future observations necessary to establish these coefficients. Such 210 information could surely bias the estimated parameters at time t and hence the forecasted values and error variances. One might consider a more realistic alternative as using Observations available only at time t to generate price forecasts, and using an Updating process to generate new parameter estimates and forecasts each period as new information from the following period is made available. However, this could entail significant computational cost. Secondly, the measurement of uncertainty via the GARCH-M model is only as good as the prediction model. If the modeler is using a different forecast model from those of economic agents, then the resulting measurements of uncertainty are clearly biased. However, this is true of any forecast model. CHAPTER 4 ESTIMATION AND RESULTS: THE CASE OF U.S.-JAPAN WHEAT TRADE 4.1 GARCH-M ESTIMATION RESULTS The GARCH-M model is premised on the notion that present uncertainty is related to the variances of past forecast errors. To examine this hypothesis, the GARCH model is first estimated without the conditional variance-covariance terms in the conditional mean equations: (13) Pfobt I 110 + 111(Pf0b-1) + X12(Pcif-1) + 113(Pf0b-2) + N14(Pcif-2) + 115(rr_1) + 115(P011-1) + ‘It Pclft . £20 + 121(Pf0b-1) + 122(Pcif-1) + H23(Pf0b-1) + 124(Pcif_2) + 125(rr-1) + 125(P011-1) + FZC (It h h .. N 0’ 1t 12t 62t h12t h2t h1t h12t c1 c12 c1 c12 + A11 A12 621t-1 €1t-1*€2t-1 h12t hzt c12 c2 c12 C2 412’422 61t-1*62t-1 €22t-1 A11 A21 + B11 B12 h1t-1 h12t-1 B11 B21 A12 A22 B21 B22 h12t-1 h2t-1 B12 B22 211 212 This is a bivariate model in which the conditional means and variances of Pfobt and Pcift are determined simultaneously. Expected prices are simply one step-ahead forecasts generated from these equations. Notice that ‘it are the current forecast errors used to influence the extent of uncertainty in the next period, via the conditional variance equations. The maximum likelihood estimates and their t statistics are presented in Table 11-5. The conditional variances Of U.S. Pacific and Japanese port prices (hl and hz) are strongly related to their past values; t-statistics are above 4.0 in several cases.10 These results are consistent with Mandelbroit’s and McNees’ earlier observations of episodic fluctuations in forecast accuracy. This can be seen graphically in Figures 2a-c, in which “It: h2t and h12t fluctuate widely through time, with high and low values tending to cluster together in contiguous time periods. However, model (13) does not allow for the conditional variances to affect the mean prices. If uncertainty presents greater transaction costs and risk-related costs associated with grain trade, then uncertainty may directly affect import and export price levels. Under this premise, the time varying risk premium has been swept into the disturbance term in (13) and represents misspecification. This 10 Simple OLS regression of the squared residuals on a constant and lagged values also provides strong indication Of ARCH effects (autoregressivs conditional heterscedasticity). The null hypothesis that E 1 and 6 were unrelated to their past values was rejected at the .05 level, For 4, 6, and 12 lag lengths. The test procedure is to regress the squared residuals on a constant and p lags. The R of the regression times the number of Observations will be asymptotically distributed as chi square with p degrees of freedom when the null hypothesis is true (Engle, 1982). 213 TABLE II-5 Maximum Likelihood Estimations Results for Equations (13)* Dependant Variable Pfob Pcif MM Explanatory Variables Intercept 6.71 5.35 -1.36 (4.28) (1.78) Pfobll - 0.71 0.29 -0.42 (7.44) (3.16) Pcif_1 0.30 0.89 0.59 (3.62) (8.52) Pfob-2 0.13 -0.14 -0.27 (1.26) (-1.05) Pcif-2 -0.21 -0.08 0.13 (-2.47) (-0.61) rr-1 -0.12 0.24 0.36 (-0.56) (0.65) Poil_1 0.004 0.007 0.003 (1.25) (1.47) *: t-statistics are in parentheses h n- 0.19 -1.78_ 0.19 -i.78- 1.06 -0.05 £2 , e ,-E _ 1 12 (0.07) (1.74) 40.07) (1.74) (5.66) <-0.23) 1‘ 1 1t ‘ 2‘ 1 I 4 4.78 -2.01 4.76 -2.01 0.19 -0.30 6 .146 , 62 , ”21 “2 {-1.741 {-0.62) 4-1.74) (-0.622J (1.36) (-l.82) 1‘ ' 2‘ ' 2‘ 1 1.06 0.19 1_0.99 0.11 h1t,‘ 6,2,,1 0.59 0.11 (5.66) (1.36) (4.06) (0.73) (4.06) (0.78) 0' -0.05 -0.30 0.16 0.80 h . _ 0.16 0.60 {-0.281 (-1.82) (1.00) (5.96) ‘2‘ ' "2‘ ' 41.00) (5.98) 214 (a) SDS/NI 3500 (b). m, 588- 2584 1972 m m m 900 902 904 904 102 (‘2) sum: m an. m. 3m. 200. 100- 0 i ..-n. 19? 97 97- 9?? 981 9B 984 98. Flgure II-2. Time Varying Conditional Variances and Covariances of U.S. Export and Japanese Import Price Forecast Errors from GARCH-M Model (13) 215 raises the question of what source of price uncertainty matters--that from the f.o.b. price, the c.i.f price, or the margin itself. Most Of the earlier studies cited (Grant et al., 1984; Brorsen et al., 1985; Brorsen et al., 1987) consider only the effect of sales pricerisk on the margin, and assume that uncertainty in the acquisition price or in the margin have no effect on the margin level. Under this assumption, the equivalent GARCH-M model is: (14) Pfobt '110 + 111(Pfob-1) + 112(Pclf-1) + N13(Pfob-2) + 114(Pcif-2) + x15(rr-1) + I15(Poil-1) + 61(h2t) + (It Pcift I120 + N21(Pfob-1) + 122(Pcif-1) + l23(PfOb-1) + l24(Pclf-2) + m25(rr-1) + 125(Poil-1) + 62(h2t) + €2t E h h It “N 0’ 1t 12t ‘2t h12t h2t h1t h12t c1 c12 C1 c12 + A11 A12 fZIt-l €1t-1*62t-1 h12t h2t c12 c2 C12 c2 A21 A22 €1t-1*62t-1 £22t-1 A11 A21 + 811 B12 h1t-1 h12t-1 B11 B21 A12 A22 B21 B22 h12t-1 h2t-1 B12 B22 216 Under this specification, the effect of c.i.f. price uncertainty on the margin is measured by 62-81.11 Regression results are presented in Table II-6. Consistent with the earlier studies, price uncertainty in the c.i.f (selling) price appears to raise the margin between the U.S. Pacific and Japanese port prices. However, a likelihood ratio test between models (13) and (14) shows that the inclusion of the conditional variance of the c.i.f. price does not improve the explanatory power of the model at the 10 percent level. This indicates that c.i.f. price uncertainty does not strongly affect either the f.o.b. or c.i.f. price level. However, a more general extension of (14) would be to examine the effects of all sources of price uncertainty and their covariance on the margin. This would be appropriate if actors involved in the acquisition, transport, and sale Of grain internationally face price uncertainty on both the buying and selling side. Such is the case for all international grain traders buying under spot or futures mechanisms from private inland handlers, elevators, and COOperatives. It would also apply for the large quantity of grain bought under competitive bidding from the U.S. government. The importance of price uncertainty depends on the actor in question. For internal suppliers in the importing country, the major source of price risk is c.i.f. price uncertainty (hz) through its direct effect on profit margins. This is especially true in countries such as Japan, where internal prices are fixed. The 11 Note that this is equivalent to examining the effect of uncertainty in the margin only if there is no uncertainty in the f.o.b. price. 217 TABLE 11-6 Maximum Likelihood Estimation Results for Equations (14) Dependent Variable Pfob Pcif MM Explanatory Variables Intercept 8.05 7.28 -0.77 (4.68)* (2.11) Pfob-1 0.71 0.28 -0.43 (8.41) (2.94) Pcif-1 (0.25) 0.81 0.56 (2.97) (6.71) Pfob-2 0.11 -0.13 -0.24 (1.23) (-0.90) Pcif-2 -0.17 -0.05 0.12 (-1.83) (-0.33) rr-1 -0.11 0.25 0.36 (-0.46) (0.64) Poil-1 0.006 0.014 0.008 (1.80) (2.30) th -0.001 0.02 0.02 (-0.07) (1.51) *: t-statistics are in parentheses 218 2 h h -O.87 -2.23 -0.87 -2.23 1.10 0.11 E , 6 , *6 , ' '2 <-0.65) <—2.48> <-0.65) <-2.48) (6.58) (0.77) 1‘ 1 1‘ 1 2‘ 1 a 4. 2 ”2‘ .02 “2.23 '3693 -203 '1693 -0618 0.29 £1t'1*€2t'1 6 Zt'T (‘2.48) (-0.99) (‘2.48) ('0.99) {-1.37) (2.06) 1.70 '0618 0.59 0.07 hTt" 7112:-1 0.59 0.09 (6.58) 4-1.37) (5.10) 40.64) (5.10) (0.71) a. 0.11 0.29 0.09 0.31 n , . 0.07 0.81 (0.77) (2.06) (0.71) (9.63) '2‘ 1 “2' 1 (0.84) (9.63) profits of grain handlers and marketing firms in exporting countries are more directly affected by f.o.b. price risk (hl). By contrast, international grain traders face price risk from both the import and export price; hence they may be most concerned with the uncertainty in the margin. It is possible that the variance in the margin is much lower than the variance of either the f.o.b. or c.i.f. price. This would occur if movements in export and import prices were highly correlated. This can be shown by formulating the identity MMt - Pcift - Pfobt in terms of forecast errors (via equation 11): E[MMt-MM*t] - E[Pcift-Pcif*t] - E[Pfobt-Pfob*t] Taking the variance of both sides gives EIMMt-MM*t]2 - E[Pcift-Pcif*t]2 + E[Pfobt-Pfob*t]2 _ 2*E[(pcift-Pcif*t)(Pfobt-Pfob*t)] 00' VAR(MMt) - hlt + hat - 2h12t 219 Therefore, the higher the conditional covariance between forecast errors in the import and export prices, the less risk incurred by international grain traders facing uncertainty from both sides of the market. This may have important implications concerning the incidence of risk bearing associated with volatile world prices. Under the assumption that both f.o.b. and c.i.f. price uncertainty influence behavior and price levels in the U.S.-Japan wheat trade, equations (14) represent misspecification, since both h1 and hz are modelled as constants in the conditional mean equation. Therefore, a more appropriate model would be (15) Pfobt - :10 + 111(Pf0b-1) + m12(Pcif-1) + «13(Pfob-2) + «14(Pcif-2) + 115(rr-1) + I15(Poil-1) + 611(h1t) + 612(h12t) + 613(h2t) + ‘It Pcift - £20 + 121(Pf0b-1) + l22(Pcif-l) + 123(Pfob-1) + 124(Pclf_2) + 125(rr-1) + 125(P011-1) + 621(h1t) + 622(h12t) + 623(h2t) + ‘2t ‘It h t h1 t ‘figN o,(: 1 2 €2t h12t h2t h1t h12t c1 c12 C1 C12 + A11 A12 £21t-1 61t-1*62t-1 h12t h2t c12 c2 c12 c2 A21 A22 e1t-1*€2t-1 ‘22t-1 A11 A21 + B11 B12 h1t-1 h12t-1 B11 821 A12 A22 B21 B22 h12t-1 h2t-1 B12 B22 220 Model (15) allows examination of how each source of price uncertainty affects price levels by direct inspection of the 5ij coefficients. The effect of each source of price uncertainty on the margin may be calculated as 62; - 61;. Maximum likelihood estimation results for (15) indicate that the presence of the conditional variance/covariance terms in the mean equations substantially increased the explained variation in the margin. A likelihood ratio test between models (13) and (15) rejects the null hypothesis that hlshlzshZ-O at well beyond the .005 level. Results indicate that forecast variances again tend to fluctuate between high and low periods, as shown by the paths of h1, h2 and “12 (Figures 3a-c). T-statistics well above 3.0 in the parameter matrix on Ht-I provide strong evidence of GARCH effects (Table II-7). Uncertainty in both the f.o.b. and c.i.f. prices again appeared highest during the early 19705 and 1985, in which the magnitude of hl and he were over five times their mean values over the sample period. High conditional variances during late 1985 coincided with legislation of major policy changes in the Food Security Act of 1985. The individual effects of hl and hz on the U.S. and Japanese prices were negative and quite large.. However, this overstates the net impact of uncertainty on the margin because these terms are both highly correlated with hlz, which has a positive effect on both prices. When uncertainty is increased from zero to the mean levels Of h1, h2, and “12 over the sample period, this is associated with a price drop Of $2.33 in the export price at U.S. Pacific ports. This is consistent with the theoretical contention that price uncertainty faced by 221 a ( ') ENS/MI 5900 «no. 300% (b) 2000 susxur 1500. (C) 680 0 , , I-,.. I- ,. - ..- H,. . 1972 _ 974 976 978 990 982 984 906 m . Figure II-3. Time Varying Conditional Variances and Covariances of U.S. Export and Japanese Import Price Forecast Errors from GARCH-M Model (15) 222 TABLE II-7 Maximum Likelihood Estimates of Equations (15) I Dependent Variable Pfob Pcif MM T Explanatory Variables Intercept 10.22 10.34 0.12 (4.76)* (2.46) Pfob-1 0.58 0.04 -0.54 (4.60) (0.23) Pcif_1 0.23 0.81 0.58 (2.77) (5.80) Pfob-2 0.23 0.03 -0.20 (1.99) (0.14) Pcif-2 -0.11 0.05 0.16 (—1.46) (0.36) rr-1 -0.002 0.35 0.35 (-0.01) (1.01) Poil-1 0.001 0.006 0.005 (0.34) (0.80) “It -0.001 -0.04 -0.04 (-0.05) (-1.30) h12t 0.06 0.16 0.10 (0.98) (2.08) th -0.09 -0.12 -0.03 (-1.95) (1.86) *: t-statistics are in parentheses 223 n n -1.09 -2.22 -1.09 -2.22 1.27 0.31 £2 - E _*E . 1 12 (-0.33) (-1.32) 1-0.33) (~1.32) (5.90) (1.60) 1‘ 1 1t 1 2t 1 I 0' -z.22 -1.66 -2.22 -1.65 -O.28 0.16 E _-e , £2 , “21 ”2 1-1.32) <-0.48) (-1.32) 1-0.66) (-l.89) 11.05) 1‘ 1 2t 1 2' 1 1.27 -0.23 0.44 -0.04 01,,, n12t_, 0.64 0.17 45.90 1-1.60) (3.25) t-0.32) (3.25) (1.18) 4' 031 0J6 0J7 (L89 11 . , -04x our (1.60) (1.05) (1.18) (0.23) 12‘ 1 “2‘ 1 (-0.32) (8.23) international marketing firms tends to depress export prices. Given that the U.S. exports approximately 3.3 million tons of wheat to Japan each year (International Wheat Council,various years), the net loss to the U.S. wheat sector may have been upwards of $7.6 million per year due to the adverse effects of price uncertainty in the U.S.-Japan margin.12 Surprisingly, the same increase in price uncertainty is associated with a $1.90 drop in the c.i.f. Japan price. This result is inconsistent with theoretical expectations only if international grain traders are assumed to be the lone actors in the system facing price risk. When the net effect of price risk on all actors in the system is considered, the results are consistent with the premise that Japanese internal suppliers react to price uncertainty by exerting downward pressure on import prices in order to protect profit margins. Their 12 This assumes that either all U.S. wheat exports to Japan originate from Pacific ports, or that uncertainty in the margin has similar effects on the export prices at other U.S. ports used to market wheat to Japan. 224 ability to do this may owe to the extent of market concentration in Japan’s wheat importing industry. This is discussed further below. When uncertainty is increased from zero to the mean levels of h1, h2, and "12 over the sample period, the U.S.-Japan marketing margin increased by 5.43 per ton, although this effect was statistically insignificant at the .05 level.13 This is because each source of price uncertainty was associated with a roughly parallel movement in both the f.o.b. and c.i.f. price. What would be the effect on U.S. wheat prices of a return to the levels of uncertainty measured during the volatile 1970s? If the levels of h1, ha, and “12 were raised 650 percent from their mean levels to those recorded during parts of 1973, 1974 and 1985, this (would be associated with a drop in U.S. Pacific wheat prices of $15, or about 10 percent of average U.S. wheat export prices over the 1972-86 period. However, during most years, the conditional variance/ covariance terms rarely deviated more than 100 percent above their mean levels. . The results also suggest that the conditional variance of the margin is quite a bit less than that of either price. This is because "12 is positive, indicating that the U.S. and Japanese prices tend to fluctuate somewhat in sync with each other. Studies that consider only output (sale) price uncertainty may therefore overstate the extent of uncertainty faced by international marketing agents. It would not, 13 The appropriate test statistic is 621-611/[(ST0521)2 + (STDg11)2] 2 where i-1,2,3 for the coefficients on h h} and hz in the wo price equations. In each case, the null hypo heSTs that 631-62;-611 could not be rejected at the .05 percent level. 225 however, overstate the uncertainty faced by internal suppliers in the importing country. Finally, the conditional variance of the US export price was more than twice as large as that of the Japanese import price (see Figures 3a and 3c). This implies that the U.S. price is more unpredictable than the Japanese price. While the model cannot ascribe a definitive cause for this finding, several conjectural hypotheses can be forwarded. Parker and Conner (1979), examining U.S. food manufacturing industries, found a strongly positive relationship between market concentration and net margins. Related research also found that U.S. food prices are in general substantially more stable (and perhaps more predictable) on the highly concentrated retail level than on the farm- first handler level, where markets are relatively competitive (Marion and Ward, 1986). If the conditional mean and variance of prices are determined simultaneously as previously suggested, then Parker and Conner’s conclusions concerning market concentration may also affect higher moments of the price distribution. The premise that market concentration affects not only price level but also price stability is consistent with the findings of Marion and Ward regarding the domestic U.S. food system as well as the findings in this study that U.S. Pacific wheat export prices are substantially more uncertain than Japanese import prices. The Japan Food Agency’s semi-monopsony status may exert downward pressure on both the level and variance of import prices. By contrast, the U.S. wheat export market, while also fairly concentrated, still appears characterized by rivalrous competition between a limited number of actors (Conklin, 1982). This conjectural 226 proposition may have implications for further research on the relationships between pricing outcomes and market structure in international markets. 4.2 COMPARISON OF RESULTS FROM ALTERNATIVE INDICATORS OF PRICE UNCERTAINTY By applying alternative measures of uncertainty to the U.S.-Japan data set, one may discern how robust these results are to the specification of uncertainty. This section applies the four alternative measures presented in Chapter 3 to the U.S.-Japan wheat margin model to measure the degree of correlation between these and the GARCH-M measures of uncertainty, and examines whether the empirical effect of uncertainty on price and margin levels are sensitive to the measure used. Monthly Japanese wheat import price uncertainty was measured over the same period as the GARCH estimation (January, 1972 to November, 1986). While the Original studies used different lengths of past prices to calculate each measure of uncertainty, 12 months are chosen here to standardize the analysis. Only import price uncertainty was examined, to be consistent with the specification used by the other marketing margin studies cited. 4 Table 11-8 contrasts the correlation coefficients between the following indicators: a. mozing average of percentage changes in price (Anderson, 19 9 ; b. weighted moving average of absolute price changes (Brorsen et al., 1985); . 227 TABLE 11-8 Correlation Coefficients of Alternative Measures of Uncertainty Applied to Monthly Japanese Wheat Import Prices, 1970-1986 (a) (b) (c) (d) (e) (f) (a) 1.00 .92 .89 .78 .74 .24 (b) 1.00 .93 .75 .74 .27 INDICATOR (C) 1.00 .68 .71 .23 (d) (e) 1.00 .77 1.00 .30 .46 (f) 1.00 Key: (a) average of last twelve month’s percentage price changes b weighted moving average of absolute price changes over past twelve months, divided by mean price moving coefficient of variation of price over past twelve months moving variance of monthly percentage price changes over past twelve months th derived from GARCH-M model (14) conditional variance (h1t+h2t-2*h12t) derived from GARCH-M model (15). 228 c. coefficient of variation of past monthly prices (Grant et al., 1984); d. variance of percentage changes (Dalziell, 1985); e. the conditional variance of the import price as measured by th in GARCH model (14); f. the conditional variance of the marketing margin as measured by h1t+h2t-2*h12t in GARCH model (15). None of the four measures (a-d) were correlated with the GARCH-M measure (f) above 0.30. This is not surprising because the GARCH model (15) measures uncertainty as a function of past forecast errors in the export and import prices, while the others measure Only import pricevariability. But not even measure (e), the conditional variance of the import price from (14), was correlated with any of the four alternativeindicators above 0.77. Correlation coefficients among the price variability indicators (a-d) generally ranged from about .70 to .90. While not an obvious sign of inconsistency in the measurement Of uncertainty across studies, several pairs appear different enough to warrant concern. If the magnitude of price uncertainty and its effects are very sensitive to the procedure used to measure it, then the development of an empirical record to understand the effects Of price uncertaint --based on inconsistent measures--must be viewed with some Skepticism. Limited correlation between the various measures of risk would suggest that empirical results concerning their effects may be sensitive to the procedures used to measure it. In spite of this, Brennan (1982) contends that: “Researchers who have compared different variables to represent risk in prices in econometric models of supply have found no superiority for the more complex variables over the 229 simpler variables." Although he does not explicitly state his criterion for superiority, it appears to be the extent to which the complex models yield different empirical results. To examine this contention, U.S. and Japanese prices were estimated using each indicator of price variability generated from the measures (a-d) in Table II-4. (15) Pfobt - 110 + 111(Pf0b-1) + I12(Pcif-1) + 113(Pfob-2) + «14(Pcif_2) + 115(rr-1) + 115(P011_1) + 01(RISKt) + Flt Pcift - £20 + 121(Pf0b-1) + 122(Pclf-l) + X23(Pf0b-1) + 124(Pcif-2) + 125(rr-1) + 125(P0il-1) + 02(RISKt) + EZt where RISKt is the representation Of c.i.f. price variability as defined by indicators (a-d) in Table II-4. Subtraction of Pfobt from Pcift yields the marketing margin equation (17) MMt - Pcift - Pfobt . 130 + N31*(Pf0b-l) + N22*(Pcif_l) + N33*(Pfob-2) + 134*(PCIF-2) + I35*(Pf0b_3) + 135*(Pclf-3) + [37*(Poil-1) + «38*(rr-1) + 03*(RISKt) + €3t where, as before, x3j - «23 - nlj; and 03 - 02 - 01. Model (16) is identical to the GARCH conditional mean models (14) and (15) except for the formulation of price uncertainty. 4 The OLS estimates of equations (16) and the computed margin coefficients using the four alternative measures are presented in Table II-9. There is a high degree of consistency across all of the results. Price uncertainty in the c.i.f. Japanese price was strongly 230 TABLE 11-9 OLS Estimates of Equations (16)* Dependent Variable U.S. Pacific Port Price Japan Import Price Uncertainty Indicator Uncertainty Indicator (1) (2) (3) (4) (1) (2) (3) (4) Constant 11.10 11.69 12.29 13.21 5.48 5.90 6.27 7.30 (3.50) (3.78) (3.99) (4.34) (1.64) (1.79) (1.92) (2.25) Pfob-1 1.03 1.02 1.06 1.03 0.57 0.57 0.60 0.58 (9.92) (9.73)(10.40)(10.07) (5.19) (5.13) (5.57) (5.30) Pcif-1 0.17 0.17 0.16 0.18 0.78 0.78 0.76 0.78 (1.76) (1.83) (1.65) (1.87) (7.76) (7.76) (7.62) (7.86) Pfob-2 -0.47 -0.45 -0.48 -O.39 -0.46 -0.50 -0.52 -0.43 (-3.6)(-4.26)(~4.50)(-3.49) (~3.3l)(-4.40)(~4.65)(-3.64) Pcif-2 -0.01 0.08 0.09 0.01 -0.09 0.06 0.07 -0.01 (-0.1) (0.85) (0.95) (0.08) (-0.79) (0.60) (0.74)(-0.12) Pfob-3 0.01 0.13 0.09 0.05 -0.11 0.06 0.04 0.05 (0.06) (0.60) (0.52) (0.34) (-0.87) (0.35) (0.21) (0.23) Pcif-3 0.05 -0.00 0.02 0.01 0.19 0.02 0.05 0.02 (0.60) (0.04) (0.05) (0.06) (2.04) (0.24) (0.02) (0.01) rr-l 0.81 0.78 0.76 0.63 0.82 0.78 0.75 0.63 (2.71) (2.60) (2.51) (2.08) (2.61) (2.42) (2.34) (1.94) Poil-l 0.00 -0.00 -0.00 -0.00 0.00 0.00 0.00 0.01 (0.05)(-0.56)(-0.78)(-0.13) (1.25) (0.78) (0.91) (1.27) RISKt 18.34 -15.74 34.06 416.08 19.58 17.30 40.18 423.29 (3.07) (2.36) (2.09) (2.70) (3.11) (2 03) (2.33) (2.58) 0 R2 .93 .93 .93 .93 .95 .95 .95 .95 OH 1.97 1.95 1.96 1.93 1.94 1.97 1.97 1.93 F 245 310 308 313 367 451 454 457 *° t-statistics are in parentheses 231 associated with higher mean c.i.f. prices, for all of the uncertainty measures examined. Import price uncertainty also appeared to increase the U.S. export price to a lesser degree. The t-statistics on the RISK coefficients exceeded 2.0 for each measure examined. However, import price variability appears to have only a weakly positive influence on the margin. This is because the results suggest that price variability raises both the f.o.b. and c.i.f. prices to a similar extent. A t-test on 03-0 could not be rejected at the 10 percent level for any of the indicators. These results are consistent with GARCH model (14), which also considers only the effect of import price risk. However, they are diametrically opposed to the results of the GARCH-M model (15). The latter suggests that uncertainty reduces both import and export prices, but especially the export price, while the other models suggest that uncertainty raises both prices. In reconciling these differences, it is necessary to emphasize that the two sets of models are measuring two different things. The GARCH model (15) examines uncertainty in both port prices and their covariance. The other models (a-d) examine variability in the import price. As demonstrated in Table II-B, these measures are not even moderately correlated with each other. The main difference appears to be the source of uncertainty measured (i.e. import price versus margin) rather than variability versus uncertainty. This is because the GARCH model (14), which measures uncertainty in the import price and therefore shares common elements to both types of models, is much more 232 highly correlated with the four alternative models (a-d) than it is with the GARCH model (15). Therefore, the relevant conclusions and policy implications of this study largely depend on which source(s) of uncertainty better reflect the conditions faced by actors in the system.14 As mentioned previously, a significant proportion of U.S. wheat marketed through export channels to Japan has the following pricing characteristics: (a) on the 0.5. side, private international traders purchasing from inland elevators, cooperatives, or government with f.o.b. prices determined from spot, futures, or competitive bidding; (0) vertical integration from first handler in the U.S. to internal supplier in Japan; (c) bilateral contracts under "supply agreement", in which the Japanese Food Agency buys a significant portion of its wheat under basis rather than flat pricing; and (d) large Japanese trading houses procuring U.S. grain through spot and/or futures market transactions. With the exception of point (b), these coordination arrangements suggest that transaction prices are uncertain both on the export and import sides of the marketing Channel, and that a model allowing for uncertainty in both prices (and their covariance) is most realistic. Therefore, the remainder of the study analyzes conclusions and implications flowing from the GARCH-M results of model (15). 1‘ Since the estimated equations are considered reduced form-- i.e., they measure the net effects of many actors’ behavior--the model is not constructed from any particular agent’s perspective, nor does it measure any particular agent’s behavior. Rather the model measures the net effect on the system. CHAPTER 5 CONCLUSIONS AND POLICY IMPLICATIONS The research results have both empirical and methodological conclusions. These are addressed in turn. Several of the empirical conclusions are somewhat conjectural and are forwarded as hypotheses subject to further research. 5.1 EMPIRICAL CONCLUSIONS (1) (2) Uncertainty in the U.S. Pacific and Japanese wheat prices was strongly correlated with a reduction in the U.S. price level. The conditional variances and covariances of the forecast errors increased the explained variation in port prices beyond the .005 level of significance. When these terms are increased from zero to their mean levels over the 1972-86 period, the U.S. price dropped by $2.33 per tOn. Given that the U.S. exports over 3 million tons of wheat to Japan annually, the loss to the U.S. wheat sector has been upwards of $7 million per year due to the adverse effects of price uncertainty in the U.S.-Japan wheat trade. This result is consistent with the theory that international marketing firms require a risk premium that depresses export prices when risk increases. However, the incidence of risk bearing is not uniform. When the conditional error variances and covariances were increased from 233 (3) 234 zero to their mean level over the sample period, the Japanese import price actually fell by $1.90 per ton. This is inconsistent with standard marketing models in which the producer (export) price should fall and consumer (import) price should rise as marketing margins increase, with the incidence of risk sharing determined by the relative elasticities of supply and demand. However, such models are based on the assumption that middlemen are the only actors in the system affected by price risk. When risk is allowed to affect the behavior of internal suppliers and exporters, the theoretical effect of price risk on price levels is less clear, especially when final consumer prices are fixed. This study presents a reduced-form model which examines the net effect of price risk on price levels, without deducing any structural interpretations. It therefore has the drawback of not providing any information about how particular actors respond to price uncertainty. However, if internal suppliers in Japan are risk averse, they may also attempt to protect operating margins by exerting downward pressure on c.i.f. prices when uncertainty increases. This premise is consistent with the empirical results stated above. Price uncertainty in the U.S. export price was over twice the magnitude of that in the Japanese price. Extending the findings of Parker and Conner (1979) regarding U.S. food manufacturing industries, it is possible that market concentration affects both the level and uncertainty of prices. If a state trading agency has strict control over import quantity and internal prices, it is (4) (5) (5) 235 not subject to the same uncertainty about the behavior of competitors and their influence on price. Thus it is possible that level of concentration may be an explanatory variable for price stability and predictability, as was found in the U.S. food system (Marion and Ward, 1986), in which prices at the relatively competitive farm level appear to be much more volatile and uncertain than on the concentrated retail level. The U.S.-Japan wheat margin increased by 5.43 per ton when uncertainty is increased from zero to the mean levels of the conditional variances and covariance; However, this effect was not significant at the .05 level. Uncertainty in the U.S.-Japan wheat margin was substantially Smaller than that of either price. This is because the U.S. and Japanese prices were somewhat correlated, reducing the overall variance in the margin. Studies that consider only output (sale) price uncertainty may therefore overstate the degree of risk incurred by actors facing both acquisition and sales price uncertainty. During periods of 1973, 1974, and 1985, the variance Of the margin was over 7 times higher than its mean level since 1980. The results suggest that a 7-f0ld increase in hl, hz and “12 from their mean levels is correlated with a $16 decline in U.S. export prices, or 10.7 percent of its mean level. Thus, a reversion to the price uncertainty of the early 19705 would impose non-marginal costs on the U.S. wheat export sector, as risk premia become 236 incorporated into price levels. However, such episodes of price uncertainty were not common. lic m ' ion If further research supports these findings for the U.S.-Japan wheat trade, they present important implications for U.S. trade and stockholding policies. Insuring against risk is not costless. A reduction in price risk borne by actors in the international grain trade may alleviate costs and distribute gains throughout the system. Policies that stabilize international grain prices may benefit U.S. wheat producers and others in the U.S. grain export system, not just by providing greater production incentives, but also directly through greater average revenues associated with reduced international marketing costs. It has frequently been articulated that freer trade would provide greater international price stability (Johnson, 1975; Shei and Thompson, 1978; Zwart and Meilke, 1979). According to this school of thought, the U.S. may achieve benefits related to risk-reduction in the marketing system, apart from other benefits attributed to freer trade. However, Part I of this study questions the extent to which a reduction in government insulation policies would stabilize world grain prices. This is because of a variety of domestic production and consumption rigidities that tend to cause inelastic national trade functions, even in the absence of government trade restrictions. More broadly, the beneficial effects of stabilized international prices for the 0.5. may enhance the desirability of trade and stock 237 policies with some undesirable features. Although U.S. wheat programs have been criticized as costly, they have indirectly tended to stabilize world markets due to massive grain stocks accumulated as a side effect. The willingness of the U.S. to accumulate large stockholdings during periods of world glut, and to release them on a sufficiently large scale during tight periods, has greatly contributed to world grain price stability (Josling and Barichello, 1984). The results of this study suggest that such policies have indirectly benefitted U.S. wheat interests through the effects of risk reduction on actors in the international marketing system. As stated by Binkley (1983, p. 63), “If the question of unstable international grain markets is as important as the amount of previous work suggests, continuing to ignore the marketing element is unjustified.“ 5.2 METHODOLOGICAL CONCLUSIONS This report has presented a measure of price uncertainty modeled as a function of past forecast errors. It is thus consistent with the definition of uncertainty as a function of the squared deviation between expected and actual price. In this regard, it differs from other commonly used empirical measures that resort to the notion of variability rather than uncertainty per se. When applied to the Japanese wheat prices, these price variability indicators were only moderately correlated with price uncertainty as measured by the GARCH-M model (ranging from .71 to .77). These measures were very weakly correlated with the GARCH-M model that considered both sources of price uncertainty and their covariance (ranging from .22 to .27). 238 Unsuprisingly, the price variability measures yielded very different empirical results concerning its effect on price levels. Econometric results using the four price variability indicators each suggested that price variability raises both U.S. export and import prices. An unfortunate general conclusion from this is that as long as analysts continue to measure uncertainty in different and inconsistent ways, the development of a reliable, useful empirical record concerning the effects of price risk will be impeded. It is little wonder that considerable disagreement persists about the effects of risk in agricultural trade and appropriate policy responses to it. 5.3 AREAS FOR FURTHER RESEARCH The model used in this study measures the effect of price uncertainty on the behavior of all actors in the international wheat marketing system. In the process, behavior responses by particular marketing agents are obscured. A structural approach would provide valuable information concerning which actors are affected by risk and in which way. This would help to interpret the reasons why price uncertainty is apparently negatively correlated with the Japanese import price. The proposition forwarded in this study--that Japanese internal suppliers require a greater risk premium to protect profit margins when uncertainty rises--may be better addressed in a structural framework. However, a structural model that examines the net effect of price uncertainty on the system would be more difficult to estimate since it must include behavioral equations for all actors affected by price risk in the system under analysis. In the U.S.-Japan wheat 239 trade, this would include farmers, domestic transporters and handlers, cooperatives, international grain traders, state trading agencies, and internal suppliers in Japan. Furthermore, a structural approach would have to account for the organizational and institutional determinants of price and margin fluctuations as addressed in Chapter 2. A larger empirical base is needed to support or refute the findings of this study. Because of the time and computational costs involved in GARCH-M estimation, only one grain trade route was analyzed. A broader sample size is necessary to clarify the importance and distributional effects of international price uncertainty on U.S. and other national grain interests. REFERENCES Anderson, M. 1989. 'U. S. Soybean Trade and Exchange Rate Volatility. n A i u . Vol. 71, No. 3, forthcoming. Aradhyula S. and M. Holt. 1988. ”Risk Behavior and Rational Expectations in the U.S. Broiler Market.” Working Paper 88-WP 33, Center for Agricultural and Rural Development, Iowa State University, Ames. Batra, R. and A. Ullah. 1974. "Competitive Firm and the Theory of Input Demand under Price Uncertainty.“ Dournal pf Pplitieal Eepnpny. Vol. 82, pp. 537-48. , Behrman, J. 1968. Supply Response in Undergevelopeg Agrieulture; Case Study pf Epur Major Annual traps in Thai lan a, 1237 -§S, Amsterdam. North- Holland Publishing Co. Binkley, J. 1983. "Marketing Costs and Instability in the International Grain Trade.” Aneriean Dpurnal of Agricultural Eepnpnies. Vol. 65, No. 1, pp. 57-63. Bollerslev, T., R. Engle, and J. Wooldridge. 1988. "A Capital Asset Pricing Model with Time-Varying Covariances." Journal of £p11tiea1_£epnpny. Vol 96, No. 1, pp. 116-31. Bollerslev, T. 1986. ”Generalized Autoregressive Conditional Heteroscedasticity.“ Dpurnal pf Eepnpnetries. Vol. 31, April, pp. 307-27. Brennan, J. 1982. "The Representation of Risk in Econometric Models of Supply: Some Observations.“ i urn A ri ult al Eepnpnies. August, pp. 151-156. Brorsen, B.W., J-P. Chavas, and W. Grant. 1987. "A Market Equilibrium Analysis of the Impact of Risk on the U.S. Rice Industry." r l l n . Vol. 69, No. 4, pp. 733-39. Brorsen, B.W., J-P. Chavas, W. Grant, and L. Schnake. 1985. “Marketing Margins and Price Uncertainty: The Case of the U. 5. Wheat Market." u al 0 omic Vol 67, No. 3, pp. 521-28. Business Week. 1978. “The Rice Industry Splits over a Futures Market." February 13, pp. 88-90. 240 241 ~ Caves, R. 1979. "Industrial Organization and the Problem of Thin Markets.“ in Pricing Prpblems in the Food Industry. Marvin Hayenga (ed.). N.C. Project 117 Monograph 7, Research Division, College of Agriculture and Life Sciences, University of Wisconsin, Madison. Conklin, N. 1982. 'An Economic Analysis of the Pricing Efficiency and Market Organization of the U.S. Grain Export System." GAO/CED-BZ- 61s, U.S. General Accounting Office, Washington, D.C. Dalziell, I. 1985. ”Sources of Agricultural Market Instability." unpublished Ph.D. dissertation, Department of Agricultural Economics, Michigan State University. Engle, R., D. Lilien, and R. Robins. 1987. “Estimating Time Varying Risk Premia in the Term Structure: The ARCH-M Model." £epnpnetri_a. Vol. 55, No. 2, pp. 391-407. Engle, R. 1982. "Autoregressive Conditional Heteroscedasticity with Estimates of the Variance of United Kingdom Inflation." Eepnpnetriea. Vol 50, No. 4, pp. 987-1007. Fisher, 8.5. 1981. "The Impact of Changing Marketing Margins on Farm Prices.” American Journal of Agrieultural Eeonomies. Vol 63, No. 2, pp. 261-63. Gardner, B.L. 1977. 'Instability in U.S. Agriculture: Discussion." ri n J urnal f A ri lt o omi . Vol 59, No. 1, pp. 185-87. Gilmore, R. 1982. A oor H rv : C h Policies and Interests in the Grain Trage. Longman Co., New York. Grant, W., J. Richardson, B.W. Brorsen, and M. Rister. 1984. Economic Impacts of Increased Price Variability: A Case Study with Rice.” ri n i R ar . Vol 36, No. 4, pp 17-27. Harrison, K. 1988. Director, K. Harrison and Associates. Personal communication. Hurt, C. and P. Garcia. 1982. “The Impact of Price Risk on Sow Farrowing, 1967-78.“ ri l r n 1 Vol. 64, pp.565-68. International Wheat Council. Various years. “World Wheat Statistics.” International Wheat Council. London. International Wheat Council. 1986. ”Ocean Freight Rates." in “Review of the World Grain Situation." International Wheat Council. London. pp. 25. 242 International Wheat Council. 1977. "Developments in the Ocean Transportation of Grain from 1947-1977." in "Review of the World Grain Situation.“ International Wheat Council. London. 1977. pp. 95-100. Johnson, D.G. 1975. “World Agriculture, Commodity Policy, and Price Yariablility-' Amer1san_J9urDal_9L_Agrisultural_£sonemiss. Vol 57, p. 823-28. Josling, T. and R. Barichello. 1984. “International Trade and World Food Security.“ Epod Ppliey. November, pp. 317-330. Just, R. 1974. 'An Investigation of the Importance of Risk in Farmers’ Decisions." Aneriean Dournal pf Agrieultural Eggnomics. Vol. 56, No. 1 pp. 14-25. Kroner, K. 1988. "The MGARCH Program.” mimeo, Department of Economics, University of California at San Diego, La Jolla, California. Lin, W. 1977. “Measuring Agricultural Supply Response under Instability." Aneriean qurnal of Agricultural Econpmies. Vol. 59,‘pp.903-7. Mandlebroit, B. 1963. ”The Variation of Certain Speculative Prices." n f ne 5. Vol 33, pp. 394-419. Marion, 8. and R. Ward. 1986. "A Comparison of Agricultural Subsectors.‘ in Ihe Drganizatipn ang Eerfprmanee at the D.S. Food System. 8. Marion (ed.). Lexington Books, Lexington, Ma. McNees, S. Sept./0ct, 1979. “The Forecasting Record for the 1970s." d mi v w. pp. 326-34. Parker, R. and J. Conner. 1979. ”Estimates of Consumer Loss due to Monopoly in the U.S. Food Manufacturing Industries." Aneriean J9urna1_9£_Agrisultural_Esonemiss. Vol. 61. pp- 628-639- Peterson, V., Vice President, U.S. Wheat Associates, personal communication, 1988. Robison. L. and P. Barry. 1986. Ihe_99mDeI111!9.E1£DL5_89509059_10 Risk. Macmillan Publishing Co., New York. Sandmo, A. 1971. 'On the Theory of the Competitive Firm under Price Uncertainty." Aneriean_£epnpnie_keyie_. Vol. 61, pp. 65-73. Shei, S. and R. Thompson. 1977. “The Impact of Trade Restrictions on Price Stability in the World Wheat Market.“ Anerisan Dournal of Agricultural_£sunemiss Vol 59. No. 4. pp 629-38- 243 Sims, C. A. 1986. "Identifying Policy Effects." Paper prepared for Brookings Institute International Models Comparison Project. Brookings Institute, Washington, D.C. Tomek, W. and K. Robinson. 1972. Agricultural Product Prices. Cornell University Press, Ithaca, NY. Traill, B. 1978. "Risk Variables in Econometric Supply Response Models.“ r lt r onomi . Vol. 29., pp. 53-61. USDA. Various Issues. "Wheat Outlook and Situation." Economic Research Service. Washington, D.C. Zwart, A. and K. Meilke. 1979. “The Influence of Domestic Pricing Policies and Buffer Stocks on Price Stability in the World Wheat Industry.” e Jo f ' ultural c nomics. Vol 61, No. 3, pp. 434-49.