WPS5355 Policy Research Working Paper 5355 Markets for Cotton By-Products Global Trends and Implications for African Cotton Producers John Baffes The World Bank Development Prospects Group & Africa Region June 2010 Policy Research Working Paper 5355 Abstract This paper analyzes and compares the structure of cotton countries, cotton by-products should be taken into by-products industries in selected countries (Uganda, consideration, both in terms of the cotton price setting Tanzania, Benin, and Burkina Faso) in the context of the mechanism and the size of the organization of the cotton global vegetable oil market. It reaches several conclusions. by-products industry. Fourth, trade policies including First, because the markets for various edible oils are export bans or import tariffs to protect the domestic highly integrated with each other, examination of each crushing industries, and policies that favor crude over oil market should be done in conjunction with all other refined oils, should be rationalized. Fifth, large cottonseed (relevant) edible oil markets. Second, the recent surge in processing operations using advanced technology, while demand for commodities used as feedstocks for biofuels efficient from a technological perspective, tend not to is unlikely to become a new source of growth for the be economically profitable in the African context. Last, cotton oil market. Third, within the context of deepening research efforts for new cotton varieties should consider the on-going reform efforts in West and Central African the value of by-products, not just lint. This paper--a joint product of the Development Prospects Group and the Africa Region--is part of a larger effort to gain a deeper understanding of the contribution of cotton by-products to the welfare of cotton growers in the context of the recent commodity price boom. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at jbaffes@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Markets for Cotton ByProducts: Global Trends and Implications for African Cotton Producers John Baffes THE WORLD BANK jbaffes@worldbank.org The views and findings of this paper reflect those of the author and not the World Bank. I would like to thank Rafiq Chaudhry, Armelle Gruère, Valerie Kelly, Patrick Labaste, Stephen Mink, Ale jandro Plastina and Marc Sadler for comments and suggestions, as well as seminar participants at the World Bank (December 2009) and the International Cotton Advisory Committee (May 2010). Rachel Weaving provided excellent editing. The results of the country studies are based on find ings from field visits by Kathryn Vasilaky and Laoura Maratou (Uganda, May 2009), Kathryn Vasi laky (Tanzania, June 2009), and Noureni Zanfongnon (Benin, June 2009 and Burkina Faso, July 2009). The author is grateful to the owners and managers of the cottonseed processing facilities as well as the heads of the various cotton organizations for their time and information provided dur ing the interviews. Financial support for the study was provided by the Development Prospects Group and the Africa Region of the World Bank, as well as the EUfunded Trust Fund, TF092488. Acronyms and Abbreviations ADF Augmented DickeyFuller test for unit roots AICB Association Interprofessionnelle du Coton du Burkina APHB Association des Producteurs dhuile des HautsBassins CFAf Communauté Financière Africaine Franc (US$ = 472 CFAf in 2009) CFDT Compagnie Française pour le Développement des Fibres Textiles) CPPOD Coopérative des Producteurs de Produits Oléagineux et Divers DAGRIS Société du Développement Agricole du Sud ESA Eastern and Southern Africa GHH Groupement des Huiliers du Houet GTPOB Groupement des Transformateurs des Produits Oléagineux du Burkina IBCG Industrie Beninoise des Corps Gras ICAGIE Industries Cotonnières AssociéesGroupement d'Intérêt Économique jc 20liter vegetable oil container (equivalent to 18 kgs) LCB Le Label Coton du Bénin NCPA National Cottonseed Products Association (of the US) SNCITEC Société Nouvelle Huilerie et Savonnerie Citec SNHB Société Nationale des Huileries du Burkina SOCOMA Société Cotonniére du Gurma SODECO Société de Développement du Coton SOFIB Société Industrielle Barro et Frères SOFITEX Société Burkinabé des Fibres et Textiles SONAPRA Société Nationale pour la Production Agricole SONICOG Société Nationale des Industries des Corps Gras SSA SubSaharan Africa TZ shs Tanzanian shillings (US$ = 1,320 TZ shs in 2009) UEMOA Union Economique et Monétaire OuestAfricaine UG shs Ugandan shillings (US$ = 2,030 UG shs in 2009) WCA West and Central Africa WTO World Trade Organization VAT Value added tax Page i Contents Acronyms and Abbreviations ..................................................................................................................................... i 1. The Value and Uses of Cotton ByProducts ................................................................................................ 2 2. Cotton Oil in the Context of the Global Edible Oil Market .................................................................... 9 3. Issues in the Use of Cotton Oil for Biofuels ............................................................................................. 14 4. The Cotton ByProducts Industry in SubSaharan Africa ................................................................. 17 5. Conclusions and Policy Implications ......................................................................................................... 22 Appendix A: Uganda ................................................................................................................................................... 28 Appendix B: Tanzania ................................................................................................................................................ 32 Appendix C: Benin ....................................................................................................................................................... 39 Appendix D: Burkina Faso ....................................................................................................................................... 43 Appendix E: United States ....................................................................................................................................... 57 Appendix F: Questionnaire for Cottonseed Processing Companies ...................................................... 61 Page ii Markets for Cotton ByProducts: Global Trends and Implications for African Cotton Producers The recent commodity boom, and especially the price increases for edible oils combined with the relatively stagnant prices for cotton, has reawakened interest in the potential of cotton byproducts as a complementary source of revenue for cotton growers. Cotton oil, like other edible oils, has received particular attention in this regard because of its potential use as a feedstock for biofuels. The expan sion of world demand for biofuels, including edibleoilbased biodiesel, has raised hopes that the development of biofuel industry might transform the Afri can edible oil industry and the prospects for African cotton producers. A key conclusion of a recent multicountry cotton study by the World Bank (Tschirley, Poulton, and Labaste 2009) is that cotton byproducts in general have a growing market and are potentially an important complementary source of revenue for cotton growers. The same study also concludes that the markets for cotton byproducts in SubSaharan Africa are not well developed and have considerable scope for improvement, especially in the context of the prices re ceived by farmers (Gergely and Poulton 2009). On the other hand, cotton by products are an area that has not received much attention.1 To assess more thoroughly the prospects of cotton byproducts industries in SubSaharan African countries, this paper analyzes the global market for cot ton oil in the context of the global vegetable oil market and demand for biofuel feedstock, and then compares the structure and recent behavior of cotton by products industries in four African countries--Uganda, Tanzania, Benin, and Burkina Faso. The paper also looks briefly at the United States, not only because its much longerestablished industry provides a useful benchmark, but also be cause of the nature of cotton ownership. Because cotton in most countries (in cluding SSA) is primarily a smallholder crop, seed cotton is typically bought by traders or ginning companies at the farm level, and thus the ownership of cotton is transferred. In the United States (as well as some other cotton producing coun tries such as Australia and to a lesser extend Brazil), cotton growers retain own ership of the crop after ginning. Further, because many cotton producing coun tries (notably several in West and Central Africa and Central Asia) have few gin ning companies, the markets for byproducts are characterized by oligopolistic 1 In a study undertaken at about the same time as the present study, Kelly and others (2010) ex amine at the cotton byproducts sectors in Benin, Burkina Faso, Chad, and Mali. That study's key objectives are to describe the evolution, organization, and performance of the byproducts indus tries as well as their institutional, political, and technological constraints. Page 1 and oligopsonistic arrangements, making the pricing of cotton byproducts less transparent. This paper uses the analytical framework developed in the World Bank's multicountry cotton study. It finds that though the SSA countries are unlikely to be able to produce biodiesel profitably from cotton oil at foreseeable levels of global energy prices, both cotton oil and meal production should feature explicit ly in these countries' cotton sector policymaking. These byproducts have in creased their contribution to the total value of cotton and seem likely to maintain their enhanced value. The paper also finds that large cottonseed processing op erations using advanced technology, while efficient from a technological pers pective, tend not to be economically profitable in the SSA context. Smaller, less technologically efficient, laborintensive operations perhaps constitute the future of the industry in SSA. The paper is structured as follows. Section 1 examines the value and uses of cotton byproducts, and recent trends in their contribution, versus that of lint, to the total value of cotton. Section 2 looks at the cotton oil market in the context of the global vegetable oil market, while Section 3 discusses issues regarding the use of cotton oil as a biofuel feedstock. Section 4 summarizes experience with the use of cotton byproducts in the five countries. Section 5 offers policy conclusions and recommendations with the goal of enhancing the contribution of cotton by products to the welfare of African cotton growers. 1. The Value and Uses of Cotton ByProducts Though cotton is often taken to be synonymous with cotton lint, the farm prod uct, seed cotton, also contains the byproduct cottonseed.2 Of lesser value than lint, cottonseed is typically crushed to separate oil from meal. Cotton oil, which competes with other vegetable oils, is typically used for human consumption and sometimes for soap manufacturing and other industrial uses such as plastics and pharmaceuticals. Cotton meal, which competes with other meals, is typically used as animal feed, while hulls or husks (the outer part of the seed) are either used to generate energy for the processing facility or are blended with meal for animal feed; less often hulls and husks are used as fertilizer. Linters--which are 2 For the rest of this paper, seed cotton will refer to the farm product. Cottonseed and lint are the two coproducts emerging from the ginning process and, depending on the context, output can be measured in terms of tons of lint or tons of cottonseed. Lint goes to textile manufacturing while cottonseed is crushed to produce cotton oil and cotton meal. To make the discussion easier, in this paper meal will refer to both meal and cake even though the two are different. Page 2 lowquality lint left over from the ginning process--are used in various manufac turing products, including yarn, plastics, and filling material (see ICAC 2000 for a detailed description of uses of cotton byproducts.) Sometimes cottonseed is used unprocessed, directly for animal feed, especially in the dairy industry. This paper will discuss the two key byproducts: oil and meal. Though the composition of seed cotton by weight depends on numerous factors, typically one ton of seed cotton yields between 35 percent and 40 percent lint, about 10 percent cotton oil, and about 30 percent cotton meal.3 The relative proportions partly depend on the ginning outturn ratio, which ranges between 30 percent and 40 percent of the weight of seed cotton, and the cotton oil extrac tion ratio, which ranges between 10 percent and 16 percent of the weight of cot tonseed depending on the method of crushing (Blasi and Drouillard 2002). Figure 1 shows the current composition of the US cotton industry. Figure 1 Weight Composition of Cotton and Potential ByProducts Cotton Lint (35%) Seed (65%) PROCESSED NOT PROCESSED (crushing facilities) (animal feed) MEAL (45%) OIL (16%) HULL (27%) OTHER (12%) [used for animal feed, [mostly human [used for animal [linters , i.e., low competes with other consumption, competes feed, often mixed quality lint (8%) and meals] with edible oils] with meal] waste (4%)] Source: National Cottonseed Products Association and author's estimates 1 On average one metric ton of seed cotton in the US generates about 350 kgs of lint and 104 kgs of 3 cotton oil. Of the remaining 546 kgs, meal accounts for more than half (293 kgs); hulls and linters account for 176 kgs and 52 kgs, respectively; and 26 kgs is waste. However, in WCA, one ton of cotton generates about 420 kgs of lint (because of the higher ginning outturn ratio) and 70 kgs of cotton oil (when less technologically advanced methods of oil extraction are used). Page 3 Cottonseed has a low oil content relative to most other oilseeds, partly be cause cotton research efforts have sought to maximize the quality and quantity of lint, rather than the oil or meal content. For example, ICAC (2000, p. 7) noted that "Unfortunately, not even a single variety on the basis of higher oil percentage has been released in any country in the world." But there are signs that a shift in research priorities may be taking place. For example, Auld and others (2006) have argued that two simple genetic modifications in cotton varieties could en hance the oil proportion by as much as 20 percent, implying that the oil content could increase from the current share of 16 percent to 19 percent. And Beltrão and Barbosa (2007) report that current research in Brazil is likely to generate new cotton varieties that will double the oil content without reducing the quality and quantity characteristics of lint. The value of cotton byproducts varies according to prices and market conditions, including the structure of the local cotton industry, the location of cotton production, demand by the dairy industry, the availability of competing oilseeds and meals, and the ginning and seedcrushing technologies employed. In the US, for example, with its large cattle industry, more than half of cottonseed is used unprocessed as animal feed. In most SSA countries, cottonseed is now typically processed with most of the resulting oil going for human consumption and the meal for animal feed. But until the mid1980s, cottonseed in most West and Central African (WCA) cotton producing countries went to waste with no value attached. Looking ahead, the outlook for the value of cotton byproducts relative to lint depends heavily on expected trends in relative prices. Figure 2 depicts the contribution of cottonseed to the total value of cotton, based on three assump tions about the ginning ratio (40 percent, 35 percent, and 30 percent) and two scenarios of lint and seed prices. It shows that at the average prices of 19702006, the contribution of seed ranged from 9.6 percent to 14.2 percent, corresponding to the high and low ginning ratios. But at the lint and meal prices that prevailed in 200709, during the commodity price boom, the contribution of seed was larg er, increasing to 15.7 percent and 22.4 percent, respectively (for details see data reported in Appendix E). During 200008 the ratio of the lint to the cotton oil price was well below the longterm average, and in 2007 and 2008 it reached record low values (thus, in these latest two years, cotton oil was at its most expensive, compared to lint) (Figure 3). Page 4 Figure 2 Seed's Contribution to the Total Value of Cotton (percent) 25% 30% 35% 40% 20% 15% 10% 5% 19702006 200709 Note: The percentages, 30%, 35%, 40% refer to ginning ratios and refer to differences in cotton varieties and ginning efficiency rather than cleanliness of harvested cotton Source: Author's calculations. 2 Figure 3 Lint to Oil Price Ratio (2000=100) 140 Average 120 100 80 60 40 20 1960 1966 1972 1978 1984 1990 1996 2002 2008 Source: World Bank 3 Page 5 The chief reason behind the large increase in the relative value of cotton byproducts during the commodity boom is that while the prices of byproducts went up, like those of most other agricultural commodities, the price of cotton (lint) did not increase much. The divergence between the price of cotton and those of other agricultural commodities began in the early 2000s (Figure 4). Be tween 2000 and 2008, the World Bank index of nominal food prices increased by 147 percent but cotton prices increased by only 21 percent (from US$1.30 to 1.57 per kg). Figure 4 Agricultural Price Index and Cotton Prices (Real, MUVdeflated, 1980=100) 160 140 120 100 Agriculture 80 60 40 20 1960 1966 1972 1978 1984 1990 1996 2002 2008 Source: World Bank 4 There are three reasons why cotton prices did not join the commodity boom. First, cotton receives more subsidies than other commodities, encouraging more production than would have taken place without the support. Second, the boom in food prices was partly aided by growth in demand for biofuel produc tion. The direct impact of biofuel demand is felt only by maize, sugarcane, and some edible oils, but the indirect impact is felt by almost all food crops, because of the strong substitutability both on the input side (mainly shifting of land from one crop to another) and on the output side (especially in edible oils, some of which are highly substitutable). In cotton, however, the indirect impact is very limited: cotton is not a substitute for any other commodity on the output side, Page 6 and its input substitutability is limited. The third, and perhaps most important, reason is the rapid and massive expansion of cotton production that took place in China and India, as a direct consequence of these countries' adoption of biotech cotton. During the fiveyear period 200207, China increased its cotton output by 56 percent (from 5.2 to 8.1 million tons) while India increased its output by 126 percent (from 2.3 to 5.2 million tons). Today, China and India account for more than half of global cotton production, mostly from biotech varieties. Econometric evidence confirms that lint prices have diverged from cotton byproduct prices, as proxied by the agricultural commodity price index. The degree of comovement between cotton prices and the agricultural commodity price index was analyzed with the following ordinary least squares regression, using annual data from 1960 to 2009: log(PtCOTTON) = + 1log(PtAG_INDEX) + 2log(MUVt) + 3t + t. [1] PtCOTTON and PtAG_INDEX denote the price of cotton and the agricultural commodity price index in year t (expressed in nominal dollar terms), MUVt denotes the def lator, t is the time trend, and t denotes the error term; , 1, 2, and 3 are para meters to be estimated. The model is expressed in logarithms to facilitate inter preting the estimated parameters as elasticities. The estimate of 1 is expected to be positive while 2 and 3 can take any sign depending on the relative impact of technological progress and inflation on the particular commodity sector. Notice that while the price of cotton is part of the agricultural commodity price index, its contribution is only 3.9 percent, thus not driving the results.4 The parameter estimates of [1] are reported in the top row of Table 1. Spe cifically, the estimate of 1 is 0.61 and highly significant with an adjustedR2 of 0.91 and an ADF statistic of 6.03, implying a very strong longrun comovement between cotton prices and the agricultural commodity price index. The inflation coefficient, 2, is highly significant as well. The time trend is highly significant and negative, with a relatively large estimate, implying that cotton prices have been declining faster than the agricultural price index, even after accounting for inflation. 4 An alternative specification would have been to deflate both PtCOTTON and PtAG_INDEX by MUVt, which would effectively restrict the sum of the price and inflation coefficients to unity (i.e. 1 + 2 = 1). The advantage of having the inflation as a separate explanatory variable is that the homo geneity restriction is relaxed and a direct estimate of the effect of inflation can be obtained (Hou thakker 1975). Page 7 Table 1 Comovement between Cotton and Agricultural Commodity Prices D1 D2 1 11D1 12D2 2 100*3 AdjR2 ADF 0.23 0.61*** 0.67*** 2.29*** 0.91 6.03*** (0.87) (6.68) (6.33) (7.45) 0.15 2.80*** 0.85*** 0.23 0.33** 1.30*** 0.93 7.16*** (0.57) (2.73) (7.53) (1.30) (2.16) (2.69) Source: Author's estimates based on World Bank price data. Notes: , 1, 2, and 3 are the estimates of the constant term and the coefficients of the price, in flation, and time trend, respectively. The numbers in parentheses denote absolute tvalues while asterisks denote parameter estimates significant at 10 percent (*), 5 percent (**), and 1 percent (***) levels, respectively. ADF is the Augmented DickeyFuller (Dickey and Fuller 1979) statistic for unit root and corresponds to the MacKinnon onesided pvalue. The lag length of the correspond ing ADF equations was determined by minimizing the Schwarzloss function. To examine whether a structural change has taken place in the relation ship between the two price series, relationship [1] was modified by adding two dummy variables (D1 and D2), taking the value of one during 19602003 and zero otherwise. D1 and D2 are applied to both the constant term and 1. Hence, the eq uation to be estimated becomes: log(PtCOTTON) = D1 + D2 + 11D1*log(PtAG_INDEX) + 12D2*log(PtAG_INDEX) + 2log(MUVt) + 3t + t, [2] If a structural break in the relationship between the price of lint and the agricultural commodity price index exists, then 11 would differ from 12. Results reported in the second row of Table 1 show that, indeed, the parameter estimate of 11 is 0.85 and highly significant [tvalue = 7.53] while that of 12 is 0.23 [tvalue = 1.30], not significant even at the 10 percent level. Furthermore, the other statistics show that model [2] outperforms model [1] (e.g., adjustedR2 = 0.93 and ADF = 7.16).Thus, the econometric evidence shows that while there was a strong rela tionship between the price of cotton lint and the agricultural commodity price index up to 2003 that relationship effectively disappeared during 200409. For the purposes of the current study, then, a key question is whether the two price series are likely to reconverge--that is, whether cotton byproducts will continue to be as profitable, vis a vis cotton lint, as they have recently be come. Part of the answer will be dictated by the factors depressing the price of cotton, discussed above, none of which appears to be reversible in the short term. It also depends on the prospects for cotton oil, and in particular, for cotton oil as a feedstock for biofuels. These are discussed in what follows. Page 8 2. Cotton Oil in the Context of the Global Edible Oil Market Ranked by oil content, cotton seeds are the lowest among key oilseeds after corn (Figure 5). Ranked by its yield per unit of land, cotton oil averages 325 liters per hectare, twice as much as corn. Other edible oils have much higher yields, how ever, the highest being palm with 6,000 liters of oil per hectare (Figure 6). The global production of cotton oil is small compared to that of other in ternationally traded edible oils, accounting for only 3.3 percent of total fats and oils supplies, and the global cotton oil market is very thin, with only 3 percent of global production traded internationally. During 200608, global production of the major 17 fats and oils averaged 147 million tons. Two vegetable oils, palm and soybean, dominated the market, accounting for 50 percent and 70 percent of global output and exports, respectively (Table 2). Cotton oil production averaged a little more than 5 million tons during the same period. Cotton oil is traded at about the same price level per ton as sunflower and corn oil, and a little higher than palm and soybean oil (Figure 7). Because cotton oil is a close substitute for other edible oils, its price depends on, and reflects, the availability and trade policies of such oils. Cotton oil prices move in concert with the edible oil index (Figure 8).5 Between the mid1980s and the early 2000s, they averaged US$600 per ton in nominal terms. They increased by 270 percent be tween January 2007 and July 2008, but declined sharply afterwards. During the first quarter of 2010 they averaged about US$700 per ton. Historically, prices of cotton and other edible oils have followed the same path as those of other agricultural commodities. They were fairly stable before the first oil shock, doubled during the oil shock, and then declined until the mid 1980s. They entered a period of relative stability until 2005, then increased as part of the overall commodity boom. The World Bank's index of nominal prices of ed ible oils reached 340 (2000=100) in July 2008 but then declined sharply to about 225 during the first quarter of 2010. Even so, it remains more than double the 2000 level. 5 Substitutability among edible oils comes from the consumption side, as various oils can be used for the same purpose. Substitutability among grains, by contrast, comes mostly from the produc tion side, where inputs, especially land, can be shifted from one crop to another. Page 9 Figure 5 Average Oil Content of Key Oilseeds (percent of seed dry weight) Corn Cotton Soybeans Rapeseed Castor Palmkernel Groundnuts Sunflower Coconut 0 10 20 30 40 50 60 70 Source: Author's calculations from various sources 5 Figure 6 Average Oil Yield of Key Oilseeds (liters of oil per hectare) Corn Cotton Soybean Sunflower Groundnut Rapeseed Olive Jatropha Coconut Palm 0 1,000 2,000 3,000 4,000 5,000 6,000 Source: Author's calculations from various sources 6 Page 10 Figure 7 Prices of Major Edible Oils (nominal, US$/ton, 2005/07 average) Tallow Lard Palm Palmkernel Soybean Coconut Corn Sunflower Cotton Linseed Rapeseed Fish Castor Groundnut 200 400 600 800 1,000 1,200 Source: Oil World, Hamburg 7 Figure 8 Cotton Oil Prices and the Edible Oil Price Index (Real, MUVdeflated, 2000=100) 450 400 350 300 Edible oil index 250 200 150 100 50 1960 1966 1972 1978 1984 1990 1996 2002 2008 Source: World Bank 8 Page 11 Table 2 Global Balance of the 17 Major Fats and Oils, 2005/062007/08 Average Production Exports Stocks Traded Production (000 tons) (000 tons) (000 tons) Share Share (percent) (percent) Palm 38,657 30,232 6,177 78.2 25.3 Soybean 36,371 10,727 3,926 29.5 23.8 Rapeseed 18,676 2,085 1,256 11.2 12.2 Sunflower 10,764 4,151 990 38.6 7.0 Tallow 8,515 2,244 502 26.4 5.6 Lard 7,672 120 438 1.6 5.0 Butter 6,882 728 639 10.6 4.5 Cotton 5,003 154 364 3.1 3.3 Palm kernel 4,516 2,523 562 55.9 3.0 Groundnut 4,360 190 306 4.4 2.9 Coconut 3,141 2,030 375 64.7 2.1 Olive 2,851 741 937 26.0 1.9 Corn 2,311 795 227 34.4 1.5 Fish 1,043 727 157 69.7 0.7 Sesame 834 34 46 4.1 0.5 Linseed 682 108 78 15.9 0.4 Castor 543 335 59 61.7 0.4 TOTAL 152,821 57,928 17,039 37.9 100.0 Source: Oil World, Hamburg. Notes: Traded share is global trade as a percentage of global production. Production share is the contribution of each type of oil to global production of all 17 fats and oils. Lard (from pig fat), tallow (from beef fat), and butter (a dairy product) are classified as fats and the remainder as oils. An econometric model similar to [1] was used to analyze the degree of co movement between the prices of various edible oils and grains. Specifically, log(Pti) = + 1log(Ptj) + 2log(MUVt) + 3t + t. [3] Ptj and Ptj denote the prices of interest while the other parameters and variable are defined as before. The results, reported in Table 3, show a high degree of co movement between the prices of three vegetable oils (cotton, soy, and palm) and the edible oil index (upper panel) as well as among the prices of the three oils (second panel). For example, all crossprice effects are significant at the one per cent level and very close to unity while the adjustedR2s averaged 0.93. These es timates indicate a high degree of substitutability among the various edible oils, Page 12 reconfirming the strength of the relationship between the cotton oil price and the edible oil price index. Table 3 Comovement among Commodity Prices 1 2 100*3 AdjR2 ADF I. COTTON OILOIL INDEX Cotton Oil ­ Oil Index 1.68 *** 0.88*** 0.11 0.01 0.90 5.25*** (5.06) (10.36) (1.05) (0.11) Soy Oil ­ Oil Index 1.30*** 1.14*** 0.17** 0.40 0.95 7.54*** (5.60) (19.21) (2.24) (1.61) Palm oil ­ Oil Index 0.99*** 1.17*** 0.12 0.11 0.94 4.83*** (3.82) (17.85) (1.51) (0.40) II. OILS Cotton Oil ­ Palm Oil 1.24*** 0.66*** 0.27** 0.03 0.88 4.67*** (3.03) (8.64) (2.37) (0.07) Cotton oil ­ Soy Oil 0.64** 0.78** 0.23** 0.35 0.93 5.40*** (2.11) (14.04) (2.97) (1.23) Soy Oil ­ Palm Oil 0.59** 0.90*** 0.01 0.46* 0.94 4.30*** (2.12) (17.35) (0.07) (1.64) III. GRAINS ­ PALM OIL Maize ­ Palm Oil 0.03 0.54*** 0.37*** 0.04 0.87 5.58*** (0.08) (6.93) (3.23) (0.10) Wheat ­ Palm Oil 0.04 0.55*** 0.29*** 0.04 0.87 4.87*** (0.12) (8.32) (2.99) (0.41) Rice ­ Palm Oil 0.37 0.66*** 0.31* 0.63 0.69 5.28*** (0.60) (5.67) (1.83) (1.01) IV. GRAINS MaizeWheat 0.31 0.86*** 0.04 0.26 0.95 4.82*** (1.54) (14.91) (0.50) (1.10) Rice Maize 0.53 1.08*** 0.04 0.52 0.77 5.80*** (1.06) (7.69) (0.23) (0.97) Rice Wheat 0.62 1.07*** 0.20 0.74 0.80 4.25*** (1.34) (8.52) (0.14) (1.45) Source: Author's estimates based on World Bank price data. Notes: , 1, 2, and 3 are the estimates of the constant term and the coefficients of the price, in flation, and time trend, respectively as specified in model [3]. For other notes see Table 1. Page 13 Such substitutability is not restricted to the edible oil market; it extends to other commodity markets as well. For example, the comovement between oils and grains is high as well as evidenced by the high adjustedR2s and ADF statis tics for the regressions between wheat and maize with palm oil (panel III). Simi larly, the comovement among grains is high (panel IV). Such findings imply that policy actions or other factors affecting one vegetable oil or grain market will eventually affect most other grain and oilseed markets. 3. Issues in the Use of Cotton Oil for Biofuels Given the growth of world demand for biofuels--which now consume an esti mated 9 percent of global vegetable oil production--vegetable oils including cot ton oil have often been mentioned as a possible new source of income for com modity producers in SubSaharan Africa. Opinions differ about the virtue of using food commodities to produce biofuels: the practice is viewed as a problem by some (consumers of food com modities who pay higher prices) and as an opportunity by others (commodity producers who receive higher prices, as well as the biofuel industry). Of the three key sources of biofuels, sugarcanebased ethanol, produced mainly in Brazil, is the most profitable. Maizebased ethanol, produced mainly in the US, ranks second. Edible oilbased biodiesel, produced mainly in the EU, ranks third. The consensus is that, with the exception of sugar canebased ethanol, production of biofuels would not have been profitable were it not for the various policies and mandates in force. Demand for biofuels can affect the cotton oil market through three chan nels. First, it may directly increase demand for cotton oil as a biofuel feedstock. Second, it may increase demand for competing oils, and so indirectly raise the demand for cotton oil. Third, it may increase the demand for other commodities and thereby, through the reallocation of resources, affect the cotton oil market as well. To analyze the effect of biofuels on the cotton oil market thus requires a look at each of these three channels. First, the country studies summarized later in this paper find that biofuel production in SSA countries, based on either cotton oil or other sources, is un likely to be profitable at currently foreseeable energy prices, at least within the next decade. Most of the African cottonoil producers surveyed believed that they could not profitably produce biodiesel from cotton oil unless it were heavily subsidized. Mitchell (2010) concluded that even sugarcanebased ethanol may not be profitable in SSA without incentives (although such technology is profita Page 14 ble in Brazil) because of high production costs due to poor infrastructure, high operating costs, and the requirement to irrigate. Second, the strong comovement among edible oil prices, demonstrated in Section 2 above, implies that policy changes in one edible oil market will affect the other markets as well. For example, policies that mandate the use of rapeseed oil for the production of biofuels in the EU, or soybean oil use in the US, will not only drive up the demand and prices of these two oils (annual crops produced mainly in Europe and North and South America, respectively) but will also in duce proportional changes in the demand and price of palm oil (a perennial crop produced mainly in Southeast Asia) and thus an expansion of palm oil produc tion. The implication is that, whether biofuel mandates are applied to one or another edible oil market, the effect will eventually be diffused among all edible oil markets, including that for cotton oil.6 Third, the effect of biofuel production on commodity markets more broad ly, and especially on commodity prices, is complex and the estimates from vari ous models differ considerably (Box 1). The issue is complicated by the fact that energy prices affect agricultural commodity prices through the cost structure, since agriculture is an energy intensive activity. Thus, the recent increase in energy prices had a large impact on agricultural commodity prices even irrespective of biofuels. Table 4 reports transmission elasticities from energy prices to the edible oil index, using the model [3]. When the sample covers the 19602005 period (first row), before the recent price boom, the estimated transmission elasticity is 0.23, implying that a 10 percent increase in energy prices is associated with a 2.3 percent increase in the prices of edible oils. But when the sample period is extended to 2009 to include the recent price boom (second row), the elasticity rises to 0.31. It is tempting to attribute such an increase to biofuels, since edible oils are one of the three key biofuel feedstocks (the other two being maize and sugarcane), but similar in creases in fact took place in all commodity subindices, including metals, fertiliz ers, and precious metals. For example, as shown in Figure 9, transmission elastic ity estimates increased for all six main nonenergy commodity price indices. (Baffes (2007 and 2010) discusses this issue in detail along with the evidence.) 6 Hence, policies favoring biofuel production in the name of environmental benefits may in fact lead to less desirable outcomes. From a global perspective, the environmental benefits from switching from fossil fuel use to, say, rapeseedbased biodiesel in Europe or soybean oilbased biodiesel in the US may be less than the environmental costs of expanding palm oil production in East Asia. Page 15 Box 1: The Effect of Biofuels on Commodity Prices--Summarizing the Literature The contribution of biofuels to the recent price boom, and especially the price spike of 2007/08, has been hotly debated, with the estimates ranging from a high to a negligible impact. Mitchell (2009) argues that biofuel production from grains and oilseeds in the US and the EU was the most important factor behind the food price increase between 2002 and 2008 -- accounting, perhaps, for as much as two thirds of the price increase. Rosegrant (2008) concludes that biofuel growth during 200007 accounted for 30 percent of the food price increases seen in that period, with the contribution varying from 39 percent for maize to 21 percent for rice. FAO (2008) concludes that if biofuel production remains at 2007 levels in 2018 (as opposed to doubling), grain prices would be 12 percent lower, wheat prices 7 percent lower, and vegetable oil prices 15 percent lower. Similarly, OECD (2008) concludes that if biofuel support policies were abolished, vegetable oil price would be 16 percent lower, while prices of feed grains and wheat would be 7 and 5 percent lower, respectively. Banse and others (2008) compared the impact of the EU's current mandate to a nomandate scenario and a mandate under which the US, Japan, and Brazil adopt tar gets for biofuel consumption. They estimate that by 2020, under the nomandate scena rio, cereal and oilseed prices will have decreased by 12 and 7 percent, respectively. By contrast, under the "global" scenario, oilseed prices will have risen by 19 percent and cereal prices by about 5 percent. The European Commission's own assessment of the longterm (2020) impacts of the 10 percent target for biofuels predicts that prices of ce reals will increase by 3 to 6 percent by 2020, with larger impacts on prices of oilseeds, particularly sunflower (+15 percent). Taheripour and others (2008), simulating the bio fuel economy during 200106, estimate the impact of biofuels on coarse grain prices in the US, EU, and Brazil at 14 percent, 16 percent, and 9.6 percent, respectively. A joint US Department of Agriculture and Department of Energy assessment (USDA/USDE 2008) concludes that the recent increase in maize and soybean prices ap pears to have little to do with the runup in prices of wheat and rice. It finds that if the amounts of corn used for ethanol and soybean oil used for biodiesel in the US had re mained unchanged at their 2005/06 levels, prices in 2007/2008 would have been 15 per cent lower for maize, 17.5 percent for soybean, and 13 percent for soybean oil. Lastly, Gilbert (2009) finds little direct evidence that demand for grains and oilseeds as biofuel feedstocks was a key cause of the price spike. At some level, energy prices may provide a floor to agricultural prices. To explore this question one must consider not only the prices of the respective commodities (energy and feedstock for biofuels), but also numerous other ele ments including subsidies, mandates, trade restrictions, and sunk costs. Page 16 Table 4 The Energy/Edible Oil Price Link 1 2 100*3 AdjR2 ADF 19602005 2.75** 0.23*** 0.34* 1.37** 0.77 2.86* (4.83) (3.55) (1.66) (2.39) 19602009 3.52** 0.31*** 0.05 0.57 0.77 2.79* (6.92) (5.00) (0.27) (1.14) Source: Author's estimates based on World Bank price data. Notes: The model structure is the same as in equation (3), where Pti denotes the nonenergy price index and Ptj denotes the energy price index. For other notes see Tables 1 and 3. Figure 9 Price Transmission Elasticities from Energy and NonEnergy Commodities (estimates based on model (3)) 0.6 19602005 19602009 0.5 0.4 0.3 0.2 0.1 0.0 Nonenergy Food Raw Materials Metals Precious Fertilizers Metals Source: Author's estimates based on World Bank price data 9 4. The Cotton ByProducts Industry in SubSaharan Africa To gain a better understanding of the structure and issues of the cotton by products industries in SSA, the experience of four countries was analyzed: Uganda and Tanzania in Eastern and Southern Africa (ESA); and Benin and Bur Page 17 kina Faso in West and Central Africa (WCA).7 Details on the historical and recent developments of the cotton byproducts industries of the four countries can be found in Appendices A through D. The analysis was supplemented by a brief discussion of the byproducts industry in the United States (Appendix E) and a short survey (see Appendix F). Table 5 provides a summary comparison of the industries in the African countries. Uganda is a small cotton producer, averaging about 25,000 tons of lint annually (very little compared to the global production of 25 million tons of lint), almost all of which is exported. It has five privatelyowned cottonseed processing companies, of which four are independent and one is associated with a ginning operation. Uganda has 50 ginneries each processing small volumes-- which explains why most of the crushing facilities are independent entities. While Uganda's cotton production is spread around the country, all the seed processing facilities are in or close to Kampala, the key oil consuming area. Tanzania's cotton sector is much larger than Uganda's, producing about 100,000 tons of lint annually. It consists of 20 privately owned cottonseed processing companies, all but one of which are attached to ginneries. (Tanzania has 73 ginneries.) Almost all Tanzania's cotton production and all its processing facilities are located in the Mwanza region. Most of the country's cotton oil out put is consumed in this region, though limited quantities go to Dar es Salaam. All cottonseed in Uganda and Tanzania is processed, with most of the oil going for human consumption and the meal for animal use. Neither Uganda nor Tanzania exports cotton oil, and both are net importers of edible oil. In both countries there are established markets in the sense that cotton seeds, oil, and meal are all tradable commodities. The survey evidence shows that seed and oil prices move in concert within each country, indicating that the marketing struc ture functions well. Cotton oil prices spiked during the 2008 boom and declined in 2009 following the decline of world edible oil prices. Cotton is a key cash crop in both Benin and Burkina Faso. Benin's cotton lint production averaged 114,000 tons during the past three seasons. Burkina Fa so is SSA's largest cotton producer; its output of cotton lint averaged more than 200,000 tons during the past three seasons and reached almost 300,000 tons in 2006/07. Both countries have undertaken many policy reforms in the cotton sec tor during the past decade, including privatizing the former parastatal that han dled all ginning and marketing. In both countries, cotton prices are set in a pan seasonal and panterritorial manner. 7 The cotton sectors of these four SSA countries are also analyzed in the multicountry cotton study (Tschirley, Poulton, and Labaste 2009). Page 18 Table 5 Characteristics of Cottonseed Processing Industries in Four SSA Countries Tanzania and Uganda Benin and Burkina Faso Organization of Competitive structure on seed and Oligopoly/oligopsony in Benin. Oli the industry the products. gopoly and mix of oligopsony and competitive structure in Burkina Faso. Ownership All privately owned. Some are in Privately owned but cotton compa dependent entities while others are nies have stake in the large compa associated with ginneries. nies. All are independent entities. Size of Daily processing capacity averages Average daily capacity of large opera operations 80 tons and labor force averages 100 tions is 500 tons with labor force of (both permanent and temporary 365. Burkina Faso's smaller opera workers). tions have daily capacity of 35 tons and labor force of 45. Input pricing Seed prices are market determined Negotiated prices between the gin and depend mostly on the size of ning and the seed processing compa the cotton crop. nies. Output pricing Oil and meal prices are market de Oil and meal prices are market de termined, with considerable intra termined, with considerable intra year and crosscompany variability. year and crosscompany variability. Interest rates for Relatively high in Uganda (average Average 13 percent for Burkina Faso's operating capital 21 percent), much lower in Tanza smaller operations; less than 10 per nia (average 13 percent). cent for the large ones in both Benin and Burkina Faso. Electricity Frequent disruptions, cited as one Only minor complaints. Not an issue supply of the industry's key problems. for the industry. Trade policies A 25 percent import duty on refined Benin imposed a shortlived export oils is imposed while crude oils en ban in 2001. Currently, it applies 45 ter dutyfree. percent import duty on palm oil and 18 percent VAT on domestic sales. Views on trade While most (not all) owners or Managers and owners favored trade policies managers favored trade policy re restriction including import duties on strictions on imported edible oils, edible oils and restrictions on seed they do not feel very strongly about exports. them. Views on Expressed strong doubts about the Most owners and managers expressed biofuels use of cotton oil for biofuel unless doubts regarding cotton oil's poten subsidies are involved. tial for biofuel. Burkina Faso's largest company runs a pilot biofuel facility. Information Information was obtained easily Difficult to convince managers to availability from managers and was, for the share information. Even when infor most part, precise. mation was shared, it was not precise. Source: Appendices A, B, C, and D. Page 19 The cotton byproduct industries in Benin and Burkina Faso share a num ber of structural similarities and, to a great extent, reflect the structure and histo ry of the cotton sector in WCA more broadly. Seed processing in WCA began on ly in the late 1980s. In fact, cottonseed was not considered a valuable byproduct up until the early 1980s. On some occasions seeds would be exported to Europe for processing or to be used by the dairy industry; on other occasions they would go to waste (in fact, ginneries invested in transport equipment to remove cotton seed from their premises.) Burkina Faso has two types of cottonoil producing company: three dominant seed processing companies (only one of which func tions at present) and a large number of small firms scattered throughout the country, with a high concentration in Bobo Dioulasso, the country's largest cot ton producing area. The official count of the smaller companies is 59; because the industry is still at its early stages, there is considerable entry and exit, and not all operations are registered with the respective associations. Most of Burkina's small cottonseed processing companies began operation during the mid2000s when the cotton sector experienced a large expansion. Benin's industry is highly concentrated with only three seed processing companies, two of which are func tioning currently (unconfirmed reports indicate that only one company processed cotton seeds during 2009/10). In both countries the seed processing companies buy cottonseed at a fixed price, but meal and oil prices move accord ing to demand and supply conditions, as in Uganda and Tanzania. The cottonseed processing facilities of Uganda, Tanzania, and Burkina Fa so`s small companies use traditional expeller pressing technology rather than the more technologically advanced solventbased approach; they currently extract about 60 percent of the oil content of the cottonseed they process, instead of the 9596 percent made possible by the solventbased technology. By contrast, Be nin's companies and Burkina Faso's large company use solventbased technolo gy. The survey respondents in Uganda and Tanzania felt that on a costbenefit basis, introducing the more advanced technology is unlikely to increase their profitability, because of: (i) the high cost of installing and operating the advanced machinery, (ii) not having enough cottonseed to utilize the machinery at full ca pacity, and (iii) the higher labor intensity of the older technology than the new-- an important advantage in countries with relatively low wage rates. Two of Bur kina Faso's large processing companies that used the solventbased technology have gone out of business. Consumers are not judged willing to pay the higher prices for the better oil that the advanced machinery would produce. In all four SSA countries, the possibility of producing biodiesel either from cotton oil or other feedstocks has been frequently discussed at the government level but not much has taken place on the ground. Only one facility in Burkina Page 20 Faso has experimented with producing biodiesel from cotton oil. The operators and managers of the cottonseed processing companies in the four countries take the view that producing biodiesel from cotton oil would not be profitable unless it were heavily subsidized. Further, because their companies face enough domes tic demand for edible oils (all countries are net importers of edible oils) they see no need to diversify their operations. In Uganda and Tanzania, all imports of refined oils are subject to a 25 per cent duty, while edible oils imported in crude form enter duty free. Most indus try representatives argued that while most oil in fact enters in refined form, it is declared as crude to escape the import duty. Benin introduced a shortlived seed export ban in 2001 that was never strictly enforced and a 45 percent import duty on palm oil (the key competing product to cotton oil). Yet in Benin, as in ESA, traders report that in most cases import duties are not collected on edible oils. Some of the owners and managers who were interviewed favored stricter protec tion of Benin's edible oil industry while others did not. The views in favor of im port restrictions were stronger in WCA than in ESA. Several other issues were identified during the interviews. Overcapacity was cited as a problem in all four countries. Electricity disruptions were cited as a key constraint in Tanzania and Uganda but not in Benin or Burkina Faso. High interest rates for operating capital were an issue in Uganda only. Attracting labor (either skilled or unskilled), was not an issue. While the surveys undertaken in the four SSA countries allow us to have a deeper understanding of the structure of the cotton byproducts industries, they do not offer enough quantitative detail for a full statistical assessment of the con tribution of cotton byproducts to the value of cotton. Furthermore, the data are not directly comparable for a number of reasons. For example, as mentioned ear lier, the ginning costs in the US are paid by the cotton growers who retain the ownership of both lint and byproducts. On the contrary, SSA cotton growers sell their seed cotton to ginners or traders, hence the ginning costs are accounted by ginneries. Furthermore, while there are reliable estimates of ginning costs in WCA, no such estimates exist for ESA. Nevertheless, the following conclusions can be drawn: In West Africa unlike in the US, the value of cotton byproducts is small com pared to the costs of ginning. In 1999/2006, the revenue from cottonseed cov ered only 23 percent of ginning costs in Benin and 28 percent in Burkina Faso. For comparison, consider that the value of byproducts in the US is expected to cover the ginning costs. This implies that not only cottonseed is valued much less in WCA than in the US but also that ginning costs are much higher in WCA than in the US. Page 21 In both Benin and Burkina Faso, the contribution of cottonseed to the total value of cotton doubled during 2007/08 compared to earlier years. This is consistent with the large increase that took place in the US, from 11.6 percent in 19702006 to 18.7 percent in 1999/2006, as well as the trend that took place in the global markets. In Burkina Faso after 2003/04, consistent with the entry of smaller companies and hence more intense competition, the contribution of cottonseed to the value of cotton increased. No such increase took place in Benin. Since 2007/08, the contribution of cottonseed to the value of cotton has been higher in Bur kina Faso, at 19 percent, than in Benin, at 13 percent. Furthermore, the contri bution doubled after 2007/08 in both countries, consistent with the increase in global markets. There have been considerable differences in the input/output price ratios among the countries examined. During 2007/08 and 2008/09, the ratio of the price of cottonseed to the price of seed cotton received by growers (in lint equivalent, properly adjusted by the respective ginning outturn ratios) was highest in Uganda (39 percent), followed by Tanzania (28 percent), Burkina Faso (25 percent), and Benin (17 percent). The ratio in the US is 19 percent. Al though such ratios may not be directly comparable because the prices have been subjected to various distortions (at least in WCA and the US) and be cause ginning costs have been excluded from the calculation in the US, one may conclude that byproducts are valued more in lintprice terms in ESA than in WCA and the US. 5. Conclusions and Policy Implications This paper has examined the cotton byproducts industries of four SSA cotton producing countries--Benin, Burkina Faso, Tanzania, and Uganda--in the con text of global market trends and the apparent divergence between lint prices and the prices of other agricultural commodities, including cotton byproducts. The paper reaches several methodological and policyrelated conclusions. Most edible oil markets, including that for cotton oil, are highly integrated with each other. Although the market for cotton oil is very thin, with only 3.3 percent of global production traded internationally, the fact the most edible oils are close substitutes for each other implies that examination of the market for any one edible oil should take into consideration all the (relevant) edible oils. Thus, policies targeting one edible oil market will inevitably affect all edible oil markets. Page 22 Industry representatives in the four SSA countries all said that they face a large demand for cotton oil for human consumption and see no potential for us ing cotton oil for biofuel production. Furthermore, biofuel production is unlikely to be profitable in SSA under current energy and nonenergy prices unless man dates and subsidies are involved. Therefore, the only channel through which bio fuels can increase profitability to African cotton growers is through their effect on commodity prices. Efforts to deepen the reforms in the cotton industries of WCA should con sider cotton byproducts as well as lint. To the extent that pricesetting mechan isms are a result of negotiations between cotton producers and ginning compa nies, the value of byproducts should play a key role in the calculation of price. Moreover, any policy actions regarding the future of the industry should cover the byproducts industry as well. The size and nature of the industries in the countries studied here show that large operations using advanced technology, while efficient from a technol ogical perspective, tend not to be economically profitable in the SSA context. Smaller, less technologically efficient, laborintensive operations perhaps consti tute the future of the industry. Wherever applicable, policies should facilitate the transition towards this model. It should also be ensured that small processors of oil for human consumption adhere to certain health standards. Because cotton oil competes with imported edible oils, there have been calls for imposing trade restrictions on such oils in African countries. Policy measures including prohibiting exports and imposing high import tariffs are common, but they go against the rules agreed within regional trade arrange ments and may conflict with the interest of consumers. It is imperative that poli cymakers reconsider these measures. Current trade policies that favor the import of crude oils over refined oils should also be reconsidered: while the idea behind such policies is to encourage domestic value addition (and, perhaps, generation of tax revenue), industry representatives believe that almost all the imported oil recorded as crude actually enters in refined form. Therefore, import duties should be uniform and small enough to give the importers an incentive to pay rather than avoid the tax. Because lint contributes most of the value of cotton, research on new cot ton varieties has traditionally focused on maximizing the quantity and quality of lint. Now that cotton byproducts are becoming increasingly important in value terms, research into new cotton varieties should focus on the properties and con tribution of cottonseed as well. Researchers see potential for substantially raising the oil content of cotton without jeopardizing the existing quantity and quality of lint. Page 23 References Ahohounkpanzon, M., and M. Y. Zakariallou (2010). "Étude sur les Mécanismes de Fixation du Prix du Coton Graine et la Prise en Compte des Coproduits du Coton au Bénin." Report prepared for the Department of Agricultural Eco nomics, Michigan State University, and the West African Cotton Improvement Program. Anderson, Kym, and Ernesto Valenzuela (2007). "The World Trade Organiza tion's Doha Cotton Initiative: A Tale of Two Issues," The World Economy 30: 12811304. Argüello, Priscilla B. (2008). "Cottonseed: Derived Demand and Price Analysis." Master's Thesis, Department of Agricultural and Applied Economics, Texas Tech University. Auld, Dick, Efrem Bechere, Jim Davis, Lindy Seip, and Jack Brown (2006). "Lint, Cottonseed Oil, or Biodiesel?--Breeding Cotton for the Next Decade." Paper presented at the ASACSSASSSA International Meetings. November 1216. Baffes, John (2010). "More on the Energy/NonEnergy Commodity Price Link," Applied Economics Letters, DOI: 10.1080/13504850903120683. ________ (2009). "The `Full' Potential of Uganda's Cotton Industry," Development Policy Review 27: 6785. ________ (2007). "Cotton Developments in West Africa: Domestic and Trade Policy Issues and the WTO." In J. Morrison and A. Sarris, eds., WTO Rules for Agricul ture Compatible with Development. Rome: Food and Agriculture Organization of the United Nations. ________ (2007). "Oil Spills on Other Commodities," Resources Policy 32: 126134. ________ (2005). "The Cotton `Problem'," World Bank Research Observer 20: 109 144. ________ (2004). "Tanzania's Cotton Sector: Reforms, Constraints, and Chal lenges," Development Policy Review 22: 7596. Beltrão, Napoleao E. M., and Sebastiao Barbosa (2007). "The Development of Bio Fuels in Brazil: Implications for Cottonseed." Paper presented at the World Cotton Research Conference, Lubbock, Texas, US. September 1014. Banse, M., H. van Meijl, and G. Woltjer (2008). "The Impact of First and Second Generation Biofuels on Global Agricultural Production, Trade, and Land Use." GTAP Conference Paper, June 2008. https://www.gtap.agecon.purdue.edu/ re sources/download/3693.pdf Bayoulou, J. (2010). "La Prise en Compte des Coproduits du Coton dans le Mécanisme de Fixation du Prix de Coton Graine au Burkina Faso." Report Page 24 prepared for the Department of Agricultural Economics, Michigan State Uni versity, and the West African Cotton Improvement Program. Blasi, Dale A., and Jim Drouillard (2002). "Composition and Feeding Value of Cot tonseed Feed Products for Beef Cattle." Kansas State University, Agricultural Experiment Station and Cooperative Extension Service. Dickey, David, and Wayne A. Fuller (1979). "Distribution of the Estimators for Time Series Regressions with Unit Roots," Journal of the American Statistical Association 74: 427431. FalckZepeda, Jose, Daniela Horna, and Melinda Smale (2007). "The Economic Impact and the Distribution of Benefits and Risks from the Adoption of Insect Resistant (Bt) Cotton in West Africa." Discussion Paper No. 718. Washington, DC: International Food Policy Research Institute,. Food and Agriculture Organization (FAO) (2008). The State of Food and Agricul ture--Biofuels: Prospects, Risks, and Opportunities. Rome: FAO. ________ (2004). The State of Food and Agriculture--Agricultural Biotechnology: Meeting the Needs of the Poor? Rome: FAO. Gergely, Nicolas (2009). "The Cotton Sector of Benin." Africa Region Working Pa per Series No. 125. Washington, DC: World Bank. Gergely, Nicolas, and Colin Poulton (2009). "Valorization of Cotton By Products." In D. Tschirley, C. Poulton, and P. Labaste, eds., Organization and Performance of Cotton Sectors in Africa: Learning from Reform Experience. Wash ington, DC: World Bank. Gilbert, Christopher (2009). "How to Understand High Food Prices." CIFREM and Department of Economics, University of Trento, Italy. Hinkle, Lawrence E., and Peter J. Montiel (1999). Exchange Rate Misalignment; Concepts and Measurements for Developing Countries. Oxford: Oxford University Press for the World Bank. Houthakker, Hendrik S. (1975). "Comments and Discussion on `The 197275 Commodity Boom' by Richard N. Cooper and Robert Z. Lawrence," Brookings Papers on Economic Activity 3: 718720. Jarvis, Michael (2009). "Faire Tache d'Huile: Cooking Oil Fortification in West Africa." Case Study prepared for Business Innovation to Combat Malnutrition. World Bank Institute. Washington, DC: World Bank. International Cotton Advisory Committee (ICAC) (2010). "Gossypol in Cotton Seed," ICAC Recorder, XXVIII: 38. ________ (2000). "The Cotton Seed," ICAC Recorder, XVII: 38. Kelly, Valerie, Sonja Perakis, Boubacar Diallo, and Niama Nango Dembéle Page 25 (2010). "Cottonseed, Oil, and Cake: CoProducts or Byproducts in the C4 Cotton Sectors?" Department of Agricultural Economics, Michigan State Uni versity. Martin, Steve D. (1990). "Gossypol Effects in Animal Feeding Can Be Con trolled." Originally published in Feedstuffs, vol. 62. Reprinted by the National Cottonseed Products Association. http://www.cottonseed.com Mitchell, Donald (2010). Biofuels in Africa: Prospects for Sustainable Development. Draft report. Washington, DC: World Bank. ________ (2009). "A Note on Rising Food Prices." Policy Research Working Paper 4682. Washington, DC: World Bank. Nixon, H. C. (1930). "The Rise of the American Cottonseed Oil Industry," Journal of Political Economy 38: 7385. Organization for Economic Cooperation and Development (2008). Economic Assess ment of Biofuel Support Policies. Directorate for Trade and Agriculture. Paris: OECD. http://www.oecd.org/dataoecd/19/62/41007840.pdf Poulton, Colin, and Wilbald Maro (2009). "The Cotton Sector of Tanzania." Africa Region Working Paper Series No. 127. Washington, DC: World Bank. Rosegrant, Mark (2008). "Biofuels and Grain Prices." Washington, DC: International Food Policy Research Institute. http://beta.irri.org/solutions/images/publications/ papers/ifpri_biofuels_grain_prices.pdf Subramanian, Arjunan, and Matin Qaim (2010). "The Impact of Bt Cotton on Poor Households in Rural India," Journal of Development Studies 46: 295311. Sumner, Daniel A. (2006). "Reducing Cotton Subsidies: The DDA Cotton Initia tive." In Kym Anderson and Will Martin, eds., Agricultural Trade Reform and the Doha Development Agenda. Washington, DC: World Bank. Taheripour, F., D. K. Birur, T. W. Hertel, and W. E. Tyner (2008). "Biofuels and Their Byproducts: Global Economic and Environmental Implications." GTAP Conference paper, Helsinki, June. Tripp, Robert (2009). Biotechnology and Agricultural Development: Transgenic Cot ton, Rural Institutions, and ResourcePoor Farmers. New York: Routledge, Taylor, and Francis Group. Tschirley, David, Colin Poulton, and Patrick Labaste (2009). Organization and Per formance of Cotton Sectors in Africa: Learning from Reform Experience. Washing ton, DC: World Bank. US Agency for International Development (USAID) (2006). Adding Value to West African Cotton: Assessment of the Cotton Textile Apparel Value Chain. Technical Report No 21. Accra: USAID, West Africa Trade Hub. Page 26 US Department of Energy/Department of Transportation (2008). "Impact of Biofu els on Food and Gasoline Prices." Washington, DC. Vasilaky, Kathryn (2009). "Cotton Byproducts in Uganda and Tanzania." Report prepared for the World Bank. June. Zanfongnon, Noureni (2009). "Study of Market for Cotton ByProducts in Benin and Burkina Faso." Report prepared for the World Bank. August. Page 27 APPENDIX A: UGANDA8 Introduced in the early 20th century, cotton was Uganda's most important cash crop for several decades. Output reached 60,000 tons of lint during the early 1930s and was sustained at this level for almost four decades, with an annual value of about US$300 million in today's terms. The success of cotton in Uganda pre1970 partly reflects the fact that this was a forcedlabor crop. Output fell drastically as the result of the political and economic turmoil of the 1970s; in just four years it dropped from 78,000 tons of lint to 14,000 tons (197276), and reached a record low of 2,000 tons in 1987. Uganda's policy reform program, begun in the late 1980s, addressed both macro and sectoral issues, including in agriculture. Reforms introduced in the cotton sector in 1993, coupled with the high cotton prices of the mid1990s, evoked a considerable supply response, with cotton production reaching 20,000 tons in two years. A wellfunctioning research program was put in place. Grow ers began receiving prompt payments and entrepreneurial activity increased enormously, following the entry of many private entities at all levels of primary processing, marketing, and trade. Despite these achievements, however, cotton output in 200509 averaged only about 25,000 tons of lint. In a recent review of Uganda's cotton sector post reform, Baffes (2009) conjectures that the fundamental issue is low profitability. This partly reflects low productivity--especially in areas with productive soils, cotton has been displaced by higher priced food crops--and partly the fact that cotton is a new crop for many of its producers, who have limited knowledge of its growing requirements. The history of cotton byproducts in Uganda mirrors that of cotton. Before the reforms of the early 1990s, all cotton seed processing facilities were owned and operated by the cooperative unions or the Cotton Board. After the reforms, all cotton seed processing facilities were privatized, as were the ginning opera tions. Currently, Uganda has five privatelyowned cotton seeds processing com panies, of which four are independent and one is associated with a ginning oper ation. It has 50 ginneries each handling very small volumes--which explains why the crushing facilities are independent entities. The ginning overcapacity reflects the large size of the sector during the 1950s and 1960s. Demand has determined the location of the cotton seeds processing indus try. While Uganda's cotton production is spread around the country, all five cot tonseed processing companies are located in or near Kampala, the key oil con 8 Most of the information in this Appendix is based on Vasilaky (2009). Page 28 suming area, implying that the seed has to be transported there for crushing. Of the five cottonseed crushing companies in Uganda, four were visited for this study in May 2009.9 Statistics related to their key characteristics are re ported in Table A1. Table A1 Summary Statistics of Uganda's Cotton ByProducts Industry (Four Compa nies) I II III IV I. Nature of operation Permanent employees 4 4 100 4 Temporary employees 21 25 100 26 Capacity (tons/day) 10 50 200 18 Year operation began 1993 1998 2000 2005 Type of operation Free standing Free standing Free standing Part of a ginnery II. Crushings (tons of seed) 2007 1,100 -- 10,000 2,900 2008 1,600 2,800 5,000 3,600 2009 2,500 5,000 10,000 4,800 III. Prices paid for cotton seeds (UG shs per kg) 2007 400 220 200 420 2008 480 475 450 465 2009 225 410 300 325 IV. Prices received from cotton oil (000 UG shs per 20liter container) 2007 37 35 37 55 2008 62 58 54 65 2009 60 58 51 58 V. Cotton meal sold (tons) 2008 800 1,400 500 1,800 2009 -- -- 350 2,400 Notes: `--` implies not reported. Source: Vasilaky (2009). Three of the companies employ about 30 workers on average (both tempo rary and permanent) while the fourth employs 200. The daily crushing capacity ranges from 10 to 200 tons of seed. Three facilities can crush other seeds (mainly 9 The four companies interviewed were: Busunju Oil Millers (Wakiso District); MM Holdings (Nakasero District); Agro Nile Industries Ltd. (Jinja District); and Agencies Ltd. Cotton Ginneries, Oil Millers, Importers, Exporters (Iganga District). Page 29 sunflower) in addition to cotton. All the facilities use expeller pressing technolo gy rather than the more efficient but much more costly solventbased approach for extracting oil. Most of the expelling machinery, like most of Uganda's ginning equipment, is imported from Indian companies. The wage rate of unskilled workers ranges from 3,000 to 4,000 UG shs/day (depending on whether other benefits are provided), which at an exchange rate of 2,000 UG shs/US$ implies earnings of US$1.50 2.00 per day. Financing for op erating the factories is obtained from local banks at interest rates ranging from 19 24 percent, which are considered very high by the owners of the operations. Almost all buying and selling is on a cash basis. On the few occasions when cot ton oil is sold on credit, the transactions reflect personal longterm relationships rather than formal credit arrangements. Prices paid by the companies for cottonseed varied widely during 2007, from 200 to 400 UG shs per kg. The following year, however, when world edible oil prices reached their peak, all four companies paid between 450 and 480 UG shs per kg for the seed. With the decline of world prices in 2009, the price paid for seed declined accordingly (Figure A1). On average for the three years pre sented here, prices paid by cotton seed processor as a ratio to the prices received by farmers (properly adjusted by the ginning outturn ratio of 0.35) gives a ratio 0.39, indicating that a substantial part of the revenues realized by the ginning companies comes from cotton byproducts. Across companies, the sale price of oil varied much less than the purchase price of seed. It increased from an average of 41,000 UG shs per 20liter container (jc) in 2007 to 61,000 UG shs per jc in 2008, and declined to 57,000 UG shs per jc in 2009 (Figure A2). The seasonal price variation in 2008 (data not reported), the year for which all four companies reported their prices was very small, at 64 62, 57 60, and 45 55 thousand UG shs per jc. Almost all cotton oil is sold locally for human consumption to wholesa lers, bakeries, and supermarkets (one factory used some cotton oil for soap man ufacturing). Only minimal stocks of cotton oil are held. The four companies also produce cotton meal, with a recovery rate from seed of about 50 percent. The meal is sold to local dairy and poultry famers. Asked what were the key prob lems facing the industry, the managers of all the companies cited electricity dis ruptions and high interest rates for operating capital. When asked about the prospects of using cotton oil as a feedstock for biodiesel production, they all dismissed the idea as either unprofitable or an unnecessary diversification for their firms, given the high demand in Uganda for cotton oil for human consump tion (indeed Uganda is a net importer of edible oils). They favored policy restric tions on imported edible oils but did not feel very strongly about them. Page 30 Figure A1 Uganda: Seed Prices Paid by Cotton Oil Processors (UG shs/kg) 600 I II III IV 500 400 300 200 100 0 2007 2008 2009 Note: The numbers represent annual average; I, II, III, and IV refer to the 4 companies interviewed. Source: Interviews with industry representatives (Vasilaky 2009). 9 Figure A2 Uganda: Oil Prices Received by Processors (UG 000 shs/20liter container) 80 I II III IV 60 40 20 0 2007 2008 2009 Note: The numbers represent annual average; I, II, III, and IV refer to the 4 companies interviewed. Source: Interviews with industry representatives (Vasilaky 2009). 10 Page 31 APPENDIX B: TANZANIA10 Early efforts to establish cotton as a plantation crop in Tanganyika bore no fruit, but during the 1920s new efforts focused on smallholder production. Cotton production rose significantly following the development of local varieties along with a better organization of the sector, and by 1966, Tanzania's cotton output was 80,000 tons of lint, almost 1 percent of global production. During the mid1960s, most aspects of the country's cotton industry, in cluding ginning, marketing, and trade, were transferred to cooperative unions and the Cotton Board. These entities soon became large bureaucracies failing to respond to the needs of the sector and, not surprisingly, the cotton sector began deteriorating. Despite various reform attempts, for the next two decades the sec tor performed poorly. Only after the privatization efforts of the early 1990s--set forth by the Cotton Act of 1994--did the sector get back on a sustainable path (see Baffes 2004 for an account of the cotton sector reforms in Tanzania and Poul ton and Maro 2009 on recent policy developments and performance.) In recent years, Tanzania has produced an average of 100,000 tons of cotton lint, almost all in the Mwanza region (districts of Mwanza, Shinyanga, and Bunda). Tanzania has 73 ginneries. Production of cotton byproducts has evolved similarly to ginning opera tions. The 1994 cotton reforms also applied to byproducts in the sense that both ginning seed processing operations were privatized, while new entrants were allowed to enter the market. Currently, there are 20 cotton crushing operations in Tanzania, ten of which were visited for this study in May 2009.11 Nine of these ten are associated with ginneries. The key characteristics of the ten facilities are reported in Table B1. The ten companies employ an average of 120 temporary and permanent workers, ranging from a minimum of 35 to a maximum of 260, and their daily crushing capacity ranges from 30 to 200 tons of seed. Most of them operate on a 24hour basis during the cotton seasons. Only three of them crush other oilseeds. 10 Most of the information in this Appendix is based on Vasilaky (2009). 11 The ten seed processing companies interviewed were: Birchand Oil Mill Ltd.; Voil/Vegetable Oil Industries, Ltd.; I. C. K. Cotton Oil Co. Ltd (Mwanza District); Afrisan Ginning Ltd.; Gaki In vestment Co. Ltd; Jambo Group of Companies Ltd.; Fresho Investment Co. Ltd. (Shinyanga Dis trict); SM Holdings Ltd./Nkalalo Ginnery; Verrian Tanzania Ltd./Bunda Oil Industries; S&C Gin nery/Balamba Ginnery (Bunda District). Page 32 Table B1 Summary Statistics of Tanzania's Cotton ByProducts Industry (Ten Companies) I II III IV V VI VII VIII IX X I. Nature of operation Permanent employees 35 60 30 15 3 12 8 9 57 3 Temporary employees 32 275 100 30 90 90 46 45 200 33 Capacity (tons/day) 75 90 100 70 100 60 60 30 200 50 Year operations began 1996 1967 2008 2006 2003 1999 2001 2008 1999 1997 II. Crushings (000 tons of seed) 2007 40.0 16.5 4.0 6.4 -- -- -- -- -- -- 2008 9.0 4.5 6.2 7.1 10.0 8.0 7.5 1.2 20.0 20.0 2009 9.0 4.2 20.0 10.0 7.5 9.0 12.0 4.5 45.0 10.0 III. Prices paid for cottonseed (TZ shs per kg) 2006 -- -- -- 120 -- 130 -- -- -- -- 2007 150 -- -- 200 250 250 -- 150 205 -- 2008 250 250 250 210 250 250 -- 250 175 ­ 2009 190 190 180 -- 150 190 -- 150 -- -- IV. Prices received from cotton oil (000 TZ shs per 20liter container 2007 25.0 -- -- -- 12.0 32.0 22.0 -- -- 28.0 2008 21.5 46.8 31.9 26.0 15.0 35.0 25.0 -- -- 30.0 2009 30.0 36.0 27.0 30.0 20.0 27.0 27.0 -- 25.0 24.0 V. Cotton meal sold (tons) 2006 -- 6,440 -- 8,500 3,150 -- -- -- -- -- 2007 -- 7,590 -- 3,692 700 -- -- -- -- -- 2008 5,000 2,070 2,000 3,200 850 4,000 450 500 10,000 -- 2009 -- 1,932 -- -- -- -- -- -- 20,000 -- Notes: Some of the 2009 figures are estimates. `--` implies not reported. Some companies did not report seed prices because most of the seed came from their own ginning operations. Source: Vasilaky (2009). Page 33 All the facilities use expeller pressing technology rather than a solvent based approach for extracting oil. Most of their machinery is imported from In dian companies: Kumar (Delhi) and Umas (Mumbai). Three quarters of them produce semirefined cotton oil, and few of them bleach or deodorize the oil. In stalling a double refinery operation to improve the quality of the oil would cost approximately US$200,000, which for an average oil refining facility would double the costs of producing cotton oil. The wage rate of unskilled workers in cotton byproducts facilities ranges from 2,500 to 3,000 TZ shs/day, which at an exchange rate of 1,200 TZ shs/US$ implies US$ 1.5 2.5 per day (very similar to Uganda). Interest rates paid by the companies for loans range from 8 percent to 16 percent, much lower than in Uganda. Sales of cotton oil are made on both a cash and a credit basis. There are no formal forward arrangements except for some longterm relationships. Prices paid for cottonseed averaged 200 TZ shs per kg during 2007, rang ing from 150 to 250 TZ shs per kg (Panel III, Table B1). They increased to 235 TZ shs per kg, on average, during 2008, consistent with the peak of world prices, and then declined to 175 TZ shs per kg in 2009. The price variability across com panies was much smaller in 2008 and 2009. Of the six companies that reported full statistics for 2008 and 2009, all but one paid 250 TZ shs per kg for seed dur ing 2008, while the 2009 price fluctuated within a narrow range of 150 and 180 TZ shs per kg (Panel III, Table B1 and Figure B1). The seasonal price variation of seed in 2008, the year for which most com panies reported prices, was quite high (Figure B2) while for cotton oil sales, the price variation among companies in 2009 was relatively low (Figure B3). The gap between the companies' highest and lowest oil prices as a percen tage of the high price--an approximate measure of variability--averaged 0.24, and is remarkably similar to the gap in Tanzania's cotton oil market (see Figure B4, which depicts similar statistics for seven companies). Most of Tanzania's cotton oil is consumed locally in the Mwanza region and is distributed through wholesalers and supermarkets. Only one company reported shipping some oil to Dar es Salaam. Two factories use limited quantities of oil for soap manufacturing. Three companies use their own oil label. Some companies hold stocks for a few months. Page 34 Figure B1 Tanzania: Seed Prices Paid by Cotton Processors (TZ shs/kg) 300 I II III IV 250 200 150 100 50 0 2007 2008 2009 Note: The numbers represent annual average; I, II, III, and IV refer to 4 of the companies interviewed. Source: Interviews with industry representatives (Vasilaky 2009). 11 Figure B2 Tanzania: 2008 Seasonal Price Variation of Seed (TZ shs/kg) 300 June December 250 200 150 100 II I III IV V Note: The numbers refer to companies that reported complete data Source: Interviews with industry representatives (Vasilaky 2009). 12 Page 35 Figure B3 Tanzania: Oil Prices Received by Processors (TZ 000 shs/20liter container) 40 I II III IV 30 20 10 0 2007 2008 2009 Note: The numbers represent annual average; I, II, III, and IV refer to 4 of the companies interviewed. Source: Interviews with industry representatives (Vasilaky 2009). 13 Figure B4 Tanzania: 2008 Seasonal Price Variation of Oil (TZ shs/20liter container) 40,000 Highest Lowest 35,000 30,000 25,000 20,000 15,000 10,000 I II III IV VI VI VII Note: The numbers refer to companies that reported complete data. Source: Interviews with industry representatives (Vasilaky 2009). 14 Page 36 The companies also produce cotton meal, obtaining a recovery rate from seed of about 55 percent--a little higher than Uganda's average of 50 percent. While some limited quantities of meal are sold locally, most is transported else where, including to Dar es Salaam and Arusha, with some exported to Uganda, Kenya, and South Africa. Only one company reported an export cake price: US$340 per ton in 2008 and US$140 per ton in 2009. Table B2 reports the weight and value composition of cottonseed, based on information received from three seed processing companies. On average, one ton of cottonseed generate 140 kgs of cotton oil, 550 kgs of husks, 260 kgs of meal, and 50 kgs of waste. In 2008 US$ value terms, the total revenue from one ton of cottonseed was US$195: of this US$129 came from cotton oil (66 percent), US$55 from cake (28 percent), and US$11 (6 percent) from husks. Table B2 Revenue Breakdown of One Ton of Cottonseed for 2008 (Four Companies) Quantity (kgs) Unit Price (TZ shs/kg) Revenue (TZ shs) Revenue (US$) Oil 140 1,100 154,000 128.72 Meal 260 120 66,000 55.17 Husk 550 150 13,000 10.87 Waste 50 0 0 0.00 Total 1,000 NA 233,000 194.76 Notes: The US$ value was calculated at an exchange rate of US$ = 1,196 TZ shs. Some husks are used for energy generation. `NA' means nonapplicable. Source: Vasilaky (2009). Following a calculation similar to Uganda for the three years presented here, prices paid by cotton seed processor as a ratio to the prices received by farmers (properly adjusted by the ginning outturn ratio of 0.35) gives a ratio 0.30, indicating that almost one third of the revenues realized by the ginning compa nies comes from cotton byproducts.12 Asked what were the key problems facing the industry, the managers and operators of all the companies cited the erratic supply of electricity and asso ciated disruptions. When asked about the prospects of using cotton oil as a feeds tock for biodiesel production, they all dismissed the idea as either unprofitable or unnecessary for their firms, given the high demand in Tanzania for cotton oil for human consumption (Tanzania, like Uganda, is a net edible oil importer.) Only Several respondents in Tanzania noted that cotton oil is the most profitable part of their opera 12 tion (including ginning). Page 37 one manager said that he had sold cotton oil to a company in Germany for pro duction of biodiesel for experimental purposes. Managers' views about policy restrictions on imported edible oils were similar to those in Uganda. While they favored such restrictions, they did not feel strongly about them because of the high local demand for edible oils. Some man agers stated that in practice most edible oil imports are labeled as crude, even though they are refined, and thus escape Tanzania's 25 percent import duty on refined oil. Page 38 APPENDIX C: BENIN13 Introduced in the mid1950s under the auspices of the French stateowned com pany CFDT (Compagnie Française pour le Développement des Fibres Textiles), cotton is an important cash crop in Benin. Seed cotton production has averaged 270,000 tons during the past three seasons (about 0.5 percent of global production), after reaching 400,000 tons in the early 2000s. Like Burkina Faso, Benin has underta ken many policy reforms in the cotton sector during the past decade, including privatization of ginning and marketing. In 1984 all aspects of the sector were transferred to a new parastatal, SO NAPRA (Société Nationale pour la Production Agricole). Changes in the structure of the sector were first contemplated in the early 1990s, mainly in response to an earlier crisis when a combination of a larger than expected crop, low world pric es, high producer price, appreciation of the CFAf, and limited ginning capacity caused SONAPRA huge financial losses.14 Most input supply activities were pri vatized and three new ginning operations were added in 1995, followed by sev eral more in 1998. Yet the new structure caused numerous conflicts resulting in frequent political interference. In response, the government created entities that assumed responsibilities for various aspects of the cotton industry. Reforms continued during the early 2000s, with the stated objective of "developing a private but nationally integrated cotton supply chain" whose management would be transferred from the government to an interprofessional body. The actual outcome was a highly regulated system without many competi tive elements. On the other hand, following the weakening of the global cotton market, producer prices fell and cotton output was halved within just four years (from 400,000 tons of seed cotton in 2001/02 to less than 200,000 tons in 2005/06). Though cotton production has recovered somewhat during the last few years, it remains far below the levels of the early 2000s. 13 Most of the information in this Appendix is based on Zanfongnon (2009). Also see Gergely (2009) for an indepth discussion of Benin's cotton sector. The CFA (Communauté Financière Africaine) franc or CFAf is a currency used in 14 countries of 14 West and Central Africa (12 former French colonies, GuineaBissau, and Equatorial Guinea). The CFAf was pegged to the French franc (FF) until 1999 and to the euro since then (1 = 655.957 CFAf). Since its creation in 1945 it has been adjusted only twice: in October 1948 (from 1 FF = 1.70 CFAf to 2.00 CFAf) and in January 1994 (from 1 FF = 100 CFAf to 200 CFAf). Although the CFAf has been praised for offering macroeconomic stability to its members, during episodes of over valuation it places the export sectors of the respective countries at a competitive disadvantage (Hinkle and Montiel 1999). It is believed that this has been the case during the early and mid 2000s. Page 39 When SONAPRA was privatized in 2008, after several postponements, three new companies emerged: SODECO (Société de Développement du Coton), with ten ginneries; ICAGIE (Industries Cotonnières AssociéesGroupement d'Intérêt Économique) an industrial group comprising five societies with five ginneries; and LCB (Le Label Coton du Bénin), with one ginnery. One third of SODECO's shares are held by the government, one third by producers, and one third by a private company. Cotton prices are still determined in a panseasonal and panterritorial manner. As in most cotton producing countries in WCA, cottonseed in Benin was left largely unprocessed before 1980. Some seed was exported to Europe where there was strong feed demand from the dairy industry. During the mid1980s SONICOG (Société Nationale des Industries des Corps Gras), a stateowned oilseed processing company, began crushing limited quantities of cottonseed obtained under favorable terms from SONAPRA. The cotton reform efforts of the 1990s affected the cotton byproducts in dustry as well. In 1996, a new seed processing company, FludorBénin, entered the market, building a new oil crushing facility with an investment of CFAf 3.4 billion and annual capacity of 90,000 tons. In 1997, SONICOG was privatized and two new companies were created: SHB (Société des Huileries du Bénin), with an initial investment of CFAf 4.2 billion and annual capacity of 120,000 tons, capable of producing 19,200 tons of oil and 45,600 tons of meal; and IBCG (Industrie Beni noise des Corps Gras), with an investment of CFAf 3.5 billion with annual capacity of 40,000 tons (Table C1). However, IBCG has been idle since 2006 (also, there have been unconfirmed reports that SHB did not operate during the 2009/10 sea son). Table C1 Quantities of Cottonseed Crushed in Benin (tons) Labor force Capacity Seeds crushed (tons/year) (total) (tons/year) 2003/04 2005/06 2006/07 2007/08 2008/09 FLUDOR 420 90,000 72,186 58,573 65,710 47,582 70,000 SHB 310 120,000 105,000 33,000 57,000 78,000 57,000 IBCG 100 40,000 0 0 0 0 0 Total 830 250,000 177,186 91,573 122,710 125,582 127,000 Source: Ahohounkpanzon and Zakariallou (2010) and Zanfongnon (2009). In effect, Benin's cotton byproducts industry now consists of two large companies: SHB and FludorBénin. SHB has 110 permanent and 200 temporary employees while Fludor has 250 permanent and 170 temporary employees. Both Page 40 companies borrow funds on a shortterm basis to run their operations, with in terest rates ranging between 8 percent and 10 percent. Their purchases of seeds are made with 40 percent 50 percent cash and the remainder on credit. All the cotton oil produced goes for human consumption. Though the two companies sometimes crush other seeds, cotton is by far their key raw material source. Unlike seed processing companies in Tanzania and Uganda, both com panies use the solvent method to extract the oil. With this advanced technology they can obtain an extraction rate of oil up to 16 percent of the weight of the cot tonseed, very similar to the US average. The price they pay for seeds is the same throughout the season (similar to the panseasonal and panterritorial pricing mechanism for lint). The price they receive for cotton oil, however, varies throughout the season depending largely on demand conditions. Sales of oil and meal are mostly for cash. Most cotton oil is domestically consumed, while some is exported to Nige ria. Detailed data on exports of cotton oil are not available since most of these exports are unrecorded.15 An estimated 90 percent of cotton meal is exported, mostly to South Africa, which has a large cattle industry. Europe used to be a key destination for Benin's cotton meal but stringent requirements for limiting afla toxin led to high rejection rates.16 Benin's trade policies on cottonseed and derived products have been con troversial. In 2001, the government banned exports of cottonseed in order to en sure adequate supplies for the domestic market, but the ban was never strictly enforced and was removed in the 2004/05 crop year. A ban on oil imports was introduced in 2006 but was later contested by neighboring Togo (most of Benin's imported oil comes through Lomé, Togo's main port) because it violated UE MOA (Union Economique et Monétaire OuestAfricaine) rules.17 In 2007 Benin con formed with the UEMOA's decision and since then has allowed edible oil im ports from UEMOA countries. Currently, Benin imposes a 45 percent import du ty on palm oil, but traders report that in most cases this is not paid. Benin also imposes an 18 percent value added tax on all domestic edible oil sales--a tax 15 The figures on trade are subject to debate. Some believe that the cotton seed processors sell cot ton oil to traders who have export permits (and thus intend to export) but in reality those traders sell it in the domestic market thus avoiding the VAT. Aflatoxin is a toxic and carcinogenic substance found often in cereals, oilseeds, and spices as 16 well as milk in animals which have been fed with contaminated feed. Most food regulatory au thorities (e.g., the Food and Drug Administration in the US) have established recommended le vels of aflatoxin that various food items should not exceed. UEMOA is a monetary union of the following West African countries: Benin, Burkina Faso, 17 Côte d'Ivoire, GuineaBissau, Mali, Niger, Senegal, and Togo. Page 41 that, many argue, forces companies to export the oil, since exports are not subject to the sales tax. As a result local demand is met by imported oil. Until 2005, the government took an active part in negotiations between the ginners and the cottonseedprocessing companies, in order to ensure that seed supplies were sufficient to keep the processing companies in business. Currently, the price of cottonseed is a result of negotiations between the ginning and seed processing industries. While it has been reported that ginners often do not honor the agreement and export part of their cottonseed if that will yield them a higher price, the figures reported in Table C1 indicate that this does not happen on a large scale. In fact, during 2009, more seed was crushed in Benin than domestic production--perhaps reflecting unrecorded imports of cottonseed from neigh boring countries, assuming that the statistics are accurate. Table C2 reports detailed price and quantity statistics for one of the two companies surveyed. Prices paid for seeds increased considerably during the past three seasons from 22,500 CFAf per ton in 2006/07 to 46,000 CFAf per ton in 2007/08 and 70,000 CFAf per ton in 2008/09 (the prices paid to farmers in these three seasons were 170, 170, and 190 CFAf per kg). However, the prices received by the company for cotton oil did not increase much: they rose from 500,000 in 2006/07 to 600,000 in 2007/08 and then declined back to 500,000 CFAf per ton in 2008/09. If the numbers are correct, they imply that the domestic market is somewhat isolated from the international markets of edible oils. Table C2 Price and Quantity Statistics for Benin and SHB, 19992009 Quantities (tons) Prices (CFAf/kg) Cotton Seed Oil Meal Cotton Seed Oil Meal 1999/00 364,000 27,000 4,000 14,700 185 26 -- -- 2000/01 336,000 62,000 12,000 29,000 200 29 390 90 2001/02 415,000 100,000 16,000 32,000 200 28 440 66 2002/03 334,000 75,000 13,000 32,000 180 33 390 56 2003/04 332,000 92,000 14,000 37,000 190 37 500 37 2004/05 427,000 105,000 9,700 26,000 201 30 445 40 2005/06 191,000 33,000 9,800 22,000 170 23 450 37 2006/07 271,000 57,000 9,600 23,000 170 23 500 54 2007/08 269,000 78,000 12,000 22,000 170 46 600 100 2008/09 216,000 57,000 8,500 22,000 221 70 500 100 Notes: The first and fifth columns refer to the entire country's seed cotton crop while the remain ing columns refer to SHB. The seed price during 2006/07 ranged from CFAf 15 to 30 per kg. `--' means not reported. Cotton prices refer to final payments (base price plus supplement). Source: Zanfongnon (2009). Page 42 The contribution of cottonseed to the total value of cotton seems to have increased in Benin consistently with the changes that took place in the global markets. When the price of cottonseed paid by the seed processing companies is expressed as a share of the total value of cotton (i.e. the sum of the price received by the producers, reported in the second column of table C2, plus the ginning costs, estimated at about CFAf 160 per kg of lint), the average contribution in creased from 6 percent (average 1999/2000 to 2006/07) to 13 percent over the past two seasons (Figure C1). However, the value of cottonseed has remained very low in relation to ginning costs. As shown in Figure C2, cottonseed prices in Benin covered only 23 percent of actual ginning costs during 1999/20002006/07. In the US, by contrast, the value of cotton byproducts is traditionally expected to cover ginning costs (see Appendix E). Figure C1 Benin: Seed's Contribution to the Value of Cotton (percent) 16% 12% 8% 4% 0% 1999/2000 2000/01 2001/02 2002/03 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 Note: This is the ratio of the price of cotton seed over the sum of the grower's price and ginning costs Source: Author's calculations based on industry data 16 Page 43 Figure C2 Benin: Seed Value to Ginning Costs Ratio (percent) 60% 50% 40% 30% 20% 10% 0% 1999/2000 2000/01 2001/02 2002/03 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 Note: This is the ratio of the price of cotton seed over the ginning costs Source: Author's calculations based on industry data 17 Page 44 APPENDIX D: BURKINA FASO18 Cotton was introduced in Burkina Faso towards the end of the colonial period, and the country is now SSA's largest cotton producer; its seed cotton output av eraged half a million tons during the past three seasons (about one percent of world output) and exceeded 700,000 tons in 2006/07. Cotton is the country's most important export item, accounting for more than half of its total merchandise trade and about 5 percent of GDP. Development of the sector was originally the responsibility of the French stateowned company CFDT, which remained in charge until 1975 when it was replaced first by a joint venture between the government and CFDT and in 1979 by a new cotton parastatal, SOFITEX (Sociètè Burkinabé des Fibres et Textiles). Reforms of the cotton sector were first considered in 1991 when the man agement responsibilities of the sector were transferred to growers and SOFITEX. In 1998, the government further reduced its stake in the cotton company by transferring 30 percent of its shares to a producer organization and 30 percent to CFDT's successor DAGRIS (Société du Développement Agricole du Sud).19 In 1999 a committee was formed to coordinate the functions of the cotton company and the producers' organization for activities such as the determination of farmgate pric es and the management of the research program. Finally, the sector was opened to (limited) competition when two private cotton companies received permission to enter the market. These companies--SOCOMA (Société Cotonniére du Gurma) and Faso Coton--were given an eightyear exclusive right to buy cotton from spe cific zones which produce about 15 percent of the country's cotton output. In 2006, an umbrella organization, the Burkina Cotton Interprofessional Association (AICB, Association Interprofessionnelle du Coton du Burkina), which represents cot ton farmers and ginners, was created to coordinate actions of all three cotton companies. In response to the CFAf devaluation by 100 percent in 1994, the policy re forms, and the high prices received by cotton growers induced considerable supply response, with seed cotton output reaching a record high of 713,000 tons in the 2004/05 season, a twofold increase from a decade earlier. Subsequently, however, the unsustainably high prices paid to cotton growers due to the deterioration of the world cotton market imposed fiscal pres sure on the cotton companies (and the government), in turn causing production 18 Most of the information in this Appendix is based on Zanfongnon (2009). In January 2008 DAGRIS was purchased by the Advens Group (51 percent) and CMACGM (49 19 percent) and was renamed to Geocoton. Page 45 to decline to less than half a million tons during the past two seasons. The ap preciation of the CFAf against the US dollar played an important role as well.20 In response, AICB introduced a new pricing mechanism supported by a smoothing fund (fonds de lissage), which takes into consideration historical and current pric es, price forecasts, and other market conditions. Burkina Faso has been a leader among SSA cotton producers in that it be gan biotech cotton trials earlier this decade in partnership with Monsanto. In 2008 the research had progressed to onfarm trials, by 2009 it had expanded to more than 100,000 hectares of commercial plantings by smallscale farmers, and by the 2010 planting season it is expected to have led to full conversion to biotech cotton varieties (apart from refuge setasides).21 (Box D1.) Box D1: Economic Effects of Biotech Cotton Varieties The pros and cons of adopting biotech cotton varieties have been extensively discussed (see Tripp 2009 for a comprehensive review of the literature). FAO's (2004) review showed that on balance, biotech cotton growers are better off than growers of conven tional varieties. Baffes (2005) argued that in addition to subsidy elimination and domes tic reforms, adoption of biotech varieties should have been a priority among policymak ers in lowincome cotton producing countries. Similarly, FalckZepeda and others (2007) concluded that the downward pressure on world cotton prices caused by the largescale adoption of biotech cotton is likely to force other countries to adopt the technology in order to compete in the global market. The data certainly confirm the presence of such pressure. Anderson and Valenzuela (2007) found that the benefits from full adoption of biotech cotton varieties by African cottonproducing countries could be even greater than the benefits of the removal of all cotton subsidies by the US and the European Un ion. Subramanian and Qaim (2010) found that the expansion of biotech cotton in India contributed to poverty reduction and rural development. Cottonseed processing in Burkina Faso goes back to the 1980s when the government acquired a private oilseed processing company under the name SNHB (Société Nationale des Huileries du Burkina), later renamed CITEC and more recently SNCITEC (Société Nouvelle Huilerie et Savonnerie Citec). 20 The euro's appreciation against the US dollar during the early to mid2000s has been a key con straint faced by the WCA cotton sectors. Between 2000 and 2007, for example, nominal world cot ton prices experienced a 7 percent increase, from US$1.30/kg to 1.40/kg. The CFAf, however, ap preciated against the US$ from CFAf 712 in 2000 to CFAf 479 in 2007. Thus, in domestic (nomin al) currency terms, WCA cotton producers experienced a 28 percent decline. Cotton Outlook (March 9, 2010, p. 14) quoted a SOFITEX researcher who asserted that "the result 21 [of biotech cotton] has been a significant gain in productivity, by virtue both of reduced costs and improved yields." Page 46 Today for all practical purposes SNCITEC is Burkina Faso's only large cottonseed processing company, coexisting with multiple much smaller, less technologically advanced, companies. Following the company's privatization in 1995, an investment of CFAf 12 billion was made to modernize its cotton crush ing facilities. Geocoton (the successor of DAGRIS) holds a 52.5 percent stake in SNCITEC while most of the remainder is held by SOFITEX, implying that SN CITEC receives preferential access to cottonseed over its competitors. This begs the question of whether SNCITEC would have been profitable if it bought cot tonseed at market prices. By most accounts, SNCITEC is the most technologically advanced cotton seed processing operation in WCA. The company's maximum annual capacity is 120,000 tons with a total labor force of almost 1,000 employees (Table D1). As much as 95 percent of its energy requirements are met by the use of husks and it achieves more than a 16 percent oil extraction rate, equivalent to the rates achieved in the US. SNCITEC produces fully refined oil and in 2008 began forti fying it with Vitamin A under its own label (Jarvis 2009). It has also installed a pilot plant for biofuel production with an annual capacity of 3,000 tons, and plans to expand this to 10,000 tons (Bayoulou 2010). Table D1 Price and Quantity Cotton Statistics for Burkina Faso and SNCITEC Quantities (tons) Prices (CFAf/kg) Cotton Seed Oil Meal Cotton Seed Oil Meal 1999/00 254,189 108,817 15,234 52,135 185 24 467 35 2000/01 275,800 100,064 16,555 47,083 170 29 449 39 2001/02 378,522 103,217 17,798 50,667 200 29 485 40 2002/03 404,419 107,320 17,252 55,013 192 32 502 45 2003/04 483,390 90,210 12,250 47,690 191 40 490 33 2004/05 632,503 97,134 14,798 47,660 215 40 461 36 2005/06 713,123 103,046 19,448 53,279 179 40 450 38 2006/07 684,958 98,268 16,244 51,781 170 40 503 45 2007/08 355,422 88,442 15,596 47,793 163 82 738 103 2008/09 451,473 115,000 19,550 63,250 179 82 602 124 Notes: The first and fifth columns refer to the entire country's cotton crop. The remaining col umns refer to SNCITEC. Cotton prices refer to final payments (base price plus supplement). Source: Zanfongnon (2009) and Bayoulou (2010). A second large company, SOFIB (Société Industrielle Barro et Frères), began cottonseed processing in 1984 with annual capacity of 36,000 tons and labor force Page 47 of 425 (Table D2). SOFIB has had very low capacity utilization during recent years while it did not process any cottonseed during the 2009/10 season. A third major company, JOSSIRA Industries, entered the market in 2001 with annual ca pacity of 45,000 tons, but has not been in operation since 2007. Both these com panies have been in arrears with SOFITEX. Table D2 Size of Major Seed Processing Companies in Burkina Faso Labor force Capacity (tons/year) (all employees) Seed Oil Meal SNCITEC 990 120,000 20,000 70,000 SOFIB 425 36,000 5,000 30,000 JOSSIRA 165 45,000 6,000 35,000 Total 1,580 201,000 31,000 135,000 Notes: JOSSIRA has not been in operation since 2007 while SOFIB did not process any cottonseed during 2009/10. Source: Zanfongnon (2009). Numerous smaller cotton seed processing companies joined the industry during the mid2000s, as the result of the policy reforms and the subsequent ex pansion of the cotton sector. These firms are scattered throughout the country, with a high concentration in BoboDioulasso, the country's key cotton producing region. The expansion of the industry, which also reflected strong demand for cotton meal by neighboring countries, was aided by investments made by expa triates who returned to Burkina Faso because of the civil conflict in Côte d'Ivoire.22 Because the smaller processors had difficulty obtaining seeds from SOFI TEX during their early years of operation, they formed an association (APHB, Association des producteurs dhuile des HautsBassins) in 2003 to facilitate their supply of seeds from SOFITEX, as well as easing their access to funds from fi nancial institutions and promoting the quality of cotton oil through a national labeling system. In 2005 a competing association was formed, GTPOB (Groupe ment des transformateurs des produits oléagineux du Burkina), which acquired most of APHB's membership and eventually replaced its competitor. Internal conflicts led some members to form two new associations in 2008: GHH (Groupement des The minimum equipment requirement (i.e., one cold pressing machine which is produced do 22 mestically) costs about 2 million CFAf (equivalent to about US$4,200, in 2010 US dollars). There fore, setting up an averagesize operation with five pressing machines costs a little more than US$20,000. Page 48 Huiliers du Houet) and CPPOD (Coopérative des Producteurs de Produits Oléagineux et Divers). As of 2009 GTPOB was the dominant one with 42 members, followed by GHH (11 members), and CPPOD (6 members.) While membership in these three associations gives the official count of Burkina Faso's small cotton seed processing companies, it should be noted that the industry is at the early stages of development, with considerable entry and exit, and also that because some companies are not members of any of these associations the actual number of companies may be much higher. For the current study, 21 of the smaller companies were surveyed, ac counting for about one third of the smaller seed processing companies in Burkina Faso.23 Summary characteristics and key statistics on quantities and prices are re ported in Table D3. The average age of the companies is three years. Their aver age daily capacity is 34 tons and their average labor force is 43 workers, mostly on temporary arrangements. The wage rate paid to temporary employees aver aged CFAf 33,750 per month (equivalent to about US$ 2.50 per day, assuming 25 working days per month). During 2007/08 and 2008/09 they crushed an average of 1,860 and 1,770 tons of cottonseed, respectively. During the 2008/09 season the combined quantity crushed by all the small companies reached 50,000 tons of seed, or one third of the total seed crushed in Burkina Faso. The 21 smaller companies surveyed here--accounting for one third of all small seed processing companies in Burkina Faso--crushed a total of 35,000 tons of seed in 2008/09. If 50,000 tons were crushed in total, this means that the re mainder (about 40) companies crushed only 15,000 tons. This may not be an un reasonable estimate given that the ones surveyed were the more established. In some sense, the smaller companies compensate for the gap left by SOFIB and JOSSIRA. On a few occasions they crush other seeds, though they find cotton by far the most profitable. Their oil extraction rate varies between 6 percent and 10 percent (much lower than SNCITEC's 16 percent). Yet, the smaller seed proces sors' lower oil extraction rates do not necessarily imply lower overall profitabili ty, because the oil remains in the meal, increasing its protein content and hence its value. The 21 companies surveyed were: AGROPA (Fada N'Gourma region); Birba et fils; COMITRA; 23 Délice Huilerie; SHS du Faso; Karamongo Lanfera; Huilerie Lodoun; NIDOR; OMORY; SOCHIB; AKAMAF; HUNOFA; DJENE DIOR; TRAORE et frères; NATILGUE; ANGOFASO (Bobo Dioulasso region); Sana Moussa; SIPAB; LIZA et fils; GENOL; SATOS (Ouagadougou region). Page 49 Table D3 Summary Statistics of 21 Cottonseed Processing Companies in Burkina Faso I II III IV V VI VII VIII IX X XI I. Nature of operation Number of employees 33 7 13 103 20 55 42 107 8 8 46 Capacity (tons/day) 30 30 25 70 20 31 50 40 3 6 60 Year operations began 2007 2006 2006 2007 2007 2006 2008 2006 2005 2007 2004 Wage rate (CFAf/month) 40,000 30,000 48,000 -- -- 32,000 38,000 32,000 25,000 25,000 31,115 II. Seed statistics Crushings in 2008 (tons) 200 100 700 5,773 -- 2,000 2,000 7,500 120 80 490 Crushings in 2009 (tons) 700 100 1,140 5,240 730 1,050 2,000 6,000 200 180 1,600 Seed price in 2008 (CFAf/ton) 92,000 90,270 75,000 75,000 -- 90,250 75,000 75,000 92,270 90,250 90,270 Seed price in 2009 (CFAf/ton) 92,000 90,270 75,000 75,000 75,000 90,250 75,000 75,000 92,270 90,250 90,270 III. Oil and meal statistics Oil price in 2008 (CFAf/jc) -- -- 13,000 12,000 -- 10,500 13,000 9,000 11,250 -- 12,500 Oil price in 2009 (CFAf/jc) -- -- 10,000 11,750 -- 13,000 12,000 9,000 11,250 -- 9,500 Meal sold in 2008 -- -- 612 -- -- -- 1,700 6,200 -- 70 320 Meal sold in 2009 -- -- 997 -- -- 918 1,700 5,020 -- 150 1,040 Meal price in 2008 (CFAf/ton) 90,000 90,000 100,000 100,000 -- -- 100,000 70,000 100,000 95,000 92,500 Meal price in 2009 (CFAf/ton) 90,000 110,000 100,000 105,000 -- 60,000 110,000 70,000 100,000 95,000 92,500 Continued next page Page 50 Table D3 (continued) Summary Statistics of 21 Cottonseed Processing Companies in Burkina Faso XII XIII XIV XV XVI XVII XVIII XIX XX XXI I. Nature of operation Number of employees 58 24 14 112 32 39 73 18 65 30 Capacity (tons/day) 36 20 12 70 50 40 60 10 40 20 Year operations began 2007 2009 2007 2005 2008 2006 2001 2008 2007 2009 Wage rate (CFAf/month) 31,000 40,000 30,000 50,000 25,000 35,000 30,000 32,000 40,000 27,000 II. Seed statistics Crushings in 2008 (tons) 1,440 -- 700 2,000 1,000 2,360 4,944 850 1,200 -- Crushings in 2009 (tons) 2,800 500 1,000 2,000 3,500 1,640 2,593 675 2,500 1,000 Seed price in 2008 (CFAf/ton) 91,368 -- 92,545 90,270 90,270 75,000 75,000 75,000 90,270 -- Seed price in 2009 (CFAf/ton) 91,368 90,250 92,545 90,270 90,270 75,000 75,000 75,000 90,270 75,000 III. Oil and meal statistics Oil price in 2008 (CFAf/jc) -- -- 11,500 14,000 13,730 10,000 13,500 12,500 14,000 -- Oil price in 2009 (CFAf/jc) -- 11,000 9,500 10,000 9,180 8,300 11,000 12,000 12,500 8,300 Meal sold in 2008 1,224 -- 595 -- 870 1,780 3,955 256 1,200 -- Meal sold in 2009 2,380 360 850 -- 3,000 1,360 2,075 263 2,500 832 Meal price in 2008 (CFAf/ton) 111,223 -- 95,000 97,500 95,000 75,775 80,000 90,000 95,000 -- Meal price in 2009 (CFAf/ton) 97,058 80,000 95,000 97,500 80,000 74,775 80,000 90,000 100,000 78,929 Notes: `--` implies data not reported or the answer was not very clear. Wage rate is for unskilled laborers; whenever specified in days or hours, the conversion to monthly wage rate was made by assuming 8 hours per day and 25 days per month. Similarly, when the oil price was reported in tons, it was divided by 55 [1,000/18] to convert it to 20liter can (jc) equivalent. Source: Zanfongnon (2009) from company statistics. Page 51 Because Burkina's smaller seed processing companies produce partly, ra ther than fully, refined oil, concerns have been expressed that their oil is of poor quality and perhaps dangerous for human consumption. SNCITEC, which pro duces fully refined (and high quality) oil with its own label, has complained re peatedly that some of the smaller companies use its label. A USAID (2006) report commented (p. 43) that: "The large processors complain that competition from smaller, more artisanal production units also undercuts them in the market. These firms operate outside the fiscal net, so they typically do not pay taxes and they do not apply the appropriate `textes' (regulations). These firms reportedly use unhygienic production methods and their unrefined oil is dirty, contami nated, and inferior to the vegetable oil of large processors." The pricing of cottonseed bought by the smaller companies in Burkina Fa so is panterritorial and panseasonal--similar to that of seed cotton. During 2008 and 2009, SOFITEX sold cottonseed at CFAf 75,000 per ton net of any taxes; when taxes are included, a ton would cost the companies more than CFAf 90,000.24 The prices charged for oil and meal, however, differed considerably among companies. Meal prices during 2009 varied from a low of CFAf 60,000 per ton to a high of CFAf 110,000 per ton (Figure D1) while oil prices varied from CFAf 8,300 per jc to 13,000 per jc (Figure D2), a range similar to those recorded in Uganda and Tanzania. There are numerous reasons for such price variability, in cluding oil quality, location of company, time of sale, etc. A comparison of the oil and meal prices received by the smaller compa nies with those received by SNCITEC shows, surprisingly, that the large firm's price premium for meal sales is larger than its premium for oil sales. SNCITEC sold its meal at CFAf 103,000 per ton in 2007/08 and 124,500 per ton in 2008/09. The average meal prices received by the smaller companies during these two seasons were CFAf 92,750 per ton and 90,300 per ton, implying that SNCITEC commanded a premium of 11 percent and 38 percent over the smaller companies. Such premium seems justifiable for a number of reasons, including that SN CITEC is an established company using advanced crushing technology and bet ter packaging, and has a larger marketing network. Table D3 (Panel II) reports prices paid for seed by the 21 companies surveyed. When the price 24 was quoted without tax, it was the same for all companies, CFAf 75,000/ton. When it was quoted with tax it was not the same (for example, company IX paid CFAf 92,270/ton while company X paid CFAf 90,270/ton.) It is not clear what explains the difference. Page 52 Figure D1 Burkina Faso: Meal Price for 20 Companies, 2009 (CFAf/ton) 110,000 100,000 90,000 80,000 70,000 60,000 50,000 I II III IV V VI VII VIII IX X XI XII XIII XIV XV XVI XVII XVII XIX XX Note: The numbers represent companies for which data were obtained (from lowest to highest) Source: Interviews with industry representatives (Zanfongnon 2009). 17 Figure D2 Burkina Faso: Oil Prices for 16 Companies, 2009 (CFAf/20liter container) 13,000 12,000 11,000 10,000 9,000 8,000 7,000 I II III IV V VI VII VIII IX X XI XII XIII XIV XV XVI Note: The numbers represent companies for which data were obtained (from lowest to highest) Source: Interviews with industry representatives (Zanfongnon 2009). 18 Page 53 The prices received for cotton oil give a different picture, however. SN CITEC received CFAf 13,400 per jc and 10,850 per jc during 2007/08 and 2008/09 while the smaller companies received CFAf 12,200 per jc and 10,500 per jc during these two seasons, implying premiums of only 10 percent and 3 percent.25 One would have expected SNCITEC to command an even larger premium for oil, given its advantages mentioned above as well as the high quality oil it produces. There are two mutually reinforcing explanations for the company's low oil premium. First, quality is not adequately priced by the consumers and, second, the edible oil market is highly integrated in Burkina Faso, in the sense that im ported oils dictate the price (in the same way that the edible oil markets are inte grated at the world market level, as the discussion in the main text established). This has important implications because it shoes that large cotton seed processing operations using advanced technology, while efficient from a tech nological perspective, may not be economically profitable in the SSA context. Unlike in Uganda and Tanzania, most companies in Burkina Faso did not report electricity problems (in fact, 18 out of 21 managers said that they have not experienced any electricity disruption). Most respondents strongly favored the imposition of high tariffs on imported edible oils as well as export restrictions on oilseeds. The interest rate paid by the companies averaged 13 percent, broadly in line with Tanzania's rates and lower than Uganda's. Lastly, when managers were asked about cotton oil's potential for biofuel use, some responded that they were not aware of such potential, and those that were aware expressed doubts for the profitability of cotton oil as a biofuel feedstock. Half of the respondents said that the cotton oil they produce is consumed domestically while the remainder said that some of their oil is exported to neigh boring countries, including Mali, Niger, and Senegal. About 12 percent of SN CITEC's cotton oil is exported, primarily within the region. Burkina Faso's trade policies on cotton byproducts are less restrictive than Benin's; in 2005/06 it banned exports of cottonseed, and it also imposed an 18 percent value added tax. Most of Burkina Faso's cottonseed is consumed domestically by the seed processing industry. An estimated 1 percent is used unprocessed as animal feed. There have been reports that an export ban on cottonseed was introduced in 2005/06. However, statistics indicate that during this season Burkina Faso ex ported an estimated 81,643 tons, equivalent to 20 percent of the country's cotton seed output, while similar level of exports were recorded for the subsequent two seasons (Bayoulou 2010). On the import side, it appears that some cottonseed is In table D2, SNCITEC's price is reported in tons; it has been converted to 20liter container by 25 dividing by 55. A similar conversion was made in table D3 for the smaller companies. Page 54 imported from the Northern regions of Côte d'Ivoire (the highest recorded figure was 30,000 tons in 2002/03). Figure D3 depicts the contribution of cottonseed to the total value of cot ton over the past decade. The contribution (calculated the same way it was calcu lated for Benin) averaged 8 percent, between 1999/2000 and 2006/07, and it in creased to 19 percent for the past two seasons. This sharp rise is very similar to that in Benin and consistent with what took place in world markets. Figure D3 Burkina Faso: Seed's Contribution to Cotton's Value (percent) 20% 16% 12% 8% 4% 0% 1999/2000 2000/01 2001/02 2002/03 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 Note: This is the ratio of the price of cotton seed over the sum of the grower's price and ginning costs Source: Author's calculations based on industry data 20 Similarly, cottonseed prices in Burkina Faso covered a little more than one quarter of actual ginning costs up to 2006/07, while during the past two seasons their share has more than doubled (Figure D4). Unlike in Benin, their share in creased slightly after the smaller companies entered the cottonseed processing industry. Although the increase is too slight to be viewed as a statistically solid result, nevertheless it is consistent with the expectation that competition drove seed prices up. Page 55 Figure D4 Burkina Faso: Seed Value to Ginning Costs Ratio (percent) 70% 60% 50% 40% 30% 20% 10% 0% 1999/2000 2000/01 2001/02 2002/03 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 Note: This is the ratio of the price of cotton seed over the ginning costs Source: Author's calculations based on industry data 21 Page 56 APPENDIX E: UNITED STATES The origins of the US cotton industry date from the 17th century when European settlers began experimenting with cotton cultivation. The expanded use of the saw gin--which was invented in 1793 and made the process of separating lint from seed commercially viable--rendered cotton a chief cash crop. Within a few decades, the US became the world's dominant cotton supplier, accounting for more than half of global cotton output, which was mostly exported to England to supply its textile industry. Apart from a short hiatus during the Civil War, the US kept its dominance in the cotton market well into the 20th century. In 2008/09 the US was the world's third largest cotton producer with 2.7 million tons of lint (following China with 8 million tons and India with 4.9 mil lion tons), down from five million tons during the mid2000s. World production of lint was 23.4 million tons. The US is still the world's dominant cotton exporter, accounting for 45 percent of the global trade of 6.4 million tons. The US cotton sector has benefited from large budgetary transfers that were instituted during the farm programs of the 1930s. It received an estimated US$2.2 billion of assistance during 2009 and is expected to receive US$2.4 billion in 2010, according to the US Department of Agriculture's February 2010 update. The transfers, which by some accounts have been a key reason for the slow progress of the Doha Development Agenda, have been the subject of two WTO (World Trade Organization) cases. In 2002 Brazil began its WTO case against the US, arguing that US cotton subsidies lowered world prices and thus lowered ex port revenues from cotton. In 2003 four WCA countries--Benin, Burkina Faso, Chad, and Mali, often referred to as Cotton4 or C4--launched the Initiative in Favor of Cotton, in effect demanding compensation for their export revenue losses caused by lower cotton prices resulting from subsidies. (See Baffes (2007) and Sumner (2006) for detailed accounts of these two WTO cases). Cottonseed processing in the US goes back to 1799 when a patent was granted for extracting oil from cottonseed. Cotton byproducts were commercia lized during the early 1830s when the first cottonseed processing facilities were built (Nixon 1930). Cotton meal was used as animal feed and on some occasions as fertilizer. Cotton oil was the key vegetable oil in the US until it was displaced by other vegetable oils (mainly soybean) during the 1950s. US cotton seed production reached a maximum of 8 million tons in 2005 (Figure E1). However, it declined to almost 4 million tons in 2008 and 2009, main ly because of the diversion of land to biofuel commodities (especially corn), and low world cotton prices (see discussion in main text). Page 57 Figure E1 Cotton Seed Production in the US (000 tons) 10,000 8,000 6,000 4,000 2,000 0 1970 1974 1978 1982 1986 1990 1994 1998 2002 2006 Source: US Department of Agriculture 21 Cottonseed prices in the US reflect the interplay of numerous factors, in cluding the size of the cotton crop, prices of competing edible oils (Figure E2), prices of competing meals (especially soybean meal), and prices of other feed grains such as corn. These factors in turn determine how much cotton seed will be used by the dairy industry in unprocessed form and how much will go for crushing. During the 2007/08 season for example, 6.5 million tons of cottonseed were produced in the US, of which 3.1 million was consumed by the diary indus try, 2.7 million tons was crushed, while 0.6 million tons was exported, a pattern that has been representative of the past ten years. 26 Nominal prices of cottonseed fluctuated around US$100 per ton up to 2006, but they have averaged US$180 per ton during the past three years, reaching a high of US$220 per ton in 2008 (Table E1). The dairy industry uses unprocessed cotton seeds because of their unique protein, energy, and 26 fiber content. However, dairy cows can only consume certain amounts of cotton seeds due to the presence of gossypol, a naturally occurring plant pigment found in the seed. Cotton seeds contain between 0.4 and 2 percent gossypol, depending on speciesvariety and growing conditions. When the cotton seeds are processed, the level of gossypol declines to 0.10.2 percent (Blasi and Drouil lard 2002). The effects of gossypol on animal health have been studied extensively and guidelines have been developed (Martin 1990). Moreover, research under way in the US indicates that cot ton varieties will be developed soon with low levels of gossypol while maintaining all other cha racteristics of the cotton plant (ICAC 2010). Page 58 Table E1 Value Composition of Cotton in the US, 19702009 Nominal prices (US$/kg) Value (US$/ton of seed cotton) Contribution (%) Year Lint Seed Lint Seed Total Lint Seed 1970 0.63 0.06 221 37 258 85.8 14.2 1971 0.74 0.06 259 37 296 87.5 12.5 1972 0.79 0.05 278 32 310 89.6 10.4 1973 1.36 0.10 474 65 539 87.9 12.1 1974 1.42 0.14 495 88 583 84.9 15.1 1975 1.16 0.10 406 63 469 86.6 13.4 1976 1.69 0.10 592 67 659 89.8 10.2 1977 1.55 0.07 544 46 590 92.2 7.8 1978 1.57 0.11 550 74 624 88.1 11.9 1979 1.69 0.12 591 79 670 88.3 11.7 1980 2.05 0.13 716 84 800 89.5 10.5 1981 1.85 0.09 646 56 702 92.0 8.0 1982 1.60 0.08 559 50 609 91.8 8.2 1983 1.85 0.17 649 108 757 85.7 14.3 1984 1.79 0.10 625 65 689 90.6 9.4 1985 1.32 0.07 461 43 504 91.5 8.5 1986 1.06 0.08 370 52 422 87.7 12.3 1987 1.65 0.08 577 54 631 91.5 8.5 1988 1.40 0.12 490 77 566 86.5 13.5 1989 1.67 0.11 586 68 654 89.6 10.4 1990 1.82 0.12 637 79 715 89.0 11.0 1991 1.68 0.07 587 46 633 92.7 7.3 1992 1.28 0.10 447 63 511 87.6 12.4 1993 1.28 0.11 448 73 522 85.9 14.1 1994 1.76 0.10 617 66 683 90.4 9.6 1995 2.13 0.11 745 69 814 91.5 8.5 1996 1.77 0.13 621 82 703 88.3 11.7 1997 1.75 0.12 612 79 690 88.6 11.4 1998 1.44 0.13 506 84 590 85.8 14.2 1999 1.17 0.09 410 58 468 87.6 12.4 2000 1.30 0.11 456 68 524 87.0 13.0 2001 1.06 0.09 370 59 429 86.3 13.7 2002 1.02 0.10 357 66 422 84.5 15.5 2003 1.40 0.12 490 76 566 86.6 13.4 2004 1.37 0.11 478 70 548 87.3 12.7 2005 1.22 0.10 426 62 488 87.2 12.8 2006 1.27 0.11 443 72 515 86.0 14.0 2007 1.40 0.16 488 105 594 82.3 17.7 2008 1.57 0.22 551 145 696 79.2 20.8 2009 1.38 0.16 484 103 587 82.4 17.6 Notes: It is assumed that one ton of seed cotton gives 350 kgs of lint and 650 kgs of cottonseed. Source: Author's calculations from US Department of Agriculture data. Page 59 Figure E2 US Prices of Fats and Oils (Nominal, US$/ton, 2006/09 average) Tallow Lard Soybean Corn Canola Cotton Sunflower Peanut 400 600 800 1,000 1,200 1,400 1,600 1,800 Source: US Department of Agriculture 22 In a study of demand for cottonseed by the US dairy industry, Argüello (2008) found that cotton meal has an ownprice demand elasticity of 0.41 but much higher crossprice elasticities from competing feed grains and meals (some even greater than unity), implying that it is these substitute and complement products that matter most. This should not be surprising since the US and global market for feed grains is much larger than the cotton meal market and therefore, it is the demand and supply conditions of these markets that drive prices of cot ton byproducts (similar to the small size of the cotton oil market compared to the global edible oil markets discussed in the main text). During the past 40 years the contribution of cottonseed to the total reve nue from cotton has fluctuated considerably, from a low of 7.8 percent in 1977 to a high of 20.8 percent in 2008, evaluated at the US average ginning outturn ratio of 35 percent (see Figure E3). The average contribution in the period up to 2006 (i.e., excluding the recent boom) was 11.6 percent while the average in the past three years was 18.7 percent. This rise is remarkably similar to that in the global cotton oil/lint price ratio, discussed in the main text. Typically cotton growers in the US expect the value of byproducts to cover their ginning costs. Unlike in most other countries (including the four SSA countries reviewed here), where Page 60 growers sell their seed cotton to ginning companies, US cotton growers use gin ning services on a fee basis thus allowing them to retain ownership of the lint and seed after ginning. US retail consumers (and in other countries) prefer cer tain vegetable oils over others. Cotton oil, being one of the lesspreferred oils, is not marketed as `cotton oil' but generically as `vegetable oil.' Paradoxically, the price of cotton oil (one way of gauging how much consumers value it) is higher than those of a number of edible oils such as corn oil that are sold at retail under their own names, (see Figures 6 and E2 for world domestic US prices). A representative from the US food industry noted the following: "The consumer attaches value to some oils--peanut oil tastes better and fries well, ca nola and sunflower oil tout health benefits. When there is no discernable con sumer reaction to the label it is easier to use a generic brand. This would allow a processor to include the best available oil at the lowest cost to the consumer and reduce the expense of constantly reworking the main elements of the label. Labeling laws enter into the picture as well." (Interview, April 22, 2010.) Cotton oil's poor reputation in the US goes as far back as the 1850s when cotton oil was mixed with domestically produced animal fats. One representative of the cotton oil industry has been quoted as saying in 1857: "We dare not call it cotton seeds oil lest it might prejudice the sale" (Nixon 1930, p. 77). Figure E3 US Cotton Seed's Share in Value Terms (percent) 21% 18% 15% 12% 9% 6% 3% 0% 1970 1974 1978 1982 1986 1990 1994 1998 2002 2006 Source: Author's calculations based on US Department of Agriculture data 23 Page 61 APPENDIX F Questionnaire for Cottonseed Processing Companies I. General questions 1. Name of operation, place, cotton region(s) it receives seeds 2. Size of operation (# of permanent and temporary employees, managers, en gineers) 3. Wage rate of daily laborers and conditions of employment (i.e., how long is the employment contract, lunch or any other benefits provided, etc.) 4. Does the operation require working capital? If yes, does it have access to credit? Under what terms (e.g., interest rate, duration of loan, amount, etc.)? 5. How old is the machinery and equipment of the crushing facility? 6. How many years is the facility in operation? 7. Ownership structure (individual, company, cooperative, other) 8. Is the operation linked to a ginnery? If yes, which one? If not, is it part of a major operation or is it a free standing operation? 9. Total crushing capacity of the facility (e.g., tons of cottonseed per day) 10. Does the facility operate during cotton harvest only or does it go beyond that period (or perhaps it operates for a shorter period by starting later)? 11. Do they crush other types of oil seeds or just cottonseed? 12. Should there be any restrictions on the imports of other competing edible oils in order to keep cotton oil prices higher? II. Cottonseed 13. Actual crushings for 2008 (and earlier years, if known) 14. Estimated (or actual) crushings for 2009 15. Who delivers the seeds (farmers, traders, middlemen, ginner representa tive?) What is the share for each group? 16. How many kgs is every seed delivery (average)? 17. Are there quality issues/problems with respect to seeds (e.g., seeds which have less oil content or lower quality oil)? 18. Does the oil facility have its own buyers that buy seeds from ginneries? 19. What is the average price paid for seed (as many years as possible)? 20. Does the price paid for seeds depend on the size of the delivery? 21. Is the price the same throughout the season? If not, what is the highest and lowest price? Page 62 22. Are there any longterm contracts between ginneries and cotton processing companies or are all transactions spot? 23. Are payments made in cash always or are deliveries made on credit? III. Cotton oil 24. What is the average quantity sold per transaction? 25. What is the average sale per day? 26. Does the operation keep stocks or is all cotton oil sold after crushing? 27. Who buys the oil (individual consumers, companies, middlemen, traders)? 28. To the best of your knowledge, where is the oil consumed (locally, regional ly, exported)? 29. Is the quality of the oil consistent or it depends on where the seeds come from? 30. Is all cotton oil going for human consumption or some goes to industrial uses? 31. Is there a standard packaging process (e.g., a 20kg plastic container)? 32. Is cotton oil sold in bulk quantities? 33. What is the average price received for oil (as many years as possible)? 34. Does the price received depend on the size of the transaction? 35. Is the cotton oil price the same throughout the season? If not, what is the highest and lowest price? 36. Are all purchases made on cash or are there credit transactions? 37. Does the cotton oil go through any treatment or bleaching process? 38. Is there any discussion/idea of using cotton oil as feedstock to biodiesel? IV. Cotton meal 39. How much cotton meal did the operation sell? 40. Was it sold in bulk? 41. Who were the buyers? 42. What is the average price received from the cotton meal? Page 63