(a) Whole Farm Costs Are Irrelevant 141. The U.S. argument that whole-farm costs are relevant to upland cotton production decisions227 is simply not logical. Why would a farmer cross-subsidize the production of upland cotton – in effect subsidizing foreign purchasers of low-priced/high-cost U.S. upland cotton – with income from other crops or from off-farm income? The United States does not answer that question. Instead, it highlights studies which find that farmers increasingly supplement their income with off-farm work.228 Brazil does not dispute these studies or this trend. Yet, the fact that U.S. upland cotton farms earn income from growing other crops or in other sectors of the economy is not relevant to these proceedings. As the original panel found, "[w]e are considering costs and market revenues in respect of upland cotton. Our examination is one of the upland cotton industry … We are not looking into the possibility of cross-subsidization or cross-financing of insufficient market revenues for upland cotton that may have come from other United States industries."229 Moreover, as the studies submitted by the United States point out, the issue of off-farm income is of particular importance for small farms.230 And as these studies also point out, most U.S. upland cotton production occurs on farms that have very big commercial operations.231 The U.S. argument is, therefore, not even relevant to the vast majority U.S. upland cotton production.
(b) MY 2003 Cost Data Is Atypical
142. Much of the distorted U.S. analysis of "costs" rests on its reliance on an unusual year – MY 2003.232 This year was unusual because it was the first – and only time – in the history of the FSRI Act of 2002 (and the first time in eight years) that U.S. farmers actually earned a profit ($65 per acre) based solely on market revenue. The United States understandably focuses its arguments on MY 2003 cost of production data in its Rebuttal Submission and subsequent submissions.233 In doing so, however, it fails to acknowledge several important points.234
143. First, the U.S. MY 2003 cost data is not materially different from what Brazil has used during this entire proceeding. The only difference is that it is broken down into three arbitrary cost groups – low, medium and high.235 When all cost groups are aggregated, the average acre planted to upland cotton made a profit of $65. Brazil's assertions regarding the seven-year $837 loss on each acre planted to upland cotton is based exclusively on USDA cost data and includes the average profit of $65 per acre achieved in MY 2003.236 Therefore, the United States assertion that "Brazil ask[ed] the Panel to simply ignore this data"237 is incorrect. Far from ignoring the data, Brazil used the sameUSDAdata in its First Written Submission, Rebuttal Submission and Oral Statement.238
144. Second, the United States failed to acknowledge that MY 2003 is the only year in the past eight in which the U.S. producers covered their total costs of production only with market revenue. In its response to this question, the United States has finally acknowledged that "prices in MY 2003 were high."239 However, the United States now argues that that most producers nevertheless would have covered their total costs in MY 2004 and 2005 because of, inter alia, increasing yields.240 USDA data squarely contradicts this statement: the data clearly show that the average acre planted to upland cotton lost $18 in MY 2004 and $88 dollars in MY 2005.241
145. Third, U.S. producers' costs are increasing significantly in MY 2006. National average yields are estimated to be 668 pounds per planted acre in MY 2006, far lower than yields of 806 pounds per planted acre in MY 2004 and 799 pounds per acre in MY 2005.242 Further, U.S. producers are faced with significantly higher seed, fertilizer, and fuel costs in MY 2006 than in MY 2003-2005.243 Total costs in MY 2006 are forecasted to be 10 percent higher than their MY 2003-2005 average.244 With market prices remaining low throughout the first eight months of MY 2006245, the loss based on market revenue is expected to significantly widen in MY 2006 compared to MY 2003-2005.246
146. Fourth, the United States makes the remarkable assertion that "there is no basis to assume that the [MY 2003] results would have been substantially different in later years"247 However, unlike MY 2003, when the average U.S. farm price was $0.62 per pound248, the average farm price was $0.416 per pound in MY 2004 and $0.477 per pound in MY 2005.249 The current average farm price for MY 2006, as of April 2007, is $0.472 per pound.250 Thus, there is no way that the average "mid cost" producer, with total costs of $0.57 per pound, covered his or her total costs of production in those years. Nor will those costs be covered in MY 2007.251 147. In any event, breaking upland cotton farms into arbitrary cost groups is neither necessary nor particularly relevant. It is far more appropriate for this compliance Panel, like the original panel, to examine national average cost of production data to get a complete picture of the U.S. upland cotton industry. Indeed, the Appellate Body in Canada Dairy (21.5 II) found that
[i]t, therefore, seems to us that the benchmark should be a single, industry-wide cost of production figure, rather than an indefinite number of cost of production figures for each individual producer. The industry-wide figure enables cost of production data for producers, as a whole, to be aggregated into a single, national standard that can be used to assess Canada's compliance with its international obligations.252
148. Using national average data certainly does not "obscure" cost of production data as the United States suggests.253 Brazil, the United States and the compliance Panel are all aware that upland cotton farmers have different cost structures. In any given year, about half of the acres planted to upland cotton have costs that are higher than the national average, while the other half has costs that are less than the national average. This is nothing more than basic statistics. It follows, therefore, that in years when the average acre planted to upland cotton lost money (six of the past seven years) more than half of the acreage planted to upland cotton suffered a loss. In this context, it is import to recall that Professor Sumner's analysis suggests that 17-19 percent of U.S. production of upland cotton would cease without marketing loan and counter-cyclical subsidies.254 This is consistent with the notion that only higher-cost U.S. producers would exit upland cotton production without these subsidies.
149. Thus, the various U.S. cost-related arguments based on MY 2003 data are entirely misplaced. The reference period of MY 2005 demonstrates significant losses by the average acre of upland cotton of $88 per acre. These losses are set to increase to an estimated $116 per acre in MY 2006.255 These are significant losses and above the average $65 per acre loss per year over the lifetime of the FSRI Act of 2002.256 (c) USDA's Total Cost of Production Data is Relevant and Appropriate 150. For the first time in this proceeding, the United States attacks the total cost of production data itself, arguing that it "provides a stylized, abstract view"257 and is used by Brazil in a way in which "it was never intended."258 These criticisms echo similar arguments made by the United States in the original proceeding.259 These old U.S. arguments attacking its own USDA cost data were properly rejected by the original panel.260 The renewed arguments should again be rejected by this compliance Panel.
151. The United States criticism of USDA's own total cost data is two-fold.261 First, the U.S. criticizes the method by which USDA values non-cash costs, such as imputed farm labor and land ownership costs.262 Second, the United States argues that USDA's total cost data is not an appropriate indicator of the financial viability of upland cotton farming because it includes non-cash costs.263 Both criticisms of USDA's cost data are without merit.
152. The United States challenge to USDA's analysis of non-cash costs squarely contradict basic economic and accounting principles. Non-cash costs, such as ownership costs and opportunity costs, are real, not imaginary, costs, as the U.S. suggests. Indeed, USDA explains that:
[c]ommodity costs and returns include estimates of both cash expenditures and noncash costs. Cash expenditures are incurred when factors of production are purchased or rented. Noncash costs occur when factors are owned. For example, if a farmer fully owns the land used to produce corn, he/she would have no expenditure for land rental or for loans to pay for the purchase of land. Yet, aneconomiccost arises. Byowningthe land andusingitto grow corn, thefarmer foregoes incomefromother uses oftheland, such asrentingittoanotherproducer. Thesecostscomeaboutbecauseproductionresourcesarelimitedandhavealternativeuses. If a farmer uses savings to pay for operating inputs, such as seed, fertilizer, chemicals, and fuel, and thus pays no interest on operating loans, the farmer still incurs an economic cost because the savings could have earned a return in another use. Likewise, the farmer has an opportunity cost of his/her labor used in the production of the commodity because it could have been used on another farm or in off-farm employment.264
153. The need to cover total costs of production, including both cash and non-cash costs, is an economic reality facing all farmers. No farmer would or could repeatedly plant his fields with a chronically loss-making crop. There is no basis for the U.S. assumption that farmers do not take into account non-cash "opportunity" costs when making production decisions. This assumption is directly contradicted by USDA economists who include both cash and non-cash costs in their cost and returns estimates.
154. The need for producers to cover total costs of production was recently highlighted by the Appellate Body in EC–Sugar. The Appellate Body found that
in the ordinary course of business, an economic operator makes a decision to produce and sell a product expecting to recover the total cost of production and to make profits. Clearly, sales below total cost of production cannot be sustained in the long term, unless they are financed from someothersources.265
155. The "some other sources" in this case that cover total costs of production are marketing loan and counter-cyclical subsidies. Without such subsidies, average seven-year losses totalled $837 per acre. With these two subsidies, the average acre showed a seven-year profit of $104 per acre.266
156. As noted by the Appellate Body, not only is an economic operator expected to cover his or her total costs of production, but he or she is also expected to make a profit. Brazil notes that USDA also recognizes the need for farmers to make a profit. USDA defines "residual return to management and risk" as follows:
[r]esidualreturnstomanagementandrisk is the difference between the gross value of production and total economic costs. The return to management and risk indicates the extent to which longrun production costs are covered by production valued at average harvest-month prices.267
157. The Appellate Body's findings in EC–Sugar and USDA's definition of residual returns to management and risk demonstrate the importance of not only covering costs, but also of generating profits from farming. The ability of upland cotton farmers to cover their variable costs, let alone total costs, is not sufficient to show that marketing loan and counter-cyclical subsidies do not impact production.
158. Brazil's cost of production analysis has not previously factored in the obvious need to make profits on the farmers' investments. This is because the huge losses based only on market revenue readily demonstrate the decisive role played by marketing loans and counter-cyclical subsidies in sustaining revenue and large amounts of upland cotton production. But if the compliance Panel were to consider the need for farmers to generate some reasonable profit, the gap between market returns and such a profit margin obviously would be even greater. For example, assuming even a fairly low profit margin of only 5 percent, the average gap between market revenue and the five percent profit would increase to $1,017 per acre in the seven year period of MY 1999-2005. For MY 2005 alone, the gap would be $115 per acre – not the $88 per acre based on a break-even methodology.268 The projected gap in MY 2006 would be $144 per acre, based on a five percent profit margin.269
159. Next, the new U.S. criticism of USDA's valuation methodologies ignores the fact that they were conceived by the American Agricultural Economics Association and implemented by USDA.270 The data are based off of surveys that are generated in the ordinary course of USDA business.271 The methods and data used to generate cost of production data are objective and appropriate. USDA explains that four approaches are used to estimate commodity costs, including (1) direct costing, (2) valuing input quantities, (3) indirect costing and (4) allocating whole-farm expenses.272 The United States singles out only the fourth element for its criticism – not the other three.
160. Brazil notes that "allocating whole-farm expenses"273 is only used by USDA to determine "general farm overhead" and "taxes and insurance."274 These costs account for about 5 percent of upland cotton producers' total costs of production.275 The other valuation methodologies used by the USDA cost survey are also properly used to address the question before this compliance Panel. USDA explains that "[u]npaid labor hours are valued using an estimate of the wages earned off-farm by farm operators"276 and that "[l]and is valued according to the average cash rental rate for land producing the commodity in the particular area."277 These relatively straightforward methodologies in no way distort or undermine an assessment of the financially viability of upland cotton farming using total costs of production.
161. Finally, the United States criticizes Brazil's use of USDA total costs of production data by highlighting the fact that the total cost of production was greater than market returns for nearly all U.S. field crops in MY 2005. Based on this fact, the U.S. argues that Brazil's planting decision analysis is "absurd"278 and "fundamentally erroneous."279 Neither Brazil nor the U.S. Congressional Research Service ("CRS"), an agency of the U.S. Government, would agree with this latest U.S. assertion. In fact, CRS confirms that "it is only with the aid of subsidies that a substantial portion of U.S. production [of all crops] is made economically viable."280 CRS's analysis relied upon a graph of total costs versus market returns in selected periods281 with and without subsidies as set out below:
Market Returns igure 5 – Excerpt from CRS Report to Congress282
Source: Calculated by CRS from USDA data.
Soybeans Corn Sorghum Peanuts Wheat Cotton Rice 62. Based on this chart calculated from USDA data, CRS concluded that "the substantial contribution of subsidies toward covering otherwise unmet production costs implies a high chance for adverse rulings for any of the major covered commodities."283
163. Indeed, data on farm income from USDA's Economic Research Service ("ERS") shows that, in MY 2005 as well as in previous years, the U.S. farm sector as a whole, i.e., including crop and livestock production, has been viable based on an assessment of costs, farm-related revenues and government subsidies.284 As the ERS Agricultural Income and Finance Outlook of November 2006 points out, this is the case primarily due to record subsidy payments of $24.3 billion in MY 2005, with a significant portion of these payments received from "ad hoc and emergency programs".285
164. USDA data shows a profit of $222 million in MY 2005 from the production of soybeans, corn, sorghum, peanuts, wheat, upland cotton and rice – taking into account total costs, market returns and revenue from the marketing loan, counter-cyclical and direct payment program as well as crop insurance subsidies.286 Production of certain crops in MY 2005, including upland cotton, resulted in losses based on market revenue alone.287 Citing low prices in MY 2005, the United States government provided additional very high emergency payments amounting to $3.2 billion.288 These additional subsidies are not accounted for in the above assessment of the overall financial viability of these seven crops.
165. Thus, the United States is incorrect when asserting that "U.S. farmers would have lost money across the board from producing any of these crops".289 166. To allow a more comprehensive assessment of the role of U.S. subsidies in ensuring the financial viability of U.S. program crop production, Brazil also provides the compliance Panel with data for the other years under the FSRI Act of 2002. Between MY 2002 and 2005, the gap between total costs and market returns for soybeans, corn, sorghum, peanuts, wheat, upland cotton and rice was $43 billion.290 Marketing loan, counter-cyclical, direct and crop insurance subsidies to the same crops amounted to $53 billion, providing a $10 billion offset to shortfalls in revenue.291 In addition, U.S. farmers received large emergency and disaster payments in totaled $8.5 billion over the MY 2002-2005 period.292 This evidence demonstrates that large portions of U.S. program crop production, including upland cotton, are profitable only due to large U.S. subsidies.
167. In sum, in view of tens of billions in various forms of subsidies provided to program crops under the FSRI Act of 2002, CRS's and Brazil's conclusion reached based on USDA's cost and returns data are neither absurd nor unexpected. The United States' claim that Brazil's and CRS's methodology of assessing USDA's cost and return data would lead to the conclusion that all of U.S. agriculture is not viable even with subsidies is simply wrong. The USDA data discussed above shows that certain sectors of U.S. agriculture are not viable but for U.S. government subsidies. It also confirms Brazil's analysis293 and the findings of the original panel294 with respect to the profitability of growing upland cotton in the United States.
60. In its Rebuttal Submission, the United States argues that Prof. Sumner's description of the model that appeared in a recent CATO publication is not "appropriate" for use in a WTO dispute involving claims of serious prejudice. Professor Sumner has since introduced "more empirical and institutional detail" to the model used in this dispute. These changes are described in paragraphs 111 117 of Brazil's Opening Statement. Does the United States view these changes as being sufficient to make the model appropriate for use in a WTO dispute involving claims of serious prejudice? If not, what modifications does the United States think should have been made to the model? 168. Brazil also notes that, in response to this question, the United States once again repeats its summary and general criticism of Professor Sumner's simulation model. However, the United States does not even attempt to rebut any of the specific arguments, explanations and rebuttals submitted by Brazil and Professor Sumner in the many submissions since the beginning of this proceeding.295 2. Increase in world market share - Article 6.3(d) of the SCMAgreement Questions to the United States 75. Could the United States explain further the textual basis of its argument that "Article 6.3(d) is not concerned with absolute market share and whether or not in any given year a member's market share would have been lower if subsidies were removed"? (Rebuttal Submission of the United States, para. 401)
169. The U.S. response provides no clear guidance with respect to the textual basis for its argument regarding the first element of Article 6.3(d), i.e., whether there has been an increase in the world market share for the year in question compared to the previous three-year average. Nor does the United States elaborate, either in this response or in any of its other submissions, on what it believes to be the required proof with respect to the first element of Article 6.3(d). Rather, the United States vaguely asserts that "what is at issue is movement, not something static like the absolute level of market share in a particular year."296 The United States then concludes that "the upward movement  itself [must be] proven to be 'the effect' of a challenged subsidy."297
170. It is difficult to respond to such vague arguments. Yet, using the terminology in the U.S. response, Brazil has demonstrated that the "movement" (i.e., the increase in MY 2005 compared to the previous three year average of MY 2002-MY 2004) was the effect of marketing loan and counter-cyclical subsidies. Brazil proved this by showing that butfor the marketing loan and counter-cyclical subsidies, the U.S. world market share in MY 2005 would not have increased, but would have decreased compared to the MY 2002-MY 2004 three-year average.298 This is not a "static" analysis, as asserted by the United States.299 Rather, this assessment constitutes a dynamic comparison between two time periods demonstrating that it was the effect of the two challenged subsidies that led to the increase in the U.S. world market share in MY 2005 compared to its previous three-year average.300 171. Finally, Brazil refers the compliance Panel to its response to question 76, in which it addresses in detail the absence of any textual basis for the U.S. argument regarding the second element of Article 6.3(d) – the consistent trend. The U.S. response to question 75 provides no additional textual basis or argument to support its new and erroneous interpretation of the "consistent trend."
C. Claim of Brazil regarding threat of serious prejudice