1. Brazil has failed to establish all of the elements necessary to establish its subsidies claims 1. For the Panel to make the findings Brazil requests, Brazil must adduce evidence and arguments sufficient for the Panel to identify the product(s) that a particular subsidy benefits. This requirement of identification can come up in a variety of ways, but the two most frequently encountered questions are: Which product(s) benefits from the subsidy? and Should the benefits of a subsidy be allocated to future production and sales of the product in question, or should such benefits be "expensed" – that is, allocated only to current production and sales during the time period in which the subsidy is received? Brazil has not provided a basis for a clear and unambiguous explanation on its conclusions for each of these points in order for the Panel to fulfil
its obligations under Article 12.7 of the DSU.
2. With respect to the first question – which product(s) benefits from the subsidy? – Annex IV to the Subsidies Agreement provides guidance. Annex IV provides guidelines for calculating total ad valorem subsidization for purposes of the now expired Article 6.1(a). A subsidy not "tied to the production or sale of" cotton ("a given product") cannot be regarded as subsidizing merely "that product"; rather, the subsidy benefits all of the "recipient firm’s sales". In the Negotiating Group on Rules, Brazil has proposed that Members adopt a "guideline" on calculating the amount of the subsidy precisely along these lines.
3. Implicit in both paragraphs 2 and 3 of Annex IV is the principle that a subsidy provides a benefit with respect to products that the recipient produces. A corollary of this principle is that a subsidy does not provide a benefit with respect to products that the recipient does not produce. Thus, a subsidy provided to a recipient who does not produce upland cotton cannot be said to provide a benefit to upland cotton. Such a subsidy cannot be regarded as having one of the effects described in Article 6.3 insofar as upland cotton is concerned.
4. The foregoing analysis suggests that, for each challenged subsidy, Brazil must identify (as would the Panel in its report) the product that benefits. In the case of product specific support – that is, a payment that is linked to production of a specific product – such as the marketing loan payments and Step 2 payments, the issue is not difficult. In the case of a payment in which the subsidy is not "tied to the production or sale of a given product", the product subsidized by that payment is all the products produced by the recipient. To determine the portion of a payment not tied to the production or sale of a given product that benefits upland cotton, the value of the payment must be allocated over the "total value of the recipient firm’ s sales".
5. With respect to the second question – how should subsidies be allocated over time? – Annex IV also provides guidance. Paragraph 7 provides that: "Subsidies granted prior to the date of entry into force of the WTO Agreement, the benefits of which are allocated to future production, shall be included in the overall rate of subsidization" (emphasis added). A corollary of this principle – that the benefits of certain subsidies should be allocated to future production – is that if subsidy benefits are notallocated to future production, they must be expensed – that is, allocated to production in the time period during which the subsidy is received. Thus, in the context of this dispute, a subsidy the benefits of which are expensed to production/sales in 2001 cannot be said to be causing serious prejudice in 2002 because the subsidy has ceased to exist. The "benefit" – one of the constituent elements of a "subsidy" under Article 1 – was used up in 2001. Once the benefit was exhausted, the subsidy ceased to exist.
6. The Subsidies Agreement does not expressly identify those subsidies "the benefits of which are allocated to future production". However, guidance is available on this question, and it suggests that subsidies that are "non recurring" should be allocated over time, while subsidies that are "recurring" should be expensed to the year of receipt.149 For example, the Informal Group of Experts recommended to the Subsidies Committee that, as a general proposition, recurring subsidies be expensed and non recurring subsidies be allocated. The Group also specifically recommended that price support payments generally be expensed. In making these recommendations, the Group follows the logic noted above: where there are not reasons to allocate subsidy benefits to future production, the subsidy must be expensed, and once the benefit was exhausted in the time period during which the subsidy is received, the subsidy ceased to exist. The analysis presented above and the conclusions and recommendations of the Group are not controversial. The domestic countervailing duty regulations of various Members, including those of Brazil and the European Communities, reflect this very approach.
7. Thus, it is appropriate for the Panel to expense the value of these payments – that is, allocate them to production in the time period during which the subsidy is received. No payment at issue is made for the acquisition of fixed assets. Rather, the challenged payments are recurring. Brazil’s own arguments endorse the notion of expensing these payments. That is, for purposes of its Peace Clause arguments, Brazil expenses these payments by allocating the total value of each of these payments to the marketing year for which the payment is received. For purposes of Brazil’s actionable subsidies claims, Brazil adopts the identical approach and expenses these payments to the marketing year for which the payment is received. Thus, despite Brazil’s silence on the issue of expensing recurring subsidies, its actions and arguments reveal that it accepts and applies the concept to the challenged US subsidies.
8. The United States has explained, and Brazil tacitly accepts, that the payments challenged in this dispute are recurring subsidies that 0are expensed – that is, allocated to production in the time period during which the subsidy is received. It follows that a recurring subsidy provided in marketing years 1999, 2000, or 2001, respectively, cannot be said to be causing serious prejudice in marketing year 2002. Because the payments in each of those prior years was allocated to production in those years, no "benefit" exists after each of those years – a benefit could only exist in a subsequent year if the payment had been allocated to future production and not expensed.
9. Because the recurring subsidies provided in each of marketing years 1999, 2000, and 2001 ceased to exist when the benefit was used up for production in those years, the effect of those subsidies cannot be the subject of subsidies claims in marketing year 2002. Under Article 5(c) and 6.3, Brazil must demonstrate what "the effect of the subsidy is". Similarly, under GATT 1994 Article XVI:3, Brazil must demonstrate that the United States grants or maintains export subsidies "which operate to increase the export of any primary product," resulting in a more than equitable share of world export trade. Subsidies that were expensed and benefited historical production in marketing years 1999, 2000, and 2001cannot also benefit current production. Thus, these past payments would not form part of Brazil’s subsidies claims nor the Panel’s analysis. Serious prejudice has to be based on findings for the 2002 marketing year.
2. Brazil's legal interpretive errors also demonstrate that it has failed to make a prima facie case on its subsidies claims 10. As complainant Brazil must identify properly the measures within the Panel’s terms of reference – that is, "subsidies provided to US producers, users and/or exporters of upland cotton" in respect of upland cotton, the subsidized product. The challenged measures are subsidies, or payments, and in order to assess their effect, one needs to know, inter alia, how large the subsidy is. Brazil has not properly identified the size of each challenged subsidy.150
11. Brazil Misinterprets Article 6.3(c) on Price Suppression or Depression. Brazil has not alleged any facts to establish that US and Brazilian cotton are found "in the same market" pursuant to Article 6.3(c) – that is, in each of the markets identified by Brazil (at various times, the United States, Brazil, Argentina, Bolivia, India, Indonesia, Italy, Paraguay, Philippines, and Portugal). Brazil has not identified the extent of subsidization of the US cotton in each market (the subsidized volume) – that is, which exports benefit from which challenged subsidy. Brazil has also not shown a price suppressing effect by those US imports in each market. Brazil simply asserts that prices in those markets are correlated to the NY futures and A index prices. This allegation of a generalized price effect cannot satisfy Brazil’s burden of showing a price effect by the subsidized product of a like product of another Member "in the same market".
12. Brazilian price quotes in fact consistently undercut US price quotes for delivery CIF Northern Europe. It is also the case that in most of the markets identified by Brazil (Argentina, Bolivia, India, Indonesia, Italy, Paraguay, Philippines, and Portugal), Brazilian prices have been consistently lower than US prices. Thus, rather than US upland cotton suppressing Brazilian prices, the data suggests that it is Brazilian cotton that is undercutting US prices.
13. Article 6.3(c) does not establish that serious prejudice may arise if the effect of the subsidy is any price suppression or depression. Indeed, were the term "significant" omitted from Article 6.3(c), it would be the case that any production subsidy that was granted on a per unit basis could be deemed to result in serious prejudice: any increase in production resulting from the subsidy would theoretically lead to some price effect. The use of the term "significant" prevents such theoretical or minor effects from rising to the level of serious prejudice.
14. Because "significant" modifies "price suppression" or "depression", it is the level of price suppression or depression itself that must be significant. One way of examining whether any alleged price suppression is significant would be to examine that degree or level in light of the price of the product itself. Another analytical tool that suggests itself is to look at the nature of the product’s price. Strong or frequent fluctuations in price would themselves tend to cut against a finding that any alleged suppression or depression is "significant", especially if the variability frequently brings the price of the product to a level at which the alleged suppression or depression (judged in light of that price) would not be significant. The United States notes that the price of upland cotton is highly variable, with frequent swings of substantial degree. Thus, this evidence relating to the price variability of upland cotton must be taken into account in any analysis of whether alleged price suppression or depression is "significant".
15. Brazil Misinterprets Article 6.3(d) on an Increase in World Market Share. Brazil misinterprets the phrase "world market share" in Article 6.3(d) as the share of world export trade. The plain meaning of the phrase "world market share" is not limited to export trade in products but includes all worldwide consumption – that is, the aggregate of all markets that make up the world. The United States is a "market" for upland cotton and part of the "world"; therefore, its domestic consumption forms part of the "world market" for upland cotton.
16. Context supports this reading of "world market share". For example, Article 6.3(a) identifies the "market of the subsidizing Member" as a relevant market from which a complaining Member’s exports can be displaced or impeded. Logically, then, the market of the subsidizing Member should also be relevant for determining the "world market share". Various provisions also provide context for not reading "world market share" as relating to "world export trade".151 Given repeated examples of the use of the terms "trade," "world trade," and "world export trade" in the covered agreements, the choice of the phrase "world market share" must be given meaning in accordance with the plain meaning of those terms.
17. The challenged US payments were only introduced in marketing year 2002; therefore, there can be no "trend" in US world market share with respect to those payments. Nonetheless, were the Panel to examine US world market share using data under the 1996 Act (consumption data, not the export data presented by Brazil), the criteria of Article 6.3(d) are not met.152
18. Brazil Has Not Demonstrated a Clear and Imminent Likelihood of Future Serious Prejudice. Although Brazil has presented evidence after the date of panel establishment (indeed, after conclusion of its three year period of investigation), it advises the Panel to consider more probative, for purposes of explaining price developments in marketing year 2003, the conditions in marketing year 1999 than the actual price developments in marketing year 2003. Brazil’s approach carries with it a high potential for erroneous findings by the Panel. Given current high market prices and the expectations embodied in futures prices that such high prices will remain through the course of the 2003 marketing year, it would appear that US price related payments (marketing loan payments and counter cyclical payments) will decline dramatically, contrary to Brazil’s assertions. In such a circumstance, it is difficult to see how challenged US payments would pose a clearly foreseen and imminent likelihood of future serious prejudice.
19. Brazil Has Misinterpreted GATT 1994 Article XVI:3. Contrary to Brazil’s arguments in this dispute, Brazil has previously agreed in a GATT plurilateral setting that GATT 1994 Article XVI:3 is limited in scope to export subsidies. Both the United States and Brazil were signatories to the Tokyo Round Agreement on Interpretation and Application of Articles VI, XVI, and XXIII of the General Agreement on Tariffs and Trade, commonly known as the Subsidies Code. Article 10 of the Subsidies Code is entitled "Export subsidies on certain primary products" and states (in paragraph 1): "In accordance with the provisions of Article XVI:3 of the General Agreement, signatories agree not to grant directly or indirectly any export subsidy on certain primary products in a manner which results in the signatory granting such subsidy having more than an equitable share of world export trade in such product . . . ." Thus, Article 10.1 of the Subsidies Code makes clear the understanding of both the United States and Brazil that GATT 1994 Article XVI:3 applies only to "export subsid[ies] on certain primary products". Therefore, Brazil has not made a prima facie case under Article XVI:3 on the basis of its arguments relating to all challenged US payments.
20. Brazil Errs in Asserting that Threat of Serious Prejudice Includes "More than an Equitable Share" under GATT 1994 Article XVI:3. There is no textual basis to assert that a claim of "threat of serious prejudice" under GATT 1994 Article XVI:1 may be founded on the "more than equitable share" language of GATT 1994 Article XVI:3. Neither Brazil nor the EC – Sugar Exports GATT panel report on which it relies cites any and that panel report does not appear to explain the basis for its decision to read the standard of Article XVI:3 into Article XVI:1. By way of contrast, footnote 13 to Article 5(c) of the Subsidies Agreement states that "[t]he term serious prejudice to the interests of another Member" is used in this Agreement in the same sense as it is used in paragraph 1 of Article XVI:1 of GATT 1994, and includes threat of serious prejudice". This footnote does not reference Article XVI:3, and as there is no "more than equitable share" prong to Article 6.3, there would not appear to be any basis to advance a threat of serious prejudice claim using that standard under Article 5(c) of the Subsidies Agreement. Footnote 13 states that "serious prejudice" in the Subsidies Agreement and GATT 1994 Article XVI:1 should be read "in the same sense". Therefore, footnote 13 provides a further textual basis for finding that a threat of serious prejudice claim under GATT 1994 Article XVI:1 may not be based on the "more than equitable share" language of Article XVI:3.
3. Brazil has failed to demonstrate the challenged US subsidies caused the effects complained of
21. The "Temporal Proximity" of US Payments and Low Cotton Prices Fails to Demonstrate that US Subsidies Caused Low Prices. Brazil has failed to make a prima facie case based on the assertion that large US outlays during marketing years with low prevailing upland cotton prices necessarily establishes causation. Brazil makes selective use of data to present a number of erroneous claims about US production or exports during a period of low and declining cotton prices. Brazil repeatedly begins the period of comparison with marketing year 1998 or ends it with marketing year 2001. Such comparisons are inappropriate for several reasons and can produce misleading results.153 The fact that high US payments were made when cotton prices were low does not establish causation.
22. Brazil Erroneously Alleges Production Effects from Decoupled Payments, Contrary to the Economic Literature. A fundamental error made by Brazil throughout its submissions and statements is to assert that decoupled payments are production distorting. Brazil’s conclusion that decoupled payments have had a large effect on cotton prices appears to be a direct consequence of Dr. Sumner’s faulty analysis – one that is inconsistent with the empirical and theoretical literature on such payments. Economic theory suggests that, if producers are seeking to maximize profits, the decision of which crop to plant is based on expected returns offered by the market or government payments above operating (variable) costs. Decoupled income support payments do not figure in this decision because such payments will be paid to the producer regardless of the programme crop that is planted or whether any crop is planted at all.154
23. The main impact of decoupled payments is likely on land values. In well-functioning markets, asset prices reflect expectations about the future returns from their ownership. The direct link between base acres for decoupled payments and the known programme benefits allowed the future stream of payments to be efficiently capitalized into land values. Thus, much of the increase in wealth from farm payments accrues to non operator landlords (Burfisher and Hopkins, 2003). Thus, the effects of increased wealth largely accrue to non operators, and any theoretical production effects are further minimized. In fact, land values set by sales and rental markets have diverged from commodity prices, suggesting that land markets have additionally capitalized the present and expected future value of government payments.
24. Data also indicate that decoupled payments, by increasing income and wealth, have allowed households to increase their leisure and reduce their work hours. If the downturn in labour comes from agricultural activities, the effect of such payments could be to decrease the household’s agricultural production, which would support world commodity prices. Data indicate that farm households that received decoupled payments in 2001 consumed more than farm households with similar incomes not participating in the programme. Thus, these data suggest that decoupled payments allow recipients to consume more out of income and may allow them to draw down savings that they typically carry as a precaution against income shortfalls.
25. Empirical studies have generally concluded that the effects of decoupled payments are minimal. For example, using an intertemporal Computable General Equilibrium model, Burfisher et al. (2003) estimate that production flexibility contract payments had "no effects on agricultural production in either the short run or the long run". These and other results are fully consistent with the fundamental requirement of Annex 2 of the Agreement on Agriculture that green box decoupled income support have no or at most minimal trade or production effects.
26. The available data also show large shifts in cotton acreage. Based on a preliminary review of a sampling of marketing year 2002 acreage reports, the United States estimates that nearly half (47 per cent) of farms receiving direct and counter cyclical payments in 2002 for upland cotton base acreage in fact planted no cotton at all. Preliminary estimates from the Farm Services Agency indicates that cotton producers enrolled upwards of 2 million acres for the 2002 Direct and Counter Cyclical Programme that had not been enrolled under the 2002 Production Flexibility Contract programme. Marketing year 1999 planted acreage deviated substantially from base acreage, both by region and by State.155 Thus, the data indicate that recipients of "upland cotton base acreage" decoupled payments plant alternative crops or no crops at all, and other farmers who do not hold upland cotton base acres choose to produce upland cotton.
27. Third Party Economic Studies Have Not Properly Modeled Cotton Production Decisions and Therefore Cannot Assist in Determining the Effect of US Subsidies on Cotton Production. Brazil has pointed to various third party economic studies which find price effects from US payments. Upon review, the United States concludes that they do not present relevant results because they generally suffer from two conceptual flaws. These fundamental flaws establish that these papers do not provide a basis to find a causal link between US payments and the effects of which Brazil complains.
28. First, several of these studies do not model the marketing loan programme appropriately.156 Simply put, if Dr. Sumner and FAPRI’s understanding of producer decisions is correct, then Brazil would have to agree that these papers do not properly model farmers’ production decisions and any potential impact of marketing loans on those decisions. As a result, these models do not provide insight into the question this Panel has been asked to examine.
29. Second, most of these studies do not distinguish between payments linked to production and payments decoupled from any requirement to produce, instead treating them as having equal impacts on production. Again, Brazil’s own expert recognizes that decoupled payments do not have the same impact as, for example, product specific marketing loan payments. Thus, Dr. Sumner’s own modelling of the impact of decoupled payments (with which the United States disagrees as contrary to the economic literature in ascribing any impact on production to these payments) indicates that these papers treat decoupled payments inappropriately.
30. Brazil’s "Total Costs of Production / Revenue Gap" is Meaningless and Cannot Establish Causation. Brazil’s so called "gap" between the average total cost of production per pound of cotton for US cotton producers and the revenue such producers received from the market is an economically meaningless measure and is based on a simplistic calculation that misstates both the revenue and cost sides of the calculation. Brazil’s revenue calculation is based on an erroneous representation of government support, especially crop insurance, decoupled payments, and Step 2 payments, and of market revenue.157 More fundamentally, the existence of a "gap" does not establish that US production would necessarily decline without the US payments Brazil has decided to challenge. For example, Brazil concedes that a substantial amount of US upland cotton in recent years was grown on non upland cotton base acreage, at the same time that government payments were allegedly "necessary" for US producers to remain in business. Brazil fails to explain how it accounts for these inconvenient facts that do not support its cost revenue gap theory.
31. On the cost side, Brazil’s use of average total cost of production for US cotton to make its revenue gap argument is the wrong figure to measure costs – it is operating costs, not total costs, that figure in production decisions. Brazil also has made no effort to update cost data that is based on a 1997 survey and so does not take into account any technological or structural changes that have occurred in the interim. Since 1997, significant technological changes have occurred in US cotton production, changes which are not reflected in the estimated costs of production, such as increased production in low cost regions and the introduction and adoption of genetically modified varieties of cotton with significantly increased yields while reducing pest control costs.
32. Finally, Brazil has used data from the International Cotton Advisory Committee (ICAC) to compare costs of production across countries arguing that the United States is a higher cost producer than many other countries. Even when good survey data are available for one country, using cost of production data to draw valid economic conclusions is fraught with difficulties. The comparison of costs across countries poses greater difficulties, rendering such comparisons invalid. The ICAC itself notes that the cost data it presents is not appropriate for making these kinds of cross country comparisons.
33. Brazil Has Failed to Make A Proper Analysis of Conditions Actually Faced by Producers in Making Production Decisions Using Futures Prices, Which Reveals No Expected Impact from Marketing Loans Except for MY2002. An analysis of the effect of marketing loan payments must begin with an understanding of farmers’ planting decisions. The United States agrees with Mr. MacDonald, Brazil’s expert on cotton markets, that the New York futures price provides the principal indicator of how market participants expect cotton prices to develop in the future. Unfortunately, Brazil’s other expert, Dr. Sumner, has ignored Mr. MacDonald’s testimony in modelling producers’ expectations of harvest season market prices by using "lagged prices" instead of futures prices. Had Dr. Sumner conferred with Mr. MacDonald, Dr. Sumner would have learned that "[t]he ‘ New York futures price’ is a key mechanism used by cotton growers . . . in determining the current market values as well as the contract prices for forward deliveries."
34. Comparing the planting time (February) New York futures price for the following harvest season (December delivery) to the marketing loan rate for upland cotton for each marketing year reveals that in every year but marketing year 2002, the planting time futures price was above the marketing loan rate. That is, New York futures prices indicated to producers that in every year but marketing year 2002 the return from the market would exceed the marketing loan rate. Thus, the marketing loan programme in marketing years 1999 2001 would not be expected to have had an effect on the decision to plant.
35. Only in marketing year 1999 does Dr. Sumner’s "lagged price" approach result in a value for producers’ expectations that equals or exceeds the futures price. In every other marketing year, the "lagged price" method significantly understates the harvest season price expected by producers and thus would distort an analysis of the effect of US subsidies. In fact, the use of "lagged prices" would lead to the erroneous conclusion that expected prices in every year but marketing year 1999 were below the applicable marketing loan rate. However, market price expectations actually were above the loan rate in every year but marketing year 2002. Thus, the use of "lagged prices" instead of futures prices to gauge producers’ price expectations at the time of planting in the specific years in which Brazil has alleged effects from US subsidies would seriously overstate the expected impact of US marketing loans. To the extent Brazil relies on Dr. Sumner’s analysis, which uses lagged prices rather than futures prices, Brazil’s analysis is fundamentally flawed.
36. The futures price data and "lagged price" data above also reveal that, despite declining market prices over the course of marketing years 1999 2002, market participants persisted in expecting prices to recover.158 Thus, Brazil’s reliance on actual market year prices to claim that US cotton plantings should have been declining ignores the fact that harvest season cotton futures prices at the time of planting were fairly stable from marketing year 1999 through marketing year 2001, even as futures for other competing crops fell in value.
37. In marketing year 2002, harvest season futures prices at the time of planting had fallen below the loan rate. In this marketing year, there is a least the possibility that producers were planting for the loan rate and not for the harvest season expected price. However, the decline in US planted cotton acreage was within the range of expected values given the decline in the harvest season futures price from the previous year. Had US producers been planting for the 52 cents per pound marketing loan rate, one would have expected to see only one tenth of the decline in planted acreage that actually occurred from marketing year 2001 to 2002.
38. Moreover, the per cent decline from marketing year 2001 to 2002 in US harvested acreage was very similar to (but larger than) the change in harvested acreage in the rest of the world. Despite the theoretical possibility that the marketing loan rate could have had some impact on planting decisions in marketing year 2002, the actual decline in US planted and harvested acreage suggests that US acreage levels were entirely consistent with price expectations and world market conditions. Thus, even in marketing year 2002, there is no evidence on this record that the marketing loan rate serves to insulate US producers’ planting decisions from market price movements. To the contrary, the evidence suggests that US producers do respond to changes in expected prices (for cotton and for other competing crops) and are as responsive if not more so than producers in other countries.