3. "Benefit" under Articles 1.1 and 3.1(a) of the SCMAgreement Questions to both parties 95. Brazil has taken the position that "different parties to a transaction involving a GSM 102 ECG derive different benefits from the GSM 102 ECG, each of which is potentially subject to assessment under Article 1.1(b) of the SCM Agreement" and has indicated that it is, in this proceeding, "primarily concerned" with the benefit received by the US exporter in the form of below-market fees (para. 404, Brazil's Rebuttal). The United States has challenged Brazil's approach of focusing on fees to the exclusion of other elements of the total cost of the loan. Please explain, referring to the provisions of the SCM Agreement and WTO jurisprudence (if any applicable), your position as to whether: (1) export credit guarantees and other types of subsidies may involve more than one type of benefit and/or recipient; (2) whether it is up to the complaining Member to decide which benefit it chooses to challenge. 203. Contrary to the United States' assertion in its response, Brazil does not ask the compliance Panel to "ignore" Article 14(c) of the SCM Agreement in its assessment of the "benefit" flowing from GSM 102 ECGs under the amended GSM 102 fee schedule.354
204. Based on Appellate Body jurisprudence, Brazil has noted the Appellate Body's ruling that Article 14 must not be applied "rigidly", even in disputes under Part V of the SCM Agreement, to which the provision explicitly applies.355 The Appellate Body has held that even in disputes under Part V of the Agreement, the provision must be applied flexibly, to account for the "factual circumstances" at hand.356
205. Preserving this flexibility is even more important in disputes under Part II of the SCM Agreement. In a dispute under Part II, only the existence of some "benefit" under Article 1.1(b) must be established, and not the quantity of that benefit. In a passage cited by the United States in these proceedings, the original panel considered that imposing the "quantitative focus and more detailed methodological obligations of Part V" of the SCM Agreement on disputes preceding under Part III of the Agreement was not appropriate, even where assessing the "magnitude" of the subsidy was required.357 For a dispute under Part II of the Agreement, where no assessment of the "magnitude" of the subsidy is required, much less the calculation of a countervailing duty rate, there is no basis whatsoever to impose the "quantitative focus and more detailed methodological obligations of Part V …. "358 206. In its 2 April response to question 100, Brazil demonstrated the proper role for Article 14(c) in these proceedings. The "factual circumstances"359 in these proceedings under Part II of the SCM Agreement are such that a "particularized showing"360, utilizing data concerning "individual loan costs and fees" or "comparable commercial loans and their terms"361, is not necessary. Proof of some "benefit" can be established without the type of evidence on which the United States insists.
207. Indeed, in proceedings involving claims against government guarantees under Parts II and V of the SCM Agreement, panels have already twice ruled that evidence of the type insisted on by the United States is not necessary to show that a benefit exists.362 Rather, evidence far short of that produced by Brazil in these proceedings (described in Brazil's 2 April response to question 100363) was accepted by these two panels as sufficient to satisfy Article 1.1(b) of the SCM Agreement, as informed by Article 14(c). Each of these two panels effectively accepted that, in principle, below-market fees for government guarantees translate into lower costs for commercial credit, and thus a lower total cost of funds – a conclusion rooted in the basic principles of financial economics addressed by Brazil in its 2 April response to question 100.364
208. This does not amount to "ignoring" Article 14(c), as the United States asserts.365 Rather, it amounts to putting Article 14(c), and the evidence required to satisfy it, into the particular context of these proceedings.
96. The parties differ as to whether different types of loans can be compared as long as they have the same "average life." What support (economic literature, etc.) exists for your position on this issue? 209. The United States' position is that two-year bullet loans and three-year amortizing loans are equivalent, as long as one averages out the lives of the two loans.
210. Brazil's position is that two-year bullet loans and three-year non-amortizing loans are not equivalent, and that averaging out the lives of the two loans does not make them so, in light of the varying risks of default.
211. In its response to question 96, Brazil provided support for its position. Instead of providing support for the U.S. position, the United States instead ascribes to Brazil positions that Brazil has not taken, and argues against those imaginary positions. The United States' response is nothing more than an effort to mislead the compliance Panel, and to deflect attention from the evidence Brazil has offered to prove that bullet and amortizing loans do not have equivalent spreads even if they have the same average lives.
212. With reference to a simplified example addressed at paragraphs 220-222 of Brazil's Oral Statement, the United States argues that Brazil's position requires the compliance Panel to assume that risk "increases over time", such that there is "much higher risk in year 3 of the GSM 102 transaction." The United States admonishes Brazil for providing no basis for this "assumption", which it argues is "not even sound as a theoretical matter."366
213. This characterization of Brazil's position is so misleading as to be absurd. Brazil has not asked the compliance Panel to assume this scenario as a matter of fact, but merely to consider it as one hypothetical example. Other examples, involving alternative patterns of default risk, are also considered in paragraphs 31-32 of a statement by Professor Sundaram included as Exhibit Bra-686 to Brazil's 2 April responses. Each of these paragraphs includes two examples; in one example in each paragraph, default risk increases over time, while in the other example in each paragraph, default risk decreases over time. As the examples show, the spread on amortizing loans could be greater or lesser than the spread on bullet loans, depending on the pattern of default risk. That is Brazil's position, and it is mathematically proven by Professor Sundaram in Exhibit Bra-686.
214. In paragraphs 221-222 of its 2 April response, the United States provides a reference that in fact supports Brazil's position. The United States cites a paper finding that for seven-year loans, the likelihood of default "in general" increases over time, rather than decreases.367 This substantiates Brazil's argument that since default risk is not constant from year to year, bullet and amortizing loans will, in general, have different spreads.368
215. Brazil also addresses three other issues raised in the United States' response to question 96: (i) the use of the average life concept in the context of mortgage-backed securities; (ii) the use of the average life concept in pricing swaps in terms of U.S. Treasuries with the same average lives; and, (iii) the use of the average life concept in pricing for an amortizing loan obtained by adding a fixed spread to an index.
Average life in the context of mortgage-backed securities 216. In paragraph 219, the United States suggests that "the concept of average life is commonly used for price comparability of investments and debt-instruments of various terms." The first example in paragraph 219 refers to mortgage-backed securities ("MBS"). Brazil makes two observations about this example.
217. First, and most importantly, the reference does not address, much less establish, the comparability of bullet and amortizing loans with the same average lives, which is the question raised by the Panel. Indeed, in discussing the relevance of the average life concept to MBS, the United States makes no reference at all to pricing. The United States does not, for example, suggest that the price of an MBS is the price of a similar bullet security with the same average life.
218. Second, the reference to MBS relies on an incorrect understanding of why average life is used for MBS. The biggest risk faced by the holder of an MBS is pre-payment risk, or in other words, the risk that some of the individual mortgages underlying the MBS will be pre-paid by the borrowers.369 Thus, the maturity of the MBS is not, in itself, adequate to describe the instrument. Two MBS may have the same maturity, but if one has higher pre-payment risk than the other (say, because the mortgage coupons underlying it are higher), then the cash flows will last for a shorter period of time. To restore some kind of comparability between different MBS, investors use models that attempt to predict the volume of pre-payments, and calculate an average life accordingly.370 Such an average life calculation is unnecessary if there is no prepayment risk.
Average life in pricing swaps 219. In paragraph 219 of its response, the United States quotes an article stating that swap dealers "'routinely price the fixed rate side of an interest rate swap as a spread over United States Treasuries of a similar average life.'"371 This statement does not provide support for (or even address) the United States' assertion that two-year bullet loans and three-year amortizing loans with the same credit quality have identical credit risk whenever they have the same "average lives".
220. The article cited by the United States describes a situation in which an institution holding fixed-rate amortizing mortgage debt funded with the issue of floating rate short-term debt seeks to hedge interest rate risk by using a "swap". A swap is an exchange of cash flows computed at a floating rate of interest, for cash flows computed at a fixed rate of interest. The institution in question looks to receive the floating-rate payment from the swap seller and make the fixed-rate payment to the swap seller so that, taken together with the mortgage that it holds and the floating-rate debt issued to fund that mortgage, the institution's interest-rate exposure is only fixed rate.
221. To identify the fixed rate in a swap, dealers generally add a spread to the Treasury yield on a bond with the same maturity as the swap. However, where the notional principal is amortizing, as in the example offered in the article quoted by the United States, a problem arises – while the notional principal on the mortgage is amortizing, Treasury bonds are bullet bonds. Some adjustment is required to equate the differing maturities of amortizing and non-amortizing debt. This adjustment is addressed in endnote 4 to the article: "[a]verage life is used by swap dealers to equate the maturities of nonamortizing Treasury debt and amortizing mortgage debt."372
222. The compliance Panel, however, has asked for support in the literature for the U.S. assertion that non-amortizing and amortizing debt with the same credit quality have identical credit risk whenever they have the same "average lives". This proposition is not even addressed by the example offered in the article quoted by the United States, because the mortgage debt and the Treasury bond in the example have different credit qualities, despite the fact that they have the same "average life". The article cited by the United States does not, as a result, provide support for the U.S. assertion that two-year bullet loans and three-year amortizing loans with the same credit quality have identical credit risk whenever they have the same "average lives".
Average life in pricing an amortizing loan
obtained by adding a fixed spread to an index 223. In paragraph 220 of its response, the United States offers another example of the "average life" concept, in this instance suggesting pricing for an amortizing loan obtained by adding a fixed spread to the appropriate index.
224. The example cited by the United States in paragraph 220 and the website cited at Exhibit US 151 do not, as requested by the Panel, offer support for the U.S. position that two-year bullet loans and three-year amortizing loans are equivalent, as long as one averages out the lives of the two loans. Indeed, Exhibit US-151 and the example included in paragraph 220 of the U.S. response do not address the correctness of the U.S. position at all.
225. The U.S. example demonstrates how a single firm in the industry, Bond Street Capital ("BSC"), approaches the pricing of its amortized loans. The example does not involve the use of "average lives" to put bullet loans of a particular maturity and amortizing loans of a different maturity on equivalent footing, therefore enabling a comparison of the prices for the two instruments. The example shows that BSC indexes its loans to average lives of U.S. Treasuries or swaps. It does not suggest that BSC considers an amortizing loan with an average life of m years to be equivalent in spread to a bullet loan with a life of m years. Indeed, the example does not even mention how BSC would price a bullet loan. Neither the example, nor the material included in Exhibit US-151, exclude that BSC would add different spreads to the index for bullet and amortizing loans.
226. Brazil also notes that the "average lives" of the loans included in the example do not comport to the common understanding of that concept. In paragraph 220 of the U.S. response and Exhibit US 151, BSC asserts that a "15/15" loan has an average life of 9 years. That is simply not accurate. A "15/15" loan has an average life of 8 years. One-fifteenth of the capital is repaid in each of the fifteen years, so the average life is (1/15) * (1+2+ . . . +15), or 8 years. Similarly, BSC asserts that the average life of a "20/20" loan is 12 years, but it is actually 10.5 years ((1/20) * (1+2+ . . . +20) = 10.5). Even if the example served to support the position for which it is offered, it is not reliable, as the average life computation is incorrect.
Questions to the United States 97. Assuming that the Panel accepts the United States' argument that "benefit" is to be assessed on the basis of the "total costs of funds", what do you consider Brazil must establish in order to meet its burden of proof in that respect? Must Brazil prove that a benefit is conferred in all instances (all transactions and all recipients)? In most instances? 227. In its comment on the U.S. response to question 44, Brazil has addressed the United States' professed confusion over whether Brazil is "challenging" the GSM 102 program "as such", or instead "particular guarantees" under the GSM 102 program.373 As noted in that comment, Brazil has shown that the amended GSM 102 program is an export subsidy. Separately, Brazil has also shown that the GSM 102 program is "applied", via individual GSM 102 ECGs for particular products, in a manner that results in circumvention of U.S. export subsidy commitments.
228. In any event, it is not necessary in WTO dispute settlement to prove that a measure violates WTO law in "all" instances. As the panel in U.S. – Export Restraints held, it suffices that a measure violates WTO laws in some instances.374
98. Does the United States dispute the accuracy of Brazil's comparison of GSM 102 fees with Exim Bank fees? Does the United States agree that ExIm Bank and GSM 102 guarantees are (at least in certain circumstances) similar or comparable? 229. In its First Written Submission, Brazil demonstrated that GSM 102 fees are below the fees charged by the United States' Export-Import Bank ("ExIm Bank") for financial instruments that are in key respects similar to GSM 102 ECGs – ExIm Bank's Letter of Credit Insurance ("LCI") (which serves as a reference point for GSM 102 ECGs issued for tenors of up to 360 days), and ExIm Bank's Medium-Term Export Credit Insurance ("MTI") (which serves as a reference point for GSM 102 ECGs issued for tenors exceeding 360 days).375 The results of this comparison between GSM 102 and ExIm Bank fees are tracked in Exhibits Bra-536 (for transactions involving annual repayment of principal) and Bra-537 (for transactions involving semi-annual repayment of principal).
230. The United States argues that GSM 102 ECGs are not similar to ExIm Bank LCI and MTI products, and are therefore not suitable for comparison, for five reasons.
231. Before addressing each of these five reasons, Brazil notes the irony of this situation. Brazil has demonstrated that there is no credit protection product available in the marketplace that is comparable to a GSM 102 ECG.376 Nonetheless, in addition to that showing, Brazil undertook a comparison between GSM 102 fees and fees for comparable non-market products offered by ExIm Bank. In rebuttal, the United States has not offered a single example of a credit protection product available in the marketplace. Brazil's fundamental position – that GSM 102 ECGs confer "benefits" per se because they have no parallel at market – remains unrebutted by the United States with anything that resembles evidence, and indeed echoes the position the United States has taken in other disputes.377
232. With that important qualification, Brazil addresses each of the five reasons offered by the United States in support of its argument that ExIm Bank LCI and MTI products are not suitable for comparison with GSM 102 ECGs.
First U.S. criticism:
No ExIm Bank MTI cover for agricultural product transactions 233. First, the United States observes that ExIm Bank's MTI is not available for agricultural products, and that it is not therefore comparable to GSM 102.378 234. Brazil has already alerted the Panel to this distinction. As Brazil noted in its First Written Submission379, ExIm Bank cover is indeed not, in normal circumstances, available for agricultural export transactions with tenors exceeding 180 days or, exceptionally, 360 days.380 The minimum tenor for a transaction to qualify for MTI cover is 360 days. Thus, MTI cover is not available for agricultural products.
235. Brazil fails to see how this criticism aids the United States' defense. All the U.S. observation establishes is that it is not even possible to identify a non-market instrument distorted by government intervention to serve as a benchmark for GSM 102 ECGs issued for tenors exceeding, at most, 360 days, let alone a market instrument. The compliance Panel will recall that in Congressional testimony, the Chairman of CoBank stated that GSM 102's three-year tenor "is critical to the program's success," and that the program "provides for tenors that are typically unavailable in the market and this is a crucial strength of the program."381
236. In any event, as Brazil has already explained and the United States has ignored, the fact that a financial instrument does not apply to agricultural products does not undermine the comparability of its fees to GSM 102 fees, and in fact reveals that the ExIm Bank comparison is overly generous to the United States.382 237. There is no reason why guarantees for agricultural products should be less risky than credit protection for transactions involving capital goods and products. If anything, making agricultural products subject to ExIm Bank's MTI cover would increase MTI fees. In an agricultural export transaction, the value of the product sold holds little value for the guarantor as a security against default by the foreign obligor. Shortly after sale, an agricultural commodity is likely to be (and short of preservation may indeed have to be) consumed; the value of any security interest in the commodity will be short-lived, and usually far less than 360 days. In contrast, an industrial product is generally more durable and lasting, and as such is more susceptible to use as security for performance by the foreign obligor, should the guarantor wish to do so. The availability of this option may lower the risk associated with default faced by the guarantor and, as a consequence, lower the fees necessary to offset that risk. Where this option is not viable, the loss in the event of default is greater, as will be the fees necessary to offset the corresponding risk.
Second U.S. criticism:
No ExIm Bank MTI cover for repayment terms between 360 days and 2 years
238. Second, the United States appears to suggest that unlike GSM 102, which applies for repayment terms of up to three years and in "hundreds of … transactions" for less than two years, ExIm Bank's MTI is only available for repayment terms of more than two years.383 While not expressed clearly, the U.S. allegation appears to be that Brazil has not offered a valid comparison for GSM 102 fees for transactions involving repayment terms of between 360 days and two years.
239. To isolate with precision the U.S. criticism, Brazil notes that the criticism does not affect the comparability of GSM 102 fees and ExIm LCI fees for transactions involving repayment terms of 360 days or less. Nor does it affect the comparability of GSM 102 fees and ExIm MTI fees for transactions involving repayment terms of between two and three years.
240. The United States assertion that MTI cover is not available for repayment terms of between 360 days and two years is, to put it bluntly, untruthful. ExIm Bank documents expressly state that MTI coverage is available for repayment terms of 1-5 years.384 The ExIm Bank calculator with which Brazil generated the MTI fees tracked in Exhibits Bra-536 and Bra-537 also specifically permits the user to control for repayment periods of between 360 days and two years (and beyond).385
241. In any event, the vast majority of GSM 102 ECGs are indeed issued for transactions involving repayment terms in excess of two years – to which the United States acknowledges ExIm Bank's MTI policies apply. The United States notes that "hundreds of GSM-102 transactions, totalling hundreds of millions of dollars are in fact for terms of less than 2 years."386 When the figures offered by the United States are compared to total GSM 102 ECGs issued in FY 2003-2006 (as well as part-year totals for FY 2007), however, the compliance Panel will note that they compose approximately 8 percent of the nearly $10 billion in GSM 102 ECGs issued in that period.387 Thus, the vast majority of GSM 102 ECGs – 92 percent – are issued for repayment terms of over two years, which even the United States acknowledges matches the repayment terms for ExIm Bank MTI cover.
242. The United States deliberately withheld its comments on Brazil's ExIm Bank comparison until this late stage of the proceedings, having foregone the opportunity to do so in the 7 submissions it made prior to its 2 April answers and subsequent to Brazil's proffer of the comparison in Brazil's First Written Submission. When it finally offers comment, it misleads the compliance Panel with inaccurate factual statements contradicted by the very U.S. government documents and data on which the United States relies. As well as being tardy, the United States' criticisms of Brazil's ExIm Bank comparison are not credible.