2.1 Raising previously litigated but unresolved issues in Article 21.5 DSU proceedings

The first scope issue in this case concerned certain US export credit guarantee programs. As noted above, the United States was found to have violated WTO rules because the fees charged for these programs did not cover their costs. As a consequence, it was found to have provided WTO-illegal export subsidies with regard to products not subject to WTO export subsidy commitments (i.e., so-called unscheduled products, including upland cotton) and with regard to one product – rice – where it had scheduled export subsidy commitments. In respect of two other scheduled products – pig meat and poultry meat – the Appellate Body reversed the Panel's conclusion that it had not exceeded its export subsidy commitments, but the Appellate Body did not complete the analysis of the issue, which meant that there were no findings regarding these two products. The Panel's conclusion that there was no violation of export subsidy commitments in other scheduled products was not challenged.

The United States claimed that it had corrected the export credit violations by eliminating two of the three export credit guarantee programs at issue and generally restructuring the remaining program. In challenging the restructured program – GSM 102 – Brazil argued that GSM 102 continued to constitute an export subsidy and requested findings in respect of unscheduled products and three scheduled products – rice, as to which a violation had been found in the original proceeding, and pig meat and poultry meat, as to which no findings were made, as described above. The United States asked the Panel to exclude pig meat and poultry meat from the scope of the proceedings. The Panel declined to do so. In its view, the fact that no violation had been found and that therefore no implementing measure needed to be taken did not mean that Brazil's claims were not related to ‘measures taken to comply’, as used in Article 21.5 DSU.Footnote ¹⁶ In reaching this conclusion, the Panel viewed the export guarantee arrangements for pig and poultry meat as measures separate from the implementing measure – the revised GSM 102 program – but it relied on US–Softwood Lumber IV Footnote ¹⁷ as supporting the notion that measures ‘with a particularly close relationship’ to an implementation measure could be considered to be within the scope of an Article 21.5 proceeding.Footnote ¹⁸ The Panel also noted that this situation was analogous to a case where the Appellate Body had ruled that a claim in respect of which a panel had exercised judicial economy in the original proceeding could be properly heard in an Article 21.5 proceeding.Footnote ¹⁹

On appeal, the Appellate Body rejected the Panel's analysis – in particular, its view that multiple measures were at issue – as inapplicable. In its view, the first step was to identify the ‘measure taken to comply’.Footnote ²⁰ In that regard, it took the view that the measure taken to comply was the revised GSM 102 program as a whole, and it noted various examples of where the United States had so described it.Footnote ²¹ The Appellate Body stressed that WTO Members have discretion to adopt either narrowly focused measures or broader measures when they implement DSB recommendations and rulings. It saw no reason why the definition of ‘measures taken to comply’ should be limited to what a Member was required to implement. Thus, since it viewed the new GSM 102 program as a single measure, it naturally encompassed the contested provisions on pig and poultry meat.Footnote ²²

In reaching its conclusion, the Appellate Body rejected the US arguments that its interpretation would lead to Members making the least possible changes to measures and would result in tangles of separate regimes, all done to avoid possible expansive challenges in Article 21.5 proceedings. The Appellate Body's position was simple: the choice of how to implement is up to the WTO Member concerned, and the DSU is neutral as to that choice.Footnote ²³ It also rejected the US argument that it was unfair to include pig and poultry meat because they would be subject to an expedited challenge and benefit from no reasonable period of time for implementation, even though they had never been subject to a finding of WTO inconsistency. Again, the Appellate Body saw these consequences as flowing from the US choice as to the scope of its implementation measure.Footnote ²⁴ Moreover, as the Appellate Body later noted, there are strong efficiency reasons for treating such related claims in one proceeding.

The Appellate Body did, however, place certain limits on the claims that could be raised. For example, it indicated that a complainant could not raise (i) claims that were raised in the original proceeding but were rejected or not pursuedFootnote ²⁵ and (ii) claims that could have been, but were not, raised in the original proceeding.Footnote ²⁶ It would be possible, however, to raise claims that had been pursued in the original proceeding but not decided for reasons of judicial economy or, as in this case, the Appellate Body's inability to complete the analysis of an issue. And, of course, to the extent that the new measure taken to comply implicates new WTO issues, those could be raised.

To us, the Appellate Body's position seems textually based and eminently reasonable, particularly for reasons of efficiency. An attempt to limit the scope of ‘measures taken to comply’ in the way suggested by the United States would likely lead to many instances of cases that belonged together being split between regular and compliance proceedings. That would obviously be inefficient, especially if different panelists were involved.Footnote ²⁷ While determination of whether a measure qualifies as a ‘measure taken to comply’ for purposes of Article 21.5 DSU may at times be difficult, that was not true in this case. The revision of the GSM 102 program was clearly a ‘measure taken to comply’.

2.2 Subsequent subsidy payments and Article 21.5 proceedings

The second Article 21.5 scope issue concerned the challengeability of subsidy payments made subsequent to 21 September 2005 (the date by which the United States was required to remove the adverse affects of or withdraw the subsidies found to be causing serious prejudice in the initial panel proceeding). As described above, in the original proceeding, payments under several US programs – the so-called Step 2 program, as well as the programs providing marketing loan, market loss assistance, and counter-cyclical payments – were found to have caused serious prejudice to Brazil. The United States terminated the Step 2 program, but continued payments under the marketing loan and counter-cyclical programs on the same terms as before. In the Article 21.5 proceeding, the United States argued that the findings in the original proceeding had been made in respect of payments made in the 1999–2002 marketing years and that it was required only to remove the adverse effects of those payments. According to the United States, later payments had to be challenged in a new proceeding.Footnote ²⁸ In this regard, it noted that Brazil had chosen to challenge the payments and not the programs ‘as such’. The Panel rejected this reasoning and took the position that the obligation in Article 5 SCM not to cause adverse effects through subsidies would not be met if a Member continued to provide payments under the same conditions as the original subsidy found to have caused adverse effects.Footnote ²⁹ The Panel also cited the ‘particularly close relationship’ language discussed in the prior section.Footnote ³⁰

On appeal, the Appellate Body reached the same result, but based it on somewhat different reasoning. It focused on the requirement in Article 7.8 SCM that ‘the Member granting or maintaining such subsidy [i.e., one found to cause adverse effects] shall take appropriate steps to remove the adverse effects or shall withdraw the subsidy’. The Appellate Body was of the view that the implementation obligation was not limited to the subsidies granted in the past that had given rise to the Panel's rulings. In its view, the word ‘maintain’ suggests that the Article 7.8 obligation is of a ‘continuous nature’, which means that it would apply to recurring annual payments subsequent to those that had been evaluated by the Panel.Footnote ³¹ Otherwise, it thought that the commitment to take appropriate steps to remove the effects would be virtually meaningless. The respondent could do nothing, and the effects of the previously granted subsidies would dissipate. while the same effects would continue as a result of the newly granted subsidies.Footnote ³²

The United States disputed this. In its view, Brazil could have obtained broader relief of the sort that it was seeking in the Article 21.5 proceeding if it had challenged the US subsidy program ‘as such’ or claimed that the program created a threat of serious prejudice in the future. The Appellate Body responded that there was little difference between an ‘as such’ challenge and that made by Brazil, because in both cases the past actual payments would have to be examined to determine if the program was causing serious prejudice.Footnote ³³ As to the possibility of alleging the threat of serious prejudice, the Appellate Body noted that that claim differed in substance from a claim of current serious prejudice. It also noted that in countervailing duty cases, a finding of serious injury entitled one to a remedy against future imports.Footnote ³⁴ Overall, the Appellate Body was concerned that accepting the US argument would make it difficult to obtain relief even after serious prejudice had been established and that the US position would undermine the DSU goal of prompt compliance.

On reflection, the Appellate Body seems to have come to the best result, although one can criticize its reasoning in some respects. For example, the use of the term ‘maintain’ in Article 7.8 SCM would seem to cover the situation where a subsidy has been granted in the past, but is still being paid out (e.g., a ten-year grant of $1 million per year), rather than the Appellate Body's interpretation, which seems to suggest that it covers future independent grants. Moreover, the relevance of its citation to CVD practice as context supporting its position is difficult to understand, as such different anti-subsidy provisions are concerned.Footnote ³⁵

Nonetheless, the US argument would unduly narrow the scope of Article 21.5 proceedings. In essence, it resembles the argument that if a measure is found to violate WTO rules, compliance is achieved by the simple act of withdrawing the contested measure and that Article 21.5 cannot be used to challenge a replacement measure. That position was long ago rejected.Footnote ³⁶ To us, it seems that if a WTO Member is found to have violated WTO rules, then it is appropriate to allow an Article 21.5 challenge to the ‘new’ regime that the Member puts in its place –whether that regime is a truly new regime, a somewhat revised regime, or essentially a continuation of the old regime.

Interestingly, the issue in this case is not so dissimilar from the issues that arise under antidumping cases involving the United States since the US system of retrospective assessment of antidumping duties on an annual basis means that by the time that a WTO proceeding – original or compliance – has been completed there will technically be a new US measure in place. The United States has to date also been unsuccessful in arguing that such ‘new’ measures are not challengeable in Article 21.5 proceedings.Footnote ³⁷

The US defense that Brazil should have challenged the US programs on an ‘as such’ basis is somewhat disingenuous. Such a challenge is a particularly difficult one to make when subsidies are involved. For example, the subsidies Brazil challenged under Article 5 SCM are not prohibited. They are a WTO problem only if they result in certain adverse effects, such as the serious prejudice to its interests that Brazil claimed in this case. Thus, an ‘as such’ challenge under Article 5 would have to show that the program at issue inherently led to serious prejudice, which would involve, as noted by the Appellate Body, looking at exactly the same material as was considered in this case. The difference would be that a stronger finding would have to be made – the program inherently causes serious prejudice, as opposed to a finding that the program has caused serious prejudice in the specific years examined. Since the existence of serious prejudice will depend on so many other market factors that will change over time, it will not be easy to conclude that a subsidy program inherently results in serious prejudice.

Basically the issue comes down to two questions: What is the nature of the obligation found in Article 5 SCM? If violated, what sort of remedial action is required? The obligation is not to cause ‘adverse effects’ such as ‘serious prejudice’. That obligation is a continuing one. Once such adverse effects are found, the respondent is required by Article 7.8 SCM to remove the adverse effects or withdraw the subsidy. Since WTO remedial obligations are commonly viewed as prospective in nature, this obligation would seem to require the subsidy program to be withdrawn or the adverse effects of the program to be removed. If the subsidy program is only modified, as here, the implementation issue will be whether the revised program will still cause serious prejudice. As such, that issue is an appropriate one for Article 21.5 DSU proceedings. And that is the result achieved here. Indeed, it would make little sense to say that the nature of the finding and the obligation to remove the adverse effects had no future implications for the programs found to have caused the problems.

2.3 Summary

In our view, the Appellate Body's resolution of the DSU Article 21.5 scope issues seems eminently reasonable.

3.1 GSM 102 export credit guarantees

The third issue on appeal was the US claim that the evidence did not support a conclusion that its continuing export credit guarantee program – GSM 102 – still operated at a loss. The Panel examined four items of what the Appellate Body characterized as quantitative evidence in its assessment of whether or not the GSM 102 program was operated at a loss.

1. (i) US estimates showing losses for the 2006–2008 GSM 102 guarantees in the 2007 and 2008 US budgets;

2. (ii) US re-estimates for the 1992–2006 period that showed no losses overall;

3. (iii) consolidated financial statements of the granting agency (CCC), which showed an overall liability for the guarantee programs; and

4. (iv) cash accounting data compiled by Brazil for the 1992–2005 period, which showed a loss for the programs.Footnote ³⁸

In its evaluation of this evidence, the Panel relied in particular on item (i), which it found was confirmed by items (iii) and (iv).Footnote ³⁹ It discounted item (ii), offered by the United States, because even though that data had shown some older cohorts (the guarantees granted in a given year) did not incur losses, the data did not demonstrate that more recent years would not result in losses. Moreover, the Panel noted that the re-estimates were only estimates.Footnote ⁴⁰ Although the Panel did not comment on it, the re-estimates seemed to continue to show recent cohorts operating at a loss (i.e., the overall profitability of the program as re-estimated by the United States fell significantly by including the 2003–2006 cohorts).Footnote ⁴¹

For the Appellate Body, the Panel's evaluation was seriously flawed. With the exception of two cohorts that had closed out profitably (1994 and 1995), all the evidence consisted of estimates and projections. Yet the Panel seemed to dismiss the US re-estimates because they were merely estimates and yet gave credence to the other estimates. Moreover, it did not attempt to reconcile the various estimates or explain why such a reconciliation was not possible. In the Appellate Body's words: ‘The Panel's internally incoherent treatment of the same class of quantitative evidence thus vitiates the conclusion it drew based on the financial data submitted by the parties.’Footnote ⁴² Thus, the Appellate Body concluded that the Panel had not made an ‘objective assessment’ of the matter and facts as required by Article 11 DSU.Footnote ⁴³

Having reversed the Panel's conclusion based on the quantitative evidence, the Appellate Body turned to whether it could complete the analysis of the issue. In this regard, it essentially concluded that the evidence offered by the United States and Brazil was of similar reliability, but supported irreconcilable results.Footnote ⁴⁴ It noted that its attempt to reconcile some of the conflicting estimates at the appellate hearing was unsuccessful.Footnote ⁴⁵ Thus, the Appellate Body concluded that the quantitative evidence did not resolve the issue of whether the GSM 102 program operated at a loss.

This conclusion did not end the matter, because the Panel had also based its finding that the US program did not cover its overall costs on the structure, design, and operation of the GSM 102 program.Footnote ⁴⁶ In that regard, the Panel had noted (i) that there was a statutory cap of 1% on GSM 102 fees; (ii) that the fees did not increase in respect of risk as fast as they did under certain other US export credit programs (probably because of the cap); and (iii) that foreign obligor risk was not reflected in the fees. The Appellate Body rejected the US criticisms of the Panel's analysis of these factorsFootnote ⁴⁷ and concluded that it was ‘not persuaded that the Panel erred in finding that “the GSM 102 program is not designed to cover its long term operating costs and losses”’.Footnote ⁴⁸

Given its foregoing analysis, the Appellate Body concluded that ‘The Panel's finding on the structure, design and operation, in light of the two plausible outcomes with similar probabilities that emerge from the quantitative evidence, provides a sufficient evidentiary basis for the conclusion that it is more likely than not that the revised GSM 102 program operates at a loss.’Footnote ⁴⁹

This conclusion is defensible – but only barely. The key evidence was the so-called quantitative evidence. The Appellate Body characterized that evidence as supporting two plausible outcomes, but since the outcomes were polar opposites, it would have been more appropriate to characterize the quantitative evidence as inconclusive. In that situation, one can legitimately ask whether indirect evidence regarding the design of the program should be sufficient to conclude that a violation had occurred. After all, evidence on design, structure, and operation is only indicative of a potential problem with meeting costs; the key evidence is the quantitative evidence, which was inconclusive. One can argue that Brazil or the Panel needed to do much more to explain the inconsistencies in that key evidence. The changes made by the United States should have reduced its losses (if any). The fact that the only cohorts that had closed in the past were profitable certainly could be taken to suggest that those changes would make it likely that the more recent cohorts would also be profitable. At a minimum, some more detailed analysis of the differences between the terms under which the profitable cohorts and the current cohorts operated should have been undertaken. These unexplored areas certainly raise some serious questions about the Panel and the Appellate Body's conclusions. Ultimately, of course, it is not the role of the Appellate Body to revisit a Panel's factual findings, but where it has reversed part of the findings supporting a specific factual conclusion, this raises questions about the conclusion itself. Thus, while the conclusion may have been correct,Footnote ⁵⁰ the issue presented a very close call and a more thorough analysis of the evidence should have been undertaken. This case is one where a remand would have been useful.

3.2 Serious prejudice

The United States also appealed the Panel's finding that its domestic subsidies to cotton producers caused serious prejudice to Brazil through the suppression of cotton prices internationally. In that respect, the United States raised two particular points.Footnote ⁵¹

First, the United States objected that the Panel did not indicate the degree of price suppression that it found to be significant. In the original proceeding, the Panel had used a ‘binary’ approach in that it first determined whether price suppression had occurred and then considered whether it was significant. In this Article 21.5 proceeding, the Panel followed a unitary approach in which it considered three issues together: the existence of price suppression, its significance, and its causation.Footnote ⁵²

The Appellate Body considered that a unitary approach was acceptable given the conceptual nature of the inquiry – which essentially involved an evaluation of a comparison of what did happen and what would have happened if there had been no US subsidization.Footnote ⁵³ In that regard, the Appellate Body noted that the Panel had used an economic model proposed by Brazil in order to undertake the counterfactual assessment. Using that model, but substituting US parameters for the Brazilian ones, the Panel found that US production was 12–18% higher with the subsidies and that elimination of the subsidies would have increased prices by 1.4–2.3% (compared to an increase of 8.2–8.9% using Brazil's parameters).Footnote ⁵⁴ The Panel did not choose between the US and Brazilian parameters, but suggested that even with the US numbers, there was significant price suppression because of the commodity status of the product involved:

Noting that the United States had not challenged the accuracy of this statement, the Appellate Body concluded: ‘Thus, the range of price effects resulting from the simulations would fall within the Panel's view of what constitutes “significant” price suppression in the specific context of the world price of upland cotton.’Footnote ⁵⁶

One would have wished for more analysis by the Panel than speculation that a small suppression of prices could be significant. Indeed, one could argue that the Panel never found that price suppression was significant in fact, but only speculated that it could be. Nonetheless, given the role of the Appellate Body, which is not to review factual questions, its decision not to disturb the Panel finding may be defensible.

Second, the United States objected to the Panel's conclusion because it had failed to consider the influence of China on cotton prices. China has become the major cotton importer. For us, as well as the Appellate Body, the US evidence that the lack of transparency about Chinese demand had a price-suppressing effect was not all that persuasive.Footnote ⁵⁷ However, the US argument was theoretically sound; it was a matter of whether there was evidence to support it. Nonetheless, the Panel's analysis of this issue was rather terse. In its entirety, it read:

Given the weakness of the US evidence, the Appellate Body essentially concluded that the Panel had done enough.Footnote ⁵⁹

In respect of both of these evidentiary issues, the difference in the level of deference accorded by the Appellate Body to panel factfinding and analysis as opposed to national authority factfinding and analysis is striking. In particular, one is struck by the stark difference between the Appellate Body's approach to non-attribution analysis here and its approach to non-attribution analysis by national authorities in trade-remedy cases. In apparent recognition of this gulf, the Appellate Body stressed that the ‘serious-prejudice’ provisions of the SCM Agreement do not contain the detailed language on causation and non-attribution applicable to CVD investigations under the same agreement.Footnote ⁶⁰ Nonetheless, there is an explicit causation requirement in Article 5(c) SCM (not mentioned by the Appellate Body) and, according to the Appellate Body, there is a causation requirement inherent in Article 6.3 SCM. Also inherent in these causation requirements is a requirement not to attribute price-suppression effects attributable to other causes than the subsidy to the subsidy. While the lack in Articles 5 and 6 SCM of the more specific language found in the trade-remedy provisions may argue for a bit more flexibility for a panel's consideration of the issue, no reasonable degree of flexibility can justify the difference. The Panel's analysis here does not at all measure up to the Appellate Body's language, for example, in Japan–Hot-Rolled Steel:Footnote ⁶¹

While one could argue that the apparent difference in approach can be justified on the ground that national investigating authorities may tend to be biased, whereas WTO panels are more objective, the Appellate Body expressly rejected the idea of a different standard, as noted below. Nonetheless, in this case the difference in approach is real (and significant under any evaluative standard).

It is noteworthy, however, that the Appellate Body's position with respect to the role of economic analysis, and in particular of economic models, has evolved significantly since the original proceeding.

In the original proceeding, Brazil submitted a quantitative simulation of an economic model in support of its ‘serious-prejudice’ arguments, which showed that, but for the US subsidies to the US upland cotton industry, world cotton prices would have been higher by an average of 12.6% during the period 1999–2002. The United States objected to the accuracy and adequacy of the simulation and to the fact that Brazil provided the model itself.

The Panel decided to ‘take note’ of the outcomes of the simulation, but refused to rely ‘upon the quantitative results of the modeling exercise – in terms of estimating any numerical value of the effects of the United States subsidies’.Footnote ⁶² It further indicated that ‘Without prejudice to the relevance or utility of such simulations generally to a serious-prejudice analysis under Part III of the SCM Agreement, we would point out our particular concern here, in ensuring procedural fairness between the parties and the reliability of the evidence, that the model was not equally accessible to the parties and, as relevant, to the Panel in these proceedings.’Footnote ⁶³

On appeal, neither party contested the Panel's decision to take the modeling results ‘into account where relevant to our analysis … and [to attribute] to them the evidentiary weight we deemed appropriate’.Footnote ⁶⁴ The United States, however, contended that the Panel failed to include in its analysis the supply response of third countries that would have resulted in the absence of US subsidies. For its part, Brazil argued that the Panel did take this factor into account, since the models it used incorporate such supply response. Both parties agreed on the features of the models, but disagreed on ‘whether the Panel took into account supply responses of third countries, as reflected in these models or otherwise.’Footnote ⁶⁵

The Appellate Body was unable to decide on what is ultimately an empirical issue. It simply indicated:

This amounted to accepting the Panel's decision, while at the same time faulting it for insufficient economic analysis. In the ALI project's examination of those reports, the authors (including one of the present coauthors) concluded:

In this Article 21.5 proceeding, the Panel seems to have heeded the criticism by the Appellate Body and the advice of the ALI project. Here Brazil again submitted an economic model to simulate the effects of US subsidies on the world market for upland cotton, a fact welcomed by the Panel. In its entirety, the relevant paragraph read:

In its evaluation, the Panel noted the advantage of the modeling approach used by Brazil, but was ‘also mindful of the criticism by the United States that Brazil's model “has no foundation within economic circles”’.Footnote ⁶⁹ Furthermore, it noted that the magnitude of the impact of US subsidies on the world cotton price depends on the parameter values assigned to the model, and that the two parties presented arguments in favor of different parameter values. The Panel decided not to take a position on which parameter values were better, but nonetheless took ‘note of the fact that price suppression has been the outcome of all the simulation results whether one uses the parameter values proposed by Brazil or [those] proposed by the United States’.Footnote ⁷⁰

Hence, contrary to the original proceeding, where the Panel simply took the model into account, here the model simulation results were central to the Panel's finding of price suppression due to the US subsidies. However, by refusing to take a position on the parameter values, the Panel was still unable to reach a judgment, simply based on the model, about the magnitude of the impact of US subsidies and, therefore, about whether price suppression was ‘significant’. For this, as noted above, it needed to invoke the fact referred to previously that upland cotton is a ‘basic and widely traded commodity’.

On appeal, the United States claimed that the Panel had failed to determine the degree of price suppression that it found to be ‘significant’. In its evaluation of the US claim, the Appellate Body started by focusing on price suppression. It credited the Panel for its extensive discussion of the model simulations presented by both parties, insisting that such simulations are essential in serious-prejudice cases: ‘Because the examination of price suppression necessarily involves an analysis of what would have been the case in the absence of an intervening event, modeling exercises are likely to be an important analytical tool that a panel should scrutinize.’Footnote ⁷¹

However, the Appellate Body faulted the Panel for refusing to take a stand on which parameters were more appropriate to measure the effect of the subsidies at issue on prices, or on the appropriateness of the model itself. In support of its view that ‘the Panel could have gone further in its evaluation and comparative analysis of the economic simulations and the particular parameters used’,Footnote ⁷² the Appellate Body noted:

The Appellate Body had already noted in an earlier footnote that ‘where a panel operates as an initial trier of facts, such as this one, it would … be expected to provide reasoned and adequate explanations and coherent reasoning’.Footnote ⁷⁴ The Appellate Body explicitly stated that this requirement was similar to what a panel is expected to do when examining determinations of domestic investigating authorities in trade-remedy cases. Perhaps this statement indicates that the Appellate Body in the future will hold panel determinations in these circumstances to the higher standard it expects of national authorities. It did not do so in this case.

Nor did the Appellate Body suggest how panel determinations in future ‘serious-prejudice’ cases could achieve a higher standard of economic expertise than in the past. Perhaps the earlier advice by the ALI project could usefully be repeated here, namely that: